Descriptive Physical Oceanography (6º Edi.) - L. D. Talley - G. L. Pickard - W. Emery - J. H. Swift

Descriptive Physical

Oceanography:

An Introduction

Sixth Edition

DESCRIPTIVE

PHYSICAL

OCEANOGRAPHY:

AN INTRODUCTION

SIXTH EDITION

LYNNE D. TALLEY

GEORGE L. PICKARD

WILLIAM J. EMERY

JAMES H. SWIFT

AMSTERDAM • BOSTON • HEIDELBERG • LONDON

NEW YORK • OXFORD • PARIS • SAN DIEGO

SAN FRANCISCO • SINGAPORE • SYDNEY • TOKYO

Academic Press is an imprint of Elsevier

Academic Press is an imprint of Elsevier

32 Jamestown Road, London NW1 7BY, UK

30 Corporate Drive, Suite 400, Burlington, MA 01803, USA

525 B Street, Suite 1800, San Diego, CA 92101-4495, USA

First published 1964

Reprinted 1966, 1968, 1970

Second edition 1975

Third edition 1979

Fourth enlarged edition 1982

Reprinted 1984, 1985, 1986, 1988, 1989

Fifth edition 1990

Reprinted 1995, 1996, 1999, 2000, 2002

Sixth edition 2011

Copyright Ó 2011. Lynne Talley, George Pickard, William Emery and James Swift. Published by Elsevier Ltd.

All rights reserved

No part of this publication may be reproduced, stored in a retrieval system or transmitted in any form or by any means

electronic, mechanical, photocopying, recording or otherwise without the prior written permission of the publisher

Permissions may be sought directly from Elsevier’s Science & Technology Rights Department in Oxford, UK:

phone (+44) (0) 1865 843830; fax (+44) (0) 1865 853333; email: permissions@elsevier.com. Alternatively, visit the Science

and Technology Books website at www.elsevierdirect.com/rights for further information

Notice

No responsibility is assumed by the publisher for any injury and/or damage to persons or property as a

matter of products liability, negligence or otherwise, or from any use or operation of any methods, products, instructions

or ideas contained in the material herein. Because of rapid advances in the medical sciences, in particular, independent

verification of diagnoses and drug dosages should be made.

British Library Cataloguing-in-Publication Data

A catalogue record for this book is available from the British Library

Library of Congress Cataloging-in-Publication Data

A catalog record for this book is available from the Library of Congress

ISBN: 978-0-7506-4552-2

For information on all Academic Press publications visit our

website at www.elsevierdirect.com

Typeset by TNQ Books and Journals Pvt Ltd

Printed and bound in the United States of America

1112131415 10987654321

Contents

Preface

Chapter 1 Introduction to Descriptive Physical Oceanography 1

Chapter 2 Ocean Dimensions Shapes and Bottom Materials 7

Chapter 3 Physical Properties of Seawater 29

Chapter 4 Typical Distributions of Water Characteristics 67

Chapter 5 Mass Salt and Heat Budgets and Wind Forcing 111

Chapter 6 Data Analysis Concepts and Observational Methods 147

Chapter 7 Dynamical Processes for Descriptive Ocean Circulation 187

Chapter 8 Gravity Waves Tides and Coastal Oceanography 223

Chapter 9 Atlantic Ocean 245

Chapter 10 Pacific Ocean 303

Chapter 11 Indian Ocean 363

Chapter 12 Arctic Ocean and Nordic Seas 401

Chapter 13 Southern Ocean 437

Chapter 14 Global Circulation and Water Properties 473

References 513

Index 545

Color Plates 557

Preface

This new edition of Descriptive Physical Oceanography:

An Introduction is dedicated to the

memory of George L. Pickard (July 15, 1913e

May 1, 2007), who was a physical oceanographer

at the University of British Columbia.

George was part of University of British

Columbia’s oceanography department from its

inception. His training was in low temperature

physics, with a Doctor of Philosophy from

Oxford in 1937. After service in WW II in the

Royal Aircraft Establishment, during which he

survived a crash in the English Channel, he

was appointed to the physics department at

UBC. As a young member of the department,

he was sent to Scripps Institution of Oceanography

for “a year’s training” in oceanography

as part of the lobbying effort by John Tully for

formation of Canada’s first academic program

in oceanography at UBC; the program was

established in 1949 (Mills, 1994). George was

director of the UBC Institute of Oceanography

from 1958 to 1978. He retired from teaching in

1982.

George wrote the first and subsequent

editions of this book as part of his teaching of

physical oceanography, bringing in Bill Emery

for the fourth and fifth editions as the material

was updated and enlarged. He also co-authored

with Stephen Pond the text Introductory Dynamical

Oceanography. In 1950, George initiated

time series measurements in many of the inlets

along the British Columbia coastline, partially

as training exercises for the UBC students;

these observations continue to the present and

constitute a tremendous source of climaterelated

information (http://www.pac.dfo-mpo.

gc.ca/science/oceans/BCinlets/index-eng.htm,

2010). His research expanded to include coral

reefs as well as fjords. A full CV is included on

the website that accompanies this edition.

When George Pickard published the original

DPO text in 1964, computers were just barely

beginning to be introduced, and courses were

taught at the blackboard and illustrated with

vii

viii

PREFACE

slides. This sixth edition of DPO stands at the

brink of fully electronic publishing and full

online support for teaching. We therefore

provide a website that includes all of the illustrations

from the print text, many in color even

if not reproduced in color herein. There are

also many additional illustrations and supporting

text on the website. Several chapters appear

on the website and not in the print text, in order

to keep the cost of the print text accessible: the

full-length version of Chapter 7 on ocean

dynamics; the final sections of Chapter 8 concerning

estuaries, coral reefs, and adjacent

seas; Chapter S15 on Climate and the oceans;

and Chapter S16 on Instruments and methods.

Secondly, with this edition we introduce a

digital set of tools and tutorials for working with

and displaying ocean property data, using Java

OceanAtlas. The software and representative

data sets are also provided online, along with a

step-by-step guide to using them and examples

associated with most chapters of the print text.

We strongly encourage students and lecturers

to make use of these web-based materials.

This edition of DPO is also dedicated by LDT

and JHS to our teachers, among them Joe

Pedlosky, Mike McCartney, Val Worthington,

Knut Aagaard and Eddy Carmack, and to our

senior colleague, Joe Reid at SIO whose work

is central to many chapters of this book, and

who preceded LDT in teaching SIO course 210.

The students of SIO 210 and colleagues who

have team taught the course e M. Hendershott

and P. Robbins e have provided annual

motivation for recalling the essentials of largescale

descriptive oceanography. A number of

colleagues and students provided invaluable

feedback on parts of the text, including J. Reid,

D. Sandwell, P. Robbins, J. Holte, S. Hautala,

L. Rosenfeld, T. Chereskin, Y. Firing, S. Gille

and the students from her data analysis laboratory

course, B. Fox-Kemper, K. Aagaard, A. Orsi,

I. Cerovecki, M. Hendershott, F. Feddersen,

M. Mazloff, S. Jayne, and J. Severinghaus, as

well as numerous comments from SIO 210

students. LDT gratefully acknowledges sabbatical

support from the SIO department, hosted

by Woods Hole Oceanographic Institution

(T. Joyce and the Academic Programs Office)

and the Université Joseph Fourier in Grenoble

(B. Barnier and the Observatoire des Sciences

de l’Univers de Grenoble).

L.D.T. and J.H.S.

Scripps Institution of Oceanography,

La Jolla, CA

C H A P T E R

1

Introduction to Descriptive Physical

Oceanography

Oceanography is the general name given to

the scientific study of the oceans. It is historically

divided into physical, biological, chemical,

and geological oceanography. This book is concerned

primarily with the physics of the ocean,

approached mainly, but not exclusively, from

observations, and focusing mainly, but not

exclusively, on the larger space and timescales

of the open ocean rather than on the near-coastal

and shoreline regions.

Descriptive physical oceanography approaches

the ocean through both observations and

complex numerical model output used to

describe the fluid motions as quantitatively as

possible. Dynamical physical oceanography seeks

to understand the processes that govern the fluid

motions in the ocean mainly through theoretical

studies and process-based numerical model

experiments. This book is mainly concerned

with description based in observations (similar

to previous editions of this text); however, in

this edition we include some of the concepts of

dynamical physical oceanography as an important

context for the description. A full treatment

of dynamical oceanography is contained in other

texts. Thermodynamics also clearly enters into our

discussion of the ocean through the processes

that govern its heat and salt content, and therefore

its density distribution.

Chapter 2 describes the ocean basins and their

topography. The next three chapters introduce the

physical (and some chemical) properties of freshwater

and seawater (Chapter 3), an overview of

the distribution of water characteristics (Chapter 4),

and the sources and sinks of heat and freshwater

(Chapter 5). The next three chapters cover data

collection and analysis techniques (Chapter 6

and supplemental material listed as Chapter S6

on the textbook Web site http://booksite.

academicpress.com/DPO/; “S” denotes supplemental

material.), an introduction to geophysical

fluid dynamics for graduate students who have

varying mathematics backgrounds (Chapter 7),

and then basic waves and tides with an introduction

to coastal oceanography (Chapter 8). The

last six chapters of the book introduce the circulation

and water properties of each of the individual

oceans (Chapters 9 through 13) ending

with a summary of the global ocean in Chapter 14.

Accompanying the text is the Web site

mentioned in the previous paragraph. It has

four aspects:

1. Textbook chapters on climate variability and

oceanographic instrumentation that do not

appear in the print version

2. Expanded material and additional figures for

many other chapters

Descriptive Physical Oceanography

1

Ó 2011. Lynne Talley, George Pickard, William Emery and James Swift.

Published by Elsevier Ltd. All rights reserved.

2

1. INTRODUCTION TO DESCRIPTIVE PHYSICAL OCEANOGRAPHY

3. A full set of tutorials for descriptive

oceanography with data and sample scripts

provided based on the Java Ocean Atlas

(Osborne & Swift, 2009)

4. All figures from the text, with many more in

color than in the text, for lectures and

presentations.

1.1. OVERVIEW

There are many reasons for developing our

knowledge of the ocean. Near-shore currents

and waves affect navigation and construction of

piers, breakwaters, and other coastal structures.

The large heat capacity of the oceans exerts

a significant and in some cases a controlling effect

on the earth’s climate. The ocean and atmosphere

interact on short to long timescales; for example,

the El NiñoeSouthern Oscillation (ENSO)

phenomenon that, although driven locally in the

tropical Pacific, affects climate on timescales of

several years over much of the world. To understand

these interactions it is necessary to understand

the coupled ocean-atmosphere system. To

understand the coupled system, it is first necessary

to have a solid base of knowledge about

both the ocean and atmosphere separately.

In these and many other applications, knowledge

of the ocean’s motion and water properties

is essential. This includes the major ocean

currents that circulate continuously but with

fluctuating velocity and position, the variable

coastal currents, the rise and fall of the tides,

and the waves generated by winds or earthquakes.

Temperature and salt content determine

density and hence vertical movement. They also

affect horizontal movement as the density affects

the horizontal pressure distribution. Sea ice has

its own full set of processes and is important for

navigation, ocean circulation, and climate. Other

dissolved substances such as oxygen, nutrients,

and other chemical species, and even some of

the biological aspects such as chlorophyll

content, are used in the study of ocean physics.

Our present knowledge in physical oceanography

rests on an accumulation of data, most of

which were gathered during the past 150 years,

with a large increase of in situ data collection

(within the actual water) starting in the 1950s

and an order of magnitude growth in available

data as satellites began making ocean measurements

(starting in the 1970s).

A brief history of physical oceanography with

illustrations is provided as supplemental material

on the textbook Web site (Chapter 1 supplement

is listed as Chapter S1 on the Web site).

Historically, sailors have always been concerned

with how ocean currents affect their ships’

courses as well as changes in ocean temperature

and surface condition. Many of the earlier navigators,

such as Cook and Vancouver, made valuable

scientific observations during their voyages

in the late 1700s, but it is generally considered

that Mathew Fontaine Maury (1855) started the

systematic large-scale collection of ocean current

data using ship’s navigation logs as his source of

information. The first major expedition designed

expressly to study all scientific aspects of the

oceans was carried out on the British H.M.S.

Challenger that circumnavigated the globe from

1872 to 1876. The first large-scale expedition

organized primarily to gather physical oceanographic

data was carried out on the German FS

Meteor, which studied the Atlantic Ocean from

1925 to 1927 (Spiess, 1928). A number of photos

from that expedition are reproduced on the

accompanying Web site. Some of the earliest

theoretical studies of the sea included the surface

tides by Newton, Laplace, and Legendre (e.g.,

Wilson, 2002) and waves by Gerstner and Stokes

(e.g., Craik, 2005). Around 1896, some of the

Scandinavian meteorologists started to turn

their attention to the ocean, because dynamical

meteorology and dynamical oceanography

have many common characteristics. Current

knowledge of dynamical oceanography owes

its progress to the work of Bjerknes, Bjerknes,

Solberg, and Bergeron (1933), Ekman (1905,

1923), Helland-Hansen (1934), and others.

SPACE AND TIMESCALES OF PHYSICAL OCEANOGRAPHIC PHENOMENA 3

The post-war 1940s through 1960s began to

produce much of the data and especially theoretical

understanding for large-scale ocean circulation.

With the advent of moored and satellite

instrumentation in the 1960s and 1970s, the

smaller scale, energetically varying part of the

ocean circulation d the mesoscale d began to

be studied in earnest. Platforms expanded from

research and merchant ships to global satellite

and autonomous instrument sampling. Future

decades should take global description and

modeling to even smaller scales (submesoscale)

as satellite observations and numerical modeling

resolution continue to evolve, and different types

of autonomous sampling within the water

column become routine. Physical oceanography

has retained an aspect of individual exploration

but large, multi-investigator, multinational

programs have increasingly provided many of

the new data sets and understanding. Current

research efforts in physical oceanography are

focused on developing an understanding of the

variability of the ocean and its relation to the

atmosphere and climate as well as continuing to

describe its steady-state conditions.

around Florida and northward along the east

coast of North America, leaves the coast at

Cape Hatteras, and moves out to sea. Its

strength and temperature contrast decay eastward.

Its large meanders and rings, with spatial

scales of approximately 100 km, are considered

mesoscale (eddy) variability evolving on timescales

of weeks. The satellite image also shows

the general decrease in surface temperature

toward the north and a large amount of smallscale

eddy variability. The permanence of the

Gulf Stream is apparent when currents and

temperatures are averaged in time. Averaging

makes the Gulf Stream appear wider, especially

after the separation at Cape Hatteras where the

wide envelope of meanders creates a wide,

weak average eastward flow.

The Gulf Stream has been known and charted

for centuries, beginning with the Spanish expeditions

in the sixteenth century (e.g., Peterson,

Stramma, & Kortum 1996). It was first mapped

accurately in 1769 by Benjamin Franklin

1.2. SPACE AND TIMESCALES OF

PHYSICAL OCEANOGRAPHIC

PHENOMENA

The ocean is a fluid in constant motion with

a very large range of spatial and temporal

scales. The complexity of this fluid is nicely represented

in the sea surface temperature image of

the Gulf Stream captured from a satellite shown

in Figure 1.1a. The Gulf Stream is the western

boundary current of the permanent, large-scale

clockwise gyre circulation of the subtropical

North Atlantic. The Gulf Stream has a width of

100 km, and its gyre has a spatial scale of thousands

of kilometers. The narrow, warm core of

the Gulf Stream (red in Figure 1.1a) carries

warm subtropical water northward from the

Caribbean, loops through the Gulf of Mexico

FIGURE 1.1 (a) Sea surface temperature from a satellite

advanced very high resolution radiometer (AVHRR)

instrument (Otis Brown, personal communication, 2009).

This figure can also be found in the color insert.

4


FIGURE 1.1

(b) Franklin-Folger map of the Gulf Stream. Source: From Richardson (1980a).

working together with whaling captain Timothy

Folger (Figure 1.1b; from Richardson, 1980a). 1

The narrow current along the coast of the

United States is remarkably accurate. The

widening envelope of the Franklin/Folger

current after separation from Cape Hatteras is

an accurate depiction of the envelope of

meandering apparent in the satellite image.

When time-mean averages of the Gulf Stream

based on modern measurements are constructed,

they look remarkably similar to this

Franklin/Folger map.

The space and timescales of many of the

important physical oceanography processes are

shown schematically in Figure 1.2. At the smallest

scale is molecular mixing. At small,

1 Franklin noted on his frequent trips between the United States and Europe that some trips were considerably quicker than

others. He decided that this was due to a strong ocean current flowing from the west to the east. He observed marked

changes in surface conditions and reasoned that this ocean current might be marked by a change in sea surface

temperature. Franklin began making measurements of the ocean surface temperature during his travels. Using a simple

mercury-in-glass thermometer, he was able to determine the position of the current.

SPACE AND TIMESCALES OF PHYSICAL OCEANOGRAPHIC PHENOMENA 5

Timescale

1,000s years

100 years

10 years

1–7 years

wks – months

12–24 hrs

El Niño

Mesoscale

eddies

Tides

Milankovitch

Thermohaline

Winddriven

circulation

circulation

FIGURE 1.2 Time and space

scales of physical oceanographic

phenomena from bubbles and

capillary waves to changes in

ocean circulation associated with

Earth’s orbit variations.

1 min – 20 hrs

0.1 sec – 1 min

0.1 sec

0.01 sec

Internal

waves

Deep-water

surface waves

Capillary

waves

Bubbles

Tsunamis

mm cm 1–100m 0.1–100km 100–1,000km 5,000 km 10,000 km

500–5,000 km

Length Scale

macroscopic scales of centimeters, microstructure

(vertical layering at the centimeter level)

and capillary waves occur. At the slightly larger

scale of meters surface waves are found, which

have rapid timescales and somewhat long-lived

vertical layers (fine structure). At scales of tens

of meters are the internal waves with timescales

of up to a day. Tides have the same timescales

as internal waves, but much larger spatial scales

of hundreds to thousands of kilometers. Surface

waves, internal waves, and tides are described

in Chapter 8.

Mesoscale eddies and strong ocean currents

such as the Gulf Stream are found at spatial

scales of tens of kilometers to several hundred

kilometers and timescales of weeks to years

(Figure 1.1a,b). The large-scale ocean circulation

has a spatial scale the size of ocean basins up to

the global ocean and a timescale ranging from

seasonal to permanent, which is the timescale

of plate tectonics that rearranges the ocean

boundaries (Chapter 2). The timescales for

wind-driven and thermohaline circulation in

Figure 1.2 are actually the same for circulation

of the flow through those systems (ten years

around the gyre, hundreds of years through the

full ocean); these are time-mean features of the

ocean and have much longer timescales. Climate

variability affects the ocean, represented in

Figure 1.2 by the El Niño, which has an interannual

timescale (several years; Chapter 10);

decadal and longer timescales of variability of

the ocean circulation and properties are also

important and described in each of the ocean

basin and global circulation chapters.

We see in Figure 1.2 that short spatial scales

generally have short timescales, and long spatial

scales generally have long timescales. There are

some exceptions to this, most notably in the

tides and tsunamis as well as in some fine-structure

phenomena with longer timescales than

might be expected from their short spatial

scales.

In Chapter 7, where ocean dynamics are discussed,

some formal non-dimensional parameters

incorporating the approximate space and timescales

for these different types of phenomena

are introduced (see also Pedlosky, 1987). A

non-dimensional parameter is the ratio of

dimensional parameters with identical dimensional

scales, such as time, length, mass, etc.,

which are intrinsic properties of the flow

phenomenon being described or modeled. Of

special importance is whether the timescale of

6


an ocean motion is greater than or less than about

a day, which is the timescale for the earth’s rotation.

Earth’s rotation has an enormous effect on

how the ocean water moves in response to

a force; if the force and motion are sustained

for days or longer, then the motion is strongly

influenced by the rotation. Therefore, an especially

useful parameter is the ratio of the timescale

of Earth’s rotation to the timescale of the

motion. This ratio is called the Rossby number.

For the very small, fast motions in Figure 1.2,

this ratio is large and rotation is not important.

For the slow, large-scale part of the spectrum,

the Rossby number is small and Earth’s rotation

is fundamental. A second very important nondimensional

parameter is the ratio of the vertical

length scale (height) to the horizontal length

scale; this is called the aspect ratio. For large-scale

flows, this ratio is very small since the vertical

scale can be no larger than the ocean depth. For

surface and internal gravity waves, the aspect

ratio is order 1. We will also see that dissipation

is very weak in the sense that the timescale for

dissipation to act is long compared with both

the timescale of Earth’s rotation and the timescale

for the circulation to move water from

one place to another; the relevant non-dimensional

parameters are the Ekman number and

Reynolds number, respectively. Understanding

how the small Rossby number, small aspect

ratio, and nearly frictionless fluid ocean behaves

has depended on observations of the circulation

and water properties made over the past century.

These are the principal subjects of this text.

C H A P T E R

2

Ocean Dimensions, Shapes,

and Bottom Materials

2.1. DIMENSIONS

The oceans are basins in the surface of the

solid earth containing salt water. This chapter

introduces some nomenclature and directs

attention to features of the basins that have

a close connection with the ocean’s circulation

and dynamical processes that are of importance

to the physical oceanographer. More detailed

descriptions of the geology and geophysics of

the ocean basins are given in Seibold and Berger

(1982), Kennett (1982), Garrison (2001), and

Thurman and Trujillo (2002), among others.

Updated data sets, maps, and information are

available from Web sites of the National

Geophysical Data Center (NGDC) of the

National Oceanic and Atmospheric Administration

(NOAA) and from the U.S. Geological

Survey (USGS).

The major ocean areas are the Atlantic Ocean,

the Pacific Ocean, the Indian Ocean, the Arctic

Ocean, and the Southern Ocean (Figure 2.1).

The first four are clearly divided from each

other by land masses, but the divisions between

the Southern Ocean and oceans to its north are

determined only by the characteristics of the

ocean waters and their circulations. The

geographical peculiarities of each ocean are

described in Section 2.11.

The shape, depth, and geographic location of

an ocean affect the general characteristics of its

circulation. Smaller scale features, such as locations

of deep sills and fracture zones, seamounts,

and bottom roughness, affect often important

details of the circulation and of mixing processes

that are essential to forcing and water properties.

The Atlantic has a very marked “S” shape while

the Pacific has a more oval shape. The Atlantic

and Indian Oceans are roughly half the eastwest

width of the Pacific Ocean, which impacts

the way that each ocean’s circulation adjusts to

changes in forcing. The Indian Ocean has no

high northern latitudes, and therefore no possibility

of cold, dense water formation. The edges

of the Pacific are ringed with trenches, volcanoes,

and earthquakes that signal the gradual

descent of the ocean bottom crustal “plates”

under the surrounding continental plates. In

contrast, the Atlantic is the site of dynamic

seafloor spreading as material added in the

center of the Mid-Atlantic Ridge (MAR) pushes

the plates apart, enlarging the Atlantic Ocean

by a few centimeters each year.

Marginal seas are fairly large basins of salt

water that are connected to the open ocean by

one or more fairly narrow channels. Those that

are connected by very few channels are sometimes

called mediterranean seas after the


7



8

2. OCEAN DIMENSIONS, SHAPES, AND BOTTOM MATERIALS

80˚N

180˚ 120˚W 60˚W 0˚ 60˚E 120˚E 180˚

Arctic Ocean

80˚N

60˚N

60˚N

40˚N

40˚N

20˚N

0˚

Pacific Ocean

20˚N

0˚

20˚S

Atlantic

Ocean

Indian

Ocean

20˚S

40˚S

40˚S

60˚S

Southern Ocean

Southern Ocean

60˚S

80˚S

180˚ 120˚W 60˚W 0˚ 60˚E 120˚E 180˚

80˚S

−6000 −4000 −2000 0 2000 4000 6000

FIGURE 2.1 Map of the world based on ship soundings and satellite altimeter derived gravity at 30 arc-second resolution.

Data from Smith & Sandwell (1997); Becker et al. (2009); and SIO (2008).

prototype, the (European) Mediterranean Sea.

The Mediterranean provides an example of

a negative water balance in a sea with less

inflow (river runoff and precipitation) than

evaporation. An excellent example of a positive

water balance marginal sea (with net precipitation)

is found in the Black Sea, which connects

with the Mediterranean Sea. Both of these seas

are discussed further in Chapters 5 and 9. Other

examples of marginal seas that are separated

from the open ocean by multiple straits or island

chains are the Caribbean Sea, the Sea of Japan,

the Bering Sea, the North Sea, the Baltic Sea,

and so forth.

The term sea is also used for a portion of an

ocean that is not divided off by land but has

local distinguishing oceanographic characteristics;

for example the Norwegian Sea, the Labrador

Sea, the Sargasso Sea, and the Tasman Sea.

More of the earth’s surface is covered by sea

than by land, about 71% sea to 29% land. (The

most recent elevation data for the earth’s surface,

used to construct Figure 2.2, show that 70.96% of

the earth is ocean; see Becker et al., 2009.)

Furthermore, the proportion of water to land in

the Southern Hemisphere is much greater (4:1)

than in the Northern Hemisphere (1.5:1). In

area, the Pacific Ocean is about as large as the

Atlantic and Indian Oceans combined. If the

neighboring sectors of the Southern Ocean are

included with the three main oceans north of it,

the Pacific Ocean occupies about 46% of the total

world ocean area, the Atlantic Ocean about 23%,

the Indian Ocean about 20%, and the rest,

combined, about 11%.

The average depth of the oceans is close to

4000 m while the marginal seas are generally

about 1200 m deep or less. Relative to sea level,

PLATE TECTONICS AND DEEP-SEA TOPOGRAPHY 9

8000

Maximum (Mt. Everest)

Elevation (m)

6000

4000

2000

Sea level

Mean height (743 m)

Depth (m)

−2000

−4000

−6000

−8000

Mean depth (3734 m)

Median depth (4093 m)

0.04%

−10000

Maximum (Mariana Trench)

0 0.5 1 1.5 2 2.5 3 3.5 4 4.5

Area %

FIGURE 2.2 Areas of Earth’s surface above and below sea level as a percentage of the total area of Earth (in 100 m

intervals). Data from Becker et al. (2009).

the oceans are much deeper than the land is high.

While only 11% of the land surface of the earth is

more than 2000 m above sea level, 84% of the sea

bottom is more than 2000 m deep. However, the

maxima are similar: the height of Mt. Everest is

about 8848 m, while the maximum depth in the

oceans is 11,034 m in the Mariana Trench in the

western North Pacific. Figure 2.2 shows the

distributions of land elevations and of sea depths

relative to sea level in 100 m intervals as the

percentage of the total area of the earth’s surface.

This figure is based on the most recent global

elevation and ocean bathymetry data from

D. Sandwell (Becker et al., 2009). (It is similar to

Figure 2.2 using 1000 m bins that appeared in

previous editions of this text, based on data

from Kossina, 1921 and Menard & Smith, 1966,

but the 100 m bins allow much more differentiation

of topographic forms.)

Although the average depth of the oceans,

4 km, is a considerable distance, it is small

compared with the horizontal dimensions of

the oceans, which are 5000 to 15,000 km. Relative

to the major dimensions of the earth, the

oceans are a thin skin, but between the sea

surface and the bottom of the ocean there is

a great deal of detail and structure.

2.2. PLATE TECTONICS AND

DEEP-SEA TOPOGRAPHY

The most important geophysical process

affecting the shape and topography of the ocean

10


basins is the movement of the earth’s tectonic

plates, described thoroughly in texts such as

Thurman and Trujillo (2002). The plate boundaries

are shown in Figure 2.3. Seafloor spreading

creates new seafloor as the earth’s plates spread

apart. This creates the mid-ocean ridge system;

the mid-ocean ridges of Figure 2.1 correspond

to plate boundaries. The ocean plates spread

apart at rates of about 2 cm/year (Atlantic) to

16 cm/year (Pacific), causing extrusion of

magma into the surface at the centers of the

ridges. Over geologic time the orientation of

the earth’s magnetic field has reversed, causing

the ferromagnetic components in the molten

new surface material to reverse. Spreading at

the mid-ocean ridge was proven by observing

the reversals in the magnetic orientations in

the surface material. These reversals permit

dating of the seafloor (Figure 2.3). The recurrence

interval for magnetic reversals is approximately

500,000 to 1,000,000 years.

The 14,000 km long MAR is a tectonic

spreading center. It is connected to the global

mid-ocean ridge, which at more than 40,000 km

long, is the most extensive feature of the earth’s

topography. Starting in the Arctic Ocean, the

mid-ocean ridge extends through Iceland

down the middle of the Atlantic, wraps around

the tip of Africa, and then winds through the

Indian and Pacific Oceans, ending in the Gulf

of California. In all oceans, the mid-ocean ridge

and other deep ridges separate the bottom

waters, as can be seen from different water

properties east and west of the ridge.

Deep and bottom waters can leak across the

ridges through narrow gaps, called fracture

zones, which are lateral jogs in the spreading

center. The fracture zones are roughly vertical

planes, perpendicular to the ridge, on either

side of which the crust has moved in opposite

directions perpendicular to the ridge. There

are many fracture zones in the mid-ocean

FIGURE 2.3 Sea floor age (millions of years). Black lines indicate tectonic plate boundaries. Source: From Müller, Sdrolias,

Gaina, and Roest (2008).


ridges. One example that is important as

a pathway for abyssal circulation in the Atlantic

is the Romanche Fracture Zone through the

MAR close to the equator. Another example is

the pair of fracture zones in the South Pacific

(Eltanin and Udintsev Fracture Zones,

Figure 2.12) that steer the Antarctic Circumpolar

Current (ACC).

At some edges of the tectonic plates, one plate

subducts (moves under) another. Subduction is

accompanied on its landward side by volcanoes

and earthquakes. Subduction creates deep

trenches that are narrow relative to their length

and have depths to 11,000 m. The deepest parts

of the oceans are in these trenches. The majority

of the deep trenches are in the Pacific: the Aleutian,

Kurile, Tonga, Philippine, and Mariana.

There are a few in other oceans such as the

Puerto Rico and the South Sandwich Trenches

in the Atlantic and the Sunda Trench in the

Indian Ocean. Trenches are often shaped like

an arc of a circle with an island arc on one side.

Examples of island arcs are the Aleutian Islands

(Pacific), the Lesser Antilles (Atlantic), and the

Sunda Arc (Indian). The landward side of

a trench extends as much as 10,000 m from the

trench bottom to the sea surface, while the other

side is only half as high, terminating at the

ocean depth of about 5000 m.

Trenches can steer or impact boundary

currents that are in deep water (Deep Western

Boundary Currents) or upper ocean boundary

currents that are energetic enough to extend to

the ocean bottom, such as western boundary

currents of the wind-driven circulation. Examples

of trenches that impact ocean circulation

are the deep trench system along the western

and northern boundary of the Pacific and the

deep trench east of the Caribbean Sea in the

Atlantic.

Younger parts of the ocean bottom are shallower

than older parts. As the new seafloor

created at seafloor spreading centers ages, it

cools by losing heat into the seawater above

and becomes denser and contracts, which

causes it to be deeper (Sclater, Parsons, & Jaupart,

1981). Ocean bottom depths range from 2

to 3 km for the newest parts of the mid-ocean

ridges to greater than 5 km for the oldest, as

can be seen by comparing the maps of seafloor

age and bathymetry (Figures 2.1 and 2.3).

The rate of seafloor spreading is so slow that it

has no impact on the climate variability that we

experience over decades to millennia, nor does it

affect anthropogenic climate change. However,

over many millions of years, the geographic

layout of Earth has changed. The paleocirculation

patterns of “deep time,” when the continents

were at different locations, differed from

the present patterns; reconstruction of these

patterns is an aspect of paleoclimate modeling.

By studying and understanding present-day

circulation, we can begin to credibly model the

paleocirculation, which had the same physical

processes (such as those associated with the

earth’s rotation, wind and thermohaline

forcing, boundaries, open east-west channels,

equatorial regions, etc.), but with different

ocean basin shapes and bottom topography.

Ocean bottom roughness affects ocean mixing

rates (Sections 7.2.4 and 7.3.2). The overall

roughness varies by a factor of 10. Roughness

is a function of spreading rates and sedimentation

rates. New seafloor is rougher than old

seafloor. Slow-spreading centers produce

rougher topography than fast spreading

centers. Thus the slow-spreading MAR is

rougher than the faster spreading East Pacific

Rise (EPR; Figure 2.4). Slow-spreading ridges

also have rift valleys at the spreading center,

whereas fast-spreading ridges have an elevated

ridge at the spreading center. Much of the

roughness on the ridges can be categorized as

abyssal hills, which are the most common landform

on Earth. Abyssal hills are evident in

Figures 2.4 and 2.5b, all along the wide flanks

of the mid-ocean ridge.

Individual mountains (seamounts) are widely

distributed in the oceans. Seamounts stand out

clearly above the background bathymetry. In

12


FIGURE 2.4 Seafloor topography for a portion of (a) the fast-spreading EPR and (b) the slow-spreading MAR. Note the

ridge at the EPR spreading center and rift valley at the MAR spreading center. This figure can also be found in the color

insert. (Sandwell, personal communication, 2009.)

the maps in Figure 2.4b there are some

seamounts on the upper right side of the panel.

In the vertical cross-section in Figure 2.5b,

seamounts are distinguished by their greater

height compared with the abyssal hills. The

average height of seamounts is 2 km.

Seamounts that reach the sea surface form

islands. A guyot is a seamount that reached

the surface, was worn flat, and then sank again

below the surface. Many seamounts and

islands were created by volcanic hotspots

beneath the tectonic plates. The hotspots are

relatively stationary in contrast to the plates

and as the plates move across the hotspots,

chains of seamounts are formed. Examples

include the Hawaiian Islands/Emperor Seamounts

chain, Polynesian island chains, the

Walvis Ridge, and the Ninetyeast Ridge in the

Indian Ocean.

Seamounts affect the circulation, especially

when they appear in groups as they do in

many regions; for instance, the Gulf Stream

passes through the New England Seamounts,

which affect the Gulf Stream’s position and

variability (Section 9.3). Seamount chains also

refract tsunamis, which are ocean waves

generated by submarine earthquakes that react

to the ocean bottom as they propagate long

distances from the earthquake source (Section

8.3.5).

back arc ridge

BACK ARC BASIN


(a)

CONTINENT

SHORE

HIGH WATER

LOW WATER

SHELF

(GRAVEL,

SAND

Ave. Slope

1 in 500)

(MUD

Ave. Slope

1 in 20)

SLOPE

RISE

BASIN

CLAY & OOZES

Mineral Organic

SEA LEVEL

OCEAN

Spreading

Center

MID-OCEAN

RIDGE

DEEP SEA

SEAMOUNT

TRENCH

ISLAND ARC

(b)

0

500

1000

1500

slope

seamount

seamount

slope

2000

2500

3000

abyssal hills

3500

4000

4500

5000

back arc basin

abyssal hills

spreading center

mid-ocean ridge

5500

6000

6500

trench

trench

(c)

20˚S

20˚S

30˚S

South Pacific

40˚S

140˚E 160˚E 180˚ 160˚W 140˚W 120˚W 100˚W 80˚W

FIGURE 2.5 (a) Schematic section through ocean floor to show principal features. (b) Sample of bathymetry, measured

along the South Pacific ship track shown in (c).

30˚S

40˚S

14


2.3. SEAFLOOR FEATURES

The continents form the major lateral boundaries

to the oceans. The detailed features of the

shoreline and of the sea bottom are important

because of the way they affect circulation. Starting

from the land, the main divisions recognized

are the shore, the continental shelf, the continental

slope and rise, and the deep-sea

bottom, part of which is the abyssal plain

(Figure 2.5a, b). Some of the major features of

the seafloor, including mid-ocean ridges, trenches,

island arcs, and seamounts, are the result

of plate tectonics and undersea volcanism

(Section 2.2 and Figure 2.3).

In some of the large basins the seafloor is very

smooth, possibly more so than the plains areas

on land. Sedimentation, which is mostly due

to the incessant rain of organic matter from the

upper ocean, covers the rough bottom and

produces large regions of very smooth topography.

Stretches of the abyssal plain in the

western North Atlantic have been measured to

be smooth within 2 m over distances of 100 km.

The ocean bottom in the northeast Indian

Ocean/Bay of Bengal slopes very smoothly

from 2000 m down to more than 5000 m over

3000 km. This smoothness is due to sedimentation

from the Ganges and Brahmaputra Rivers

that drain the Himalayas. Bottom sediments

can be moved around by deep currents; formation

of undersea dunes and canyons is common.

Erosional features in deep sediments have

sometimes alerted scientists to the presence of

deep currents.

Bottom topography often plays an important

role in the distribution of water masses and the

location of currents. For instance, bottom water

coming from the Weddell Sea (Antarctica) is

unable to fill the eastern part of the Atlantic

basin directly due to the height of the Walvis

Ridge (South Atlantic Ocean). Instead, the

bottom water travels to the north along the

western boundary of the South Atlantic, finds

a deep passage in the MAR, and then flows

south to fill the basin east of the ridge. At shallower

depths the sills (shallowest part of

a channel) defining the marginal seas strongly

influences both the mid-level currents and the

distribution of water masses associated with

the sea. Coastal upwelling is a direct consequence

of the shape of the coast and its related

bottom topography. Alongshore currents are

often determined by the coastal bottom topography

and the instabilities in this system can

depend on the horizontal scales of the bottom

topography. Near the shore bottom topography

dictates the breaking of surface gravity waves

and also directly influences the local tidal

expressions.

Much of the mixing in the ocean occurs near

the boundaries (including the bottom). Microstructure

observations in numerous regions,

and intensive experiments focused on detection

of mixing and its genesis, suggest that flow of

internal tides over steep bottom slopes in the

deep ocean is a major mechanism for dissipating

the ocean’s energy. Ocean bottom slopes

computed from bathymetry collected along

ship tracks show that the largest slopes tend to

occur on the flanks of the fastest spreading

mid-ocean ridges. With bathymetric slopes

computed from the most recent bathymetric

data (Figure 2.6) and information about the

ocean’s deep stratification, it appears the flanks

of the mid-ocean ridges of the Atlantic,

Southern Ocean, and Indian Ocean could be

the most vigorous dissipation sites of ocean

energy (Becker & Sandwell, 2008).

2.4. SPATIAL SCALES

Very often some of the characteristics of the

ocean are presented by drawing a vertical

cross-section of a part of the oceans, such as

the schematic depiction of ocean floor features

in Figure 2.5a. An illustration to true scale

would have the relative dimensions of the

edge of a sheet of paper and would be either

SHORE, COAST, AND BEACH 15

FIGURE 2.6 Mean slope of the ocean bottom, calculated from shipboard bathymetry and interpolated to a 0.5 degree

grid. Source: From Becker and Sandwell (2008).

too thin to show details or too long to be convenient.

Therefore, we usually distort our crosssection

by making the vertical scale much larger

than the horizontal one. For instance, we might

use a scale of 1 cm to represent 100 km horizontally

while depths might be on a scale of 1 cm

to represent 100 m (i.e., 0.1 km). In this case

the vertical dimensions on our drawing would

be magnified 1000 times compared with the

horizontal ones (a vertical exaggeration of

1000:1). This gives us room to show the detail,

but it also exaggerates the slope of the sea

bottom or of contours of constant water properties

(isopleths) drawn on the cross-section

(Figure 2.5b). In reality, such slopes are far less

than they appear on the cross-section drawings;

for instance, a line of constant temperature

(isotherm) with a real slope of 1 in 10,000 would

appear as a slope of 1 in 10 on the plot.

2.5. SHORE, COAST, AND BEACH

The shore is defined as that part of the landmass,

close to the sea, that has been modified

by the action of the sea. The words shore and

coast have the same meaning. Shorelines (coasts)

shift over time because of motion of the land

over geologic time, changes in sea level, and

erosion and deposition. The sedimentary record

16


shows a series of marine intrusions and retreats

corresponding to layers that reflect periods

when the surface was above and below sea

level. Variations in sea level between glacial

and interglacial periods have been as much as

120 m. The ability of the coast to resist the

erosional forces of the ocean depends directly

on the type of material that makes up the coast.

Sands are easily redistributed by the ocean

currents whereas granitic coasts are slow to

erode. Often sea level changes are combined

with the hydrologic forces of an estuary, which

dramatically change the dynamical relationship

between the ocean and the solid surface.

The beach is a zone of unconsolidated particles

at the seaward limit of the shore and extends

roughly from the highest to the lowest tide

levels. The landward limit of the beach might

be vegetation, permanent sand dunes, or human

construction. The seaward limit of a beach,

where the sediment movement on- and offshore

ceases, is about 10 m deep at low tide.

Coasts can be classified in many different

ways. In terms of long timescales (such as those

of plate tectonics; Section 2.2), coasts and continental

margins can be classified as active or

passive. Active margins, with active volcanism,

faulting, and folding, are like those in much of

the Pacific and are rising. Passive margins, like

those of the Atlantic, are being pushed in front

of spreading seafloor, are accumulating thick

wedges of sediment, and are generally falling.

Coasts can be referred to as erosional or depositional

depending on whether materials are

removed or added. At shorter timescales, waves

and tides cause erosion and deposition. At

millennial timescales, changes in mean sea level

cause materials to be removed or added.

Erosional coasts are attacked by waves and

currents, both of which carry fine material that

abrades the coast. The waves create alongshore

and rip currents (Section 8.3) that carry the

abraded material from the coastline along and

out to sea. This eroded material can be joined

by sediments discharged from rivers and form

deltas. This type of erosion is fastest on highenergy

coasts with large waves, and slowest on

low-energy coasts with generally weak wave

fields. Erosion occurs more rapidly in weaker

materials than in harder components. These

variations in materials allow erosive forces to

carve characteristic features on coastlines such

as sea cliffs and sea caves, and to create an alternation

between bays and headlands.

Beaches result when sediment, usually sand,

is transported to places suitable for continued

deposition. Again these are often the quiet

bays between headlands and other areas of

low surf activity. Often a beach is in equilibrium;

new sand is deposited to replace sand that is

scoured away. Evidence for this process can be

seen by how sand accumulates against new

structures built on the shore, or by how it is

removed from a beach when a breakwater is

built that cuts off the supply of sand beyond it.

On some beaches, the sand may be removed

by currents associated with high waves at one

season of the year and replaced by different

currents associated with lower waves at another

season. These currents are influenced by

seasonal and interannual wind variations.

Sea level, which strongly affects coasts, is

affected by the total amount of water in the

ocean, changes in the containment volume of

the world’s ocean, and changes in the temperature/salinity

characteristics of the ocean that

alter its density and hence cause the water to

expand or contract. Changes in the total amount

of water are due primarily to changes in the

volume of landfast ice, which is contained in

ice sheets and glaciers. (Because sea ice floats

in water, changes in sea ice volume, such as

that in the Arctic or Antarctic, do not affect sea

level.) Changes in containment volume are

due to tectonics, the slow rebound of continents

(continuing into the present) after the melt of

landfast ice after the last deglaciation, and

rebound due to the continuing melt of glacial

ice. Changes in heat content cause seawater to

expand (heating) or contract (cooling).

DEEP OCEAN 17

Sea level rose 20 cm from 1870 to 2003,

including 3 cm in just the last 10 years

(1993e2003). Because good global observations

are available for that last 10 years, it is possible

to ascribe 1.6 cm to thermal expansion, 0.4 cm

to Greenland and Antarctic ice sheet melt, and

0.8 cm to other glacial melt with a residual of

0.3 cm. Sea level is projected to rise 30 10 cm

in the next 100 years mainly due to warming

of the oceans, which absorb most of the anthropogenic

heat increase in the earth’s climate

system. (See Bindoff et al., 2007 in the 4th assessment

report of the Intergovernmental Panel on

Climate Change.)

2.6. CONTINENTAL SHELF, SLOPE,

AND RISE

The continental shelf extends seaward from

the shore with an average gradient of 1 in 500.

Its outer limit (the shelf break) is set where the

gradient increases to about 1 in 20 (on average)

to form the continental slope down to the deep

sea bottom. The shelf has an average width of

65 km. In places it is much narrower than this,

while in others, as in the northeastern Bering

Sea or the Arctic shelf off Siberia, it is as much

as ten times this width. The bottom material is

dominantly sand with less common rock or

mud. The shelf break is usually clearly evident

in a vertical cross-section of the sea bottom

from the shore outward. The average depth at

the shelf break is about 130 m. Most of the

world’s fisheries are located on the continental

shelves for a multitude of reasons including

proximity of estuaries, depth of penetration of

sunlight compared with bottom depth, and

upwelling of nutrient-rich waters onto some

shelves, particularly those off western coasts.

The continental slope averages about 4000 m

vertically from the shelf to the deep-sea bottom,

but in places extends as much as 9000 m vertically

in a relatively short horizontal distance.

In general, the continental slope is considerably

steeper than the slopes from lowland to highland

on land. The material of the slope is

predominantly mud with some rock outcrops.

The shelf and slope typically include submarine

canyons, which are of worldwide occurrence.

These are valleys in the slope, either V-shaped

or with vertical sides, and are usually found

off coasts with rivers. Some, usually in hard

granitic rock, were originally carved as rivers

and then submerged, such as around the Mediterranean

and southern Baja, California. Others,

commonly in softer sedimentary rock, are

formed by turbidity currents described in the

next paragraph. The lower part of the slope,

where it grades into the deep-sea bottom, is

referred to as the continental rise.

Turbidity currents (Figure 2.7) are common on

continental slopes. These episodic events carry

a mixture of water and sediment and are driven

by the unstable sediments rather than by forces

within the water. In these events, material builds

up on the slope until it is no longer stable and

the force of gravity wins out. Large amounts of

sediment and bottom material crash down the

slope at speeds up to 100 km/h. These events

can snap underwater cables. The precise conditions

that dictate when a turbidity current

occurs vary with the slope of the valley and

the nature of the material in the valley. Turbidity

currents carve many of the submarine canyons

found on the slopes. Some giant rivers, such as

the Congo, carry such a dense load of suspended

material that they form continuous

density flows of turbid water down their

canyons.

2.7. DEEP OCEAN

From the bottom of the continental slope, the

bathymetric gradient decreases down the continental

rise to the deep-sea bottom, the last and

most extensive area. Depths of 3000e6000 m

are found over 74% of the ocean basins with

1% deeper. Perhaps the most characteristic

18


FIGURE 2.7 Turbidity current evidence south of Newfoundland resulting from an earthquake in 1929. Source: From

Heezen, Ericson, and Ewing (1954).

aspect of the deep-sea bottom is the variety of its

topography. Before any significant deep ocean

soundings were available, the sea bottom was

regarded as uniformly smooth. When detailed

sounding started in connection with cable

laying, it became clear that this was not the

case and there was a swing to regarding the

sea bottom as predominantly rugged. Neither

view is exclusively correct, for we now know

that there are mountains, valleys, and plains

on the ocean bottom, just as on land. With the

advent of satellite altimetry for mapping ocean

topography, we now have an excellent global

view of the distribution of all of these features

(e.g., Figure 2.1; Smith & Sandwell, 1997), and

can relate many of the features to plate tectonic

METHODS FOR MAPPING BOTTOM TOPOGRAPHY 19

processes (Section 2.2) and sedimentation sources

and processes.

2.8. SILLS, STRAITS, AND

PASSAGES

Sills, straits, and passages connect separate

ocean regions. A sill is a ridge, above the average

bottom level in a region, which separates one

basin from another or, in the case of a fjord

(Section 5.1), separates the landward basin from

the sea outside. The sill depth is the depth from

the sea surface to the deepest part of the ridge;

that is, the maximum depth at which direct

flow across the sill is possible. An oceanic sill is

like a topographic saddle with the sill depth

analogous to the saddlepoint. In the deep ocean,

sills connect deep basins. The sill depth controls

the density of waters that can flow over the ridge.

Straits, passages, and channels are horizontal

constrictions. It is most common to refer to a strait

when considering landforms, such as the Strait of

Gibraltar that connects the Mediterranean Sea

and the Atlantic Ocean, or the Bering Strait that

connects the Bering Sea and the Arctic Ocean.

Passages and channels can also refer to submarine

topography, such as in fracture zones that

connect deep basins. Straits and sills can occur

together, as in both of these examples. The

minimum width of the strait and the maximum

depth of the sill can hydraulically control the

flow passing through the constriction.

2.9. METHODS FOR MAPPING

BOTTOM TOPOGRAPHY

Our present knowledge of the shape of the

ocean floor results from an accumulation of

sounding measurements (most of which have

been made within the last century) and, more

recently, using the gravity field measured by

satellites (Smith & Sandwell, 1997). The early

measurements were made by lowering a weight

on a measured line until the weight touched

bottom, as discussed in Chapter S1, Section

S1.1 located on the textbook Web site http://

booksite.academicpress.com/DPO/; “S” denotes

supplemental material. This method was slow;

in deep water it was uncertain because it

was difficult to tell when the weight touched

the bottom and if the line was vertical.

Since 1920 most depth measurements have

been made with echo sounders, which measure

the time taken for a pulse of sound to travel

from the ship to the bottom and reflect back to

the ship. One half of this time is multiplied by

the average speed of sound in the seawater

under the ship to give the depth. With presentday

equipment, the time can be measured very

accurately and the main uncertainty over a flat

bottom is in the value used for the speed of

sound. This varies with water temperature and

salinity (see Section 3.7), and if these are not

measured at the time of sounding an average

value must be used. Research and military ships

are generally outfitted with echo sounders and

routinely report their bathymetric data to data

centers that compile the information for bathymetric

mapping. The bathymetry along the

research ship track in Figure 2.5b was measured

using this acoustic method.

The modern extension of these single echo

sounders is a multi-beam array, in which many

sounders are mounted along the bottom of the

ship; these provide two-dimensional “swath”

mapping of the seafloor beneath the ship.

Great detail has been added to our knowledge

of the seafloor topography by satellite measurements.

These satellites measure the earth’s

gravity field, which depends on the local mass

of material. These measurements allow mapping

of many hitherto unknown features, such as fracture

zones and seamounts in regions remote

from intensive echo sounder measurements,

and provide much more information about these

features even where they had been mapped

(Smith & Sandwell, 1997). Echo sounder

measurements are still needed to verify the

20


gravity-based measurements since the material

on the ocean bottom is not uniform; for instance,

a bottom shaped by extensive sediment cover

might not be detected from the gravity field.

The bathymetry shown in Figure 2.5c, as well

as in the global and basin maps of Figures 2.1

and Figures 2.8e2.12, isablendedproductof

all available shipboard measurements and satellite-based

measurements.

2.10. BOTTOM MATERIAL

On the continental shelf and slope most of the

bottom material comes directly from the land,

either brought down by rivers or blown by the

wind. The material of the deep-sea bottom is

often more fine-grained than that on the shelf

or slope. Much of it is pelagic in character, that

is, it has been formed in the open ocean. The

two major deep ocean sediments are “red”

clay and the biogenic “oozes.” The former has

less than 30% biogenic material and is mainly

mineral in content. It consists of fine material

from the land (which may have traveled great

distances through the air before finally settling

into the ocean), volcanic material, and meteoric

remains. The oozes are over 30% biogenic and

originate from the remains of living organisms

(plankton). The calcareous oozes have a high

percentage of calcium carbonate from the shells

of animal plankton, while the siliceous oozes

have a high proportion of silica from the shells

of silica-secreting planktonic plants and animals.

The siliceous oozes are found mainly in

the Southern Ocean and in the equatorial

Pacific. The relative distribution of calcareous

and siliceous oozes is clearly related to the

nutrient content of the surface waters, with

calcareous oozes common in low nutrient

regions and siliceous oozes in high nutrient

regions.

Except when turbidity currents deposit their

loads on the ocean bed, the average rate of deposition

of the sediments is from 0.1 to 10 mm per

1000 years, and a large amount of information

on the past history of the oceans is stored in

them. Samples of bottom material are obtained

with a “corer,” which is a 2e30 m long steel

pipe that is lowered vertically and forced to

penetrate into the sediments by the heavy

weight at its upper end. The “core” of sediment

retained in the pipe may represent material

deposited from 1000 to 10 million years per

meter of length. Sometimes the material is

layered, indicating stages of sedimentation of

different materials. In some places, layers

of volcanic ash are related to historical records

of eruptions; in others, organisms characteristic

of cold or warm waters are found in different

layers and suggest changes in temperature of

the overlying water during the period represented

by the core. In some places gradations

from coarse to fine sediments in the upward

direction suggest the occurrence of turbidity

currents bringing material to the region with

the coarser material settling out first and the

finer later.

Large sediment depositions from rivers create

a sloping, smooth ocean bottom for thousands of

miles from the mouths of the rivers. This is

called a deep-sea sediment fan. The largest, the

Bengal Fan, is in the northeastern Indian Ocean

and is created by the outflow from many rivers

including the Ganges and Brahmaputra. Other

examples of fans are at the mouths of the Yangtze,

Amazon, and Columbia Rivers.

Physical oceanographers use sediments to

help trace movement of the water at the ocean

floor. Some photographs of the deep-sea bottom

show ripples similar to a sand beach after the

tide has gone out. Such ripples are only found

on the beach where the water speed is high,

such as in the backwash from waves. We

conclude from the ripples on the deep-ocean

bottom that currents of similar speed occur

there. This discovery helped to dispel the earlier

notion that all deep-sea currents are very slow.

Sediments can affect the properties of

seawater in contact with them; for instance,

OCEAN BASINS 21

silicate and carbonate are dissolved from sediments

into the overlying seawater. Organic

carbon, mainly from fecal pellets, is biologically

decomposed (remineralized) into inorganic

carbon dioxide in the sediments with oxygen

consumed in the process. The carbon dioxiderich,

oxygen-poor pore waters in the sediments

are released back into the seawater, affecting

its composition. Organic nitrogen and phosphorus

are also remineralized in the sediments,

providing an important source of inorganic

nutrients for seawater. In regions where all

oxygen is consumed, methane forms from

bacterial action. This methane is often stored

in solid form called a methane hydrate. Vast

quantities (about 10 19 g) of methane hydrate

have accumulated in marine sediments over

the earth’s history. They can spontaneously

turn from solid to gaseous form, causing submarine

landslides and releasing methane into the

water, affecting its chemistry.

2.11. OCEAN BASINS

The Pacific Ocean (Figure 2.8) is the world’s

largest ocean basin. To the north there is a physical

boundary broken only by the Bering Strait,

which is quite shallow (about 50 m) and 82 km

wide. There is a small net northward flow

from the Pacific to the Arctic through this strait.

At the equator, the Pacific is very wide so that

tropical phenomena that propagate east-west

take much longer to cross the Pacific than across

the other oceans. The Pacific is rimmed in the

west and north with trenches and ridges. This

area, because of the associated volcanoes, is

called the “ring of fire.” The EPR, a major topographic

feature of the tropical and South Pacific,

is a spreading ridge that separates the deep

waters of the southeast from the rest of the

Pacific; it is part of the global mid-ocean ridge

(Section 2.2). Fracture zones allow some

communication of deep waters across the ridge.

Where the major eastward current of the

Southern Ocean, the ACC (Chapter 13), encounters

the ridge, the current is deflected.

The Pacific has more islands than any other

ocean. Most of them are located in the western

tropical regions. The Hawaiian Islands and their

extension northwestward into the Emperor

Seamounts were created by motion of the Pacific

oceanic plate across the hotspot that is now

located just east of the big island of Hawaii.

The Pacific Ocean has numerous marginal

seas, mostly along its western side. In the North

Pacific these are the Bering, Okhotsk, Japan,

Yellow, East China, and South China Seas in

the west and the Gulf of California in the east.

In the South Pacific the marginal seas are the

Coral and Tasman Seas and many smaller

distinct regions that are named, such as the

Solomon Sea (not shown). In the southern South

Pacific is the Ross Sea, which contributes to the

bottom waters of the world ocean.

The Atlantic Ocean has an “S” shape

(Figure 2.9). The MAR, a spreading ridge

down the center of the ocean, dominates its

topography. Deep trenches are found just east

of the Lesser Antilles in the eastern Caribbean

and east of the South Sandwich Islands. The

Atlantic is open both at the north and the south

connecting to the Arctic and Southern Oceans.

The northern North Atlantic is one of the two

sources of the world’s deep water (Chapter 9).

One of the Atlantic’s marginal seas, the Mediterranean,

is evaporative and contributes high

salinity, warm water to the mid-depth ocean.

At the southern boundary, the Weddell Sea is

a major formation site for the bottom water

found in the oceans (Chapter 13). Other

marginal seas connecting to the Atlantic are

the Norwegian, Greenland, and Iceland Seas

(sometimes known collectively as the Nordic

Seas), the North Sea, the Baltic Sea, the Black

Sea, and the Caribbean. The Irminger Sea is

the region southeast of Greenland, the Labrador

Sea is the region between Labrador and Greenland,

and the Sargasso Sea is the open ocean

region surrounding Bermuda. Fresh outflow

22


120°E 180° 120°W 60°W

80°N 80°N

60°N 60°N

Okhotsk

Sea

Bering

Sea

40°N

East Sea

40°N

North Pacific

20°N Philippine

20°N

South

China

Sea

Yellow

Sea

East

China Sea

Basin

Japan or

Izu-Ogasawara Ridge

Mariana T.

Kuril Islands

Shatsky

Rise

Kamchatka

Emperor Seamounts

Wake

Island Psg.

Bering

Strait

Aleutian Islands

Hawaiian Islands

Gulf of

Alaska

Gulf of

California

FIGURE 2.8 Map of

the Pacific Ocean. Etopo2

bathymetry data from

NOAA NGDC (2008).

Celebes

Sea

0° 0°

Banda

Sea

Solomon Sea

Galapagos

Coral

Sea

20°S 20°S

South Pacific

Tasman

40°S 40°S

Sea

Fiji

Campbell

Plateau

Samoan

Psg.

Southern Ocean

60°S Drake Passage 60°S

Ross Sea

80°S 80°S

120°E 180° 120°W 60°W

East Pacific Rise

Bellingshausen

Basin

Depth (m)

1000 2000 3000 4000 5000 6000 7000

from large rivers such as the Amazon, Congo,

and Orinoco Rivers form marked low-salinity

tongues at the sea surface.

The Indian Ocean (Figure 2.10) is closed off by

land just north of the tropics. The topography of

the Indian Ocean is very rough because of the

ridges left behind as the Indian plate moved

northward into the Asian continent creating

the Himalayas. The Central Indian Ridge and

Southwest Indian Ridge are two of the slowest

spreading ridges on Earth. (As discussed previously,

seafloor roughness from abyssal hills and

fracture zones is highest at slower spreading

rates, which is necessary in understanding the

spatial distribution of deep mixing in the global

ocean.) The only trench is the Sunda Trench

where the Indian plate subducts beneath Indonesia.

The eastern boundary of the Indian Ocean

is porous and connected to the Pacific Ocean

through the Indonesian archipelago. Marginal

seas for the Indian Ocean include the Andaman

Sea, the Red Sea, and the Persian Gulf. The open

ocean region west of India is called the Arabian

Sea and the region east of India is called the Bay

of Bengal.

The differential heating of land and ocean in

the tropics results in the creation of the monsoon

weather system. Monsoons occur in many places,

but the most dramatic and best described

monsoon is in the northern Indian Ocean

(Chapter 11). From October to May the Northeast

Monsoon sends cool, dry winds from the

OCEAN BASINS 23

20°N

0°

40°N

60°N

G. Mexico

Cape

Hatteras

Caribbean

Sea

80°N

Hudson Bay

Bahamas

Sargasso

Sea

Orinoco R.

New England

Seamounts

Antilles

Baffin

Bay

60°W 0°

North Atlantic

Amazon R.

Labrador

Sea

Grand

Banks

Mid-Atlantic

Ridge

Charlie Gibbs FZ

Vema FZ

Denmark

Strait

Irminger

Sea

Iceland B.

Reykjanes R.

Azores

Nordic Seas

Iceland-

Faroe R.

Rockall

Plateau

Rockall

Trough

Bay of

Biscay

Canary Isl.

Cape Verde Isl.

Romanche FZ

North

Sea

80°N

Baltic

Sea

Mediterranean Sea

Strait of

Gibraltar

60°N

Black

Sea

Congo R.

40°N

20°N

0°

Angola

Basin

20°S

40°S

Argentine

Basin

Brazil B.

South Atlantic

Mid-Atlantic

Ridge

Walvis Ridge

Cape

Basin

40°S

20°S

60°S

Drake Passage

Scotia Sea

Southern Ocean

60°S

Weddell

Sea

Depth (m)

80°S

60°W 0°

80°S

1000 2000 3000 4000 5000 6000 7000

FIGURE 2.9 Map of the Atlantic Ocean. Etopo2 bathymetry data from NOAA NGDC (2008).

continental land masses in the northeast over

the Indian subcontinent to the ocean. Starting

in June and lasting until September, the system

shifts to the southwest monsoon, which brings

warm, wet rains from the western tropical

ocean to the Indian subcontinent. While these

monsoon conditions are best known in India,

they also dominate the climate in the western

tropical and South Pacific.

Most of the rivers that drain southward from

the Himalayas d including the Ganges, Brahmaputra,

and Irawaddy d flow out into the

24


20°E 40°E 60°E 80°E 100°E 120°E 140°E

40°N 40°N

Mediterranean Sea

Red

Sea

Persian

Gulf

20°N 20°N

Gulf of Aden

Somali Basin

Arabian

Sea

Bay of

Bengal

Andaman

Sea

Pacific

Ocean

0° Indonesian

0°

Archipelago

Amirante Pa.

Carlsberg Ridge

Mascarene

Basin

Gulf of

Oman

Chagos-Laccadives Ridge

Central Indian

Basin

Indian

Ocean

N. Australia Basin

Banda Sea

Mozambique

20°S Channel

W. Australia Basin

20°S

Ninety-East Ridge

Central Indian Ridge

Mozambique

Basin

Madagascar

Basin

Crozet

Basin

Broken Plateau

Perth

Basin

Naturaliste

Plateau

40°S 40°S

Agulhas Basin

Crozet Plateau

Southwest Indian Ridge

Southeast Indian Ridge

S. Australia Basin

Southern Ocean

Enderby Basin Agulhas

Kerguelen Plateau

Plateau

Australian-Antarctic Basin

60°S 60°S

Depth (m)

80°S 80°S

10°E 40°E 80°E 120°E 150°E

1000 2000 3000 4000 5000 6000 7000

FIGURE 2.10 Map of the Indian Ocean. Etopo2 bathymetry data from NOAA NGDC (2008).

Bay of Bengal, east of India rather than into the

Arabian Sea, west of India. This causes the

surface water of the Bay of Bengal to be quite

fresh. The enormous amount of silt carried by

these rivers from the eroding Himalayan Mountains

into the Bay of Bengal creates the subsurface

geological feature, the Bengal Fan, which

slopes smoothly downward for thousands of

kilometers. West of India, the Arabian Sea, Red

Sea, and Persian Gulf are very salty due to the

dry climate and subsequent high evaporation.

Similar to the Mediterranean, the saline Red

Sea water is sufficiently dense to sink to middepth

in the Indian Ocean and affects water

properties over a large part of the Arabian Sea

and western Indian Ocean.

OCEAN BASINS 25

180˚

150˚W

Bering

Strait

150˚E

90˚W

120˚W

Hudson

Bay

60˚W

Hudson Strait

Mackenzie R.

Fury &

Hecla St.

Baffin I.

Davis

Strait

Lancaster Sound

Baffin

Bay

30˚W

Beaufort

Sea

Canada

Basin

Canadian

Archipelago

Arctic Ocean

Ellesmere I.

Nares Strait

Denmark

Strait

Chukchi

Sea

Canadian

Basin

Fram

Strait

Greenland

Sea

Jan Mayen

Iceland

Sea

Iceland-Faroe

Ridge

East Siberian

Sea

Makarov

Basin

Lomonosov Ridge

Amundsen Basin

Norwegian Sea

Faroe-Shetland

Channel

Laptev Sea

Eurasian

Basin

Nansen Basin

Franz Josef

Land

Svalbard

Severnaya

Zemlya

Novaya

Zemlya

Barents Sea

Kara

Sea

Lena R.

30˚E

Yenisei R.

Ob R.

120˚E

60˚E

90˚E

0˚

Depth (m)

1000 2000 3000 4000 5000

FIGURE 2.11 The Arctic Ocean. Etopo2 bathymetry data from NOAA NGDC (2008).

The Arctic Ocean (Figure 2.11) is sometimes

not regarded as an ocean, but rather as a mediterranean

sea connected to the Atlantic Ocean.

It is characterized by very broad continental

shelves surrounding a deeper region, which is

split down the center by the Lomonosov Ridge.

These shelf areas around the Arctic are called

the Beaufort, Chukchi, East Siberian, Laptev,

Kara, and Barents Seas. The Arctic is connected

to the North Pacific through the shallow Bering

Strait. It is connected to the Nordic Seas (Norwegian

and Greenland) through passages on either

side of Svalbard, including Fram Strait between

Svalbard and Greenland. The Nordic Seas are

separated from the Atlantic Ocean by the

submarine ridge between Greenland, Iceland,

26


90°W

120˚W

60˚W

East Pacific Rise

180˚

150˚W

SW Pacific

Basin

South Pacific

Eltanin Fracture Zone

Udintsev Fracture Zone

Ross Sea

Amundsen Abyssal

Plain

Drake Passage

Antarctic

Penin.

Falkland

Plateau

Scotia Sea

Weddell Sea

Argentine

Basin

30˚W

Mid-Atlantic Ridge

South

Atlantic

0˚

150˚E

Campbell

Plateau

Tasman

Sea

Macquarie Ridge

S. Australia Basin

120˚E

Adélie

Land

Australian-Antarctic

Basin

Southeast Indian Ridge

Wilkes Land

Prydz Bay

Kerguelen Plateau

Indian

Enderby Basin

Crozet

Basin

Southwest Indian Ridge

60˚E

Cape

Basin

30˚E

Depth (m)

1000 2000 3000 4000 5000 6000 7000

FIGURE 2.12 The Southern Ocean around Antarctica. Etopo2 bathymetry data from NOAA NGDC (2008).

90°E

and the UK, with a maximum sill depth of about

620 m in the Denmark Strait, between Greenland

and Iceland. Dense water formed in the

Nordic Seas spills into the Atlantic over this

ridge. The central area of the Arctic Ocean is

perennially covered with sea ice.

The Southern Ocean (Figure 2.12) is not

geographically distinct from the Atlantic,

Indian, and Pacific Oceans, but is often considered

separately since it is the only region

outside the Arctic where there is a path for eastward

flow all the way around the globe. This

occurs at the latitude of Drake Passage between

South America and Antarctica and allows

the three major oceans to be connected.

The absence of a meridional (north-south)

OCEAN BASINS 27

boundary in Drake Passage changes the

dynamics of the flow at these latitudes

completely in comparison with the rest of the

ocean, which has meridional boundaries.

Drake Passage also serves to constrict the

width of the flow of the ACC, which must

pass in its entirety through the passage. The

South Sandwich Islands and trench east of

Drake Passage partially block the open circumpolar

flow. Another major constriction is the

broad Pacific-Antarctic rise, which is the

seafloor spreading ridge between the Pacific

and Antarctic plates. This fast-spreading ridge

has few deep fracture zones, so the ACC must

deflect northward before finding the only two

deep channels, the Udintsev and Eltanin Fracture

Zones.

The ocean around Antarctica includes

permanent ice shelves as well as seasonal sea

ice (Figures 13.11 and 13.19). Unlike the Arctic

there is no perennial long-term pack ice; except

for some limited ice shelves and all of the firstyear

ice melts and forms each year. The densest

bottom waters of the world ocean are formed in

the Southern Ocean, primarily in the Weddell

and Ross Seas as well as in other areas distributed

along the Antarctic coast between the

Ross Sea and Prydz Bay.

C H A P T E R

3

Physical Properties of Seawater

3.1. MOLECULAR PROPERTIES

OF WATER

Many of the unique characteristics of the

ocean can be ascribed to the nature of water

itself. Consisting of two positively charged

hydrogen ions and a single negatively charged

oxygen ion, water is arranged as a polar molecule

having positive and negative sides. This

molecular polarity leads to water’s high dielectric

constant (ability to withstand or balance an

electric field). Water is able to dissolve many

substances because the polar water molecules

align to shield each ion, resisting the recombination

of the ions. The ocean’s salty character is

due to the abundance of dissolved ions.

The polar nature of the water molecule

causes it to form polymer-like chains of up to

eight molecules. Approximately 90% of the

water molecules are found in these chains.

Energy is required to produce these chains,

which is related to water’s heat capacity. Water

has the highest heat capacity of all liquids

except ammonia. This high heat capacity is the

primary reason the ocean is so important in

the world climate system. Unlike the land and

atmosphere, the ocean stores large amounts of

heat energy it receives from the sun. This heat

is carried by ocean currents, exporting or

importing heat to various regions. Approximately

90% of the anthropogenic heating

associated with global climate change is stored

in the oceans, because water is such an effective

heat reservoir (see Section S15.6 located on the

textbook Web site http://booksite.academic

press.com/DPO/; “S” denotes supplemental

material).

As seawater is heated, molecular activity

increases and thermal expansion occurs,

reducing the density. In freshwater, as temperature

increases from the freezing point up to about

4 C, the added heat energy forms molecular

chains whose alignment causes the water to

shrink, increasing the density. As temperature

increases above this point, the chains break

down and thermal expansion takes over; this

explains why fresh water has a density maximum

at about 4 C rather than at its freezing point. In

seawater, these molecular effects are combined

with the influence of salt, which inhibits the

formation of the chains. For the normal range of

salinity in the ocean, the maximum density

occurs at the freezing point, which is depressed

to well below 0 C(Figure 3.1).

Water has a very high heat of evaporation (or

heat of vaporization) and a very high heat of

fusion. The heat of vaporization is the amount

of energy required to change water from a liquid

to a gas; the heat of fusion is the amount of

energy required to change water from a solid

to a liquid. These quantities are relevant for

our climate as water changes state from a liquid

in the ocean to water vapor in the atmosphere

and to ice at polar latitudes. The heat energy


29



30

3. PHYSICAL PROPERTIES OF SEAWATER

30

Salinity

0 10 20 30 40

FIGURE 3.1 Values of density s t

(curved lines) and the loci of maximum

density and freezing point (at atmospheric

pressure) for seawater as functions of

temperature and salinity. The full density r

is 1000 þ s t with units of kg/m 3 .

Temperature (°C)

20

10

0

5

10

15

20

25

30

90%

of ocean

0

–2

Temp. of max. density

Freezing point

Mean

T&S

Most

abundant

involved in these state changes is a factor in

weather and in the global climate system.

Water’s chain-like molecular structure also

produces its high surface tension. The chains

resist shear, giving water a high viscosity for

its atomic weight. This high viscosity permits

formation of surface capillary waves, with wavelengths

on the order of centimeters; the restoring

forces for these waves include surface tension as

well as gravity. Despite their small size, capillary

waves are important in determining the

frictional stress between wind and water. This

stress generates larger waves and propels the

frictionally driven circulation of the ocean’s

surface layer.

3.2. PRESSURE

Pressure is the normal force per unit area

exerted by water (or air in the atmosphere) on

both sides of the unit area. The units of force

are (mass length/time 2 ). The units of pressure

are (force/length 2 ) or (mass/[length time 2 ]).

Pressure units in centimeters-gram-second (cgs)

are dynes/cm 2 and in meter-kilogram-second

(mks) they are Newtons/m 2 . A special unit for

pressure is the Pascal, where 1 Pa ¼ 1 N/m 2 .

Atmospheric pressure is usually measured in

bars where 1 bar ¼ 10 6 dynes/cm 2 ¼ 10 5 Pa.

Ocean pressure is usually reported in decibars

where 1 dbar ¼ 0.1 bar ¼ 10 5 dyne/cm 2 ¼

10 4 Pa.

The force due to pressure arises when there is

a difference in pressure between two points. The

force is directed from high to low pressure.

Hence we say the force is oriented “down the

pressure gradient” since the gradient is directed

from low to high pressure. In the ocean, the

downward force of gravity is mostly balanced

by an upward pressure gradient force; that is,

the water is not accelerating downward.

Instead, it is kept from collapsing by the upward

pressure gradient force. Therefore pressure

increases with increasing depth. This balance

of downward gravity force and upward pressure

gradient force, with no motion, is called

hydrostatic balance (Section 7.6.1).

The pressure at a given depth depends on the

mass of water lying above that depth. A pressure

change of 1 dbar occurs over a depth change of

slightly less than 1 m (Figure 3.2 and Table 3.1).

Pressure in the ocean thus varies from near

zero (surface) to 10,000 dbar (deepest). Pressure

PRESSURE 31

Depth (m)

6000

4000

2000

TABLE 3.1

0

0 2000 4000 6000

Pressure (dbar)

FIGURE 3.2 The relation between depth and pressure,

using a station in the northwest Pacific at 41 53’N, 146

18’W.

Comparison of Pressure (dbar) and Depth

(m) at Standard Oceanographic Depths

Using the UNESCO (1983) Algorithms

Pressure (dbar) Depth (m) Difference (%)

0 0 0

100 99 1

200 198 1

300 297 1

500 495 1

1000 990 1

1500 1453 1.1

2000 1975 1.3

3000 2956 1.5

4000 3932 1.7

5000 4904 1.9

6000 5872 2.1

Percent difference ¼ (pressure depth)/pressure 100%.

is usually measured in conjunction with other

seawater properties such as temperature,

salinity, and current speeds. The properties are

often presented as a function of pressure rather

than depth.

Horizontal pressure gradients drive the horizontal

flows in the ocean. For large-scale

currents (of horizontal scale greater than a kilometer),

the horizontal flows are much stronger

than their associated vertical flows and are

usually geostrophic (Chapter 7). The horizontal

pressure differences that drive the ocean

currents are on the order of one decibar over

hundreds or thousands of kilometers. This is

much smaller than the vertical pressure

gradient, but the latter is balanced by the downward

force of gravity and does not drive a flow.

Horizontal variations in mass distribution

create the horizontal variation in pressure in

the ocean. The pressure is greater where the

water column above a given depth is heavier

either because it is higher density or because it

is thicker or both.

Pressure is usually measured with an electronic

instrument called a transducer. The accuracy

and precision of pressure measurements

is high enough that other properties such as

temperature, salinity, current speeds, and so

forth can be displayed as a function of pressure.

However, the accuracy, about 3 dbar at depth, is

not sufficient to measure the horizontal pressure

gradients. Therefore other methods, such as the

geostrophic method, or direct velocity measurements,

must be used to determine the actual

flow. Prior to the 1960s and 1970s, pressure

was measured using a pair of mercury thermometers,

one of which was in a vacuum

(“protected” by a glass case) and not affected

by pressure while the other was exposed to the

water (“unprotected”) and affected by pressure,

as described in the following section. More

information about these instruments and

methods is provided in Section S6.3 of the

supplementary materials on the textbook Web

site.

32


3.3. THERMAL PROPERTIES OF

SEAWATER: TEMPERATURE,

HEAT, AND POTENTIAL

TEMPERATURE

One of the most important physical characteristics

of seawater is its temperature. Temperature

was one of the first ocean parameters to be

measured and remains the most widely

observed. In most of the ocean, temperature is

the primary determinant of density; salinity is

of primary importance mainly in high latitude

regions of excess rainfall or sea ice processes

(Section 5.4). In the mid-latitude upper ocean

(between the surface and 500 m), temperature

is the primary parameter determining sound

speed. (Temperature measurement techniques

are described in Section S6.4.2 of the supplementary

materials on the textbook Web site.)

The relation between temperature and

heat content is described in Section 3.3.2. As

a parcel of water is compressed or expanded,

its temperature changes. The concept of “potential

temperature” (Section 3.3.3) takes these

pressure effects into account.

3.3.1. Temperature

Temperature is a thermodynamic property of

a fluid, due to the activity or energy of molecules

and atoms in the fluid. Temperature is

higher for higher energy or heat content. Heat

and temperature are related through the specific

heat (Section 3.3.2).

Temperature (T) in oceanography is usually

expressed using the Celsius scale ( C), except

in calculations of heat content, when temperature

is expressed in degrees Kelvin (K). When

the heat content is zero (no molecular activity),

the temperature is absolute zero on the Kelvin

scale. (The usual convention for meteorology is

degrees Kelvin, except in weather reporting,

since atmospheric temperature decreases to

very low values in the stratosphere and above.)

A change of 1 C is the same as a change of 1 K. A

temperature of 0 C is equal to 273.16 K. The

range of temperature in the ocean is from the

freezing point, which is around 1.7 C (depending

on salinity), to a maximum of around 30 C

in the tropical oceans. This range is considerably

smaller than the range of air temperatures. As

for all other physical properties, the temperature

scale has been refined by international

agreement. The temperature scale used most

often is the International Practical Temperature

Scale of 1968 (IPTS-68). It has been superseded

by the 1990 International Temperature Scale

(ITS-90). Temperatures should be reported in

ITS-90, but all of the computer algorithms

related to the equation of state that date from

1980 predate ITS-90. Therefore, ITS-90 temperatures

should be converted to IPTS-68 by multiplying

ITS-90 by 0.99976 before using the 1980

equation of state subroutines.

The ease with which temperature can be

measured has led to a wide variety of oceanic

and satellite instrumentation to measure ocean

temperatures (see supplementary material in

Section S6.4.2 on the textbook Web site).

Mercury thermometers were in common use

from the late 1700s through the 1980s. Reversing

(mercury) thermometers, invented by Negretti

and Zamba in 1874, were used on water sample

bottles through the mid-1980s. These thermometers

have ingenious glasswork that cuts off the

mercury column when the thermometers are

flipped upside down by the shipboard observer,

thus recording the temperature at depth. The

accuracy and precision of reversing thermometers

is 0.004 and 0.002 C. Thermistors are now

used for most in situ measurements. The best

thermistors used most often in oceanographic

instruments have an accuracy of 0.002 C and

precision of 0.0005e0.001 C.

Satellites detect thermal infrared electromagnetic

radiation from the sea surface; this radiation

is related to temperature. Satellite sea

surface temperature (SST) accuracy is about

0.5e0.8 K, plus an additional error due to the

THERMAL PROPERTIES OF SEAWATER: TEMPERATURE, HEAT, AND POTENTIAL TEMPERATURE 33

presence or absence of a very thin (10 mm) skin

layer that can reduce the desired bulk (1e2 m)

observation of SST by about 0.3 K.

3.3.2. Heat

The heat content of seawater is its thermodynamic

energy. It is calculated using the

measured temperature, measured density, and

the specific heat of seawater. The specific heat

is a thermodynamic property of seawater

expressing how heat content changes with

temperature. Specific heat depends on temperature,

pressure, and salinity. It is obtained from

formulas that were derived from laboratory

measurements of seawater. Tables of values or

computer subroutines supplied by UNESCO

(1983) are available for calculating specific

heat. The heat content per unit volume, Q, is

computed from the measured temperature

using

Q ¼ rc p T (3.1)

where T is temperature in degrees Kelvin, r is

the seawater density, and c p is the specific heat

of seawater. The mks units of heat are Joules,

that is, units of energy. The rate of time change

of heat is expressed in Watts, where 1 W ¼ 1

J/sec. The classical determinations of the

specific heat of seawater were reported by

Thoulet and Chevallier (1889). In 1959, Cox

and Smith (1959) reported new measurements

estimated to be accurate to 0.05%, with values

1 to 2% higher than the old ones. A further study

(Millero, Perron, & Desnoyers, 1973) yielded

values in close agreement with those of Cox

and Smith.

The flux of heat through a surface is defined

as the amount of energy that goes through the

surface per unit time, so the mks units of heat

flux are W/m 2 . The heat flux between the atmosphere

and ocean depends in part on the

temperature of the ocean and atmosphere.

Maps of heat flux are based on measurements

of the conditions that cause heat exchange

(Section 5.4). As a simple example, what heat

loss from a 100 m thick layer of the ocean is

needed to change the temperature by 1 Cin30

days? The required heat flux is rc p DT V/Dt.

Typical values of seawater density and specific

heat are about 1025 kg/m 3 and 3850 J/(kg C).

V is the volume of the 100 m thick layer, which

is 1 m 2 across, and Dt is the amount of time

(sec). The calculated heat change is 152 W. The

heat flux through the surface area of 1 m 2 is

thus about 152 W/m 2 . In Chapter 5 all of

the components of ocean heat flux and their

geographic distributions are described.

3.3.3. Potential Temperature

Seawater is almost, but not quite, incompressible.

A pressure increase causes a water

parcel to compress slightly. This increases the

temperature in the water parcel if it occurs

without exchange of heat with the surrounding

water (adiabatic compression). Conversely if

a water parcel is moved from a higher to a lower

pressure, it expands and its temperature

decreases. These changes in temperature are

unrelated to surface or deep sources of heat. It

is often desirable to compare the temperatures

of two parcels of water that are found at

different pressures. Potential temperature is

defined as the temperature that a water parcel

would have if moved adiabatically to another

pressure. This effect has to be considered

when water parcels change depth.

The adiabatic lapse rate or adiabatic temperature

gradient is the change in temperature per unit

change in pressure for an adiabatic displacement

of a water parcel. The expression for the

lapse rate is

GðS; T; pÞ ¼ vT

vp (3.2)

heat

where S, T, and p are the measured salinity,

temperature, and pressure and the derivative

34


is taken holding heat content constant. Note that

both the compressibility and the adiabatic lapse

rate of seawater are functions of temperature,

salinity, and pressure. The adiabatic lapse rate

was determined for seawater through laboratory

measurements. Since the full equation of

state of seawater is a complicated function of

these quantities, the adiabatic lapse rate is also

a complicated polynomial function of temperature,

salinity, and pressure. In contrast, the lapse

rate for ideal gases can be derived from basic

physical principles; in a dry atmosphere the

lapse rate is approximately 9.8 C/km. The lapse

rate in the ocean, about 0.1 to 0.2 C/km, is much

smaller since seawater is much less compressible

than air. The lapse rate is calculated using

computer subroutines based on UNESCO

(1983).

The potential temperature is (Fofonoff, 1985):

qðS; T; pÞ ¼T þ

Z pr

p

GðS; T; pÞdp (3.3)

where S, T, and p are the measured (in situ)

salinity, temperature, and pressure, G is the

adiabatic lapse rate, and q is the temperature

that a water parcel of properties (S, T, p) would

have if moved adiabatically and without

change of salinity from an initial pressure p

to a reference pressure p r where p r may be

greater or less than p. The integration above

can be carried out in a single step (Fofonoff,

1977). An algorithm for calculating q is given

by UNESCO (1983), using the UNESCO adiabatic

lapse rate (Eq. 3.2); computer subroutines

in a variety of different programming

languages are readily available. The usual

convention for oceanographic studies is to

reference potential temperature to the sea

surface. When defined relative to the sea

surface, potential temperature is always lower

than the actual measured temperature, and

only equal to temperature at the sea surface.

(On the other hand, when calculating potential

density referenced to a pressure other than sea

surface pressure, potential temperature must

also be referenced to the same pressure; see

Section 3.5.)

As an example, if a water parcel of temperature

5 C and salinity 35.00 were lowered adiabatically

from the surface to a depth of 4000 m,

its temperature would increase to 5.45 C due

to compression. The potential temperature

relative to the sea surface of this parcel is

always 5 C, while its measured, or in situ,

temperature at 4000 m is 5.45 C. Conversely,

if its temperature was 5 C at 4000 m depth

and it was raised adiabatically to the surface,

its temperature would change to 4.56 C due

to expansion. The potential temperature of

this parcel relative to the sea surface is thus

4.56 C. Temperature and potential temperature

referenced to the sea surface from a profile in

the northeastern North Pacific are shown in

Figure 3.3. Compressibility itself depends on

temperature (and salinity), as discussed in

Section 3.5.4.

3.4. SALINITY AND

CONDUCTIVITY

Seawater is a complicated solution containing

the majority of the known elements. Some

of the more abundant components, as percent

of total mass of dissolved material, are chlorine

ion (55.0%), sulfate ion (7.7%), sodium ion

(30.7%), magnesium ion (3.6%), calcium ion

(1.2%), and potassium ion (1.1%) (Millero, Feistel,

Wright, & McDougall, 2008). While the total

concentration of dissolved matter varies from

place to place, the ratios of the more abundant

components remain almost constant. This “law”

of constant proportions was first proposed

by Dittmar (1884), based on 77 samples of

seawater collected from around the world

during the Challenger Expedition (see Chapter

S1, Section S1.2, on the textbook Web site),

confirming a hypothesis from Forchhammer

(1865).

SALINITY AND CONDUCTIVITY 35

(a) (b) (c)

0 5 10 15 20 30 35 40

33.5 34.0 34.5

0

0

0

1000

1000

1000

Pressure (decibar)

2000

3000

4000

0

1000

2000

3000

4000

5000

θ

T

2000

3000

4000

2000

3000

4000

T

0 1 2 3

5000

θ

5000

5000

0 5 10 15 20

Temperature/potential temperature (°C)

30 35 40

Conductivity (mmho)

33.5 34.0 34.5

Salinity

FIGURE 3.3 (a) Potential temperature (q) and temperature (T) ( C), (b) conductivity (mmho), and (c) salinity in the

northeastern North Pacific (36 30’N, 135 W).

The dominant source of the salts in the ocean

is river runoff from weathering of the continents

(see Section 5.2). Weathering occurs very slowly

over millions of years, and so the dissolved

elements become equally distributed in the

ocean as a result of mixing. (The total time for

water to circulate through the oceans is, at

most, thousands of years, which is much shorter

than the geologic weathering time.) However,

there are significant differences in total concentration

of the dissolved salts from place to place.

These differences result from evaporation and

from dilution by freshwater from rain and river

runoff. Evaporation and dilution processes

occur only at the sea surface.

Salinity was originally defined as the mass in

grams of solid material in a kilogram of

seawater after evaporating the water away;

this is the absolute salinity as described in

Millero et al. (2008). For example, the average

salinity of ocean water is about 35 grams of salts

per kilogram of seawater (g/kg), written as

“S ¼ 35 &” oras“S¼ 35 ppt” and read as

“thirty-five parts per thousand.” Because evaporation

measurements are cumbersome, this

definition was quickly superseded in practice.

In the late 1800s, Forch, Knudsen, and Sorensen

(1902) introduced a more chemically based definition:

“Salinity is the total amount of solid

materials in grams contained in one kilogram

of seawater when all the carbonate has been

converted to oxide, the bromine and iodine

replaced by chlorine, and all organic matter

completely oxidized.”

This chemical determination of salinity was

also difficult to carry out routinely. The method

used throughout most of the twentieth century

was to determine the amount of chlorine ion

(plus the chlorine equivalent of the bromine

and iodine) referred to as chlorinity, by titration

36


with silver nitrate, and then to calculate salinity

by a relation based on the measured ratio of

chlorinity to total dissolved substances. (See

Wallace, 1974, Wilson, 1975, or Millero et al.,

2008 for a full account.) The current definition

of salinity, denoted by S &, is “the mass of silver

required to precipitate completely the halogens

in 0.3285234 kg of the seawater sample.” The

current relation between salinity and chlorinity

was determined in the early 1960s:

Salinity ¼ 1:80655 Chlorinity (3.4)

These definitions of salinity based on chemical

analyses were replaced by a definition based

on seawater’s electrical conductivity, which

depends on salinity and temperature (see Lewis &

Perkin, 1978; Lewis & Fofonoff, 1979; Figure 3.3).

This conductivity-based quantity is called practical

salinity, sometimes using the symbol psu

for practical salinity units, although the preferred

international convention has been to use no units

for salinity. Salinity is now written as, say,

S ¼ 35.00 or S ¼ 35.00 psu. The algorithm that is

widely used to calculate salinity from conductivity

and temperature is called the practical

salinity scale 1978 (PSS 78). Electrical conductivity

methods were first introduced in the

1930s (see Sverdrup, Johnson, & Fleming, 1942

for a review). Electrical conductivity depends

strongly on temperature, but with a small

residual due to the ion content or salinity. Therefore

temperature must be controlled or

measured very accurately during the conductivity

measurement to determine the practical

salinity. Advances in the electrical circuits and

sensor systems permitted accurate compensation

for temperature, making conductivitybased

salinity measurements feasible (see

supplemental materials in Chapter S6, Section

S6.4.3 on the textbook Web site).

Standard seawater solutions of accurately

known salinity and conductivity are required

for accurate salinity measurement. The practical

salinity (S P ) of a seawater sample is now given

in terms of the ratio of the electrical conductivity

of the sample at 15 C and a pressure of one standard

atmosphere to that of a potassium chloride

solution in which the mass fraction of KCl is

32.4356 10 3 at the same temperature and

pressure. The potassium chloride solutions

used as standards are now prepared in a single

laboratory in the UK. PSS 78 is valid for the

range S ¼ 2 to 42, T ¼ 2.0 to 35.0 C and pressures

equivalent to depths from 0 to 10,000 m.

The accuracy of salinity determined from

conductivity is 0.001 if temperature is very

accurately measured and standard seawater is

used for calibration. This is a major improvement

on the accuracy of the older titration

method, which was about 0.02. In archived

data sets, salinities that are reported to three

decimal places of accuracy are derived from

conductivity, while those reported to two places

are from titration and usually predate 1960.

The conversion from conductivity ratio to

practical salinity is carried out using a computer

subroutine based on the formula from Lewis

(1980). The subroutine is part of the UNESCO

(1983) routines for seawater calculations.

In the 1960s, the pairing of conductivity

sensors with accurate thermistors made it

possible to collect continuous profiles of salinity

in the ocean. Because the geometry of the

conductivity sensors used on these instruments

change with pressure and temperature, calibration

with water samples collected at the same

time is required to achieve the highest possible

accuracies of 0.001.

An example of the relationship between

conductivity, temperature, and salinity profiles

in the northeastern North Pacific is shown in

Figure 3.3. Deriving salinity from conductivity

requires accurate temperature measurement

because the conductivity profile closely tracks

temperature.

The concept of salinity assumes negligible

variations in the composition of seawater.

However, a study of chlorinity, density relative

to pure water, and conductivity of seawater

DENSITY OF SEAWATER 37

carried out in England on samples from the

world oceans (Cox, McCartney, & Culkin,

1970) revealed that the ionic composition of

seawater does exhibit small variations from

place to place and from the surface to deep

water. It was found that the relationship

between density and conductivity was a little

closer than between density and chlorinity.

This means that the proportion of one ion to

another may change; that is, the chemical

composition may change, but as long as the total

weight of dissolved substances is the same, the

conductivity and the density will be unchanged.

Moreover, there are geographic variations in

the dissolved substances not measured by the

conductivity method that affect seawater

density and hence should be included in absolute

salinity. The geostrophic currents computed

locally from density (Section 7.6.2), based on the

use of salinity PSS 78, are highly accurate.

However, it is common practice to map properties

on surfaces of constant potential density or

related surfaces that are closest to isentropic

(Section 3.5). On a global scale, these dissolved

constituents can affect the definition of these

surfaces.

The definition of salinity is therefore undergoing

another change equivalent to that of

1978. The absolute salinity recommended by the

IOC, SCOR, and IAPSO (2010) is a return to

the original definition of “salinity,” which is

required for the most accurate calculation of

density; that is, the ratio of the mass of all dissolved

substances in seawater to the mass of

the seawater, expressed in either kg/kg or g/kg

(Millero et al., 2008). The new estimate for absolute

salinity incorporates two corrections over

PSS 78: (1) representation of improved information

about the composition of the Atlantic

surface seawater used to define PSS 78 and

incorporation of 2005 atomic weights, and (2)

corrections for the geographic dependence of

the dissolved matter that is not sensed by

conductivity. To maintain a consistent global

salinity data set, the IOC, SCOR, and IAPSO

(2010) manual strongly recommends that observations

continue to be made based on conductivity

and PSS 78, and reported to national

archives in those practical salinity units. For

calculations involving salinity, the manual indicates

two corrections for calculating the absolute

salinity S A from the practical salinity S P :

S A ¼ S R þ dS A ¼ð35:16504gkg 1 =35ÞS P þ dS A

(3.5)

The factor multiplying S P yields the

“reference salinity” S R , which is presently the

most accurate estimate of the absolute salinity

of reference Atlantic surface seawater. A

geographically dependent anomaly, dS A , is

then added that corrects for the dissolved

substances that do not affect conductivity; this

correction, as currently implemented, depends

on dissolved silica, nitrate, and alkalinity. The

mean absolute value of the correction globally

is 0.0107 g/kg, and it ranges up to 0.025 g/kg

in the northern North Pacific, so it is significant.

If nutrients and carbon parameters are not

measured along with salinity (which is by

far the most common circumstance), then

a geographic lookup table based on archived

measurements is used to estimate the anomaly

(McDougall, Jackett, & Millero, 2010). It is

understood that the estimate (Eq. 3.5) of absolute

salinity could evolve as additional measurements

are made.

All of the work that appears in this book

predates the adoption of the new salinity scale,

and all salinities are reported as PSS 78 and all

densities are calculated according to the 1980

equation of state using PSS 78.

3.5. DENSITY OF SEAWATER

Seawater density is important because it

determines the depth to which a water parcel

will settle in equilibrium d the least dense on

top and the densest at the bottom. The distribution

of density is also related to the large-scale

38


circulation of the oceans through the

geostrophic/thermal wind relationship (see

Chapter 7). Mixing is most efficient between

waters of the same density because adiabatic

stirring, which precedes mixing, conserves

potential temperature and salinity and consequently,

density. More energy is required to

mix through stratification. Thus, property distributions

in the ocean are effectively depicted by

maps on density (isopycnal) surfaces, when

properly constructed to be closest to isentropic.

(See the discussion of potential and neutral

density in Section 3.5.4.)

Density, usually denoted r, is the amount of

mass per unit volume and is expressed in kilograms

per cubic meter (kg/m 3 ). A directly

related quantity is the specific volume anomaly,

usually denoted a, where a ¼ 1/r. The density

of pure water, with no salt, at 0 C, is 1000 kg/m 3

at atmospheric pressure. In the open ocean,

density ranges from about 1021 kg/m 3 (at the

sea surface) to about 1070 kg/m 3 (at a pressure

of 10,000 dbar). As a matter of convenience, it

is usual in oceanography to leave out the first

two digits and use the quantity

s stp ¼ rðS; T; pÞ 1000 kg=m 3 (3.6)

where S ¼ salinity, T ¼ temperature ( C), and

p ¼ pressure. This is referred to as the in situ

density. In earlier literature, s s,t,0 was commonly

used, abbreviated as s t . s t is the density of the

water sample when the total pressure on it has

been reduced to atmospheric (i.e., the water

pressure p ¼ 0 dbar) but the salinity and

temperature are as measured. Unless the analysis

is limited to the sea surface, s t is not the

best quantity to calculate. If there is range of

pressures, the effects of adiabatic compression

should be included when comparing water

parcels. A more appropriate quantity is potential

density, which is the same as s t but with temperature

replaced by potential temperature and

pressure replaced by a single reference pressure

that is not necessarily 0 dbar. Potential density is

described in Section 3.5.2.

The relationship between the density of

seawater and temperature, salinity, and pressure

is the equation of state for seawater. The

equation of state

rðS; T; pÞ ¼rðS; T; 0Þ=½l p=KðS; T; pÞŠ (3.7)

was determined through meticulous laboratory

measurements at atmospheric pressure. The

polynomial expressions for the equation of state

r(S, T, 0) and the bulk modulus K(S, T, p) contain

15 and 27 terms, respectively. The pressure

dependence enters through the bulk modulus.

The largest terms are those that are linear in S,

T, and p, with smaller terms that are proportional

to all of the different products of these.

Thus, the equation of state is weakly nonlinear.

Today, the most common version of Eq. (3.7) is

“EOS 80” (Millero & Poisson, 1980; Fofonoff,

1985). EOS 80 uses the practical salinity scale

PSS 78 (Section 3.4). The formulae may be found

in UNESCO (1983), which provides practical

computer subroutines and are included in

various texts such as Pond and Pickard (1983)

and Gill (1982). EOS 80 is valid for T ¼ 2to

40 C, S ¼ 0 to 40, and pressures from 0 to 10,000

dbar, and is accurate to 9 10 3 kg/m 3 or better.

A new version of the equation of state has been

introduced (IOC, SCOR, and IAPSO, 2010), based

on a new definition of salinity and is termed

TEOS-10. Only EOS 80 is used in this book.

Historically, density was calculated from

tables giving the dependence of the density

on salinity, temperature, and pressure. Earlier

determinations of density were based on

measurements by Forch, Jacobsen, Knudsen,

and Sorensen and were presented in the Hydrographical

Tables (Knudsen, 1901). Cox et al.

(1970) found that the s 0 values (at T ¼ 0 C) in

“Knudsen’s Tables” were low by about 0.01 (on

average) in the salinity range from 15 e 40, and

by up to 0.06 at lower salinities and temperatures.

To determine seawater density over a range

of salinities in the laboratory, Millero (1967)

used a magnetic float densimeter. A Pyrex glass

float containing a permanent magnet floats in a


250 ml cell that contains the seawater and is surrounded

by a solenoid, with the entire apparatus

sitting in a constant temperature bath. The float is

slightly less dense than the densest seawater and

is loaded with small platinum weights until it just

sinks to the bottom of the cell. A current through

the solenoid is then slowly increased until the

float just lifts off the bottom of the cell. The

density of the seawater is then related to the

current through the solenoid. The relation

between current and density is determined by

carrying out a similar experiment with pure

water in the cell. The accuracy of the relative

density determined this way is claimed to

be 2 10 6 (at atmospheric pressure). But as

the absolute density of pure water is known to

be only 4 10 6 , the actual accuracy of seawater

density is more limited. The influence of

pressure was determined using a high pressure

version of the previously mentioned densimeter

to measure the bulk modulus (K). K has also been

determined from measurements of sound speed

in seawater because sound speed depends on

the bulk modulus and seawater compressibility.

The following subsections explore how

seawater density depends on temperature,

salinity, and pressure, and discusses concepts

(such as potential and neutral density) that

reduce, as much as possible, the effects of

compressibility on a given analysis.

3.5.1. Effects of Temperature and

Salinity on Density

Density values evaluated at the ocean’s

surface pressure are shown in Figure 3.1 (curved

contours) for the whole range of salinities and

temperatures found anywhere in the oceans.

The shaded bar in the figure shows that most

of the ocean lies within a relatively narrow

salinity range. More extreme values occur only

at or near the sea surface, with fresher waters

outside this range (mainly in areas of runoff or

ice melt) and the most saline waters in relatively

confined areas of high evaporation (such as

marginal seas). The ocean’s temperature range

produces more of the ocean’s density variation

than does its salinity range. In other words,

temperature dominates oceanic density variations

for the most part. (As noted previously,

an important exception is where surface waters

are relatively fresh due to large precipitation or

ice melt; that is, at high latitudes and also in the

tropics beneath the rainy Intertropical Convergence

Zone of the atmosphere.) The curvature

of the density contours in Figure 3.1 is due to

the nonlinearity of the equation of state. The

curvature means that the density change for

a given temperature or salinity change is

different at different temperatures or salinities.

To emphasize this point, Table 3.2 shows the

change of density (Ds t ) for a temperature

change (DT) of þ1 K (left columns) and the

value of Ds t for a salinity change (DS) of þ0.5

(right columns). These are arbitrary choices for

changes in temperature and salinity. The most

important thing to notice in the table is how

density varies at different temperatures and

salinities for given changes in each. At high

temperatures, s t varies significantly with T at

all salinities. As temperature decreases, the

rate of variation with T decreases, particularly

at low salinities (as found at high latitudes or

in estuaries). The change of s t with DS is about

the same at all temperatures and salinities, but

is slightly greater at low temperature.

3.5.2. Effect of Pressure on Density:

Potential Density

Seawater is compressible, although not

nearly as compressible as a gas. As a water

parcel is compressed, the molecules are pushed

closer together and the density increases. At

the same time, and for a completely different

physical reason, adiabatic compression causes

the temperature to increase, which slightly

offsets the density increase due to compression.

(See discussion of potential temperature in

Section 3.3.)

40

TABLE 3.2


Variation of Density (Ds t ) with Variations of Temperature (DT) and of Salinity (DS) as Functions of

Temperature and Salinity

Salinity 0 20 35 40 0 20 35 40

Temperature ( C) Ds t for DT [ D1 C Ds t for DS [ D0.5

30 0.31 0.33 0.34 0.35 0.38 0.37 0.37 0.38

20 0.21 0.24 0.27 0.27 0.38 0.38 0.38 0.38

10 0.09 0.14 0.18 0.18 0.39 0.39 0.39 0.39

0 þ0.06 0.01 0.06 0.07 0.41 0.40 0.40 0.40

Density is primarily a function of pressure

(Figure 3.4) because of this compressibility. Pressure

effects on density have little to do with the

initial temperature and salinity of the water

parcel. To trace a water parcel from one place

to another, the dependence of density on pressure

should be removed. An early attempt was

to use s t , defined earlier, in which the pressure

effect was removed from density but not from

temperature. It is now standard practice to use

potential density, in which density is calculated

using potential temperature instead of temperature.

(The measured salinity is used.) Potential

Pressure (dbar)

0

1000

2000

3000

4000

5000

1030 1035 1040 1045 1050

Density (kg m –3 )

FIGURE 3.4 Increase in density with pressure for

a water parcel of temperature 0 C and salinity 35.0 at the

sea surface.

density is the density that a parcel would have

if it were moved adiabatically to a chosen reference

pressure. If the reference pressure is the sea

surface, then we first compute the potential

temperature of the parcel relative to surface

pressure, then evaluate the density at pressure

0 dbar. 1 We refer to potential density referenced

to the sea surface (0 dbar) as s q , which signifies

that potential temperature and surface pressure

have been used.

The reference pressure for potential density

can be any pressure, not just the pressure at

the sea surface. For these potential densities,

potential temperature is calculated relative to

the chosen reference pressure and then the

potential density is calculated relative to the

same reference pressure. It is common to refer

to potential density referenced to 1000 dbar as

s 1 , referenced to 2000 dbar as s 2 , to 3000 dbar

as s 3 and so on, following Lynn and Reid (1968).

3.5.3. Specific Volume and Specific

Volume Anomaly

The specific volume (a) is the reciprocal of

density so it has units of m 3 /kg. For some

purposes it is more useful than density. The in

situ specific volume is written as a s,t,p . The

1 The actual pressure at the sea surface is the atmospheric pressure, but we do not include atmospheric pressure in many

applications since pressure ranges within the ocean are so much larger.


specific volume anomaly (d) is also sometimes

convenient. It is defined as:

d ¼ a s;t;p a 35;0;p (3.8)

The anomaly is calculated relative to a 35,0,p ,

which is the specific volume of seawater of

salinity 35 and temperature 0 C at pressure p.

With this standard d is usually positive. The

equation of state relates a (and d) to salinity,

temperature, and pressure. Originally all calculations

of geostrophic currents from the distribution

of mass were done by hand using

tabulations of the component terms of d,

described in previous editions of this book.

With modern computer methods, tabulations

are not necessary. The computer algorithms for

dynamic calculations (Section 7.5.1) still use

specific volume anomaly d, computed using

subroutines, rather than the actual density r,

to increase the calculation precision.

3.5.4. Effect of Temperature and

Salinity on Compressibility: Isentropic

Surfaces and Neutral Density

Cold water is more compressible than warm

water; it is easier to deform a cold parcel than

a warm parcel. When two water parcels with

the same density but different temperature and

salinity characteristics (one warm/salty, the

other cold/fresh) are submerged to the same

pressure, the colder parcel will be denser. If

there were no salt in seawater, so that density

depended only on temperature and pressure,

then potential density as defined earlier, using

any single pressure for a reference, would be

adequate for defining a unique isentropic surface.

An isentropic surface is one along which water

parcels can move adiabatically, that is, without

external input of heat or salt.

When analyzing properties within the ocean

to determine where water parcels originate, it

is assumed that motion and mixing is mostly

along a quasi-isentropic surface and that mixing

across such a surface (quasi-vertical mixing)

is much less important (Montgomery, 1938).

However, because seawater density depends

on both salinity and temperature, the actual

surface that a water parcel moves along in the

absence of external sources of heat or freshwater

depends on how the parcel mixes along that

surface since its temperature and salinity will

be altered as it mixes with adjacent water

parcels on that surface. This quasi-lateral mixing

alters the temperature (and salinity) and

therefore, the compressibility of the mixture.

As a result, when it moves laterally, the parcel

will equilibrate at a different pressure than if

there had been no mixing. This means that there

are no closed, unique isentropic surfaces in the

ocean, since if our water parcel were to return

to its original latitude and longitude, it will

have moved to a different density and hence

pressure because its temperature and salinity

will have changed due to mixing along that

surface. Note that these effects are important

even without diapycnal mixing between water

parcels on different isentropic surfaces (quasivertical

mixing), which also can change temperature,

salinity, and compressibility.

The density differences associated with these

differences in compressibility can be substantial

(Figure 3.5). For instance, water spilling out of

the Mediterranean Sea through the Strait of

Gibraltar is saline and rather warm compared

with water spilling into the Atlantic from the

Nordic Seas over the Greenland-Iceland ridge

(Chapter 9). The Mediterranean Water (MW)

density is actually higher than the Nordic Sea

Overflow Water (NSOW) density where they

flow over their respective sills, which are at about

the same depth. However, the warm, saline MW

(13.4 C, 37.8 psu) is not as compressible as the

much colder NSOW (about 1 C, 34.9 psu; Price &

Baringer, 1994). The potential density relative to

4000 dbar of MW is lower than that of the more

compressible NSOW. The NSOW reaches the

bottom of the North Atlantic, while the MW

does not. (As both types of water plunge

42


(a)

Potential temperature

(b)


30

20

10

0

30

20

10

0

Salinity

32 34 36 38 40

38

22

(1)

(1)

40

24

(2)

(2)

32 34 36 38 40

Salinity

downward, they entrain or mix with the waters

that they pass through. This also has an effect

on how deep they fall, so the difference in

compressibility is not the only cause for different

outcomes.)

Restating this more generally, changing the

reference pressure for potential density alters

the density difference between two water parcels

(Figure 3.5). For the pair labeled 1, the densities

are the same at the sea surface (upper panel).

Because the cold parcel compresses more than

the warm one with increasing pressure, the

26

42

44

28

30

0 dbar

46

48

4000 dbar

FIGURE 3.5 Potential density relative to (a) 0 dbar and

(b) 4000 dbar as a function of potential temperature (relative

to 0 dbar) and salinity. Parcels labeled 1 have the same

density at the sea surface. The parcels labeled 2 represents

Mediterranean (saltier) and Nordic Seas (fresher) source

waters at their sills.

cold parcel is denser than the warm one at higher

pressure (lower panel). The pair labeled 2 illustrates

the MW (warm, salty) and NSOW (cold,

fresh) pair with their properties at the sills where

they enter the North Atlantic. At the sea surface,

which neither parcel ever reaches, the Mediterranean

parcel would actually be denser than the

Nordic Seas parcel. Near the ocean bottom, represented

by 4000 dbar (Figure 3.5b), the colder

Nordic Seas parcel is markedly denser than the

Mediterranean parcel. Therefore, if both parcels

dropped to the ocean bottom from their respective

sills, without any mixing, the Nordic Seas parcel

would lie under the Mediterranean parcel. (In

actuality, as already mentioned, there is a large

amount of entrainment mixing as these parcels

drop down into the North Atlantic.)

The surfaces that we use to map and trace

water parcels should approximate isentropic

surfaces. Early choices, that were an improvement

over constant depth surfaces, included

sigma-t surfaces (Montgomery, 1938) and even

potential temperature surfaces (Worthington

and Wright, 1970). A method, introduced by

Lynn and Reid (1968), that produces surfaces

that are closer to isentropic uses isopycnals

with a reference pressure for the potential

density that is within 500 m of the pressure of

interest. Therefore when working in the top

500 m, a surface reference pressure is used.

When working at 500 to 1500 m, a reference

pressure of 1000 dbar is used, and so forth.

Experience has shown this pressure discretization

is sufficient to remove most of the problems

associated with the effect of pressure on density.

When isopycnals mapped in this fashion move

into a different pressure range, they must be

patched onto densities at the reference pressure

in the new range. Reid (1989, 1994, 1997, 2003)

followed this practice in his monographs on

Pacific, Atlantic, and Indian Ocean circulations.

It is less complicated to use a continuously

varying surface rather than one patched from

different reference pressures, although in practice

there is little difference between them.


“Neutral surfaces,” introduced by Ivers (1975),

a student working with J.L. Reid, use a nearly

continuously varying reference pressure. If

a parcel is followed along its path from one

observation station to the next, assuming the

path is known, then it is possible to track its pressure

and adjust its reference pressure and density

at each station. McDougall (1987a) refined this

neutral surface concept and introduced it widely.

Jackett and McDougall (1997) created a computer

program for computing their version of this

neutral density, based on a standard climatology

(average temperature and salinity on a grid for

the whole globe, derived from all available observations;

Section 6.6.2), marching away from

a single location in the middle of the Pacific.

The Jackett and McDougall neutral density is

denoted g N with numerical values that are

similar to those of potential density (with units

of kg/m 3 ). Neutral density depends on latitude,

longitude, and pressure, and is defined only for

ranges of temperature and salinity that occur in

the open ocean. This differs from potential

density, which is defined for all values of temperature

and salinity through a well-defined equation

of state that has been determined in the

laboratory and is independent of location.

Neutral density cannot be contoured as a function

of potential temperature and salinity analogously

to Figure 3.5 for density or potential

density.

The advantage of neutral density for

mapping quasi-isentropic surfaces is that it

removes the need to continuously vary the reference

pressure along surfaces that have depth

variation (since this is already done in an

approximate manner within the provided software

and database). Neutral density is a convenient

tool. Both potential and neutral density

surfaces are approximations to isentropic

surfaces. Ideas and literature on how to best

approximate isentropic surfaces continue to be

developed; neutral density is currently the

most popular and commonly used approximation

for mapping isentropes over large distances

that include vertical excursions of more than

several hundred meters.

3.5.5. Linearity and Nonlinearity

in the Equation of State

As described earlier, the equation of state

(3.6) is somewhat nonlinear in temperature,

salinity, and pressure; that is, it includes products

of salinity, temperature, and pressure. For

practical purposes, in theoretical and simple

numerical models, the equation of state is sometimes

approximated as linear and its pressure

dependence is ignored:

rzr 0 þ aðT T 0 ÞþbðS S 0 Þ;

a ¼ vr=vT andb

¼ vr=vS

(3.9)

where r 0 , T 0, , and S 0 are arbitrary constant

values of r, T, and S; they are usually chosen

as the mean values for the region being

modeled. Here a is the thermal expansion coefficient,

which expresses the change in density

for a given change in temperature (and should

not be confused with specific volume, defined

with the same symbol in Section 3.5.3), and

b is the haline contraction coefficient, which is

the change in density for a given change in

salinity. The terms a and b are nonlinear functions

of salinity, temperature, and pressure;

their mean values are chosen for linear models.

Full tables of values are given in UNESCO

(1987). The value of ar (at the sea surface and at

a salinity of 35 psu) ranges from 53 10 6 K 1 at

a temperature of 0 C to 257 10 6 K 1 at a

temperature of 20 C. The value of br (at the

sea surface and at a salinity of 35 psu) ranges

from 785 10 6 psu 1 (at a temperature of 0 C)

to 744 10 6 psu 1 (at a temperature of 20 C).

Nonlinearity in the equation of state leads to

the curvature of the density contours in Figures

3.1 and 3.5. Mixing between two water parcels

must occur along straight lines in the temperature/salinity

planes of Figures 3.1 and 3.5.

Becauseoftheconcavecurvatureofthedensity

44


contours, when two parcels of the same density

but different temperature and salinity are

mixed together, the mixture has higher density

than the original water parcels. Thus the

concavity of the density contours means that

there is a contraction in volume as water

parcels mix. This effect is called cabbeling

(Witte, 1902). In practice, cabbeling may be of

limited importance, having demonstrable

importance only where water parcels of very

different initial properties mix together. Examples

of problems where cabbeling has been

a factor are in the formation of dense water in

the Antarctic (Foster, 1972) and in the modification

of intermediate water in the North Pacific

(Talley & Yun, 2001).

There are two other important mixing effects

associated with the physical properties of

seawater: thermobaricity and double diffusion.

Thermobaricity (McDougall, 1987b) is best

explained by the rotation with depth of

potential density contours in the potential

temperatureesalinity plane (Section 3.5.4). As

in Figure 3.5, consider two water parcels of

different potential temperature and salinity in

which the warmer, saltier parcel is slightly

denser than the colder, fresher one. (This is

a common occurrence in subpolar regions

such as the Arctic and the Antarctic.) If these

two water parcels are suddenly brought to

a greater pressure, it is possible for them to

reverse their relative stratification, with the

colder, fresher one compressing more than

the warmer one, and therefore becoming the

denser of the two parcels. The parcels would

now be vertically stable if the colder, fresher

one were beneath the warmer, saltier one. Thermobaricity

is an important effect in the Arctic,

defining the relative vertical juxtaposition of

the Canadian and Eurasian Basin Deep Waters

(Section 12.2).

Double diffusion results from a difference in

diffusivities for heat and salt, therefore, it is

not a matter of linearity or nonlinearity. At the

molecular level, these diffusivities clearly differ.

Because double diffusive effects are apparent in

the ocean’s temperatureesalinity properties, the

difference in diffusivities scales up in some way

to the eddy diffusivity. Diffusivity and mixing

are discussed in Chapter 7, and double diffusion

in Section 7.4.3.2.

3.5.6. Static Stability and Brunt-Väisälä

Frequency

Static stability, denoted by E, is a formal

measure of the tendency of a water column to

overturn. It is related to the density stratification,

with higher stability where the water

column is more stratified. A water column is

statically stable if a parcel of water that is

moved adiabatically (with no heat or salt

exchange) up or down a short distance returns

to its original position. The vigor with which

the parcel returns to its original position

depends on the density difference between

theparcelandthesurroundingwatercolumn

at the displaced position. Therefore the rate of

change of density of the water column with

depth determines a water column’s static

stability. The actual density of the parcel

increases or decreases as it is moved down or

up because the pressure on it increases or

decreases, respectively. This adiabatic change

in density must be accounted for in the definition

of static stability.

The mathematical derivation of the static

stability of a water column is presented in detail

in Pond and Pickard (1983) and other texts. The

full expression for E is complicated. For very

small vertical displacements, static stability

might be approximated as

Ez ðl=rÞ ðvr=vzÞ (3.10a)

where r is in situ density. The water column is

stable, neutral, or unstable depending on

whether E is positive, zero, or negative, respectively.

Thus, if the density gradient is positive

downwards, the water column is stable and

there is no tendency for vertical overturn.


For larger vertical displacements, a much

better approximation uses local potential

density, s n :

E ¼ ðl=rÞðvs n =vzÞ (3.10b)

Here the potential density anomaly s n is

computed relative to the pressure at the center

of the interval used to compute the vertical

gradient. This local pressure reference approximately

removes the adiabatic pressure effect.

Many computer subroutines for seawater properties

use this standard definition. An equivalent

expression for stability is

E ¼ ðl=rÞðvr=vzÞ ðg=C 2 Þ (3.10c)

where r is in situ density, g ¼ acceleration due

to gravity, and C ¼ in situ sound speed. The

addition of the term g/C 2 allows for the

compressibility of seawater. (Sound waves are

compression waves; Section 3.7.)

A typical density profile from top to bottom

of the ocean has a surface mixed layer with

low stratification, an upper ocean layer with

an intermediate amount of stratification, an

intermediate layer of high stratification (pycnocline),

and a deep layer of low stratification

(Section 4.2). The water in the pycnocline is

very stable; it takes much more energy to

displace a particle of water up or down

than in a region of lesser stability. Therefore

turbulence, which causes most of the mixing

between different water bodies, is less able to

penetrate through the stable pycnocline than

through less stable layers. Consequently, the

pycnocline is a barrier to the vertical transport

of water and water properties. The stability of

these layers is measured by E. In the upper

1000 m in the open ocean, values of E range

from 1000 10 8 m 1 to 100 10 8 m 1 ,with

larger values in the pycnocline. Below 1000 m,

E decreases; in abyssal trenches E may be as

low as 1 10 8 m 1 .

Static instabilities may be found near the interfaces

between different waters in the process of

mixing. Because these instabilities occur at a small

vertical scale, on the order of meters, they require

continuous profilers for detection. Unstable

conditions with vertical extents greater than tens

of meters are uncommon below the surface layer.

The buoyancy (Brunt-Väisälä) frequency associated

with internal gravity waves (Chapter 8) is

an intrinsic frequency associated with static

stability. If a water parcel is displaced upward

in a statically stable water column, it will sink

and overshoot the original position. The denser

water beneath its original position will force it

back up into lighter water, and it will continue

oscillating. The frequency of the oscillation

depends on the static stability: the more stratified

the water column, the higher the static

stability and the higher the buoyancy frequency.

The Brunt-Väisälä frequency, N, is an intrinsic

frequency of internal waves:

N 2 ¼ gEzg½ ðl=rÞðvs n =vzÞŠ (3.11)

The frequency in cycles/sec (hertz) is

f ¼ N/2p and the period is s ¼ 2p/N. In the

upper ocean, where E typically ranges from

1000 10 8 to 100 10 8 m 1 , periods are

s ¼ 10 to 33 min (Figure 3.6). For the deep ocean,

E ¼ 1 10 8 m 1 and s z 6h.

The final quantity that we define based on

vertical density stratification is the “stretching”

part of the potential vorticity (Section 7.6). Potential

vorticity is a dynamical property of a fluid

analogous to angular momentum. Potential

vorticity has three parts: rotation due to Earth’s

rotation (planetary vorticity), rotation due to

relative motions in the fluid (relative vorticity,

for instance, in an eddy), and a stretching

component proportional to the vertical change

in density, which is analogous to layer thickness

(Eq. 7.41). In regions where currents are weak,

relative vorticity is small and the potential

vorticity can be approximated as

Qz ðf=rÞðvr=vzÞ (3.12a)

This is sometimes called “isopycnic potential

vorticity.” The vertical density derivative is

46


0

Period (minutes)

30 15

10

0

FIGURE 3.6 (a) Potential

density and (b) Brunt-Väisälä

frequency (cycles/h) and

period (minutes) for a profile

in the western North Pacific.

1000

1000

Pressure (dbar)

2000

3000

2000

3000

4000

4000

5000

North Pacific

24.258°N, 147.697°W

5000

24 26 28

Potential density

0 2 4 6

Brunt−Väisälä Frequency

(cycles per hour)

calculated from locally referenced potential

density, so it can be expressed in terms of

Brunt-Väisälä frequency:

Q ¼ðf=gÞN 2

3.5.7. Freezing Point of Seawater

(3.12b)

The salt in seawater depresses the freezing

point below 0 C(Figure 3.1). An algorithm for

calculating the freezing point of seawater is

given by Millero (1978). Depression of the

freezing point is why a mixture of salt water

and ice is used to make ice cream; as the ice

melts, it cools the water (and ice cream) below

0 C. At low salinities, below the salinity of

most seawater, cooling water reaches its

maximum density before freezing and sinks

while still fluid. The water column then overturns

and mixes until the whole water column

reaches the temperature of maximum density.

On further cooling the surface water becomes

lighter and the overturning stops. The water

column freezes from the surface down, with

the deeper water remaining unfrozen.

However, at salinities greater than 24.7 psu,

maximum density is achieved at the freezing

point. Therefore more of the water column

must be cooled before freezing can begin, so

freezing is delayed compared with the freshwater

case.

3.6. TRACERS

Dissolved matter in seawater can help in

tracing specific water masses and pathways of

flow. Some of these properties can be used for

dating seawater (determine the length of time

since the water was last at the sea surface;

Section 4.7). Most of these constituents occur in

such small concentrations that their variations

TRACERS 47

do not significantly affect density variations or

the relationship between chlorinity, salinity,

and conductivity. (See Section 3.5 for comments

on this.) These additional properties of seawater

can be: conservative or non-conservative;

natural or anthropogenic (man-made); stable

or radioactive; transient or non-transient. The

text by Broecker and Peng (1982) describes the

sources and chemistry of many tracers in detail.

For a tracer to be conservative there are no

significant processes other than mixing by

which the tracer is changed below the surface.

Even salinity, potential temperature, and hence

density, can be used as conservative tracers

since they have extremely weak sources within

the ocean. This near absence of in situ sources

and sinks means that the spreading of water

masses in the ocean can be approximately traced

from their origin at the sea surface by their characteristic

temperature/salinity values. Near the

surface, evaporation, precipitation, runoff, and

ice processes change salinity, and many surface

heat-transfer processes change the temperature

(Section 5.4). Absolute salinity can be changed

only very slightly within the ocean due to

changes in dissolved nutrients and carbon

(end of Section 3.4). Temperature can be raised

very slightly by geothermal heating at the ocean

bottom. Even though water coming out of

bottom vents at some mid-ocean ridges can be

extremely hot (up to 400 C), the total amount

of water streaming out of the vents is tiny, and

the high temperature quickly mixes away,

leaving a miniscule large-scale temperature

increase.

Non-conservative properties are changed by

chemical reactions or biological processes

within the water column. Dissolved oxygen is

an example. Oxygen enters the ocean from the

atmosphere at the sea surface. It is also

produced through photosynthesis by phytoplankton

in the sunlit upper ocean (photic

zone or euphotic zone) and consumed by respiration

by zooplankton, bacteria, and other creatures.

Equilibration with the atmosphere keeps

ocean mixed layer waters at close to 100% saturation.

Below the surface layer, oxygen content

drops rapidly. This is not a function of the

temperature of the water, which generally is

lower at depth, since cold water can hold more

dissolved oxygen than warm water. (For

example, for a salinity of 35: at 30 C, 100%

oxygen saturation occurs at 190 mmol/kg; at

10 C it is 275 mmol/kg; and at 0 C it is 350

mmol/kg.) The drop in oxygen content and saturation

with depth is due to respiration within

the water column, mainly by bacteria feeding

on organic matter (mostly dead plankton and

fecal pellets) sinking from the photic zone. Since

there is no source of oxygen below the mixed

layer and photic zone, oxygen decreases with

increasing age of the subsurface water parcels.

Oxygen is also used by nitrifying bacteria,

which convert the nitrogen in ammonium

(NH 4 ) to nitrate (NO 3 ).

The rate at which oxygen is consumed is

called the oxygen utilization rate. This rate

depends on local biological productivity so it

is not uniform in space. Therefore the decrease

in oxygen from a saturated surface value is not

a perfect indication of age of the water parcel,

especially in the biologically active upper ocean

and continental shelves. However, below the

thermocline, the utilization rate is more uniform

and changes in oxygen following a water parcel

correspond relatively well to age.

Nutrients are another set of natural, nonconservative,

commonly observed properties.

These include dissolved silica, phosphate, and

the nitrogen compounds (ammonium, nitrite,

and nitrate). Nutrients are essential to ocean

life so they are consumed in the ocean’s surface

layer where life is abundant; consequently,

concentrations there are low. Nutrient content

increases with depth and age, as almost a mirror

image of the oxygen decrease. Silica is used by

some organisms to form protective shells. Silica

re-enters the water column when the hard parts

of these organisms dissolve as they fall to the

ocean floor. Some of this material reaches the

48


seafloor and accumulates, creating a silica source

on the ocean bottom as well. Some silica also

enters the water column through venting at

mid-ocean ridges. The other nutrients (nitrate,

nitrite, ammonium, and phosphate) re-enter

the water column as biological (bacterial)

activity decays the soft parts of the falling

detritus. Ammonium and phosphate are immediate

products of the decay. Nitrifying bacteria,

which are present through the water column,

then convert ammonium to nitrite and finally

nitrate; this process also, in addition to respiration,

consumes oxygen. Because oxygen is

consumed and nutrients are produced, the ratios

of nitrate to oxygen and of phosphate to oxygen

are nearly constant throughout the oceans. These

proportions are known as “Redfield ratios,” after

Redfield (1934) who demonstrated the nearconstancy

of these proportions. Nutrients are

discussed further in Section 4.6.

Other non-conservative properties related to

the ocean’s carbon system, including dissolved

inorganic carbon, dissolved organic carbon,

alkalinity, and pH, have been widely measured

over the past several decades. These have both

natural and anthropogenic sources and are

useful tracers of water masses.

Isotopes that occur in trace quantities are also

useful. Two have been widely measured:

14 C

and 3 He. 14 C is radioactive and non-conservative.

3 He is conservative. Both have predominantly

natural sources but both also have anthropogenic

sources in the upper ocean. Isotope concentrations

are usually measured and reported in terms

of ratios to the more abundant isotopes. For 14 C,

the reported unit is based on the ratio of 14 Cto

12 C. For 3 He, the reported unit is based on the

ratio of 3 He to 4 He. Moreover, the values are often

reported in terms of the normalized difference

between this ratio and a standard value, usually

taken to be the average atmospheric value (see

Broecker & Peng, 1982).

Most of the 14 C in the ocean is natural. It is

created continuously in the atmosphere by

cosmic ray bombardment of nitrogen, and enters

the ocean through gas exchange. “Bomb” radiocarbon

is an anthropogenic tracer that entered

the upper ocean as a result of atomic bomb tests

between 1945 and 1963 (Key, 2001). In the ocean,

14 C and 12 C are incorporated by phytoplankton

in nearly the same ratio as they appear in the

atmosphere. After the organic material dies

and leaves the photic zone, the 14 C decays radioactively,

with a half-life of 5730 years. The ratio of

14 Cto 12 C decreases. Since values are reported as

anomalies, as the difference from the atmospheric

ratio, the reported oceanic quantities

are generally negative (Section 4.7 and

Figure 4.24). The more negative the anomaly,

the older the water. Positive anomalies

throughout the upper ocean originated from

the anthropogenic bomb release of 14 C.

The natural, conservative isotope 3 He originates

in Earth’s mantle and is outgassed at vents

in the ocean floor. It is usually reported in terms

of its ratio to the much more abundant 4 He

compared with this ratio in the atmosphere. It

is an excellent tracer of mid-depth circulation,

since its sources tend to be the tops of mid-ocean

ridges, which occur at about 2000 m. The

anthropogenic component of 3 He is described

in the last paragraph of this Section.

Another conservative isotope that is often

measured in seawater is the stable (heavy)

isotope of oxygen, 18 O. Measurements are again

reported relative to the most common isotope

16 O. Rainwater is depleted in this heavy isotope

of oxygen (compared with seawater) because it

is easier for the lighter, more common isotope

of oxygen, 16 O, to evaporate from the sea and

land. A second step of reduction of 18 O in atmospheric

water vapor relative to seawater occurs

when rain first forms, mostly at warmer atmospheric

temperatures, since the heavier isotope

falls out preferentially. Thus rainwater is

depleted in 18 O relative to seawater, and rain

formed at lower temperatures is more depleted

than at higher temperatures. For physical oceanographers,

18 O content can be a useful indicator

in a high latitude region of whether the source of

SOUND IN THE SEA 49

freshwater at the sea surface is rain/runoff/

glacial melt (lower 18 O content), or melted sea

ice (higher content). In paleoclimate records, it

reflects the temperature of the precipitation

(higher 18 O in warmer rain); ice formed during

the (cold) glacial periods is more depleted in

18 O than ice formed in the warm interglacials

and hence 18 O content is an indicator of relative

global temperature.

Transient tracers are chemicals that have been

introduced by human activity; hence they are

anthropogenic. They are gradually invading

the ocean, marking the progress of water from

the surface to depth. They can be either stable

or radioactive. They can be either conservative

or non-conservative. Commonly measured

transient tracers include chlorofluorocarbons,

tritium, and much of the upper ocean 3 He and

14 C. Chlorofluorocarbons (CFCs) were introduced

as refrigerants and for industrial use.

They are extremely stable (conservative) in

seawater. Their usage peaked in 1994, when

recognition of their role in expanding the ozone

hole in the atmosphere finally led to international

conventions to phase out their use.

Because different types of CFCs were used

over the years, the ratio of different types in

a water parcel can yield approximate dates for

when the water was at the sea surface. Tritium

is a radioactive isotope of hydrogen that has

also been measured globally; it was released

into the atmosphere through atomic bomb

testing in the 1960s and then entered the ocean,

primarily in the Northern Hemisphere. Tritium

decays to 3 He with a half-life of 12.4 years, which

is comparable to the circulation time of the

upper ocean gyres. When 3 He is measured along

with tritium, the time since the water was at the

sea surface can be estimated (Jenkins, 1998).

3.7. SOUND IN THE SEA

In the atmosphere, we receive much of our

information about the material world by means

of wave energy, either electromagnetic (light) or

mechanical (sound). In the atmosphere, light in

the visible part of the spectrum is attenuated

less than sound; we can see much farther

away than we can hear. In the sea the reverse

is true. In clear ocean water, sunlight may be

detectable (with instruments) down to 1000 m,

but the range at which humans can see details

of objects is rarely more than 50 m, and usually

less. On the other hand, sound waves can be

detected over vast distances and are a much

better vehicle for undersea information than

light.

The ratio of the speed of sound in air to that

in water is small (about 1:4.5), so only a small

amount of sound energy starting in one medium

can penetrate into the other. This contrasts with

the relatively efficient passage of light energy

through the air/water interface (speed ratio

only about 1.33:1). This is why a person

standing on the shore can see into the water

but cannot hear any noises in the sea. Likewise,

divers cannot converse underwater because

their sounds are generated in the air in the

throat and little of the sound energy is transmitted

into the water. Sound sources used in

the sea generate sound energy in solid bodies

(transducers), for example, electromagnetically,

in which the speed of sound is similar to that

in water. Thus the two are acoustically

“matched” and the transducer energy is transmitted

efficiently into the sea.

Sound is a wave. All waves are characterized

by amplitude, frequency, and wavelength

(Section 8.2). Sound speed (C), frequency (n),

and wavelength (l) are connected by the wave

equation C ¼ nl. The speed does not depend

on frequency, so the wavelength depends on

sound speed and frequency. The frequencies of

sounds range from 1 Hz or less (1 Hz ¼ 1 vibration

per second) to thousands of kilohertz

(1 kHz ¼ 1000 cycles/sec). The wavelengths of

sounds in the sea cover a vast range, from about

1500 m for n ¼ 1 Hz to 7 cm for n ¼ 200 kHz.

Most underwater sound instruments use

50


a more restricted range from 10 to 100 kHz, for

which the wavelengths are 14 to 1.4 cm.

There are many sources of sound in the sea.

A hydrophone listening to the ambient sound

in the sea will record a wide range of frequencies

and types of sounds, from low rumbles to

high-frequency hisses. Some sources of undersea

sounds are microseisms (10 e 100 Hz); ships

(50 e 1500 Hz); the action of wind, waves, and

rain at the surface (1 e 20 kHz); cavitation of

air bubbles and animal noises (10 e 400 Hz);

and fish and crustaceans (1 e 10 kHz). Noises

associated with sea ice range from 1 e 10 kHz.

Sound is a compressional wave; water molecules

move closer together and farther apart as

the wave passes. Therefore sound speed

depends on the medium’s compressibility.

The more compressible a medium is for a given

density, the slower the wave since more

activity is required to move the molecules.

The speed of sound waves in the sea, C, is

given by

C ¼ ðbrÞ 1=2 where b ¼ r 1 ðvr=vpÞ q;S:

(3.13)

b is the adiabatic compressibility of seawater

(with potential temperature and salinity

constant), r is the density, p is the pressure, q

is the potential temperature, and S is the

salinity. Since b and r depend (nonlinearly)

on temperature and pressure, and to a lesser

extent, salinity, so does the speed of sound

waves. There are various formulae for the

dependence of Eq. (3.13) on T, S, and p; all

derived from experimental measurements.

The two most accepted are those of Del Grosso

(1974) and of Chen and Millero (1977); Del

Grosso’s equation is apparently more accurate,

based on results from acoustic tomography and

inverted echo sounder experiments (e.g.,

Meinen & Watts, 1997). Both are long and

nonlinear polynomials, as is the equation of

state. We present a simpler formula, which

itself is simplified from Mackenzie (1981) and

is similar to Del Grosso (1974), to illustrate

features of the relationship:

C ¼ 1448:96 þ 4:59T 0:053T 2

þ 1:34ðS

35Þþ0:016p

(3.14)

in which T, S, and p are temperature, salinity,

and depth, and the constants have the correct

units to yield C in m/s. The sound speed is

1449 m/s at T ¼ 0 C, S ¼ 35, and p ¼ 0. The

sound speed increases by 4.5 m/s for DT ¼þ

1 K, by 1.3 m/s for DS ¼þ1, and by 16 m/s

for Dp ¼ 1000 dbar.

Sound speed is higher where the medium is

less compressible. Seawater is less compressible

when it is warm, as noted in the previous potential

density discussion and apparent from the

simplified equation (3.14). Seawater is also less

compressible at high pressure, where the fluid

is effectively more rigid because the molecules

are pushed together. Salinity variations have

a negligible effect in most locations. In the

upper layers, where temperature is high, sound

speed is high, and decreases downward with

decreasing temperature (Figure 3.7). However,

pressure increases with depth, so that at middepth,

the decrease in sound speed due to

cooler water is overcome by an increase in

sound speed due to higher pressure. In most

areas of the ocean, the warm water at the surface

and the high pressure at the bottom produce

maximum sound speeds at the surface and

bottom and a minimum in between. The

sound-speed minimum is referred to as the

SOund Fixing And Ranging (SOFAR) channel.

In Figure 3.6, the sound-speed minimum is at

about 700 m depth. In regions where temperature

is low near the sea surface, for instance at

high latitudes, there is no surface maximum in

sound speed, and the sound channel is found

at the sea surface.

Sound propagation can be represented in

terms of rays that trace the path of the sound

(Figure 3.8). In the SOFAR channel, at about

1100 m in Figure 3.8, sound waves directed at

SOUND IN THE SEA 51

FIGURE 3.7 For station Papa in the Pacific Ocean at 39 N, 146 W, August, 1959: (a) temperature ( C) and salinity (psu)

profiles, (b) corrections to sound speed due to salinity, temperature, and pressure, (c) resultant in situ sound-speed profile

showing sound-speed minimum (SOFAR channel).

moderate angles above the horizontal are

refracted downward, across the depth of the

sound-speed minimum, and then refracted

upward; they continue to oscillate about the

sound-speed minimum depth. (Rays that travel

steeply up or down from the source will not be

channeled but may travel to the surface or

bottom and be reflected there.) Low frequency

sound waves (hundreds of hertz) can travel

considerable distances (thousands of kilometers)

along the SOFAR channel. This permits

detection of submarines at long ranges and has

been used for locating lifeboats at sea. Using

the SOFAR channel to track drifting subsurface

floats to determine deep currents is described

in Chapter S6, Section S6.5.2 of the supplemental

materials located on the textbook

Web site.

The deep SOFAR channel of Figure 3.8b is

characteristic of middle and low latitudes,

where the temperature decreases substantially

as depth increases. At high latitudes where the

temperatures near the surface may be constant

or even decrease toward the surface, the sound

speed can have a surface minimum (Figure 3.8a).

The much shallower sound channel, called

a surface duct, may even be in the surface layer.

In this case, downward directed sound rays

from a shallow source are refracted upward

while upward rays from the subsurface source

are reflected downward from the surface and

then refracted upward again. In this situation,

detection of deep submarines from a surface

ship using sonar equipment mounted in the

hull may not be possible and deep-towed sonar

equipment may be needed. In shallow water

(e.g., bottom depth <200 m), reflection can occur

both from the surface and from the bottom.

A pulse transmitted from a source near the

SOFAR channel axis does not appear to

52


FIGURE 3.8 Sound ray

diagrams: (a) from a shallow

source for a sound-speed

profile initially increasing

with depth in upper mixed

layer to a shallow minimum

and then decreasing, and (b)

from a sound source near

the speed minimum in the

sound channel for a typical

open ocean sound-speed

profile.

distant receivers as a sharp pulse but as

a drawn-out signal rising slowly to a peak followed

by a sharp cutoff. The peak before the

cutoff is the arrival of the sound energy along

the sound channel axis (direct signal), while

the earlier arrivals are from sound that traveled

along the refracted ray routes. It might

appear in Figure 3.8b that the refracted rays

have to travel a greater distance than the

direct ray and would thus be delayed, but

this is an illusion. Figure 3.8b is drawn with

gross vertical exaggeration to enable the rays

to be shown clearly, but the differences in

distances traveled by refracted rays and the

direct rays are very small; the greater speed

in the refracted ray paths compensates for

the greater distance they travel, so the direct

ray arrives last.

Sound is used widely to locate and observe

solid objects in the water. Echo sounders are

used to measure bottom depths to the ocean’s

maximum depth of more than 11,000 m. SONAR

(SOund Navigation And Ranging) can determine

the direction and distance to a submarine

at ranges of hundreds of meters and to schools

of fish at somewhat lesser ranges. Sidescan

sonars determine the structure of the ocean

bottom and can be used to locate shipwrecks.

Acoustically tracked Swallow floats (see

Chapter S6, Section S6.5.2 of the supplemental

SOUND IN THE SEA 53

material on the textbook Web site) provided

some of the first direct observations of deep

currents. Current speeds (or the speed of

a ship relative to the water) are often measured

using the reflection of sound waves from small

particles moving with the water, applying the

principle of Doppler shift. Because temperature

and density affect the sound velocity, sound can

be used to infer ocean water characteristics and

their variations. Sound is used to measure

surface processes such as precipitation, a measurement

that is otherwise nearly impossible to

determine.

In echo sounding, short pulses of sound

energy are directed vertically downward where

they reflect off of the bottom and return to the

ship. (Echo sounders are also used to detect

shoals of fish, whose air bladders are good reflectors

of sound energy. Modern “fish finders” are

simply low-cost echo sounders designed to

respond to the fish beneath the vessel.) The

acoustic travel time, t, yields the depth D ¼

C o t/2, where C o is the mean sound speed

between the surface and the bottom. Transducers

in ordinary echo sounders are not much

larger than the wavelength of the sound, so the

angular width of the sound beam is large. Wide

beams cannot distinguish the details of bottom

topography. For special sounding applications,

much larger sound sources that form a narrower

beam are used. It is also possible to improve the

resolution by using higher frequencies (up to 100

kHz or even 200 kHz), but the absorption of

sound energy by seawater increases roughly as

the square of the frequency, so higher frequency

echo sounders cannot penetrate as deeply.

Inhomogeneities distort an initial sharp

sound pulse so that the signal received at

a hydrophone is likely to have an irregular tail

of later arrival sounds. This is referred to as

reverberation. One source of reverberation is the

“deep scattering layer,” which is biological in

nature. This layer is characterized by diel

(day/night) vertical migrations of several

hundreds of meters; the organisms migrate

toward the sea surface at dusk to feed and

back down at dawn. This layer was first identified

because of the scattering produced by the

plankton and (gas-filled) fish bladders.

Sound is used to determine the speed of

ocean currents and of ships, using a technique

called acoustic Doppler profiling (see Chapter

S6, Section S6.5.5.1 of the supplemental material

located on the textbook Web site). Sound is

transmitted from a source and reflects off the

particles (mainly plankton) in the water and

returns back to a receiver. If the source is moving

relative to the particles, then the received sound

wave has a different frequency from the transmitted

wave, a phenomenon called Doppler

shifting. Doppler shift is familiar to anyone

who has listened to the sound of a siren when

an emergency vehicle first approaches (Doppler

shifting the sound to a higher frequency and,

therefore, a higher pitch) and then retreats

away (Doppler shifting the sound to a lower

frequency and lower pitch). Acoustic Doppler

speed logs are common on ships and give a relatively

accurate measure of the speed of the ship

through the water. If the ship’s speed is tracked

very precisely using, for instance, GPS navigation,

then the ship speed can be subtracted

from the speed of the ship relative to the water

to yield the speed of the water relative to the

GPS navigation, providing a measure of current

speeds. Acoustic Doppler current profilers are

also moored in the ocean to provide long-term

records of current speeds.

Sound can be used to map the ocean’s

temperature structure and its changes, through

a technique called acoustic tomography (see

Chapter S6, Section S6.6.1 of the supplemental

material located on the textbook Web site). Since

sound speed depends on temperature, temperature

changes along a ray path result in travel

time changes. With extremely accurate clocks,

these changes can be detected. If multiple ray

paths crisscross a region, the travel time changes

can be combined using sophisticated data analysis

techniques to map temperature changes in

54


the region. This technique has been especially

useful in studying the three-dimensional structure

of deep convection in regions and seasons

that are virtually impossible to study from

research ships. Similar techniques have been

applied to very long distance monitoring of

basin-average ocean temperature, which is

possible because of the lack of attenuation of

sound waves over extremely long distances

(Munk & Wunsch, 1982). However, large-scale

monitoring of ocean temperature changes

using sound has been eclipsed by the global

temperatureesalinity profiling float array, Argo,

which provides local as well as basin-average

information.

Much more information about ocean acoustics

can be found in textbooks such as Urick (1983).

3.8. LIGHT AND THE SEA

This is a very brief introduction to a complex

subject. Full treatments are available in various

sources; some suggestions are Mobley (1995)

and Robinson (2004).

Sunlight with a range of wavelengths enters

the sea after passing through the atmosphere.

Within the upper layer of the ocean, up to 100

m depth or more, the visible light interacts

with the water molecules and the substances

that are dissolved or suspended in the water.

The light provides energy for photosynthesis

and also heats the upper layer. Processes of backscattering,

absorption, and re-emission result in

the visible light (ocean color) that emerges back

from the ocean surface into the atmosphere.

This emerging radiation is then measured with

instruments above the sea surface, including

satellites. For satellite observations, the atmosphere

again affects the signal from the sea.

Observations of ocean color by satellites can

then be related to the processes within the ocean

that affect the emerging light, including an abundance

of phytoplankton, particulate organic

carbon, suspended sediment, and so forth.

Absorption (attenuation) of the sun’s energy

in the upper layer depends on the materials

within the water; therefore these materials affect

how heating is distributed in the surface layer

and affects mixed layer processes. General

circulation models that are run with observed

forcing sometimes use information about light

attenuation, affecting mixed layer formation

and, consequently, sea surface temperature in

the model.

Section 3.8.1 describes the optical properties

of seawater and Section 3.8.2 describes the

quantity that is observed as ocean color. Examples

of observations are shown in Chapter 4.

3.8.1. Optical Properties

The sun irradiates the earth with a peak in the

visible spectrum (wavelengths from about 400

to 700 nm, from violet to red, where 1 nm ¼

10 9 m). Sunlight behaves differently in water

and air. The ocean absorbs light in much shorter

distances than the atmosphere. When this shortwave

energy penetrates the sea, some of it is

scattered, but much is absorbed, almost all

within the top 100 m. The energy is attenuated

approximately exponentially. This is the photic

(euphotic) zone, where photosynthesis occurs.

This penetration of solar energy into the ocean’s

upper layer is also important in the ocean’s heat

budget (Chapter 5).

A schematic overview of the ocean’s optical

processes is shown in Figure 3.9, after Mobley

(1995), who provides much greater detail and

precise expressions for each of the quantities

in the diagram. Each of the quantities can be

observed, with greater or lesser difficulty. At

the top of the diagram, the external environmental

quantities that determine the amount

of radiation entering the ocean are the sun’s

radiance distribution, which depends on its

position and on sky conditions; the sea state,

since this determines how much radiation is

reflected without entering the sea; and the ocean

bottom, if it is shallow enough to intercept the

LIGHT AND THE SEA 55

Inherent optical

properties of seawater

Absorption

Scattering

Environment

Incident radiance (sun

position and sky conditions)

Sea State

Bottom condition

Radiometric quantities

Downwelling and upwelling

irradiance

Photosynthetically available

radiation (PAR)

Reflectance

(ocean color)

Apparent optical

properties of the ocean

Reflectance

Downwelling and upwelling

irradiance attenuation

PAR attenuation

FIGURE 3.9 Schematic

of optical processes in

seawater. Adapted and

simplified from Mobley

(1995), with added indicators

of seawater heating

and photosynthesis, as

well as satellite observation

of ocean color.

Upper layer heating

Photosynthesis

light. The inherent optical properties of the

seawater determine how it absorbs and scatters

radiation, as a function of wavelength; this

depends on the matter that is dissolved, suspended,

or active (in the case of phytoplankton).

The environmental conditions and inherent

optical properties work together through a radiative

transfer equation to set the radiometric

quantities of the medium. Here it is useful to

provide some definitions of the radiometric

quantities listed in the middle box of Figure 3.9.

First, we note that light from a source, which

could be at any point in a medium in which light

is diffused or scattered, illuminates a complete

sphere around the source. Therefore the solid

angle, measured in “steradians” (sr), is a useful

measure, similar to area. Next, the flux of energy

from the light is measured in Watts (J/sec). The

radiance is the flux of energy per unit area and

per unit steradian; it is measured in units of

W/(sr m 2 ). If the radiance is measured as a function

of wavelength of the light (i.e., spectral

radiance), then its units are W/(sr m 2 nm) if

wavelength is measured in nanometers.

The irradiance is the total amount of radiance

that reaches a given point (i.e., where your

optical measurement is made), so it is the sum

of radiance coming in from all directions to the

observation point; therefore it is the integral of

radiance over all solid angles, and has units of

W/m 2 for total irradiance, or W/(m 2 nm) for

spectral irradiance (which is a function of wavelength).

Next, upwelling irradiance is defined as

the irradiance from all solid angles below the

observation point; downwelling irradiance

would come from all angles above that point.

Reflectance is the ratio of upwelling irradiance

to downwelling irradiance, defined at a point;

reflectance defined this way has no units. It is

not the same as actual reflected light from the

sea surface. Rather, reflectance is the light

emerging from the ocean. For remote sensing,

in which the radiation from the ocean’s surface

is being measured from a specified location,

rather than from all directions, reflectance can

be defined alternatively as the ratio of upwelling

radiance to downwelling irradiance; in this case,

reflectance has units of (sr 1 ).

Finally, the amount of radiation available for

photosynthesis (photosynthetically available

radiation; PAR) is measured in photons s 1 m 2 .

Returning to Figure 3.9, the rightmost bottom

box lists the apparent optical properties of the

seawater. These include the rate at which light

is attenuated within the water column, and

how much light returns back out through the

56


sea surface (indicated as reflectance). The irradiance

and PAR are attenuated with increasing

depth as the radiation is absorbed, scattered,

and used for photosynthesis by phytoplankton.

Attenuation is often approximately exponential.

If attenuation were exactly exponential, of the

form I(z) ¼ I o e Kz , where I o is the radiation

intensity at the sea surface, I the intensity at

a depth z meters below the surface, and K the

vertical attenuation coefficient of the water,

then the apparent optical properties would be

expressed in terms of the e-folding depth, K.

The actual attenuation is not exponential, so

the attenuation coefficient, K, is proportional

to the depth derivative of the radiation intensity

(and would be equal to the e-folding depth if the

dependence were exponential).

The effects of depth and constant attenuation

coefficient on light intensity are illustrated in

Table 3.3, from Jerlov (1976). The coefficient K

depends mainly on factors affecting absorption

of light in the water and to a lesser extent on

scattering. The last two columns of Table 3.3

indicate the range of penetrations found in

actual seawater.

The smallest attenuation coefficient in Table

3.3 (K ¼ 0.02 m 1 ) represents the clearest ocean

water and deepest penetration of light energy.

Energy penetrates coastal waters less readily

because of the extra attenuation due to suspended

particulate matter and dissolved materials.

The largest attenuation coefficient listed

in the table, K ¼ 2m 1 , represents very turbid

water with many suspended particles.

In seawater, the attenuation coefficient K

also varies considerably with wavelength.

Figure 3.10b shows the relative amounts of

energy penetrating clear ocean water to 1, 10,

and 50 m as a function of wavelength (solid

curves). Light with blue wavelengths penetrates

deepest; penetration by yellow and red is much

less. That is, blue light, with wavelength of

about 450 nm, has the least attenuation in clear

ocean water. At shorter and longer wavelengths

(in the ultraviolet and red), the attenuation is

much greater. The increased attenuation in the

ultraviolet is not important to the ocean’s heat

budget, because the amount of energy reaching

sea level at such short wavelengths is small.

Much more solar energy is contained in and

beyond the red end of the spectrum. Virtually

all of the energy at wavelengths shorter than

the visible is absorbed in the top meter of water,

while the energy at long wavelengths (1500 nm

or greater) is absorbed in the top few

centimeters.

All wavelengths are attenuated more in

turbid water than in clear water. In clear ocean

TABLE 3.3

Amount of Light Penetrating to Specified Depths in Seawater as a Percentage of that Entering Through

the Surface

Vertical Attenuation Coefficient K (m L1 ) Clearest Ocean Water Turbid Coastal Water

Depth (m) K ¼ 0.02 K ¼ 0.2 K ¼ 2

0 100% 100% 100% 100% 100%

1 98 82 14 45 18

2 96 67 2 39 8

10 82 14 0 22 0

50 37 0 0 5 0

100 14 0 0 0.5 0

Jerlov, 1976.


FIGURE 3.10 (a) Attenuation

coefficient k l , as a function of

wavelength l (mm) for clearest

ocean water (solid line) and turbid

coastal water (dashed line). (b)

Relative energy reaching 1, 10, and

50 m depth for clearest ocean water

and reaching 1 and 10 m for turbid

coastal waters.

water, there is enough light at 50 to 100 m to

permit a diver to work, but in turbid coastal

waters almost all of the energy may have been

absorbed by a depth of 10 m. K is larger for

turbid water than clear (Figure 3.10a), and the

least attenuation is in the yellow part of the

spectrum. In turbid water, less energy penetrates

to 1 m and 10 m, and the maximum penetration

is shifted to the yellow (Figure 3.10b).

(The energy reaching 50 m in this turbid

water is too small to show on the scale of this

graph.)

In clear ocean water, the superior penetration

of blue and green light is evident both

visually when diving and also in color photographs

taken underwater by natural light.

Red or yellow objects appear darker in color

or even black as they are viewed at increasing

depths because the light at the red end of the

spectrum has been absorbed in the upper

layers and little is left to be reflected by the

objects. Blue or green objects retain their colors

to greater depths.

The presence of plankton in seawater also

changes the penetration depth of solar radiation,

and hence the depth at which the sun’s

heat is absorbed. This changes the way the

surface mixed layer develops, which can in

turn impact the plankton, leading to a feedback.

There are significant and permanent geographical

variations in this vertical distribution of

absorption, since some regions of the world

58


ocean have much higher biological productivity

than others.

3.8.2. Ocean Color

To the eye, the color of the sea ranges from

deep blue to green to greenish yellow (Jerlov,

1976). Broadly speaking, deep or indigo blue

color is characteristic of tropical and equatorial

seas, particularly where there is little biological

production. At higher latitudes, the color

changes through green-blue to green in polar

regions. Coastal waters are generally greenish.

Two factors contribute to the blue color of

open ocean waters at low latitudes. Because

water molecules scatter the short-wave (blue)

light much more than the long-wave (red) light,

the color seen is selectively blue. In addition,

because the red and yellow components of

sunlight are rapidly absorbed in the upper few

meters, the only light remaining to be scattered

by the bulk of the water is blue. Looking at the

sea from above, sky light reflected from the

surface is added to the blue light scattered

from the body of the water. If the sky is blue,

the sea will still appear deep blue, but if there

are clouds, the white light reflected from the

sea surface dilutes the blue scattered light

from the water and the sea appears less

intensely blue.

If there are phytoplankton in the water, their

chlorophyll absorbs blue light and also red light,

which shifts the water color to green. (This is

also why plants are green.) The organic products

from plants may also add yellow dyes to

the water; these will absorb blue and shift the

apparent color toward the green. These shifts

in color generally occur in the more productive

high-latitude and coastal waters. In some

coastal regions, rivers carry dissolved organic

substances that emphasize the yellowish green

color. The red color that occurs sporadically in

some coastal areas, the so-called red tide, is

caused by blooms of reddish brown phytoplankton.

Mud, silt, and other finely divided

inorganic materials carried into the ocean by

rivers can impart their own color to the water.

In some fjords, the low-salinity surface layer

may be milky white from the finely divided

“rock flour” produced by abrasion in the

glaciers and carried down by the melt water.

The sediment can be kept in suspension by

turbulence in the upper layer for a time, but

when it sinks into the saline water, it flocculates

(forms lumps) and sinks more rapidly. When

diving in such a region the diver may be able

to see only a fraction of a meter in the upper

layer but be able to see several meters in the

saline water below.

The color of seawater and depth of penetration

of light were traditionally judged using

a white Secchi disk (see Chapter S6, Section

S6.8 of the supplemental materials located on

the textbook Web site) lowered from the ship.

This method has been superseded by a suite of

instruments that measure light penetration at

different wavelengths, transparency of the

water at various wavelengths, and fluorescence.

Most important, color observations are now

made continuously and globally by color

sensors on satellites.

Ocean color is a well-defined quantity, related

to reflectance (Figure 3.9 and definition in

Section 3.8.1). Reflectance, or ocean color, can

be measured directly above the ocean’s surface.

Observations of ocean color since the 1980s

have been made from satellites, and must be

corrected for changes as the light passes

upward through the atmosphere. Ocean color

observations are then converted, through

complex algorithms, to physically useful quantities

such as the amount of chlorophyll

present, or the amount of particulate organic

carbon, or the amount of “yellow substance”

(gelbstoff) that is created by decaying vegetation.

With global satellite coverage, these quantities

can be observed nearly continuously and

in all regions.

Robinson (2004) provided a complete treatment

of the optical pathways involved in ocean


color remote sensing, starting with consideration

of the total radiance observed by the satellite

sensor. Many pathways contribute to the

observed radiance. These can be grouped into

an atmospheric path radiance (L p ), a “waterleaving

radiance” from just below the sea

surface (L w ), and a radiance due to all surface

reflections (L r ) within the instantaneous field

of view of the satellite sensor. The radiance L s

received at the satellite sensor is

L s ¼ L p þ TL w þ TL r (3.15)

T is the transmittance, which gives the

proportion of radiance that reaches the sensor

without being scattered out of the field of view.

The water-leaving radiance provides the

information about ocean color, so it is the

desired observed quantity. It is closely related

to the reflectance; the ratio of water-leaving

radiance just above the sea surface to downwelling

irradiance incident on the sea surface is the

“remote sensing reflectance,” or “normalized

water-leaving radiance.” The three net radiance

terms depend on the wavelength and on the

turbidity of the seawater. The largest

contribution is from the atmospheric pathway

L p . The water-leaving and reflected radiances

are much smaller. Because of the weak signal

strength for the ocean pathways (water-leaving

radiance), ocean color remote sensing requires

very precise atmospheric correction of the

visible light sensed by the satellite. Complex

radiative transfer models are invoked to carry

out this correction and often the accuracy of

the chlorophyll estimates depends critically on

this atmospheric correction. After correction,

the resultant radiances are analyzed for various

components related to biological activity, particularly

chlorophyll.

The biggest effect of chlorophyll on the spectrum

of reflectance (normalized water-leaving

radiance) is to reduce the energy at the blue

end of the spectrum compared with the spectrum

for clear water. This is demonstrated in

Figure 3.11 (H. Gordon, personal communication,

2009). Here the spectrum of radiance is

shown with and without correction for the

atmosphere. When the atmosphere is not

removed, there is virtually no difference

between the spectra for low and high chlorophyll

waters. When the atmospheric signal is

FIGURE 3.11 Example of observations

of water-leaving radiance observed by the

Multi-angle Imaging SpectroRadiometer

(MISR), with bands observed by satellite

color sensors indicated. Solid curves: low

chlorophyll water (0.01 mg/m 3 ). Dotted

curves: high chlorophyll water (10.0 mg/m 3 ).

The two lower curves have the atmospheric

signal removed. (H. Gordon, personal

communication, 2009.)

60


removed, the desired difference emerges. The

high chlorophyll spectrum is depressed at

the blue end of the spectrum and elevated at

the green and red.

Ocean color observations from satellites can

also be used as a proxy for the attenuation properties

of seawater, which can be used in mixed

layer models that are set up to be run with

observed atmospheric forcing. Absorption of

solar radiation in the ocean heats the upper

layer. The time and space distribution of absorption

is directly related to the substances in the

water column, and this affects ocean color. In

practice, at present most mixed layer models

in ocean general circulation models use a sum

of two exponentially decaying functions that

are proxies for attenuation of red light (quickly

absorbed) and blue-green light (penetrating

much deeper), with coefficients based on

assumption of a particular Jerlov (1976) water

type (Paulson & Simpson, 1977). However,

explicit incorporation of biological effects on

attenuation (incorporation of ocean color observations

and of bio-optical models along with

mixed layer models) is being tested widely

and is a likely direction for the future, since

the effects on modeled mixed layer temperature

are clear (Wu, Tang, Sathyendranath, & Platt,

2007) and can in fact affect the temperature

of the overlying atmosphere (Shell, Frouin,

Nakamoto, & Somerville, 2003).

3.9. ICE IN THE SEA

Ice in the sea has two origins: the freezing of

seawater and the ice broken off from glaciers.

The majority of ice comes from the first of these

sources and is referred to as sea ice; the glaciers

supply “pinnacle” icebergs in the Northern

Hemisphere and flat "tabular" icebergs in the

Southern Hemisphere. Sea ice alters the heat

and momentum transfers between the atmosphere

and the ocean, is a thermal insulator,

damps surface waves, changes the temperature

and salinity structure in the upper layer by

melting and freezing, and is a major hindrance

to navigation. Ice cover is an important part of

Earth’s climate feedbacks because of its high

reflectivity, that is, its high albedo (Section

5.4.3). The iceealbedo feedback, which affects

climate, is described in Section 5.4.5, and is especially

important in the Arctic (Section 12.8).

3.9.1. Freezing Process

When water loses sufficient heat (by radiation,

conduction to the atmosphere, convection,

or evaporation) it freezes to ice, in other words,

it changes to the solid state. Initial freezing

occurs at the surface and then the ice thickens

by freezing at its lower surface as heat is conducted

away from the underlying water

through the ice to the air.

The initial freezing process is different for

fresh and low-salinity water than for more

saline water because the temperature at which

water reaches its maximum density varies

with salinity. Table 3.4 gives the values of the

freezing point and temperature of maximum

density for water of various salinities. (Note

that the values are for freezing, etc., at atmospheric

pressure. Increased pressure lowers

the freezing point, which decreases by about

0.08 K per 100 m increase in depth in the sea.)

To contrast the freezing process for freshwater

and seawater, first imagine a freshwater

lake where the temperature initially decreases

from about 10 C at the surface to about 5 C

TABLE 3.4

Temperatures of the Freezing Point (t f ) and

of Maximum Density (t rmax ) for Fresh and

Salt Water

S 0 10 20 24.7 30 35 psu

t f 0 0.5 1.08 1.33 1.63 1.91 C

t rmax þ3.98 þ1.83 0.32 1.33 d C

Note that the values for freezing and so forth are at atmospheric

pressure. Increased pressure lowers the freezing point, which

decreases by about 0.08 K per 100 m increase in depth in the sea.

ICE IN THE SEA 61

at about 30 m depth. As heat is lost through the

surface, the density of the water increases and

vertical convective mixing (overturn) occurs

with the temperature of the surface water layer

gradually decreasing. This continues until the

upper mixed layer cools to 3.98 C and then

further cooling of the surface water causes its

density to decrease and it stays near the top.

The result is a rapid loss of heat from a thin

surface layer, which soon freezes. For seawater

of salinity ¼ 35 psu of the same initial temperature

distribution, surface cooling first results

in a density increase and vertical mixing by

convention currents occurs through a gradually

increasing depth, but it is not until the

whole column reaches 1.91 C that freezing

commences. As a much greater volume of

water has to be cooled through a greater

temperature range than in the freshwater

case, it takes longer for freezing to start in

salt water than in freshwater. A simple calculation

for a column of freshwater of 100 cm depth

and 1 cm 2 cross-section initially at 10 C shows

that it takes a heat loss of l63 J to freeze the top

1 cm layer, whereas for a similar column of

seawater of S ¼ 35 psu it takes a loss of 305 J

to freeze the top 1 cm because the whole

column has to be cooled to 1.91 C rather

than just the top 1 cm to 0 C for the freshwater.

Note that as seawater of salinity <24.7 psu

has a higher temperature of maximum density

than its freezing point, it will behave in a manner

similar to freshwater, although with a lower

freezing point. For seawater of salinity >24.7

psu, (in high latitudes) the salinity generally

increases with depth, and the stability of the

water column usually limits the depth of

convection currents to 30e50 m. Therefore ice

starts to form at the surface before the deep

water reaches the freezing point.

Generally, sea ice forms first in shallow water

near the coast, particularly where the salinity is

reduced by river runoff and where currents are

minimal. When fully formed, this sea ice connected

to the shore is known as “fast ice.” The

first process is the formation of needle-like

crystals of pure ice, which impart an “oily”

appearance to the sea surface (grease or frazil

ice). The crystals increase in number and form

a slush, which then thickens and breaks up

into pancakes of approximately one meter

across. With continued cooling, these pancakes

grow in thickness and lateral extent, eventually

forming a continuous sheet of floe or sheet ice.

Once ice has formed at the sea surface, when

the air is colder than the water below, freezing

continues at the lower surface of the ice and

the rate of increase of ice thickness depends on

the rate of heat loss upward through the ice

(and any snow cover). This loss is directly

proportional to the temperature difference

between top and bottom surfaces and inversely

proportional to the thickness of the ice and snow

cover.

With very cold air, a sheet of sea ice of up to

10 cm in thickness can form in 24 hours, the

rate of growth then decreasing with increasing

ice thickness. Snow on the top surface insulates

it and reduces the heat loss markedly, depending

on its degree of compaction. For instance,

5 cm of new powder snow may have insulation

equivalent to 250e350 cm of ice, while 5 cm of

settled snow can be equivalent to only 60e100

cm of ice, and 5 cm of hard-packed snow can

be equivalent to only 20e30 cm of ice.

As an example of the annual cycle of the

development of an ice sheet at a location in the

Canadian Arctic, ice was observed to start to

form in September, was about 0.5 m thick in

October, 1 m in December, 1.5 m in February,

and reached its maximum thickness of 2 m in

Maydafter which it started to melt.

3.9.2. Brine Rejection

In the initial stage of ice-crystal formation, salt

is rejected and increases the density of the neighboring

seawater, some of which then tends to

sink and some of which is trapped among the

ice crystals forming pockets called “brine cells.”

62


The faster the freezing, the more brine is trapped.

Sea ice in bulk is therefore not pure water-ice but

has a salinity of as much as 15 psu for new ice

(and less for old ice because gravity causes the

brine cells to migrate downward in time). With

continued freezing, more ice freezes out within

the brine cells leaving the brine more saline, in

a process called brine rejection. Someofthesalts

may even crystallize out.

Because of brine rejection, the salinity of firstyear

ice is generally 4e10 psu, for second-year

ice (ice that has remained frozen beyond the first

year) salinity decreases to 1e3 psu, and for

multiyear ice salinity may be less than 1 psu. If

sea ice is lifted above sea level, as happens

when ice becomes thicker or rafting occurs, the

brine gradually trickles down through it and

eventually leaves almost salt-free, clear old ice.

Such ice may be melted and used for drinking

whereas melted new ice is not potable. Sea ice,

therefore, is considered a material of variable

composition and properties that depends very

much on its history. (For more detail see

Doronin & Kheisin, 1975.)

As a result of brine rejection, the salinity of

the unfrozen waters beneath the forming sea

ice increases. When this occurs in shallow

regions, such as over continental shelves, the

increase in salinity can be marked and can result

in formation of very dense waters. This is the

dominant mechanism for formation of the

deep and bottom waters in the Antarctic

(Chapter 13), and for formation of the densest

part of the North Pacific Intermediate Water

in the Pacific (Chapter 10). Brine rejection is

a central process for modification of water

masses in the Arctic as well (Chapter 12).

3.9.3. Density and Thermodynamics

of Sea Ice

The density of pure water at 0 C is 999.9

kg/m 3 and that of pure ice is 916.8 kg/m 3 .

However the density of sea ice may be greater

than this last figure (if brine is trapped among

the ice crystals) or less (if the brine has escaped

and gas bubbles are present.) Values from 924 to

857 kg/m 3 were recorded on the Norwegian

Maud Expedition (Malmgren, 1927).

The amount of heat required to melt sea ice

varies considerably with its salinity. For S ¼ 0

psu (freshwater ice) it requires 19.3 kJ/kg from

2 C and 21.4 kJ/kg from 20 C, while for sea

ice of S ¼ 15 psu, it requires only 11.2 kJ/kg

from 2 C but 20.0 kJ/kg from 20 C. The small

difference of heat (2.1 kJ/kg) needed to raise the

temperature of freshwater ice from 20 C to

2 C is because no melting takes place; that is,

it is a true measure of the specific heat of pure

ice. However, for sea ice of S ¼ 15 psu, more

heat (8.8 kJ/kg) is required to raise its temperature

through the same range, because some ice

near brine cells melts and thus requires latent

heat of melting as well as heat to raise its temperature.

Note also that less heat is needed to melt

new ice (S ¼ 15 psu) than old ice, which has

a lower salinity.

3.9.4. Mechanical Properties of Sea Ice

Because of the spongy nature of first-year sea

ice (crystals þ brine cells) it has much less

strength than freshwater ice. Also, as fast freezing

results in more brine cells, the strength of ice

formed this way is less than when freezing occurs

slowly; in other words, sea ice formed in very

cold weather is initially weaker than ice formed

in less cold weather. As the temperature of ice

decreases, its hardness and strength increase,

and ice becomes stronger with age as the brine

cells migrate downward. When ice forms in

calm water, the crystals tend to line up in a pattern

and such ice tends to fracture along cleavage

planes more easily than ice formed in rough

water where the crystals are more randomly

arranged and cleavage planes are not formed.

The mechanical behavior of sea ice is

complex when temperature changes. As the ice

temperature decreases below its freezing point,

the ice expands initially, reaches a maximum

ICE IN THE SEA 63

expansion, and then contracts. For instance, an

ice floe of S ¼ 4 psu will expand by 1 m per

1 km length between 2 and 3 C, reaches its

maximum expansion at 10 C, and thereafter

contracts slightly. Ice of S ¼ 10 psu expands

4 m per 1 km length from 2to 3 C, and

reaches its maximum expansion at 18 C. The

expansion on cooling can cause an ice sheet to

buckle and “pressure ridges” to form, while

contraction on further cooling after maximum

expansion results in cracks, sometimes wide,

in the ice sheet.

Pressure ridges can also develop as a result of

wind stress on the surface driving ice sheets

together. The ridges on top are accompanied

by a thickening of the lower surface of the ice

by four to five times the height of the surface

ridges. Sea ice generally floats with about fivesixths

of its thickness below the surface and

one-sixth above, so relatively small surface

ridges can be accompanied by deep ridges

underneath d depths of 25 to 50 m below the

sea surface have been recorded. Thickening of

an ice sheet may also result from rafting when

wind or tide forces one ice sheet on top of

another or when two sheets, in compression,

crumble and pile up ice at their contact. Old

ridges, including piled up snow, are referred

to as hummocks. As they are less saline than

newer pressure ridges, they are stronger and

more of an impediment to surface travels than

the younger ridges.

3.9.5. Types of Sea Ice and its Motion

Sea ice can be categorized as fast ice (attached

to the shore), pack ice (seasonal to multiyear ice

with few gaps), and cap ice (thick, mostly multiyear

ice), as described in Section 12.7.1. Several

forces determine the motion of sea ice if it is

not landfast:

(a) Wind stress at the top surface (the

magnitude depending on the wind speed

and the roughness of the ice surface as

ridges increase the wind stress). Typical ice

speeds are 1 to 2% of the wind speed.

(b) Frictional drag on the bottom of an ice sheet

moving over still water tends to slow it

down, while water currents (ocean and

tidal) exert a force on the bottom of the ice in

the direction of the current. Because current

speeds generally decrease with increase in

depth, the net force on deep ice and icebergs

will be less than on thin ice, and pack ice will

move past icebergs when there is significant

wind stress.

(c) In the cases of (a) and (b), the effect of the

Coriolis force (Section 7.2.3) is to divert

the ice motion by 15e20 degrees to the

right of the wind or current stress in the

Northern Hemisphere (to the left in the

Southern Hemisphere). (It was the

observation of the relation between wind

direction and ice movement by Nansen,

and communicated by him to Ekman, that

caused the latter to develop his wellknown

theory of wind-driven currents.)

It is convenient to note that as surface

friction causes the surface wind to blow at

about 15 degrees to the left of the surface

isobars, the direction of the latter is

approximately that in which the ice is

likely to drift (Northern Hemisphere).

(d) If the ice sheet is not continuous, collisions

between individual floes may occur with

a transfer of momentum (i.e., decrease of

speed of the faster floe and increase of

speed of the slower). Energy may go into

ice deformation and building up ridges

at impact. This is referred to as internal

ice resistance and increases with ice

concentration, that is, the proportion of area

covered by ice. The effect of upper surface

roughness (R on a scale of 1 to 9) and ice

concentration (C on a scale of 1 to 9) on the

speed of the ice V (expressed as a percentage

of the wind speed) is given by: V ¼ R(1

0.08 C) (taken to only one decimal place),

so that the speed of the ice increases with

64


roughness but decreases with increased

ice concentration. Note that for very close

pack ice, stresses of wind or current are

integrated over quite large areas and the

local motion may not relate well to the local

wind.

3.9.6. Polynyas and Leads

Regions of nearly open water within the ice

pack are often found where one might expect

to find ice. These open water areas are critical

for airesea heat exchange, since ice is a relatively

good insulator. Small breaks between ice

floes are called leads; these are created by motion

of the ice pack and have random locations.

Larger recurrent open water areas are called

polynyas. There are two types of polynyas,

depending on the mechanism maintaining

the open water (Figure 3.12; see also Barber &

Massom, 2007):

1. Latent heat polynyas are forced open by winds,

often along coasts or ice shelf edges. New ice

soon forms; latent heat from the forming sea

ice is discharged to the atmosphere at a rate

of as much as 200e500 W/m 2 .

2. Sensible heat polynyas result from relatively

warm water upwelling to the surface

and melting the ice there. Another term

often encountered is flaw polynya, which

means that the polynya occurs at the

boundary between fast ice and pack ice.

Because most polynyas include a mixture

of these forcings, nomenclature is tending

toward being more specific about the

forcing (mechanicalewind; convectivee

melting: Williams, Carmack, & Ingram,

2007).

Wind-forced polynyas are usually near coastlines

or the edges of ice shelves or fast ice, where

winds can be very strong, often forced by strong

landesea temperature differences (katabatic

winds). The open water is often continually

freezing in these polynyas since they are kept

open through mechanical forcing. These windforced

polynyas act as ice factories, producing

larger amounts of new ice than regions where

the ice is thicker and airesea fluxes are minimized

by the ice cover. If these polynyas occur

over shallow continental shelves, the brine

rejected in the ongoing sea ice formation

process, together with temperatures at the

(a)

Wind-forced (latent heat) polynya

Wind

(b)

Heat loss

Polynya

Brine

rejection

Dense shelf water

Sea ice

Cold, fresher

Cold,

saltier

Melting (sensible heat) polynya, driven by mixing

Heat loss

Polynya

Tidal mixing Melting

Vertical

excursion

Sea ice

Cold, fresher

Warm, saltier

FIGURE 3.12 Schematics of polynya formation: (a) latent heat polynya kept open by winds and (b) sensible heat polynya

kept open by tidal mixing with warmer subsurface waters (after Hannah et al., 2009).

ICE IN THE SEA 65

freezing point, can produce especially dense

shelf waters. This is one of the major mechanisms

for creating very dense waters in the

global ocean (Section 7.11), particularly around

the coastlines of Antarctica (Chapter 13) and

the Arctic (Chapter 12), as well as the densest

(intermediate) water formed in the North Pacific

in the Okhotsk Sea (Chapter 10).

Polynyas that are maintained by melting

within the ice pack result from mixing of the

cold, fresh surface layer with underlying

warmer, saltier water. These polynyas might

also produce sea ice along their periphery

since the airesea fluxes will be larger than

through the ice cover, but the upward heat

flux from the underlying warmer water means

that they produce new ice much less efficiently

than wind-forced polynyas. The vertical mixing

can result from convection within the polynya,

which can occur in deep water formation sites

such as the Odden-Nordbukta in the Greenland

Sea (Section 12.2.3). In shallow regions, the mixing

process can be greatly enhanced by tides

moving the waters over undersea banks

(Figure 3.12b). A number of the well-known

polynyas in the Canadian Archipelago in

the Arctic Ocean are tidally maintained

(Figure 12.23 from Hannah, Dupont, & Dunphy,

2009), as is a recurrent polynya over Kashevarov

Bank in the Okhotsk Sea (Figure 10.29).

3.9.7. Ice Break-up

Ice break-up is caused by wave action, tidal

currents, and melting. Melting of ice occurs

when it gains enough heat by absorption of

solar radiation and by conduction from the

air and from nearby seawater to raise its

temperature above the melting point. The

absorption of radiation depends on the albedo

of the surface (proportion of radiation

reflected), which varies considerably; for

example, the albedo for seawater is from 0.05

to 0.10 (it is a very good absorber of radiation),

for snow-free sea ice it is from 0.3 to 0.4, while

for fresh snow it is 0.8 to 0.9. Dark materials,

like dirt and dust, have a low albedo of 0.1

to 0.25 and absorb radiation well. Such material

on ice can form a center for the absorption

of radiation and consequent melting of ice

around it, so puddles can form. These can

absorb heat because of the low albedo of

water and may even melt right through an

ice sheet. When any open water forms, it

absorbs heat and causes rapid melting of ice

floating in it.

C H A P T E R

4

Typical Distributions of Water

Characteristics

4.1. INTRODUCTION

In this chapter, we describe the typical distributions

of water properties such as temperature,

salinity, oxygen, and nutrients. The properties

were introduced in Chapter 3. Here we highlight

distributions that are common, for

instance, to the Atlantic, Pacific, and Indian

Oceans, or to all subtropical regions, or to all

three equatorial regions, and so on. The overview

provides an essential framework for the

heat and freshwater budgets of Chapter 5 and

for the detailed descriptions of properties and

circulation in each ocean basin presented in later

chapters. Summaries of some of the large-scale

water masses are included in Chapter 14.

Several central concepts are useful for

studying large-scale water properties. First,

most water properties are initially set at the sea

surface and are then modified within the ocean

through a process called ventilation. Ventilation

is the connection between the surface and the

ocean interior (similar to breathing). Second, the

ocean is vertically stratified in density, and flow

within the ocean interior is primarily along isentropic

(isopycnal) surfaces rather than across

them. That is, flow within the ocean interior is

nearly adiabatic (without internal sources of

heat and freshwater). Third, as a result, water

properties are helpful for identifying flow paths

from the surface into the interior, and for identifying

forcing and mixing processes and locations.

This is related to the usefulness of the concept of

water masses, defined in the next section.

Most water characteristics have large and

typical variations in the vertical direction, which

encompasses an average of 5 km in the deep

ocean, whereas variations of similar magnitude

in the horizontal occur over vastly greater

distances. For instance, near the equator, the

temperature of the water may drop from 25 C

at the surface to 5 C at a depth of 1 km, but it

may be necessary to go 5000 km north or south

from the equator to reach a latitude where the

surface temperature has fallen to 5 C. The

average vertical temperature gradient (change

of temperature per unit distance) in this case is

about 5000 times the horizontal one. However,

the more gradual horizontal variations are

important: the horizontal density differences

are associated with horizontal pressure differences

that drive the horizontal circulation,

which is much stronger than the vertical circulation.

To illustrate the three-dimensional distributions

of water properties and velocities, we

use a number of one- and two-dimensional

representations, such as profiles, vertical sections,

and horizontal maps.

Much of the geographic variation in properties

in the oceans and atmosphere occurs in


67



68

4. TYPICAL DISTRIBUTIONS OF WATER CHARACTERISTICS

the north-south (meridional) direction. Properties

are often much more uniform in the eastwest

(zonal) direction. A principal exception to

the latter is the important zonal variation near

boundaries, especially on the west sides of

ocean basins. In addition to the major ocean

basins, we also refer to general regions that are

mainly distinguished by latitude ranges.

The equatorial region refers to the zone within

several degrees of the equator, while tropical

refers to zones within the tropics (23 Nor S

of the equator). In the equatorial region, the

effect of the earth’s rotation on currents is

minimal, leading to very distinctive currents

and water property distributions compared

with other regions. Within the tropics, there is

net heating at the sea surface. The distinction

between equatorial and tropical is often significant,

but when the two are to be lumped

together, they are referred to as the low latitudes.

In contrast, the regions near the poles, north and

south, are called the high latitudes. Subtropical

refers to mid-latitude zones poleward of the

tropics, characterized by atmospheric high pressure

centers. Polar is used for the Arctic and

Antarctic regions, where there is net cooling

and usually sea ice formation. Subpolar refers

to the region between the strictly polar conditions

and those of the temperate mid-latitudes.

The most marked seasonal changes take

place in the temperate zones (approximately

30e60 Nor S).

Throughout this chapter and in subsequent

chapters we refer to the concept of a water

mass, which is a body of water that has had its

properties set by a single identifiable process.

This process imprints properties that identify

the water mass as it is advected and mixed

through the ocean. Most water masses are

formed at the sea surface where their identifying

characteristics are directly related to

surface forcing, but some water masses acquire

their characteristics (e.g., an oxygen minimum)

through subsurface processes that might be

biogeochemical as well as physical. Some water

masses are nearly global in extent while other

water masses are confined to a region, such as

a gyre in a specific ocean basin. Water masses

have been given names that are usually capitalized.

Some water masses have several names,

simply because of the history of their study.

Water type and source water type are useful

related concepts; a water type is a point in property

space, usually defined by temperature and

salinity, and a source water type is the water

type at the source of the water mass (e.g.,

Tomczak & Godfrey, 2003).

Each water mass is introduced in terms of

(1) its identifying characteristic(s) and (2) the

ocean process that creates that specific characteristic.

Descriptive physical oceanographers

often first identify an extremum or interesting

central characteristic. They then seek to find

the process that created that characteristic.

Once the process is identified, additional

information about the process is used to refine

that water mass’s definition, for example, the

full density range might be assigned to the

water mass. Information about the process and

water mass distribution assists in studying the

circulation.

The Mediterranean Water (MW) (Chapter 9)

is an example of a water mass with a simple

identifying characteristic. MW is a salinity

maximum layer in the North Atlantic at middepth

(1000e2000 m) and a lateral salinity

maximum on any quasi-horizontal surface

cutting through the layer (e.g., Figure 6.4). Its

source is the saline outflow of water from the

Mediterranean through the Strait of Gibraltar.

Its high salinity results from excess evaporation

and internal dense water formation within

the Mediterranean Sea (see the textbook Web

site, which contains supplementary materials,

http://booksite.academicpress.com/DPO/ to

view Section S8.10.2; “S” denotes supplemental

material). The MW density range within the

North Atlantic is a function of both its high

density at the Strait of Gibraltar and also

intense mixing with ambient (stratified) North

TEMPERATURE DISTRIBUTION OF THE OCEANS 69

Atlantic water as it plunges down the continental

slope after it exits the Strait of Gibraltar.

Subtropical Mode Water (STMW) is another

example of a water mass with a simple vertical

extremum; in this case its thickness (vertical

homogeneity) is compared with waters above

and below it. A type of STMW is found in

each ocean’s subtropical gyre (Sections 9.8.2,

10.9.1, and 11.8.1). STMW originates in a thick

surface winter mixed layer that is then advected

down along isopycnals into the ocean interior.

STMW retains its signature of relative thickness,

just as the MW retains its signature of high

salinity. Slow mixing within the ocean interior

eventually erodes these extrema, but they

persist far enough from their sources to be

useful tracers of flow.

Many other major world water masses are

introduced in this chapter. Detailed descriptions

of them and of their formation processes are

provided in the ocean basin chapters (9 through

13), with a final summary in Chapter 14.

Taking into account the whole set of ocean

properties and information about water masses,

it is useful to think of the vertical structure

in terms of four layers: upper, intermediate,

deep, and bottom. The upper layer contains

a surface mixed layer, thermocline and/or

halocline, pycnocline, and other structures

embedded in these (see descriptions with

respect to temperature and density in Sections

4.2 and 4.4). The upper layer is in contact

with the atmosphere, either directly or through

broad flow (relatively directly) into the upper

ocean through the subduction process

described in Sections 4.4.1 and 7.8.5. The intermediate,

deep, and bottom layers are all below

the pycnocline, or at most, embedded within

the bottom of it. These layers are identified by

water masses that indicate surface origins,

with respect to location and formation processes,

and relative age.

Before describing some typical distributions

of each of the water properties, the following

information on ocean water temperatures

and salinities is given for orientation (see

Figure 3.1):

1. 75% of the total volume of the ocean water

has a temperature between 0 and 6 C and

salinity between 34 and 35 psu,

2. 50% of the total volume of the oceans has

properties between 1.3 and 3.8 C and

between 34.6 and 34.7 psu,

3. The mean temperature of the world ocean is

3.5 C and the mean salinity is 34.6 psu.

4.2. TEMPERATURE

DISTRIBUTION OF THE OCEANS

The ocean and atmosphere interact at the sea

surface. Surface forcing from the atmosphere

and sun sets the overall pattern of sea surface

temperature (SST) (Figure 4.1). High SST in the

tropics is due to net heating, and low SST at

high latitudes is due to net cooling. Beyond this

simple meridional variation, the more complex

features of SST result from ocean circulation

and spatial variations in atmospheric forcing.

The ocean’s surface, which could include sea

ice, provides the forcing at the bottom of the

atmosphere through various kinds of heat

forcing and as a source of water vapor.

SST ranges from slightly more than 29 Cin

the warmest regions of the tropics, to freezing

temperature (about 1.8 C; Figure 3.1) in iceforming

regions, with seasonal variations especially

apparent at middle to high latitudes.

Below the sea surface, we refer only to potential

temperature so that the pressure effect on

temperature is removed (Section 3.3 and

Figure 3.3). The vertical potential temperature

structure can usually be divided into three

major zones (Figure 4.2): (1) the mixed layer,

(2) the thermocline, and (3) the abyssal layer.

This structure is typical of low and midlatitudes

with high SST. Relative to the fourlayer

structure introduced in Section 4.1, the

first two zones are within the upper layer and

70


(a)

40˚

20˚

0˚

20

18

20˚

60˚ 2

0

26

26

22 24

20

80˚N

4

16

28

8

10

20

60˚W

Winter surface temperature

0˚ 60˚E 120˚E 180˚ 120˚W

0

4

2

6

8

4

8

16

10

18

20

22

20

24

26

28

28

29

26

24

22 20

80˚N

60˚

40˚

20˚

0˚

20

20˚

40˚

60˚

6

10 1214

16

10

4

2

0

-1.5

10

60˚

40˚

80˚S

60˚W

0˚ 60˚E 120˚E 180˚ 120˚W

80˚S

(b)

FIGURE 4.1 (a) Surface temperature ( C) of the oceans in winter (January, February, March north of the equator; July,

August, September south of the equator) based on averaged (climatological) data from Levitus and Boyer (1994). (b) Satellite

infrared sea surface temperature ( C; nighttime only), averaged to 50 km and 1 week, for January 3, 2008. White is sea ice.

(See Figure S4.1 from the online supplementary material for this image and an image from July 3, 2008, both in color). Source:

From NOAA NESDIS (2009).


0

500

Potential temperature (°C)

0 10 20 300 10 20 300

10

Mixed Layer

Thermocline

Thermocline

Thermocline

Thermostad

Dichothermal

layer

Depth (m)

1000

Abyssal

Abyssal

1500

2000

(a)

Low latitude

(tropical North Pacific)

(b)

Eastern N. Pacific

Western N. Pacific

Mid-latitude

(subtropical North Pacific)

(c)

High

latitude

(subpolar

N. Pac.)

FIGURE 4.2 Typical potential temperature ( C)/depth (m) profiles for the open ocean in (a) the tropical western North

Pacific (5 N), (b) the western and eastern subtropical North Pacific (24 N), and (c) the western subpolar North Pacific

(47 N). Corresponding salinity profiles are shown in Figure 4.16.

the third temperature zone contains the intermediate,

deep, and bottom layers.

In high latitudes where SST is low, this structure

differs, and can have a mixed layer,

a vertical temperature minimum and underlying

maximum near the sea surface, and then

the thermocline and abyssal layer.

The mixed layer (Section 4.2.2) is a surface

layer of relatively well-mixed properties. In

summer in low latitudes, it can be very thin or

non-existent. In winter at middle to high latitudes,

it can be hundreds of meters thick, and

in isolated deep convection regions, the mixed

layer can be up to 2000 m thick. Mixed layers

are mixed by both wind and surface buoyancy

forcing (air-sea fluxes). The thermocline (Sections

4.2.3 and 4.2.4) is a vertical zone of rapid

temperature decrease with a depth of roughly

1000 m. In the abyssal layer, between the thermocline

and ocean bottom, potential temperature

decreases slowly. At high latitudes,

a near-surface temperature minimum (dichothermal

layer) is often found, a holdover from a cold

winter mixed layer that is “capped” with

warmer waters in other seasons (Figure 4.2c);

the underlying temperature maximum (mesothermal

layer) results from advection of waters

from somewhat warmer locations. This temperature

structure is stable because there is strong

salinity stratification, with fresher water in the

surface layer.

Typical temperatures at subtropical latitudes

are 20 C at the surface, 8 C at 500 m, 5 Cat

1000 m, and 1e2 C at 4000 m. All of these values

and the actual shape of the temperature profile

are a function of latitude, as shown by the three

different profiles in Figure 4.2.

There are some notable additions to this basic

three-layered structure. In all regions, spring

and summer warming produces a thin warm

layer overlying the winter’s mixed layer. In

the western subtropical regions as well as

other regions, there are often two thermoclines

with a less stratified (more isothermal) layer

(thermostad) between them, all within the upper

1000 m (Figure 4.2b). In some regions another

72


mixed layer is found at the very bottom

(“bottom boundary layer”) and can be up to

100 m thick.

In many parts of the ocean, density is a strong

function of temperature (Chapter 3), and has the

same layered structure as temperature; that is,

an upper layer, a pycnocline with rapidly

increasing density, and an abyssal zone. Salinity

usually has a more complicated vertical structure

(Section 4.3). In regions of high precipitation

and/or runoff (such as subpolar and high

latitude regions and parts of the tropics), salinity

may be more important than temperature in

setting the vertical density structure, especially

in the upper layer, since the water column

must be vertically stable on average. A typical

vertical salinity profile in these regions includes

a relatively fresh surface layer with a halocline

separating the surface layer from the higher

salinity water below. The higher underlying

salinity is an indication of a sea-surface source

of water in a less rainy area. On the other

hand, in the subtropics where the sea surface

salinity is dominated by evaporation, surface

water is usually more saline than the underlying

water. Here temperature clearly dominates the

vertical stability.

This three-layered structure is simpler than

our simplest description of overall water

mass structure, for which at least four layers

are usually required (Section 4.1). The abyssal

layer, in terms of temperature, usually includes

at least two or possibly three separate water

mass layers: intermediate, deep, and bottom

waters. However, potential temperature is relatively

low in all of these water mass-based

layers, declining toward the bottom, and is

not a useful indicator of these water mass

layers.

4.2.1. Surface Temperature

The temperature distribution at the surface of

the open ocean is approximately zonal, with the

curves of constant temperature (isotherms)

running roughly east-west (Figure 4.1). Near

the coast where the currents are diverted by

the boundaries, the isotherms may swing more

nearly north and south. Also, along the eastern

boundaries of the oceans, surface temperatures

are often lower due to upwelling of subsurface

cool water, for example, along the west coast

of North America in summer, causing the

isotherms to trend equatorward. Upwelling

also causes lower surface temperatures in the

eastern equatorial Pacific and Atlantic.

The open ocean SST, averaged over all longitudes

and displayed as a function of latitude

(Figure 4.3), decreases from as high as 28 C

just north of the equator to nearly 1.8 C near

sea ice at high latitudes. This distribution corresponds

closely with the input of short-wave

radiation (mainly from the sun), which is highest

in the tropics and lowest at high latitudes

(Section 5.4.3). The corresponding mean zonal

surface salinity and density are also shown.

Salinity and density are discussed in Sections

4.3 and 4.4. Density is dominated by temperature.

Salinity has subtropical maxima in both

the Northern and Southern Hemispheres and

a minimum just north of the equator.

Because many satellites observe SST and SSTrelated

quantities, many different SST products

are available, providing daily and longer term

average maps with higher spatial and temporal

resolution than the climatology based on

in situ data shown in Figure 4.1a. Global SST

based on infrared imagery for one week in

January (boreal winter, austral summer) is

shown in Figure 4.1b. (The equivalent image

for July is included in the online supplementary

materials as Figure S4.1.) The structures of ocean

currents, fronts, upwelling regions, eddies, and

meanders are more apparent in these nearly

synoptic SST images.

Non-zonal features of global SSTs that are

most apparent and important to note in Figure

4.1 include the warm pool and the cold tongue.

The warm pool is the warmest SST region,

located in the western tropical Pacific, through


(a)

Temperature (°C)

(b)

Salinity

Density – 1000 (kg/m 3 )

30

25

20

15

10

5

0

36

35

34

33

32

31

30

(c)

30

28

26

24

22

20

90°S 60°S 30°S 0° 30°N 60°N 90°N

Latitude

FIGURE 4.3 Variation with latitude of surface (a)

temperature, (b) salinity, and (c) density averaged for all

oceans for winter. North of the equator: January, February,

and March. South of the equator: July, August, and

September. Based on averaged (climatological) data from

Levitus and Boyer (1994) and Levitus et al. (1994b).

the Indonesian passages, and into the tropical

Indian Ocean. The cold tongue is the narrow

tongue of colder water along the equator in

both the eastern Pacific and Atlantic. This

forms due to upwelling of thermocline water

along the equator. Because the thermocline is

shallower in the eastern Pacific and Atlantic

than in the west, upwelling brings up colder

water in the east.

In each ocean, warm regions are centered in

the west, off the equator. Cooler waters cycle

equatorward in the central and eastern parts of

each ocean. These SST patterns reflect the

anticyclonic circulation of the subtropical gyres

(clockwise in the Northern Hemisphere, counterclockwise

in the Southern Hemisphere),

which advects warm water away from the

tropics and cooler water toward the equator.

There are also regions of warmer water in the

eastern tropical North Pacific and North

Atlantic. These are found east of the subtropical

circulation and north of the cold tongue;

high temperatures are not suppressed by

either the anticyclonic circulation or equatorial

upwelling.

In the subpolar North Pacific and North

Atlantic, there is, again, evidence of the circulation

in the SST pattern. Here the gyres

are cyclonic (counterclockwise in the Northern

Hemisphere). Warmer waters are advected

northward in the eastern parts of these circulations

(along the coast of British Columbia and

along northern Europe). Warmer water extends

far to the north in the Atlantic toward the Arctic,

along the Norwegian coast. Cold waters are

found in the western parts of these circulations,

along the Kamchatka/Kuril region in the Pacific

and Labrador/Newfoundland region in the

Atlantic.

In the Southern Ocean, SST is not exactly

zonal. This reflects excursions in the Antarctic

Circumpolar Current (ACC), which is also not

zonal. Colder waters are farther north in the

Atlantic and Indian Oceans and pushed southward

in the Pacific (Section 13.4).

In the satellite SST maps (Figures 4.1b and

S4.1 from the online supplementary material),

eddy-scale (100e500 km) features are apparent

even with global maps, particularly where the

color scaling provides large contrasts. Especially

visible are the large wavelike structures

in the equatorial regions; length scales of tropical

waves are longer than at higher latitudes

so they are better resolved in this map. The

waves around the Pacific’s equatorial cold

tongue are the Tropical Instability Waves

(TIWs), with timescales of about a month

(Section 10.7.6).

74


4.2.2. Upper Layer Temperature and

Mixed Layer

Within the ocean’s near-surface layer, properties

are sometimes very well mixed vertically,

particularly at the end of the night (diurnal

cycle) and in the cooling seasons (seasonal

cycle). This is called the mixed layer. This layer

is mixed by the wind and by buoyancy loss

due to net cooling or evaporation at the sea

surface. It is unmixed by warming and precipitation

at the sea surface and by circulations

within the mixed layer that move adjacent

mixed waters of different properties over

each other. Processes that create and destroy

the mixed layer are described in much greater

detail in Section 7.4.1. Here we focus on the

observed structure and distribution of mixed

layers.

As a rule of thumb, wind-stirred mixed

layers do not extend much deeper than 100 or

150 m and can reach this depth only at the

end of winter. On the other hand, infrequent

vigorous cooling or evaporation at the sea

surface can cause the mixed layer to deepen

locally to several hundred meters, or even

briefly in late winter to more than 1000 m in

isolated deep convection locations. Mixed

layers in summer may be as thin as 1 or 2 m,

overlying a set of remnant thin mixed layers

from previous days with storms, and thicker

remnant mixed layers from winter. Because

the mixed layer is the surface layer that

connects the ocean and atmosphere, and

because sea-surface temperature is the main

way the ocean forces the atmosphere, observations

of the mixed layer and understanding

how it develops seasonally and on climate timescales

is important for modeling and understanding

climate.

A given vertical profile will not usually

exhibit a thick, completely mixed layer of

uniform temperature, salinity, and density.

Most often, there will be small steps, nearly

discontinuities, in the profiles due to daily

restratification and remixing with layers sliding

in from nearby. For a careful study of the mixed

layer, the investigator assigns the mixed layer

depth based on examination of every vertical

profile. However, for general use (e.g., with the

growing profiling float data set, or for use in

upper ocean property mapping for fisheries,

climate prediction, or navigational use), it is

not feasible to examine each profile, and it is

important to have consistent criteria for assigning

the mixed layer depth. Functional definitions

of mixed layer depth have been

developed, mostly based on finding a set

temperature or density difference between the

surface observation and deeper observations;

this is the so-called “threshold method.” In

tropical and mid-latitudes, temperature-based

definitions are adequate, but at higher latitudes,

it is common to find a subsurface temperature

maximum lying underneath a low salinity

surface layer. Currently, the most commonly

used criterion is a density difference of s q ¼

0.03 kg/m 3 or temperature difference of 0.2 C,

as used in the mixed layer maps shown in

Figure 4.4a,b (deBoyer Montégut et al., 2004).

Other treatments have employed larger thresholds

(e.g., 0.8 C in Kara, Rochford, & Hurlburt,

2003) or more detailed criteria that fit the

observed vertical profiles rather than relying

on a threshold (Holte & Talley, 2009). A global

map of the maximum mixed layer depth, using

the latter method, is shown in Figure 4.4c (Holte,

Gilson, Talley, & Roemmich, 2011).

In all regions, winter mixed layers are much

thicker than summer mixed layers. The main

features of the global winter mixed layer maps

are the thick mixed layers in the northern North

Atlantic and in a nearly zonal band in the

Southern Ocean. These regions correspond to

maxima in anthropogenic carbon uptake

(Sabine et al., 2004), so they have practical implications

for global climate. These thick winter

mixed layers are the main source of Mode

Waters, which are identified as relatively thick

layers in the upper ocean (Section 4.2.3).


Mixed layer development is affected by the

amount of turbulence in the surface layer. This

turbulence is generated by breaking surface

and internal waves generated by the wind,

decreasing with increasing depth. Mixed layer

development can also be affected by Langmuir

cells, which are transient helical circulations

(in the vertical plane) aligned parallel to the

wind (Section 7.5.2). These create the “wind

rows” sometimes seen at the sea surface under

the wind, where the water is pushed together,

or converges, in the Langmuir cells, which reach

to about 50 m depth and 50 m width, and can

create turbulence that affects mixing in the

mixed layer.

Another dynamical phenomenon present in

the near-surface layer is the Ekman response

to wind forcing, which forces flow in the

ocean’s surface layer off to the right of the

wind in the Northern Hemisphere (and to

the left in the Southern Hemisphere), because

of the Coriolis force (Section 7.5.3). Turbulence

in the surface layer acts like friction. In the

Northern Hemisphere, each thin layer within

the surface layer pushes the one below it a little

more to the right, and with a little smaller

FIGURE 4.4 Mixed layer depth in (a) January and (b) July, based on a temperature difference of 0.2 C from the nearsurface

temperature. Source: From deBoyer Montégut et al. (2004). (c) Averaged maximum mixed layer depth, using the 5

deepest mixed layers in 1 1 bins from the Argo profiling float data set (2000e2009) and fitting the mixed layer structure

as in Holte and Talley (2009). This figure can also be found in the color insert.

76


velocity than the layer above. This creates an

“Ekman spiral” of decreasing velocities with

increasing depth. The whole spiral occurs

within the top 50 m of the ocean. If all of the

velocities are added together to calculate the

total transport in the Ekman layer, the net effect

is that this Ekman transport moves at exactly

right angles to the wind direction d to the

right in the Northern Hemisphere and to the

left in the Southern Hemisphere. Ekman velocities

are small and do not generate turbulence.

Thus they have no direct effect on mixed layer

development and are affected by the upper

layer turbulence but not by the mixed layer

stratification. The Ekman response is crucial,

however, for conveying the effect of the wind

to the ocean, for development of the large-scale

and long timescale ocean circulation, as also

described in Chapter 7.

4.2.3. Thermocline, Halocline, and

Pycnocline

Below the surface layer, which can be well

mixed or can include messy remnants of local

mixing and unmixing, temperature begins to

decrease rapidly with depth. This rapid decrease

ceases after several hundred meters,

with only small vertical changes in temperature

in the deep or abyssal layer that extends

on down to the bottom. The region of higher

vertical temperature gradient (rate of decrease

of temperature with increasing depth) is called

the thermocline. The thermocline is usually

a pycnocline (high vertical density gradient). It

is often hard to precisely define the depth

limits, particularly the lower limit, of the thermocline.

However, in low and middle latitudes,

a thermocline is always present at

depths between 200 and 1000 m. This is

referred to as the main or permanent thermocline.

In polar and subpolar waters, where the

surface waters may be colder than the deep

waters, there is often no permanent thermocline,

but there is usually a permanent halocline

(high vertical salinity gradient) and associated

permanent pycnocline.

The continued existence of the thermocline

and pycnocline requires explanation. There are

two complementary concepts, one based on

vertical processes only, and the other based on

horizontal circulation of the waters that form

the thermocline away from where they outcrop

as mixed layers in winter. Both concepts are

important and work together.

The vertical processes that affect the thermocline

are downward transfer of heat from the sea

surface and either upwelling or downwelling

(these depend on the location in the ocean and

on what creates the vertical motion). One might

expect that as the upper waters are warmest,

heat would be transferred downward by diffusion

despite the inhibiting effect of the stability

in the pycnocline/thermocline, and that the

temperature difference between the upper and

lower layers would eventually disappear.

However, the deeper cold waters are fed continuously

from the sea surface at higher latitudes

(deep and bottom water formation regions,

mainly in the northernmost North Atlantic and

Greenland Sea and in various regions around

Antarctica). These deep inflows maintain the

temperature difference between the warm

surface waters and cold deep waters. The deep

waters upwell and warm up through downward

diffusion of heat. If upwelling from the

bottommost layers to near the surface occurs

through the whole ocean, the upward speed

would be 0.5e3.0 cm/day. Unfortunately these

speeds are too small to accurately measure

with current instruments, so we are unable to

test the hypothesis directly. The result of the

downward vertical diffusion of heat balanced

by this persistent upwelling of the deepest

cold waters results in an exponential vertical

profile of temperature (Munk, 1966), which

approximates the shape of the permanent

thermocline.

This simplified vertical model of the thermocline

is depicted in Figure 4.5, which shows


Depth (z)

Temperature (T)

Thermocline

Downward diffusion

Upward advection

wT

FIGURE 4.5 Vertical processes that can maintain the

thermocline in a simplified one-dimensional model.

a typical vertical temperature profile in the

upper ocean containing the thermocline. The

result of downward diffusion of heat is labeled

as A vT

vz

and the result of upward vertical advection

of colder, deeper water is labeled “wT”.

(Equation (7.46) shows these two terms are

the vertical integrals of the vertical diffusion

and vertical advection terms, assuming constant

eddy diffusivity A and constant vertical

velocity w. In this simplest of thermocline

models, it is assumed that downward diffusion

of heat is entirely balanced by upward advection.)

If we assume that the difference between

these two terms is a constant, we have an equation

with an exponential solution for temperature

T, which in many cases approximates the

shape of the thermocline. We can use similar

arguments relative to the vertical distribution

of tracers like dissolved oxygen except that

such tracers can have both sources and sinks

within the water column, ultimately resulting

in subsurface maxima or minima.

A second, more horizontal, adiabatic and

complementary process for maintaining the

thermocline/pycnocline was suggested by

Iselin (1939) and further developed by Luyten,

Pedlosky, and Stommel (1983; Section 7.8.5). Iselin

observed that the surface temperaturesalinity

relation along a long north-south swath

in the North Atlantic strongly resembled the T-S

relation in the vertical (Figure 4.6). He hypothesized

that the waters in the subtropical thermocline

therefore originate as surface waters

farther to the north. As they move south, the

colder surface waters subduct beneath the

warmer surface waters to the south (using

the term from Luyten et al., borrowed from plate

tectonics). Subduction of many layers builds up

the temperature, salinity, and density structure

of the main pycnocline (thermocline) in the

subtropical gyre. This process is adiabatic, not

requiring any mixing or upwelling across isopycnals.

Such one-dimensional diapycnal

processes would then modify the thermocline

structure, smoothing it out.

Double diffusion (Section 7.4.3) is another

vertical mixing process that might affect the

thermocline (pycnocline). This process might

modify the relation between temperature and

salinity within the pycnocline, smoothing the

profile that results from adiabatic subduction

(Schmitt, 1981).

The main thermoclines/pycnoclines of the

world’s subtropical gyres are permanent features.

The temperature-salinity relation in the

thermocline of each subtropical gyre is shown

in Figure 4.7. The main thermoclines are identifiable

in temperature/salinity relations, and

they have a common formation history that is

some combination of subduction and vertical

upwelling/diffusion. Therefore, the waters in

the thermocline can be identified as a water

mass. This is the first water mass that we introduce

systematically, rather than as an example.

The thermocline water mass is Central Water.

Central Water differs from typical water masses

because it has a large range of temperature,

salinity and density.

So far, we have referred to the “main,” or

permanent, thermocline. There are also permanent,

double thermoclines in some large but

78


FIGURE 4.6 Temperature-salinity along surface swaths in the North Atlantic (dots and squares), and in the vertical (solid

curves) at stations in the western North Atlantic (Sargasso Sea) and eastern North Atlantic. Source: From Iselin (1939).

geographically restricted regions. For instance,

two thermoclines are found in the Sargasso

Sea just south of the Gulf Stream. A layer of

lower vertical stratification separates the two

thermoclines. The layer of lower stratification

is called a thermostad (or pycnostad for the equivalent

density layer).

The thermostad/pycnostad is often given the

water mass name, Mode Water. This is the second

water mass that we introduce. Mode Water is

FIGURE 4.7 Potential temperature-salinity

relation in the thermocline of each subtropical

gyre. These are the Central Waters. R is the best

fit of a parameter associated with double

diffusive mixing (Section 7.4.3). Source: From

Schmitt (1981).

considered a water mass because it is identified

by a particular characteristic (a vertical

extremum in layer thickness), and because it

has a specific formation process (subduction of

thick mixed layers). The name “Mode Water”

was introduced by Masuzawa (1969). Volumetrically

there is more water in a particular

temperature/salinity range than in the thermoclines

above and below it, so Mode Water

appears as a mode in the distribution of volume

in temperature/salinity space.

In the region where the Mode Water outcrops

as a thick mixed layer, the overlying thermocline

is actually a seasonal thermocline that disappears

in late winter. After Mode Water subducts,

its thermostad is embedded in the permanent

thermocline, creating a double thermocline.


4.2.4. Temporal Variations of

Temperature in the Upper Layer

and Thermocline

The temperature in the upper zone and into the

thermocline varies seasonally, particularly in

mid-latitudes. In winter the surface temperature

is low, waves are large, and the mixed layer is

deep and may extend to the main thermocline.

In summer the surface temperature rises, the

water becomes more stable, and a seasonal thermocline

often develops in the upper layer.

The growth and decay of the seasonal thermocline

is illustrated in Figure 4.8a using

monthly mean temperature profiles from

March 1956 to January 1957 taken at Ocean

Weather Station P (“Papa”) in the northeastern

(subpolar) North Pacific. From March to August,

the temperature gradually increases due to

absorption of solar energy. A mixed layer from

the surface down to 30 m is evident all the

time. After August there is a net loss of heat

and continued wind mixing; these erode away

the seasonal thermocline until the isothermal

condition of March is approached again. Note

that March does not have the maximum heat

FIGURE 4.8 Growth and decay of the seasonal thermocline

at 50 N, 145 W in the eastern North Pacific as (a)

vertical temperature profiles, (b) time series of isothermal

contours, and (c) a time series of temperatures at depths

shown.

loss; rather, it is the last month of cooling before

seasonal heating begins. Therefore total heat

content is lowest in March. In tropical and

subtropical locations, the summer mixed layer

may be even thinner.

These same data may be presented in alternative

forms; for instance, as a time series showing

80


the depths of the isotherms during the year

(Figure 4.8b). (The original data include the

alternate months, which were omitted from

Figure 4.8a to avoid crowding.) In Figure 4.8c

the temperatures are plotted at selected depths.

The different forms in which the thermocline

appears in these three presentations should be

noted. In Figure 4.8a, the permanent thermocline

appears as a maximum gradient region in

the temperature/depth profiles. In Figure 4.8b,

the thermocline appears as a crowding of the

isotherms, which rises from about 50 m in

May to 30 m in August and then descends to

100 m in January. In Figure 4.8c, the thermocline

appears as a wide separation of the 20- and 60-m

isobaths between May and October, and

between the 60- and 100-m isobaths after that

as the thermocline descends.

At the highest latitudes, the surface temperatures

are much lower than at lower latitudes,

while the deep-water temperatures are little

different. As a consequence, the main

thermocline might not be present at high latitudes,

and only a seasonal thermocline might

occur. In high northern latitudes, there is often

a layer of cold water at 50e100 m (Figure 4.2c),

with temperatures as low as 1.6 C, sandwiched

between the warmer surface and

deeper layers. As described at the beginning of

Section 4.2, this cold layer is referred to as the

dichothermal layer. The warmer surface water

is often just seasonal, and the thermocline overlying

the dichothermal layer is therefore

seasonal.

Figure 4.9 shows the annual range of surface

temperature over the globe. Annual variations

at the surface rise from 1e2 C at the equator

to between 5 and 10 Cat40 latitude in the

open ocean, then decrease toward the polar

regions (due to the heat required in the melting

or freezing processes where sea ice occurs).

Near the coast, larger annual variations

(10e20 C) occur in sheltered areas and in the

western subtropical regions of the Northern

80˚N

Maximum sea surface temperature difference (°C)

0˚ 60˚E 120˚E 180˚ 120˚W 60˚W 0˚

4

80˚N

60˚

8

60˚

40˚

20˚

0˚

6

20˚

4

2

18

10

1

2

4

1

6

4

2

10

8

1

2

4

40˚

20˚

0˚

20˚

40˚

60˚

4

6 6

4

4

2

8

6

60˚

40˚

80˚S

0˚ 60˚E 120˚E 180˚ 120˚W 60˚W 0˚

80˚S

T (°C)

0 2 4 6 8 10 12 14 16 18 20

FIGURE 4.9 Annual range of sea surface temperature ( C), based on monthly climatological temperatures from the

World Ocean Atlas (WOA05) (NODC, 2005a, 2009).


Hemisphere, where the Kuroshio and Gulf

Stream are located and where surface heat loss

is highest (Section 5.5, Figure 5.12). These

annual variations in temperature decrease with

depth and are rarely perceptible below

100e300 m. The maximum temperature at the

surface occurs at the end of the warming season,

in August/September in the Northern Hemisphere,

and the minimum at the end of the cooling

season, during February/March. Below the

surface, the times of occurrence of the maxima

and minima are delayed by as much as two

months relative to the surface.

Diurnal variations of SST had been thought to

be small (<0.4 C) prior to satellite observations.

Such measurements, verified by in situ observations

from a moored buoy in the Sargasso Sea

over a period of two years (Stramma et al.,

1986), have shown that diurnal variations to

1 C are common with occasionally higher

values, up to 3e4 C. The larger diurnal variations

of 1 C or more are observed in conditions

of high insolation (solar radiation) and low

wind speed, and are generally limited to the

upper few meters of water. Similar diurnal variation

has been observed elsewhere in the North

Atlantic and in the Indian Ocean. In sheltered

and shallow waters along the coast, values of

2e3 C are common.

4.2.5. Deep-Water Temperature and

Potential Temperature

Below the thermocline, the temperature

slowly decreases with increasing depth. (This

vertical temperature change is much smaller

than through the thermocline.) In the deepest

waters, temperature can rise toward the bottom,

almost entirely because the high pressure that

compresses the water and raises its temperature

adiabatically (Section 3.3.3, Figure 3.3). To interpret

variations in temperature, even in shallow

waters over a continental shelf as well as from

the surface to thousands of meters, potential

temperature (q) should always be used. Potential

temperature reflects the original temperature of

the water when it was near the sea surface.

An example of this difference between in situ

and potential temperature is shown in Table 4.1

and in Figure 4.10 using data collected in 1976

by the R/V T. Washington from the Mariana

Trench (the deepest trench in the world ocean).

While temperature (T) reaches a minimum at

about 4500 m and thereafter increases toward

the bottom, potential temperature is almost

uniform. (Salinity also is almost uniform

between 4500 m and the deepest observation

as are potential densities relative to any reference

pressure.) Uniform properties from

TABLE 4.1

Comparison of in situ and Potential Temperatures and Potential Densities Relative to the Sea Surface

(s q ), 4000 dbar (s 4 ) and 10,000 dbar (s 10 ) in the Mariana Trench in the Western North Pacific

Depth (m) Salinity (psu) Temperature ( C) q ( C) s q (kg m L3 ) s 4 (kg m L3 ) s 10 (kg m L3 )

1487 34.597 2.800 2.695 27.591 45.514 69.495

2590 34.660 1.730 1.544 27.734 45.777 69.903

3488 34.680 1.500 1.230 27.773 45.849 70.015

4685 34.697 1.431 1.028 27.800 45.898 70.090

5585 34.699 1.526 1.004 27.803 45.904 70.099

6484 34.599 1.658 1.005 27.803 45.904 70.099

9940 34.700 2.266 1.007 27.804 45.904 70.099

Data from R/V T. Washington, 1976.

82


FIGURE 4.10 Mariana Trench:

(a) in situ temperature, T, and

potential temperature, q ( C); (b)

salinity (psu); (c) potential density

s q (kg m 3 ) relative to the sea

surface; and (d) potential density

s 10 (kg m 3 ) relative to 10,000 dbar.

0

2000

Temperature (°C)

0

Salinity

Depth (m)

4000

6000

θ

T

5000

8000

10000

T

0 10 20

10000

0

(a)

0 1 2 3 4 5

(b)

34.5 35.0

2000

Depth (m)

4000

6000

8000

(c)

(d)

10000

26.5 27.0 27.5 28.0 69.0 69.5 70.0 70.5

Potential density -1000 Potential density 10 -1000

4500e9940 m imply that the trench is filled with

water that passes over the sill into the trench,

and that there is no other source of water. The

slight increase in potential temperature with

depth might be due to the weak geothermal heating

acting on this nearly stagnant thick layer.

It is not necessary to go to the deepest part of

the ocean to see the important differences

between in situ temperature and potential

temperature. Through most of the deep ocean,

there is a temperature minimum well above

the ocean bottom, with higher temperature at

the bottom. However, potential temperature

decreases to the ocean bottom almost everywhere.

This is because the densest waters that

fill the oceans are also the coldest, since salinity

variations are mostly too weak to control the

density stratification in the deep waters. There

SALINITY DISTRIBUTION 83

are some limited exceptions to the monotonic

decrease with depth: in localized regions of

densest water formation, at some mid-ocean

ridges where geothermal heating slightly

warms waters right on the ridges, and in the

central South Atlantic where there is significant

vertical salinity variation at mid-depth

(Figure 4.11b).

A global map of potential temperature at the

ocean bottom in the deep ocean (>3500 m

depth) is shown in Chapter 14 (Figure 14.14b).

The bottom temperature distribution is mostly

set by the two sources of bottom water, from

the Antarctic and the Nordic Seas. (Mid-ocean

ridges also result in bottom temperature variations

since they jut upward into warmer

waters.) Bottom waters of Antarctic origin are

the coldest; bottom temperatures are near the

freezing point near Antarctica, with tongues of

water colder than 0 C extending northward

into the deep basins of the Southern Hemisphere.

Bottom waters of northern Atlantic

origin (which arise from overflows from the

Nordic Seas) are considerably warmer with

temperatures around 2 C.

4.2.6. Vertical Sections of Potential

Temperature

We now view potential temperature using

meridional cross-sections through each of the

three oceans (Figures 4.11a, 4.12a, and 4.13a) to

identify common and typical features. Salinity

and potential density sections are also shown

to keep the vertical sections from each ocean

together. The salinity and density distributions

are described in Sections 4.3 and 4.4.

In all oceans, the warmest water is in the

upper ocean with the highest temperatures in

the tropics. In the subtropics, the warm water

fills bowl-shaped regions. These bowls define

the upper ocean circulations, with westward

flow on the equatorward side of the bowls

and eastward flow on the poleward side of

the bowls. Potential temperature decreases

downward through the thermocline into much

more uniform, colder temperatures at depth.

The coldest water is found at the surface at

high latitudes (and is vertically stable because

of low salinity surface water). The coldest water

in these sections is in the Antarctic, since the

northern ends of the sections do not extend

into the Arctic. In the Antarctic, the cold

isotherms slope steeply downward between 60

and 50 S. This marks the eastward flow of the

ACC (Chapter 13).

There are distinct differences in potential

temperature distributions between the Northern

and Southern Hemispheres. The cold surface

waters are much more extensive in the south.

Even the two bowls of higher temperature are

not symmetric; the southern bowl is more extensive

than its northern counterpart. In the deep

part of the Atlantic, Pacific, and Indian Oceans,

the coldest waters are in the south (in the

Antarctic) and the potential temperatures are

slightly higher in the north.

4.3. SALINITY DISTRIBUTION

The mean salinity of the world ocean is 34.6

psu, based on integrating the climatological

data in Java Ocean Atlas (Osborne & Swift,

2009; see the online supplementary materials

located on the Web site for this text). There are

significant differences between the ocean

basins. The Atlantic, and especially the North

Atlantic, is the saltiest ocean and the Pacific is

the freshest (excluding the Arctic and Southern

Ocean, which are both fresher than the Pacific).

These basin differences are illustrated in

Figure 4.14, which shows the mean salinity along

well-sampled hydrographic sections, averaged

zonally, and from top to bottom of the ocean.

Salinity sections from south to north in each

ocean are included in Figures 4.11, 4.12, and

4.13. The following descriptions refer back to

these sections. It is apparent after comparing

salinity, potential temperature, and potential

(a)

0

1000

2000

3000

4000

5000

5000

0

1

Atlantic

Atlantic

θ

Salinity 34.7

34.7 34.7

6000

6000

0 2000 4000 6000 8000 10000 12000 14000 km 0 2000 4000 6000 8000 10000 12000 14000 km

60°S 40°tS 20°S 0° 20°N 40°N 60°N

60°S 40°tS 20°S 0° 20°N 40°N 60°N

(c)

0

1000

2000

3000

4000

5000

27.8

46.1

1

0

1

0

46

2

46.1

1

3

4

2

27.5

5

27

10

3

15 20

60°S 40°tS 20°S 0° 20°N 40°N 60°N

27.8

26

45.7

45.8 45.8

45.9

46

46

2

4

10

5

3

4

5

0

1000

2000

3000

4000

0

1000

2000

3000

4000

5000

200

220

240

180

34.7

180

220

34.3

200

220

34.7

240

220

140

200 180

200

Atlantic

Atlantic

σ θ and σ 4

46.1

Oxygen

6000

6000

0 2000 4000 6000 8000 10000 12000 14000 km 0 2000 4000 6000 8000 10000 12000 14000 km

(b)

(d)

240

34.5

34.9

260

80

240

37

36

220

35

240

34.9

36

34.9

260

220

240

240260

60°S 40°tS 20°S 0° 20°N 40°N 60°N

FIGURE 4.11 (a) Potential temperature ( C), (b) salinity (psu), (c) potential density s q (top) and potential density s 4 (bottom) (kg m 3 ), and (d)

oxygen (mmol/kg) in the Atlantic Ocean at longitude 20 to 25 W. Data from the World Ocean Circulation Experiment. This figure can also be found in

the color insert.

260

280

84



(a)

0

1000

2000

3000

4000

5000

Pacific

θ

6000

0 2000 4000 6000 8000 10000 12000

(c)

0

1000

1

2 2

1

1.5

5

10

15 20

3 3

60°S 40°S 20°S 0° 20°N 40°N

27

27.5

1

26

5

4

1.5

2

km

(b)

1000

0

1000

0

34.6

2000 34.73

3000

4000

34.4

34.65

34.7

34.68

34.5

34.6

34.4

34.65

34.7

5000

Pacific

Salinity

34.7

6000

0 2000 4000 6000 8000 10000 12000

(d)

34.7

34.6

34.6

34.65

34.5

34.4

34.3

34.68

60°S 40°S 20°S 0° 20°N 40°N

180

260

220

180

160

40

100

80

40

34

km

2000

3000

4000

46.05

45.95

45.85

45.9

45.8

45.75

5000

45.95

Pacific

σ θ and σ 4 45.95

45.9

6000

0 2000 4000 6000 8000 10000 12000

60°S 40°S 20°S 0° 20°N 40°N

45.9

45.7

45.8

45.85

2000

3000

4000

200

160

180

80

100

160

5000

180

Pacific

Oxygen

6000

km 0 2000 4000 6000 8000 10000 12000

140

160

180

60°S 40°S 20°S 0° 20°N 40°N

120

140

160

140

km

FIGURE 4.12 (a) Potential temperature ( C), (b) salinity (psu), (c) potential density s q (top) and potential density s 4

(bottom; kg m 3 ), and (d) oxygen (mmol/kg) in the Pacific Ocean at longitude 150 W. Data from the World Ocean Circulation

Experiment. This figure can also be found in the color insert.

density sections for each ocean that the salinity

distribution is more complex than temperature

and density. While potential temperature

decreases monotonically to the bottom in most

places, salinity has marked vertical structure;

from the simplicity of the density field, it is

apparent that it is dominated by potential

temperature. Salinity therefore functions in part

as a tracer of waters, even as it affects density

in a small way.

More detailed depictions of the global

salinity distribution and seasonal changes are

available in the climatological (seasonally

averaged) data set from Levitus, Burgett, and

Boyer (1994b). They also showed the data used

as the basis for the climatologies. There are

far more observations (~90%) in the Northern

Hemisphere than in the Southern Hemisphere

(~10%), and far more observations in summer

than in winter (e.g., Figure 6.13). (This is also

true of temperature observations.) This sampling

bias is rapidly being corrected in the upper

1800 m by the global profiling float program

(Argo) that began in the 2000s.

(a)

0

1000

2000

2

2

3

10

2

15 20 25

5

4

3

(b)

0

1000

2000

34.73

34.3

34.6

34.7

34.5

35.4

34.6

34.9

35

34.8

86

3000

4000

5000

0

1000

2000

3000

4000

27.8

0

46.1

46.0

1

46

45.9

0

27

46.1

5000

46

Indian

s q and s 4

6000

0 2000 4000 6000 8000 km

60°S 50°S 40°S 30°S 20°S 10°S 0° 10°N

27.5

45.8

1

Indian

Indian

q

Salinity

6000

6000

0

2000 4000 6000 8000 km 0 2000 4000 6000 8000 km

60°S 50°S 40°S 30°S 20°S 10°S 0° 10°N 60°S 50°S 40°S 30°S 20°S 10°S 0° 10°N

(c)

45.9

26

27

24

3000

4000

5000

(d)

0

1000

2000

3000

4000

5000

6000

200

0

220

Indian

Oxygen

34.7

220

240

180

40 80 0

200

200

180

180

34.73

240

40 80 0

34.7

220

200

200

180

120

100 80

180

140

140

34.73

0 2000 4000 6000 8000 km

60°S 50°S 40°S 30°S 20°S 10°S 0° 10°N

FIGURE 4.13 (a) Potential temperature ( C), (b) salinity (psu), (c) potential density s q (top) and potential density s 4 (bottom; kg m 3 ), and (d) oxygen

(mmol/kg) in the Indian Ocean at longitude 95 E. Data from the World Ocean Circulation Experiment. This figure can also be found in the color insert.

80

120

160

80 40 0

40

100



Mean salinity (psu)

35.4

35.2

35.0

34.8

34.6

34.4

Mean salinity: 34.83

Indian

Atlantic

Pacific

FIGURE 4.14 Mean salinity,

zonally averaged and from top to

bottom, based on hydrographic

section data. The overall mean

salinity is for just these sections and

does not include the Arctic,

Southern Ocean, or marginal seas.

Source: From Talley (2008).

40°S 30 20 10 0 10 20 30 40 50 60°N

Latitude

4.3.1. Surface Salinity

Surface salinity in the open ocean ranges

from 33 to 37. Lower values occur locally near

coasts where large rivers empty and in the

polar regions where the ice melts. Higher values

occur in regions of high evaporation, such as

the eastern Mediterranean (salinity of 39) and

the Red Sea (salinity of 41). On average, the

North Atlantic is the most saline ocean at

the surface (35.5 psu), the South Atlantic and

South Pacific are less so (about 35.2 psu),

and the North Pacific is the least saline (34.2

psu), which reflects the ocean basin differences

in salinity over the whole ocean depth

(Figure 4.14).

The salinity distribution at the ocean’s surface

is relatively zonal (Figure 4.15), although not as

strongly zonal as sea-surface temperature. Unlike

SST, which has a tropical maximum and polar

Winter surface salinity

80˚N

60˚W

0˚ 60˚E 120˚E 180˚ 120˚W

30

80˚N

40˚

60˚

33

34

35

36

353031

32 34

36.5

37

37.5

33

34.5

32

60˚

40˚

20˚

0˚

32

37

32

34

34.5

36

33

34

35

35

35.5

34

20˚

0˚

20˚

37

36

35

35.5

36

20˚

40˚

60˚

34

35

34

35

34

34

60˚

40˚

80˚S

80˚S

60˚W 0˚ 60˚E 120˚E 180˚ 120˚W

FIGURE 4.15 Surface salinity (psu) in winter (January, February, and March north of the equator; July, August, and

September south of the equator) based on averaged (climatological) data from Levitus et al. (1994b).

88


minima, salinity has a double-lobed structure,

with maxima in the subtropics in both hemispheres

and minima in the tropics and subpolar

regions. This meridional variation is also

apparent in the global zonal average of surface

salinity (Figure 4.3b). In that figure, the salinity

maximum just north of 60 N (with corresponding

density deviation) results from dominance

of the North Atlantic waters over North Pacific

at these latitudes. This is a combination of geography

and the higher overall salinity of the North

Atlantic; as the North Pacific closes off at these

latitudes, the zonal average mainly includes

more saline North Atlantic waters, even though

internally the subpolar North Atlantic waters

are fresher than its subtropical waters.

Surface salinity is set climatologically by the

opposing effects of evaporation (increasing it)

and precipitation, runoff, and ice melt (all

decreasing it), mostly captured by the map of

evaporation minus precipitation (Figure 5.4a).

The meridional salinity maxima of Figures 4.3

and 4.15 are in the trade wind regions and

subtropical high pressure regions where the

annual evaporation (E) exceeds precipitation

(P), so that (E P) is positive. On the other hand,

the surface temperature maximum is near the

equator because the balance of energy into the

sea has a single maximum there. Just north of

the equator, precipitation is high and surface

salinity is lower because of the Intertropical

Convergence Zone (ITCZ) in the atmosphere.

Generally the regions of high positive evaporation

minus precipitation (E P) are displaced

to the east of the subtropical salinity maxima.

This lateral displacement results from the circulation

(advection) of the surface waters, so that

salinity is highest at the downstream end of

the flow of upper ocean waters through the

evaporation maxima.

4.3.2. Upper Layer Salinity

The vertical salinity distribution (Figure 4.16

and sections in Figures 4.11, 4.12, and 4.13) is

more complicated than the temperature distribution.

In the upper ocean, in the tropics, and

subtropics and parts of the subpolar regions,

temperature dominates the vertical stability

(density profile). In the deep ocean, beneath

the pycnocline, temperature also dominates

over salinity. Therefore, warmer water (lower

density) is generally found in the upper layers

and cooler water (higher density) in the deeper

layers. Salinity can have much more vertical

structure, ranging from low to high, without

creating vertical overturn. (In subpolar and

high latitudes, where surface waters are quite

fresh and also cold, salinity does dominate the

vertical stability.) As a consequence of its less

important role in dictating the density structure,

salinity is a more passive tracer than temperature.

Thus, salinity can often be used as a marker

of the flow directions of water masses (minima

or maxima).

In the subtropics, salinity is high near the sea

surface due to subtropical net evaporation.

Salinity decreases downward to a minimum in

the vertical at 600e1000 m. Below this, salinity

increases to a maximum, with the exact depths

of the vertical minimum and maximum depending

on the ocean. In the Atlantic and Indian

Oceans, the salinity maximum is at depths of

1500e2000 m. In the Pacific, the maximum

salinity is at the bottom.

In the tropics and southernmost part of the

subtropical gyres, salinity is often slightly lower

at the sea surface than in the main part of the

subtropics. Salinity increases to a sharp subsurface

maximum at depths of 100e200 m, close to

the top of the thermocline. This maximum arises

from the high salinity surface water in each

subtropical gyre (Figures 4.7, 4.11b, 4.12b,

4.13b, and 4.15). This high salinity water

subducts and flows equatorward and downward

beneath the fresher, warmer tropical

surface water, thus forming a salinity maximum

layer. This shallow salinity maximum is found

in the equatorward part of every subtropical

gyre, merging into the tropics. Because it has an


(a) (b) (c)

Depth (m)

0

500

1000

Salinity

34 35

5°N, 148°E

Salinity

34 35

Eastern

24°N, 147°W

Western

24°N, 147°E

Salinity

33 34

Eastern

47°N, 137°W

Western

47°N, 162°E

FIGURE 4.16 Typical

salinity (psu) profiles for

the tropical, subtropical,

and subpolar regions of

the North Pacific. Corresponding

temperature

profiles are shown in

Figure 4.2.

1500

2000

Low

latitude

(tropical

N. Pacific)

Mid

latitude

(subtropical

N. Pacific)

High latitude

(subpolar N. Pac.)

identifiable characteristic (salinity maximum)

and common formation history (subduction

from the high salinity surface water at midlatitudes),

it has acquired status as a water

mass. Several names are used for this water

mass. Our preference is Subtropical Underwater,

following Worthington (1976). It is also referred

to as “salinity maximum water.”

Low salinity layers also result from subduction,

in this case from the fresher but denser

northern outcrops of the subtropical gyres.

Advection of these waters southward results

in subduction and a low salinity layer that is

found around the eastern and into the southern

side of the anticyclonic gyre. In the North and

South Pacific, these are extensive features called

the Shallow Salinity Minimum in each ocean

(Reid, 1973). In the subpolar North Atlantic,

there is a much less-extensive shallow salinity

minimum associated with the subarctic front

(part of the North Atlantic Current); it is called

Subarctic Intermediate Water.

In subpolar and high-latitude regions, with

high precipitation, runoff, and seasonal ice

melt, there is generally low salinity at the

sea surface. The halocline, with a rapid downward

increase of salinity, lies between the

surface low-salinity layer and the deeper,

saltier water. In such regions, the pycnocline

is often determined by the salinity distribution

rather than by temperature, which

remains relatively cold throughout the year,

and may have only a weak thermocline or

even none at all. This condition, associated

with runoff and precipitation, occurs throughout

the subpolar North Pacific. In the Arctic

and Antarctic and other regions of sea ice

formation, ice melt in spring creates a similarly

freshened surface layer.

This low salinity surface layer in regions

like the subpolar North Pacific and around

Antarctica permits a vertical temperature

minimum near the sea surface, with a warmer

layer below (the dichothermal and mesothermal

layers, described in Section 4.2).

4.3.3. Intermediate Depth Salinity

At intermediate depths (around 1000e1500 m)

in many regions of the world, there are horizontally

extensive, vertically broad layers of

either low salinity or high salinity. These are

easily identified in Figures 4.11, 4.12, and 4.13

because of their vertical salinity extrema. In the

North Pacific and Southern Hemisphere, the

salinity minimum layer is at about a depth of

90


1000 m. The subpolar North Atlantic salinity

minimum is at about a depth of 1500 m. The

low salinity layers are located near the base of

the pycnocline, with temperatures of 3e6 C.

The two major intermediate-depth salinity

maximum layers are in the North Atlantic

and northern Indian Ocean (not to be confused

with the deeper salinity maximum associated

with North Atlantic Deep Water; NADW). They

are considerably warmer than the low salinity

intermediate waters. The vertical salinity

extrema reflect specific formation processes,

described briefly here and in more detail in later

chapters. These layers are therefore labeled as

water masses and called the “intermediate

waters.”

A map of the locations of the major intermediate

water masses is provided in Chapter 14

(Figure 14.13). Their low salinity and their

temperature ranges indicate that they originate

at the sea surface at subpolar latitudes

where surface waters are relatively fresh, but

where surface waters are warmer than freezing.

The North Pacific Intermediate Water (NPIW) originates

in the northwest Pacific and is found

throughout the North Pacific. Labrador Sea Water

(LSW) originates in the northwest Atlantic and

is found through the North Atlantic. LSW is

also marked by high oxygen and chlorofluorocarbons,

and retains these signatures even as it

loses its salinity minimum as it becomes part

of the NADW in the tropical and South Atlantic.

Antarctic Intermediate Water (AAIW) originates

in the Southern Ocean near South America

and is found throughout the Southern Hemisphere

and tropics. In these three ventilation

regions, surface salinity is lower but density is

higher than the upper ocean and thermocline

waters in the subtropics and tropics. The

ventilated intermediate waters spread equatorward

and carry their low salinity signature with

them.

The two major salinity maximum intermediate

waters result from high salinity outflows

from the Mediterranean and Red Seas. The

source of these high salinity waters is surface

inflow into these seas; high evaporation within

the seas increases the salinity and cooling

reduces their temperature, thus dense water is

formed. When these saline, dense waters flow

back into the open ocean, they are dense enough

to sink to mid-depths.

Other, more local, intermediate waters are

also identified by vertical salinity extrema. For

instance, in the tropical Indian Ocean, a middepth

salinity minimum originates from fresher

Pacific Ocean water that flows through the Indonesian

Passages (Chapter 11). This intermediate

salinity minimum has been called Indonesian

Intermediate Water or Banda Sea Intermediate

Water (Rochford, 1961; Emery & Meincke,

1986; Talley & Sprintall, 2005).

Each of these intermediate waters is discussed

in greater detail in the relevant ocean

basin chapter (9e13).

4.3.4. Deep-Water Salinity

The deep waters of the oceans exhibit

salinity variations that mark their origin. The

North Atlantic is the saltiest of all of the oceans

at the sea surface, so dense waters formed in

the North Atlantic carry a signature of high

salinity as they move southward into the

Southern Hemisphere and then eastward and

northward into the Indian and Pacific Oceans.

This overall water mass is referred to as North

Atlantic Deep Water. Dense waters formed in

the Antarctic are colder and denser than North

Atlantic dense waters, so they are found

beneath waters of North Atlantic origin. The

dense Antarctic waters are also fresher than

North Atlantic waters; their progress northward

into the Atlantic can be tracked through

their lower salinity, where they are referred to

as Antarctic Bottom Water (AABW). The vertical

juxtaposition of the salty NADW and fresher

AABW is apparent in the Atlantic vertical

salinity section (Figure 4.11b). This NADW/

AABW structure is also apparent in the


southern Indian Ocean since both NADW and

AABW enter the Indian Ocean from the south

(Figure 4.13b).

The northern Indian Ocean is tropical so no

dense waters are formed there, but the high

salinity from the intermediate waters of the

Red Sea penetrates and mixes quite deep,

making northern Indian Ocean deep waters

relatively saline (Figure 4.13b). The North

Pacific does not form dense, abyssal waters

because the sea surface in the subpolar North

Pacific is too fresh to allow formation of waters

as dense as those from the Antarctic and North

Atlantic. Therefore, the salinity structure in the

deep North Pacific is determined by the inflow

of the mixture of Antarctic and North Atlantic

deep waters from the south; this mixture is

more saline than the local North Pacific waters

so salinity increases monotonically to the

bottom in the North Pacific (Figure 4.12b).

A global map of bottom salinity is shown in

Chapter 14 (Figure 14.14c). Globally, the salinity

variation in the deep waters is relatively small,

with a range from 34.65 to 35.0 psu. Like bottom

temperatures, the bottom salinities reflect the

Antarctic and Nordic Seas origins of the waters.

The Antarctic bottom waters are freshest, with

salinities lower than 34.7 psu. The bottom

waters of Nordic Sea origin are the saltiest,

with salinity up to 35.0 psu. Full interpretation

of the bottom salinity map also requires consideration

of the varying bottom depth d as ridges

cut up into overlying deep waters d and of

downward diffusion of properties from the

overlying deep waters, which are beyond the

scope of this book.

Thus both the deep water temperature and

deep water salinity have small ranges. The

deep water environment is relatively uniform

in character compared with the upper ocean

and thermocline and even the intermediate

layer. This relative uniformity is the result of

the small number of distinct sources of dense

waters, and the great distance and time that

these waters travel, subjected to mixing with

each other and to downward diffusion from

layers above them.

4.3.5. Temporal Variations of Salinity

Salinity variations at all timescales are less

well documented than temperature variations,

because temperature is more easily measured.

Annual variations of surface salinity in the

open ocean are less than 0.5 psu. In regions

of marked annual variation in precipitation

and runoff, such as the eastern North Pacific

and the Bay of Bengal and near sea ice, there

are large seasonal salinity variations. These

variations are confined to the surface layers

because in such regions the effect of reduced

salinity overrides the effect of temperature in

reducing the seawater density. This keeps the

low salinity water in the surface layer. Diurnal

variations of salinity appear to be small, but

againthisisaconclusionbasedonveryfew

observations. Local rainstorms produce fresh

surface waters even in the open ocean that

mix into the surrounding waters after several

weeks.

Temporal salinity variations at a given location

can be large at large-scale fronts between

waters of different properties. These fronts

are sometimes termed water mass boundaries.

Temperature variations can also be quite large

across these fronts. The fronts move about their

mean location, on weekly, seasonal, and longer

timescales. Meandering of the fronts and creation

of eddies of the different types of waters

can cause large salinity and temperature variations

at a given location.

Interannual and long-term changes in largescale

salinity are observed and are part of the

documentation of climate change. With the

advent of the global profiling float array, it is

becoming possible to document salinity

changes in all regions of the non-ice-covered

ocean; significant patterns of surface salinity

change have already been detected (Hosoda,

Suga, Shikama, & Mizuno, 2009; Durack &

92


Wijffels, 2010). Salinity variations in the North

Atlantic and Nordic Seas are associated with

changes in mixed layer convection and with

changesinwatermassformationintheLabrador

and Greenland Seas (Chapters 9 and 12).

LSW has dramatic decadal salinity variations

that correspond with changes in its formation.

See Figure S15.4 (Yashayaev, 2007) from the

online supplementary material. Decades long

freshening of the subpolar North Atlantic

and Nordic Seas (see Chapter S15 from the

online supplementary material), followed by

salinification in recent years, has caused

much interest in terms of NADW production

rates. Large-scale, coherent salinity changes

over several decades have been documented

(Boyer, Antonov, Levitus, & Locarnini, 2005;

Durack & Wijffels, 2010) and can be associated

with large-scale changes in precipitation and

evaporation that might be related to overall

warming of the atmosphere (Bindoff et al.,

2007).

4.3.6. Volumetric Distribution

of Potential Temperature and Salinity

A classic (and typical) approach to looking at

the water mass structure is to display various

properties as a function of each other; a more

modern statistical approach describes the water

masses in terms of all of their properties (see

Section 6.7). Potential temperature-salinity diagrams

are used throughout the basin chapters

(9e13) to illustrate the water masses. A volumetric

q-S diagram from Worthington (1981) is

introduced as our first global summary of

water properties (Figure 4.17). The method is

described in Section 6.7.2.

The underlying q-S in the upper panel

(Figure 4.17a) shows three separate branches

stretching from low q-S to higher q-S; these are

the Central Waters of the pycnocline (as in

Figure 4.7). The saltiest branch is the North

Atlantic; the freshest branch is the North Pacific.

The intermediate branch, with larger volumes,

is the three Southern Hemisphere basins (South

Atlantic, South Pacific, and Indian). The importance

of the Southern Ocean connection between

these latter three basins is immediately

apparent, as the three have properties that are

more similar than the two Northern Hemisphere

basins.

In the deep water (Figure 4.17b), the largest

peak is the Pacific Deep Water (or Common

Water); the large volume in a single q-S class

indicates how well mixed this water mass is,

which is a direct result of its great age (Section

4.7). The coldest waters are the AABW, with

the single ridge again indicating Southern

Ocean circumpolar connectivity. Above about

0 C, the diagram splits into three branches d

the Pacific Ocean, Indian/Southern Ocean, and

Atlantic Ocean, from freshest to saltiest. The

salty Atlantic ridge has a long portion of high

volume, without a huge, single peak such as is

found in the Pacific. This reflects the multiple

sources of NADW and its relatively young,

unmixed character.

4.4. DENSITY DISTRIBUTION

Potential density must increase with depth in

a system in equilibrium. To be more precise, the

water column must be statically stable, using the

definition of static stability (Eq. 3.9) in Section

3.5.6. This means that potential density, using

a local reference pressure, must increase with

depth. While potential temperature and salinity

together determine density, individually they can

have maxima and minima in thevertical,aslong

as the water density increases with depth. The

only exceptions to the monotonic increase occur

at very short timescales, on the order of hours or

less, which is the timescale for overturn. As soon

as denser water flows over lighter water, or surface

layer density is increased above that of the underlying

water, the water column becomes unstable

and will overturn and mix, removing the

instability.

DENSITY DISTRIBUTION 93

FIGURE 4.17 Potential temperature-salinity-volume

(q-S-V) diagrams

for (a) the whole water

column and (b) for waters colder

than 4 C. The shaded region in (a)

corresponds to the figure in (b).

Source: From Worthington (1981).

In Chapter 3, we discussed the use of different

reference pressures for reporting potential

density, or equivalently for use of an empirically

defined type of density such as neutral density

(Section 3.5.4). The potential density that is used

should best approximate the local vertical

stability and isentropic surfaces. Profiles of potential

density relative to both the sea surface and

4000 dbar are used in constructing the potential

density sections of Figures 4.11, 4.12, and 4.13.

When spatial variability in temperature and

salinity is very small, any type of potential density

will increase monotonically with depth; an

example is the potential density relative to both

the sea surface and 10,000 dbar in the Mariana

Trench (Figure 4.10). The North Pacific has little

variation in temperature and salinity below the

pycnocline, which is the vertical region of large

94


(a)

0

1000

2000

3000

4000

27.85

27.88

27.8

27.5

27.88

27.85

27

27.88

27.8

27.5

26

27.88

27.8

27.5

5000

5000

Atlantic

σ θ

Atlantic

γ Ν

6000

6000

0 2000 4000 6000 8000 10000 12000 14000 km 0 2000 4000 6000 8000 10000 12000 14000 km

60°S 40°S 20°S 0° 20°N 40°N 60°N

(b)

0

1000

2000

3000

4000

28

28.2

28.3

28.4

28.2

28.3

28.2

27.9

28.1

27

27.5

60°S 40°S 20°S 0° 20°N 40°N 60°N

27.8

28

26

28.1

FIGURE 4.18 (a) Potential density s q (kg m 3 ) and (b) neutral density g N in the Atlantic Ocean at longitude 20 to 25 W.

Compare with Figure 4.12c. Data from the World Ocean Circulation Experiment.

density change. Therefore all potential density

choices yield stable-appearing vertical profiles

(Figure 4.20).

In the South Atlantic, on the other hand, there

are large-scale salinity inversions where the

saline NADW is layered between the fresher

AAIW and AABW (Figure 4.11b). Here the

differences in compressibility of warmer and

cooler waters begin to matter. Figure 4.11c

emphasizes the local potential density structure,

which is decidedly stable in the vertical. To illustrate

the main drawback of using a surface reference

pressure for deep-water density, a vertical

section through the Atlantic Ocean of potential

density relative to the sea surface, s q , for the

full water column is shown in Figure 4.18a.

There is a large-scale inversion of s q in the South

Atlantic, most pronounced just south of the

equator at a depth of about 3700 m. This is the

base of the high salinity NADW layer. Potential

temperature contours are compressed below the

NADW (Figure 4.11a). Potential density referenced

to 4000 dbar, s 4 , hence locally referenced,

has no inversion (Figure 4.11c).

Neutral density g N (Section 3.5.4; Jackett &

McDougall, 1997) is commonly used to represent

the stable increase of “potential” density

with depth. 1 Like choosing appropriate locally

referenced potential densities, neutral density

eliminates the apparent density inversions of

Figure 4.18a and also removes the need to use

multiple pressure reference levels such as in

the use of 0 and 4000 dbar references in Figures

4.11c, 4.12c, and 4.13c. The neutral density g N

section for the Atlantic is clearly monotonic,

with g N increasing from top to bottom. The

deep contours resemble those of s 4 (Figure 4.11c),

and the distortions of s q in the region of the

Mediterranean salinity maxima at about 2000

m in the North Atlantic are removed.

4.4.1. Density at the Sea Surface and in

the Upper Layer

The density of seawater at the ocean surface

increases from about s q ¼ 22 kg/m 3 near the

equator to s q ¼ 26 e 28 kg/m 3 at 50e60 latitude,

and beyond this it decreases slightly (Figures 4.3

1 There continues to be energetic discussion of the most appropriate variable for density for constructing the most isentropic

surfaces in the sense of the direction of motion of water parcels and the directions of along-isopycnal and diapycnal

mixing.


Winter surface density

80˚N

60˚W

0˚ 60˚E 120˚E 180˚ 120˚W

80˚N

20˚

0˚

20˚

40˚

40˚

60˚

21

22

25

26

60˚

25

24

26

80˚S

24

25

27

27.5

21

60˚W

21

22 23 24

25

23

27.5

28

26.5 27

25

22

26

24

26.5

21

27

23

23

20 21

22

24

26.5

22

23

27.5

0˚ 60˚E 120˚E 180˚ 120˚W

FIGURE 4.19 Surface density s q (kg m 3 ) in winter (January, February, and March north of the equator; July, August, and

September south of the equator) based on averaged (climatological) data from Levitus and Boyer (1994) and Levitus et al.

(1994b).

22

26

25

23

80˚S

25

60˚

60˚

40˚

40˚

20˚

0˚

20˚

and 4.19), due to lower salinity at higher latitudes.

Surface densities at high latitudes in the

Antarctic and North Atlantic are higher than in

the North Pacific even at the freezing point.

North Pacific surface water must be less dense

since its surface water is fresher.

In Figure 4.3, we see that the surface density

averaged for all oceans follows surface temperature

rather than surface salinity in the tropics

and mid-latitudes. At the highest northern and

southern latitudes, poleward of 50 , surface

density follows salinity more than temperature,

because temperature is close to the freezing

point there, with little variation in latitude.

Surface density and the vertical stratification

determine the depth to which surface waters

will sink as they move away from their ventilation

(“outcrop”) region. The combination of

surface temperature and salinity for a given

density also affects the sinking because of their

effect on compressibility, with warmer, saltier

water compressing less than colder, fresher

water at the same surface density. Thus the

colder, fresher parcel will become more dense

and, consequently, deeper than the warmer,

saltier parcel as they move into the ocean even

though they start with the same surface density.

See Section 3.5.4.

In late winter, surface waters reach their local

density maximum as the cooling season draws to

a close. (Cooling in many regions is also associated

with evaporation, so both temperature and

salinity may change together to create dense

water, depending on the local amount of precipitation.)

Late winter density is associated with

the deepest mixed layers. As the warming season

begins (March in the Northern Hemisphere,

September in the Southern Hemisphere), the

dense winter mixed layer is “capped” by warmer

water at the surface. The capped winter waters

move (advect) away from the winter ventilation

region. If they move into a region where the

96


winter surface waters are less dense, they sink

beneath the local surface layers, and will not be

reopened to the atmosphere during the next

winter. This subduction process is a primary

mechanism for moving surface waters into the

ocean interior (Luyten et al., 1983; Woods, 1985

and Sections 4.2.3 and 7.8.5).

Longer timescale variations in surface

density can affect the amounts of intermediate

and deep waters that form and the overall size

of the regions impacted by them. During major

climate changes associated with glacial/interglacial

periods, surface density distributions

must have been strongly altered, resulting in

very different deep water distributions.

Winter mixed layer depths vary from tens of

meters to hundreds of meters, depending on

the region (Figure 4.4). Because they have

usually been detected using temperature

criteria, we discussed mixed layers in some

detail in Section 4.2.2. In the tropics, winter

mixed layer depths may be less than 50 m.

Winter mixed layer depths are greatest in the

subpolar North Atlantic, reaching more than

1000 m in the Labrador Sea, and in the Southern

Hemisphere around the northern edge of the

major current that circles Antarctica at a latitude

of about 50 S, reaching up to about 500 m

thickness.

4.4.2. Pycnocline

Like potential temperature, potential density

does not increase uniformly with depth

(Figure 4.20). The vertical structure of density

is similar to that of potential temperature. There

is usually a shallow upper layer of nearly

uniform density, then a layer where the density

increases rapidly with depth, called the pycnocline,

analogous to the thermocline (Section

4.2.3). Below this is the deep zone where the

density increases more slowly with depth

(Figures 4.10, 4.11, 4.12, 4.18, and 4.20). There

is much smaller variation with latitude of the

deep-water density compared with upper ocean

density. As a consequence, in high latitudes,

where the surface density rises to s q ¼ 27 kg/m 3

or more, there is a smaller increase of density

FIGURE 4.20 Typical density/

depth profiles for low and high

latitudes (North Pacific).

(a)

(b)

21 22 23 24 25 26 27 28 40 41 42 43 44 45 46

0

1 1

2 3 2 3

1000

Depth (m)

2000

(1) 47°N, 170°E

(2) 30°N, 149°E

(3) 5°N, 150°E

3000

4000

21 22 23 24 25 26 27 28 40 41 42 43 44 45 46

Potential density s q -1000 Potential density s 4 -1000


with depth than in the low latitudes, and the

pycnocline is much weaker.

The double thermocline structure that occurs

in some broad regions (described in Section 4.2),

is mirrored in density because of the strong

dependence of ocean density on temperature.

Layers of lower vertical density gradients are

called pycnostads.

In all regions, there is a seasonal pycnocline

in the warm seasons. This results from seasonal

warming and/or ice melt, overlying the

remnant of the winter mixed layer, which forms

a pycnostad in non-winter seasons. A permanent

double pycnocline, with a pycnostad lying

between the pycnoclines, is a common feature of

subtropical regions. Mode Waters (Section 4.2.3)

are pycnostads, and are best identified in terms

of density stratification rather than temperature

stratification; that is, a minimum in vertical

stability is the best identifier of a Mode Water

on a given vertical profile. Often Mode Waters

and other water masses are tracked in terms of

their potential vorticity (Eq. 3.11 and Section 7.7);

the dominant contribution to potential vorticity

in most of the ocean (except in strong currents)

is proportional to the vertical stability. Potential

vorticity is a useful tracer because it is a

conserved dynamical quantity in the absence

of mixing.

4.4.3. Depth Distribution of Potential

Density

Potential density structure is simpler than

potential temperature and salinity simply

because the water column must be vertically

stable. Potential density, appropriately defined,

must increase downward. Below the pycnocline,

vertical potential density variations are

much smaller, similar to potential temperature

structure. There are no large-scale inversions in

density if the appropriate reference pressures

are used, as described in Section 3.6 and as

seen in the vertical section through the Atlantic

Ocean (Figure 4.18 compared with Figure 4.11c

in Section 4.2.6). Horizontal variations in

potential density are associated with horizontal

pressure gradients and therefore with largescale

currents (Section 7.6).

Potential density structure is displayed

along vertical sections through the north-south

length of each ocean (Figures 4.11, 4.12, and

4.13). The main features are downward bowls

in the upper to intermediate ocean in the

subtropics, and a large upward slope toward

the southern (Antarctic) end of the sections.

Below about 2000 m, the total range of potential

density is small, from about s q ¼ 27.6 to 27.9

kg/m 3 (or s 4 ¼ 45.6 to 46.2 kg/m 3 ,whichis

potential density relative to 4000 dbar).

Because mixing is greatly inhibited by vertical

stratification, there is a strong preference

for flow and mixing to occur along isentropic

surfaces, which are approximately isopycnals

(surfaces of constant potential density). In the

upper ocean, surfaces of constant s q are useful.

For instance, the processes that give ocean waters

their particular properties act almost exclusively

at the surface, and the origin of even the deepest

watercan be tracedbacktoaregion offormation at

the surface somewhere. Because deep ocean

water is of high density, it must form at high latitudes

where cold, dense water is found at the

surface. After formation, it spreads down almost

isopycnally (reference pressures should be

adjusted to account for temperature dependence

of the compressibility). The sinking is combined

with horizontal motion so that the water actually

moves in a direction only slightly inclined to the

horizontal. Even in the regions of largest isopycnal

slopes, for instance in the southern part of

Figures 4.11c, 4.12c, and 4.13c, the slopes are no

more than several kilometers down over several

hundred kilometers horizontally.

Even though there is large-scale structure in

the deep ocean’s salinity field (e.g., in the

Atlantic in Figure 4.11b), temperature dominates

the density structure in the deep ocean.

Salinity is important for the density structure

near the sea surface at high latitudes where

98


precipitation or ice melt creates a low salinity

surface layer, such as in the Arctic, in the region

next to Antarctica, and in the subpolar North

Pacific and coastal subpolar North Atlantic. In

shallow coastal waters, fjords, and estuaries,

salinity is often the controlling factor in determining

density at all depths, while the temperature

variations are of secondary importance

(e.g., Table 3.1).

In much of the ocean, the density profile with

increasing depth appears nearly exponential,

asymptoting to a nearly constant value in the

deep ocean (Figure 4.20). However, in some

regions, where deep waters from very different

sources are juxtaposed, there is a weak pycnocline

(higher density variation) between them.

An obvious example is between the NADW

and AABW in the South Atlantic, which is

where we illustrated the necessity for local referencing

of the potential density. Such regions are

most common in the subtropical Southern

Hemisphere where dense Antarctic waters

flow northward in a thick layer under slightly

less dense deep waters flowing southward

from the North Atlantic, Pacific, and Indian

Oceans (Figures 4.11 and 4.18).

4.5. DISSOLVED OXYGEN

Seawater contains dissolved gases, including

oxygen and carbon dioxide. Some of the transient

tracers are dissolved gases (Section 4.6).

The ocean is an important part of the global

(atmosphere/land/ocean) cycle of carbon

dioxide, which is a greenhouse gas. However,

because of its complexity, we do not describe

the ocean’s carbon chemistry in this book, and

instead refer readers to texts on biogeochemical

cycles (e.g., Broecker & Peng, 1982; Libes, 2009).

Dissolved oxygen content is used as an

important tracer of ocean circulation and an

indicator of the time passed since a water parcel

was at the sea surface (ventilated) (Section 3.6).

The range of oxygen values found in the sea is

from 0 to 350 mmol/kg (0 to 8 ml/L), but a large

proportion of values fall within the more limited

range from 40 to 260 mmol/kg (0 to 6 ml/L). The

atmosphere is the main source of oxygen dissolved

in seawater. At the sea surface, the water

is usually close to saturation. Sometimes, in the

upper 10e20 m, the water is supersaturated

with oxygen, a by-product of photosynthesis

by marine plants. Supersaturation also occurs

near the sea surface if the water warms up

through solar radiation that penetrates tens of

meters into the ocean. (If the actual sea surface

becomes warmer, it will lose its excess oxygen

to the atmosphere, so supersaturation is not

found right at the sea surface.) Sometimes

surface waters are undersaturated; this occurs

rarely in winter if mixing of the surface layer

is especially intense, entraining underlying

older, less saturated waters. (The equilibration

time of surface waters d time required to

restore the waters to 100% saturation d is

a few days to a few weeks and is a function of

wind speed and temperature.) Below the

surface layers, the oxygen saturation is less

than 100% because oxygen is consumed by

living organisms and by the bacterial oxidation

of detritus. Low values of dissolved oxygen in

the sea are often taken to indicate that the water

has been away from the surface for a long time,

the oxygen having been depleted by the biological

and detrital chemical demands.

Figure 4.21 shows typical dissolved oxygen

profiles for the Atlantic and the Pacific for three

latitude zones. Figures 4.11d, 4.12d, and 4.13d

show oxygen along a south-north section for

each ocean. Common features of the Atlantic

and Pacific are (1) high oxygen close to the surface,

(2) an oxygen minimum at 500e2500 m, (3) relatively

high values below 1500 m in the Atlantic

(NADW), (4) low values in the North Pacific

beneath the surface layer, and (5) more similar

subsurface distributions in the southern latitudes

in both oceans. Distributions in the Indian

Ocean are similar to those in the Pacific (south

and tropics). The lower values in the deep water

NUTRIENTS AND OTHER TRACERS 99

(a) (b) (c)

Oxygen (mmol/kg)

0 100 200 300 0 100 200 300 0 100 200 300

0

FIGURE 4.21 Profiles of dissolved

oxygen (mmol/kg) from the Atlantic

(gray) and Pacific (black) Oceans.

(a) 45 S, (b) 10 N, (c) 47 N. Data

from the World Ocean Circulation

Experiment.

1000

Depth (m)

2000

3000

PAC

45°S

ATL

45°S

PAC

10°N

ATL

10°N

4000

PAC

47°N

ATL

47°N

5000

of the Pacific compared with the Atlantic indicate

that the Pacific water has been away from the

surface for a much longer time. In some regions

of extremely low oxygen, such as the Black Sea

and the bottom of the Cariaco Trench (off Venezuela

in the Caribbean), hydrogen sulfide is

present, created from the reduction of sulfate

ion by bacteria. This indicates that the water

has been stagnant there for a long time.

The oxygen minimum through the world

oceans at mid-depth, overlying higher oxygen

at the bottom, results from at least several mechanisms.

One is that minimal circulation and mixing

do not replace the oxygen consumed. A

second is that the increase of density with depth

(stability) allows biological detritus to accumulate

in this region, which increases the oxidation

rate. A third is that the bottom waters in each of

the oceans are relatively high in oxygen because

of their surface source in the Antarctic. In the

Pacific and Indian Oceans, a three-layer structure

is obtained, with high oxygen at the surface

decreasing through the pycnocline, an oxygen

minimum layer in the intermediate and deep

water, and higher oxygen in the abyss. The

Atlantic has a four-layer structure because of

the juxtaposition of high oxygen content in the

NADW onto this three-layer structure (the thick

layer of higher oxygen between 2000 and 4000

dbar in Figure 4.11d corresponds with high

salinity in Figure 4.11b).

A pronounced vertical oxygen minimum is

found in the upper ocean in the tropical Atlantic

(Figure 4.11d), eastern tropical Pacific

(Figure 4.12d), and in the northern Indian Ocean

(Figure 4.13d). These shallow oxygen minima

result from very high biological productivity

in the surface waters in these regions. Bacterial

consumption of the large amount of sinking

detritus from these surface waters is large and

consumes almost all of the dissolved oxygen

within the upper 300e400 m of the ocean.

The production and utilization of oxygen in

the sea are essentially biogeochemical matters

(Section 3.6). Oxygen is a useful tracer, broadly

indicative of a water parcel’s age, but since

it is non-conservative, it must be used carefully.

4.6. NUTRIENTS AND OTHER

TRACERS

Other common water properties used as flow

tracers or for identifying water masses include

100


the nutrients (phosphate, nitrate, nitrite, silicic

acid, and ammonium), dissolved gases other

than oxygen and carbon dioxide, and plankton,

which are small organisms (both plant and

animal) that drift with the water. These water

mass characteristics must be used with caution

because, like oxygen, they are non-conservative;

in other words, they may be produced or

consumed within the water column. Other

chemical and radioactive tracers are now

measured widely (e.g. Broecker and Peng,

1982). Here we concentrate on the main features

of the nutrient distributions, which are of

interest to marine biologists as well as physical

and chemical oceanographers.

Nutrient values are low in the upper few

hundred meters with higher values in the deeper

water (Figure 4.22). In the North Pacific, these

deeper distributions are in the form of mid-depth

maxima, extending from north to south, with

highest values at 1000/2000 m for nitrate (NO 3 )

and phosphate (PO 4 ), and at 2000/3000 m for

silicate. Additionally, there are maxima in the

south in the dense water formed in the Antarctic.

In the Atlantic, the mid-depth low nutrient

tongues extending from north to south are associated

with the NADW (Section 9.8). Maximum

values are found in the south and along the

bottom in the dense waters formed in the

Antarctic (AABW).

The low values of nutrients in the upper

layers result from utilization by phytoplankton

in the surface layer (euphotic zone, exposedto

sunlight), while the increase in deeper waters

is because of their release back to solution by

biological processes (respiration and nitrification,

mostly microbial) during the decay of

detrital material sinking from the upper layers.

Therefore, nutrient distributions are approximately

mirror images of the oxygen distribution.

Phosphate and nitrate have similar

distributions because they are governed by

almost the same biological cycle (see discussion

of Redfield ratios in Section 3.6). (Therefore

only nitrate sections are included in

Figure 4.22.) The dissolved silica (silicic acids)

distribution is not as closely similar. Silica has

an additional source at the ocean bottom, as

it can be dissolved into the seawater from

the sediments or injected by hydrothermal

sources.

Nutrient replenishment in the surface layers

is strongly influenced by the physical processes

of vertical diffusion, overturning, and upwelling.

These bring nutrients from below the

euphotic zone up to the surface. The impact of

upwelling on surface nutrients is illustrated by

the nitrate distribution at the sea surface

(Figure 4.23). Nitrate is nearly zero in the

subtropical regions where surface waters downwell

(Chapters 9e11). Surface nitrate is non-zero

(although small) where there is upwelling from

just below the euphotic zone, which occurs in

subtropical eastern boundary regions (Section

7.9.1), along the equator, and in the subpolar

regions. These are regions of high biological

productivity because of the nutrient supply to

the euphotic zone (see map of depth of the

euphotic zone in Figure 4.29).

At mid-depth, in the nutrient vertical maxima,

the Pacific nutrients are higher than Atlantic

values by a factor of about two for phosphate

and nitrate, and by a factor of three to ten for silicate.

This is due to the much greater age of the

mid-depth and deep waters in the Pacific than

in the North Atlantic. The lower dissolved

oxygen values in the Pacific than in the Atlantic

are attributed to the same cause.

Taken together, the oxygen and nutrient

distributions, along with salinity, provide our

principal identification for water masses below

the pycnocline. The high oxygen, low nutrient,

high salinity deep layer in the Atlantic Ocean

is the NADW. The low oxygen, high nutrient

layer in the Pacific Ocean is the Pacific Deep

Water, and the same layer in the Indian Ocean

is the Indian Deep Water. The higher oxygen,

lower nutrient, very cold bottom layer in all

oceans is the AABW. When considering more

carefully the east-west distributions of

(a)

0

1000

2000

3000

4000

130

100

5000

30

5000

Atlantic

120

Atlantic

20

Nitrate

6000

6000

Silticate

0 2000 4000 6000 8000 10000 12000 14000 km 0 2000 4000 6000 8000 10000 12000 14000 km

(c)

0

1000

2000

3000

4000

32.5

32.5

32.5

32.5

32.5

5000

Pacific

35

Nitrate

6000

0 2000 4000 6000 8000 10000 12000 km

60°S 40°S 20°S 0° 20°N 40°N

(e)

0

1000

2000

3000

35

30

30

32.5

25

32.5

25

4000

32.5

-10

20 30 1

10 20 30 1

32.5 -10

20 30 1

5000

Indian

Nitrate

6000

0 2000 4000 6000 8000 km

60°S 50°S 40°S 30°S 20°S 10°S 0° 10°N

20

20

35

35

32.5 30

25

20

60°S 40°S 20°S 0° 20°N 40°N 60°N

10 1

20 35

10

20

40

32.5

35

1

32.5

30

30

35

32.5

35

32.5

37.5

10

42.5

1

20

40 42.5

35

1

20

NUTRIENTS AND OTHER TRACERS 101

30

40

37.5

35

10

15

37.5

0

1000

2000

3000

4000

0

1000

2000

3000

4000

5000

6000

0

1000

2000

3000

4000

120

120

130

120

100

Pacific

Silicate

80

120

60

80

5

10

60°S 40°S 20°S 0° 20°N 40°N 60°N

100

5

10

80

40

60

120

130

130

10 20 40 60 80 50

50

100 120 40 80 60

5

5000

Indian

Silicate

6000

0 2000 4000 6000 8000 km

60°S 50°S 40°S 30°S 20°S 10°S 0° 10°N

40

60

140

0 2000 4000 6000 8000 10000 12000 km

60°S 40°S 20°S 0° 20°N 40°N

130

50

80

100

120

130

60

80

120

100

5

10

20

40

80

100

120

140

20

80

100

130

150

150

130

40

40

160

150

20

10

20

170

5 20

40

FIGURE 4.22 Nitrate (mmol/kg) and dissolved silica (mmol/kg) for the Atlantic Ocean (a, b), the Pacific Ocean (c, d), and

the Indian Ocean (e, f). Note that the horizontal axes for each ocean differ. Data from the World Ocean Circulation

Experiment. This figure can also be found in the color insert.

(b)

(d)

(f)

140

160

10

5

20

102


80˚N

60˚W

Surface nitrate

0˚ 60˚E 120˚E 180˚ 120˚W

80˚N

40˚

60˚

5

1

15

5

10

1

60˚

40˚

20˚

1

1

1

20˚

0˚

1

5

0˚

20˚

1

40˚

60˚

FIGURE 4.23

(1994).

20˚

1

1

5

10

5 40˚

15

20

20

25

25

60˚

80˚S

80˚S

60˚W 0˚ 60˚E 120˚E 180˚ 120˚W

Nitrate (mmol/L) at the sea surface, from the climatological data set of Conkright, Levitus, and Boyer

properties in the Southern Hemisphere

subtropics, we can also distinguish between

higher oxygen, lower nutrient waters that

come from the Antarctic compared with the

low oxygen, high nutrient Pacific and Indian

Deep Waters. These Antarctic deep waters are

not as dense as AABW and are referred to as

Circumpolar Deep Water. Different types of

Circumpolar Deep Water are described in

Chapter 13.

4.7. AGE, TURNOVER TIME, AND

VENTILATION RATE

Estimates of age and overturning rates of

ocean waters assist in understanding the overall

distribution of temperature and salinity in the

ocean, the replenishment rate of nutrients in

the upper layers, and the exchange of gases

between the atmosphere and ocean. The

effectiveness and safety of the deep ocean as

a dump for noxious materials depends on the

turnover time of the deep waters. For these

and many other applications, it is useful to estimate

timescales for ocean ventilation.

Age as applied to ocean water is the time

since a parcel of water was last at the sea

surface, in contact with the atmosphere. The

ventilation rate or production rate is the transport

of water that leaves a surface formation site

and moves into the ocean interior. Turnover

time is the amount of time it takes to replenish

a reservoir, such as an ocean basin or a layer

or water mass in the ocean. The “reservoir”

can also be construed in terms of a tracer rather

than water particles (e.g., molecules of CO 2 ,or

zooplankton, etc.). Residence time is the time

a particle spends in a reservoir.

The ages of waters can be estimated using

tracers (Section 3.6). Tracers that are biologically

inert are more straightforward than those that

AGE, TURNOVER TIME, AND VENTILATION RATE 103

are biologically active. Anthropogenic transient

tracers that have measured histories in the atmosphere

are useful in the upper ocean and wellventilated

parts of the deep ocean. The penetration

of chlorofluorocarbons (CFCs; Pacific section

shown in Figure 4.24a) and tritium (Pacific map

in Figure 4.25b) is evidence of recent ventilation;

absence of these tracers is clear evidence of age

that is greater than 50e60 years.

Pairs of tracers whose concentration ratio

changes with time can be used to estimate age,

including pairs of CFCs with different atmospheric

time histories that result in a changing

ratio in surface source waters (Figure 4.25a). Similarly,

because tritium ( 3 H) decays to 3 He with

a half-life of about 12 years, the 3 H/ 3 He pair

reflects age (ignoring the smaller amounts of

natural 3 He injected in the deep ocean at the

mid-ocean ridges). This tracer ratio method is

straightforward only if the surrounding waters

are free of the tracers because mixing between

waters with different ratios complicates the age

calculation. The tropical Pacific and North Pacific

are ventilated only in the upper ocean, with no

deep water sources except far to the south, so

the CFC and tritium/ 3 He pair “ages” are especially

useful for estimating water age there.

For the deep ocean, where water is too old to

be dated using anthropogenic tracers, and also

as an alternate method of estimating age in the

better ventilated parts of the ocean, natural

tracers such as oxygen, nutrients, and 14 C

(Figure 4.24b) are useful. Biological activity

reduces oxygen and increases nutrient content

once the water moves away from the sea

surface. If the oxygen consumption rate or

nutrient remineralization rates are known as

a function of location and temperature, then

the age of a water mass as it moves away from

its source can be estimated. As with anthropogenic

tracers, simplifying assumptions about

mixing with waters of different oxygen and

nutrient content are required.

Radiocarbon can be used for dating just

as with terrestrial organic matter (Section 3.6).

14 C is created in the atmosphere by cosmic

rays and quickly becomes part of the atmospheric

CO 2 . It enters the ocean with the CO 2

that dissolves in surface water. When the

surface water is subducted or incorporated in

deeper waters, the decay of 14 C at a rate of 1%

every 83 years results in increasingly negative

values (deficits) along the pathways into the

deep ocean. The largest deficits globally are

found in the deep North Pacific, reflecting the

great age of the waters there (Figure 4.24b).

Use of 14 C deficits to precisely date ocean water

is subject to caveats about mixing and also

complications due to local sinking of organic

matter from the surface and other sources of 14 C

including nuclear testing. The gross estimate

of ages of deep waters based directly on 14 C

deficits is 275 years for the Atlantic, 250 years

for the Indian, and 510 years for the Pacific. It

is easy to see these age estimates are biased since

the oldest waters always mix with younger

waters and vice versa. Thus the age of the

deep northern Pacific waters is likely higher

than their 14 C age (around 1000 years), while

the age of the deep northern Atlantic waters is

lower, as evidenced by invasion of CFCs to the

bottom (Broecker et al., 2004).

The ventilation rate (production rate) of a water

mass or layer can be defined in several different

ways, which can lead to somewhat different

quantitative estimates. In all cases, the objective

is to estimate the rate of injection of new water

into a reservoir. One approach is to estimate

production rate from the volume of the reservoir

divided by its age:

R P ¼ Volume=Age

(4.1a)

which has units of transport (m 3 /sec). This is

a straightforward concept, but difficult to implement

since the ocean is not composed of simple

boxes filled with waters of uniform age; therefore

somewhat complex calculations and simple

models are used to obtain ventilation rates from

the continuous distribution of ages. If Eq. (4.1a)

(a)

0

500

5.7

4.0 4.0

2.0

2.0

1.0

0.5

0.25

104

1000

0.02

1500

2000

0.02

Depth (m)

2500

3000

3500

0.25

4000

4500

5000

5500

6000

6500

0

(b)

Depth (m)

0

500

1000

1500

2000

2500

3000

3500

Pacific

CFC-11

500 1000 1500 2000 2500 3000 3500 4000 4500 5000 5500 6000 6500 7000 7500 8000 8500 9000 9500 10000 10500 11000 11500 12000 12500 13000 13500

Distance (km)

–160

–180

–40

0

100

–210

–200

–100

–230

–235

–240


4000

4500

5000

5500

6000

6500

0

Pacific

Δ 14 C

500 1000 1500 2000 2500 3000 3500 4000 4500 5000 5500 6000 6500 7000 7500 8000 8500 9000 9500 10000 10500 11000 11500 12000 12500 13000 13500

Distance (km)

FIGURE 4.24 (a) Chlorofluorocarbon content (CFC-11; pmol/kg) and (b) D 14 C (/mille) in the Pacific Ocean at 150 W. White areas in (a) indicate

undetectable CFC-11. From the WOCE Pacific Ocean Atlas. Source: From Talley (2007).

AGE, TURNOVER TIME, AND VENTILATION RATE 105

A related approach to estimating a ventilation

rate using transient tracers starts with the total

amount (inventory) of the tracer and concentration

of the tracer at its source (sea surface). For

instance, using CFC’s with an inventory I CFC

(units of moles) and surface concentration

C source (units of moles/kg), the ventilation rate

is given by (Smethie & Fine, 2001):

I CFC ¼ X R P C source Dt

(4.1b)

(b)

Tritium (TU) 500 m

120°E 180° 120°W 60°W

80°N 80°N

60°N 1.0

60°N

0.5

0.3

2.01.0

1.0

40°N

0.5

1.5

40°N

2.0

1.0

20°N 1.5

20°N

0.4

0.5

0.3

0.1

0.2

0.05

0.02

0.1

0°

0.02

0°

1

0.05

0.05

20°S 0.2

20°S

0.1

0.3

0.4

0.5

40°S 40°S

0.2

0.1

60°S 60°S

0.05

0.02

80°S 80°S

120°E 180° 120°W 60°W

0.02 0.2 0.5 2

1 3

FIGURE 4.25 (a) Age (years) of Pacific Ocean waters on

the isopycnal surface 27.2 s q , using the ratio of chlorofluorocarbon-11

to chlorofluorocarbon-12. Source: From Fine,

Maillet, Sullivan, and Willey (2001). (b) Tritium concentration

at 500 m in the Pacific Ocean from the WOCE Pacific Ocean

Atlas. Source: From Talley (2007).

is written in terms of turnover time (Eq. 4.2

below) instead of age, the rate that is obtained

could differ since age and turnover time are

usually not identical.

Since both the source concentration and inventory

vary with time, this ventilation rate is

obtained iteratively.

Ventilation rates, R P , are also estimated from

observations of the transport of newly ventilated

waters very close to the source of the water

mass. Farther from the source, quantitative

water mass identification techniques (Section

6.7) can be used to estimate the portion of

observed transport that can be attributed to

the source versus the portion due to mixing

with other waters. Indirect estimates are also

frequently used, based on measuring the buoyancy

forcing that results in ventilation with

simple or complex models to compute the ventilation

rate; an approach using airesea fluxes of

heat and freshwater within surface outcropping

regions of isopycnal layers was introduced by

Walin (1982) and has been employed in numerous

calculations.

Turnover time is the time it takes to replenish

a reservoir. If in reference to water rather than

a tracer, it is equal to the volume V of the water

mass or layer, in units of m 3 , divided by its

outflow transport R out in m 3 /sec. If in reference

to a tracer, it is the inventory of the tracer, in

moles, divided by the transport of the tracer

out of the reservoir in mole/sec. Turnover

time, which has been defined generally for use

in biogeochemistry, is written in terms of

the exit flow because reservoirs are usually

well-mixed, unlike the inflow sources, resulting

in a simpler (proportional) relation between

outflow volume transport and turnover time.

106


Turnover times for volume and for a tracer are

given by:

T turnover ¼

V

RRR RRR

dV dV

¼ RR

R out vout dA / RR

vin dA (4.2a)

T Cturnover ¼

RRR

rCdV

RR

rCvout dA

(4.2b)

where v out and v in are velocities out of and in

to the reservoir, C is the concentration of

a tracer (e.g., in mmol/kg) and r is density.

Eq. (4.2a) can also be written in terms of

mass rather than volume transport by

including r in both the numerator and

denominator. The rightmost term with inflow

velocity in Eq. (4.2a) yields turnover time if

the system is in steady state. In a steady state,

Eq. (4.2b) could also be written in terms of the

inflow.

Residence time is the time an individual water

parcel spends in a reservoir. The average residence

time is obtained by averaging over all of

the water parcels passing through the reservoir.

The average residence time is equal to the turnover

time if the system is in steady state. The

average residence time is twice the age if water

is moving steadily through the reservoir, since

the age is the average of newest to oldest waters

in the reservoir.

4.8. OPTICAL PROPERTIES

OF SEAWATER

The transparency of the ocean depends on

how much suspended or living material is contained

in it, as described in Section 3.8. If the

water is very transparent, then solar radiation

penetrates to greater depth than if there is

much suspended material. Therefore optical

properties of the surface waters affect surface

layer heating, thereby affecting surface temperature

and hence ocean-atmosphere interaction.

Ocean color depends on suspended materials,

especially including chlorophyll-producing

phytoplankton, so large-scale observations of

color and other optical properties can be used

to study biological productivity. Optical observations

of ocean color using satellite remote

sensing are used routinely to quantify the

amount of chlorophyll-a (green pigment;

McClain, Hooker, Feldmand, & Bontempi,

2006), and, more recently, particulate organic

carbon (POC; Gardner, Mishonov, & Richardson,

2006; Stramski et al., 2008), and euphotic zone

depth (Lee et al., 2007).

Prior to invention of electronic optical

devices, transparency was measured using

a Secchi disk (see Section S16.8 in the supplemental

materials on the text Web site for information

about this instrument). This was done

by visually observing when the specially

painted disk could no longer be seen from the

ship’s deck. An enormous number of Secchi

disk depths (>120,000) were collected and are

archived at the U.S. National Oceanographic

Data Center (Lewis, Kuring, & Yentsch, 1988).

The majority of the values were for the northern

oceans and taken in the summer. There are large

areas of the Southern Hemisphere open ocean

where there are no values at all, but coastal areas

were generally well sampled. Large Secchi

depths are found in the open oceans at low

and middle latitudes with lower values in

higher latitudes and along most coasts. The latitudinal

variation is apparent in Figure 4.26,

which shows averages of Secchi depths along

180 20 W for the Pacific, and along 35 10 W

for the Atlantic. Lewis et al. (1988) concluded

that the prime source of variability in the open

ocean is attenuating material in the water. The

smaller Secchi depths correspond to higher

chlorophyll-a values. The most marked feature

in Figure 4.26 is the sharp decrease in Secchi

depths beyond about 30 latitude, corresponding

to higher productivity in the higher latitudes.

The large Secchi depths in the Atlantic

are in the Sargasso Sea, a region of notably low

biological productivity. In a polynya in the

OPTICAL PROPERTIES OF SEAWATER 107

FIGURE 4.26

(1988).

Mean Secchi disk depths as functions of latitude in the Pacific and Atlantic Oceans. Source: From Lewis et al.

Weddell Sea in 1986, a Secchi disk was visible to

four observers at 79 m and disappeared at 80 m.

This was claimed as a record: the Secchi depth

calculated for distilled water is 80 m, so a greater

depth is not possible. In coastal waters, values of

2e10 m are common, and in silty waters near

rivers and in estuaries, values of less than 1 m

are observed.

Modern in situ optical observations are made

with instruments that measure many different

aspects of seawater optical properties, which

are affected by suspended materials, including

sediments and plankton (Section 3.8; Figure 3.9).

Fluorescence provides a measure of chlorophyll

concentration and therefore, phytoplankton.

Within the water column, light transmission,

beam attenuation, and fluorescence, among

other properties, are measured at different

wavelengths to quantify the amount and

type of suspended material (Gardner, 2009). As

an example, beam attenuation measured with

a transmissometer, at a visible wavelength

(660 nm), is shown for the equatorial Pacific

and the eastern subpolar North Pacific (Ocean

Weather Station P or Papa; Figure 4.27). This

instrument provides its own light as it is lowered

through the water column, so the observation

is related to the local amount of scattering

and absorption by particles and absorption by

water, and not to the actual penetration of

sunlight. This particular beam attenuation can

then be related to the amount of POC, which is

measured from actual samples of seawater.

The high beam attenuation in the uppermost

layer indicates high POC.

Using ocean color remote sensing and in situ

observations of chlorophyll-a and POC, algorithms

have been developed to map the latter

quantities. Chlorophyll-a maps are now standard

remote sensing products. Seasonal maps

of chlorophyll from remote sensing are shown

in Figure 4.28. Notable features of the northern

summer chlorophyll distribution include very

low values throughout the subtropical gyres,

high values in the equatorial regions and along

parts of the ACC, very high values in the high

northern latitudes and Arctic, and high values

in coastal regions. In austral summer, the high

latitude patterns reverse somewhat, with

increased chlorophyll along the margins of

108


FIGURE 4.27 Profile of beam attenuation coefficient at 660 nm, from a transmissometer, converted to POC (solid line)

and in situ measurements of POC (circles): (a) equatorial Pacific and (b) northeast Pacific at OWS Papa. Source: From Bishop

(1999).

FIGURE 4.28 Global images of chlorophyll derived from the Coastal Zone Color Scanner (CZCS). Global phytoplankton

concentrations change seasonally, as revealed by these three-month “climatological” composites for all months between

November 1978eJune 1986 during which the CZCS collected data: JanuaryeMarch (upper left), AprileJune (upper right),

JulyeSeptember (lower left), and OctobereDecember (lower right). Note the “blooming” of phytoplankton over the entire

North Atlantic with the advent of Northern Hemisphere spring, and seasonal increases in equatorial phytoplankton

concentrations in both Atlantic and Pacific Oceans and off the western coasts of Africa and Peru. Figure 4.28 will also be

found in the color insert. See Figure S4.2 from the online supplementary material for maps showing the similarity between

particulate organic carbon (POC) and chlorophyll. Source: From NASA (2009a).

OPTICAL PROPERTIES OF SEAWATER 109

FIGURE 4.29 Euphotic zone depth (m) from the Aqua MODIS satellite, 9 km resolution, monthly composite for

September 2007. (Black over oceans is cloud cover that could not be removed in the monthly composite.) See Figure S4.3

from the online supplementary material for the related map of photosynthetically available radiation (PAR). This figure can

also be seen in the color insert. Source: From NASA (2009b).

Antarctica (now ice free) and reduced chlorophyll

in the high northern latitudes. The POC

distribution derived from ocean color is closely

related to the chlorophyll-a distribution (Gardner

et al., 2006; see Figure S4.2 in the online

supplement).

The solar radiation that affects the upper

ocean is quantified as photosynthetically

available radiation (PAR; Section 3.8.1), and is

mapped routinely from ocean color sensors

(NASA, 2009b). An example (Figure S4.3) is

included in the online supplementary materials.

(In the NASA images, 1 Einstein ¼ 1 mole of

photons.) The reader is encouraged to visit the

NASA Web site where images are continually

posted and where the large seasonal variability

is readily apparent.

The euphotic zone depth (Figure 4.29), which

is defined as the depth of 1% light penetration,

is also mapped from satellite color information

using algorithms based on in situ observations

(Lee et al., 2007). The euphotic zone depth is

related to the historical Secchi disk depths

(Figure 4.26 and Section S16.8 of the supplementary

online materials); the features that

were described previously for the zonally averaged

Secchi depths are apparent in the satellitebased

map.

Ocean color and derived products are mapped

at a resolution of 4e9 km (as in Figure 4.29).

110


Color then complements remotely sensed SST

data of a similar spatial resolution. Chlorophyll-a

is somewhat independent of SST, so

the two products provide powerful information

about local circulation (advection) and

upwelling (Simpson et al., 1986). The two fields

are used extensively in studies of regional circulation

and ecosystems. Examples of ocean color

maps to illustrate regional circulation are

included throughout later chapters.

C H A P T E R

5

Mass, Salt, and Heat Budgets

and Wind Forcing

Conservation principles, such as the

conservation of energy, of mass, of momentum,

and so forth, are important in all of the sciences

because these simple principles have very farreaching

results and valuable applications.

This chapter discusses conservation of volume

(or mass; Section 5.1), conservation of salt and

freshwater (Sections 5.2 and 5.3), and conservation

of heat energy (Section 5.4), as applied to

the oceans. The ocean’s heat budget and heat

transports are described in Sections 5.5 and

5.6. Because heat and freshwater fluxes combine

to make the buoyancy fluxes that are applied to

the surface ocean, airesea buoyancy fluxes

are presented in Section 5.7. To complete the

presentation of the principal drivers of the

ocean circulation (prior to chapters describing

the dynamics and circulation), wind forcing is

included in Section 5.8.

5.1. CONSERVATION OF VOLUME

AND MASS

The conservation of volume principle (or, as

it is often called, the equation of continuity) is

based on the fact that the compressibility of

water is small. If water is flowing into a closed,

full container at a certain rate, it must be flowing

out somewhere else at the same rate or the level

in the container must increase. “Containers”

(such as bays, fjords, etc.) in the oceans are not

closed in the sense of having lids (except when

frozen over), but if, say, the mean observed sea

level in a bay remains constant (after averaging

out the waves and tides), then the bay is equivalent

to a closed container.

For example, many of the fjords of Norway,

western Canada, and Chile have large rivers

flowing into their inland ends, but the mean

sea level in them remains constant. We conclude

from the continuity of volume that there must be

outflow elsewhere, since evaporation is very

unlikely to be large enough to balance the inflow.

The only likely place for outflow is at the

seaward end; if we measure the currents in

fjords, we usually find a net outflow of the

surface layer. However, when we actually

measure the outflow, we might find a much

greater volume flowing out to the sea than is

coming in from the river. Because volume must

be conserved, there must be another inflow; fjord

current measurements usually show inflow in

a subsurface layer. The river water, being fresh

and therefore less dense than the seawater of

the fjord, stays in the surface layers as it flows

toward the sea. The subsurface inflow is freshened

by mixing with the river water, and upwells

into the surface layer where it flows out with the

river water. (This is estuarine circulation; see


111



112

5. MASS, SALT, AND HEAT BUDGETS AND WIND FORCING

Chapter S8, Section S8.8 in the online

supplementary material located at http://


supplemental material.)

5.1.1. Conservation of Volume in a

Closed Box

If we represent flows in and out of a closed

region such as a fjord (Figure 5.1), and add

precipitation (P) and river runoff (R), and

subtract evaporation (E) from the water surface,

conservation of volume may be stated as:

V i þ R þ AP ¼ V o þ AE (5.1)

or, rearranged slightly, as

V o V i ¼ðR þ APÞ AEhF: (5.2)

Here V is the volume transport, which expresses

flow in terms of volume per second (m 3 /sec)

rather than as a speed (m/sec). The subscripts

o and i represent outward and inward transports,

respectively. The symbol R represents

the river runoff as a volume transport that

adds water into the basin. The symbols P and

E are the precipitation and evaporation for

each point and are therefore expressed in

m/sec or an equivalent, such as cm/year. To

calculate the total volume transport into the

box due to precipitation and evaporation, they

must be integrated over the surface area of the

box. For the simple, illustrative approximation

of Eqs. (5.1) and (5.2), if P and E have the same

values at all points in the box, they can just be

FIGURE 5.1 Schematic diagram of basin inflows and

outflows for conservation of volume discussion.

multiplied by the surface area, A, of the box. F

(for freshwater) is defined by the right side of

Eq. (5.2), which is why the symbol h is used.

The left side of Eq. (5.1) is the volume transport

into the fjord. The right side of Eq. (5.1) is the

volume transport out. The second equation

simply says that the net volume flow of salt

water balances the net volume flow of freshwater

(when averaged over a suitable time

period). This is an example of a steady-state situation

in which some or all of the parts of

a system may be in action, but at no point is

there any change of motion (or of a property)

with time.

(To be more precise, the principle expressed

in Eqs. (5.1) and (5.2) should also include the

density of water, and becomes a statement of

mass conservation rather than of volume

conservation. This is because simple heating of

water will expand it slightly without adding

any mass, so the true conservation principle is

for mass. However, for most ocean applications,

seawater density has such a small range of

values that we can usually assume the density

is uniform.)

Even though this conservation principle was

discussed using an example of a nearly closed

region, such as a fjord, it also applies just as

well to any other closed “box” that might be

drawn in the ocean. If our closed box includes

the sea surface, then it will include P and E. If

it has a coastline, then it will include R. If it

has sea ice flowing into it and melting, or vice

versa, then it includes yet another term for the

water volume in the ice. Also, our box could

be completely within the ocean somewhere, in

which case the flows into the box must balance

flows out of the box, as described next.

5.1.2. Open Ocean Continuity

Thinking about flow into and out of a closed

box can be extended into the open ocean. Here

we think of a hypothetical closed box, with

sides, a top, and a bottom (Figure 5.2). We

CONSERVATION OF VOLUME AND MASS 113

V i = v in A in

A in

A out

V o = v out A out

FIGURE 5.2 Continuity of mass for a small volume of

fluid. By continuity, V o ¼ V i .

then apply the same balance (Eq. 5.1) to this box.

If none of the sides are next to the coast, then the

runoff term R is zero. If the top of the box is

inside the ocean and is not the sea surface, the

precipitation and evaporation terms are also

zero. Then the volume balance for the box

becomes:

V o V i ¼ 0 (5.3)

This says that the transport into the box must

equal the transport out of the box. (The full

“continuity equation,” expressed in partial

derivative form, is given in Section 7.2.) In practice,

in all open ocean areas, the volume transports

into and out of boxes of interest are

usually much larger than any precipitation or

evaporation flux across the sea surface, so we

use an approximate version of Eq. (5.3) even

for boxes that include the sea surface (Section

5.1.3).

This principle of continuity is so fundamental

that it might not seem very interesting, but it is

the one law that applies in all situations, regardless

of how complex the system becomes.

5.1.3. Radiation, Flux, and Diffusion

Before we can talk further about conservation

of volume and salt, we need to understand how

heat, water, salt, and other dissolved materials

move around within the ocean and how they

can be changed by physical processes (as

opposed to chemical or biological processes).

There are three ways to physically change

things inside the ocean: radiation, advection,

and diffusion.

Radiation is how electromagnetic waves e

heat and light e move. Radiation is most important

in the atmosphere and less important in the

ocean, since water is not very transparent.

However, light does penetrate the ocean’s

surface layer (“euphotic zone”; Section 3.8),

which is how the sun actually heats the ocean

(Section 5.4). The ocean also radiates heat

(infrared electromagnetic waves) out to the

atmosphere (Section 5.4.2).

Advection is how the movement of a fluid

“parcel” carries properties such as heat and

salt. Sometimes we use the word convection

when referring to vertical motion. As already

introduced in Section 5.1.1, the basic concept

here is velocity, which has units of length

divided by time (m/sec) and a direction. A fluid

is made up of countless molecules that move

more or less together. If we draw some sort of

surface in our minds through a part of the

ocean, the surface will have an area (A;

Figure 5.2). The volume transport of water

through the area is equal to the velocity of the

fluid (v) through the surface multiplied by the

area A, or vA. The units of volume transport

are m 3 /sec. We can talk about transport of

mass as well. Water, including seawater, has

density r, which has units of mass/volume.

Mass transport though our area is then density

multiplied by velocity and area (r v A) and

has units of kg/sec.

Seawater has dissolved matter in it, which

has a concentration (C) of mass or molecules

of matter per unit mass of seawater. (Recall

our definition of salinity in Chapter 3.) We can

talk about any dissolved matter, including salts.

The transport of the dissolved matter becomes

this concentration times density times velocity

times area (C r v A) and has units of (mass of

matter)/time or molecules/time, depending

on how you write the concentration. For salt

transport, we use salinity written as units of

114


grams of salt per kilogram of seawater, and so

salt transport has units of g/sec, which can

also be written in terms of kg/sec by multiplying

by 1000.

For specific dissolved substances, the concentration

can be written in terms of moles per unit

mass of seawater (mol/kg). Because concentrations

of many common dissolved materials

(such as oxygen and nutrients) are on the order

of 10 6 mol/kg, the concentration unit of

mmol/kg (micromoles per kilogram) is often

used. Transports of these substances are then

expressed as mmol/sec.

Heat transport uses the definition of heat Q in

Eq. (3.1), which has units of energy (Joules). In

place of “concentration,” we use the specific

heat multiplied by temperature (on the absolute

temperature scale, Kelvin), and the transport is

(Q v A), with units of J/sec. The unit for J/sec

is a Watt, where 1 W ¼ 1 J/sec.

Flux is directly related to transport. Flux is

the transport per unit area. It can be thought

of as “stuff” per unit time per unit area. For

instance, heat flux is expressed as W/m 2 ,

which is Joules per unit time per unit area.

Flux and transport are important when there

is a difference between the transport of a property

into a closed volume and the transport

out of the same closed volume (Figure 5.2).

The change in transport has to be related to

a change in properties within the closed box.

For example, if there is a larger flux of heat

into a box than out of the box, then the water

coming out of the box is cooler and must have

been cooled inside the box. This could happen

through a loss of heat out of the sea surface if

one side of the box is at the sea surface. As

another example, if there is a higher transport

of oxygen into the box than out of the box,

then oxygen could have been consumed

(usually by bacteria) within the box. This change

in transport through a box is called transport

divergence (if more comes out than goes in) or

transport convergence (if less comes out than

goes in).

Advection is similar to flux, but advection

occurs at a point rather than through the side

of a volume. When a property is carried along

by the flow, it is “advected.” The equations in

fluid mechanics that describe the change in

a property at a point in a fluid include “advective

terms,” which indicate how the divergence

or convergence of the flux of the property

changes the property at that location.

Diffusion is the third way properties can

change in a fluid. Diffusion is like flux convergence

and divergence, but it happens at

extremely small spatial scales. Molecules or

tiny parcels of water bump around randomly

(turbulence) and carry their properties with

them. If there is a difference in heat or salt

from one side of a region to the other side,

then the random jostling will gradually smooth

out the difference (or “gradient,” which is the

property difference divided by the separation

distance, if you consider this distance as

becoming very small).

Fick’s law of diffusion says that the diffusive

flux of “stuff” is proportional to its concentration

gradient. Therefore diffusion will move

stuff down a gradient (from high concentration

to low concentration). If there is no

gradient (no variation in concentration), there

is no flux and hence no impact of diffusion. If

the gradient is constant (meaning that the

property difference centered at one location

is the same as at another location), diffusion

also has no effect on the property distribution

because there must be a flux divergence or

convergence for the property to change. In

mathematical terms, the second spatial derivative

of the concentration must be non-zero

for diffusion to cause a change in concentration.

More simply said, if there is more stuff

at one location than at another, it will flux

toward the lesser concentration. But the

concentration will only change if there is

a flux convergence or divergence. Therefore

diffusion only acts if the property concentration

gradient varies.

CONSERVATION OF SALT 115

In turbulent flows, such as water in the

ocean or air, we are sometimes more interested

in properties and velocity changes over scales

of many meters to many kilometers, or even

thousands of kilometers, than over scales of

centimeters or less. It is almost impossible to

consider all ranges of motion at once, even

though a single, limited set of fluid dynamics

equations describes all of them. Therefore fluid

dynamicists and modelers almost always make

simplifying assumptions about the scales of

motion (spatial and temporal) they are interested

in, and often treat smaller scales (subgrid

scale) as if they obey random, molecular

motions.

Fluids such as water and air are highly turbulent,

meaning that they are not very viscous.

Turbulence at small scales is often considered

to act on the larger scales of interest in a way

that is analogous to random, molecular, microscopic

motions. Thus, fluid dynamicists introduce

the concept of eddy diffusivity, in which

turbulent “eddies” at smaller scales accomplish

the diffusion. Eddy diffusivity is much higher

than molecular diffusivity since turbulent eddies

carry properties much farther than molecular

motions. Eddy diffusivity (and eddy viscosity)

are discussed again in Section 5.4.7 and more

formally and in more depth in Section 7.2.4.

5.2. CONSERVATION OF SALT

The principle of conservation of salt is based

on the nearly accurate assumption that the total

mass of dissolved salts in the ocean is constant.

Rivers contribute a total of about 3 10 12 kg of

dissolved solids per year to the sea, which may

sound like a lot, but it has negligible effect on

salinity. The ocean is very large. The total

amount of dissolved salt in the ocean is 5

10 19 kg. Therefore the amount of salt brought

into the ocean each year by the world’s rivers

increases the average ocean salinity by about

one part in 17 million per year. We can only

measure salinity to an accuracy of about

0.001, or about 500 parts in 17 million if we

assume the mean ocean salinity to be about 35.

In other words, the oceans would increase in

salinity each year by an amount which is only

1/500 of our best accuracy of measurement if

we neglect removal of salt for the moment.

Salt is actually removed from the ocean in the

form of evaporites, so salinity actually increases

even more slowly, if at all, over geological time.

For all practical purposes, it is assumed that the

average salt content of the oceans is constant, at

least over periods of tens or even hundreds of

years.

Salinity, which is the dilution of the salts

(Section 3.4), could vary at a barely measurable

level with change in the total amount of water

in the ocean, which depends on how much

water is locked up in ice, especially in ice sheets

(Greenland and Antarctica). If 1 m of water were

added to the ocean from melting ice sheets

and glaciers, the salinity change would be

1/4000 ¼ 0.0003, since the mean depth of the

oceans is about 4000 m (Section 2.1). The

maximum sea level change that might be

expected from complete melting of Greenland

(which is not an unreasonable possibility) is 7

m, leading to a mean salinity decrease of 0.002,

which would be barely observable.

The early Greek philosophers were confused

about the salinity of the ocean as compared to

the freshwater of the rivers feeding into them.

Rather than realizing that the very long-term

accumulation of salts from these rivers caused

the ocean’s salinity, they postulated “salt fountains”

at the bottom of the ocean. As recently

as the nineteenth century, Maury (1855)

believed that salt had been present in the oceans

since “creation,” in contrast to what he refers to

as the “Darwin theory” (which is now understood

to be mostly correct), that the salt is

washed in by rivers.

The conservation of salt is usually applied to

bodiesofwaterthataresmallerthantheworld

ocean. Salt conservation in smaller bodies, such

116


as the Mediterranean Sea, is a reasonable

hypothesis. Salinity, however, can vary when

the freshwater balance changes the dilution,

mainly through changes in precipitation and

evaporation patterns and amplitude. While

the mean salinity of the ocean does not measurably

change, the salinity of one region can

increase while that of another region decreases,

due to redistribution of freshwater. The salt

conservation principle may be expressed

symbolically as:

V i r i S i ¼ V o r o S o (5.4)

where V i and V o are the volume transports of

the inflowing and outflowing seawater for an

enclosed region, S i and S o are their salinities,

and r i and r o their densities. This is the equation

for salt transport; it states that no salt is gained or

lost inside the region. (For complete accuracy,

Eq. 5.4 should be applied pointwise, using

velocity rather than transport, and then added

up over the whole area surrounding the volume

being considered.) The left side of Eq. (5.4) is the

rate at which salt is transported into the box,

and the right side is the rate of transport out.

Because the two densities will be the same

within 3% (the difference between seawater

and freshwater) the r’s nearly cancel, leaving:

V i S i ¼ V o S o (5.5)

This equation can be combined with the equation

for conservation of volume (5.2) to give

Knudsen’s relations (Knudsen, 1901):

V i ¼ F S o =ðS i S o Þ and V o ¼ F S i =ðS i S o Þ

(5.6)

where F is due to runoff, precipitation, and

evaporation (F ¼ R þ P E), integrated over

the full area. F is the freshwater volume input

measured in m 3 /sec. Equation (5.6) is useful

for calculating volume transport if we know F

and measure the salinities.

Conversely, if we know the transports

and salinities around the perimeter of the

region through measurements, we can calculate

F:

F ¼ V i ðS i =S o 1Þ or F ¼ V o ð1 S o =S i Þ

(5.7)

This is the equation for freshwater transport and

expresses how much freshwater is gained or

lost inside the box. Equations (5.6) and (5.7) can

be applied to any region, especially including

marginal seas, estuaries, and fjords where inflow

and outflow salinities are easily assigned. If F is

positive (more runoff and precipitation than

evaporation), then the marginal sea is considered

to be “positive.” If F is negative (net evaporation),

then the marginal sea is called “negative.”

Qualitative conclusions can be drawn from

Eqs. (5.6) and (5.7). If both S o and S i are large,

they must be similar because there is an upper

limit to S in the ocean. Therefore (S i S o )must

be small and both S o /(S i S o ) and S i /(S i S o )

must be large. Therefore V i and V o must be large

compared with F, the excess of freshwater

inflow over evaporation. That is, for large

volume exchanges (large flushing rate), the

salinity change will be small for a given amount

of evaporation or precipitation. On the other

hand, if the salinity difference between inflow

and outflow is large (S o much less or much

more than S i ), then the exchange rate (V i and

V o ) is small for the same size F. Thus a body of

water with large volume exchange will be better

flushed and less likely to be stagnant than one

with small volume exchange.

For the open ocean, where salinity and

velocity vary continuously, it is more useful

and accurate to calculate salt and freshwater

transports as integrals of vS and v (1 S/S o )

around the whole region being considered,

where v and S are point observations of velocity

and salinity (Wijffels, Schmitt, Bryden, & Stigebrandt,

1992; Wijffels, 2001; Talley, 2008). (The

integration is in depth and horizontal distance

around the region.) S o is an arbitrary constant.

The net volume transport F into the whole

region should balance the amount gained and

THREE EXAMPLES OF THE TWO CONSERVATION PRINCIPLES 117

lost by runoff, precipitation, and evaporation,

hence be very small.

5.3. THREE EXAMPLES OF THE

TWO CONSERVATION PRINCIPLES

5.3.1. The Mediterranean Sea: An

Example of Negative Water Balance

The Mediterranean Sea has a negative water

balance e evaporation exceeds precipitation

plus river runoff. There is a small net loss of

volume due to net evaporation (i.e., for the

volume transport Eq. 5.2, E > (R þ P) and F is

negative). Because salt is conserved, the salinity

increases. The saltier water is denser and sinks

within the Mediterranean. This denser water

flows out of the Mediterranean at the bottom

of the sill at the Strait of Gibraltar, injecting

this saltier water into the North Atlantic at

depth (Section S8.10.2 in Chapter S8 located on

the textbook Web site). The outflow, with

salinity 38.4 psu, is balanced by inflow of less

salty (36.1 psu) water from the North Atlantic

in the upper layer (Figure 5.3a). The ratios of

salinities in Eq. (5.6) both have values of about

16, which means that the inflow and outflow

volume transports V i and V o are both greater

by this factor than the airesea freshwater loss, F.

Direct measurements of the upper layer

inflow at the Strait of Gibraltar (Section S8.10.2

on the textbook Web site) give an average inflow

transport of V i ¼ 0.72 Sv, where 1 Sv ¼ 1 10 6

m 3 /sec. Then, from Eq. (5.6), V o ¼ 0.68 Sv and

F ¼ (R þ AP) AE ¼ 0.04 Sv; in other words,

total evaporation exceeds freshwater input by

0.04 10 6 m 3 /sec. The units for inflow V i can

be converted to 2.3 10 4 km 3 /year. At this

rate it would take about 165 years to fill the

Mediterranean, which has a volume of 3.8

10 6 km 3 . (The Mediterranean does not “fill”

since outflow balances inflow.) This “filling

rate” is a measure of the mean turnover time,

that is, the time required for replacement of all

the Mediterranean water (sometimes called

flushing time or residence time) (Section 4.7).

The deep salinity within the Mediterranean

Sea is between 38 and 39 psu (Section S8.10.2

on the textbook Web site). The outflow salinity

at the Strait of Gibraltar is lower than this

because the outflow entrains (mixes with) lower

salinity inflow as it passes through the strait into

the North Atlantic.

5.3.2. The Black Sea: An Example

of Positive Water Balance

Even though it is adjacent to the Mediterranean,

the Black Sea (Section S8.10.3 located on

the textbook Web site) is a “positive” basin, in

which there is a net gain of freshwater from

the atmosphere and runoff (Figure 5.3b). The

salinity of the inflow (bottom layer) is approximately

35 psu. The salinity of the outflow

(upper layer) is much lower at 17 psu. The ratios

of salinities in Eq. (5.6) are 1 and 2, respectively,

indicating that the transports V i and V o , which

are the Black Sea’s exchange with the Mediterranean,

are of the same order as the airesea

(a)

ATLANTIC

for MEDITERRANEAN

V i

S i = 36.1 psu

V o

S o = 38.4 psu

290 m

Strait of Gibraltar

100 m

(b)

MEDITERRANEAN

for BLACK SEA

V o

S o = 17 psu

V i

S i = 35 psu

35 m

Bosporus

25 m

BLACK SEA

FIGURE 5.3 Schematic diagrams

of inflow and outflow characteristics

for (a) Mediterranean Sea (negative

water balance; net evaporation), (b)

Black Sea (positive water balance; net

runoff/precipitation).

118


freshwater balance flux F. Measured values are

approximately V i ¼ 9.5 10 3 m 3 /s (300 km 3

yr 1 of saline water) and V o ¼ 19 10 3 m 3 /s

(600 km 3 yr 1 of fresher water), giving F ¼

(R þ P) E ¼ 9.5 10 3 m 3 /sec (Oguz et al.,

2006). That is, there is a net flux of freshwater

into the Black Sea from an excess of runoff and

precipitation compared with evaporation. (The

mean salinity of the deep Black Sea is about

22.4 psu and the surface layer is much fresher,

which means there is a net gain of freshwater.)

Using the Black Sea volume of 0.6 10 6 km 3 ,

a turnover time of 1000e2000 years can be calculated.

These turnover, or flushing-time calculations,

are very rough, but the contrast with the

165 years turnover time for the Mediterranean

is notable given that these seas are connected.

Independent oceanographic measurements

support the contrast in turnover time between

the Mediterranean and Black Seas, as the bulk

of the Mediterranean water has an oxygen

content of over 160 mmol/kg (>4 ml/L) whereas

the Black Sea water below 200 m has no dissolved

oxygen but a large amount of hydrogen

sulfide (over 6 ml/L), indicative of great age.

The Mediterranean is described as well flushed

or well ventilated, whereas the Black Sea is stagnant

below 95 m. As described in Chapter 9, the

physical reason for the ventilation of the Mediterranean

is that deep water is formed by winter

evaporation and cooling at the surface in the

north. In the Black Sea, precipitation and river

runoff decrease the salinity and density so

much that even severe winter cooling cannot

make the water dense enough to sink. Thus,

regional climate dictates turnover time.

5.3.3. Salt and Freshwater Transports in

the Open Ocean

The concepts of salt and freshwater transports

are important for global water balances.

It rains more in some regions than in others,

and there is more evaporation from the sea

surface in some regions than in others; yet, on

the whole, the salinity distribution of the world

oceans is mostly in steady state. The ocean does

not become saltier over time in evaporation

regions or fresher in net precipitation regions.

(This is not to say that there are no small daily

or seasonal changes, or small and perhaps

important climate changes over the course of

years to decades. Rather, the general distribution

observed in the 1990s, described in Chapter

4, applies as well to several hundred years

ago and perhaps even several hundred years

hence.)

Evaporation, precipitation, and runoff (see

map in Figure 5.4a) affect only the total water

content (freshwater) and not the total amount

of salt. Salt remains, by and large, in the ocean.

(The amounts flung into the air, where they

might become important condensation nuclei

for clouds, are infinitesimal, and have no

impact on ocean salt budgets; the input rates

from weathering are also miniscule.) However,

evaporation, precipitation, and runoff do

change the concentration of salt, that is, the

salinity. Globally there is net evaporation (red

regions in Figure 5.4a) reaching more than

150 cm/year in each of the southeastern

subtropical regions. Net precipitation (blue

regions) is high in the tropics beneath the

ascending air of the atmosphere’s Hadley

circulation (Intertropical Convergence Zone).

Net precipitation is also found in the subpolar

regions of both hemispheres, in the Antarctic

and Arctic.

For the steady-state salinity distribution in

the ocean, freshwater must be transported

within the ocean from regions of net precipitation

to regions of net evaporation. (The rest of

the freshwater cycle is completed through the

atmosphere, which must transport moisture

from regions of net evaporation to those of net

precipitation. The net freshwater transport into

each area of the ocean must exactly balance

the net freshwater transport in the atmosphere

over the same area.) The total volume transports

associated with open ocean freshwater transport

THREE EXAMPLES OF THE TWO CONSERVATION PRINCIPLES 119

(a)

40˚

20˚

0˚

20˚

40˚

60˚

-100

-50

50

100

150

-100 -150

50

0

150

100

0

60˚

180˚

80˚N

-50

120˚W 60˚W 0˚ 60˚E 120˚E 180˚

80˚N

-50

-50

0

-50

0

50 50

150

50

0

100 150

0

-100

-150

50

-50

0

-100

-50

150

150 200

100

100

100 100

0

50

-50

-100

-50

60˚

60˚

0.4 Sv

Northern

40˚

40˚

20˚

–1.0 Sv

0˚ Subtropics/

Tropics

20˚

0.6 Sv

Southern

80˚S

180˚ 120˚W 60˚W 0˚ 60˚E 120˚E

0.3 Sv –0.1 Sv

–0.2 Sv

Pacific

Atlantic

Indian

180˚

80˚S

(b)

FIGURE 5.4 (a) Net evaporation and precipitation (E P) (cm/yr) based on climatological annual mean data (1979e2005)

from the National Center for Environmental Prediction. Net precipitation is negative (blue), net evaporation is positive (red).

Overlain: freshwater transport divergences (Sverdrups or 1 10 9 kg/sec) based on ocean velocity and salinity observations.

This figure can also be found in the color insert. After Talley (2008). (b) Meridional (south to north) freshwater mass transport

(Sverdrups), positive northward, based on ocean velocity and salinity observations (direct) and based on atmospheric

analyses (continuous curves). Source: From Wijffels (2001).

120


are very small compared with the net volume

transports. That is, while the circulation regularly

transports volume at rates of 10 10 6 to

100 10 6 m 3 /sec from one location to another,

freshwater transport into large ocean regions

(gain or loss to the atmosphere in that region)

is on the order of 0.1 to 1.0 10 6 m 3 /sec. As

an example, consider the total freshwater transport

into the central North Pacific between latitudes

25 N and 35 N. This region has

a surface area of 16.2 10 12 m 2 . This is a region

of net evaporation, resulting in higher surface

salinities than in the tropics and subpolar

region. The net freshwater flux F into the ocean

is 0.11 10 6 m 3 /sec, based on climatology

(Figure 5.4a). The circulation in this region is

dominated by a strong western boundary

current called the Kuroshio, which flows northward

along the western boundary in a band that

is about 100 km wide (Section 10.3.1). Most of

this water turns around and flows back southward

across the width of the North Pacific,

mainly within the upper ocean. If we apply

Eq. (5.6) to this situation, using an inflow

Kuroshio volume transport for the upper ocean

of 25 10 6 m 3 /sec and a freshwater gain F of

0.11 10 6 m 3 /sec, we calculate that the salinity

of the southward flow should be about 0.15 psu

lower than the salinity of the northward

Kuroshio. If we go to the actual data for the

upper 1000 m of the ocean, we find that the

average salinity of the Kuroshio is 34.73 psu,

and the average salinity of the southward return

flow is 34.60 psu, which substantiates our

estimate.

Global estimates of meridional (south to

north) ocean freshwater transports (Figure 5.4b)

have been constructed from the total distribution

of evaporation/precipitation/runoff (as in

Figure 5.4a). These freshwater transports are all

less than 1 Sverdrup (1 Sv ¼ 1 10 6 m 3 /sec,

which is equivalent to the units of 1 10 9 kg/

sec in Figure 5.4b). Even the weakest ocean

currents transport much more total water

volume than this. The freshwater transport of

Eq. (5.6) is the excess amount of freshwater at

one location compared with another. Thus, the

freshwater transport is the amount of dilution

or evaporation required to change the salinity

in a given region. In other words, what we are

really calculating (and what we can really

compare with the precipitation, evaporation,

and runoff) is the divergence or convergence

of freshwater transport into a given region.

When calculating these transports over

complete ocean basins, an arbitrary reference

salinity is chosen such that all other salinities

are compared with it and the freshwater transports

calculated accordingly. That is, the arbitrary

constant salinity S o in the denominator in

Eq. (5.6) must be a single number for the whole

global calculation.

The freshwater divergences (net freshwater

transport into the indicated areas) labeled in

Figure 5.4a show more graphically the pattern

of these differences in freshwater transports

with latitude. Where the freshwater transport

increases toward the north, freshwater is being

added to the ocean. This occurs in the rainy

belts from 80 S to about 40 S, from 10 S to

10 N, and from 40 N to 80 N (also see

Figure 5.4b). Where the freshwater transport

decreases toward the north, freshwater is being

removed. These are the evaporation regions of

the subtropics, from 40 S to 10 S and from

10 Nto40 N.

The total freshwater transport for the globe

must balance to nearly zero when averaged

over several years, given that the ocean’s

mean salinity is constant (Section 5.2). Thus

the freshwater transport curves of Figure 5.4b

should start at zero in the south at Antarctica

and end at zero at the North Pole. The “indirect”

estimates of freshwater transports, based

on precipitation and evaporation (Comprehensive

Ocean Atmosphere Data Set, or COADS

and from the National Oceanography Centre,

Southampton or NOCS), do not balance

because they are based on surface observations

of rainfall and evaporation, which have large

CONSERVATION OF HEAT ENERGY; THE HEAT BUDGET 121

errors, especially in the Southern Ocean. The

“direct” estimates, which are calculated from

ocean velocity and salinity observations, fall

along the curves from the indirect estimates.

This indicates that both estimates are detecting

a similar signal. Both panels of Figure 5.4 show

net precipitation at high southern and northern

latitudes, and net evaporation in the subtropics.

Net precipitation at the equatorial latitudes is

evident in Figure 5.4b. The map (Figure 5.4a)

shows, additionally, that the Atlantic and

Indian Oceans are net evaporative, while the

Pacific has net precipitation. This accounts for

the relative saltiness of the Atlantic and Indian

Oceans compared with the Pacific. The

Southern Ocean, south of 30 S, is fresher than

all of these.

Higher evaporation in the Atlantic compared

with the Pacific is associated with the trade

winds. In the Atlantic, they originate from the

dry continent (Mideast and northern Africa),

whereas in the Pacific they have only the narrow

Central American landmass to cross; that is,

there is a zonal atmospheric transport of moisture

from the Atlantic to Pacific (Zaucker &

Broecker, 1992).

5.4. CONSERVATION OF HEAT

ENERGY; THE HEAT BUDGET

5.4.1. Heat Budget Terms

The spatial and temporal variations of ocean

temperatures are indications of heat transfer by

currents, absorption of solar energy, loss by

evaporation, and so forth. The size and character

of the temperature variations depend on

the net rate of heat flow (transport) into or out

of a water body. Heat budgets quantify these

balances. In the following list, the symbol Q

represents the rate of heat flow measured in

Joules per second (Watts) per square meter,

W/m 2 . Subscripts are used to distinguish the

different components of the heat budget. These

components include:

Q s ¼ rate of inflow of solar energy through

the sea surface (shortwave radiation)

Q b ¼ net rate of heat loss by the sea as

longwave radiation to the atmosphere and

space (back radiation)

Q h ¼ rate of heat loss/gain through the sea

surface by conduction (the sensible heat flux)

Q e ¼ rate of heat loss/gain by evaporation/

condensation (the latent heat flux)

Q v ¼ rate of heat loss/gain by a water body

due to currents (the advective term)

Other sources of heat inflow, such as that

from the earth’s interior, change of kinetic

energy of waves into heat in the surf, heat

from chemical or nuclear reactions, and so forth,

are all small and can mostly be neglected relative

to the previously listed terms. The heat

budget for a particular body of water is:

Q T ¼ Q s þ Q b þ Q h þ Q e þ Q v (5.8)

where Q T is the total rate of gain or loss of heat

of the body of water (T refers to total). A schematic

of average values of these terms is shown

in Figure 5.5. Q v , the advective heat flux, is not

shown in Figure 5.5. The advective heat flux,

which is internal to the ocean and is the product

of velocity and temperature gradient (Section

7.3.1), can range from 1 to over 20 units on the

scale of Figure 5.5.

When Eq. (5.8) is used for heat-budget calculations,

numerical values have a positive sign if

the water gains heat and a negative sign if they

represent a heat loss from the sea. Solar heat

flux Q s values are always positive (heat gain)

and longwave back radiation Q b values are

almost always negative (heat loss). Latent heat

fluxes, Q e , are almost always negative. Sensible

heat flux, Q h , can be negative or positive

depending on the sign of the temperature difference

between the air and water. Advective heat

flux, Q v , depends on the difference in temperature

between the water flowing into the region

and water flowing out of the region. These

volume transports are equal by the Conservation

122


SCATTERED

TO SPACE

SHORT-WAVE

RADIATION

FROM SUN

LONG-WAVE

RADIATION

FROM EARTH

TO SPACE

100

TO SPACE

ABSORBED

IN ATOMS

6

TO ATMOSPHERE

FROM

AIR &

CLOUDS

29

19

13 21 7

52

TO SURFACE

4 REFLECTED

Q b

Q e Q h

48

1 PHOTOCHEMICAL

Q s PROCESSES

19

FIGURE 5.5 Distribution of 100 units of incoming shortwave radiation from the sun to Earth’s atmosphere and surface:

long-term world averages.

of Volume (Eq. 5.3), and in actuality differ only

by the very small freshwater gain or loss within

a region. Therefore Q v may be positive (inflow of

warmer water and outflow of colder water) or

negative (opposite case).

Observations of solar radiation, back radiation,

and latent and sensible heat flux are at

points on the sea surface, with units of W/m 2 .

To obtain their total impact on the heat

content in a body of water in Watts, they must

be multiplied by the sea surface area (m 2 )of

the body. (For continuously varying values,

this is actually a sum of the heat fluxes at

each unit area of the sea surface, or equivalently,

an area integral.) Advection through the

sides of the water body must likewise be

calculated at all points on the sides and

summed for each unit area (equivalent to an

area integral).

If the temperature of a body of water is not

changing with time, this does not mean that

there is no heat exchange. It simply means that

the algebraic sum of the terms on the right

side of the heat-budget equation (5.8) is zeroe

net heat inflow equals net heat outflow, an

example of a steady-state condition. If we apply

the heat-budget equation to the world ocean as

a whole, Q v must be zero because then all the

advection is internal and must add up to zero.

Also, if we average over a whole year or number

of years then the seasonal changes average out

and Q t becomes zero. The equation for the

oceans in this case simplifies to

Q s þ Q b þ Q h þ Q e ¼ Q sfc ¼ 0: (5.9)

The global distribution of each of the four

components is examined next. Typical relative

values in Figure 5.5 are only intended as an


indication of the general range and must not

be used for specific calculations. The largest

component is the shortwave radiation Q s

and it is always positive (input of heat into

ocean). The other three components usually

represent a loss of heat from the ocean. The

sensible heat flux Q h varies with time and

place, having maximum values in the northwestern

North Atlantic and North Pacific,

but is generally the smallest term. Latent

heat flux Q e is the second largest term in the

heat-balance equation and has large seasonal

variations. Longwave radiation Q b has the

smallest variations.

The following sections explain how each of

these heat flux components is calculated. The

observed quantities are temperature, humidity,

wind speed, cloud cover, and surface reflectivity.

These are measured from routine observation

stations, ships, ocean buoys, and,

increasingly, from satellites. The heat fluxes are

calculated from these observations, based on

empirical approximations called “bulk

formulae,” with basic physical principles only

loosely at the core. While there has been modest

progress made in our understanding of the

physical principles of turbulent heat exchanges,

this progress has not transitioned into a more

formal analytical description of the individual

heat flux terms. The only alternatives to these

bulk estimates are precise observations of the

individual heat fluxes. Such observations

are sufficiently complex that they cannot be

routinely made.

Local experiments have been carried out at

island stations, moorings, and research ships,

which have provided time series of accurate

measurements of heat exchange, including

diurnal components, and also provided data to

improve the bulk estimates. A long-term goal

is to improve airesea heat exchange estimates

to have errors of less than 10 W/m 2 . Hopefully,

at some point, satellite measurements will

provide global, accurate coverage of each

component of the heat exchange.

Our discussion of the commonly used bulk

estimates closely follows Josey, Kent, and Taylor

(1999), with a summary of satellite techniques

given by Liu and Katsaros (2001).

Maps of annual (and seasonal) averages of

each of the heat flux components as well as

descriptions of their patterns are provided in

Section 5.5 (and in the online supplement to

this chapter).

5.4.2. Shortwave and Longwave

Radiation: Elements of Radiation

Theory

Before discussing the shortwave and longwave

radiation terms, Q s and Q b , certain aspects

of electromagnetic radiation theory must first be

reviewed. First, Stefan’s Law states that all bodies

radiate energy at a rate proportional to the

fourth power of their absolute temperature

T (expressed in Kelvin as K ¼ C þ 273.15).

This energy is in the form of electromagnetic

radiation over a range or spectrum of wavelengths.

Second, the concentration of energy is

not the same at all wavelengths but has

a marked peak at a wavelength of l m given by

Wien’s Law:l m • T ¼ 2897 mm K, where T is again

the absolute temperature (in degrees Kelvin) of

the radiating body. Therefore bodies at higher

temperatures radiate energy at preferentially

shorter wavelengths and vice versa.

The sun has a surface temperature T of

approximately 6000 K and radiates energy

in all directions at a rate proportional to

T 4 ¼ 6000 4 . According to Wien’s Law this energy

is concentrated around a wavelength of 0.5 mm

(1 mm ¼ 10 6 m); 50% of this energy is in the

visible part of the electromagnetic spectrum

(about 0.35 to 0.7 mm), whereas 99% is of wavelengths

shorter than 4 mm. This energy is

referred to as shortwave radiation and is the

source of the Q s term in the heat budget. The

shortwave energy that reaches the ocean (after

passing through the atmosphere and clouds,

but not including the portion that is reflected)

124


is absorbed by the water where it is converted

into heat energy. This increases the temperature

of the water, consistent with its specific heat,

which relates temperature to heat energy

(Section 3.3.2).

The longwave radiation term Q b represents

the electromagnetic energy that is radiated

outward by the earth (land and sea) at a rate

depending on the absolute temperature of the

local surface. Taking an average temperature

of 17 C ¼ 290 K for the sea, it is radiating energy

at a rate proportional to 290 4 . This is a much

smaller rate than for the sun. Wien’s Law then

says that the earth’s peak radiation wavelength

is longer since the temperature is lower. The

wavelength at which the sea radiation reaches

its maximum is about 10 mm (i.e., in the thermal

infrared). About 90% of the sea radiation is in

the wavelength range from 3 to 80 mm.

5.4.3. Shortwave Radiation (Q s )

The sun is the dominant source of energy for

Earth. Most of the sun’s energy is in the visible

(short) wavelength part of the electromagnetic

spectrum. Because of absorption and scattering

in the atmosphere and reflection, only 50% or

less of this radiation reaches the earth’s surface.

In Figure 5.5, this loss of shortwave radiation is

represented by the 29 units lost to space by

scattering from the atmosphere and clouds, 19

units absorbed in the atmosphere and clouds,

and 4 units reflected from the sea surface. The

remaining 48 units enter the sea as the Q s

term of the heat budget. Of these 48 units,

about 29 units reach the sea as direct radiation

from the sun and 19 units as indirect scattered

radiation from the atmosphere (sky radiation).

This distribution represents a long-term worldarea

average; instantaneous values vary diurnally,

seasonally, and with locality and cloud

cover.

Shortwave radiation input to the sea is typically

calculated using two different methods:

from bulk formulae using in situ observations

and from a suite of satellite observations.

Direct measurements of the energy arriving

at the sea surface are made with a pyranometer

(see Section S6.8 located on the textbook Web

site),butitisnotpracticaltodothisoverlarge

areas, or for prediction. Such direct observations

are used to derive the bulk formulae

and develop the satellite algorithms and

calibrations.

The following bulk formula is in general use

for the shortwave radiation flux penetrating the

ocean’s surface, using traditional surface-based

observations of cloud cover:

Q s ¼ð1 aÞQ c ð1 0:62C þ 0:0019q N Þ; (5.10)

in which Q c is the incoming clear-sky solar radiation

(measured above the atmosphere in units

of W/m 2 , and often referred to as the “solar

constant,” even though the value is not constant

in time or space), C is the monthly mean fractional

cloud cover, a is the albedo (fraction of

radiation that is reflected), and q N is the noon

solar elevation in degrees (Taylor, 2000). In practical

calculations, Q s is not allowed to exceed Q c .

The terms in Eq. (5.10) are explained in the next

subsection. Absorption of the radiation by the

sea is discussed in Section 5.4.3.2. Annual

mean values for Q s are shown below in Section

5.5 and Figure 5.11.

Satellite-based shortwave radiation calculations

include observing the incident solar radiation

at the top of the atmosphere, composition of

the atmosphere including water vapor content

and clouds, and information on surface conditions

including atmospheric reflectivity. A major

effort has been put into observing cloud conditions

from satellites (International Satellite

Cloud Climatology Project or ISCCP, and the

Atmospheric Radiation Monitoring or ARM

program). The top of the atmosphere radiation

is measured through the Earth Radiation

Budget Experiment (ERBE). These products

are combined in the Surface Radiation Budget

Program at NASA. An example of a map of

the shortwave radiation from ERBE is shown


in Figure S5.1 located on the textbook Web site.

These shortwave radiation estimates are still

bulk estimates since they involve observations

of the external parameters that affect radiation

rather than being direct measurements of the

radiation penetrating the ocean’s surface.

5.4.3.1. Factors Affecting Shortwave

Radiation Reaching Earth’s Surface

In the expression for shortwave radiation (Eq.

5.10), the central quantity is the incoming clearsky

solar radiation Q c . The rate at which energy

reaches the outside of the atmosphere from the

sun is called the solar constant and, as obtained

from satellite measurements above most of the

earth’s atmosphere, is about 1365e1372 W/m 2 ,

perpendicular to the sun’s rays. In Figure 5.5

this penetration of shortwave radiation is represented

at the top left as 100 units of incoming

shortwave radiation. In addition to direct

sunlight, the sea also receives a significant

amount of energy from the sky, such as sunlight

scattered or absorbed and re-radiated by the

atmosphere, clouds, and so forth. The skylight

component increases in importance at high latitudes.

For instance, at Stockholm (59 N), for

a clear sky in July, about 80% of Q s will be direct

sunlight and only 20% skylight. In December,

only 13% will be direct sunlight and 87%

skylight. However, the total amount of energy

reaching the ground will be less in December

than in July, so 87% of skylight in December

represents a smaller energy flow than the 20%

in July.

The incoming shortwave radiation is

partially reflected upward both from the atmosphere

(clouds and water vapor) and from

Earth’s surface. The albedo, a, in Eq. (5.10) is

the ratio of the radiation reflected from the

surface to the incoming radiation, expressed in

percent. The albedo is also called “reflectance,”

and depends on the properties of the reflecting

surface. The albedo of water (most of the ocean)

is about 10e12% but can be higher or lower

depending on the suspended matter and sea

state. The albedo of sea ice can be much higher

but depends strongly on ice type and whether

it has snow cover. The albedo of new ice can

be as low as 5e20% (see Section 3.9 for sea ice

formation stages), while the albedo for first

year ice can be as high as 60%. The albedo for

multiyear ice without snow cover is about

70%. The albedo of snow is between 60 and

90%. Land surface albedo depends on vegetation

and ranges between 0.5 and 30%. Clouds

also reflect solar radiation and contribute

greatly to the albedo of the whole earth system.

Some values for albedo, extracted from

Payne’s (1972) tables, are given in Table 5.1,

assuming complete transmittance through the

atmosphere (no clouds) and an average sea state

(neither flat nor extremely rough). The smoother

the sea state, the higher the reflection, therefore

the albedo has a (small) wind speed dependence.

The albedo also depends on the sun’s

elevation since direct sunlight is reflected more.

The reflection characterized by albedo is

diffuse. It is not the same as reflection from

a mirror, which is called “specular reflection.”

Specular reflection from the ocean surface is

known as sun glint. Sun glint patterns

(Figure 5.6) are likely caused by variations in

the specular reflection of sunlight from the

ocean’s surface due to variations in the surface

waves caused by variations in the winds.

The bulk formula (5.10) also depends on

cloud cover C, which is the fraction of the sky

covered by clouds. Part of the incoming radiation

is reflected, absorbed, or scattered by

TABLE 5.1

Reflection Coefficient (Albedo) and

Transmission Coefficient (100) for Sea

Water

Sun’s Elevation: 90 60 30 20 10 5

Amount reflected (%): 2 3 6 12 35 40

Amount transmitted

into water (%):

Payne, 1972

98 97 94 88 65 60

126


partially absorbs the radiation. A beam of radiation

of one square meter cross-section covers

an area of one square meter of calm sea surface

when the sun is vertically overhead. At lower

elevations, the beam strikes the sea surface

obliquely and is distributed over a larger area.

The energy density, or amount per square meter

of sea surface, therefore decreases as the sun

moves further from the vertical. (This explains

why the equatorial regions are warm and the

polar regions are cold, and explains why the

seasonal radiation changes are greatest at midlatitudes.)

The absorption is due to the

combined effect of gas molecules, dust in the

atmosphere, water vapor, and so forth. When

the sun is directly overhead (q N ¼ 90 ), the radiation

passes through the atmosphere by the

shortest possible path and the absorption is at

a minimum. When the sun elevation is less

than 90 degrees, the radiation path is longer

and the absorption is greater.

FIGURE 5.6 Sun glint in the Mediterranean Sea. Source:

From NASA Visible Earth (2006a).

clouds. The factor, 0.062, multiplying C was

worked out empirically from direct observations

of shortwave radiation reaching the

surface. One of the biggest sources of error in

computing the shortwave radiation, and in the

whole airesea heat exchange budget, is the

cloud cover estimate. Cloud cover estimates

prior to the 1980s were mainly subjective. Automated

techniques for measuring cloud cover

include satellite observations and radar observations

from land-based networks for weather

prediction and satellite observations; both

methods were introduced in the 1980s.

Finally, the solar radiation reaching the sea

surface in Eq. (5.10) depends on the sun’s elevation,

q N , for two reasons: (1) dependence of the

sea surface area of intersection of a “beam” of

sunlight and (2) dependence of the path length

of the beam through the atmosphere, which

5.4.3.2. Absorption of Shortwave Radiation

in the Sea

Shortwave radiation is not absorbed in the

ocean’s surface skin layer (approximately 10

mm), but instead penetrates to 1e100 m, depending

on wind stirring and incident shortwave

flux magnitude. The absorption decreases exponentially

with depth (Section 3.8). The shortwave

penetration affects the way the mixed

layer restratifies after being mixed by wind or

cooling. Shortwave radiation also penetrates

below the mixed layer in many regions, particularly

at low latitudes. The solar energy penetration

allows for growth of phytoplankton, the

ocean’s chlorophyll-producing plants, in the

near-surface euphotic zone.

The penetration depth of absorption depends

on both the wavelength of the light and the

optical properties of the water. The water’s

optical properties and attenuation of solar radiation

also depend on particle concentration in

the water, which can be composed of sediment

(near-coastal) and plankton (everywhere). In


clear water, the e-folding depth for attenuation

of light is about 50 m (Table 3.2, Figure 5.7). In

water with a heavy load of sediments or biological

particles, for instance during major

plankton blooms, the radiation is absorbed

much closer to the sea surface with an e-folding

scale of less than 5 m.

When more of the solar radiation is

absorbed close to the surface, the surface

temperature increases faster than where the

water is clear. The heating rate can differ by

a factor of 100.

5.4.4. Longwave Radiation (Q b )

The radiation term, Q b , in the heat budget

(Eq. 5.9) is the amount of energy lost or gained

by the sea as longwave (thermal infrared) radiation.

The back radiation is the difference

between the energy radiated outward from the

sea surface and the longwave radiation received

by the sea from the atmosphere. Both the sea

surface and the atmosphere radiate approximately

as “black bodies,” at a rate proportional

to the fourth power of their absolute temperature,

according to Stefan’s Law (Section 5.4.2).

The outward radiation from the sea at these

wavelengths is generally greater than the

inward longwave thermal radiation from the

atmosphere, so Q b usually presents a loss of

energy from the sea (hence the subscript “b”

for back radiation). Q b is expressed through

the following empirical bulk formula, evaluated

by Josey et al. (1999) as the most accurate of

several differing formulations:

Q b ¼ 3s SB T 4 w ð0:39 0:05e1=2 Þð1 kC 2 Þ

þ 43s SB T 3 w ðT w T A Þ: (5.11)

Here 3 is the emittance of the sea surface (0.98);

s SB is the Stefan-Boltzmann constant (5.67

10 8 W m 2 K 4 ); T w is the surface water

temperature in Kelvin; T A is the air temperature

in Kelvin, which is usually measured on a ship

near the bridge that may be as much as 8e10

m above the sea surface; e is the water vapor

pressure; k is a cloud cover coefficient that is

determined empirically and that increases with

latitude; and C is the fractional cloud cover (as

in Section 5.4.3).

Maps of longwave radiation from the ocean’s

surface are shown below in Section 5.5 (mean in

Figure 5.11 in this chapter and seasonal variations

in Figure S5.4 of the online supplement seen on

the textbook Web site).

FIGURE 5.7 Absorption of shortwave radiation as

a function of depth (m) and chlorophyll concentration,

C (mg m 3 ). The vertical axis is depth (m). The horizontal

axis is the ratio of the amount of radiation at depth z to the

amount of radiation just below the sea surface, at depth “0.”

Note that the horizontal axis is a log axis, on which exponential

decay would appear as a straight line. ÓAmerican

Meteorological Society. Reprinted with permission. Source:

From Morel and Antoine (1994).

5.4.4.1. Factors Affecting Longwave

Radiation

The first term in the longwave heat flux (Eq.

5.11) is essentially the black body radiation

term (product of the Stefan-Boltzmann constant

and the fourth power of the water temperature).

Pure Stefan’s Law assumes a perfect black body.

Each actual object has its own “gray-body”

128


emittance 3. Emittance is always a fraction less

than 1, depending on the molecular structure

of the body. Water has a relatively high emittance,

hence the value of 0.98 given earlier.

While Q b basically follows the fourth power

of temperature, there are several parts to Eq.

(5.11) that must be determined empirically. It is

difficult to directly measure Q b over large areas,

although it can be measured locally with a radiometer

(see the description in Chapter S6,

Section S6.8 located on the textbook Web site).

Early studies estimated the heat loss using

data published by Ångström (1920). He showed

that the net loss depends upon the absolute

temperature of the sea surface and upon the

water-vapor content of the atmosphere immediately

above it. The surface temperature, T W ,

determines the rate of outward flow of energy.

The water vapor pressure e effectively determines

the inward flow from the atmosphere

because the water vapor in the atmosphere is

the main source of its longwave radiation.

The two empirical terms multiplying the

T W 4 term include water vapor pressure e and

cloud cover C. Water vapor in the atmosphere

radiates longwave energy back to the sea, thus

reducing the net longwave radiation from the

sea. Clouds reduce the longwave back radiation

from the sea to space. This effect of cloud cover

is familiar on land where the frost that results

from radiative cooling is more frequent on clear

nights than on cloudy ones. With the sky

completely covered with substantial cloud (C ¼ 1),

in a region where the factor k ¼ 0.2, this cloud

exponent factor is 0.2. The reason for the big

difference between clear and cloudy conditions

is that the atmosphere, particularly its watervapor

content, is relatively transparent to radiation

in the range from about 8 to 13 mm, which

includes the peak of the radiation spectrum for

a body at the temperature of the sea. In clear

weather, energy at 8 13 mm passes through the

more transparent “wavelength window” in the

atmosphere and out into space where it is lost

from the earth system.

Global cloud cover is represented by an

image from NASA’s MODIS instrument

(Figure 5.8). (Climatological cloud cover for

the four seasons is also available as Figure S5.3

from the online supplement located at the textbook

Web site.) Over the oceans, cloud cover is

high in the polar regions and in zonal stripes

in the Intertropical Convergence Zone. Cloud

cover is low in the subtropical regions, where

FIGURE 5.8 Cloud fraction (monthly average for August, 2010) from MODIS on NASA’s Terra satellite. Gray scale

ranges from black (no clouds) to white (totally cloudy). Source: From NASA Earth Observatory (2010).


we will see that evaporation greatly dominates

precipitation.

Returning to the expression for longwave

radiation (Eq. 5.11), the second term is proportional

to the difference between the water

temperature and the air temperature just

above the water. This represents the atmospheric

feedback to the longwave radiation

radiated at the sea surface due to atmospheric

moisture (Thompson & Warren, 1982). The

temperature difference is generally small, so

the correction is only important in a few

special regions. An example is where warm

surface ocean currents flow under a cold overlying

atmosphere, such as in the western

boundary currents in the North Pacific and

North Atlantic.

5.4.4.2. Sea Surface Temperature and

Penetration Depth of Longwave Radiation

Longwave radiation depends mainly on sea

surface temperature (SST). But what is the appropriate

measure of SST? From what depth ranges

is the sea surface emitting longwave radiation?

Water is nearly opaque to longwave radiation.

The incoming longwave radiation from the

atmosphere is absorbed in the top millimeters,

unlike incoming shortwave radiation that penetrates

much deeper (Section 5.4.3.2). Thus, the

outward longwave radiation is determined by

the temperature of the literal surface or skin

temperature of the sea, which is less than one

millimeter thick.

The bulk surface temperature, characterizing

the upper few meters of the ocean and

measured with in situ instruments (such as

thermistors on buoys or in engine intake water)

is not the skin temperature. Instead it represents

the temperature about 0.5 to 1 m beneath

the surface. Skin and bulk temperatures are

equivalent only when the bulk layer is well

mixed vertically, as in the presence of breaking

surface waves and strong surface winds.

Models of skin layer physics (Castro, Wick, &

Emery, 2003; Wick, Emery, Kantha, & Schluessel,

1996; Wick, 1996) suggest that the difference

between the skin and bulk temperatures is

proportional to the wind speed, which affects

surface waves, and net airesea heat flux, which

affects mixing.

Regardless of the actual physical process, the

empirical bulk formula (5.11) was developed to

be used with the traditional bulk SST and not

with the skin layer temperature.

5.4.4.3. Outgoing Longwave

Radiation (OLR)

“Outgoing longwave radiation” (OLR) is the

total infrared radiation at wavelengths of 5 to

100 mm that escapes from the top of Earth’s

atmosphere back into space. Most of this longwave

energy is emitted from the surface of the

ocean while some of it is emitted by the land

and the atmosphere. The OLR can be computed

from infrared satellite data and is generated as

a standard product by the National Oceanic

and Atmospheric Administration (NOAA)

from their polar-orbiting satellites. It was also

a product of the ERBE program (Section 5.4.3).

An example of satellite-derived OLR is shown

in Figure 5.9. Maximum OLR dominates the midlatitudes

in all ocean regions, where cloudiness is

low, broken only by equatorial minima that are

related to the cloudy regions of the atmosphere’s

Intertropical Convergence Zone along 5 N to

10 N and to the Walker circulation (Section

7.9.2). Associated with the latter, an equatorial

minimum dominates the Indonesian archipelago

and extends into the western Pacific.

5.4.5. Effect of Ice and Snow Cover on

the Radiation Budget

When the sea surface becomes covered with

a layer of ice, and especially if snow covers the

ice, there is a marked change in the heat-radiation

budget as described in the section on shortwave

radiation. First of all, ice cover significantly

reduces heat exchange between the ocean and

atmosphere d 1 meter of ice will almost totally

130


FIGURE 5.9 Outgoing Longwave

Radiation (OLR) for Sept.

15eDec. 13, 2010. This figure can

also be found in the color insert.

Source: From NOAA ESRL (2010).

insulate the ocean. However, sea ice is always

moving and is full of leads (breaks that expose

open water). Ocean heat loss in leads is intense,

and new ice forms quickly. Therefore heat

budgets in ice-covered areas must take into

account ice type, thickness, and concentration

(percent coverage). Where there is ice cover, the

heat flux into the water column becomes

Q ¼ kðT w T s Þ=h (5.12)

where k is the ice conductivity, T w is the water

temperature just below the ice, T s is the temperature

at the upper surface of the ice, and h is the

ice thickness. The water temperature is assumed

to be at the freezing point. The surface temperature

of the ice and its thickness are more difficult

to determine without local measurements. The

availability of satellite-based observations of ice

cover using microwave imagery has greatly

improved knowledge and regular mapping of

ice cover. The type and thickness of ice is harder

to estimate from satellite observations, but there

are various approaches for obtaining information

about the distribution of new, first-year,

and multiyear ice that are then translated to estimates

of ice thickness. Remote sensing of ice

thickness remains one of the major hurdles in

airesea flux estimation at high latitudes.

A second important effect of ice is that it is

highly reflective (high albedo), much more so

than water (lower albedo). This mostly impacts

the incoming shortwave radiation (Section 5.4.3).

Sea ice and snow also reflect most incident solar

radiation, so they have a high albedo in comparison

with open water (Figure 5.10). However,

the back radiation (Q b ) heat loss is much the

same for ice as for water (due to the relative

similarity of surface temperature). Therefore

there is a smaller net gain (Q s Q b ) by ice and

snow surfaces than by water. Thus as sea ice

melts back, more solar radiation is absorbed

by the water, which then warms and causes

more ice to melt. This is a positive feedback,

which is called ice-albedo feedback.

The ice balance in a region such as the Arctic is

somewhat delicate (Section 12.7). If the sea ice

were melted all the way at a given time, the

increased net heat gain (Q s Q b )couldmaintain

an ice-free Arctic Ocean. On the other hand, this

could increase evaporation, which would

increase precipitation in the high northern latitudes,

increasing snow cover and high latitude

albedo, which would have a cooling effect. The

present situation of ever-decreasing Arctic Ocean

ice cover (Section 12.8) suggests that the icealbedo

feedback effect dominates.


Ice formation;

Reduced % of

open water

Surface albedo (reflectivity)

Absorbed radiation

(+ positive feedback)

Ice-albedo feedback

Incident radiation

Ocean

Incident radiation

Reflected

radiation

Ice

Absorbed radiation

warms water

FIGURE 5.10 Ice-albedo feedback. In

the feedback diagram, arrowheads

(closed circles) indicate that an increase

in one parameter results in an increase

(decrease) in the second parameter. The

net result is a positive feedback, in

which increased sea ice cover results in

ocean cooling that then increases the ice

cover still more.

Ocean temperature

5.4.6. Evaporative or Latent Heat Flux

(Q e )

Evaporation requires a supply of heat from

an outside source or from the remaining liquid.

(This is why one feels cold when wet after

swimming, as evaporation of the water requires

heat.) Therefore evaporation, besides implying

loss of water volume, also implies loss of heat.

The rate of heat loss is

Q e ¼ F e $L (5.13)

where F e is the rate of evaporation of water in kg

sec 1 m 2 and L is the latent heat of evaporation

(vaporization) in kilojoules (1 kJ ¼ 10 3 J). For pure

water, L depends on the temperature of the water

Tin C: L ¼ (2494e 2.2 T) kJ/kg. At 10 C, the

latent heat is about 2472 kJ/kg, which is larger

than its value of 2274 kJ/kg (540 cal/gm) at the

boiling point. While one can see steam after the

boiling point is reached, more volume is being

evaporated at temperatures well below boiling.

The average amount of evaporation F e from

the sea surface is about 120 cm/yr, in other

words, the equivalent of the sea surface sinking

more than 1 m per year. Local values range

from an annual minimum of as little as 30 to

40 cm/yr in high latitudes to maxima of 200

cm/yr in the tropics associated with the trade

winds. This decreases to about 130 cm/yr at

the equator where the mean wind speeds are

lower.

How is the evaporation rate F e determined?

Therateofwaterlossfromapanofwatercan

be measured, but this has serious practical

difficulties. For large area estimates and for

prediction, a formula using easily measured

parameters is desirable. Evaporation is basically

a diffusive process that depends on how

water vapor concentration changes with height

above the sea surface and on the processes that

cause diffusion. In Section 5.1.3, we discussed

eddy diffusion, which is analogous to molecular

diffusion, except that the turbulence in

the air or water is the process that diffuses

properties rather than movement of individual

molecules. Air turbulence controls the diffusion

that creates evaporation. Air turbulence

depends on wind speed, therefore we expect

the evaporation rate to depend on wind speed.

(This explains why we feel cooler when the

wind is blowing.)

A semi-empirical (“bulk”) formula for evaporation

that depends on wind speed and the

132


vertical change in water vapor content is

frequently used:

F e ¼ r C e uðq s q a Þ: (5.14)

Here r is the density of air, C e is the transfer

coefficient for latent heat, u is the wind speed

in meters per second at 10 m height, q s is 98%

of the saturated specific humidity at the sea

surface temperature, and q a is the measured

specific humidity. The factor of 98% for saturated

humidity over seawater compensates for

the salinity. The saturated specific humidity

over distilled water may be obtained from tables

of physical or meteorological constants.

Specific humidity is the mass of water vapor

per unit mass of air, in g/kg. Saturated specific

humidity is the maximum weight of water

vapor that the air can hold for a given temperature.

Relative humidity is the amount of water

vapor divided by the saturated water vapor

expressed in percent. Therefore it is equal to

the specific humidity divided by the saturated

specific humidity.

The empirical (“bulk”) formula for heat flux

due to evaporation (latent heat flux) is therefore,

in W/m 2 ,

Q e ¼ F e L ¼ r C e uðq s q a ÞL: (5.15)

In most regions of the ocean, the saturated

specific humidity q s is greater than the actual

specific humidity q a . Since all of the other terms

in the practical formula (5.15) are positive, evaporative

heat loss occurs in these regions. As long

as the sea temperature is more than about 0.3 K

greater than the air temperature, there will be

a loss of heat from the sea due to evaporation.

Only in a few regions is the reverse the case,

when the air temperature is higher than the

sea temperature and the humidity is sufficient

to cause condensation of water vapor from the

air into the sea. This results in a loss of heat

from the air into the sea. The Grand Banks off

Newfoundland and the coastal seas off northern

California are examples of regions where the

latent heat flux Q e is into the sea (numerically

positive). The fog in these regions is the result

of cooling the atmosphere.

The heat loss due to evaporation occurs from

the topmost layer of the sea, like longwave radiation

and unlike shortwave radiation heat gain.

In models of airesea heat exchange, the evaporative

heat loss is applied to the ocean surface

element. It should also be noted that this latent

heat flux term is usually the largest of the heat

flux terms other than the incoming shortwave

radiation. Unlike the short- and longwave terms

discussed previously, there is no straightforward

estimate of latent heat exchange from

satellite observations. Therefore the latent flux

is best estimated from in situ measurements,

which will be presented later.

5.4.7. Heat Conduction or Sensible

Heat Flux (Q h )

The last process that we discuss for heat

exchange between the sea surface and atmosphere,

sensible heat flux, arises from a vertical

difference (gradient) in temperature in the air

just above the sea. This is perhaps the simplest

of the heat flux terms in Eq. (5.9) to understand.

If temperature decreases upward, heat will be

conducted away from the sea, resulting in an

ocean heat loss. If the air temperature increases

upward, heat will be conducted into the sea.

The rate of loss or gain of heat is proportional

to the air’s vertical temperature gradient, and

to the heat conductivity (for which we use an

eddy diffusivity or conductivity, A h ):

Q h ¼ A h c p dT=dz (5.16)

As described for eddy diffusion of water vapor,

the eddy conductivity A h depends on wind

speed. The vertical gradient of temperature is

measured as a difference between the sea

surface temperature and the air temperature.

The bulk formula for sensible heat flux is

written, with units of W/m 2 , as:

Q h ¼ r c p C h uðT s ðT a þ gzÞÞ (5.17)

GEOGRAPHIC DISTRIBUTION AND TEMPORAL VARIATION OF THE HEAT-BUDGET TERMS 133

where r is the air density; C h is the transfer coefficient

for sensible heat (derived from the eddy

conductivity); u is the wind speed in meters

per second at 10 m height; T s is the surface

temperature of the ocean (assumed to be equal

to the air temperature immediately above the

ocean surface); T a is the air temperature; z is

the height where T a is measured; and g is the

adiabatic lapse rate of the air, which accounts

for changes in air temperature due to simple

changes in height and pressure.

The sensible heat flux cannot be estimated

from satellite measurements and must be estimated

from in situ data using these bulk

formulae. Global maps of such computations

are presented in Section 5.5.

5.4.8. Dependence of the Latent and

Sensible Heat Transfer Coefficients on

Stability and Wind Speed

Latent and sensible heat transfers are

computed using bulk formulae like Eqs. (5.15)

and (5.17) using in situ observations. Values for

the transfer coefficient for sensible heat for

various airesea temperature differences and

different wind speeds are given in Table 5.2

(from Smith, 1988). The transfer coefficients in

the two expressions, C e and C h , depend on

whether the ocean is warmer or colder than the

atmosphere, and whether the atmosphere is

undergoing deep or vigorous convection. If the

sea is warmer than the air above it, there will

be a loss of heat from the sea because of the direction

of the temperature gradient. However,

larger scale atmospheric convection will increase

the heat transfer away from the sea surface.

Convection occurs because the air near the

warm sea is heated, expands, and rises, carrying

heat away rapidly. In the opposite case, where

the sea is cooler than the air, convection does

not occur. Therefore, for the same temperature

difference between sea and air, the rate of heat

TABLE 5.2

(T s L T a ) (K)

Some Values for the Sensible Heat Transfer

Coefficient, C h , as Functions of (T s T a )

and Wind Speed u

Wind Speed u in m/sec

2 5 10 20

10 d d 0.75 0.96

3 d 0.62 0.93 0.99

1 0.34 0.87 0.98 1.00

þ1 1.30 1.10 1.02 1.00

þ3 1.50 1.19 1.06 1.01

þ10 1.87 1.35 1.13 1.03

Smith, 1988

loss when the sea is warmer is greater than the

rate of gain when the sea is cooler.

For example, for (T s T a ) ¼ 1 K, that is, for

the sea cooler than the air (T s < T a ), the stability

in the air is positive. When the sea is warmer

than the air, for example, (T s T a ) ¼þ1 K, the

air is unstable and heat conduction away from

the sea is promoted, so the transfer coefficient

is larger than 1. The blank areas in the table

are for highly stable conditions (unusual) where

Smith’s analysis breaks down.

For the transfer coefficient for evaporation,

Smith commented that measurements in open

sea conditions are relatively rare, particularly

for high wind speeds. After reviewing the available

data, he recommended C e ¼ 1.20 C h . That

is, the physical process causing the transfer coefficient

is similar for both evaporation and heat

conduction.

5.5. GEOGRAPHIC DISTRIBUTION

AND TEMPORAL VARIATION OF

THE HEAT-BUDGET TERMS

Maps and description of the four components

of the surface heat flux are given in this section.

A monthly climatology 1 of fluxes from the

1 A monthly climatology is the average of values from a given month over all the years of analysis (see Section 6.6.2).

134


National Oceanography Center, Southampton

(NOCS; Grist & Josey, 2003) is used here, but we

could use other available climatologies for

description of the basic patterns and magnitudes

as well. The NOCS fluxes are based on carefully

quality-controlled ship observations covering

more than a century, from the COADS database.

The annual mean heat flux components are

shown in the text; four monthly maps representing

the seasonal variations for each component

and the net heat flux are provided in the online

supplement as Figures S5.2eS5.7 along with

seasonal cloud cover over the oceans (see

Figure S5.3 based on data from da Silva, Young,

& Levitus, 1994) since it has a large impact on

the shortwave and longwave radiation.

Large and Yeager (2009) created a product in

which input fields from reanalysis and satellite

observations are systematically adjusted and

combined to balance heat and freshwater

budgets. Their fluxes are used for the mean

buoyancy flux map in Figure 5.15. The mean

heat and freshwater airesea flux maps that are

combined for the buoyancy flux are shown in

Figure S5.8 seen on the textbook Web site.

Other commonly used global airesea flux

products are from weather prediction models

that have been systematically “reanalyzed” to

create consistent data sets over many years of

runs. The two major reanalyses are from the

National Centers for Environmental Prediction

(NCEP; Kalnay et al., 1996) and from the European

Centre for Medium-range Weather Forecasts

(ECMWF).

5.5.1. Annual Mean Values of the

Components of the Heat Budget

The four components of the airesea heat flux

are shown in Figure 5.11 and their sum, the net

heat flux, in Figure 5.12. Basically heat is added

to the ocean through shortwave radiation

(incoming sunlight) and mostly lost from the

ocean through the other three components.

The shortwave radiation, Q s , (Figure 5.11a)

depends mainly on latitude. It adds 50 to

150 W/m 2 of heat to the ocean in subpolar latitudes,

and 150 to almost 250 W/m 2 in the

subtropics and tropics (Figures 5.11a and

Figure S5.2 located on the textbook Web site).

Shortwave radiation does vary from exact

dependence on latitude. The highest shortwave

flux, of almost 250 W/m 2 , is in patches in the

eastern tropical Pacific and the western tropical

Indian Ocean along the Arabian Peninsula.

Lower tropical shortwave heat gain is found in

wide regions in the eastern parts of the oceans.

For instance, in the eastern Pacific, the 200 W/m 2

contours bulge toward the equator in both the

North and South Pacific. These variations in

shortwave radiation are due to spatially varying

cloud cover (Figure 5.8), which partially blocks

incoming shortwave radiation.

Longwave radiation, Q b , (Figure 5.11b)

results in net heat loss from the ocean, even

though there is some longwave radiation into

the ocean from the atmosphere. The radiation

heat loss centers around 50 W/m 2 over much

of the earth. Longwave radiation does not

have a large range of values because it depends

on the absolute temperature (Kelvin and not

Celsius). The relative changes in temperature

are just a small fraction of the total temperature.

The relative humidity also does not change

much over the sea. For instance, a seasonal

change of sea temperature from 10 to 20 C

changes the outward radiation proportional to

the ratio 293 4 /283 4 or about 1.15, only a 15%

increase. At the same time the atmospheric radiation

inward would increase and reduce the net

rate of loss below this figure. The small seasonal

and geographic changes of Q b contrast with the

large seasonal changes of Q s . Variations in longwave

radiation with latitude follow cloud cover

rather than surface temperature. Longwave

radiative heat loss is highest in the subtropics

(>50 W/m 2 ) where the cloud cover is smaller

than in the equatorial and subpolar regions.

Latent heat flux, Q e , (Figure 5.11c) is the

largest heat loss component at all latitudes. It


(a)

180 90 W 0 90 E 180

180 90 W 0 90 E 180

60 N

Short

Long

100

100

<

wave

-50

30 N

-50

> wave

200

<

0

> > >

200

>

200

30 S

200

>

-50

100

-50

<

60 S

30 S -100 -150 -100

0 -15

(c)

(d)

60 N

Latent -15 Sensible

30 N -100

0

60 S

(b)

60 N

-50

> 30 N

0

30 S

60 S

60 N

-15 30 N

0

30 S

60 S

180 90 W 0 90 E 180

180 90 W 0 90 E 180

–200 –150 –100 –50 0 50 100 150 200

Mean heat fluxes (W/m 2 ) (SOC)

FIGURE 5.11 Annual average heat fluxes (W/m 2 ). (a) Shortwave heat flux Q s . (b) Longwave (back radiation) heat flux

Q b . (c) Evaporative (latent) heat flux Q e . (d) Sensible heat flux Q h . Positive (yellows and reds): heat gain by the sea. Negative

(blues): heat loss by the sea. Contour intervals are 50 W/m 2 in (a) and (c), 25 W/m 2 in (b), and 15 W/m 2 in (d). Data are from

the National Oceanography Centre, Southampton (NOCS) climatology (Grist and Josey, 2003). This figure can also be found

in the color insert.

is strongest (more than 100 W/m 2 heat loss) in

the subtropical regions of low cloud cover,

where dry air descends from aloft onto the

oceans. Variations in Q e from west to east in

the stormy regions of the western North

Atlantic and western North Pacific are associated

with dry winds blowing off the continents,

creating greater latent heat flux. Latent heat loss

also depends mildly on temperature since

warmer water evaporates more easily than

colder water.

Sensible heat flux, Q h , (Figure 5.11d) is

usually the smallest of all of the components

over most of the ocean ( 15 to 0 W/m 2 ). It is

slightly larger in the western North Atlantic

and western North Pacific, where latent heat

loss is also large. This is because the airesea

temperature contrast is large in these regions,

where cold air blows off the continent over the

warm western boundary currents. Sensible

heat loss is much larger in winter in some

regions than is apparent from these maps of

the mean components. A small amount of heat

gain from sensible heat exchange is shown in

the Antarctic, but the reader should understand

that these data are especially poor.

The total airesea heat flux based on the NOCS

climatology (Figures 5.12 and 5.13) is the sum of

the four components of Figure 5.11. (Total

airesea heat flux from a different climatology is

136


180 120 W 60 W 0 60 E 120 E 180

80 N 80 N

-100

60 N 60 N

-50

0 0

-0.1

-50 0.6

40 N -150 0.8

40 N

-0.2

-100

0.8

1.2

20 N 0 20 N

0.7 0

1.2

0 100 50

50

0

50

0.6

0

0

20 S 0

-0.4

20 S

0

0.3

-1.3

-50 -50

40 S 0.1

0.5

-50 0

40 S

0

0

0

0

60 S 60 S

80 S 80 S

180 120 W 60 W 0 60 E 120 E 180

–200 –150 –100 –50 0 50 100 150 200

Annual mean net heat flux (W/m 2 ) (NOCS, 2003)

FIGURE 5.12 Annual average net heat flux (W/m 2 ). Positive: heat gain by the sea. Negative: heat loss by the sea. Data

are from the NOCS climatology (Grist and Josey, 2003). Superimposed numbers and arrows are the meridional heat

transports (PW) calculated from ocean velocities and temperatures, based on Bryden and Imawaki (2001) and Talley (2003).

Positive transports are northward. The online supplement to Chapter 5 (Figure S5.8) includes another version of the annual

mean heat flux, from Large and Yeager (2009). This figure can also be found in the color insert.

shown in Figure S5.8 on the textbook Web site,

and is combined with freshwater fluxes to

produce the total buoyancy flux seen in

Figure 5.15 in this chapter. Comparison of these

two maps provides a useful indication of uncertainty

in the total flux.) The ocean gains heat in

the tropics and loses heat at higher latitudes.

The most heat is gained along the equator, especially

in the eastern Pacific. Regions of net heat

gain spread away from the equator on the

eastern sides of the oceans, in the regions where

colder water is upwelled to the surface. Patches

of heat gain are found in the Antarctic, corresponding

to regions where the sensible heat

flux is into the ocean in Figure 5.11d, but where

there are also almost no winter data to balance

observations of summer heat gains.

The greatest mean annual heat losses occur in

the Gulf Stream region of the North Atlantic, the

Kuroshio of the North Pacific, and in the Nordic

Seas north of Iceland and west of Norway. In the

Southern Hemisphere, the Agulhas/Agulhas

Return Current is the region of largest heat

loss. The Brazil and East Australian Currents

are marked by heat loss, as is the Leeuwin

Current, which is the only southward-flowing

eastern boundary current. Each of these regions

is characterized by fast poleward flow of warm

water that loses its heat locally rather than over

a large region; the highest heat losses are where


Heat input per 1° latitude band (PW)

0.8

0.6

0.4

0.2

–0.0

–0.2

–0.4

Shortwave

Latent

80°S 60°S 40°S 20°S 0° 20°N 40°N 60°N 80°N

Latitude

these warm waters encounter dry, cold continental

air in winter (Gulf Stream and Kuroshio).

The local values of each term in the heat

budget and the total heat flux can be summed

all the way around the earth in each latitude

band (Figure 5.13). The numbers used in this

figure are the heat gain or loss in each 1 latitude/longitude

square multiplied by the area

of the square. Then all the heat gains or losses

for a single latitude are added together to get

the total heat gain or loss in Watts for each latitude

band. The total (net) heat gain or loss in

each latitude band (Figure 5.13) is the sum of

these four. As is apparent in the maps in

Figure 5.11, shortwave radiation heats the ocean

while the other three components cool it. Latent

heat loss is the largest of these three heat losses,

but longwave radiation is also significant. The

sensible heat contribution is very small.

All of the heat budget components are larger

in the Southern Hemisphere than in the

Northern Hemisphere. Part of the reason may

be there is more ocean area in the Southern

Hemisphere. Shortwave radiation is also

slightly skewed because Earth is closer to the

sun in January, which is the summertime in

the Southern Hemisphere (Section 5.5.2). The

net heat exchange is positive (heating) in the

Net

Sensible

Longwave

FIGURE 5.13 Heat input through the sea surface (where

1PW¼ 10 15 W) (world ocean) for 1 latitude bands for all

components of heat flux. Data are from the NOCS climatology

(Grist and Josey, 2003).

low latitudes and negative (cooling) at higher

latitudes. The net heat flux is also skewed,

with slightly more heating in the low-latitude

Southern Hemisphere. The net heat flux distribution

requires a transport of excess heat from

low to high latitudes in order to maintain

a climatologically steady state (Section 5.6).

5.5.2. Seasonal Variations in the

Components of the Heat Budget

Each component of the heat budget varies in

time. Components can vary on short (diurnal) to

long (decades to millennia) timescales, but at

mid-latitudes, seasonal variation has the largest

amplitude and impact on weather. Seasonal

maps for each of the airesea heat flux components,

and also cloud cover, are shown in

Figures S5.2eS5.7 located on the textbook Web

site. Only a short summary of salient results is

presented here.

The march of the seasons from summer to

winter is apparent in the shortwave radiation

maps (see Figure S5.2 located on the textbook

Web site), with much more shortwave radiation

reaching the summertime hemisphere.

Northern Hemisphere winter radiation is higher

than Southern Hemisphere winter radiation

because Earth is closer to the sun in January

than in July, so the winter seasons are not identical.

(This has paleoclimate ramifications, as it

highlights the importance of the exact orbit of

Earth, which varies slowly, changing the distribution

of incoming radiation.)

For longwave radiation, seasonal variations

are small (see Figure 5.4 of the online supplemental

material), just as geographical variations

are small (Section 5.5.1 of this chapter), because

the radiation variations depend only weakly on

temperature. Within these small variations, longwave

radiation is larger in the winter hemisphere

than in the summer hemisphere, mainly because

of greater cloud cover in summer.

Latent heat loss through the seasons is strongest

(most negative) in the winter (Figure S5.5

138


located on the textbook Web site). The Northern

Hemisphere’s western boundary currents (Gulf

Stream and Kuroshio) are clearly marked by

their large latent heat loss in fall and winter.

Southern Hemisphere latent heat loss is less

associated with the western boundary currents

and more associated with the central subtropical

gyres where evaporation is high.

Despite the relatively small contribution of

sensible (conductive) heat flux to the annual

mean net heat flux, its seasonal variations

(Figure S5.6 located on the textbook Web site)

are striking because of the sign change in the

temperature difference between the ocean and

overlying air. Sensible heat flux causes heat

loss in winter when the overlying air is colder

than the ocean. Significant heat loss is found

in the western boundary current regions,

more than 100 W/m 2 in the climatological

maps, and much higher in individual storms.

Sensible heat flux heats the ocean in higher latitudes

in summer, when the air is warmer than

the ocean.

5.6. MERIDIONAL HEAT

TRANSPORT

The ocean gains heat in the tropics between

30 S and 30 N and loses heat at higher latitudes,

when zonally averaged (along latitudes) and

over the year (Figure 5.13). There is net radiative

heat gain by the ocean at lower latitudes

because the solar radiation Q s is greater than

the longwave radiation Q b between the equator

and about 30 to 40 N(Figures 5.11 and 5.13). At

higher latitudes, longwave heat loss is greater

than shortwave heat gain; evaporative heat

loss is also higher, and therefore, overall, there

is a net heat loss. These zonally averaged

patterns of heat gain and loss also apply to the

atmosphere.

Because the oceans as a whole are not warming

or cooling (except for the very small rates

that are indeed significant for climate studies),

we expect a nearly exact balance between heat

gain and loss when summed over all of the

ocean area. This requires a net flux of heat

toward both poles, from the lower latitudes of

net heat gain to the higher ones of net heat

loss. This poleward heat flux is carried by both

the ocean and atmosphere. Both transport

warm water or air toward the pole and cooler

water or air toward the equator, although not

symmetrically in all oceans (see the following

section). The atmosphere carries much more of

this heat than the ocean (see the following

section), but the ocean’s role in heat transport

is important, especially at low to mid-latitudes.

The meridional (north-south) heat transports

by currents within the oceans are calculated in

three separate and independent ways. The first

two methods are indirect, in which the ocean’s

heat transport is inferred from heat balances

rather than from measurements of ocean

velocity and interior temperature.

The first indirect method uses the surface

heat fluxes (as in Figures 5.12 and 5.13), which

are summed within latitude bands. The ocean

must then transport enough heat into or out of

each latitude band to balance the heat lost or

gained through the sea surface in that band.

The second indirect method starts with the

heat exchange of the whole Earth’s system

with outer space; that is, at the top of the

atmosphere (TOA). Then heat transports are

calculated for the atmosphere from meteorological

data. The ocean’s heat transport is the

TOA flux minus the atmosphere’s flux. The

first such estimates (Oort & Vonder Haar,

1976), based on 9 years of radiation measurements

from satellites, showed the ocean heat

transport to be a maximum of 60% of the total

at 20 N, 25% at 40 N, and 9% at 60 N.

However, as observations have improved,

and especially with the addition of a special

satellite mission to measure radiation at the

TOA (ERBE), estimates of the total heat transport

and atmospheric heat transport have

become higher, leaving the ocean heat

MERIDIONAL HEAT TRANSPORT 139

(b)

(a)

2

Northwatrd heat transport (PW)

1

0

–1

–2

NOCS (2003) climatology

with 29 W/m 2 adjustment

80° 40° 0° 40° 80°

Latitude

FIGURE 5.14 Poleward heat transport (W) for the world’s oceans (annual mean). (a) Indirect estimate (light curve)

summed from the net airesea heat fluxes of Figures 5.12 and 5.13. Data are from the NOCS climatology, adjusted for net zero

flux in the annual mean. Data from Grist and Josey (2003). A similar figure, based on the Large and Yeager (2009) heat fluxes

is reproduced in the online supplement (Figure S5.9). (b) Summary of various direct estimates (points with error bars) and

indirect estimates. The direct estimates are based on ocean velocity and temperature measurements. The range of estimates

illustrates the overall uncertainty of heat transport calculations. ÓAmerican Meteorological Society. Reprinted with

permission. Source: From Ganachaud and Wunsch (2003).

transport at about the same original values, but

a smaller fraction of the whole (Trenberth &

Caron, 2001).

The third type of ocean heat transport calculation

is direct, based on measuring velocity

and temperature across a whole cross-section

of the ocean through which there is zero net

mass transport. (If there is net transport, then

additional cross-sections forming a “box” with

balanced mass must be included.). The net

heat transported through the section (or box)

can then be calculated, and must balance the

total heat gain or loss through the sea surface

on either side of the section (or within the

box). That is, if the section is located at 30 Nin

Figure 5.12, then there is net gain of heat to the

south and net loss of heat to the north. The

section’s velocities and temperature should

then show a net flow of warmer waters northward

and colder waters southward.

Global annual mean ocean heat transport,

calculated using all three methods, is from the

heat-absorbing tropics to the cooling regions at

mid to high latitudes (Figures 5.12 and 5.14).

The maximum rate is 1 to 2 10 15 W at about

20 to 30 latitude in each hemisphere. This is

20 to 30% of the total global energy transport

of about 6 10 15 W; the atmosphere transports

more heat than the ocean everywhere (Trenberth

& Caron, 2001). The Pacific Ocean exports heat

poleward in both hemispheres. The Indian

Ocean exports heat southward because of the

absence of a high latitude Northern Hemisphere

region.

140


As previously mentioned, the global heat

budget must almost exactly balance; that is,

there is almost no net heat gain or loss for the

whole Earth, and the same for the ocean alone. 2

The meridional ocean heat transports based on

Grist and Josey’s (2003) fluxes are nearly, but

not exactly, in balance. In Figure 5.14b we also

show an “adjusted” meridional heat transport

curve that balances exactly, with zero transport

in both the north and south, achieved by adding

2.5 W/m 2 at every grid point. The adjusted

curve is within the error of Grist and Josey’s

(2003) calculation. Discussion in the next paragraphs

is in reference to this adjusted transport.

A counterintuitive result, found in every heat

transport product, is that the heat transport

throughout the Atlantic, including the South

Atlantic, is northward (arrows and annotation

in Figure 5.12). This is because there is so

much heat loss in the subpolar North Atlantic

and Nordic Seas. To feed this heat loss, there

must be a net northward flow of upper ocean

water throughout the length of the Atlantic,

which is returned southward by deeper colder

water. In all oceans, the subtropical gyre circulation

in just the upper ocean carries heat poleward

(Section 14.2.2) (Talley, 2003). This part of

the heat transport in the South Atlantic is not

strong enough to overcome the northward

heat transport due to the top-to-bottom overturn

that produces North Atlantic Deep Water.

The Pacific Ocean does not have a top-to-bottom

overturn with associated meridional heat transport,

so there is an asymmetry between the

Pacific and Atlantic heat transports.

Regarding the apparently odd northward heat

transport throughout the Atlantic even while the

Pacific and Indian Oceans follow the expected

pattern of poleward heat transport in both hemispheres,

Henry Stommel (personal communication)

told an interesting story. One of the last

studies by Georg Wüst (1957) was a study of

north-south transports in the South Atlantic.

Although he computed and published the transports

of oxygen, salinity, nutrients, and so forth,

he did not publish the meridional heat flux,

which is the easiest to compute since only

temperature profiles are required. Stommel suspected

that Wüst computed the heat transport

but found that it appeared to go in the wrong

direction, namely from the south to north

(toward the equator), as we see in Figure 5.12.

This violated Wüst’s intuition, which required

the heat to flow from the tropical north to the

colder polar south. To verify his suspicion, Stommel

sought out Wüst’s former students. He

managed to locate a German Admiral Noodt

who wrote to say that, yes, Professor Wüst did

not publish the heat transport because it

appeared that it “flew in the wrong direction”

(sic). This view did not change until the 1970s

when new studies (Bennett, 1976) clearly displayed

that the meridional heat transport in the

South Atlantic is northward.

5.7. BUOYANCY FLUXES

Buoyancy forcing changes the density of

seawater. External forcing is due to heat fluxes

(heating and cooling) and freshwater fluxes

(evaporation and precipitation plus runoff

from land, see the preceding sections). Almost

all of these forcings are from (or through) the

atmosphere, with only a very minor component

from Earth’s crust below. 3 Brine rejection due to

2 Major climate change on the order of 1 to 5 C, such as global warming or a shift into an ice age, would be associated with

a net ocean heat gain or loss on the order of 1 to 10 W/m 2 , calculated for a 1000 m thick layer of water over 100 or 10 years,

respectively. It is also well known that global warming associated with a doubling of CO 2 in the atmosphere corresponds

to a net change in heat flux of 4 W/m 2 .

3 Geothermal heat flux is typically 0.05 W/m 2 , in comparison to typical solar heating of 250 W/m 2 , weaker by a factor

of 5000.

BUOYANCY FLUXES 141

sea ice formation is an effective direct means of

fractionating (redistributing) the water column

density, by freshening the sea water locked up

in sea ice and increasing the seawater density

below the ice by release of the salt into the water

column. Sea ice maps and brine rejection are

described in the Arctic and Southern Ocean

chapters (Chapters 12 and 13).

A global airesea buoyancy flux map (annual

mean) is shown in Figure 5.15, based on Large

and Yeager (2009). It is the sum of the mean

airesea heat flux and mean freshwater flux

(evaporation minus precipitation/runoff). While

the units of buoyancy are inverse density, or

m 3 /kg, the mapped flux is converted to heat

flux units (W/m 2 ); this is simply because most

present-day depictions of airesea fluxes are in

terms of heat, and thus intuition is largely based

on heat. The airesea heat and freshwater

fluxes from Large and Yeager (2009) used for

Figure 5.15 are shown in the online supplement

(Figure 5S.8). In terms of a textbook description,

these fluxes differ only slightly from the NOCS

fluxes shown earlier.

The buoyancy flux map strongly resembles

the heat flux map because freshwater forcing,

while essential to the salinity balance, is weak.

Buoyancy loss (density gain) is most vigorous

in subtropical western boundary current separation

regions, where heat loss is large (Section

5.5.1). The other region of large buoyancy loss,

due to heat loss, is the subpolar North Atlantic

and Nordic Seas. The associated northward

transport of heat and hence buoyancy in the

Atlantic is related to the meridional overturning

circulation (Section 14.2).

Buoyancy gain is largest in the tropics,

particularly over the cool surface waters in the

eastern equatorial Pacific (Section 10.7.2). In

this “cold tongue,” the sea surface temperature

is persistently lower than the air temperature,

leading to heat gain. The equatorial Atlantic is

also a region of high buoyancy gain, in part

due to freshwater contributions from the

40˚

20˚

0˚

20˚

40˚

60˚N

50

60˚S

0

0

0

0

50

0

100 0 0

-25

-50

-100

0

0

0˚ 40˚E 80˚E 120˚E 160˚E 160˚W 120˚W 80˚W 40˚W 0˚

100

0

0

0

0

-50

0

Buoyancy flux (LY09)

(equivalent W/m 2 )

0

-25

0

25

0

0

0

0

0

-100

0

0

0

0 0

0 0 0

0

0

25

-100

5 -100

75 -150

-125

-50

0

0

0

-100

0

50

25

-25

-50

-100

75

0

-25

50

25

0

100

0

125

0

0

0

50

-25

100

-100

0˚ 40˚E 80˚E 120˚E 160˚E 160˚W 120˚W 80˚W 40˚W 0˚

0

25

25

125

-50

-25

0

60˚N

40˚

20˚

50

75 0˚

0

20˚

40˚

60˚S

FIGURE 5.15 Annual mean airesea buoyancy flux converted to equivalent heat fluxes (W/m 2 ), based on Large and

Yeager (2009) airesea fluxes. Positive values indicate that the ocean is becoming less dense. Contour interval is 25 W/m 2 .

The heat and freshwater flux maps used to construct this map are in the online supplement to Chapter 5 (Figure S5.8).

142


Amazon, Orinoco, Congo, and Niger River

outflows. The equatorward eastern boundary

currents (Peru-Chile, Benguela, California, and

Canary) are also regions of buoyancy gain, associated

with heating, and dynamically associated

with upwelling.

Perhaps most counterintuitive are the high

latitude regions where the seawater actually

becomes less dense due to air-sea fluxes, rather

than being cooled and becoming denser. The

two large regions that stand out are the subpolar

North Pacific and the Southern Ocean within

and south of the Antarctic Circumpolar Current.

Both are open ocean upwelling regions, and are

also regions of equatorward Ekman transport.

These two processes supply cooler water to the

sea surface, which is apparently met with net

heat gain.

Although freshwater fluxes generally contribute

much less than heat fluxes to the total

buoyancy flux, freshwater fluxes in the Southern

Ocean and subpolar North Pacific tip the balance

toward stronger and broader regions of buoyancy

gain by the ocean. Freshwater fluxes also

make a difference where heat fluxes are small,

such as throughout the subtropical gyres outside

the western boundary currents, where contributions

from both heat and freshwater fluxes (net

evaporation) are on the order of 10 W/m 2 .

Intense heat losses in ice formation regions

are not represented well in Figure 5.15. Brine

rejection is the main agent for producing very

dense water, but it is not included at all as it

only redistributes buoyancy internally. The

heat losses in coastal polynyas where large

amounts of sea ice are formed have small spatial

scales. For instance, the Weddell and Ross Seas

both appear as regions of net buoyancy gain,

whereas this is where very dense water is

formed through cooling and brine rejection;

cooling in cavities under the ice shelves is also

a factor here and cannot be represented in these

airesea flux maps (Chapter 13). In the North

Pacific, the brine rejection region of the Okhotsk

Sea does not appear in this map, which instead

shows a net buoyancy gain driven by runoff

from the Amur River.

5.8. WIND FORCING

Surface wind stress is the principal means for

forcing the ocean circulation, through a nearsurface

frictional (turbulent) layer and the

mass convergences and divergences in that

layer (see Chapter 7). The convergences/divergences

are directly related to the wind stress

curl. Global wind stress and wind stress curl

are shown in Figure 5.16. Seasonal variation is

important, especially in monsoonal regions, so

mean August and February winds are also

shown.

The largest scale annual mean wind patterns

are the easterly trade winds in the tropics and

westerly winds poleward of 30 latitude in both

hemispheres. Annual mean winds and wind

stress are strongest in the westerly wind belt of

the Southern Hemisphere (40 Sto60 S). In the

summer hemispheres in the tropics, the summer

monsoon with winds blowing from the ocean to

the continent is apparent in all three ocean basins,

but is most pronounced in the northwestern

Indian Ocean. The opposing monsoon is also

evident in the winter hemispheres (represented

by February in the Northern and August in

the Southern Hemisphere). In the Northern

Hemisphere winter, the westerly winds are

strongly developed around low pressure centers

in the North Pacific (Aleutian Low) and North

Atlantic (Iceland Low). In the Southern Hemisphere

winter, strong southerlies are apparent

around the Antarctic; these are the wintertime

katabatic winds (gravity-driven flow down the

sloping ice sheet). Similar winds are apparent

off the Greenland ice cap in Northern Hemisphere

winter.

Global mean wind stress curl (Figure 5.16d)

from the QuikSCAT satellite is an extraordinary

recent result (Chelton, Schlax, Freilich, & Milliff,

2004), with important detail that is not resolved

WIND FORCING 143

in the coarser resolution NCEP winds displayed

in the other panels of Figure 5.16. Wind stress

curl is related to ocean circulation because the

curl indicates Ekman convergence/divergence

that then drives interior equatorward/poleward

Sverdrup transport (Section 7.8 and maps in

Figure 5.17). Ekman downwelling is present

throughout the subtropical regions. Ekman

upwelling is present in the subpolar regions

and Antarctic, and in long zonal bands in the

tropics. These features are evident in any map

of mean wind stress curl, including those with

much coarser spatial resolution as shown in

the basin chapters.

With the high resolution winds, persistent

smaller scale features in wind stress curl are

(a)

40˚

20˚

0˚

0˚ 60˚ 120˚ 180˚ 240˚ 300˚ 0˚

60˚

60˚

Annual mean

0.1 N/m

40˚

20˚

0˚

–20˚ –20˚

–40˚ –40˚

–60˚ –60˚

0˚ 60˚ 120˚ 180˚ 240˚ 300˚ 0˚

(b)

60˚

February

40˚

0.1 N/m

60˚

40˚

FIGURE 5.16 Mean wind stress

(arrows) and zonal wind stress (color

shading) (N/m 2 ): (a) annual mean, (b)

February, and (c) August, from the NCEP

reanalysis 1968e1996 (Kalnay et al., 1996).

(d) Mean wind stress curl based on 25 km

resolution QuikSCAT satellite winds

(1999e2003). Downward Ekman pumping

(Chapter 7) is negative (blues) in the

Northern Hemisphere and positive (reds)

in the Southern Hemisphere. Source: From

Chelton et al. (2004). This figure can also be

found in the color insert.

20˚

0˚

20˚

0˚

–20˚ –20˚

–40˚ –40˚

–60˚ –60˚

(c)

40˚

60˚

0˚ 60˚ 120˚ 180˚ 240˚ 300˚ 0˚

August

0.1 N/m

60˚

40˚

20˚

20˚

0˚

0˚

–20˚ –20˚

–40˚ –40˚

–60˚ –60˚

0˚ 60˚ 120˚ 180˚ 240˚ 300˚ 0˚

–0.20 –0.16 –0.12 –0.08 –0.04 0 0.04 0.08 0.12 0.16 0.20

Zonal

wind

stress

144


FIGURE 5.16

(Continued).

260˚

280˚

300˚

320˚

340˚

0˚

20˚

40˚

60˚

80˚

100˚

120˚

140˚

160˚

180˚

200˚

220˚

240˚

260˚

60˚

30

10

0

60˚

40˚

20˚

-20

0

-5

-15 -10

0

30 20

-40 -30 -20

10

-10

0

40˚

20˚

0˚

0˚

0

-10

0

–20˚ 10

10

–20˚

0

25

20

20

10 0

30 20 30

40

20

–40˚ 10

15

65 80

70

50

–40˚

15

20

30

0

260˚

280˚

300˚

320˚

340˚

0˚

20˚

40˚

60˚

80˚

Sverdrup transport

–50 –45 –40 –35 –30 –25 –20 –15 –10 –5 0 5 10 15 20 25 30 35 40 45

(Sv)

50

FIGURE 5.17 Sverdrup transport (Sv), where blue is clockwise and positive is counterclockwise circulation. Wind stress

data are from the NCEP reanalysis 1968e1996 (Kalnay et al., 1996). The mean annual wind stress and wind stress curl used

in this Sverdrup transport calculation are shown in Figure 5.16a and in the online supplement, Figure S5.10.

100˚

120˚

140˚

160˚

180˚

200˚

220˚

240˚

260˚

WIND FORCING 145

apparent in the lee of large islands and mountain

gaps; examples include the Hawaiian

Islands among many others, and west of Central

America, where strong winds force eddy generation

in the Gulf of Tehuantepec (Chapter 10).

Also apparent in the altimetric product (and

not in the coarser reanalysis products) are

wind stress curl patterns that follow the major

western boundary currents such as the Gulf

Stream, Kuroshio, and Agulhas, suggesting

that these ocean fronts affect the position of

the winds, constituting a feedback (Chelton

et al., 2004).

The general circulation of the upper ocean is

mainly driven by wind stress through the

Sverdrup balance (Section 7.8). The Sverdrup

transport computed from NCEP reanalysis

wind stress curl is shown in Figure 5.17. (The

NCEP wind stress curl that is the basis for

Figure 5.17 is shown in the online supplement,

Figure S5.10). Its pattern and magnitude are

similar to that calculated from the mean

QuikSCAT winds shown in Figure 5.16d

(Risien & Chelton, 2008). This global map is

mainly described in the later basin chapters as

context for the circulation. The Sverdrup transport

is computed as the zonal integral of the

wind stress curl, integrated westward from the

eastern meridional boundary in a given basin.

For the Southern Ocean, the eastern boundary is

the Chilean coast of South America and the integration

extends westward across all three oceans

until reaching the Argentine and Brazil coast of

South America. Sverdrup transport is not

computed at the latitudes of the Drake Passage

because there is no meridional boundary there.

It is also not shown for the equatorial region

because the dynamics there are more complex.

C H A P T E R

6

Data Analysis Concepts

and Observational Methods

Our basic information about the oceans

comes from observations and, increasingly,

from numerical model output. To assist with

reading later chapters and other observational

oceanography literature, this chapter provides

an overview of commonly used methods for

analyzing observations. The basis of the statistical

methods that are commonly used for

oceanographic observations and numerical

model output is not fully and mathematically

described here (to a point where linear algebra

is required), as is necessary in many modern

approaches. There are many good starting

points for a full course on modern data analysis;

some useful texts are Bendat and Piersol (1986),

von Storch and Zwiers (1999), Chatfield (2004),

Emery and Thomson (2001), Bevington and

Robinson (2003), Wunsch (1996), and so on.

The Wolfram (2009) Web site provides demonstrations

of basic statistical concepts. Press,

Flannery, Teukolsky, and Vetterline’s (1986)

Numerical Recipes is useful for moving from

concepts to the practice of data analysis.

In data analysis, we begin with observations or

determinations of the value of a variable, such as

pressure, time, temperature, conductivity,

oxygen content, and so forth. These are collected

using oceanographic instruments at particular

times and locations that are chosen through

a sampling strategy (Section 6.1). From these

imperfect observations, containing both instrumental

and sampling error (Section 6.2), we estimate

the true field and its statistical properties as

a function of time and/or space (Sections 6.3

through 6.7). Sources of error are crucial to identify

and are expressed in terms of statistical

quantities. Errors arise from the accuracy of

the instrumental measurements and from

sampling that is discretized in time or space

and finite in duration.

A large amount of supplementary material

for this chapter appears on the textbook Web

site http://booksite.academicpress.com/DPO/

as Chapter S16 (“S” denotes supplemental material).

This supplement consists of an extended,

fully illustrated description of instrumentation

and methods for collecting information about

the ocean, including accuracies and sources of

error. Chapter S16 includes some of the sampling

issues for physical oceanography (Section S16.1),

platforms for observations (research and

merchant ships; Section S16.2), instruments for

in situ (within the water column) observations

(Sections S16.3eS16.8), an overview of satellite

remote sensing (Section S16.9), and oceanographic

archives (Section S16.10).

As discussed in the supplementary historical

materials for Chapter 1 (located on the textbook


147



148

6. DATA ANALYSIS CONCEPTS AND OBSERVATIONAL METHODS

Web site as Chapter S1) and the supplementary

materials on instrumentation for this chapter

(Chapter S16), observations in large-scale

physical oceanography through the 1950s

were designed to resolve the mean or longterm,

large-scale structure of the ocean. Observations

of time-variable phenomena were

focused on smaller spatial scales and higher

frequencies, including waves and tides.

Because of the relatively small volume of data

prior to the era of electronic sampling, satellite

instruments, and computer analysis, individual

observations were given greater consideration

than they often are today. Sparse

sampling still characterizes parameters that

are difficult to measure, for instance, some of

the chemical properties that require seawater

samples and specialized laboratories for analysis.

Error detection for such observations

relies on good laboratory practice, including

use of standards; comparisons with previous,

possibly sparse, observations; and careful

review of sample collection logs and analysis

procedures.

In contrast, modern instruments that

measure nearly continuous vertical profiles d

underway sampling systems such as expendable

bathythermographs (XBTs) and acoustic

Doppler current profilers (ADCPs), moored

current meters, autonomous drifting and

guided systems, and satellites d can generate

large volumes of digital data. These large data

sets can be treated statistically to identify data

errors, to map fields, to generate statistical information

such as means and trends, and to detect

embedded time and space patterns and correlations

among different observed parameters

(Section 6.4). Some of the basic concepts of

time series analysis, including brief introductions

to spectral and empirical orthogonal function

methods, are included at a rudimentary

level (Sections 6.5 and 6.6).

Oceanographic data are, by nature, threedimensional

in space and have time variation.

Spatial sampling is almost always irregular,

making good statistical techniques for mapping

and analysis beneficial. Objective mapping is

one common approach and is based on minimizing

the difference between the mapped field

and the observations in a least squares sense

(Section 6.4).

Least squares methods are central to many

common data analysis techniques, including

those that seek to estimate the absolute

velocity field from vertical profiles of temperature

and salinity (see Section 7.6 for information

on the derivation of dynamic height).

Geostrophic velocity calculations require an

accurate estimate of a quantity that is not

measured: the geostrophic velocity at just one

depth (any depth). Estimating this unknown

“reference” velocity is important because

transport calculations and budgets for all

parameters depend on having an absolute

velocity field. “Inverse methods” based on

least squares estimation have been developed

over the past several decades to yield the

optimal estimate of these unknowns (Wunsch,

1996). Large-scale observed velocity fields

from drifters or floats with climatological

hydrographic data have also been merged

using least squares techniques. Most recently,

oceanographic data assimilation (or state estimation)

methods, in which observations are

incorporated in computer ocean models, are

moving toward providing these absolute

velocity fields since the ocean is now becoming

sampled enough for practical data assimilation.

Inverse and data assimilation methods

are not introduced here, but we do provide

some acquaintance with the principles of least

squares methods on which they are based

(Section 6.3.4).

Methods for the presentation and analysis of

hydrographic (water property) data are also

provided (Section 6.7). This traditional study

has evolved to include statistical analysis as

well. We introduce a reasonably common recent

method, optimum multiparameter analysis

(OMP), to illustrate the possibilities for modern

OCEANOGRAPHIC SAMPLING 149

water mass analysis. OMP is also based on least

squares methods.

Many terms and concepts are introduced in

this chapter, so a glossary of terms with short

definitions of many of the concepts is found at

the end.

The most important rule in data analysis is

that there are no absolutely fixed methods.

Data should be plotted, played with, replotted,

combined with other data, and so forth, until

the objective is achieved, which for scientists

usually means discovering something new

about the ocean. On the other hand, common

understanding and application of data analysis

techniques, and especially those that involve

estimating error, are absolutely essential when

combining and comparing results from different

instruments, different properties, and different

scientists.

6.1. OCEANOGRAPHIC SAMPLING

Modern physical oceanographic data are

collected using many different platforms and

instruments (see Chapter S16 located at the textbook

Web site). Research ships have provided

a long historical data set and continue to

provide important modern observations. Analysis

of these data sets often requires dealing

with irregular temporal sampling and inhomogeneous

spatial sampling (much higher resolution

in some spatial directions than others).

Instrumentation has also evolved and sampling

philosophies have changed over time; in order

to combine historical and modern data, changes

in instrumentation, measurement error, and

sampling have to be considered. Modern in

situ data sets increasingly include large data

sets from autonomous samplers such as floats

and drifters and of vertical profiles collected

from merchant ships.

Satellites (and sometimes aircraft) collect

remotely sensed measurements of surface

parameters such as sea surface height (SSH)

and sea surface temperature (SST). Relative to

ship and buoy measurements, satellites sample

the ocean’s surface so quickly that the observations

can be regarded as almost synoptic. The

word “synoptic” comes from meteorology,

where it refers to weather in both space and

timescales. In practical physical oceanography,

the word is usually applied to observations

taken at nearly a single time relative to the timescale

of interest, similar to a “snapshot,” and

that are interpreted to contain only spatial information.

For the ocean, the synoptic timescale is

about two weeks, that is, the timescale of evolution

of an eddy. For the non-seasonal part of

large-scale ocean circulation, which varies on

interannual and longer timescales, synoptic

could be as long as a season or even a year

or two.

All observations are collected with finite

sampling intervals and over finite lengths of

time. Observations that might be considered

synoptic or nearly simultaneous for a given

phenomenon might not resolve faster-evolving

motions very well. For instance, sampling at

“eddy timescales” demonstrably misses faster

motions, such as tides or barotropic waves,

and will contaminate interpretation of the

desired timescale. This contamination is called

aliasing, in which the actual underlying fast

motion, being badly sampled in time, takes on

the appearance of a much longer timescale

(Section 6.5.3).

Moored instruments collect temporally

continuous information on currents, floattracks,

temperature, salinity, and other chemical

quantities; these can be treated using more

straightforward time series methods than the

ship- and satellite-based data sets. Now equipped

with satellite transmission systems, these

buoys report data in near real time for processing

and incorporation into model studies.

Because of the cost of individual moorings,

these data sets tend to be spatially isolated;

experiments and observation systems that

incorporate multiple moorings have to be

150


designed carefully based on whether it is advantageous

to have observations at adjacent moorings

correlated (or not) over the timescales of

interest.

Sampling and analysis strategies for each of

these various systems are based on the space

and timescale of the phenomenon observed. Is

the observing system looking at capillary and

surface gravity waves or at changes in the North

Atlantic’s meridional overturning circulation?

Sampling resolution must be sufficient to

measure variations at the space and timescales

of interest. This means that samples must be sufficiently

frequent in space or time to resolve the

highest frequency of variability, and the entire

record must be long enough in space or time to

contain a minimum number of cycles of the

important fluctuations of the variable of interest.

For instance, in vertical profiling, the vertical

variation in most properties is larger through

the pycnocline than through the abyssal waters.

Therefore vertical sample spacing or resolution

through the pycnocline and in the upper ocean

should be much closer than in the abyss (unless

all are oversampled, as with a nearly continuous

profiling instrument such as a conductivitytemperature-depth

profiler; CTD). Another

example is horizontal sampling in regions that

contain a mixture of spatial scales; for example,

the western North Atlantic contains both the

Gulf Stream and the ocean interior offshore of

the Gulf Stream. Good sampling strategy would

include better spatial resolution across the

narrow currents than in the broader flow

regimes. (The best sampling might be considered

to be the highest possible horizontal

spacing throughout, but ship time and cost are

also important factors, so sampling schemes

are based on prior knowledge of the time and

space scales of interest.)

For time or space coverage from an instrument

that records nearly continuously, these

data are averaged and/or filtered to yield the

time/space scale of interest. Thus, the data processing

for, say, a surface wave experiment is

very different from the data processing for

seasonal variability.

6.2. OBSERVATIONAL ERROR

Errors in observations, due to imperfections

in the actual measured values, can be characterized

in terms of accuracy and precision. In

Chapter 3 and in the supplementary Web site

materials Chapter S16, we report the accuracy

and precision of different instruments measuring

common ocean properties. Accuracy is

how well the observation reproduces a welldefined

standard that is usually set by an international

group. High accuracy means that the

difference between the observation and the

international standard is small. Every instrument

or observational technique has to be calibrated

to match these international standards

to the level required by the user and, therefore,

the manufacturer or engineer. Accuracy of

a measurement or data set is reported in terms

of offset and standard deviation of the offset.

Bias error (see following text) is directly affected

by the accuracy of an observation.

Precision is the repeatability of an observation

using a given instrument or observing system.

An instrument or system could be highly inaccurate

(e.g., due to lack of calibration), but

highly precise, meaning that its variations due

to instrument noise are very small. Precision is

related to random error (see the next paragraphs).

When observations are reported, the

level of precision affects the number of significant

places in the report; for instance, highly

precise ocean temperature observations are

listed out to four decimal places (10 4 C).

There are two basic types of error in all data

sets: systematic or bias error and random error.

Bias errors are an offset of the measured values

from the true values. Such errors can result

from poor sampling strategy, failure of the

sensor, error in the recording system, measurement

inaccuracy, or insufficient record length

BASIC STATISTICAL CONCEPTS 151

for a time series that is averaged. For example,

sampling choices can inadvertently create

a “fair weather” bias. Many more ship-based

observations are made in summer than in

winter, especially at high latitudes; this biases

the cumulative historical data set toward

warm conditions. Another example is infrared

satellite sensing of SST, which requires clear

sky (cloud-free) conditions, thus biasing these

observations.

The bias error associated with the measurements

is separate from the bias errors that can

be introduced by the statistical methods used

to analyze the data set. Statistical methods

(e.g., how an average is weighted) can introduce

or offset bias. It is usually desirable to use statistical

estimates that are unbiased. The estimators

given in Section 6.3 are unbiased.

Random error or noise arises from variations at

different time or space scales than the process of

interest for the particular experiment. These can

be both intrinsic to the observed variable (hence

the desired true statistical property) or due to

instrument or sampling error. The root-meansquare

(rms) standard deviation (Section 6.3) is

the calculated quantity associated with this

noise. To minimize noise, data are averaged or

filtered. For time series, this means collecting

a record that is long enough to cover many

cycles of the process of interest. These cycles

are then averaged together. Figure 9.6 shows

an example with snapshots of the path of the

Gulf Stream and the mean value of this path.

The noise or variance is a measure of the envelope

of all of the meanders around this mean.

Most time or space series analysis techniques

have been developed for long and continuous

data sets. Many oceanographic data sets are

sampled irregularly, which leads to problems

with analyzing the variability in the data sets.

A source of irregular sampling in time may be

the failure of an instrument or its replacement

by another instrument with clearly different

sampling characteristics. Irregular sampling is

also a consequence of combining historical

data for a particular analysis. While individual

cruises or experiments might have been organized

for a specific task, a combination of these

different sampling programs will not have

regular temporal spacing. The resulting gaps

in the time or space series mean that temporal

variability is often not well resolved.

Even if sampling is at regular intervals, the

true field cannot be continuously sampled. The

discrete time interval (or distance) between

samples leads to error in estimating processes

that have short time (or space) scales. The

Nyquist frequency (Section 6.5.3) is the highest

frequency that can be resolved with a given

sampling interval. Anything happening at

a frequency higher than the Nyquist frequency

is then very badly sampled, but is still in the

record. These higher frequency signals appear

as much lower frequency signals; this is called

aliasing. It is highly desirable to design the

observing strategy, specifically the sampling

interval, to minimize aliasing.

6.3. BASIC STATISTICAL

CONCEPTS

Every variable (such as temperature, salinity,

pressure, velocity, etc.) has a set of true statistical

behaviors; every set of observations of

the variable is an imperfect representation of

these statistical behaviors. In-depth data analysis

courses and textbooks carefully cover the

differences and similarities between the true

statistics and estimation of these statistics and

the associated error that arises simply because

of the always imperfect sampling. The estimated

statistics are called “sample statistics.”

Here, the sample statistics (mean, variance,

standard deviation, etc.), rather than the true

ones, are presented. We assert but do not derive

the important relations that show that these

expressions provide “unbiased estimates” of

the true mean, variance, covariance, and so

forth. An unbiased estimate is one that

152


preserves the true value without introducing

bias through the estimation method. Sometimes

biased estimators can be more useful or more

practical.

6.3.1. Mean, Variance, Standard

Deviation, and Standard Error

The abundance of data in modern physical

oceanography means that most analyses use

averages of data values rather than individual

samples. Modern instruments such as CTDs,

current meters, satellite instruments, and so

forth, collect many samples per second. Data

processing usually starts with averages over

many samples, for example, over 1 second or

some other time interval. The sample mean x

of a data set x that has been measured N equally

spaced times is

x ¼ 1 N

X N

i ¼ 1

x i (6.1)

With an increasing number of observations,

Eq. (6.1) approaches the true mean if there is

no external source of bias error. (That is, in the

absence of instrument calibration problems,

Eq. 6.1 is an unbiased estimate of the true mean.)

An anomaly is the difference between a

measured value and the mean value (6.1):

x 0 ¼ x x: (6.2)

The quantity x 0 is also often referred to as a deviation

from the mean; it is also referred to as an

anomaly. Because many observational studies

are concerned with time or space variation,

calculation and display of anomalies is common

in oceanography, meteorology, and climate

science. Depending on the study, it might also

be common to remove a seasonal cycle from

the original data set by computing monthly or

seasonal means rather than the overall mean,

and then displaying the anomalies relative to

the monthly or seasonal means. For instance,

a time series of surface pressure in the North

Atlantic can be averaged over its approximately

50-year record, the average (mean) removed,

and the anomaly time series analyzed to search

for signals like the North Atlantic Oscillation

or El Niño-Southern Oscillation (ENSO)

influence.

The variance of the data set x is the mean

value of the squared deviations of each measurement

from the mean. The variance gives

the inherent variability of the data set including

variability of the true field and variability

due to random sampling or instrument error.

For sampled values, the best (unbiased) estimate

of variance is the sum of the squared

deviations, divided by (N 1), rather than N:

X N

s 2 ¼ 1 ðx i xÞ 2

N 1

i ¼ 1

¼ 1 X N

ðx i Þ 2 1 X N

N 1

N

i ¼ 1

i ¼ 1

x i

2

(6.3)

The last expression provides a computationally

efficient way to compute the variance, using

only one pass through the data. The square

root of Eq. (6.3) is the standard deviation, s. Variance

and standard deviation are intrinsic properties

of the variable; Eq. (6.3) approaches the

true variance with an increasing number of

observations.

The rms error or standard error s 3 of the

observed data set is the square root of the

mean value of the difference between the true

mean and the sample mean, averaged over

many realizations of the sample mean. The standard

error is related to the standard deviation

from the mean as

s 3 ¼ p s ffiffiffiffi

(6.4)

N

Thus, the rms error of the mean, x, is smaller

by p1ffiffiffi

than the standard deviation, s, of

N

an individual measurement x. The standard

error decreases with increasing numbers of


(a)

Eastward wind velocity (m/s)

20

15

10

5

0

−5

−10

FIGURE 6.1 Example of time

series and probability density

functions (pdfs). (a) Eastward

wind speed (m/sec) from an

ocean buoy in Santa Monica

Basin. (b) pdf of eastward wind

velocity. (c) pdf of northward

wind velocity. (d) pdf of wind

speed. (Constructed from Gille,

2005).

−15

−20

2000 2001 2002 2003 2004 2005

(b)

0.18

(c)

0.35

Time (years)

(d)

0.25

0.16

0.14

0.3

0.2

Probability density

0.12

0.1

0.08

0.25

0.2

0.15

0.15

0.1

0.06

0.04

0.1

0.05

0.02

0.05

0

−20 0 20

Eastward

wind velocity (m/s)

0

−20 0 20

Northward

wind velocity (m/s)

0

0 10 20

Wind speed (m/s)

154


observations; it is not an inherent property of

the field that is being measured, but is a property

of the sampling and instrumentation.

The profound difference between standard

deviation (which is property of the true field)

and standard error (which is a property of the

sampling) is illustrated in the salinity climatology

in Figure 6.13, in which the standard

error is large in undersampled regions, while

the standard deviation is large in regions of

high oceanographic variability.

6.3.2. Probability Density Function

The probability density function (pdf) is the most

basic building block for statistical description of

a variable. Although the reader is much more

likely to encounter spectral analysis or empirical

orthogonal functions (see the following sections)

in various publications, it is best to introduce

pdfs first in order to develop intuition about estimates

and confidence intervals.

The pdf of the true field is a measure of how

likely the variable is to have a certain value. The

probability of falling somewhere in the entire

range of possible values is 1. The observed pdf

is basically a histogram, that is, counts of the

number of occurrences of a value in a given

range. The histogram is then “normalized” to

produce the pdf by dividing by the total number

of observations and the bin widths. The more

observations there are, the closer the histogram

comes to the pdf of the true field, assuming

that observational bias error is low (accuracy

of observations is high).

Probability distribution functions can have

many different shapes, depending on the variable

and on the physical processes. As an

example, from Gille (2005), a time series of

wind velocity from an ocean buoy off the coast

of southern California is shown in Figure 6.1.

The data are hourly samples for four years. To

compute the pdfs, the number of samples of

velocities/speeds in each 0.1 m/sec bin was

counted to create a histogram; for the pdf, the

values in each bin were normalized by dividing

by the total number of hourly samples (43,797)

and by the bin width (0.1). The two wind

velocity pdfs are somewhat symmetric about 0,

but they are not quite Gaussian (bell-shaped,

Eq. 6.5). The wind speed pdf cannot be centered

at 0 since wind speeds can only be positive; this

pdf resembles a Rayleigh distribution, which

has positive values only, a steep rise to a

maximum and then a more gradual fall toward

higher values.

A pdf with a uniform distribution would

have equal likelihoods of any value within

a given range. The pdf would look like a “block.”

Random numbers generated by a random

number generator, for instance, could have

a uniform distribution (the same number of

occurrences for each value).

One special form of pdf has a “bell shape”

around the mean value of the variable. Such a

pdf is called a Gaussian distribution or a normal

distribution. Expressed mathematically, a pdf of

the variable x with a normal distribution is

pdf ¼ p 1

s

ffiffiffiffiffi e ðx xÞ2 =2s 2 (6.5)

2p

where the mean x is defined in Eq. (6.1) and the

standard deviation s in Eq. (6.2). A field that is

the sum of random numbers has a normal

distribution. The pdf associated with calculating

the mean value has a normal distribution.

Thus if we measure a large number of

sample means of the same variable, the distribution

of these mean values would be normal.

The pdf associated with a sum of squared

random variables is called a chi-squared distribution.

Squared variables show up in basic statistics

in the variance (6.2), so the chi-squared

distribution is important for estimates of variance.

Gaussian and chi-squared distributions

have a special place in statistical analysis, especially

in assessing the quality of an estimate

(confidence intervals), as described at the end

of the next section.


6.3.3. Covariance, Auto-Covariance,

Integral Timescale, Degrees of Freedom,

and Confidence Intervals

If two or more variables are measured, it is

useful to quantify how closely they depend on

each other. For instance, we might want to

know how closely temperature and velocity

are correlated with each other. The sample

covariance is the statistical relation between two

observed variables, for example x and y, each

sampled N equally spaced times:

covðx; yÞ ¼ 1

N 1

X N

i ¼ 1

ðx i xÞðy i

yÞ (6.6)

With an infinite number of samples, this

approaches the true covariance. The sample

correlation is covariance divided by the sample

standard deviations:

covðx; yÞ

r x;y ¼ (6.7)

s x s y

in which the standard deviations for both x and

y are defined as in Eq. (6.3).

The autocovariance and autocorrelation are

the same expressions as (6.6) and (6.7), but

with the two variables replaced by a single variable

measured at different times. In this case the

sum is over all pairs separated by the same time

difference within the time series. For instance, if

velocity (indicated here by the variable x) is

measured every hour for four years (e.g.,

Figure 6.1), then time lags, denoted by s, of

1 hour up to 4 years are available. If the record

length is T, with a total of N samples at

a sampling interval of Dt (so T ¼ NDt), the autocorrelation

for a given lag s ¼ nDt is

r x;x ðsÞ ¼ 1 s 2 x

1

M

ðNX

nÞDt

i ¼ 1

x 0 ðt i nDtÞx 0 ðt i Þ (6.8)

The anomaly x 0 was defined in Eq. (6.2). The

autocorrelation is typically calculated for all

time lags. The value of M can be either N (total

number of samples), or N n (total number of

pairs at lag s). If the total number of pairs is

chosen, then Eq. (6.8) is an unbiased estimate

of the autocorrelation, because the estimated

autocorrelation is not offset from the true autocorrelation.

However, this unbiased estimate

becomes very large at large lags, where N n

becomes very small. If the total number of

samples N is chosen, then Eq. (6.8) is a biased

estimate of the autocorrelation, but it has good

behavior at large lags. The unbiased estimate

is best for looking at behavior at small lags,

such as for finding decorrelation timescales

(see the next paragraph).

Using a simulated temperature record for

a Pacific island, the unbiased and biased autocorrelation

functions are calculated and plotted

as a function of lag (Figure 6.2; Gille, 2005).

The autocorrelation is 1 at zero lag, as it should

be, since the values should be perfectly correlated

with themselves. The unbiased estimate

blows up at large time lag, but is well-behaved

at small lag. The biased estimate (which is the

default in the Matlab software package used

by many oceanographers) is well-behaved at

large lag. At small lags (Figure 6.2d), the autocorrelation

decreases to a zero crossing at about

6 months. The time lag for the zero crossing is

one measure of the “decorrelation timescale”

for the variable, that is, the time lag for which

samples become uncorrelated. Since the autocorrelation

hovers around zero for several

months in Figure 6.2d, the decorrelation timescale

is somewhat ambiguous, but is in the

range 6e14 months.

An integral timescale T int for the observed variable

is defined as the time integral of the autocorrelation

(e.g., Gille, 2005; Rudnick, 2008).

The integral timescale is another measure of

the decorrelation timescale. For the sample

autocorrelation in Eq. (6.8), T int is the sum of

the autocorrelations multiplied by the time lag

bin width; this is the area under the autocorrelation

function in Figure 6.2b and c. In practice,

the sum is computed starting with just a

small total time interval; this is increased

156


FIGURE 6.2 (a) Time series of temperature at Fanning Island (Pacific Ocean) from the NCAR Community Ocean Model.

(b) Autocorrelation normalized to a maximum value of 1 (biased estimate with averages divided by N). (c and d) Autocorrelation

(unbiased estimate with averages divided by N n). Source: From Gille (2005).

incrementally until reaching the entire length of

the record. There will usually be a maximum

value for the integral at one of the intermediate

integration limits. This maximum value is the

integral timescale. (Thus the misbehavior of

the unbiased estimate at large lags can be

ignored, so either the biased or unbiased autocorrelation

function can be used.) For the time

series in Figure 6.2, the computed integral timescale

is 9.3 months, which can be compared with

the crude decorrelation timescale of >6 months

estimated from the first zero crossing.

The effective degrees of freedom of an estimate

are related to the integral timescale. The N

measurements are not necessarily independent

of each other. How many independent samples

do we really have, for instance, in a nearly

continuously sampled time series? The number

of independent samples, which is the

same as the “degrees of freedom,” is the total


length of the time series divided by the integral

timescale:

N dof ¼ T=T int (6.9)

How many degrees of freedom are desirable

for a good estimate? There are textbook answers

to this question, but the real answer lies in how

well you want or need to know the answer;

that is, how large are the errors? If they are too

large, little can be learned from the data set. The

“error bars” are formally the confidence intervals

calculated from the data set. Confidence intervals

are central to most data-oriented analyses.

Confidence intervals depend on the number

of degrees of freedom, and they also depend

on the standard error, hence on the standard

deviation of the time series. (Again, a full textbook

description and derivation is recommended;

see for instance Bendat & Piersol, 1986.)

Suppose we have a set of averaged, observed

values X of the variable x. We have already

found the standard deviation and the number

of degrees of freedom. Therefore we already

know the standard error s 3 ¼ s/ON dof . Suppose

we are looking for the probability that the true

average X exists within a given interval. What

is that interval? The statement of probability P is

P½X

s 3 t Ndof ða=2Þ X

X þ s 3 t Ndof ða=2ÞŠ ¼ 1

a

(6.10)

If, for instance, we wish to find 95% confidence

intervals, then (1 a) ¼ 95% and a ¼ 5%.

The statement (Eq. 6.10) is then read as “there

is a 95% probability that the true mean X lies

within the interval from X s 3 t N to X þ s 3 t N

where X is the sample mean.” The factors t N

are the “Student t variables” with N degrees of

freedom, and which depend on the choice of

confidence interval. Here the N is the calculated

degrees of freedom N dof . (The Student t-test is

appropriate for a variable that has a Gaussian

distribution; because we craftily started out

with a variable X that was already an average,

we can be pretty certain that the distribution is

Gaussian. This is a consequence of the Central

Limit Theorem. (See Bendat &Piersol, 1986 for

a discussion of this important theorem.) Once

you have estimated the degrees of freedom

and chosen an a, the t-variables are found

from a lookup table, available in most statistics

textbooks or online; they can also be obtained

through functions in Matlab or Mathematica.

For a 95% confidence interval and with

10 degrees of freedom, the t variable is 2.23.

For 10 degrees of freedom and a 90% confidence

interval, the t variable is 1.81, whereas for a 99%

confidence interval, it is 3.16. As the number of

degrees of freedom increases, the t variables

become smaller and the confidence interval

shrinks.

An example of a plot with confidence intervals,

in this case at the 90% level, is shown in

Figure 6.3. This is a graph of global ocean heat

content in the upper 700 m since the 1950s, constructed

from all available temperature profile

data at the time of the analysis. Because there

are confidence intervals on the plot, it is possible

to conclude that the upper ocean has warmed

since the 1950s, and that the warming is “significant”

in a formal sense. However, there is an

important limitation to the use of confidence

intervals, for which it is assumed that error is

random. When there is also a problem of accuracy

(formally, bias error), then confidence intervals

are simply not adequate. The graph in

Figure 6.3 should be compared with the more

recent version of global ocean heat content in

Figure S15.15 on the textbook Web site, from

Domingues et al. (2008). There was a protracted

episode of low quality temperature data in

the 1970s, due to error in assigning depths to

falling XBT profilers (see instrument description

in Section S16.4.2.5 in the online supplementary

materials); this led to artificially high temperatures

in the 1970s in Figure 6.3. Domingues

et al. (2008) recognized and corrected for

this accuracy problem; the improved estimate

of heat content change is much more monotonic

over the full record from the 1950s to the 2000s.

158


FIGURE 6.3 Example of time series with confidence intervals. Global ocean heat content (10 22 J) for the 0 to 700 m layer,

based on Levitus et al. (2005a; black curve), Ishii et al. (2006; full record gray curve and larger error bar), and Willis et al.

(2004; darker gray after 1993 and shorter error bar). Shading and error bars denote the 90% confidence interval. Compare

with Figure S15.15 seen on the textbook Web site from Domingues et al. (2008) which uses improved observations. Source:

From the IPCC AR4, Bindoff et al., 2007; Climate Change 2007: The Physical Science Basis. Working Group I Contribution to the

Fourth Assessment Report of the Intergovernmental Panel on Climate Change, Figure 5.1. Cambridge University Press.)

This discussion about confidence intervals is

very limited because this text is not primarily

about statistics. The subject is large and subtle.

Most actual properties are not normally distributed,

although they might be close. Techniques

for choosing confidence intervals can therefore

vary. Several other important methods are used

for assessing uncertainty, including bootstrap

and Monte Carlo methods. Those who wish to

pursue serious analysis of data sets are urged

to study this subject in much greater depth.

6.3.4. Least Squares Analysis

When two or more variables are measured,

we can use correlation to see how closely they

are related (Section 6.3.3). This gives us a single

number. Beyond correlation, we can make a

guess at the functional dependence of one

variable on the other. To determine parameters

in the relation between the variables, we perform

some kind of fit based on assumptions

or a model of how the variables are related.

One of the most prominent methods of this

type is least squares analysis. Wunsch (1996)

provided an excellent, deep introduction to least

squares and applications to inverse modeling

and related topics in physical oceanography.

As the simplest common example of least

squares, consider two time series, x(t) and y(t).

Suppose that we believe that one depends linearly

on the other, for whatever reason, so we


have an equation that we assume is a pretty

good relation between them, but with parameters

that we do not yet know. That is, suppose:

yðt i Þ¼axðt i Þþb (6.11)

We then determine the parameters a and b by

minimizing the difference between the two time

series in a least squares sense:

3 ¼ XN

i ¼ 1

y i x i

=

i ¼ 1

ðy i ax i bÞ 2 (6.12)

where a and b are unknown parameters to be

determined and 3, which is the sum of the

squared differences between the two expressions,

is the squared “misfit” we seek to minimize.

(3 is also referred to as a “cost function,”

which leads into the topics of inverse modeling

and data assimilation based on least squares.) To

find the best values for a and b, take the partial

derivatives of 3 with respect to a and separately

with respect to b, set the derivatives to zero, and

solve for a and b. Just to show how this works

with an even simpler model, in which b ¼ 0,

the solution for parameter a comes from solving

v3

va ¼ v X N

2ay i x i þ a 2 x 2 i ¼ 0

va

i ¼ 1

(6.13)

P N P N

a ¼

x 2 i

i ¼ 1

As the “model” becomes more complex, but

still linear, solving for the parameters becomes

much simpler if basic linear algebra is

employed. That is, the parameters in a matrix

“A” that relate vectors x and y might be

expressed as, y[Ax, where the bold type indicates

vectors and matrices. Solution of even

the simplest of these problems is beyond this

chapter. However, it is straightforward to carry

through and to actually calculate the parameters

in a problem like Eq. (6.12) using software packages

like Matlab that include linear algebra and

matrix operations.

Simple least squares fits are often employed

in calibrations. For instance, for CTD conductivity

calibration, a number of highly accurate

salinity values might have been obtained externally

using bottle samples. These can be converted

to conductivity (see Chapter 3), and

then the measured CTD conductivity can be

compared with the sample conductivities at

the same locations. The differences between

the two data sets (bottle and CTD) would be

expressed as a sum of squared differences, and

the CTD conductivity fits, using least squares,

to the bottle samples. The fits can be linear,

quadratic, cubic, and so forth, to provide the

best possible calibration, which depends on

the underlying physical response of the conductivity

sensor (it usually also has pressure and

temperature dependence, adding more

complexity to the fitting process).

In more general circumstances, beyond the

simple time series observations used in the

previous paragraphs, the size of the vectors y

and x can differ so much that A is not a square

matrix. Their meanings can also differ, where x

can be a set of observations, and y the field being

sought. These become “overdetermined” problems

if there are fewer unknowns than equations,

and “underdetermined” problems if

there are more unknowns than equations. It is

not necessary to use the squared difference as

the ideal “norm” for minimization; a more

educated choice would depend on the statistics

of the differences being minimized. For least

squares, it is assumed that the differences have

a Gaussian distribution.

More advanced linear least squares analysis

can also add external constraints to the assumed

functional relationship between two

datasets.Thisenterstherealmofinverse models

as applied to estimation of the unknown reference

velocity for a geostrophic velocity profile

that is first calculated from vertical density

profiles using dynamic or steric height

(Wunsch, 1996). It also enters the realm of data

assimilation, in which the proper dynamical

160


relations between different fields are presumed

and all of the fields are adjusted, in a dynamically

consistent manner, to most closely fit

the observations using least squares. But the

basic assumptions and practices are the same:

the assumption that there are linear relationships,

and the practice of minimizing the differences

between two functions based on some

assumed norm.

But the real question is whether even complicated

linear models are valid. In many cases

they are. The assumption of a simple linear relationship

is like the origin of spectral analysis in

the late nineteenth century, when the assumption

was that a given time series could be fit or

explained in terms of just one or two sinusoidal

functions. Spectral analysis moved far beyond

this by generalizing the valid representation of

a given time series in terms of a very large

number of sinusoidal functions that can completely

represent the data (Section 6.5). Spectral

analysis is not always ideal since the underlying

processes may not be sinusoidal, in which case

a spectrum can be obtained but might not be

as easy to interpret as an analysis using more

appropriate orthogonal basis functions. Similarly,

in many cases we know that linear

relationships are valid, because we know independently

what the dynamical relationship

between two data sets is expected to be, at least

after components that have little to do with the

dynamical relationship of interest are filtered

out. If the relationship is linear (and the differences

are Gaussian), then linear least squares

methods are a valuable place to start.

6.4. VARIATION IN SPACE:

PROFILES, VERTICAL SECTIONS,

AND HORIZONTAL MAPS

Observations that sample the ocean spatially

are often combined to produce vertical crosssections

and quasi-horizontal maps. These data

are almost never collected on a regular grid.

Temporal coverage within spatially sampled

data sets also varies greatly.

Beyond simply plotting the data as profiles or

values along a float track, for example,

a frequent step in working with spatially distributed

data is to map it to a regular grid. The most

simplistic method is to average the data within

a given grid “box” defined by latitude and

longitude. This “bin-average” method produces

a field that can be described and can be

adequate for a given study, especially if the

data set is very large. But a bin-averaged field

is often not the optimal field for quantifying

and studying the dynamics.

Objective mapping is a more complex and

common method for mapping randomly spaced

data to a specified set of locations. Objective

mapping for oceanography was introduced by

Bretherton, Davis, and Fandry (1976) for analysis

of eddy-scale observations in the Gulf

Stream region; their paper remains the definitive

basic treatment. Objective mapping

provides the “least square error linear estimate”

of the field (Bretherton et al., 1976) and, importantly,

an error field that depends on sampling

locations. Objective mapping methods are

most useful if the estimates are unbiased, and

there are different approaches to achieving this

(e.g., Le Traon, 1990). With objective mapping

techniques, external constraints on the fields

can be incorporated (e.g., the mapped velocity

field is geostrophic), and different types of

data can be combined.

We do not present any details of the objective

mapping method, as linear algebra is not

included in this text. Objective maps are basically

weighted averages of the data in the neighborhood

of the grid points. But the weighting

requires information on the horizontal shape

and scale of smoothing as a function of distance

from the grid point. The more influence given to

data from farther away, the smoother and larger

scale the mapped field. This weighting information

should come from the spatial covariance,

but this is not always (or usually) known since

VARIATION IN SPACE: PROFILES, VERTICAL SECTIONS, AND HORIZONTAL MAPS 161

it is also obtained from the actual observations.

In practice, simplified functions for weighting

are often chosen a priori; these are often either

exponentials or Gaussians (squared exponentials).

The weighting can be anisotropic, with

different horizontal decay scales in different

directions. This is useful for studies of frontal

or coastal regions where correlations are larger

in one direction than in the other. Anisotropic

scales are also useful for mapping data onto

vertical sections since the vertical and horizontal

scales differ enormously.

6.4.1. Variation in the Vertical

Direction

6.4.1.1. Sampling

Many ocean observations are collected as

vertical profiles. Because the ocean varies more

strongly in the vertical than in the horizontal,

sampling strategies usually provide far more

vertical resolution than horizontal resolution.

In addition, the upper ocean is more strongly

stratified than the deeper ocean in most places,

and most properties and currents reflect the

stratification. So sampling strategies often

include higher vertical resolution in and above

the pycnocline than in the deep ocean.

Instruments that are typically used for

vertical profiling, like a CTD, XBT, profiling

float, or lowered ADCP (LADCP), measure

“continuously.” For instance, commonly used

CTDs sample at 24 Hz (24 samples per second).

Profile processing usually involves averaging

over short pieces of the continuous sample to

produce a series of data at regularly spaced

pressures (e.g., 1 or 2 dbar) or times (e.g.,

1 second). The averaging reduces the profile

noise resulting from smaller scale processes,

such as microstructure on scales of centimeters.

These smoothed data series then resolve most

of the phenomena of interest for a large-scale or

mesoscale study, since ocean layering is often

on scales of 10 m and more. (Obviously

a different sampling and averaging approach is

necessary for studying the much smaller vertical

scales of the microstructure and fine structure

associated with mixing processes.)

Bottle sampling requires a choice of observation

depths, using prior knowledge of the basic

field and nearby or concurrent CTD profiles,

which provide temperature/salinity information

that can assist choice of bottle-sampling

depths. In the past, it was common to sample

at standard depths. This is no longer considered

good practice for hydrographic sampling since

we now focus on mapping the three-dimensional

property distributions, or properties on

surfaces that cut through the ocean, for example,

isopycnal surfaces. A hydrographic property

field is best mapped if there is some randomness

in the vertical sampling from station to station,

but always with enough samples to define the

vertical gradients in the property.

6.4.1.2. Vertical Profiles

The distribution of properties with depth is

illustrated with temperature/depth, salinity/

depth, and oxygen/depth profiles (e.g., Figure

4.2, etc.), and is usually the first step in examining

hydrographic data. It is useful to plot the data as

soon as possible so that problems with equipment,

sampling, or laboratory analyses can be

identified and remedied before many more

samples or stations are collected. Data from

multiple stations, geographical positions, or

times can be displayed together, to differentiate

between true ocean structure and biased or noisy

data. When analysts work with large data sets,

collected over many years, rather than with

a data set that is newly acquired, they often

average the data after interpolation to either

standard depths or standard densities, and examine

or reject data points that are outliers because

they are outside the range of the standard

deviation of the data set. (This involves some iteration

if additional data continue to be collected.)

With nearly continuous and regularly sampled

vertical profile data, such as from CTDs,

XBTs, LADCPs, and so forth, it is easy to

162


interpolate to any desired vertical level (depth,

pressure, density, temperature, etc.). With

sampling intervals on the order of 10 dbar or

less, simple linear interpolation can be adequate.

Some properties are more sparsely sampled, such

as chemical samples from a rosette water sampler,

where sampling intervals can range up to several

hundred decibars. Vertical interpolation using

smoother methods, such as a cubic spline, can

then be advantageous; care should be used to

choose procedures that do not introduce spurious

(unmeasured) maxima or minima in the profiles.

The Akima cubic spline has been found to be

especially good in this respect.

6.4.1.3. Vertical Sections

To examine the geographical distribution of

properties across a basin and in the vertical direction,

cross-sections through the ocean (“vertical

sections”) are produced from a set of stations,

often located along a substantially straight path

(e.g., Figures 4.11 through 4.13 and many other

examples through the text). Such sections are

an invaluable step in data quality control. For

quality control and to communicate the accuracy

of a given sampled field, it is useful to include

station locations and also sample locations if

the property is sampled with discrete bottles

(e.g. vertical sections of chlorofluorocarbon and

D 14 C in Figure 4.24).

To construct vertical sections from profile or

water sample data, interpolation methods are

usually required. Objective mapping (Section

6.4.1) is a commonly used method. Roemmich

(1983) introduced a useful method for mapping

vertical section data that relies on the station

separation to set horizontal decorrelation scales,

and very crude information about vertical stratification

to set vertical decorrelation scales. The

assumptions are that the scientists collecting

the data will know to sample more closely

across strong dynamical features such as

western boundary currents, the equator, and

fronts, and that this information should not be

lost in the process of objectively mapping the

data. Almost all of the vertical sections presented

in this text were objectively mapped

using Roemmich’s method.

Since flow is mostly along isopycnal surfaces,

with only a small diapycnal component, it can

be useful to display data as a function of density

rather than pressure or depth. An Atlantic

salinity section is plotted in Figure 9.17 as a function

of both depth and neutral density (Section

3.5.4). (The profile/sample data were first interpolated

to a large number of neutral densities

using cubic splines and then objectively mapped.)

The contours are much “flatter” in the isopycnal

coordinate than they are in the depth

coordinate. This suggests that flow is more

along isopycnal surfaces than along surfaces of

constant depth and helps to justify analysis of

large-scale properties along isopycnal surfaces

(Section 6.4.2).

6.4.2. Variation in the Horizontal

Direction

Some types of measurements lend themselves

exclusively to horizontal mapping. Examples

include temperature, velocity, surface height,

and airesea flux fields from surface drifter and

satellite observations. Within the ocean, because

of strong vertical stratification, flow mostly

follows isopycnal surfaces, which are substantially

horizontal. Therefore velocity and water

property data from within the ocean are often

mapped on quasi-horizontal surfaces, including

constant depth or isopycnal surfaces. Examples

are found throughout the chapters of this text.

Horizontal maps are usually created by

choosing the desired surface, interpolating the

data in the vertical to the surface, and then

mapping the vertically interpolated data.

Various methods are used for vertical interpolation

(Section 6.4.1.2). Horizontal mapping of

the data to a latitude/longitude or distance/

distance grid is also an interpolation exercise.

It is often carried out using objective mapping

(see beginning of Section 6.4), or some other

VARIATION IN SPACE: PROFILES, VERTICAL SECTIONS, AND HORIZONTAL MAPS 163

procedure that chooses data within a given

radius of the grid point and then creates

a weighted mean of the data depending on

distance from the grid point.

Maps that illustrate the influence of the high

salinity Mediterranean Water (MW; Section

9.8.3.2) are shown in Figure 6.4. Three kinds of

maps are shown: at a constant depth, on an

isopycnal surface, and at a “core layer.” All three

have their uses. The isopycnal surface is expected

to be the most representative of the actual flow.

The choice of pressure reference for the isopycnal

is important for creating a surface that best

follows the flow; in this figure, a reference pressure

at 1000 dbar was chosen since the core of

the MW is around 1200 dbar. A neutral density

surface would also be an effective choice (see

Figure 6.4 caption). Core layers (surfaces defined

by vertical extrema of properties like salinity)

were introduced and used extensively in the

1930s by German oceanographers. The core layer

in Figure 6.4c is the MW salinity maximum.

For all three maps, care must be taken in

interpreting “tongues” of high or low salinity

as indicating flow direction. If mixing is relatively

strong, then a horizontal tongue may indicate

flow direction. On the other hand, if there

were no mixing, then flow would have to follow

contours of the mapped property, and go

around the tongue. Therefore the main usefulness

of a core layer is to show the area of influence

of a particular water mass.

The three maps in Figure 6.4 are complementary,

and all show that the highest salinity in this

depth range originates at the Strait of Gibraltar,

where the high salinity MW exits into the North

Atlantic. The idea of the core layer method is

that the high salinity pool in the North Atlantic

indicates movement of the water away from the

Strait of Gibraltar. Because of mixing, a core

gradually weakens along its length.

Horizontal velocity mapping is important for

studying circulation. At horizontal length scales

ranging from mesoscale (ten to hundreds of km)

to global scale (thousands of km), the horizontal

velocities are nearly geostrophic and therefore

non-divergent. 1 The horizontal velocities can

then be represented by a streamfunction

(Section 7.6). Consequently, it is desirable to

map continuous contours that align with the

velocity vectors; error maps can be produced

as the difference between the mapped nondivergent

velocities and the original velocity

data, as well as the usual error estimates due

to the mapping procedure and due to measurement

error and variance.

Large-scale velocity and streamfunction

maps are shown in other chapters to illustrate

the sea surface and 900 m circulation. These

are based on surface drifter plus altimeter data

(Niiler, Maximenko, & McWilliams, 2003) and

subsurface float data (Davis, 2005). (These

instruments are described in the supplementary

online material in Chapter S16, Section S16.5.)

These two treatments of “Lagrangian” data

(Section 7.2) followed somewhat different

routes to produce the non-divergent fields,

and both illustrate the sensible creation and

application of mapping techniques based on

the desired product and the types of available

data. Niiler et al. (2003) combined the surface

drifter with satellite altimeter data, using a least

squares procedure and dynamical constraints

to produce the mean surface streamfunction.

Davis (2005) produced mean velocity vector

maps constrained to produce a non-divergent

field; he then used objective mapping to

produce the geostrophic streamfunction.

1 Typical horizontal current speeds range from 1 cm/sec up to 200 cm/sec (about 200 km/day or about 2 knots) in the swift

western boundary currents (Gulf Stream, Kuroshio), in the Antarctic Circumpolar Current and in the upper ocean

equatorial currents, to a fraction of 1 cm/sec in much of the surface layer and in the deep waters. The vertical speeds

associated with the large-scale circulation are much less, on the order of 10 5 cm/sec or 1 cm/day; these are essentially

unmeasurable except with extremely good instruments and data filtering.

164


FIGURE 6.4 Different types of surfaces for mapping. The Mediterranean Water salinity maximum illustrated using:

(a) a standard depth surface (1200 m); (b) an isopycnal surface (potential density s 1 ¼ 32.2 kg/m 3 relative to 1000 dbar,

s q ~ 26.62 kg/m 3 relative to 0 dbar, and neutral density ~ 26.76 kg/m 3 ); (c) at the salinity maximum of the Mediterranean

Water and North Atlantic Deep Water (white areas are where there is no deep salinity maximum); and (d) data locations

used to construct these maps. This figure can also be found in the color insert.

VARIATION IN TIME 165

Davis and also Gille (2003), who mapped the

velocity field at 900 m in the Southern Ocean

from subsurface floats, provided in-depth information

about their mapping methods, the

details of which are beyond this text.

Horizontal mapping of velocity fields has

been carried out on regional scales using density

profile and ADCP velocity data; the sampling

strategies here have been quasi-grids with

observations in three dimensions rather than

just along a single section. The density information

provides the vertical shear of the

geostrophic velocity through the thermal wind

balance. The ADCP data provide information

for the geostrophic velocity referencing, but

contain all timescales of motion, including

ageostrophic velocity as well as geostrophic.

An example from the California Current is

shown in Figure 6.5 (Chereskin & Trunnell,

1996); other similar maps have been produced

FIGURE 6.5 Objective mapping of velocity data,

combining density and ADCP velocity measurements.

California Current: absolute surface streamfunction and

velocity vectors in April, 1999, using the method from

Chereskin and Trunnell (1996). This figure can also be found

in the color insert. Source: From Calcofi ADCP (2008).

for the Azores Front (Rudnick, 1996) and

the Antarctic Circumpolar Current in Drake

Passage (Figure 13.9 from Lenn, Chereskin,

Sprintall, & Firing, 2008). These publications

include extensive information on the mapping

techniques created for these specific data sets.

6.5. VARIATION IN TIME

All ocean flows and properties vary in time.

Here we introduce some basic ways of displaying

and analyzing time series (data display,

spectral methods), as well as some common

but more advanced methods (empirical orthogonal

functions).

6.5.1. Time Series Data Display

Examples of time series are shown in Figures

6.6 through 6.9. Others appear throughout the

chapters of this book. The first step in working

with a time series is usually a simple plot of

the property versus time (e.g., Figure 6.6a). For

data collected as profiles, overlays of all, or

a subset of, the profiles are useful for seeing

the variability and variance in the time series

(Figure 6.6b). It can be useful to make a “waterfall”

plot, with the profiles offset from each

other by a fixed increment of the observed property,

to see individual features and how they

might propagate through the time series.

Profile data, such as from profiling floats, are

often contoured like a vertical section with time

and depth (or pressure or density) as the axes

rather than distance and depth. Similarly, if

time series data are collected from a number of

locations along a repeated track, display of the

data as a contour plot as a function of time

and the spatial dimension can be useful. This

type of plot is called a Hovmöller diagram. This

type of display is used in this book to show

the evolution of Arctic sea ice (Figure 12.22)

and the westward propagation at mid-latitudes

typical of Rossby wave behavior (Figure 14.18).

166


(a)

PROPERTY

TIME

SERIES

SALINITY %

33

32

1972

1973

1974

TIME

(b)

0

TEMP. SCALE : 0 5 10 15 °C

1.

2. 3. 4. 5. 6.

TIME

SERIES

OF

PROFILES

DEPTH m

50

100

5.0° 5.0° 5.0° 5.1° 5.1° 5.1°

0 20 40 60 80 100

TIME days

(c)

CURRENT

SPEED

AND

DIRECTION

TIME

SERIES

SPEED cm/s

DIRECTION (° TRUE)

60

40

20

0

360°

180°

0

0

1

2

TIME

3

days

4

5

(d)

0

1

SPEED SCALE

2

3

4

5

STICK

DIAGRAM

0 20 40

cm/s

60

(FOR

DATA

IN (c))

FIGURE 6.6 Examples of time series plots: (a) property/time, (b) time series of profiles, (c) current speed and direction, and (d) stick

diagram for data of (c).


FIGURE 6.7 Example of time series,

spectra, and spectral confidence intervals.

(a) Velocity (cm/sec) stick plot, lowpassed

at 100 hours, from 5 deep current

meters at different depths on one

mooring in the Deep Western Boundary

Current in Samoan Passage (see

Figure 10.16). The vertical direction is

along the passage axis. (b) Spectra from

the same current meters, offset by one

decade. The 95% confidence intervals are

shown at the bottom. Source: From

Rudnick (1997).

6.5.2. Velocity (Vector) Data

Time-Series Analysis

Vector fields such as velocity are slightly

more complicated to present than scalar time

series because they include two quantities:

magnitude (i.e., speed) and direction. The stick

plot is a useful vector display method, representing

speed and direction by a line drawn to scale

out from a time axis at each observation time

(Figures 6.6d and 6.7). Currents from a set of

instruments on a mooring can be plotted above

each other to give a visual idea of the correlation

of the records at different depths (Figure 6.7).

Another alternative for displaying velocity

data is a progressive vector diagram (see example

in Figure S7.14a from Chereskin, 1995 on the

textbook Web site), in which the displacements

from each time step are added to produce an

apparent particle track in space. This is not the

track followed by an actual particle, but it is

a useful visual representation of the velocity

time series.

Vector fields with geographic coverage (latitude

and/or longitude) can be plotted as vectors

on a map (Figure 6.5). A Hovmöller diagram

with a time axis and a position axis can also be

used, with vectors plotted as a function of time

and distance.

6.5.3. Spectral Analysis

In Section 6.3, we introduced some concepts

for quantifying basic properties of a time series.

In Section 6.5.2 we described some simple

approaches to viewing the data. More in-depth

analysis of a time series, for instance using spectral

analysis techniques, can yield much more

information about ocean processes such as their

timescales, repeatability, and evolution.

168


FIGURE 6.8 Example of a time series, spectrum, and

spectral aliasing. (a) Tide record at Victoria, British

Columbia (July 29 to September 27, 1975). The heavy dots

are a once per day subsampling of the record. (b) Power

spectrum of the complete tidal record (dashed) and the

subsampled record (solid), showing how the diurnal and

semi-diurnal tidal energy are aliased to periods of 10 days

and longer. Source: From Emery and Thompson (2001).

Spectral analysis, or “Fourier analysis,” is

a straightforward approach to extracting more

information from a time series. Spectral analysis

is useful for determining tidal components

in a time series of sea-surface height or currents

(Figures 6.7a and 6.8a), or for deciding if there

is significant variability at a seasonal or interannual

frequency. “Significance” is a formal

concept for spectral analysis, based on an estimate

of error for the energy at each frequency

and related to the confidence intervals described

in Section 6.3 for probability density functions.

Here we briefly describe some of the basic

concepts. Further information can be obtained

from a number of textbooks and Web sites.

Examples can also be found in Press et al.

(1986), Emery and Thomson (2001), Chatfield

(2004), von Storch and Zwiers (1999), and

Wolfram (2009). The following paragraph draws

partially on Emery and Thomson and notes from

Gille (2005).

Each true time-varying process in the ocean

can be represented as an integral (continuous

sum) of an infinite number of orthogonal basis

functions such as sines and cosines; that is, an

infinite time series can be fit to specified functions

using least squares. If the functions

are orthogonal to each other and if there are

enough of them (e.g., an infinite series), then

the time series can be completely represented

as a sum of these functions. Because many of

the external forcings for the ocean recur regularly,

(orthogonal) periodic functions that describe

both the forcing and the ocean are a

sensible place to start. For instance, the tidal

record in Figure 6.8a clearly includes periodic

components.

In spectral analysis, the orthogonal basis

functions are sines and cosines for a range of

frequencies. With actual data sets, we must

also deal with the finiteness and discrete

sampling of the observed time series and

produce error estimates for the contribution of

each frequency to the overall process.

Spectral analysis is often as useful in the spatial

domain as in the time domain, that is, for yielding

information about spatial scales (wavelengths).

This is especially helpful when analyzing

spatial-temporal data to study waves, which are

characterized by wavelengths and frequencies.

On the other hand, for large-scale oceanography,

other techniques such as empirical orthogonal

functions (Section 6.6) that do not presume


2

1.5

1

0.5

0

Southern Annular Mode index (CPC/NCEP)

FIGURE 6.9 Lowpass filtering

by averaging the time series:

Southern Annular Mode monthly

index from the NCEP Climate

Prediction Center (thin black)

with 1- and 5-year running means

(mid-weight and heavy, respectively),

with uniform weighting.

Data from Climate Prediction

Center Internet Team (2006).

Source: From Roemmich et al.

(2007).

−0.5

−1

−1.5

−2

1980 1985 1990 1995 2000 2005

periodicity can be more useful than spectral

analysis to study spatial structures.

The simplest approach to representing a time

series in terms of sines and cosines projects the

time series onto these functions, using a least

squares fit or Fourier transform (and at first

ignoring the important issues with discrete

sampling and a finite length time series). The

Fourier transform yields an amplitude for

each frequency. Mathematically, the Fourier

transform X(f) of a sample time series x(t), as

a function of frequency f, and its spectral

density S(f) are

Xðf j Þ¼ XN

i ¼ 1

xðt i Þe i2pf jt i

¼ XN

i ¼ 1

x i e

i2pðj 1Þði 1Þ=N

(6.14a)

Sðf j Þ¼jXðf j Þj 2 ; j ¼ 0; .; N 1 (6.14b)

The frequency distribution of the squared

amplitudes (6.14b) is known as a “periodogram.”

The periodogram does not have statistical

value (having large error at each

frequency) because it includes no averaging.

The “power spectrum” is calculated by averaging

the periodogram, and is therefore a statistical

quantity. The averaging is sometimes taken

over multiple realizations of the periodogram,

which can be calculated from multiple realizations

of the time series. (In practice, this means

taking a long time series and chopping it into

shorter time series, calculating the periodogram

for each piece, and averaging them.) Averaging

to create the power spectrum can also be over

a frequency range in the periodogram, thus

reducing the frequency resolution. (Later we

see that this is equivalent to reducing the length

of the time series, so the two types of averaging

are equivalent.)

170


It turns out that the spectrum obtained from

the periodogram is identical to that obtained

as the Fourier transform of the autocovariance

function (which already includes the squared

amplitudes); this latter is the Blackman-Tukey

approach. Since it is now efficient to compute

Fourier transforms using readily available fast

Fourier transform (fft) software (Press et al.,

1986; Matlab, Mathematica, etc.), the periodogram

approach is now the most commonly

used.

The total “energy” in the power spectrum is

equal to the total variance in the time series

(Parseval’s theorem). Thus the area under the

spectrum (in a graph of the spectral amplitude

vs. frequency) is the total variance of the time

series. If the spectrum is normalized by

dividing by the variance, then the area under

the spectrum is 1, and the spectral values give

the fraction of the total variance at each

frequency.

The power spectrum is often displayed

divided by the frequency interval at each

frequency. This is called the power spectral

density (“density” since it is divided by the

interval). The units of power spectral density

are spectral energy/frequency. For instance, for

a sea level spectrum, the units might then be

m 2 /cps (where cps is cycles/second). For a

velocity record, the units would be (m/s) 2 /cps.

For a temperature record, the units would be

( C) 2 /cps.

There are a number of important details to be

considered when working with real, discretely

sampled data from a finite length time series.

We never have an infinite time series. Turning

on and then turning off the sampling of

a continuing process means that a “box car”

has been applied to the actual process (the box

car has multiplied the true time series). The

calculated spectrum then includes the box car;

the sudden jump up and drop down in amplitude

at the beginning and end of sampling has

unfortunate, unwanted spectral properties,

that is, the “ringing” of the Gibbs phenomenon,

which creates unphysical energy at high

frequencies. To avoid this, sample time series

are multiplied by a window that “tapers”

smoothly to zero amplitude at the beginning

and end of the time series. The references listed

at the beginning of this section describe the

commonly used windows.

The actual frequencies that can be analyzed

for a sample time series depend on the total

length of the time series and on the discrete

sampling interval for the time series. (If the

time series is sampled irregularly, there is an

additional set of considerations that are not discussed

here.) In Eq. (6.15), the length of the time

series is T, the sampling interval is 6t, and the

total length of the time series is T ¼ N6t, where

N is the total number of samples.

The lowest frequency, f 0 , that can be resolved

by a given time series is called the “fundamental

frequency”:

f 0 ¼ 1=T ¼ 1=ðNDtÞ (6.15)

in units of Hertz (or u ¼ 2p/T if frequency u is

in terms of radians). However, such a low

frequency relative to the record length is

sampled only once in the record, so confidence

in the amplitude estimate is low. Energy at

frequencies lower than the fundamental frequency

will appear as a trend in the record. It

is common practice to first fit a linear trend to

the time series and remove the trend from the

time series prior to performing the Fourier analysis.

(Recognition of problems with spectral

estimation when very few cycles of an oscillation

have been sampled is vitally important for

large-scale oceanographic data analysis if timescales

of interest are tens to hundreds of years,

especially when looking at climate variability

and change. Oceanographic time series are not

long enough to resolve these low frequencies

very well.)

In the spectrum, the fundamental frequency

is also the difference in frequency between adjoining

frequency components, f 1 and f 2 . That is,


Df ¼jf 2 f 1 j¼1=NDt (6.16)

The two frequencies are well resolved for

6f ¼ 2/N6t and 3/2N6t, just resolved for

6f ¼ 1/N6t, and not resolved for 6f ¼

1/2N6t.

The highest frequency that can be observed

depends on the sampling interval, 6t, because

two samples are required to sample a given

frequency (the “sampling theorem”). The

maximum resolved frequency is

f N ¼ 1=2Dt: (6.17)

This is the Nyquist frequency. As with the fundamental

frequency, the estimate of spectral amplitude

at the Nyquist frequency is poor since the

sampling does not resolve the sinusoidal character.

Note that if f N is the highest frequency

we can measure and if f 0 is the limit of frequency

resolution, then the Nyquist frequency also gives

the maximum number of Fourier components

that can be estimated in any analysis:

f N =f 0 ¼ð1=2DtÞ=ð1=NDtÞ ¼N=2: (6.18)

What happens if there is energy in the time

series at frequencies that are higher than the

Nyquist frequency? Energy from the undersampled

higher frequencies appears in the spectrum

at much lower frequencies. This is called

aliasing, as mentioned in Sections 6.1 and 6.2.

For any actual time series, there will always be

higher frequencies that are not sampled; this

presents a problem only if they have a significant

amount of energy. An example of aliasing from

Emery and Thompson (2001) is shown in

Figure 6.8b. If the well-measured tidal record

in the top panel is subsampled with just one

observation per day, a much lower, erroneous,

frequency appears in the spectrum. When

the spectra of the original record and the subsampled

records are computed, the correct

spectrum has peaks at the well-known tidal

frequencies, while the subsampled record

produces a spectrum without these peaks, but

also with the energy folded back (aliased) into

lower frequencies. This erroneously boosts the

spectral amplitude at lower frequencies.

Figure 6.8b also illustrates the sampling

theorem: the highest frequency that is resolvable

with sampling once per day is 1/(2 days). Thus

the solid curve ends at this frequency.

As another example of aliasing, satellite

altimeters measure SSH. Their orbits sample

a given location every 10 days. However,

at almost all locations in the ocean there is

significant tidal energy at semi-diurnal and/or

diurnal frequencies. There is also undersampled,

high-frequency SSH variability due

to fluctuations in atmospheric pressure and barotropic

motions in the ocean. (Barotropic variability

is solely due to dynamical changes in

surface height without compensation, or “baroclinicity,”

in the ocean interior. Barotropic

variability has much shorter timescales than baroclinic;

see Sections 7.6 and 7.7.) These energetic,

higher frequency signals must be managed when

analyzing altimetric spectra, which can be

approached using models of tides and the barotropic

variability and analyses of atmospheric

pressure (Chelton et al., 2001; Stammer, Wunsch,

& Ponte, 2000). Stammer and Wunsch (1999)

showed a spectrum of SSH from the altimeter,

with an aliased semi-diurnal tidal peak.

The significance of a spectral estimate is

measured with a confidence interval. These are

calculated similarly to those for the basic time

series (Eq. 6.10). The spectral density at a given

frequency is like an energy, meaning that it is the

square of an amplitude. To obtain useful (significant)

spectral estimates, there must be some

averaging. This can be done by either averaging

spectral estimates from many independent time

series sampling the same process, or by averaging

together spectral estimates for a range of

adjoining frequencies. Since the spectral estimate

is then a sum of squares, it has a chi-square

probability density function. (See Emery and

Thomson (2001) to learn about chi-square

172


distributions.) The chosen confidence intervals

are then determined by the chi-square distribution

and knowledge of the effective number of

degrees of freedom in each spectral estimate.

Again this is very similar to Eq. (6.10), but

instead of a Student t-test, the function used to

evaluate is the chi-square distribution.

It is common practice to show the 95% confidence

interval with a spectral estimate. An

example from Rudnick (1997) is shown in

Figure 6.7. In this spectrum, the 95% confidence

interval varies with frequency because the averaging

in the spectral estimate and therefore the

number of degrees of freedom differ with the

frequency.

In spectral analysis, it is assumed that the

physical processes that have been sampled are

stationary. In a stationary process, observations

from one time period yield the same spectrum

as observations from another time period.

However, the underlying processes can change.

Ocean currents that produce instabilities, and

hence eddies, can change with time; for

example, seasonally or in response to climate

variability. The spectra of the eddies would

then change. A more complicated method,

called wavelet analysis, recognizes that the

processes underlying the spectrum might be

changing, and thus the spectrum might vary

with time. Empirical orthogonal functions

(Section 6.6) also do not require stationarity of

the time series, so they can be generally more

useful than spectra for some large-scale (spatial

and time) oceanographic processes.

6.5.4. Filtering Data

We are often interested in isolating

phenomena with a particular frequency in a

data record, whether a time series or spatial

data set. For instance, if the study is focused

on tides, the lower and higher frequencies

present in the data set might be removed. If

the study is focused on decadal variation, the

internal waves or tides might be removed. To

do this, the data set is filtered to remove frequencies

that are not of interest. Filters can be applied

in either the time domain or the frequency

domain. As for other aspects of data analysis

presented in this chapter, only the rudimentary

concepts are provided. Among the many treatments

of filtering, Press et al. (1986) provided

general, practical advice about filtering; Bendat

and Piersol (1986) provided much more of the

complete background and mathematics; and

Emery and Thomson (2001) provided a thorough

treatment as commonly practiced in physical

oceanography. The Matlab signal processing

toolbox includes many different filters as well

as the capability to design filters.

In the time domain, the output from a filter at

a given time is a weighted sum of the input data

from a range of times. For example, to filter in

the time domain,

yðt j Þ¼ XN

i ¼ 1

wðt i t j Þxðt i Þ j ¼ 1; N (6.19a)

where x is the original data, y is the output, and

the w’s are the weights, which depend on the

difference in time between the data point and

the output. What does a given filter (choice of

weights) do to the frequencies in the time series?

To answer this, the filter weights can be Fourier

transformed to the frequency domain and

plotted as a function of frequency. This is called

the frequency response of the filter.

To filter in the frequency domain, the time

series of data are first Fourier transformed to

form a periodogram (unaveraged spectrum).

Then the periodogram is filtered, with weights

that depend on frequency, and the results are

Fourier transformed back to the time domain.

For example:

Yðf i Þ¼ XN

i ¼ 1

Wðf i ÞXðf i Þ i ¼ 1; N (6.19b)

where f is frequency, X is the Fourier transformed

data, Y is the filtered spectrum, and W

are the weights in the frequency domain. An

MULTIDIMENSIONAL SAMPLING 173

equivalent method is to design a filter shape (the

weights, W) in the frequency domain and Fourier

transform this filter to the time domain; the

resulting time series then becomes the time

domain weights, w.

A “lowpass” filter removes high frequencies

above a cut-off frequency, retaining only the

low frequencies. A “highpass” filter is the opposite,

retaining high frequencies only. A “bandpass”

filter retains frequencies in the middle of

the record, removing both the low and high

frequencies. This filter is characterized by its

central frequency and bandwidth.

Lowpass filtering is equivalent to smoothing

a time series. This is the easiest type of filtering

to understand within the time domain. Box car

averaging is a simple lowpass method, in which

a segment of the data record is averaged, with

equal weight given to each point in the segment;

the uniform weights resemble a box. The box car

can be moved through the record, with overlapping

segments; this is called a running mean.

The weights in Eq. (6.19a) for a box car filter

are all the same size up to the length of the

segment (summing to 1, producing an average

value of the data over the chosen time interval)

and 0 for all other times. An example of oneand

five-year running means, with box car

weighting, for the climate index called the

Southern Annular Mode (Section 13.8) is shown

in Figure 6.9 (after Roemmich et al., 2007).

Box car averaging is often all that is necessary

for a given purpose. However, the frequency

response of a box car filter can be undesirable

because of the Gibbs phenomenon: the sudden

drop to 0 for the weights means that there is

high-frequency ringing in the filtered data set.

Weights chosen for a low pass filter can be

tapered to zero at the ends of the filter, much

like windowing in spectra.

Lowpass filtering can also be done spectrally

in the frequency domain rather than the time

domain. The time series can first be Fourier transformed.

All undesired high frequencies can then

be set to zero amplitude (or tapered to zero

amplitude). Then the filtered data record can be

reconstituted using an inverse Fourier transform.

Bandpass and highpass filtering are conceptually

easiest to understand in the frequency

domain, since the objective is to remove certain

frequencies. The crudest method is to take the

Fourier transform of the time series, then set

the amplitude of all undesired frequencies to

zero, and then take the inverse Fourier transform

to reconstitute the time series, which will now be

missing all the undesired frequencies. However,

such simple removal of undesired frequencies

also creates problems in the inverse Fourier

transform similar to the Gibbs phenomenon.

Therefore, it is desirable to taper (window)

when removing undesired frequencies.

Bandpass and highpass filtering are often

carried out in the time domain. The crudest

method of highpass filtering is to subtract the

lowpass record from the original data record.

Bandpass filtering can be produced by successive

application of low and highpass filtering

(Emery & Thomson, 2001). However, it is more

desirable to design these filters to diminish the

Gibbs phenomenon. The time domain filter

can be constructed as the Fourier transform of

a frequency domain filter. For bandpass filters,

the narrower the desired band of frequencies,

the longer the time series must be to produce

the narrow band; this is readily understood

from the wide shape of the Fourier transform

of a very narrow signal.

There are many subtleties associated with

filtering that are not described here.

6.6. MULTIDIMENSIONAL

SAMPLING

The ocean is often sampled in time and in at

least two dimensions in space. This is especially

true in the present era of satellite programs,

which collect data over large parts of the ocean

surface at regular intervals. With regular data

for the whole surface, observers often wish

174


to extract signals that indicate processes. For

instance, ENSO is a time-varying climate process

with an underlying quasi-periodicity of

3 to 7 years (see Figure 10.28b). Over the course

of several years, different parts of the sea surface

in the tropical Pacific have changed in height

and temperature at different times. There is

also variability at the sea surface, in the atmosphere,

and in vegetation far from the tropical

Pacific; observers are interested in knowing

how much of the variability can be traced to or

participates in ENSO.

Covariance and correlation are the most basic

calculations used to analyze observations of

different properties in different locations and/

or at different times (Sections 6.1 and 6.3). These

calculations show how temperature in the

western Pacific varies with temperature in the

eastern Pacific; or how the correlation of time

series of some property such as surface temperature

at all locations on the globe with the time

series of, say, a climate index can provide useful

displays of the spatial distribution of that

climate variation (see the figures in Chapter

S15 on the textbook Web site).

Beyond point-to-point correlation analysis, it

is useful to look at the large-scale geographic

“modes” of variability. This can be approached

by extracting wavelike signals from the data

sets, using spectral analysis, or by allowing the

data set to define its own spatial patterns. The

latter approach, which uses empirical orthogonal

functions, can produce patterns that can

clearly be identified as ENSO, the Southern

Annular Mode, and so forth. The time series

of the identified climate pattern can then be

correlated with observed time series at different

locations to begin to identify sources of local

variability.

We first briefly describe common methods

of dealing with both spatially and temporally

sampled variability (spectral analysis and empirical

orthogonal functions), and then common

methods of displaying time-averaged data with

spatial distribution (climatologies and atlases).

6.6.1. Multidimensional Time Series

Data and Empirical Orthogonal

Functions

Spectral analysis can be applied to spatial and

temporal sampling. Wavenumber-frequency

spectra (bispectra), which represent both the

spatial and temporal aspects of the data, are

very useful in studying waves that are typically

described theoretically in terms of dispersion

relations that relate frequency and wavenumber

(Section 8.2). Wave fields are typically at least

two-dimensional in space, so wavenumberwavenumber

spectra are also useful. Figure 6.10

shows two recent examples of frequency-wavenumber

spectra: (a) for large-scale, much lower

frequency equatorial waves in the Pacific Ocean

(Shinoda, Kiladis, & Roundy, 2009) and (b) for

very high frequency surface gravity waves

(Herbers, Elgar, Sarap, & Guza, 2002). In both

panels, theoretical dispersion relations are overlaid.

In Figure 6.10b, the observed spectrum can

then be used to determine if a given theory

accounts for the observations.

A drawback of spectral analysis is its underlying

assumption that the processes are periodic

(in space or time, depending on how it is

applied). While many ocean processes indeed

satisfy this assumption d ocean surface waves,

tides, and large-scale waves such as Kelvin and

Rossby waves (Figure 6.10a) d many largescale

ocean and climate processes do not. This

is especially true of the spatial patterns for

large-scale ocean responses to changing forcing,

where the geography begins to dominate the

patterns. Therefore it is useful to move beyond

spectral analysis to find basis functions that

better represent the underlying ocean processes.

Empirical orthogonal functions (EOFs) are regularly

used in oceanography, meteorology, and

climate science for analyzing space-time data

sets such as satellite or SST time series. EOF

analysis was introduced for meteorology by

Lorenz (1956). It is similar to principal component

analysis used in other sciences (and


FIGURE 6.10 Examples of frequency-wavenumber spectra. (a) Equatorial waves (Kelvin and Rossby) from SSH

anomalies, compared with theoretical dispersion relations (curves). Figure 6.10a can also be found in the color insert. Source:

From Shinoda et al. (2009). (b) Surface gravity waves: observed two-dimensional spectrum ( ) ) averaged over wavenumber at

each frequency, and compared with several theoretical dispersion relations. Source: From Herbers et al. (2002).

occasionally in meteorology and oceanography).

Unlike spectral analysis, in which the

temporal and spatial dependence are represented

as sines and cosines, the EOF procedure

defines its own set of functions that can be

used to describe the process most efficiently.

Each EOF is “orthogonal” to the others, which

means that each one represents something

unique about the process. (In spectral analysis,

each sine and cosine function is orthogonal to

all the others.)

EOFs are determined through a linear least

squares process, minimizing the difference

between the observations and the EOFs; that

is, the basic ideas presented in Section 6.3.4 for

least squares apply to finding these much

more complex functions. We do not present

any of the method here, but refer to texts such

as Wilks (2005) and von Storch and Zwiers

(1999). Following the procedures in these texts

and in basic publications on EOFs, calculation

of EOFs is straightforward because the useful

linear algebra software is easily available (Matlab,

Mathematica). Other multivariate approaches

are also described in these texts,

including canonical correlation analysis, in

which the pattern with the highest possible

correlation between two time series is sought.

EOFs are shown in this text in reference to

modes of climate variability: the Pacific Decadal

Oscillation (PDO), North Pacific Gyre Oscillation

(NPGO), the Arctic Oscillation, the Southern

Annular Mode, and so forth (see Chapter S15

on the textbook Web site). Other climate modes

introduced throughout, such as the ENSO and

the North Atlantic Oscillation, are also easily

and often described in terms of EOFs.

EOFs are typically ordered by amplitude

(percentage of variance of the observations

explained by that EOF); that is, the first mode

explains the most variance of the signal (by

design because of the least squares approach),

176


the second mode explains as much of the remaining

variance as possible, and so on. Typically,

only the two or three EOFs with the largest

amplitudes are significant while the others might

be below the noise level of the observations. The

major climate modes listed in the previous paragraphtendtobefirstorsecondEOFs.

In a classic oceanographic application of EOF

analysis, including the reasoning for an EOF

approach and an extensive appendix describing

the EOF method, Davis (1976) analyzed historical

(1947e1974) SST and sea level pressure

(SLP) anomalies in the North Pacific Ocean.

Using data that had already been gridded to

latitudes and longitudes for each month, he constructed

monthly anomalies by removing the

long-term monthly mean at each grid point,

and then calculated the EOFs from the anomalies

(Figures 6.11 and 6.12). These figures are

typical of EOF displays, including the spatial

FIGURE 6.11 Example of empirical orthogonal functions (EOFs): the eight principal EOFs describing the sea surface

temperature anomalies. ÓAmerican Meteorological Society. Reprinted with permission. Source: From Davis (1976).


1.00

0.75

0.50

0.25

EOF presentations often include the computed

time series of the amplitude of the dominant

EOF modes. These time series can then be

analyzed using spectral analysis or some other

approach, or simply employed to study correlations

with other fields and other modes. Davis

(1976) presented the frequency spectrum but

not the actual time series. Time series of climate

indices that are actually EOF amplitudes are

commonly used; for instance, the Southern

Annular Mode index seen in Figure 6.9.

6.6.2. Climatologies and Atlases

0 5 10

pattern of the principal EOFs, ranked in terms of

variance of the overall signal explained by that

EOF, and a plot of the variance for each mode.

The spatial patterns in Figure 6.11 illustrate

non-sinusoidal EOFs and how each successive

mode has more spatial complexity and is visually

orthogonal to the other modes.

Davis (1976) showed that the SLP variability

could be explained by fewer EOFs than the SST

variability (Figure 6.12), likely due to the inherently

smoother SLP field. A second important

result was the spatial distributions of the modes

(Figure S6.3 on the textbook website). The first

EOF of SLP looks like the Aleutian Low pattern

and the first EOF of SST is the temperature variability

that accompanies variations in the Aleutian

Low strength. This first EOF is essentially

the North Pacific part of the PDO, also called

the North Pacific Index (see online materials

that include Chapter S15). The second EOF is

now associated with the North Pacific Gyre

Oscillation (NPGO) (DiLorenzo et al., 2008).

M

15 25

FIGURE 6.12 The cumulative fraction of total sea

surface temperature (circles, o) and sea level pressure

(triangles, 6) anomaly variance accounted for by the first M

empirical orthogonal functions. ÓAmerican Meteorological

Society. Reprinted with permission. Source: From Davis

(1976).

In oceanography and meteorology, it is useful

to use mean values distributed geographically

to produce a climatology. A climatology is generally

understood to be the mean value over many

years, usually including at least several decades.

The mean can be over all months, but often the

mean values are constructed for individual

months or seasons, in which case they can be

referred to as monthly or seasonal climatologies.

Climatologies are usually constructed from

observations that are irregularly sampled in

time. Therefore some sort of weighting of each

data point is important so that, for instance,

well-sampled summers and poorly-sampled

winters do not bias the mean value toward

summer. It can be useful to first construct short

period mean values, such as daily averages,

before constructing the monthly to annual

mean. Data gaps are often filled by some interpolation

or mapping procedure before

averaging.

Observations used for climatologies are also

usually irregularly sampled in space; the same

weighting and data gap issues that apply for

time are applied here. Objective mapping is

a common method for producing a spatially

gridded data set. Simple geographic binning is

also used if there are many observations.

Construction of a climatology almost always

involves a data quality step, which can be very

extensive. Published climatologies, therefore,

178


are often accompanied by carefully qualitycontrolled

data sets.

Climatologies are essential for large-scale numerical

modeling and also for data assimilation,

which is based on numerical modeling. A good

mean field is essential as a startup condition in

these models since “spinup” from a state of

rest without stratification would be hopelessly

inefficient.

Commonly used ocean climatologies include

the National Oceanographic Data Center’s

(NODC) World Ocean Atlas (WOA05; NODC,

2005a) and its antecedents, beginning with

Levitus (1982), which are based on the archived,

quality-controlled original data World Ocean

Data (most recent version WOD05; NODC,

2005b). An example from WOA05 that illustrates

the climatological salinity at 500 m with several

types of available indications of error is shown

in Figure 6.13; the online climatology from

NODC includes many other properties and

depths as well. Another hydrographic climatology

in general use, along with its qualitycontrolled

data set (Hydrobase), was produced

for the Atlantic, Pacific, and Indian Oceans by

Lozier, Owens, and Curry (1995), Macdonald,

FIGURE 6.13 Illustration of a climatology: Salinity at 500 m. (a) Climatological annual mean, (b) data distribution,

(c) standard deviation, and (d) standard error. Many other properties and depths are also available online (NODC, 2005a).

Source: From Antonov et al. (2006).

WATER PROPERTY (WATER MASS) ANALYSES 179

Suga, and Curry (2001), and Kobayashi and Suga

(2006), respectively. Hydrographic data in this

climatology were averaged along isopycnal

surfaces, producing useful improvements in

regions of large isopycnal slopes (strong

currents) over those averaged on constant pressure

surfaces.

Atlases are more vaguely defined than climatologies.

Traditionally atlases were books filled

with maps d or for oceanography, vertical

sections of ocean properties in addition to

maps d for visualization of mostly the mean

fields, or at most, seasonal fields. Atlases using

data from the International Geophysical Year

(1957e1958) were published in the 1960s (Fuglister,

1960; Worthington & Wright, 1970). Atlases of

vertical sections from the Geosecs expeditions of

the 1970s were also published and are widely

used (Bainbridge et al., 1981 to 1987). The World

Ocean Circulation Experiment atlases of vertical

sections and maps are now published both in

print and online (Orsi &Whitworth, 2005; Talley,

2007, 2011; Koltermann, Jancke, & Gouretski,

2011). Many figures from these atlases are reproduced

in other chapters of this text.

Modern atlases include graphics based on

averaged data (climatologies) and most are

now digital. Perhaps the first of these, a print

atlas based on all available NODC data, was

published by Levitus (1982); this was produced

specifically to provide climatological fields for

general ocean circulation models. The latest

version of the NODC atlases is the World Ocean

Atlas 2005 (WOA05), which is exclusively

digital (NODC, 2005a). Figure 6.13 is from this

NODC atlas. Some modern digital atlases also

include software for the user to produce their

own graphics based on individual data. Java

Ocean Atlas (JOA; Osborne & Swift, 2009) is

the basis for the DVD distributed with this

text. Ocean Data View (ODV, 2009), developed

by R. Schlitzer in the 1990s, is a widely used

display package and database that is easily

adapted to optimum multiparameter analysis

(Section 6.7.3).

6.7. WATER PROPERTY (WATER

MASS) ANALYSES

Much of the descriptive oceanography in

other chapters of this text is associated with

the large-scale ocean circulation and water

mass distributions. Data sets that describe this

circulation extend back more than a century.

Techniques to work with the data to discern

sources and influence of water masses have

been based on using several different characteristic

properties of the water masses in addition

to its density to trace them. This section

describes some of these traditional techniques,

which remain effective for studying water

mass distributions and the associated ocean

circulation (Sections 6.7.1 and 6.7.2). These

methods are being replaced by more statistical

techniques as the data sets grow. A technique

that has been widely adopted in recent years is

optimum multiparameter analysis (OMP), which

is a least squares approach to estimating the

fraction of a given source water (Section 6.7.3).

6.7.1. Analysis Using Two

Characteristics

Observed properties such as potential

temperature, salinity, dissolved oxygen, and so

forth, may have important correlations with

each other. (Density is not included because it

is derived from temperature and salinity.) These

dependencies may be regional or have time

variations. The high correlations between properties

arise because most ocean water masses

(Section 4.1) acquire their characteristics at the

surface of the sea in particular localities.

The water properties are determined there by

the local climate, and when the water sinks

along density surfaces it carries these properties

with it. The characteristic and unique combination

of different water properties that arises

from a given source or process in the ocean

provides the definition of a given water mass

(Section 4.1). Water masses are useful because

180


they can be recognized by these combinations

and hence the associated processes deduced

from their distributions and modifications.

To show these combinations, characteristic

diagrams of the water properties were first introduced

by Helland-Hansen (1916), who plotted

temperature against salinity (T-S plots) for individual

oceanographic stations. Potential temperature

rather than temperature is almost always

used now to remove the adiabatic effect of pressure

on temperature. (Use of q is important even

in water as shallow as 100 m, if one is attempting

to identify the water with its surface source or

follow it down into the ocean or deduce its mixing

properties.) Potential temperature-salinity

diagrams are used throughout this text to illustrate

water mass distributions.

Each point on the q-S diagram corresponds

to a particular potential density, so potential

density is often contoured (Figure 6.14). Different

reference pressures for potential density can be

used, as explored with Figure 3.5. (Neutral

density is not contoured on q-S plots since it is

defined empirically based on the actual temperature

and salinity properties of the water column,

rather than based on the equation of state; therefore

it is not defined for the full two-dimensional

potential temperature/salinity plane. It could,

however, be indicated for given observations.)

The q-S diagram in Figure 6.14 illustrates water

types, water masses and mixing diagrams

(Section 4.1.1). Water types are points in property-property

space, and represent source waters.

As the water advects away from its source and

(a)

SALINITY

34 35

A

(b)

% SALINITY %

36

34 35

20

t = 26 ATL UPPER/WATER

36

26

27

15

27

B

28

10

NORTH

ATL.

28

C

5

TYPES A,B,C : O

ANTARCTIC

0

MASSES A - B :

INTERMEDIATE ANT. BOTTOM

B - C :

–2

C - A

WATER TYPES :

A - B - C :

2 - TYPE MIXTURES :

ABC = MIXING TRIANGLE

WATER MASS :

FIGURE 6.14 Example of a potential temperature (q)-salinity diagram. (a) Schematic showing three water types and their

mixing products. (b) q-S diagram from the central North Atlantic with water masses labeled, illustrating how mixing

connects the extrema. The contoured field on the diagrams is the density s t since this figure is reproduced from an earlier

version of this text, although as indicated in Chapter 3, it is advisable to use a potential density parameter.


mixes with other waters from other sources, its

identifying properties spread to a range of properties.

If the water type originates as a vertical

extremum such as a salinity minimum, then as

it mixes downstream, the vertical extremum

might remain, marking the influence of the original

water type. The overall envelope of these

gradually mixing properties identifies the water

mass. However, the sources of waters below the

pycnocline are so well separated in space and

hence in properties that the water types and

masses are relatively easily defined.

It has long been the practice to compute fractional

mixing rates of end points (source waters)

for a given water parcel, based on q-S properties.

This has been readily extended to other

conservative and possibly non-conservative

tracers. A good assumption for conservative

tracers is that mixing occurs along straight lines;

for mixing of q and S, this is along straight

lines in the q-S plane (Section 3.5.5). Nonconservative

tracers, such as oxygen and nutrients,

are more problematic since they depend

on water parcel age. Extended methods include

use of conservative parameters such as “PO,”

“NO,” and “N ) ” that take advantage of the

linear mixing assumption (Broecker, 1974;

Gruber & Sarmiento, 1997). Redfield ratios

(Section 3.6) are based on fixed proportions of,

say, phosphate (“P”) and nitrate (“N”) production

while oxygen (“O”) is consumed. As soon

as additional properties are included, error estimates

are desired, and it is understood that end

points are not necessarily well defined, and

a more quantitative approach becomes useful.

This end point mixing practice has therefore

evolved into the more statistical approach of

OMP analysis (Section 6.7.3).

The additional independent information

about source waters and mixing available from

tracers other than potential temperature and

salinity is evident in two-parameter plots for

a multitude of tracers (Figure 6.15). When

more than two parameters are available, one

can think in terms of multi-dimensional space,

and the relationships between the parameters

that reveal the source waters and mixing rapidly

become more easily utilized through OMP.

Finally, property-property plots such as

Figure 6.15 are useful visual tools for checking

data quality. Since locally envelopes of profiles

in property-property space can often be quite

“tight,” or have small variance, outliers can be

identified and flagged for additional quality

checking. In Figure 6.15, for instance, slightly

low salinity values in the nitrate and silicate

versus salinity plots suggest that each step

in obtaining these values be checked, often by

going back to original log sheets and laboratory

notebooks to see if there were any issues or

uncertainties in data collection or analysis procedures.

(These particular values were found to be

accurate and were therefore retained.)

6.7.2. Volumetric q-S Characteristics of

Ocean Waters

The volume of water with a given property or

set of properties can be a useful diagnostic of

relative quantity, or reservoir size, of the ocean’s

water masses. This technique for potential

temperature-salinity was pioneered by Montgomery

(1958) and subsequently reworked for

the world oceans by Worthington (1981; see

Figure 4.17 in this text). In principle, volumetric

assays for any set of properties, not just potential

temperature and salinity, can be produced

and displayed. Column inventories of properties

such as chlorofluorocarbons and CO 2 have

emerged as important tools for understanding

the ocean’s role in the global carbon system.

The volumetric q-S diagram in Figure 4.17

was produced by choosing a “bin” size for

potential temperature and salinity (e.g., 0.1 C

and 0.01 psu) and the volume was calculated

for each bin from observed oceanographic

data. When done originally without benefit of

computer interpolation, this was an extremely

tedious exercise. It can be done relatively easily

now, and is a feature of the JOA package

182


FIGURE 6.15 Example of property-property plots for a variety of different properties, for the Japan/East Sea. Source:

From Talley et al. (2004).

distributed as part of this text, in the supplemental

materials on the Web site.

The last form of traditional, three-dimensional

characteristic diagram we show is the

temperature-salinity-time (T-S-t) diagram. This

is a compact way of showing the sequence of

combinations of water properties with time.

As an example, Figure 6.16 shows monthly

mean values for three zones of the Australian

Great Barrier Reef lagoon (Pickard, 1977). In


30°

32 33

SALINITY %

34 35 36

FEB

CENTRE

JAN

MAR

FEB

NORTH

JAN

DEC

s t = 20

MAR

21

APR

MAY

DEC

FEB

MAR

JAN

NOV

TEMPERATURE °C

25

10°

20°

30°

40°

120°

130°

140°

GT

BARRIER

REEF

AUSTRALIA

150° E

NORTH

22

CENTRE

SOUTH

APR

MAY

23

JUN

JUN

24

APR

JUL

MAY

JUL

AUG

JUN

DEC

NOV

OCT

SEP

AUG

SEP

NOV

SOUTH

OCT

SEP

OCT

50° S

AUG

JUL

20

25

FIGURE 6.16 Temperature-salinity-time (T-S-t) diagrams for shallow lagoon waters inside the Great Barrier Reef.

Source: From Pickard (1977).

the south, the annual variation is mainly in

temperature. In the north there are large variations

of both temperature and salinity, while

in the center zone there is an extreme salinity

variation. The reason for the differences is that

the north and center zones are subject to heavy

monsoonal rains in the austral summer (January

to April) while the south zone escapes these.

The very low salinity in the center is due to

the rivers that drain much larger inland areas

than the smaller rivers in the north. In situ

temperatures used for these T-S-t diagrams,

which is appropriate for these surface data.

6.7.3. Optimum Multiparameter

Analysis

All waters in the interior of the ocean are

understood to be a mixture of waters that have

some well-defined source at the sea surface.

Those sources have associated water properties

that depend on location. All measured chemical

properties can be used in some way to

help define the source waters and the relative

mixing. Mixing in the ocean is mostly linear

and hence proportional. For a given water

parcel with several measured properties, there

184


are likely to be multiple source waters. If there

are more properties than source waters, then

formally the problem of determining the relative

mixture of the source waters in the parcel

is “over-determined,” meaning there are more

equations (one for each property) than unknowns

(proportion of each source water).

A formal method for determining the relative

proportions of the source waters was introduced

by Tomczak (1981), and developed

formally with a least squares approach by

Mackas, Denman, and Bennett (1987) and

others, including Tomczak. Software written in

Matlab was developed for general use by M.

Tomczak and has been further developed and

provided for general use by Karstensen (2006),

whose Web site also contains practical information

about OMP and a bibliography. The output

of OMP analysis for a given observed water

parcel is the fraction of each assumed source

water, which should add in total to 1.0. OMP

thus first requires selection of at least two source

waters, with as many observed parameters as

possible. If, for instance, two source waters are

assumed, with six parameters each (temperature,

salinity, oxygen, nitrate, silicate, and potential

vorticity), then there would be six equations

with two unknowns for each water parcel;

hence this is an overdetermined system. OMP

finds the best least squares solution, that is, the

best choice of fractions of the source waters for

each observed water parcel.

Mathematically, for an example of three

conservative parameters and two source waters,

the linear equations expressing the fractions of

each source water for a given water parcel are

x 1 q 1 þ x 2 q 2 ¼ q obs þ R q

x 1 S 1 þ x 2 S 2 ¼ S obs þ R S

(6.20)

x 1 PV 1 þ x 2 PV 2 ¼ PV obs þ R PV

x 1 þ x 2 ¼ 1 þ R M

where the conservative parameters chosen here

are potential temperature (q), salinity (S), and

potential vorticity (PV). The fourth equation is

conservation of mass. (It is not necessary to

include PV; if included, the system is overdetermined.

It is only necessary that the system

not be underdetermined.) The x’s are the

mass fractions of the two source waters in

the observed (“obs”) water parcel. The R’s on

the right-hand side are residuals that permit

solution in a least squares manner of this overdetermined

system of four equations and two

unknowns. Solution of the system proceeds

by minimizing the squared residuals R (difference

between the left-hand sides and the

observed values) for each equation simultaneously.

This linear algebra step is omitted

here, since we are not introducing this level of

mathematics in this textbook. A complete

description of the remaining steps and procedures

is available in the cited references (Mackas

et al., 1987; Karstensen, 2006) as well as in

numerous papers that use OMP (e.g., Tomczak

& Large, 1989; Maamaatuaiahutapu et al., 1992;

Poole & Tomczak, 1999).

OMP analysis often includes a constraint that

the fractions of the source waters be non-negative

(Mackas et al., 1987). A measured water

parcel can fall outside the a priori range of the

source water characteristics ending up with

a non-physical negative fraction. (This does

not necessarily mean that the source waters

need to be redefined, unless so many observations

yield negative fractions that the a priori

choice of source waters is clearly inadequate.)

For instance, in the Kuroshio-Oyashio region,

the obvious source waters are “pure” Kuroshio

and “pure” Oyashio water. However, there

could be fresh, near-coastal water parcels that

might lie outside the assumed ranges. If a nonnegativity

constraint is enforced, then such

a parcel would be assigned a 1.0 fraction of Oyashio

and 0.0 fraction of Kuroshio water. It might

sometimes be better not to impose a constraint,

because information about whether water properties

can be explained by chosen source waters

can be lost.

GLOSSARY 185

We show an example of OMP application

in the southwestern Atlantic, where numerous

water masses meet in the Brazil-Malvinas

confluence (Maamaatuaiahutapu et al., 1992).

Their analysis used six properties (temperature,

salinity, oxygen, phosphate, nitrate, and

silicate) plus mass conservation, and seven

source water types (point sources). This is an

exactly determined rather than overdetermined

system. Three of the water mass fraction

sections are shown in Figure 6.17. The source

waters here are well separated in properties,

and also dominate different vertical layers.

The result is quantitative information on the

mount of each water mass at each location on

the section, as opposed to simply subjective

labeling, or a more traditional attempt at calculating

water mass fractions using just temperature

and salinity.

The global maps of North Atlantic Deep

Water and Antarctic Bottom Water fractions

shown in Chapter 14 (Figure 14.15), from Johnson

(2008), are results of OMP analysis.

Stn 308 307 306 305

0

2500

5000

Stn 308 307 306 305

0

2500

5000

AAIW

UCDW

Stn 308 307 306 305

0

2500

5000

WSDW

ABOVE 0.90

0.75 - 0.90

0.50 - 0.75

0.25 - 0.50

0.10 - 0.25

BELOW 0.10

FIGURE 6.17 Example of optimum multiparameter

(OMP) water mass analysis. Southwestern Atlantic about

36 S, showing the fraction of three different water masses.

Antarctic Intermediate Water, AAIW; Upper Circumpolar

Deep Water, UCDW; and Weddell Sea Deep Water, WSDW.

This figure can also be found in the color insert. Source: From

Maamaatuaiahutapu et al. (1992).

OMP analysis can be carried out along quasiisopycnals

if desired. If isopycnal mixing is

assumed, then temperature and salinity are

not independent. Since potential density,

regardless of how it is referenced in pressure,

is not conserved when two water parcels mix

(because of cabbeling), OMP can reveal the

extent of cabbeling (Yun & Talley, 2003).

Approaches to determining the distribution

of different source waters through the ocean

are continually being updated and improved.

Thus this section is just an introduction to the

general topic, and creative approaches are

strongly encouraged.

GLOSSARY

The following list summarizes a number of

the basic terms introduced in this chapter.

Accuracy Difference between an estimate and the “true”

value. High accuracy means that this difference is small.

Aliasing Folding of spectral energy above the Nyquist

frequency back into the frequencies below the Nyquist,

creating higher spectral energy at these frequencies than

is actually in the time series.

Anomaly Difference between an observation and the mean

value, regardless of how the mean value is defined.

Climatology Time mean values of a geographically mapped

field.

Correlation Normalized version of the covariance. It is

equal to the covariance divided by the product of the

standard deviations of the two variables. Correlation

ranges from e1 toþ1.

Covariance A measure of the covariability of two variables,

computed as the averaged sum of the cross-product of the

variations from the respective means of the two variables.

Determination or observation Actual direct measurement

of a variable, e.g., the length of a piece of wood using

a ruler. Synonyms include observation, measurement, or

sample.

Empirical orthogonal functions (EOFs) Set of orthogonal

basis functions that can completely describe (sum to)

a given field. EOFs are often used to describe the spatial

structure of a time-varying field in place of spectral

analysis in the spatial domain.

Estimation Value for one variable derived from one or more

determinations (either of the variable of interest or of

other related variables), for example, the estimation of

salinity from the determination of conductivity and

186


temperature. This also refers to the use of repeated

“determinations” to define a statistical parameter such as

the mean or standard deviation. Thus, we can speak of an

“estimate of the mean”.

Filter To output a data set as a weighted sum of the original

data.

Gaussian population or distribution Probability density

function (pdf) characterized by a symmetric bell curve

defined by a mean and a variance (or standard deviation).

It is also known as a “normal” population or

distribution.

Inverse methods Ways to find the best estimate of an

unknown quantity in an underdetermined system,

usually using a least squares approach. In large-scale

physical oceanography, this has most often been applied

to estimating geostrophic reference velocities.

Least squares methods Ways to fit one function or data

set to another function and/or data set by minimizing the

sum of the squared differences between the two series.

Mean Average of a series of measurements over a fixed time

interval such as a week, a month, a year, and so forth, or

over a specific spatial interval (square kilometer,

a 1 degree square, a five degree square, etc.).

Nyquist frequency Highest detectable frequency in a time

series, equal to half the sampling frequency of the time

series.

Objective mapping A statistically unbiased method of

mapping irregularly spaced observations.

Precision Difference between one estimate and the mean of

several obtained by the same method, that is, reproducibility

(includes random errors only).

Probability density function (pdf) Sampling population

from which the data are collected. This can be depicted

by a histogram showing the frequency of occurrence of

each data value.

Random error This results from basic limitations in the

method, for example, the limit to one’s ability to read the

temperature of a thermometer. It is possible to determine

a value for this type of error by statistical analysis of

a sufficient number of measurements because it affects

precision. Truly random errors have a Gaussian distribution

with zero mean.

Standard deviation Square root of the variance.

Synoptic sampling A way of sampling the conditions as

they exist at a given time over a broad area (a snapshot).

Systematic error or bias Error that results from a basic (but

unrealized) fault in the method that causes values to be

consistently different from the true value. Systematic

error cannot be detected by statistical analysis of values

obtained and affects accuracy.

Variance Mean square difference between a sample value

and the sample mean.

C H A P T E R

7

Dynamical Processes for Descriptive

Ocean Circulation

The complete version of this chapter

(Chapter S7) appears on the textbook Web

site http://booksite.academicpress.com/DPO/.

The sections and equations are identical, but

the explanatory text and figures are greatly truncated

in this book. Figures, chapters, and

sections that appear only on the Web site are

denoted by “S” in their name such as

Figure S7, Chapter S7, and so forth. Tables

mentioned in this chapter appear only on the

Web site.

7.1. INTRODUCTION:

MECHANISMS

Ultimately, motion of water in the ocean is

driven by the sun, the moon, or tectonic processes.

The sun’s energy is transferred to the ocean

through buoyancy fluxes (heat fluxes and water

vapor fluxes) and through the winds. Tides create

internal waves that break, creating turbulence

and mixing. Earthquakes and turbidity currents

create random, irregular waves including

tsunamis. Geothermal processes heat the water

very gradually with little effect on circulation.

Earth’s rotation profoundly affects almost all

phenomena described in this text. Rotating

fluids behave differently from non-rotating

fluids in ways that might be counterintuitive.

In a non-rotating fluid, a pressure difference

between two points in the fluid drives the fluid

toward the low pressure. In a fluid dominated

by rotation, the flow can be geostrophic, perpendicular

to the pressure gradient force, circling

around centers of high or low pressure due to

the Coriolis effect.

Ocean circulation is often divided conceptually

into wind-driven and thermohaline (or

buoyancy-dominated) components. Wind causes

waves, inertial currents, and Langmuir cells. At

longer timescales, which involve the Coriolis

effect, wind drives the near-surface frictional

layer and, indirectly, the large-scale gyres and

currents that are usually referred to as the

wind-driven circulation. Thermohaline circulation

is associated with heating and cooling

(“thermo”), and evaporation, precipitation,

runoff, and sea ice formation, all of which

change salinity (“haline”). Thermohalinedominated

circulation is mostly weak and

slow compared with wind-driven circulation.

In discussing thermohaline effects, it is common

to refer to the overturning circulation, which

involves buoyancy changes. The energy source

for thermohaline circulation importantly

includes the wind and tides that produce the

turbulence that is essential for the diffusive

upwelling across isopycnals that closes the thermohaline

overturning. Both the wind-driven


187



188

7. DYNAMICAL PROCESSES FOR DESCRIPTIVE OCEAN CIRCULATION

and thermohaline circulations are almost

completely in geostrophic balance, with the

forcing that drives them occurring at higher

order.

7.2. MOMENTUM BALANCE

Fluid flow in three dimensions is governed by

three equations expressing how velocity (or

momentum) changes, one for each of the three

physical dimensions. Each of the three

momentum equations includes an acceleration

term (how velocity changes with time), an advection

term (see Section 5.1.3), and forcing terms:

Density ðAcceleration þ AdvectionÞ

¼ Forces per unit volume (7.1)

Forces per unit volume ¼ Pressure

gradient force þ Gravity þ Friction (7.2)

Expressions (7.1) and (7.2) are each three equations,

one for each of the three directions (e.g.,

east, north, and up). The terms in Eqs. (7.1)

and (7.2) are illustrated in Figure 7.1.

The inclusion of advection means that

Eq. (7.1) is the expression of momentum change

in a Eulerian framework, where the observer sits

at a fixed location relative to Earth. Equation

(7.1) can be written without the advection

term, in a Lagrangian framework, where the

observer drifts along with the fluid flow. (See

also Section S16.5 in the online Supplement.)

For a rotating geophysical flow, we, as

observers, sit within a rotating “frame of reference”

attached to the rotating Earth. For this

reference frame, the acceleration term on the

left-hand side of Eq. (7.1) is rewritten to separate

local acceleration due to an actual local force

from the effects of rotation. The effects that are

separated out are the centrifugal and Coriolis

accelerations (Section 7.2.3).

The frictional force in Eq. (7.1) leads to dissipation

of energy due to the fluid’s viscosity.

7.2.1. Acceleration and Advection

Acceleration is the change in velocity with

time. If the vector velocity is expressed in Cartesian

coordinates as u ¼ (u, v, w) where the bold

u indicates a vector quantity, and u, v, and w are

the positive eastward (x-direction), northward

(y-direction) and positive upward (z-direction)

velocities, then

x-direction acceleration ¼ vu=vt

(7.3a)

with similar expressions for the y- and

z-directions.

Advection is defined in Section 5.1.3. Advection

is how the flow moves properties

(including scalars such as temperature or

salinity) and vectors (such as the velocity).

Advection can change the flow property if there

is a gradient in the property through which the

fluid moves. In the x-momentum equation, the

advection term is

x-direction advection

¼ u vu=vx þ v vu=vy þ w vu=vz

(7.3b)

The substantial derivative is the sum of the acceleration

and advection terms:

Du=Dt ¼ vu=vt þ u vu=vx þ v vu=vy

þ w vu=vz (7.4)

7.2.2. Pressure Gradient Force and

Gravitational Force

Pressure is defined in Section 3.2. The flow of

fluid due to spatial variations in pressure is also

described. In mathematical form, the pressure

gradient force is

x-direction pressure gradient force

¼ vp=vx (7.5)

The gravitational force between Earth and the

object or fluid parcel is directed toward the

center of mass of Earth. Gravitational force is

MOMENTUM BALANCE 189

(a)

Acceleration

(b)

Advection V T

x

x 2 x 3 x 4

t 1 t 2 t 3

x 1 x 2 x 3

v 1

a

v 2

Time

Position

Velocity

Acceleration

x 1

Time t 1

T = 2° 3° 4° 5°

Time t 2

2° 3° 4° 5°

(c)

Pressure gradient force

(d)

Gravitational force – g

High

pressure

pressure gradient

Low

pressure

– g

x A

P A

dp P B – P

=~ A

dx x B – x A

x B

P B

pressure gradient

force

(e)

Acceleration associated with friction and viscosity

z

Moving plate, speed u = u o



fluid

velocity u(z)

x-momentum flux

= u/ z

Fixed plate, speed u = 0

time: just after top plate starts

High flux divergence

High acceleration

x


time: later

Lower flux divergence

Lower acceleration


time: -->

No flux divergence

No acceleration

FIGURE 7.1 Forces and accelerations in a fluid: (a) acceleration, (b) advection, (c) pressure gradient force, (d) gravity, and

(e) acceleration associated with viscosity y.

mass of the object gravitational acceleration g,

equal to 9.780318 m 2 /sec (at the equator). The

gravitational force per unit volume is

z-direction gravitational force per unit

volume ¼ rg (7.6)

7.2.3. Rotation: Centrifugal and

Coriolis Forces

Centrifugal force is the apparent outward force

on a mass when it is rotated. Since Earth rotates

around a fixed axis, the direction of centrifugal

force is always outward away from the axis,

190


opposite to the direction of gravity at the

equator; at Earth’s poles it is zero. (Centripetal

force is the necessary inward force that keeps

the mass from moving in a straight line; it is

the same size as centrifugal force, with the

opposite sign. Centripetal force is real; centrifugal

force is just an apparent force.) The mathematical

expression for centrifugal acceleration

(force divided by density) is

centrifugal acceleration ¼ U 2 r (7.7)

where U istherotationrateofEarth,equalto

2p/T where T is the length of day, and r is

Earth’s radius. Because the centrifugal acceleration

is nearly constant in time and points

outward, away from Earth’s axis of rotation,

we usually combine it formally with the gravitational

force, which points toward Earth’s

center. We replace g in Eq. (7.6) with an effective

gravity g, which has a weak dependence

on latitude. Hereafter, we do not refer separately

to the centrifugal force. The surface

perpendicular to this combined force is called

the geoid. If the ocean were not moving relative

to Earth, its surface would align with the

geoid.

The second term in a rotating frame of reference

included in the acceleration equation (7.1)

is the Coriolis force. When a water parcel, air

parcel, bullet, hockey puck, or any other body

that has little friction moves, Earth spins out

from under it. By Newton’s Law, the body

moves in a straight line if there is no other force

acting on it. As observers attached to Earth, we

see the body appear to move relative to our

location. In the Northern Hemisphere, the

Coriolis force causes a moving body to appear

to move to the right of its direction of motion

(Figure S7.2b). In the Southern Hemisphere, it

moves to the left.

The Coriolis force is non-zero only if the body

is in motion, and is important only if the body

travels for a significant period of time. Coriolis

force is larger for larger velocities as well.

Mathematically, the Coriolis force is

x-momentum equation : 2Usin4 vh fv

(7.8a)

y-momentum equation : 2Usin4 uhfu

Coriolis parameter : fh2Usin4

(7.8b)

(7.8c)

where “h” denotes a definition, U is the rotation

rate, 4 is latitude, u is velocity in the x-

direction, v is velocity in the y-direction, and

where the signs are appropriate for including

these terms on the left-hand side of Eq.

(7.1). The Coriolis parameter, f, is a function

of latitude and changes sign at the equator,

and it has units of sec 1 . (The nondimensional

parameter called the Rossby

number introduced in Section 1.2 is Ro ¼ 1/

fT or Ro ¼ U/fL, where U, L, and T are characteristic

velocity, length, and timescales for

the flow.)

7.2.4. Viscous Force or Dissipation

Fluids have viscous molecular processes that

smooth out variations in velocity and slow

down the overall flow. These molecular

processes are very weak, so fluids can often be

treated, theoretically, as “inviscid” rather than

viscous. However, it is observed that turbulent

fluids like the ocean and atmosphere actually

act as if the effective viscosity were much larger

than the molecular viscosity. Eddy viscosity is

introduced to account for this more efficient

mixing (Section 7.2.4.2).

7.2.4.1. Molecular Viscosity

We can think of molecular viscosity by

considering two very different types of coexisting

motion: the flow field of the fluid, and,

due to their thermal energy, the random motion

of molecules within the flow field. The random

molecular motion carries (or advects) the larger

scale velocity from one location to another, and

then collisions with other molecules transfer

their momentum to each other; this smoothes

MOMENTUM BALANCE 191

out the larger-scale velocity structure

(Figure S7.3).

The viscous stress within a Newtonian fluid

is proportional to the velocity shear. The proportionality

constant is the dynamic viscosity,

which has meter-kilogram-second (mks) units

of kg/m-sec. The dynamic viscosity is the

product of fluid density times a quantity called

the kinematic viscosity, which has mks units of

m 2 /sec. For water, the kinematic viscosity is

1.8 10 6 m 2 /sec at 0 Cand1.0 10 6 m 2 /

sec at 20 C (Table S7.1). Flow is accelerated

or decelerated if there is a variation in viscous

stress from one location to another

(Figure 7.1e).

Formally, for a Newtonian fluid, which is

defined to be a fluid in which stress is proportional

to strain (velocity shear), and if viscosity

has no spatial dependence, viscous stress enters

the momentum equations as

x-momentum dissipation

¼ yðv 2 u=vx 2 þ v 2 u=vy 2 þ v 2 u=vz 2 Þ (7.9)

where y is the molecular (kinematic) viscosity.

(The dynamic viscosity is ry.) Molecular

viscosity changes flow very slowly. Its effectiveness

can be gauged by a non-dimensional

parameter, the Reynolds number, which is the

ratio of the dissipation timescale to the advective

timescale: Re ¼ UL/y. When the Reynolds

number is large, the flow is nearly inviscid

and most likely very turbulent; this is the

case for flows governed by molecular viscosity.

When Earth’s rotation and hence the Coriolis

term is important, the non-dimensional parameter

of most interest for judging the effectiveness

of dissipation is the Ekman number:

E ¼ y/fH 2 . Nearly inviscid rotating flows

have very small Ekman number. From matching

observations and theory we know that

the ocean currents dissipate energy much

more quickly than we can predict using molecular

viscosity. How this happens is described

next.

7.2.4.2. Eddy Viscosity

Mixing at spatial scales larger than those

quickly affected by molecular viscosity is generally

a result of turbulence in the fluid. Turbulent

motions stir the fluid, deforming and pulling it

into elongated, narrow filaments. A stirred fluid

mixes much faster than one that is calm and

subjected only to molecular motion. We refer

to the effect of this turbulent stirring/mixing

on the fluid as eddy viscosity. For large-scale

ocean circulation, the “turbulent” motions are

mesoscale eddies, vertical fine structure, and so

on, with spatial scales smaller than the larger

scales of interest. Like molecular viscosity,

eddy viscosity should be proportional to the

product of turbulent speed and path length.

Therefore, horizontal eddy viscosity is generally

much larger than vertical eddy viscosity (Table

S7.1).

To mathematically include eddy viscosity, the

viscous terms in Eqs. (7.1) and (7.9) are replaced

by the eddy viscosity terms:


¼ A H ðv 2 u=vx 2 þ v 2 u=vy 2 ÞþA V ðv 2 u=vz 2 Þ

(7.10)

where A H is the horizontal eddy viscosity and

A V is the vertical eddy viscosity. A H and A V

have units of kinematic viscosity, m 2 /sec in

mks units. (Although we often use these Cartesian

coordinates, the most relevant stirring/

mixing directions are along isopycnals (isentropic

surfaces) and across isopycnals (diapycnal

mixing), so the coordinate system used in Eq.

(7.10) is better modeled by rotating it to have

the “vertical” direction perpendicular to isopycnal

surfaces, and replace A H and A V with eddy

viscosities that are along and perpendicular to

those surfaces.)

Although eddy viscosity is much larger than

molecular viscosity, the ocean is nevertheless

nearly inviscid, in the sense that the Reynolds

number is large and the Ekman number is small

even when eddy viscosities are used.

192


7.2.5. Mathematical Expression of

Momentum Balance

The full momentum balance with spatially

varying eddy viscosity and rotation is:

Du=Dt fv ¼ vu=vt þ u vu=vx

þ v vu=vy þ w vu=vz fv

¼ ð1=rÞvp=vx þ v=vxðA H vu=vxÞ

þ v=vyðA H vu=vyÞþv=vzðA V vu=vzÞ

(7.11a)

Dv=Dt þ fu ¼ vv=vt þ u vv=vx þ v vv=vy

þ w vv=vz þ fu ¼ ð1=rÞvp=vy

þ v=vxðA H vv=vxÞþv=vyðA H vv=vyÞ

þ v=vzðA V vv=vzÞ

(7.11b)

Dw=Dt ¼ vw=vt þ u vw=vx þ v vw=vy

þ w vw=vz ¼ ð1=rÞvp=vz g

þ v=vxðA H vw=vxÞþv=vyðA H vw=vyÞ

þ v=vzðA V vw=vzÞ

(7.11c)

Here the standard notation “D/Dt” is the

substantial derivative defined in Eq. (7.4).

The full set of equations describing the

physical state of the ocean must also include

the continuity (or mass conservation) equation

(Section 5.1):

Dr=Dt þ rðvu=vx þ vv=vy þ vw=vzÞ ¼0

(7.11d)

If density changes are small, Eq. 7.11d is

approximated as

vu=vx þ vv=vy þ vw=vz ¼ 0

(7.11e)

which is known as the continuity equation.

The set is completed by the equations

governing changes in temperature, salinity,

and density, which are presented in the

following section.

7.3. TEMPERATURE, SALINITY,

AND DENSITY EVOLUTION

Evolution equations for temperature and

salinity d the equation of state that relates

density to salinity, temperature, and pressure,

and thus an evolution equation for density d

complete the set of equations (7.11aed) that

describe fluid flow in the ocean.

7.3.1. Temperature, Salinity, and

Density Equations

Temperature is changed by heating, cooling,

and diffusion. Salinity is changed by addition

or removal of freshwater, which alters the dilution

of the salts. Density is then computed

from temperature and salinity using the equation

of state of seawater. The “word” equations

for temperature, salinity, and density forcing

include:

temperature change

þ temperature advection=convection

¼ heating=cooling term þ diffusion

(7.12a)

salinity change þ salinity advection=convection

¼ evaporation=precipitation=runoff

=brine rejection þ diffusion (7.12b)

equation of stateðdependence of density on

salinity; temperature; and pressureÞ (7.12c)

density change þ density advection=convection

¼ density sources þ diffusion (7.12d)

Written in full, these are

DT=Dt ¼ vT=vt þ u vT=vx þ v vT=vy

þ w vT=vz ¼ Q H =rc p þ v=vxðk H vT=vxÞ

þ v=vyðk H vT=vyÞþv=vzðk V vT=vzÞ

(7.13a)

TEMPERATURE, SALINITY, AND DENSITY EVOLUTION 193

DS=Dt ¼ vS=vt þ u vS=vx þ v vS=vy þ w vS=vz

r ¼ rðS; T; pÞ

¼ Q S þ v=vxðk H vS=vxÞ

þ v=vyðk H vS=vyÞþv=vzðk V vS=vzÞ

Dr=Dt ¼ vr=vt þ u vr=vx þ v vr=vy

þ w vr=vz ¼ðvr=vSÞ DS=Dt

(7.13b)

(7.13c)

þðvr=vTÞ DT=Dt þðvr=vpÞ Dp=Dt

(7.13d)

where Q H is the heat source (positive for heating,

negative for cooling, applied mainly near

the sea surface), c p is the specific heat of

seawater, and Q S is the salinity “source” (positive

for evaporation and brine rejection, negative

for precipitation and runoff, applied at or

near the sea surface). k H and k v are the horizontal

and vertical eddy diffusivities, analogous

to the horizontal and vertical eddy viscosities in

the momentum equations (7.11aed) (Table S7.1

located on the textbook Web site). The full equation

of state appears in Eq. (7.13c), from which

the evolution of density in terms of temperature

and salinity change can be computed (Eq.

7.13d). The coefficients for the three terms in

Eq. (7.13d) are the haline contraction coefficient,

the thermal expansion coefficient, and the

adiabatic compressibility, which is proportional

to the inverse of sound speed (Chapter 3).

7.3.2. Molecular and Eddy Diffusivity

The molecular diffusivity k for each substance

depends on the substance and the fluid. The

molecular diffusivity of salt in seawater is

much smaller than that for heat (Table S7.1).

This difference results in a process called

“double diffusion” (Section 7.4.3). Eddy diffusivity

is the equivalent of eddy viscosity for properties

like heat and salt. A globally averaged vertical

eddy diffusivity of k v ¼ 1 10 4 m 2 /sec

accounts for the observed average vertical

density structure (Section 7.10.2; Munk, 1966).

However, the directly observed vertical (or diapycnal)

eddy diffusivity in most of the ocean is

a factor of 10 lower: k v ~ 1 10 5 m 2 /sec,

implying that there are regions of much higher

diffusivity to reach the global average. Measurements

show huge enhancements of diapycnal

eddy diffusivity in bottom boundary regions,

especially where topography is rough (Figure 7.2)

(Polzin, Toole, Ledwell, & Schmitt, 1997; Kunze

et al., 2006), and on continental shelves where

tidal energy is focused (Lien & Gregg, 2001).

In the surface layer, eddy diffusivities and

eddy viscosities are also much greater than the

Munk value (e.g., Large, McWilliams, & Doney,

1994) (Section S7.4.1). Horizontal eddy diffusivities

k H are estimated to be between 10 3 and

z (m)

1000

2000

4000

5000

Mozambique Plateau

Madagascar Plateau

SW Indian Ridge

6000

0 2000 4000 6000 8000

r (km)

SE Indian Ridge

Ninety-East Ridge

Diamantina FZ

Perth

-3 log(K) m 2 /s 2

-4

-5

GM IW

-6

FIGURE 7.2 Observed diapycnal

diffusivity (m 2 /s 2 )

along 32 S in the Indian Ocean,

which is representative of other

ocean transects of diffusivity.

See Figure S7.4 for diffusivity

profiles. This figure can also be


ÓAmerican Meteorological Society.

Reprinted with permission. Source:

From Kunze et al. (2006).

194


10 4 m 2 /sec, with large spatial variability (e.g.,

Figure 14.17). k H is much larger than k V .

7.4. MIXING LAYERS

Mixing occurs throughout the ocean. While

weak, mixing is essential for maintaining the

observed stratification and can regulate the

strength of some parts of the circulation.

7.4.1. Surface Mixed Layer

The surface layer (Section 4.2.2) is forced

directly by the atmosphere through surface

wind stress and buoyancy (heat and freshwater)

exchange. For a surface layer that is initially

stably stratified (Figure 7.3a), sufficiently large

wind stress will create turbulence that mixes

and creates a substantially uniform density or

mixed layer (Figure 7.3b). This typically results

in a discontinuity in properties at the mixed

layer base.

The upper layer can also be mixed by buoyancy

loss through the sea surface, increasing

the density of the top of the surface layer and

causing it to overturn (convect) to a greater depth

(Figure 7.3cee).

This type of mixed layer typically has no

discontinuity in density at its base. Heat or

freshwater gain decreases the density of the

top of the surface layer, resulting in a more

stably stratified profile. If the wind then mixes

it, the final mixed layer is shallower than the

initial mixed layer (Figure 7.3feh). (Mixed layer

observations typically show much more vertical

structure than might be expected from these

simple ideas.)

The thickest mixed layers occur at the end of

winter (Figure 4.5), after an accumulation of

months of cooling that deepens the mixed layer

and increases its density. For large-scale oceanographic

studies, these end-of-winter mixed

layers set the properties that are subducted

into the ocean interior (Section 7.8.5).

7.4.2. Bottom Mixed Layers

Near the ocean bottom turbulence, and hence

mixing, can be generated by currents or current

shear caused by the interaction with the bottom.

In shallow (e.g., coastal) waters, complete mixing

of the water column occurs if the depth is shallow

enough and the tidal currents are fast enough

(see reviews in Simpson, 1998 and Brink, 2005

and more extended discussion in Section S7.4.2

on the textbook Web site). At longer timescales

on the shelf, a bottom Ekman layer can develop

in which frictional and Coriolis forces balance

(Ekman, 1905 and Section 7.5.3), with the bottom

slope also affecting the layer.

Enhanced turbulence in a bottom boundary

layer can be created by movement of water across

rough topography and by breaking of internal

waves that reflect off the topography and result

in higher eddy diffusivity values (Figure 7.2).

This can create “steppy” vertical profiles near

the bottom some distance from the mixing site

(Figure S7.6a located on the textbook Web site).

Bottom currents due to density differences

can also cause mixing. One example is

a turbidity current down an underlying bottom

slope (Section 2.6). Another example is the overflow

of dense water across a sill, as seen at the

Strait of Gibraltar (Chapter 9). The dense water

flows down the continental slope as a plume,

mixing vigorously with the lighter water

around it (Figure S7.6b). This turbulent process

is called entrainment. Density differences due

to the injection of lighter water into the ocean

also cause mixing and entrainment. An example

is hot hydrothermal water injected at mid-ocean

ridges and hotspots that entrain ambient waters

as the plumes rise.

7.4.3. Internal Mixing Layers

In the interior of the ocean (i.e., away from

boundaries), continuous profiling instruments

have shown that vertical profiles of water properties

d temperature and salinity, and hence

density d are often not smooth (Figure 7.3i)

MIXING LAYERS 195

FIGURE 7.3 Mixed layer

development. (a, b) An

initially stratified layer mixed

by turbulence created by

wind stress; (c, d, e) an initial

mixed layer subjected to heat

loss at the surface which

deepens the mixed layer; (f, g,

h) an initial mixed layer subjected

to heat gain and then to

turbulent mixing presumably

by the wind, resulting in

a thinner mixed layer; (i, j) an

initially stratified profile subjected

to internal mixing,

which creates a stepped

profile. Notation: s is wind

stress, Q is heat (buoyancy).

but “stepped” (Figure 7.3j). Turbulence and/or

double diffusion mix the water column internally

and can create such steps.

Breaking internal waves (Chapter 8) can create

internal mixing (Section 8.4; Rudnick et al., 2003).

Vertical shear from other sources can also result

in turbulence. On the other hand, vertical stratification

stabilizes the mixing. One way to express

this trade-off is through a non-dimensional

parameter called the Richardson number (Ri):

Ri ¼ N 2 =ðvu=vzÞ 2 (7.14)

196


N 2 ¼ g ðvr=vzÞ=r 0 (7.15)

where N is the Brunt-Väisälä frequency

(Section 3.5.6) and the vertical shear of the horizontal

speed is (vu/vz). If the Richardson number

is small, the stratification is weak and the shear is

large, so we expect mixing to be vigorous.

Stirring between two horizontally adjacent

waters with strongly contrasting temperature

and salinity results in interleaving or fine structure,

with layering of one to tens of meters on

both sides of the front. A much smaller scale

of vertical structure, on the order of centimeters

(microstructure), is associated with the actual

mixing at the interfaces between the interleaving

layers.

Heat diffuses about 100 times faster than salt

(Table S7.1). Double diffusion arises from these

differing molecular diffusivities, acting at scales

of centimeters to meters, and can also create

well-mixed internal layers. When warm, salty

water lies above cold, fresh water, the saltier

water becomes denser and tends to sink into

the lower layer and vice versa (Figure S7.7a).

The alternating columns are called salt fingers.

Lateral diffusion occurs between the “fingers”

and produces a uniform layer that may be

meters to tens of meters thick in the ocean.

When cold, fresh water lies above warm, salty

water (Figure S7.7b), the fresher upper layer

becomes warmer and rises within the upper

layer; this is called the diffusive form of double

diffusion. Salt fingering effects are observed in

the ocean where there are strong contrasts in

salinity, for instance, where salty Mediterranean

Water enters the Atlantic (Figure S7.7c). Diffusive

interfaces are observed in high latitude

regions with dichothermal layers (Sections 4.2,

4.3.2 and Figure S7.7d).

7.5. RESPONSE TO WIND FORCING

The wind blows over the sea surface exerting

stress and causing the water to move within the

top 50 m. Initially the wind excites small capillary

waves that propagate in the direction of

the wind. Continued wind-driven momentum

exchange excites a range of surface waves

(Chapter 8). The net effect of this input of atmospheric

momentum is a stress on the ocean (wind

stress). For timescales of about a day and longer,

Earth’s rotation becomes important and the

Coriolis effect enters in, as described in the

following subsections.

7.5.1. Inertial Currents

At timescales of a day or so (after build-up of

surface waves and possibly Langmuir circulation),

the ocean responds to a wind stress impulse

with transient motions known as “inertial

currents.” These are a balance of the Coriolis force

and the time derivatives of the initial horizontal

velocities caused by the wind stress. In the

Northern Hemisphere, the water particles trace

out clockwise circles (Figure S7.8a). In the

Southern Hemisphere, inertial currents are

counterclockwise.

(Mathematically, inertial currents are the

solution of

vu=vt ¼ fv

(7.16a)

vv=vt ¼ fu (7.16b)

which is taken from Eq. (7.11a and b) assuming

that advection, pressure gradient forces, and

dissipation are very small.)

Inertial currents are often observed in surface

drifter trajectories and surface velocity moorings

in the wake of a storm (Figure 7.4). Inertial

periods are often very close to tidal periods, so

separating tidal and inertial effects in time series

is sometimes difficult.

After the wind starts to blow impulsively, the

current will initially oscillate around and then,

after several days, settle frictionally to a steady

flow at an angle to the wind (Figure S7.8b

from Ekman, 1905). This becomes the surface

Ekman velocity (Section 7.5.3).

RESPONSE TO WIND FORCING 197

FIGURE 7.4 Observations of

near-inertial currents. Surface drifter

tracks during and after a storm.


Reprinted with permission. Source: From

d’Asaro et al. (1995). SeeFigureS7.8

for schematics of inertial currents

and Ekman’s (1905) original

hodograph.

7.5.2. Langmuir Circulation

“Langmuir circulation” (LC) is another

transient response to impulsive wind forcing

(Langmuir, 1938). A lengthier discussion with

illustrations is provided in Section S7.5.2 in the

online supplementary material. LCs are visually

evident as numerous long parallel lines or

streaks of flotsam (“windrows”) that are mostly

aligned with the wind (Figure S7.9). The streaks

are formed by the convergence caused by helical

vortices with a typical depth and horizontal

spacing of 4e6 m and 10e50 m, but they can

range up to several hundred meters horizontal

separation and up to two to three times the

mixed layer depth (Figure S7.10). Alternate cells

rotate in opposite directions, causing convergence

and divergence between alternate pairs

of cells. The cells can be many kilometers long.

Langmuir circulations generally occur only for

wind speeds greater than 3 m/sec and appear

within a few tens of minutes of wind onset.

The mechanism for producing Langmuir circulation

is still a matter of study and beyond the

scope of this text. See Smith (2001) and Thorpe

(2004) for further discussions.

7.5.3. Ekman Layers

Wind stress is communicated to the ocean

surface layer through viscous (frictional)

processes that extend several tens of meters into

the ocean. For timescales longer than a day, the

response is strongly affected by Coriolis acceleration.

This wind-driven frictional layer is called the

Ekman layer. The physical processes in an Ekman

layer include only friction (eddy viscosity) and

Coriolis acceleration. Velocity in the Ekman layer

is strongest at the sea surface and decays exponentially

downward, disappearing at a depth of

about 50 m. It coexists with, but is not the same

as, the mixed layer depth or euphotic zone depth.

The two most unusual characteristics of an

Ekman layer (compared with a frictional flow

that is not rotating) are (1) the horizontal

velocity vector spirals with increasing depth

(Figure 7.5) and (2) the net transport integrated

through the Ekman layer is exactly to the right

of the wind in the Northern Hemisphere (left

in the Southern Hemisphere).

ThesurfacewaterinanEkmanlayermoves

at an angle to the wind because of Coriolis

acceleration. If eddy viscosity is independent

198


D E

“Ekman flow” - Water velocities decreasing

and rotating with increasing depth:

Wind

Surface

V o

V 1

V 2

Using a constant eddy viscosity of 0.05 m 2 /sec

from within the observed range (Section 7.5.5),

the Ekman layer depths at latitudes 10, 45, and

80 degrees are 63, 31, and 26 m, respectively.

The vertically integrated horizontal velocity in

the Ekman layer is called the Ekman transport:

Z

U E ¼ u E ðzÞ dz (7.18a)

V 7

V 6

V 9

45°

V 5

V 4

V 3

V E ¼

Z

v E ðzÞ dz

(7.18b)

V 9

V 7

V 6

V 5

W

V 4

Ekman spiral

V 3

Horizontal

Plane

Resultant volume transport

at right angles to wind

FIGURE 7.5 Ekman layer velocities (Northern Hemisphere).

Water velocity as a function of depth (upper

projection) and Ekman spiral (lower projection). The large

open arrow shows the direction of the total Ekman transport,

which is perpendicular to the wind.

of depth, the angle is 45 degrees to the right of

the wind in the Northern Hemisphere; otherwise

it differs somewhat from 45 degrees.

Due to the frictional stress proportional to the

eddy viscosity A V , each layer from the surface

on down accelerates the next layer below to

the right (Northern Hemisphere) and has

a weaker velocity than the layer above it. The

complete structure is a decaying “spiral.” If

the velocity arrows are projected onto a horizontal

plane, their tips form the Ekman spiral

(Figure 7.5).

The Ekman layer depth is the e-folding depth

of the decaying velocity:

V 2

V 1

V o

D E ¼ð2A v =fÞ 1=2 (7.17)

where u E and v E are the eastward and northward

velocities in the Ekman layer, and U E

and V E are the associated Ekman transports.

(Ekman “transport” has units of depth times

velocity, hence m 2 /sec, rather than area times

velocity.) Ekman transport in terms of the

wind stress is derived from Eq. (7.11):

U E ¼ s ðyÞ =ðrfÞ

(7.19a)

V E ¼ s ðxÞ =ðrfÞ (7.19b)

where s (x) and s (y) are the wind stresses positive in

the east and north directions, assuming no time

acceleration, advection, or pressure gradient

force, and setting the eddy friction stress at the

sea surface equal to the wind stress. The Ekman

transport is exactly perpendicular and to the right

(left) of the wind in the Northern (Southern)

Hemisphere (large arrow in Figure 7.5). For applications

of Ekman layers to general circulation

(Sections 7.8 and 7.9), only the Ekman transport

matters. Thus, the actual eddy viscosity and

Ekman layer thickness are unimportant.

Bottom Ekman layers that are 50 to 100 m

thick can develop if there is a flow along the

bottom. In shallow water, the top and bottom

Ekman layers can overlap, so that the rightturning

tendency in the top layer (Northern

Hemisphere) will overlap the left-turning

tendency in the bottom layer. If there is a wind

stress at the top surface that would produce an

Ekman layer of depth D E in deep water, then


in water of depth h, the approximate angle

a between the wind and the surface flow is as

listed in Table S7.2. As water depth decreases,

the net flow is more in the direction of the wind.

7.5.4. Ekman Transport Convergence

and Wind Stress Curl

When the wind stress varies with position so

that Ekman transport varies with position, there

can be a convergence or divergence of water

within the Ekman layer. Convergence results

in downwelling of water out of the Ekman layer.

Divergence results in upwelling into the Ekman

layer. This is the mechanism that connects the

frictional forcing by wind of the surface layer

to the interior, geostrophic ocean circulation

(Section 7.8).

The vertical velocity w E at the base of the

Ekman layer is obtained from the divergence of

the Ekman transport, by vertically integrating

the continuity equation Eq. (7.11e) over the depth

of the Ekman layer:

ðvU E =vx þ vV E =vyÞ ¼V,U E

¼ ðw surface w E Þ¼w E

(7.20)

where U E is the horizontal vector Ekman transport

(Eq. 7.18) and it is assumed that the vertical

velocity at the sea surface, w surface , is 0. When

Eq. (7.20) is negative, the transport is convergent

and there must be downwelling (w E at the base

of the Ekman layer is negative). The relation of

Ekman transport divergence to the wind stress

from Eq. (7.19a,b) is

V,U E ¼ v=vxðs ðyÞ =ðrfÞÞ

v=vyðs ðxÞ =ðrfÞÞ

¼ k,V ðs=rfÞ (7.21)

where s is the vector wind stress and k is the

unit vector in the vertical direction. Therefore,

in the Northern Hemisphere (f > 0), upwelling

into the Ekman layer results from positive

wind stress curl, and downwelling results

from negative wind stress curl. Downwelling

is referred to as Ekman pumping. Upwelling is

sometimes referred to as Ekman suction.

A global map of wind stress curl was shown

in Figure 5.16d, and is referred to frequently in

subsequent chapters because of its importance

for Ekman pumping/suction, although the

mapped quantity should include the Coriolis

parameter, f, to be related directly to upwelling

and downwelling.

Equatorial upwelling due to Ekman transport

results from the westward wind stress (trade

winds). These cause northward Ekman transport

north of the equator and southward Ekman

transport south of the equator. This results in

upwelling along the equator, even though the

wind stress curl is small, more or less because

of the Coriolis parameter dependence in Eq.

(7.21).

The coastline is the other place where Ekman

transport divergence or convergence can occur,

and it is not included in Eq. (7.21) because this

divergence is due to the boundary condition at

the coast and not wind stress curl. If the wind

blows along the coast, then Ekman transport is

perpendicular to the coast, so there must be

either downwelling or upwelling at the coast

to feed the Ekman layer (Figure 7.6). This is

one mechanism for creation of coastal upwelling

and subtropical eastern boundary current

systems (Section 7.9).

7.5.5. Observations of Ekman Response

and Wind Forcing

The Ekman theory has major consequences

for wind-driven ocean circulation. Thus it is

important to confirm and refine Ekman’s theory

with ocean observations. Observations of

Ekman response are difficult because of the

time dependence of the wind. California

Current observations produced an easily visible

Ekman-like response because the wind direction

was relatively steady (Figure 7.7 and

Chereskin, 1995).

200


(a)

Ekman transport (northward)

Northern

Hemisphere

Trade Winds

Equator

Ekman transport (southward)

Southern

Hemisphere

(b)

Sea surface

warm

Ekman

transport

Thermocline

Southern

Hemisphere

(c)

cold

Upwelling

Equator

North

Trade

Winds

warm

Ekman

transport

Northern

Hemisphere

Alongshore wind

Northern

Hemisphere

Upwelling region

Eastern boundary curr.

Isopycnal

Ekman transport

East

Isopycnal

Onshore

transport

Poleward undercurrent

FIGURE 7.6 (a) Ekman transport divergence near the equator driven by easterly Trade Winds. (b) The effect of equatorial

Ekman transport divergence on the surface height, thermocline, and surface temperature. (c) Coastal upwelling system due

to an alongshore wind with offshore Ekman transport (Northern Hemisphere).

An Ekman response to the wind for a large

part of the Pacific Ocean is apparent in the

average 15 m velocity from surface drifters

(Figure 7.8). Velocities are to the right of the

wind stress in the Northern Hemisphere and

to the left in the Southern Hemisphere.

7.6. GEOSTROPHIC BALANCE


Coriolis Force Balance

Throughout most of the ocean at timescales

longer than several days and at spatial scales

GEOSTROPHIC BALANCE 201

North Velocity (cm/s)

8

6

4

2

0

−2

−4

−6

OBSERVATIONS

(slab extrapolation)

Average currents and wind (6 Jun − 4 Oct 1993)

8

12 16

Mean wind

(m/s)

EKMAN THEORY

De = 25 m

2

A = 274.2 cm /s

16

12

−8

−8 −6 −4 −2 0 2 4 6 8

East Velocity (cm/s)

FIGURE 7.7 Observations of an Ekman-like response in

the California Current region. Observed mean velocities (left)

and two theoretical Ekman spirals (offset) using different

eddy diffusivities (274 and 1011 cm 2 /s). The numbers on the

arrows are depths. The large arrow is the mean wind. See

Figure S7.14 for the progressive vector diagram. Source: From

Chereskin (1995).

8

4

0

De = 48 m

2

A = 1011 cm /s

longer than several kilometers, the balance of

forces in the horizontal is between the pressure

gradient and the Coriolis force. This is called

“geostrophic balance” or geostrophy.

In a “word” equation, geostrophic balance is

horizontal Coriolis acceleration

¼ horizontal pressure gradient force (7.22)

This is illustrated in Figure 7.9. The pressure

gradient force vector points from high pressure

to low pressure. In a non-rotating flow, the

water would then move from high to low pressure.

However, with rotation, the Coriolis force

exactly opposes the pressure gradient force, so

that the net force is zero. Thus, the water parcel

does not accelerate (relative to Earth). The parcel

moves exactly perpendicular to both the pressure

gradient force and the Coriolis force. A

heuristic way to remember the direction of

geostrophic flow is to think of the pressure

gradient force pushing the water parcel from

high to low pressure, but Coriolis force moves

the parcel off to the right (Northern Hemisphere)

or the left (Southern Hemisphere).

FIGURE 7.8 Ekman response. Average wind vectors (blue) and average ageostrophic current at 15 m depth (red). The

current is calculated from 7 years of surface drifters drogued at 15 m, with the geostrophic current based on average density

data from Levitus et al. (1994a) removed. (No arrows were plotted within 5 degrees of the equator because the Coriolis force is

small there.) This figure can also be found in the color insert. ÓAmerican Meteorological Society. Reprinted with permission. Source:

From Ralph and Niiler (1999).

202


v (in)

Low

pressure

CF

PGF

z

x

High

pressure

The vertical force balance that goes with

geostrophy is hydrostatic balance (Section 3.2).

The vertical pressure gradient force, which

points upward from high pressure to low pressure,

is balanced by gravity, which points

downward.

The mathematical expression of geostrophy

and hydrostatic balance, from Eq. (7.11), is

fv ¼ ð1=rÞvp=vx (7.23a)

fu ¼ ð1=rÞvp=vy (7.23b)

0 ¼ vp=vz rg (7.23c)

An alternate form for Eq. (7.23c), used for

dynamic height calculations (Section 7.6.3), is

0 ¼ avp=vz g (7.23d)

where a is specific volume. From Eq. (7.23a and

b), if the Coriolis parameter is approximately

constant (f ¼ f o ), the geostrophic velocities are

approximately non-divergent:

vu=vx þ vv=vy ¼ 0

PGF

Low

pressure

(7.23e)

Formally in fluid dynamics, such a non-divergent

velocity field can be written in terms of

a streamfunction j:

CF

v (out)

FIGURE 7.9 Geostrophic balance: horizontal forces and

velocity. PGF ¼ pressure gradient force. CF ¼ Coriolis

force. v ¼ velocity (into and out of page). See also

Figure S7.17.

u ¼ vj=vy and v ¼ vj=vx (7.23f)

From Eqs. (7.23a and b) the streamfunction for

geostrophic flow is j ¼ p/(f o r o ). Therefore,

maps of pressure distribution (or its proxies

like dynamic height, steric height, or geopotential

anomaly; Section 7.6.2), are maps of the

geostrophic streamfunction, and flow approximately

follows the mapped contours.

Geostrophic balance is intuitively familiar to

those with a general interest in weather reports.

Weather maps show high and low pressure

regions around which the winds blow. Low

pressure regions in the atmosphere are called

cyclones. Flow around low pressure regions is

thus called cyclonic (counterclockwise in the

Northern Hemisphere and clockwise in the

Southern Hemisphere). Flow around highpressure

regions is called anticyclonic.

In the ocean, higher pressure is caused by

a higher mass of water lying above the observation

depth. At the “sea surface,” pressure differences

are due to an actual mounding of water

relative to the geoid. Over the complete width

of the Atlantic or Pacific Ocean anticyclonic

gyres, the total contrast in sea surface height is

about 1 m.

The geostrophic velocities at the sea surface

could be calculated if the appropriately timeaveraged

sea surface height were known (as

yet not possible for the time mean, but definitely

possible from satellite altimetry for variations

from the mean). The geostrophic velocity at

the sea surface in terms of sea surface height h

above a level surface is derived from Eqs.

(7.23a and b):

fv ¼ gvh=vx (7.24a)

fu ¼ gvh=vy (7.24b)

To calculate the horizontal pressure difference

below the sea surface, we have to consider

both the total height of the pile of water above

our observation depth and also its density, since

the total mass determines the actual pressure at

our observation depth (Figure 7.10 in this


A

z

x

v

v

v

p 1

PGF

r 1

p 2

r 2

p 3

r 3

FIGURE 7.10 Geostrophic flow and thermal wind

balance: schematic of change in pressure gradient force

(PGF) with depth. The horizontal geostrophic velocity v is

into the page for this direction of PGF and is strongest at the

top, weakening with depth, as indicated by the circle sizes.

Density (dash-dot) increases with depth, and isopycnals are

tilted. With the sea surface at B higher than at A, the PGF at

the sea surface (h 1 ) is to the left. The PGF decreases with

increasing depth, as indicated by the flattening of the

isobars p 2 and p 3 .

chapter and Figure S7.19 on the textbook Web

site). The variation in geostrophic flow with

depth (the geostrophic velocity shear) is therefore

proportional to the difference in density of the

two water columns on either side of our observation

location. The relation between the

geostrophic velocity shear and the horizontal

change (gradient) in density is called the thermal

wind relation.

The thermal wind relation is illustrated in

Figure 7.10. The sea surface is sloped, with

surface pressure higher to the right. This creates

a pressure gradient force to the left, which

drives a surface geostrophic current into the

page (Northern Hemisphere). The density r

increases with depth, and the isopycnals are

tilted. Therefore the geostrophic velocity

changes with depth because the pressure

gradient force changes with depth due to the

tilted isopycnals.

B

A useful rule of thumb for geostrophic flows

that are surface-intensified is that, when facing

downstream in the Northern Hemisphere,

the “light/warm” water is to your right. (In

the Southern Hemisphere, the light water is

to the left when facing downstream.) It can be

useful to memorize the example for the Gulf

Stream recalling that the current flows eastward

with warm water to the south. Geostrophic flow

with vertical shear, which requires sloping isopycnals,

is often called baroclinic. Geostrophic

flow without any vertical shear is often called

barotropic. Barotropic flow is driven only by

horizontal variations in sea surface height.

Most oceanic geostrophic flows have both barotropic

and baroclinic components.

Mathematically, the thermal wind relations

are derived from the geostrophic and hydrostatic

balance Eq. (7.23):

fvv=vz ¼ðg=r 0 Þvr=vx

fvu=vz ¼ðg=r 0 Þvr=vy

(7.25a)

(7.25b)

(Here we have used the Boussinesq approximation,

where r can be replaced by the constant

r o in the x and y momentum equations, whereas

the fully variable density r must be used in the

hydrostatic balance equation.)

To calculate geostrophic velocity, we must

know the absolute horizontal pressure difference

between two locations. If we have only

the density distribution, we can calculate only

the geostrophic shear. To convert these relative

currents into absolute currents, we must determine

or estimate the absolute current or pressure

gradient at some level (reference level). A

common, but usually inaccurate, referencing

approach has been to assume (without

measuring) that the absolute current is zero at

some depth (level of no motion). In the next

subsection, we introduce the “dynamic”

method widely used to calculate geostrophic

velocities (shear), and continue the discussion

of reference velocity choices.

204


7.6.2. Geopotential and Dynamic

Height Anomalies and Reference Level

Velocities

Historically and continuing to the present, it

has been too difficult and too expensive to instrument

the ocean to directly observe velocity

everywhere at every depth. Density profiles,

which are much more widely and cheaply

collected, are an excellent data set for estimating

geostrophic velocities using the thermal wind

relations and estimates of a reference level

velocity. To calculate geostrophic velocity shear

from density profiles, oceanographers have

created two closely related functions, geopotential

anomaly and dynamic height, whose horizontal

gradients represent the horizontal pressure

gradient force. Another closely related concept,

steric height, is used to study variations in sea

level. The gradient of the geopotential, F, isin

the direction of the local force due to gravity

(modified to include centrifugal force). Geopotential

is defined from hydrostatic balance

(Eq. 7.23c) as

dF ¼ gdz ¼ a dp (7.26a)

where a is specific volume. The units of geopotential

are m 2 /sec 2 or J/kg. For two isobaric

surfaces p 2 (upper) and p 1 (lower), the geopotential

is

Z

Z

F ¼ g dz ¼ gðz 2 z 1 Þ¼ adp (7.26b)

Geopotential height is defined as

Z

Z ¼ð9:8 ms 2 Þ 1 gdz

Z

¼ ð9:8 ms 2 Þ 1 a dp

(7.26c)

and is nearly equal to geometric height. This

equation is in mks units; if centimetergram-second

(cgs) units are used instead, the

multiplicative constant would change from 9.8

ms 2 to 980 cm s 2 . Most practical calculations,

including common seawater computer subroutines,

use the specific volume anomaly

d ¼ aðS; T; pÞ að35; 0; pÞ (7.26d)

to compute the geopotential anomaly

Z

DF ¼ d dp:

(7.26e)

The geopotential height anomaly is then defined

as

Z

Z 0 ¼ ð9:8 ms 2 Þ 1 d dp: (7.26f)

Geopotential height anomaly is effectively identical

to steric height anomaly, which is defined by

Gill and Niiler (1973) as

Z

h 0 ¼ ð1=r o Þ r dz (7.27a)

in which the density anomaly r’ ¼ r r o . Using

hydrostatic balance and defining r o as r(35,0,p),

Eq. (7.27a) is equivalent to Tomczak and

Godfrey’s (1994) steric height (anomaly)

Z

h 0 ¼ dr o dz (7.27b)

which can be further manipulated to yield

Z

h 0 ¼ð1=gÞ d dp: (7.27c)

This is nearly identical to the geopotential

height anomaly in Eq. (7.26f), differing only in

the appearance of a standard quantity for g. In

SI units, steric height is in meters.

Dynamic height, D, is closely related to geopotential,

F, differing only in sign and units of

reporting. Many modern publications and

common computer subroutines do not distinguish

between dynamic height and geopotential

anomaly. The unit traditionally used for

dynamic height is the dynamic meter:

1 dyn m ¼ 10 m 2 =sec 2 : (7.28a)

Therefore dynamic height reported in dynamic

meters is related to geopotential anomaly as

DD ¼

Z

DF=10 ¼ d dp=10:

(7.28b)

Its relation to the geopotential height and steric

height anomalies is

10 DD ¼ 9:8 Z 0 ¼ gh 0 : (7.28c)

The quantities DD and Z 0 are often used

interchangeably, differing only by 2%. With

use of the dynamic meter, maps of dynamic

topography are close to the actual geometric

height of an isobaric surface relative to a level

surface; a horizontal variation of, say, 1 dyn m,

means that the isobaric surface has a horizontal

depth variation of about 1 m. Note that the geopotential

height anomaly more closely reflects

the actual height variation, so a variation of 1

dyn m would be an actual height variation

closer to 1.02 m.

Geostrophic velocities at one depth relative to

those at another depth are calculated using Eq.

(7.25) with geopotential anomalies, steric height

anomalies, or dynamic heights. In SI units, and

using dynamic meters for dynamic height, the

difference between the northward velocity v

and eastward velocity u at the pressure surface

p 2 relative to the pressure surface p 1 is

fðv 2 v 1 Þ¼10 vDD=vx ¼ vDF=vx

¼ gvh 0 =vx

fðu 2 u 1 Þ¼ 10 vDD=vy ¼ vDF=vy

(7.29a)

¼ gvh 0 =vy (7.29b)

where the dynamic height or geopotential

anomalies are integrated vertically from p 1 to

p 2 . The surface p 1 is the reference level. (Comparison

of Eq. 7.29 with Eq. 7.23 shows that the

dynamic height and geopotential anomalies

are streamfunctions for the difference between

geostrophic flows from one depth to another.)

How is the velocity at the reference level

chosen? Since the strength of ocean currents


decreases from the surface downward in

many (but not all) regions, for practical

reasons, a deep level of no motion has often

been presumed. A much better alternative is

to use a “level of known motion” based on

direct velocity observations. Satellite altimetry

by itself is insufficient for the ocean’s mean

flow field because the spatial variations of

Earth’s geoid are vastly larger than the

ocean’s sea-surface height variations; the

GRavity and Earth Climate Experiment

(GRACE) is helping to resolve this geoid

problem. Modern practice requires that the

flow field that is defined by many density

profiles must satisfy overall constraints such

as mass conservation. The constraints then

help narrow the choices of reference level

velocities, which can be done formally (see

Wunsch, 1996). Ocean state estimation (data

assimilation), which merges observations

with an ocean model, is currently the focus

of most activity for construction of velocity

fields from density profiles.

As an example of the geostrophic method, we

calculate dynamic height and velocity profiles

from density profiles across the Gulf Stream

(Figure 7.11 in this chapter and Table S7.3 on

the textbook Web site). The isopycnals sloping

upward toward the north between 38 and

39 N mark the Gulf Stream (Figure 7.11a). The

geostrophic velocity profile is calculated

between stations “A” and “B” relative to an

arbitrary level of no motion at 3000 m. Station

A has lower specific volume (higher potential

density) than station B (Figure 7.11b). The

surface dynamic height at A is therefore lower

than at B and the surface pressure gradient force

is toward the north, from B to A. Therefore, the

geostrophic velocity at the midpoint between

the stations (Figure 7.11d) is eastward and is

largest at the sea surface. This means that the

sea surface must tilt downward from B to A.

The vertical shear is largest in the upper 800 m

where the difference in dynamic heights is

largest.

206


(a)

0

Stations

26.0

B

A

(b)

A

(c)

A

(d)

500

26.5

B

B

27.0

1000

27.5

27.7

1500

Depth (meters)

2000

2500

3000

27.8

3500

27.88

4000

4500

5000

Potential

density

(kg/m 3 )

38°N 39°N 40°N

100 200 300

Distance (km)

0 200 400 0 2 0 50 100

Specific volume anomaly

(x 10 –8 m 3 /kg)

Dynamic height

(dyn m)

Geostrophic velocity

(cm/sec)

FIGURE 7.11 Geostrophic flow using observations. (a) Potential density section across the Gulf Stream (66 W in 1997).

(b) Specific volume anomaly d (10 8 m 3 /kg) at stations A and B. (c) Dynamic height (dyn m) profiles at stations A and B,

integrated from 3000 m depth. (d) Eastward geostrophic velocity (cm/sec), assuming zero velocity at 3000 m. This figure is

described in detail in Section S7.6.2 of the online supplement.

7.6.3. Dynamic Topography and Sea

Surface Height Maps

Dynamic height at one surface relative to

another is the streamfunction for the geostrophic

flow at that surface relative to the other, as an

extension of Eq. (7.23f). Flows are along the

contours with the high “hills” to the right of

the flow in the Northern Hemisphere (to the

left in the Southern Hemisphere). The speed at

any point is proportional to the steepness of the

slope at that point, in other words, inversely

proportional to the separation of the contours.

Dynamic topography maps (equivalently,

steric height) are shown in Chapter 14 and

throughout the ocean basin chapters (9e13)

to depict the geostrophic flow field. At the

sea surface, all five ocean basins have highest

dynamic topography in the west in the

subtropics. The anticyclonic flows around

these highs are called the subtropical gyres.

The Northern Hemisphere oceans have low

dynamic topography around 50 to 60 N; the

cyclonic flows around these lows are the

subpolar gyres. Tightly spaced contours along

VORTICITY, POTENTIAL VORTICITY, ROSSBY AND KELVIN WAVES, AND INSTABILITIES 207

the western boundaries indicate the swift

western boundary currents for each of the

gyres. Low values are found all the way

around Antarctica; the band of tightly spaced

contours to its north marks the eastward

Antarctic Circumpolar Current. The contrast

in dynamic height and sea-surface height

from high to low in a given gyre is about 0.5

to 1 dynamic meters.

7.6.4. A Two-Layer Ocean

It is frequently convenient to think of the

ocean as composed of two layers in the vertical,

with upper layer of density r 1 and lower layer of

density r 2 (Figure S7.21). The lower layer is

assumed to be infinitely deep. The upper layer

thickness is h þ H, where h is the varying height

of the layer above the sea level surface and H is

the varying depth of the bottom of the layer. We

sample the layers with stations at “A” and “B.”

Using the hydrostatic equation (7.23c), we

compute the pressure at a depth Z at the

stations:

p A

¼ r 1 gðh þ HÞþr 2 gðZ HÞ (7.30a)

p B

¼ r 1 gðh B þ H B Þþr 2 gðZ H B Þ (7.30b)

Here Z represents a common depth for both

stations, taken well below the interface. If we

assume that p A ¼ p B , which amounts to

assuming a “level of no motion” at Z, we can

compute a surface slope, which we cannot

measure in terms of the observed density interface

slope:

h A

Dx

h B

¼ r 2 r 1 H A H B

r 1 Dx

(7.31a)

We then use Eq. (7.30a) to estimate the surface

velocity v:

fv ¼ g h A

Dx

h B

¼ g r 2 r 1 H A H B

r 1 Dx

(7.31b)

7.7. VORTICITY, POTENTIAL

VORTICITY, ROSSBY AND KELVIN

WAVES, AND INSTABILITIES

An apparent “problem” with the geostrophic

balance (Eq. 7.23a,eb) is that it does not include

any of the external forces that make the ocean

flow; it has only pressure gradient and Coriolis

force. How do we insert external forces such as

the wind? In formal geophysical fluid dynamics,

we would show that these forces are in the

momentum equations, but are so weak that we

safely consider the flows to be geostrophic (to

lowest order). To reinsert the external forces,

we have to consider the “vorticity” equation,

which is formally derived from the momentum

equations by combining the equations in a way

that eliminates the pressure gradient force terms.

(It is straightforward to do.) The resulting equation

gives the time change of the vorticity, rather

than the velocities. It also includes dissipation,

variation in Coriolis parameter with latitude,

and vertical velocities, which can be set externally

by Ekman pumping.

The text that follows in this section is a greatly

truncated version of the full text found at the

textbook Web site (Section S7.7), which includes

numerous figures and examples. For a more

thorough treatment, it is recommended that

the full text be used.

7.7.1. Vorticity

Vorticity is twice the angular velocity at a point

in a fluid. It is easiest to visualize by thinking of

a small paddle wheel immersed in the fluid

(Figure 7.12). If the fluid flow turns the paddle

wheel, then it has vorticity. Vorticity is a vector,

and points out of the plane in which the fluid

turns. The sign of the vorticity is given by the

“right-hand” rule. If you curl the fingers on

your right hand in the direction of the turning

paddle wheel and your thumb points upward,

then the vorticity is positive. If your thumb

points downward, the vorticity is negative.

208


(a)

Right-hand rule, thumb up:

North positive vorticity

(b)

North

Paddlewheel circulation


Up

Current

Up

Current

East

East

Right-hand rule, thumb down:

negative vorticity

FIGURE 7.12 Vorticity. (a) Positive and (b) negative vorticity. The (right) hand shows the direction of the vorticity by the

direction of the thumb (upward for positive, downward for negative).

The vorticity vector u is the curl of the

velocity vector v:

u ¼ V v ¼ iðvv=vz

vw=vyÞ

þ jðvw=vx vu=vzÞþkðvv=vx vu=vyÞ

(7.32)

Vorticity, therefore, has units of inverse time,

(sec) 1 .

Fluids (and all objects) have vorticity

simply because of Earth’s rotation. This is

called planetary vorticity. The vector planetary

vorticity points upward, parallel to the rotation

axis of Earth, which has an angular rotation

rate of U:

u planetary ¼ 2U (7.33)

where U ¼ 2p/day ¼ 2p/86160 sec ¼ 7.293

10 5 sec 1 ,sou planetary ¼ 1.4586 10 4 sec 1 .

The vorticity of the fluid motion relative to

Earth’s surface (Eq. 7.32) is called the relative

vorticity. The total or absolute vorticity of a piece

of fluid is the sum of the relative vorticity and

planetary vorticity.

For large-scale oceanography, only the local

vertical component of the total vorticity is

used because the fluid layers are thin compared

with Earth’s radius, so flows are nearly horizontal.

The local vertical component of the

planetary vorticity is exactly equal to the

Coriolis parameter f (Eq. 7.8c) and is therefore

maximum and positive at the North Pole (4 ¼

90 N), maximum and negative at the South

Pole (4 ¼ 90 S), and 0 at the equator.

The local vertical component of the relative

vorticity from Eq. (7.32) is

2 ¼ðvv=vx vu=vyÞ ¼curl z v (7.34)

where v is the horizontal velocity vector. The

local vertical component of the absolute

vorticity is therefore (z þ f). The geostrophic

velocities calculated from Eq. (7.23) (Section 7.6)

are often used to calculate relative vorticity.

7.7.2. Potential Vorticity

Potential vorticity is a dynamically important

quantity related to relative and planetary

vorticity. Conservation of potential vorticity is

one of the most important concepts in geophysical

fluid dynamics, just as conservation of

angular momentum is a central concept in solid

body mechanics. Potential vorticity takes into

account the height of a water column as well

as its local spin (vorticity). If a column is shortened

and flattened (preserving mass), then it

must spin more slowly. On the other hand, if

a column is stretched and thinned (preserving


mass), it should spin more quickly similar to

a spinning ice skater or diver who spreads his

or her arms out and spins more slowly (due to

conservation of angular momentum). Potential

vorticity, when considering only the local

vertical components, is

Q ¼ðz þ fÞ=H (7.35)

where H is the depth, if the fluid is unstratified.

When the fluid is stratified, the equivalent

version of potential vorticity is

Q ¼ ðz þ fÞð1=rÞðvr=vzÞ: (7.36)

When there are no forces (other than gravity)

on the fluid and no buoyancy sources that

can change density, potential vorticity Q is

conserved:

DQ=Dt ¼ 0 (7.37)

where “D/Dt” is the substantial derivative

(Eq. 7.4). Figures S7.24eS7.26 and text describing

the trade-offs between the relative, planetary, and

stretching vorticity are found on the textbook Web

site. All that we note here is that f varies with latitude,

with huge consequences for ocean currents

and stratification. Therefore, a special symbol is

introduced to denote the change in Coriolis

parameter with northward distance y,orinterms

of latitude f and Earth’s radius R e :

b ¼ df=dy ¼ 2U cos f=R e (7.38)

We often refer to the “b-effect” when talking

about how changes in latitude affect currents,

or the very large-scale, mainly horizontal

Rossby waves for which the b-effect is the

restoring force, described next.

7.7.3. Rossby Waves

The adjustment of any fluid to a change in

forcing takes the form of waves that move out

and leave behind a steady flow associated

with the new forcing. We describe some general

properties of waves in Chapter 8. The largescale,

almost geostrophic circulation adjusts to

changing winds and buoyancy forcing mainly

through “planetary” or Rossby waves and Kelvin

waves (Section 7.7.6). Pure Rossby (and Kelvin)

waves are never found except in simplified

models and lab experiments. However, much

of the ocean’s variability can be understood in

terms of Rossby wave properties, particularly

the tendency for westward propagation relative

to the mean flow.

Most of the physical motivation for these

waves, including illustrations (Figures

S7.27eS7.29), are in the full online version

located at the textbook Web site. Only the most

basic facts are in the following list.

1. Rossby waves have wavelengths of tens to

thousands of kilometers. Therefore particle

motions in Rossby waves are almost

completely transverse (horizontal, parallel

to Earth’s surface).

2. The restoring force for Rossby waves is the

variation in Coriolis parameter f with

latitude, so all dispersion information

includes b (Eq. 7.38). As a water column is

shoved off to a new latitude, its potential

vorticity must be conserved (Eq. 7.37).

Therefore, the water column height (long

Rossby waves) or relative vorticity (short

Rossby waves) begins to change. As with

all waves, the column overshoots, and

then has to be restored again, creating the

wave.

3. All Rossby wave crests and troughs move

only westward (relative to any mean flow,

which could advect them to the east) in both

the Northern and Southern Hemispheres;

that is, the phase velocity is westward (plus

a northward or southward component).

4. The group velocity of Rossby waves is

westward for long wavelengths (more

than about 50 km) and eastward for short

wavelengths (even though the phase velocity

is westward).

210


5. Velocities in Rossby waves are almost

geostrophic. Therefore, they can be calculated

from variations in pressure, for instance, as

measured by a satellite altimeter, which

observes the sea-surface height (e.g., Rossby

wave-like behavior in Figures 14.18 and

14.19).

7.7.4. Rossby Deformation Radius and

Rossby Wave Dispersion Relation

The length scale that separates long from

short Rossby waves is called the Rossby deformation

radius. It is the intrinsic horizontal length

scale for geostrophic or nearly geostrophic flows,

relative to which all length scales are compared.

Again, see the full online version at http://

booksite.academicpress.com/DPO/ Chapter S7

for more information.

The Rossby deformation radius in an unstratified

ocean is

R E ¼ðgHÞ 1=2 =f

(7.39a)

where H is the ocean depth scale. R E is

called the barotropic Rossby deformation radius or

“external” deformation radius. Barotropic deformation

radii are on the order of thousands of

kilometers.

The Rossby deformation radius associated

with the ocean’s stratification is

R I ¼ NH s =f

(7.39b)

where N is the Brunt-Väisälä frequency (7.15),

and H s is an intrinsic scale height for the flow.

R I is called the baroclinic deformation radius (or

“internal” deformation radius). The vertical

length scale H s associated with the first baroclinic

mode is about 1000 m, which is the typical

pycnocline depth. R I for the first baroclinic

mode varies from more than 200 km in the

tropics to around 10 km at high latitudes

(Figure S7.28a) (Chelton et al., 1998).

The dispersion relation (Section 8.2) for first

mode baroclinic Rossby waves is

u ¼

bk

k 2 þ l 2 þð1=R I Þ 2 (7.40)

where u is the wave frequency, k and l are the

wavenumbers in the east-west (x) and northsouth

(y) directions, b is as in Eq. (7.38), and R I

is as given in Eq. (7.39b). Highest frequency

(shortest period) occurs at the wavelength associated

with the Rossby deformation radius

(Figure S7.29). The shortest periods vary from

less than 50 days in the tropics to more than

2 to 3 years at high latitudes (Figure S7.28b

from Wunsch, 2009). Poleward of about 40 to

45 degrees latitude there is no first baroclinic

mode at the annual cycle, so seasonal atmospheric

forcing cannot force the first baroclinic

mode at these higher latitudes.

7.7.5. Instability of Geostrophic Ocean

Currents

Almost all water flows are unsteady. When

gyre-scale flows break up, they do so into large

eddies, on the order of tens to hundreds of

kilometers in diameter or larger (see Section

14.5). The size of the eddies is often on the

order of the Rossby deformation radius. The

eddies usually move westward, like Rossby

waves.

Instabilities of flows are often studied by

considering a mean flow and then finding the

small perturbations that can grow exponentially.

This approach is called “linear stability

theory”; it is linear because the perturbation is

always assumed to be small relative to the

mean flow, which hardly changes at all. When

perturbations are allowed to grow to maturity,

when they might be interacting with each other

and affecting the mean flow, the study has

become nonlinear.

We define three states: stable, neutrally stable,

and unstable. A stable flow returns to its original

state after it is perturbed. A neutrally stable flow

remains as is. In an unstable flow, the perturbation

grows.

WIND-DRIVEN CIRCULATION: SVERDRUP BALANCE AND WESTERN BOUNDARY CURRENTS 211

The two sources of energy for instabilities are

the kinetic energy and the potential energy of the

mean flow. Recall from basic physics that kinetic

energy is ½ mv 2 where m is mass and v is speed;

for a fluid we replace the mass with density r,or

just look at the quantity ½ v 2 . Also recall from

basic physics that potential energy comes from

raising an object to a height; the work done in

raising the object gives it its potential energy. In

a stratified fluid like the ocean, there is no available

potential energy if isopycnals are flat, which

means that nothing has been moved and nothing

can be released. For there to be usable or available

potential energy, isopycnals must be tilted.

Barotropic instabilities feed on the kinetic

energy in the horizontal shear of the flow. Baroclinic

instabilities draw on the available potential

energy of the flow. Baroclinic instability is peculiar

to geostrophic flows, because Earth’s rotation

makes it possible to have a mean

geostrophic flow with mean tilted isopycnals.

On the other hand, barotropic instability is

similar to instabilities of all sheared flows

including those without Earth’s rotation.

7.7.6. Kelvin Waves

Coastlines and the equator can support

a special type of hybrid wave called a “Kelvin

wave,” which includes both gravity wave and

Coriolis effects. Kelvin waves are “trapped” to

the coastlines and trapped at the equator, which

means that their amplitude is highest at the coast

(or equator) and decays exponentially with

offshore (or poleward) distance. Kelvin waves

are of particular importance on eastern boundaries

since they transfer information poleward

from the equator. They are also central to how

the equatorial ocean adjusts to changes in wind

forcing, such as during an El Niño (Chapter 10).

Kelvin waves propagate with the coast to the

right in the Northern Hemisphere and to the left

in the Southern Hemisphere. At the equator,

which acts like a boundary, Kelvin waves propagate

only eastward. In their alongshore

direction of propagation, Kelvin waves behave

just like surface gravity waves and obey the

gravity wave dispersion relation (Section 8.3).

However, unlike surface gravity waves, Kelvin

waves can propagate in only one direction.

Kelvin wave wavelengths are also very long,

on the order of tens to thousands of kilometers,

compared with the usual surface gravity waves

at the beach. Although the wave propagation

speed is high, it can take days to weeks to see

the transition from a Kelvin wave crest to

a Kelvin wave trough at a given observation

point.

In the across-shore direction, Kelvin waves

differ entirely from surface gravity waves. Their

amplitude is largest at the coast. The offshore

decay scale is the Rossby deformation radius

(Section 7.7.4).

Lastly, Kelvin wave water velocities in the

direction perpendicular to the coast are exactly

zero. The water velocities are therefore exactly

parallel to the coast. Moreover, the alongshore

velocities are geostrophic, so they are associated

with pressure differences (pressure gradient

force) in the across-shore direction.

7.8. WIND-DRIVEN

CIRCULATION: SVERDRUP

BALANCE AND WESTERN

BOUNDARY CURRENTS

The large-scale circulation in the ocean basins

is asymmetric, with swift, narrow currents

along the western boundaries, and much gentler

flow within the vast interior, away from the side

boundaries. This asymmetry is known as

westward intensification of the circulation; it

occurs in both the Northern and Southern

Hemispheres and in the subtropical and

subpolar gyres. Sverdrup (1947) first explained

the mid-ocean vorticity balance now called

the “Sverdrup interior” solution. Stommel

(1948) and Munk (1950) provided the first (frictional)

explanations for the western boundary

212


currents, and Fofonoff (1954) showed how very

different the circulation would be without

friction.

7.8.1. Sverdrup Balance

The gentle interior flow of the (non-equatorial)

oceans can be described in terms of its

meridional (north-south) direction. In the

subtropical gyres, the interior flow is toward

the equator in both the Northern and Southern

Hemispheres. In the subpolar gyres, the interior

flow is poleward in both hemispheres. These

interior flow directions can be understood

through a potential vorticity argument introduced

by Sverdrup (1947), so we call the applicable

physics the “Sverdrup balance.”

Consider a schematic of the subtropical

North Pacific (Figure 7.13). The winds at the

sea surface are not spatially uniform (Figures

5.16a c and S5.10a). South of about 30 N, the

Pacific is dominated by easterly trade winds.

North of this, it is dominated by the westerlies.

This causes northward Ekman transport under

the trade winds, and southward Ekman transport

under the westerlies. As a result, there is

Ekman convergence throughout the subtropical

North Pacific (Figures 5.16d and S5.10a).

The convergent surface layer water in the

subtropics must go somewhere so there is

downward vertical velocity at the base of the

(50 m thick) Ekman layer. At some level

between the surface and ocean bottom, there

is likely no vertical velocity. Therefore there is

net “squashing” of the water columns in the

subtropical region (also called Ekman pumping).

This squashing requires a decrease in either

planetary or relative vorticity (Eq. 7.35). In the

Subpolar gyre

Westerlies

FIGURE 7.13 Sverdrup balance circulation

(Northern Hemisphere). Westerly

and trade winds force Ekman transport,

creating Ekman pumping and suction

and hence Sverdrup transport. See also

Figure S7.12.

North

Subtropical gyre

Trades

Ekman transport

East

Ekman

upwelling

Tropical gyre

Thermocline

Ekman

downwelling

Ekman transport

Ekman

upwelling

Northern Hemisphere

Sverdrup transport


ocean interior, relative vorticity is small, so planetary

vorticity must decrease, which results in

the equatorward flow that characterizes the

subtropical gyre (Figure S7.26).

The subpolar North Pacific lies north of

the westerly wind maximum at about 40 N.

Ekman transport is therefore southward, with

a maximum at about 40 N and weaker at higher

latitudes. Therefore there must be upwelling

(Ekman suction) throughout the wide latitude

band of the subpolar gyre. This upwelling

stretches the water columns (Eq. 7.35), which

then move poleward, creating the poleward

flow of the subpolar gyre.

The Sverdrup transport is the net meridional

transport diagnosed in both the subtropical

and subpolar gyres, resulting from planetary

vorticity changes that balance Ekman pumping

or Ekman suction.

All of the meridional flow is returned in

western boundary currents, for reasons

described in the following sections. Therefore,

subtropical gyres must be anticyclonic and

subpolar gyres must be cyclonic.

Mathematically, the Sverdrup balance is

derived from the geostrophic equations of

motion with variable Coriolis parameter

f (Eq. 7.23a,b). The x-andy-momentum equations

are combined to form the vorticity equation:

fðvu=vx þ vv=vyÞþbv ¼ 0 (7.41)

Using the continuity equation

vu=vx þ vv=vy þ vw=vz ¼ 0 (7.42)

Eq. (7.41) becomes the potential vorticity balance

bv ¼ f vw=vz: (7.43)

This important equation states that water

column stretching in the presence of rotation is

balanced by a change in latitude (Figure S7.26).

In Eq. (7.43), the vertical velocity w is due to

Ekman pumping. From Eqs. (7.20) and (7.21):

w ¼ v=vxðs ðyÞ =rfÞ v=vyðs ðxÞ =rfÞ ¼}curl s}

(7.44)

where s is the vector wind stress, s (x) is the zonal

wind stress, and s (y) is the meridional wind

stress. Assuming that the vertical velocity w is

zero at great depth, Eq. (7.43) can be vertically

integrated to obtain the Sverdrup balance:

bðM ðyÞ ðs ðxÞ =fÞÞ ¼ v=vxðs ðyÞ Þ v=vyðs ðxÞ Þ

¼ }curl s}

(7.45)

where the meridional (south-north) mass transport

M (y) is the vertical integral of the meridional

velocity v times density r. The second term on

the left side is the meridional Ekman transport.

Thus, the meridional transport in the Sverdrup

interior is proportional to the wind stress curl

corrected for the Ekman transport.

The meridional transport M (y) is the Sverdrup

transport. A global map of the Sverdrup transport

integrated from the eastern to the western

boundary is shown in Figure 5.17. The size of

the integral at the western boundary gives the

western boundary current transport since

Sverdrup’s model must be closed with a narrow

boundary current that has at least one additional

physical mechanism beyond those in the

Sverdrup balance (a shift in latitude because of

water column stretching driven by Ekman

transport convergence). Physics of the boundary

currents are discussed in the following sections.

7.8.2. Stommel’s Solution: Westward

Intensification and Western Boundary

Currents

Because the Sverdrup balance applies to the

whole ocean basin, the return flow must be in

a narrow, swift meridional jet where the potential

vorticity balance is different from the

Sverdrup balance. Stommel (1948) included

dissipation of potential vorticity Q on the

right-hand side of Eq. (7.37), and showed that

the returning flow must be along the western

boundary (Figure S7.31). His potential vorticity

balance is change in planetary vorticity

214


balanced by bottom friction. Stommel’s idealized

circulation resembles the western-intensified

Gulf Stream and Kuroshio subtropical

gyres. Much more discussion is provided in

Section S7.8.2 in the textbook Web site http://

booksite.academicpress.com/DPO/.

7.8.3. Munk’s Solution: Western

Boundary Currents

Like Stommel, Munk (1950) also showed

western intensification of the gyres, but used

a more realistic type of dissipation that could

work equally well with a stratified ocean. To

the potential vorticity conservation (Eq. 7.37),

Munk added friction between the currents and

the side walls. A narrow, swift jet along the

western boundary returns the Sverdrup interior

flow to its original latitude (Figure S7.32).

We use Munk’s model to understand why the

returning jet must be on the western rather than

the eastern boundary (Figure 7.14). In the

Sverdrup interior of a subtropical gyre, Ekman

pumping squashes the water columns, which

then move equatorward to lower planetary

vorticity. To return to a higher latitude (i.e.,

increase the planetary vorticity), higher vorticity

must be put back into the fluid through either

stretching or relative vorticity. Stretching

through wind stress curl in very narrow regions

does not occur where and when it would be

needed. Therefore, there must be an input of

relative vorticity. Relative vorticity in a narrow

boundary current is high because the horizontal

shear is high. At the side wall, the velocity

parallel to the wall is zero; it increases offshore,

so the boundary current has positive relative

vorticity if it is on a western boundary (Figure

7.14a). This vorticity is injected into the fluid

by the friction at the wall, which then allows the

planetary vorticity to change and the fluid to

return to its original latitude; such a circulation

closure would not be possible if the boundary

current were on the eastern boundary

(Figure 7.14b). The online version of this text

includes a more detailed explanation (see textbook

Web site http://booksite.academicpress.

com/DPO/).

7.8.4. Fofonoff’s Solution: Large-Scale

Inertial Currents

In addition to interior flows created from

Ekman pumping or other external sources of

vorticity, there are free, unforced modes of circulation,

as shown by Fofonoff (1954), in a model

with no wind input and no friction

(Figure S7.34). This type of circulation is called

an “inertial circulation”. Without an external

vorticity input, the interior flow is exactly zonal

(east-west), because there is no way to change

its planetary vorticity (due to the b-effect).

Suppose there is westward flow across the

middle of the ocean. When it reaches the

western boundary, it gets back to the eastern

boundary to feed back into the westward flow

by moving either northward or southward

along the western boundary in a very narrow

current that can have as much relative vorticity

as needed. Suppose it is northward. Then the

relative vorticity of this frictionless current is

positive, allowing it to move to a higher latitude.

It then jets straight across the middle of the

ocean, reaches the eastern boundary, and forms

another narrow jet and moves southward,

feeding into the westward flow in the interior.

Aspects of the Fofonoff inertial solution are

found in highly energetic regions, such as near

the separated Gulf Stream where its transport

increases far above that predicted by the

Sverdrup balance. The recirculation gyres associated

with this energetic part of the Gulf Stream

can be partially thought of in terms of Fofonoff

gyres.

7.8.5. Wind-Driven Circulation in

a Stratified Ocean

What happens to the wind-driven circulation

theories in a stratified ocean? Water moves


(a)

North

Western boundary (coastline)

Input

positive

relative

vorticity

Frictional

boundary

layer

Western boundary current

Frictional western boundary

layer (Munk, 1950): input of

positive relative vorticity allows

northward boundary current

(increasing planetary vorticity)

East

Southward interior

(Sverdrup) flow

FIGURE 7.14 (a) Vorticity

balance at a western boundary, with

sidewall friction (Munk’s model).

(b) Hypothetical eastern boundary

vorticity balance, showing that only

western boundaries can input the

positive relative vorticity required

for the flow to move northward.

(b)

What happens if the boundary current is on

the eastern boundary? Input of negative

relative vorticity cannot allow northward boundary

current. This solution is not permissible as a

balance for southward Sverdrup interior flow.

West

Southward interior

(Sverdrup) flow

Input

negative

relative

vorticity

Frictional

boundary

layer

Impermissible eastern boundary current

North

Eastern boundary (coastline)

down into the ocean, mostly along very gradually

sloping isopycnals. Where streamlines of

flow are connected to the sea surface, we say

the ocean is directly ventilated (Figure 7.15).

Where there is Ekman pumping (negative

wind stress curl), the Sverdrup interior flow is

equatorward (Section 7.8.1). Water columns at

the local mixed layer density move equatorward

and encounter less dense water at the

surface. They slide down into the subsurface

along isopycnals, still moving equatorward.

This process is called subduction (Luyten,

Pedlosky, & Stommel, 1983), using a term borrowed

from plate tectonics. The subducted

waters then flow around the gyre and enter

the western boundary current if they do not

first enter the tropical circulation. The details

are beyond the scope of this text.

216


SURFACE

OUTCROP

EASTERN BOUNDARY

FIGURE 7.15 Subduction

schematic (Northern Hemisphere).

See Figure S7.35 for additional

schematics, including obduction.

WESTERN

UNVENTILATED

POOL

WIND

SUBDUCTED

REGION

WIND

EASTERN

SHADOW

ZONE

N

W

WESTERN BOUNDARY

ABYSSAL

OCEAN

In each subducted layer, there can be three

regions (Figure 7.15): (1) a ventilated region connected

from the sea surface as just described,

(2) a western unventilated pool with streamlines

that enter and exit from the western

boundary current without entering the surface

layer, and (3) an eastern quiet (shadow) zone

between the easternmost subducting streamline

and the eastern boundary. A continuous

range of surface densities is found in the

subtropical gyre; the water column is directly

ventilated over this full range, with waters at

each density coming from a different seasurface

location depending on the configuration

of streamlines on that isopycnal. This is called

the “ventilated thermocline”; in water mass

terms, this process creates the Central Water.

The maximum density of the ventilated thermocline

is set by the maximum winter surface

density in the subtropical gyre (Stommel,

1979).

The opposite of subduction is obduction, borrowed

again from plate tectonics by Qiu and

Huang (1995). In obducting regions, waters

from subsurface isopycnals come up and into

the surface layer. These are generally upwelling

regions such as the cyclonic subpolar gyres and

the region within and south of the Antarctic

Circumpolar Current.

Wind-driven circulation occurs in nonventilated

stratified regions as well. It is most

vigorous in regions connected to the western

boundary currents where water can enter and

exit the western boundary. In these regions, the

western boundary currents and their separated

extensions usually reach to the ocean bottom.

These dynamics are also beyond our scope.

7.9. WIND-DRIVEN

CIRCULATION: EASTERN

BOUNDARY CURRENTS AND

EQUATORIAL CIRCULATION

7.9.1. Coastal Upwelling and Eastern

Boundary Currents

The eastern boundary regions of the

subtropical gyres have strong but shallow

flow that is dynamically independent of the

open ocean gyre regimes. Upper ocean eastern

boundary circulation is driven by alongshore

wind stress that creates onshore (or offshore)

Ekman transport that creates upwelling (or

WIND-DRIVEN CIRCULATION: EASTERN BOUNDARY CURRENTS AND EQUATORIAL CIRCULATION 217

downwelling; Section 7.5.4). Beneath or inshore

of the equatorward eastern boundary currents

there is a poleward undercurrent or countercurrent.

Coastal upwelling systems are not

restricted to eastern boundaries; the southern

coast of the Arabian peninsula has the same

kind of system. These circulations are fundamentally

different from western boundary

currents, which are tied to potential vorticity

dynamics (Section 7.8).

The classical explanation of eastern boundary

currents is that equatorward winds force Ekman

flow offshore, which drives a shallow upwelling

(on the order of 200 m deep) in a very narrow

region adjacent to the coast (on the order

10 km; Figure 7.6c). The upwelling speed is

about 5e10 m/day. Because of stratification,

the source of upwelled water is restricted to

layers close to the sea surface, usually between

50 and 300 m.

Thezoneofcoastalupwellingcanbeextended

to more than 100 km offshore by an increase in

longshore wind strength with distance offshore;

this is observed in each eastern boundary

upwelling system due to topographic steering

of the winds by the ocean-land boundary. The

offshore Ekman transport therefore increases

with distance offshore, which requires upwelling

through the whole band (Bakun & Nelson, 1991).

The zone is identified by positive wind stress

curl (Figure 5.16d).

Upwelled water is cooler than the original

surface water. It originates from just below the

euphotic zone and therefore is also rich in nutrients.

Cool surface temperatures and enhanced

biological productivity are clear in satellite

images.

Upwelling is strongly seasonal, due to seasonality

in the winds. Onset of upwelling can

be within days of arrival of upwelling-favorable

winds. In one example, off the coast of Oregon,

the surface temperature dropped by 6 C in two

days after a longshore wind started.

Coastal upwelling is accompanied by a rise

in upper ocean isopycnals toward the coast

(Figure 7.6). This creates an equatorward

geostrophic surface flow, the eastern boundary

current. These currents are narrow (<100 km

width and near the coast), shallow (upper

100 m), strong (40 to 80 cm/sec), and strongly

seasonal. The actual flow in an eastern

boundary current system includes strong,

meandering eddies and offshore jets/filaments

of surface water, often associated with coastline

features such as capes (Figure 10.6). Actual

eastern boundary currents are some distance

offshore, at the axis of the upwelling front

created by the offshore Ekman transport.

Poleward undercurrents are observed at about

200 m depth beneath the equatorward surface

currents. They are driven by a poleward pressure

gradient force along the eastern boundary.

When upwelling-favorable winds weaken or

disappear, the equatorward flow also disappears

and the poleward undercurrent extends

up to the surface (there is no longer an

undercurrent).

7.9.2. Near-Surface Equatorial

Currents and Bjerknes Feedback

Circulation within about 2 latitude of the

equator is very different from non-equatorial

circulation because the Coriolis parameter

f vanishes at the equator. Equatorial circulation

is driven by easterly trade winds in the

Pacific and Atlantic and by the seasonally

reversing monsoonal winds in the Indian

Ocean.

Since the Coriolis parameter vanishes and

there is no frictional Ekman layer at the equator,

the easterly trade winds drive equatorial surface

flow due westward in a frictional surface layer

(“normal panel” in Figure 10.27). The westward

surface current is shallow (50 to 100 m) and of

medium strength (10 to 20 cm/sec). In each of

the three oceans, this westward surface flow is

a part of the South Equatorial Current. The water

piles up gently in the west (to about 0.5 m

height) and leaves a depression in the east.

218


This creates an eastward pressure gradient force

(from high pressure in the west to low pressure

in the east). The pressure gradient force drives

an eastward flow called the Equatorial Undercurrent

(EUC). The EUC is centered at 100 to 200 m

depth, just below the frictional surface layer.

The EUC is only about 150 m thick. It is among

the strongest ocean currents (>100 cm/sec). The

pileup of waters in the western equatorial

region also results in a deepened pycnocline

there, called the warm pool, and a shoaling of

the pycnocline in the eastern equatorial region

(upwelling into the cold tongue). Coriolis effects

set in a small distance from the equator, and

off-equatorial Ekman transport enhances

upwelling in the equatorial band. This enhances

the cold tongue. The east-west contrast in

temperature along the equator maintains the

atmosphere’s Walker circulation, which has

ascending air over the warm pool and descending

over the eastern colder area.

The Walker circulation is an important part of

the trade winds that create the warm pool and

cold tongue, so there can be a feedback between

the ocean and atmosphere; this is called the

Bjerknes feedback (Bjerknes, 1969) (Figure S7.36b

located on the textbook Web site). If something

weakens the trade winds, as happens at the

beginning of an El Niño event (Chapter 10),

the westward flow at the equator weakens and

upwelling weakens or stops. Surface waters in

the eastern regions therefore warm. Water in

the deep warm pool in the west sloshes eastward

along the equator, thinning the pool. The

change in SST weakens the Walker circulation/

trade winds even more, which further exacerbates

the ocean changes. This is an example of

a positive feedback.

In the Indian Ocean, the prevailing equatorial

winds are monsoonal, meaning that trade winds

are only present for part of the year. This creates

seasonally reversing equatorial currents and

inhibits the formation of the warm pool/cold

tongue structure. The Indian Ocean sea surface

temperature is high at all longitudes.

7.10. BUOYANCY

(THERMOHALINE) FORCING AND

ABYSSAL CIRCULATION

Heating and cooling change the ocean’s

temperature distribution, while evaporation,

precipitation, runoff, and ice formation change

the ocean’s salinity distribution (Chapters 4

and 5). Collectively, these are referred to as

buoyancy, or thermohaline, forcing. Buoyancy

processes are responsible for developing the

ocean’s stratification, including its abyssal properties,

pycnocline, thermocline, halocline, and

upper layer structure (other than in windstirred

mixed layers).

Abyssal circulation refers to the general category

of currents in the deep ocean. The overturning

circulation, also called the thermohaline

circulation, is the part of the circulation associated

with buoyancy changes, and overlaps

spatially with the wind-driven upper ocean

circulation; it also includes shallow elements

that are independent of the abyssal circulation.

In the overturning circulation, cooling and/or

salinification at the sea surface causes water to

sink. This water must rise back to the warm

surface, which requires diffusion of heat (buoyancy)

downward from the sea surface. The

source of eddy diffusion is primarily wind and

tidal energy.

7.10.1. Buoyancy Loss Processes

(Diapycnal Downwelling)

Water becomes denser through net cooling,

net evaporation, and brine rejection during sea

ice formation. We have already described brine

rejection (Section 3.9); it is responsible for

creating the densest bottom waters in the global

ocean (Antarctic Bottom Water and parts of the

Circumpolar Deep Water) and also in the

regional basins where it is operative (Arctic

Ocean, Japan Sea, etc.). Here we focus on convection

created by net buoyancy loss in the open

ocean, when surface water becomes denser

BUOYANCY (THERMOHALINE) FORCING AND ABYSSAL CIRCULATION 219

than water below, and advects and mixes downward.

Diurnal (daily) convection occurs at night

in areas where the surface layer restratifies

strongly during the day. During the annual cycle,

cooling usually starts around the autumnal

equinox and continues almost until the spring

equinox. The resulting convection eats down

into the surface layer, reaching maximum depth

and density at the end of winter when the cumulative

cooling reaches its maximum. A convective

mixed layer can be hundreds of meters

thick by the end of winter, whereas a windstirred

mixed layer is limited to about 150 m by

the depth of wind-driven turbulence.

Ocean convection is usually driven by

surface cooling. Excess evaporation can also

create convection, but the latent heat loss associated

with evaporation is usually dominant.

“Deep” convection is a loose term that usually

refers to creation of a surface mixed layer that

is thicker than about 1000 m. Deep convection

has three phases: (1) preconditioning (reduction

in stratification), (2) convection (violent mixing),

and (3) sinking and spreading. (See Killworth,

1983 and Marshall & Schott, 1999.)

Convective regions have a typical structure

(Figure S7.37). These include: (a) a chimney,

which is a patch of 10 km to more than 100 km

across within which preconditioning can allow

convection and (b) convective plumes that are

about 1 km or less across. The plumes are about

the same size across as they are deep.

Deep convection occurs in only a very few

special locations around the world: Greenland

Sea, Labrador Sea, Mediterranean Sea, Weddell

Sea, Ross Sea, and Japan (or East) Sea. These

sites, with the exception of the isolated Japan

Sea, ventilate most of the deep waters of the

global ocean.

7.10.2. Diapycnal Upwelling (Buoyancy

Gain)

The structure of the basin and global scale

overturning circulations depends on both the

amount of density increase in the convective

source regions and the existence of a buoyancy

(heat) source at lower latitudes that is at least

as deep as the extent of the cooling (Sandström,

1908; Figure S7.38). Since there are no significant

local deep heat sources in the world ocean,

waters that fill the deep ocean can only return

to the sea surface as a result of diapycnal eddy

diffusion of buoyancy (heat and freshwater)

downward from the sea surface (Sections 7.3.2

and 5.1.3).

Munk’s (1966) diapycnal eddy diffusivity

estimate of k v ¼ 1 10 4 m 2 /sec (Section 7.3)

was based on the idea of isolated sources of

deep water and widespread diffusive upwelling

of this deep water back to the surface. From all

of the terms in the temperature and salt equations

(7.12) and (7.13), Munk assumed that

most of the ocean is dominated by the balance

vertical advection ¼ vertical diffusion (7.46a)

w vT=vz ¼ v=vzðk v vT=vzÞ

(7.46b)

Munk obtained his diffusivity estimate from an

average temperature profile and an estimate of

about 1 cm/day for the upwelling velocity w,

which can be based on deep water formation

rates and an assumption of upwelling over the

whole ocean. The observed diapycnal eddy

diffusivity in the open ocean away from boundaries

is much smaller than Munk’s estimate,

which must be valid for the globally averaged

ocean structure. This means that there must be

much larger diffusivity in some regions of the

ocean, now thought to be at the boundaries, at

large seamount and island chains, and possibly

the equator (Section 7.3).

7.10.3. Stommel and Arons’ Solution:

Abyssal Circulation and Deep Western

Boundary Currents

Deep ocean circulation has been explained

using potential vorticity concepts that are very

familiar from Sverdrup balance (Section 7.8;

220


Stommel, 1958; Stommel & Arons, 1960a,b). The

sources of deep water are very localized. The

deep water fills the deep ocean layer, which

would raise the upper interface of this layer if

there were no downward eddy diffusion (Section

7.10.2). This upwelling stretches the deep ocean

water columns. Stretching requires a poleward

shift of the water columns to conserve potential

vorticity (Eq. 7.36). The predicted interior flow

is therefore counterintuitive d it runs toward

the deepwater source! (Actual abyssal flow is

strongly modified from this by the major topography

that modifies the b-effect, by allowing

stretched columns to move toward shallower

depths rather than toward higher latitude.)

Deep Western Boundary Currents (DWBCs)

connect the isolated deepwater sources and

the interior poleward flows. Whereas unambiguous

poleward flow is not observed in the deep

ocean interior (possibly mostly because of

topography), DWBCs are found where they

are predicted to occur by the Stommel and

Arons abyssal circulation theory (Figure 7.16;

Warren, 1981). One such DWBC runs southward

beneath the Gulf Stream, carrying dense waters

from the Nordic Seas and Labrador Sea.

Swallow and Worthington (1961) found this

current after being convinced by Stommel to

go search for it.

7.10.4. Thermohaline Oscillators:

Stommel’s Solution

An entirely different approach to understanding

the meridional overturning circulation

considers changes in overturn due to changes

in dense water production by reducing the ocean

to just a few boxes (Stommel, 1961). Such box

models show how even the simplest model of

climate change, for example, can lead to complex

results. In this case, multiple equilibria result, that

is, the system can jump suddenly between quite

different equilibrium states. Stommel reduced

the ocean to two connected boxes representing

dense, cold, fresh high latitudes and light,

warm, saltier low latitudes (Figure S7.41). In

each box, the temperature and salinity are set

by: (1) flux of water between the boxes (thermohaline

circulation) that depends on the density

difference between the boxes and (2) restoring

temperature and salinity to a basic state over

some set time period.

Stommel found that several different thermohaline

circulation strengths exist for a given set

FIGURE 7.16 Global abyssal

circulation model, assuming two

deep water sources near Greenland

and Antarctica (filled circles), filling

a single abyssal layer. (These sources

are actually at different densities.)

Source: From Stommel (1958).


of choices of model parameters (externally

imposed temperature and salinity, restoration

timescales for temperature and salinity, and

factor relating the flow rate to the density difference

between the boxes). As the basic state was

slowly changed, perhaps by reduction of the

basic high-latitude salinity (which reduces its

density), the flow rate slowly changed and

then suddenly jumped to a different equilibrium

rate. When the basic state salinity was

then slowly increased, the system jumped back

to a higher flow rate but at a very different basic

salinity than during its decreasing phase. Thus

this system exhibits hysteresis: it has different

equilibrium states depending on whether the

state is approached from a much higher salinity

or a much lower salinity.

The coupled atmosphere-sea-ice-land-physicsbiology-chemistry

climate system is far more

complex than the two simple boxes in this very

simple Stommel oscillator model. Yet its

multiple equilibria and hysteresis behavior

have been useful in demonstrating the potential

for abrupt and relatively large changes in

climate and, more specifically, for interpretation

of numerical models of the changes in overturning

circulation that could result from changes in

external forcing.

C H A P T E R

8

Gravity Waves, Tides, and Coastal

Oceanography

8.1. INTRODUCTION

This chapter continues the dynamical discussion

of Chapter 7, starting with an overview of

the properties of waves (Section 8.2), and moving

to surface and internal gravity waves and tides

(Sections 8.3 to 8.6). These sections are truncated

here, but appear in full in Chapter S8 of the

online supplement located at the textbook Web

site http://booksite.academicpress.com/DPO/;

“S” denotes supplemental material.

This chapter then continues in supplementary

form on the textbook Web site, covering

several aspects of the coastal regime: coastal

runoff, estuaries, and coral reefs (Sections S8.7

to S8.9). The supplement ends with descriptions

of circulation and water properties in various

adjacent seas of the Atlantic (Mediterranean,

Black, Baltic and North Seas), Pacific (Bering,

Okhotsk, Japan, Yellow, East and South China

Seas, Gulf of California), and Indian Ocean

(Red Sea and Persian Gulf) (Section S8.10).

Relevant advanced treatments of waves

include Phillips (1977), Lighthill (1978),

Pedlosky (2003) and Mei, Stiassnie, and Yue

(2005). Some suggestions for coastal oceanography

texts are Komar (1998), Van Dorn (1993),

Open University (1999), Tomczak (2002), and

Stewart (2008). Comprehensive reviews of

coastal oceanography and adjacent seas are

found in many volumes of The Sea (Brink &

Robinson, 1998; Robinson & Brink, 1998, 2005,

2006; Bernard & Robinson, 2009).

8.2. GENERAL PROPERTIES

OF WAVES

Waves are the displacement of parcels in

a medium, such as water, that has a force that

pushes the parcel back to its initial position,

where it overshoots and is then restored back

again, overshoots, and is then restored. For

example, for surface gravity waves (Section 8.3),

the medium is water (the airesea interface)

and the restoring force is gravity acting on

parcels displaced vertically at the interface. All

types of waves are generated by some external

force that creates the initial displacement of

particles away from their equilibrium position.

For surface gravity waves, the most common

external force is the wind, although undersea

earthquakes can also generate them (tsunamis).

Waves are described in terms of their wavelength,

period, amplitude, and direction

(Figure 8.1). The wavelength (L) is the distance

from one wave crest to the next or from one

trough to the next. Another quantity used to

describe the length of waves is the wavenumber

(k), where k ¼ 2p/L and has units of radians


223



224

8. GRAVITY WAVES, TIDES, AND COASTAL OCEANOGRAPHY

Wave crest

FIGURE 8.1

Wavenumber = 2π __

L

Wavelength L

Wave

height

Wave trough

Schematic of a sinusoidal wave.

Wave

amplitude

per unit length. The wave period (T) is the time

between observing successive crests (or troughs)

passing a fixed point. The wave frequency (u) is

2p/T with units of radians per unit time.

(Frequency given in hertz would be 1/T.) The

wave amplitude is conventionally one-half the

height of the wave from crest to trough.

Two different types of velocity describe how

all waves travel: phase velocity and group

velocity. Phase velocity is the velocity of individual

wave crests. The phase velocity c p is

c p ¼ L=T ¼ u=k (8.1)

where L is the wavelength, T is the wave period,

u is the frequency, and k is the wavenumber.

Waves are non-dispersive if the phase velocity is

a single constant for all wavelengths. If the

phase velocity is not constant, waves are dispersive

and separate from each other. The dispersion

relation for a given type of wave (e.g., surface

waves, internal waves, or acoustic waves)

expresses the wave frequency in terms of the

wavelength or wavenumber, that is, u ¼ u(k).

For waves that move in several different directions,

a wavenumber is defined for each direction.

For the (x, y, z) directions, these

wavenumbers are often called (k, l, m). The

phase speed (Eq. 8.1) is defined in each of these

directions.

Wave energy moves at a different speed from

the wave crests for most types of waves. This

speed is called the group velocity. In readily

recognized examples (such as the waves in

a boat wake), a wave group (packet) moves

out from the source, and individual waves propagate

through the packet. The packet moves at

the group velocity. In deep-water surface waves,

the phase velocity is faster than the group

velocity, so it looks like the waves just appear

from one side of the packet, move through,

and disappear out the other side. Formally,

group velocity (c g ) is the derivative of frequency

with respect to wavenumber. In one dimension,

this is

c g ¼ vu=vk (8.2)

For two and three dimensions, the group

velocity is a vector:

c g ¼ðvu=vk; vu=vl; vu=vmÞ (8.3)

For non-dispersive waves, the group velocity

must be the same constant as the phase velocity.

8.3. SURFACE GRAVITY WAVES

8.3.1. Definitions and Dispersion

Relation

The restoring force for surface gravity waves

is gravity, assisted by the large difference

between the density of air and that of water,

which acts against any disturbance of the free

surface. Any external forcing that can momentarily

mound up the water causes surface

gravity waves: wind, a passing boat, slumping

of the ocean bottom caused by an earthquake,

and so forth.

Surface gravity waves with wavelengths that

are much shorter than the water depth are

referred to as deep-water waves or “short waves.”

Water particle motion in a deep-water wave is

nearly circular in the vertical plane parallel to

the direction of wave propagation. The diameter

of the circles decreases exponentially with

depth and the wave does not “feel” the bottom.

At the other extreme are shallow-water waves or

“long waves”, whose wavelengths are greater

SURFACE GRAVITY WAVES 225

than the water depth. The water particles move

elliptically in the vertical plane rather than in

circles.

The dispersion relation for an ideal (linear,

sinusoidal), short (deep-water), surface gravity

wave is

p

u ¼

ffiffiffiffiffi

gk

(8.4a)

Therefore the phase velocity (Eq. 8.1) of a short

surface gravity wave is

c p ¼ u rffiffiffi

k ¼ g

(8.4b)

k

and the group velocity (Eq. 8.2) is

c g ¼ vu

vk ¼ 1 rffiffiffi

g

2 k

(8.4c)

Therefore, short surface gravity waves (large k)

move more slowly than longer surface gravity

waves (smaller k), and energy propagates at

a different, slower, speed than phase.

For shallow-water (long) gravity waves, in

water of depth (d), the dispersion relation is

p

u ¼ k

ffiffiffiffiffiffi

gd

(8.5a)

Their phase and group velocities are

c p ¼ u k ¼

c g ¼ vu

vk ¼

p ffiffiffiffiffiffi

gd

p ffiffiffiffiffiffi

gd

(8.5b)

(8.5c)

When the group speed is a constant (and therefore

equal to the phase speed), as in Eq. (8.5c), the

waves are non-dispersive; that is, energy moves

at the same speed for all wavelengths.

8.3.2. Wind-Forced Surface Gravity

Waves

On an extremely calm day, the ocean surface

appears glassy, with no visible short waves. As

the wind starts to blow, small capillary waves

form and the water surface begins to appear

slightly rough. The wind-forced waves grow

and change through differences in air pressure

created by the wind between the front and backside

of the waves. The pressure differences

become larger as the surface gravity waves

grow. Nonlinear interactions between the waves

spread energy to longer wavelengths and lower

frequencies.

The resulting wave state produced by local

winds is called the wind-sea. These wind-forced

surface gravity waves have periods and wavelengths

that range from about 1 to 25 sec and

about 1 to 1000 m. The amplitude and

frequency/wavelength of waves generated

locally by the wind depends on the wind duration

(time over which the wind blows), fetch

(distance over which the wind blows), and

strength. In a storm, with wind gusts in many

different directions, the rough sea surface

becomes choppy with waves traveling in all

directions (confused sea). Whitecaps appear

when the wind strength exceeds about 10 knots

(3 m/sec). A fully developed sea arises after the

wind blows for many days with a very long

fetch. The white caps and foam in the photograph

in Figure 8.2a are characteristic of strong

wind conditions, in this case, in the Gulf of

Tehuantepec in the eastern Pacific (Section

10.7.6 and wind curl map in Figure 5.16d).

When the wind slows, the shorter waves are

damped out, leaving behind longer, slower,

smoother waves called swell. Swell wavelengths

are tens of meters. Swell can propagate exceedingly

long distances with little damping. Using

the dispersion relation (8.4), swell with a period

of 14 sec travels at a phase speed of 22 m/sec

and group speed of 11 m/sec, taking approximately

5 days to propagate from the Gulf of

Alaska to the north shore of Hawaii, a distance

of about 4500 km. The swell arriving on distant

beaches often is fairly narrow-band in

frequency, so it consists of well-defined sets

separated by relatively quiescent intervals.

226


(a)

(b)

Spectral density (m 2 /Hz)

12

10

8

6

4

2

Swell

Separation frequency

Seas

(c)

(d)

0

0 0.05 0.1 0.15 0.2 0.25 0.3 0.35

Frequency (Hz)

3

Period (sec)

20 15 10 9 8 7 6 5 4 3

0.4

Spectral density (m 2 /Hz)

2

1

46006

2009/05/16 12z

0

0.1

0.2 0.3 0.4

Frequency (Hz)

FIGURE 8.2 Wind waves. (a) Open ocean waves in the Gulf of Tehuantepec (February 7, 2009), under wind speeds of

20e25 m/sec, including actively breaking waves, old foam patches, and streaks of foam (K. Melville, personal communication,

2009). (b) Example of a surface wave spectrum (spectral density) in which ocean swell and wind-seas are well

separated in frequency. Source: From National Data Buoy Center (2006). (c, d) Directional wave spectrum (spectral density) and

spectrum, without clear separation between swell and the wind-sea, from the NE Pacific (station 46006, 40 53’ N 137 27’ W,

May 16, 2009. In (c), wave periods are from about 25 sec at the center of the ring to 4 sec at the outer ring. Blue is low energy,

purple is high. Direction of the waves is the same as direction relative to the center of the circle. Gray arrow in center

indicates wind direction. “Hs” indicates significant wave height. Figure 8.2c can also be found in the color insert. Source:

part c is from NOAA Wavewatch III (2009) and part d is from National Data Buoy Center (2009).

Since waves of many different frequencies

and wavelengths are present in the open ocean

at the same time, the surface gravity wave field

is often described using spectral analysis

(Section 6.5.3). The spectra can often have two

separate peaks associated with the local windseas

and with the swell (Figure 8.2b). However,

many spectra show no clear separation between

swell and wind seas (Figure 8.2d). Directional

wave spectra that show energy as a function of


frequency and the direction of the waves

(Figure 8.2c) can clarify what is happening in

the non-directional spectrum.

Description and forecast of the wave state in

the open ocean is crucial for shipping. Wave

observations from buoys and satellites are

analyzed using global wave models (Figure 8.3).

The significant wave height in the left panel of

Figure 8.3 is the average height of the highest

one-third of the waves. For the right panel,

wave spectra at each location were used to

determine the wave period with maximum

(peak) energy; vectors show the direction of

propagation of the peak energy waves.

8.3.3. Beaches, Breaking Waves and

Associated Set-Up, and Near-Shore

Currents

Surface gravity waves move from offshore

generation regions to the near-shore region

where they impact the beach and coastline. We

distinguish between the beach, the surf zone

(where waves break), and the swash zone (where

water from the broken waves runs up the

beach). Offshore sand bars and reefs are also

important for how waves break and how

currents are set up in the surf zone. Beaches

exist in a delicate balance between variable

FIGURE 8.3 (a) Significant wave height (m) and (b) peak wave period (s) and direction (vectors) for one day (May 16,

2009). Figure 8.3a and 8.3b can also be found in the color insert. Source: From NOAA Wavewatch III (2009).

228


waves, tides, and near-shore currents. The

waves and local currents often have strong seasonality

with resulting variations in beach structure

and composition (Yates, Guza, O’Reilly, &

Seymour, 2009).

Based on their interaction with impinging

waves, beaches can be classified as dissipative

or reflective. Dissipative beaches remove much

of the wave energy. Dissipation is enhanced by

a mild bottom slope and rough material. Reflective

beaches reflect much of the wave energy;

these are more steeply sloped and/or composed

of smooth material.

As a surface wave approaches the shore, it

“feels” the bottom at a depth that depends

upon its wavelength. The wave slows down,

becoming shorter and steeper while retaining

the same period. Its height increases to the point

where it breaks. Observations suggest that

waves typically break when H/d ¼ 0.8, where

H ¼ wave height and d ¼ depth to the bottom.

If the incident waves approach the beach at

an angle, their direction changes to be more

perpendicular as they shoal; this causes the

wave crests to become more parallel to the

shore. This is called refraction. Refraction occurs

because the phase speed of the waves decreases

as the depth decreases. The offshore part of the

crest, in deeper water, moves more quickly

toward shore and the whole crest pivots. (This

is Snell’s law of refraction.) Some of the energy

of the incident waves also reflects from the

shoaling bottom as they come ashore.

If the bottom depth has alongshore variation,

as it does on almost every beach, then refraction

and reflection of incoming waves vary alongshore.

This can result in focusing of wave

energy in some locations. This is illustrated in

Figure 8.4a, which shows a large swell

approaching the shore at La Jolla, California.

A major underwater canyon is situated toward

the front of the photograph, on the north side

of the pier; this reflects much of the swell to

the left, while some swell continues onshore

(Thomson, Elgar, & Herbers, 2005).

Breaking waves are typically classified as: (a)

spilling breakers, (b) plunging breakers, or (c)

surging breakers (Figure 8.5). Spilling breakers

occur on the mildest sloping beach, plunging

breakers on a moderately sloping beach, and

surging breakers on a steep beach where the

wave reaches the beach before it has a chance

(a)

(b)

FIGURE 8.4 (a) Surf zone, looking toward the south at the Scripps Pier, La Jolla, CA. Source: From CDIP (2009). (b) Rip

currents, complex pattern of swell, and alongshore flow near the head of a submarine canyon near La Jolla, CA, Photo

courtesy of Steve Elgar (2009).


I. Spilling breakers

foam

II. Plunging breakers

III. Surging breakers

foam

nearly horizontal beach

steep beach

very steep

beach

FIGURE 8.5

Komar (1998).

Types of breaking waves: (I) spilling breaker, (II) plunging breaker, and (III) surging breaker. Source: From

horizontal swash

FIGURE 8.6 Features of the surf zone.

Source: From Komar and Holman (1986).

vertical

swash

set-up

still-water level

swash excursion

set-up slope

sea bed

to break. The type of breaker also depends on

wave steepness (ratio of wave height to wavelength).

The greatest dissipation in the surf

zone occurs for spilling breakers, and the greatest

reflection of incident waves back to deep

water occurs for surging breakers. A given

surf zone may include a combination of these

different types of breakers. Breaking wave

heights are reported similarly to open ocean

wave heights, in terms of significant breaker

height (average height of the one-third highest

breakers) and maximum breaker height.

Breaking waves transport momentum to the

near shore region. This creates wave set-up,

which is a rise in mean water level above the

mean still water line (Figure 8.6). There is

a complementary set-down as an incoming

wave trough reaches the beach. Waves that are

3 m high offshore can produce a set-up of

50 cm (Guza & Thornton, 1982; Komar &

Holman, 1986). The total run-up on the beach

is the sum of set-up, swash (landward flow of

water) of individual larger waves, and swash

due to longer period (>20 sec) surf beat. Swash

on a reflective beach is more strongly affected by

the incident waves. Swash on a dissipative

beach is more affected by longer period edge

waves. Edge waves are surface gravity waves

with relatively long periods (>20 sec) that travel

along, and are trapped to, the shoaling beach

with amplitude decreasing offshore. They are

forced by incident surface gravity waves.

The transport of mass onshore resulting from

breaking waves must be compensated by offshore

flow. There are two types of offshore flow:

undertow and rip currents. Undertow balances

the mass two-dimensionally: the onshore transport

near the sea surface in the breaking wave

zone is balanced by offshore transport in a layer

at the bottom. Rip currents, on the other hand,

230


return the mass back out to sea in horizontally

limited jets (Figure 8.4b). Rip currents occur

when alongshore flow is generated by alongshore

variability in wave breaking. This leads to variability

in set-up, which leads to an alongshore

pressure gradient that drives alongshore flow.

The location of rip currents can be controlled by

bathymetry or shoreline shape, or can be transient,

depending on instability of the alongshore

current. Rip current intensity varies with incident

wave amplitude, and is weak to absent in low

wave conditions.

8.3.4. Storm Surge

Sea level is affected by local storm systems that

drive water onshore. Storms have both very low

atmospheric pressure and strong winds. The

low pressure raises the sea surface locally within

the storm. The winds create large waves, which

can generate significant set-up at the coastline.

The winds can also push water onshore. Both

result in a rise in local sea level, called a storm

surge.

The size of a storm surge depends on the

strength of the storm and on the slope of the

bottom. For gradually sloping shelves with

shallow water far offshore, storm surges can

be large, as in the North Sea. When the shelf

depth increases quickly offshore, storm surge

can be quite small, as on the west coast of North

America, where storm surge is usually dwarfed

by the tides. Many storm surges pass quickly

and unremarkably, but when they coincide

with maximum tidal height they can be disastrous;

hurricane-force winds in conjunction

with a high spring tide flooded low-lying areas

of the North Sea in 1953.

Low-lying areas with tropical cyclones, hence

extremely strong winds and low atmospheric

pressure, are particularly susceptible to storm

surges. In Bangladesh, storm surges in 1970

(Bhola cyclone) and 1991 (Bangladesh cyclone)

reached 10 and 6 m, respectively, with enormous

loss of life (World Meteorological

Organization, 2005a,b). A storm surge of about

9 m in the Gulf of Mexico resulting from Hurricane

Katrina (2005) created the most destructive

natural disaster in U.S. history (Figure 8.7).

8.3.5. Tsunamis

Surface gravity waves can be forced by

seismic shifts in submarine topography and

other large, abrupt forcing events such as underwater

landslides, meteorite impacts, and underwater

volcanic eruptions. If there is a sudden

submarine earthquake in which the bottom

drops on one side of the fracture, the result is

a displacement of seawater from top to bottom

above the fracture of the same amplitude as

the bottom shift (see Gonzalez, 1999). The

sudden seawater displacement creates a surface

gravity wave called a tsunami, which is the Japanese

word for harbor wave.

Tsunami wavelengths are hundreds to thousands

of kilometers. Since this is much greater

than the ocean depth, the tsunami is a shallow

water wave (Eq. 8.5). Therefore, the speed and

time for a tsunami to propagate from one point

to another in the ocean are set by the ocean

depth. Frequencies are 10 minutes to about

2 hours (Mei et al., 2005). In the open ocean,

where the depth is 4000 to 5000 m, the speed

is 200 to 220 m/sec (17,280 km/day), so

tsunamis take up to a day to cross a large ocean

basin like the Pacific or Indian Ocean.

Tsunamis propagate with little decay across

vast ocean expanses. Most of the energy is

concentrated in the initial packet (Figure 8.7a

and b). The initial arrival may be either a rise

or a fall in sea level. The shape and separation

of the peaks, as well as the dispersion, depend

on the shape of the initial deformation due to

the earthquake and on the bottom geometry.

All of the energy in an idealized tsunami in

a flat-bottomed ocean is initially distributed

around a circle centered on the earthquake. As

the tsunami front moves out, the circle radius

increases and the energy per unit length along


FIGURE 8.7 Sumatra Tsunami (December 26, 2004). (a) Tsunami wave approaching the beach in Thailand. Source: From

Rydevik (2004). (b) Simulated surface height two hours after earthquake. Source: From Smith et al. (2005). (c) Global reach:

simulated maximum sea-surface height and arrival time (hours after earthquake) of wave front. Figure 8.7c can also be

found in the color insert. Source: From Titov et al. (2005).

the circumference of the circle decreases. The

tsunami refracts and scatters as it crosses deep

topographic features, resulting in less energy

density in some regions and more in others.

Mid-ocean ridges may act as waveguides for

the tsunami waves (Figure 8.7c).

When a tsunami reaches the shoaling continental

slope, its wave speed decreases and the

232


wave refracts to approach the coastline in a more

normal direction, just like any surface gravity

wave. Part of its energy can reflect from the shelf

and part can generate waves, which can be either

coastally trapped or reflected. Because its wavelengths

are very long, the wave steepness of

a tsunami is small. Thus shoaling tsunamis

behave as surging breakers, with little loss of

energy through breaking until reaching the

beach. Amplitudes of run-up from submarine

earthquakes can reach 10 to 30 m. Because of their

large run-up, tsunamis can flood large coastal

regions in a short time (half the period of the

wave, which is about half an hour or less).

Tsunami energy can be focused by mid-ocean

features and also by the natural resonance of

specific continental shelves and harbors. For

instance, Crescent City, California, is particularly

susceptible to large tsunamis due to

a combination of the offshore Mendocino Fracture

Zone that focuses tsunami energy as it

crosses the ocean, and the natural resonances

of the local continental shelf and the harbor

(Horrillo, Knight, & Kowalik, 2008).

8.4. INTERNAL GRAVITY WAVES

This section is a brief introduction to the

internal gravity waves, or internal waves, that

ride the internal stratification in the ocean, and

how they are affected by Earth’s rotation. The

ocean is stably stratified almost everywhere.

Therefore a water parcel that is displaced, say,

upward, encounters water of lower density

and falls back downward and vice versa. This

results in an oscillation, hence a wave. The

restoring force is the buoyancy force, which is

the product of gravity and the difference in

density between the displaced water parcel

and its neighbors at the same pressure. Internal

gravity waves are similar in this respect to the

surface gravity waves on the strong airesea

density interface. Because the stratification

within the ocean is much weaker than that

between the air and water, the restoring action

is weaker and the waves have much lower

frequencies than surface waves of comparable

wavelengths. For the same reason, water particles

can travel large distances up and down in

internal waves: amplitudes of tens of meters

are common for internal waves.

Internal waves are mostly generated by tides,

which interact with topography and generate

internal tides (baroclinic tides), and by the

wind, which stirs the mixed layer and generates

internal waves with frequencies close to the

inertial frequency (associated with Earth’s rotation).

Following Gill (1982), we take two

approaches to considering internal waves: (a)

waves on an interface between two layers of

different density and (b) waves in a continuously

stratified ocean. These two types of waves

have quite different behaviors.

8.4.1. Interfacial Internal Gravity

Waves

An interfacial internal wave is illustrated in

Figure 8.8. This kind of internal wave is strikingly

similar to a surface gravity wave. It propagates

horizontally and involves heaving up

and down of the sharp interface between the

two layers, whose densities are r 1 and r 2 . The

principal modification from surface gravity

Layer 1 ρ 1

Layer 2 ρ 2

Propagation direction

Surface convergence

(possible slick)

w

Wavelength L = 2π/k

FIGURE 8.8 Schematic of a simple interfacial internal

wave in a two-layer flow. Source: After Gill (1982).

U 1

U 2

INTERNAL GRAVITY WAVES 233

waves is that the density difference, Dr ¼ r 1 r 2 ,

between the two layers is much smaller than

the density difference between air and water.

The phase and group speeds of the interfacial

internal wave are like those of shallow water

surface waves (Eq. 8.5):

c 2 p zgDr r H 1hgH 1 (8.6)

where H 1 is the mean thickness of the upper

layer, r is the mean density and g 0 is called the

“reduced gravity.” It is assumed in Eq. (8.6)

that the upper layer (1) is much shallower than

the deeper layer (2); if they are of comparable

depth, then the factor H 1 becomes a more

complicated combination of both layer depths.

In Figure 8.8, the wave is propagating to the

right. The water at the node in the center of

the diagram (zero between the crest and the

trough) is moving downward. The horizontal

velocities are highest at the crests and troughs.

There is a convergence at the node behind the

crest; if the wave has very large amplitude, it

can produce a surface slick (Figure 8.9c).

An example of internal waves that are nearly

like interfacial waves is shown in Figure 8.9. The

temperature fluctuations are due to internal

waves heaving the thermocline up and down.

The amplitude is up to 8 m, even in this very

shallow water (15 m depth). Just to the north,

surface slicks parallel to the coastline are often

observed on calm days (Figure 8.9c); these are

due to internal waves similar to those at the

Mission Bay site. (Surface slicks are also caused

by convergent surface flow in Langmuir circulations,

Section 7.5.2, but these typically occur on

windy days when surface wave activity is high.)

8.4.2. Internal Gravity Waves in

a Continuously Stratified Ocean

Now consider waves within a continuously

stratified ocean (or atmosphere), ignoring the

upper and lower boundaries. Vertical stratification

is the most important external ocean

property for characterizing these waves. The

Brunt-Väisälä (buoyancy) frequency, N, introduced

in Section 3.5.6, is the maximum

frequency for internal gravity waves. The

maximum frequency is higher for higher stratification

(higher N). The wave periods range from

several minutes in the well-stratified upper

ocean, to hours in the weakly stratified deep

ocean. Waves at the Brunt-Väisälä frequency

propagate entirely horizontally, with water particles

moving exactly vertically with maximum

exposure to the stratification (Figure 8.10).

Because internal waves can have periods on

the order of hours, low frequency internal

gravity waves are influenced by Earth’s rotation

(Eq. 7.8). The lowest frequency waves are pure

inertial waves, whose frequency is equal to

the Coriolis parameter, f. These have particle

motions that are entirely in the horizontal plane,

with no vertical component that can feel the

vertical stratification. The full range of internal

wave frequencies, u, is

f u N (8.7)

Because f depends on latitude (0 at the equator

and maximum at the poles), the allowable range

of frequencies depends on latitude as well as on

stratification.

The complete dispersion relation for internal

waves in a continuously stratified flow is given

here, without derivation, in terms of horizontal

and vertical wavenumbers k, l, and m:

u 2 ¼ ðk2 þ l 2 ÞN 2 þ m 2 f 2

k 2 þ l 2 þ m 2 (8.8)

This has been simplified by assuming that N has

no variation and that f is constant (constant latitude).

Even if more complicated stratification is

included, Eq. (8.8) can still be a good approximation

to the local behavior of the internal

waves.

The internal wave frequencies from f to N are

set entirely by the angle of the wave vector with

234


FIGURE 8.9 Internal wave observations. (a) Temperature as a function of time and depth on June 16, 1997 at location

shown in (b) (Lerczak, personal communication, 2010). (b) Map of mooring location in water of 15 m depth west of Mission

Bay, California. Source: From Lerczak (2000). (c) Ocean surface west of Scripps Institution of Oceanography (map in b) on

a calm day; the bands are the surface expression of internal waves propagating toward shore. (Shaun Johnston, personal

communication, 2010).

the vertical (q in Figure 8.10). As the wave vector

tilts from horizontal toward the vertical, the

water particles feel less and less stratification,

and the frequency decreases until finally reaching

its lowest value, f. Manipulation of the

dispersion relation (8.8) (not derived here)

shows that the frequencies do not depend on

the actual wavenumber, only on the angle of

the wave vector with the horizontal. This differs

entirely from surface gravity waves and from

interfacial waves (Section 8.4.1).

The group velocity (c g ) of internal gravity

waves is exactly at right angles to the phase

velocity (c p ). Thus the energy propagation direction,

which is always given by the group

velocity, is in the direction that the particles

move. And finally, the group velocity for both

the highest frequency (N) and lowest frequency

INTERNAL GRAVITY WAVES 235

z (up)

Low frequency

K = (0,m) (m > 0)

ω = f c g =0

K = (k,m) (m > 0)

f < ω < N

θ

High frequency

K = (k,0)

ω = N c g =0

c g

f < ω < N

K = (k,m) (m < 0)

ω = f c g =0

K = (0,m) (m < 0)

Lighter

x (horizontal)

Denser

(f) internal waves is 0 in all directions (upward

and horizontally).

At near-inertial frequencies (close to f),

downward group velocity from the mixed layer

is accompanied by upward phase velocity, and

the particles move in clockwise ellipses that

are almost circular. Because the Coriolis parameter,

f, is 0 at the equator, internal waves of very

low frequency can be found in the equatorial

region, with periods of many days (10 days at

3 degrees latitude to infinite at the equator).

8.4.3. Internal Wave Generation and

Observations

c g

FIGURE 8.10 Schematic of properties of internal waves.

The direction of phase propagation is given by the wavevector

(k, m) (heavy arrows). The phase velocity (c p ) is in the

direction of the wavevector. The group velocity (c g ) is exactly

perpendicular to the wavevector (shorter, lighter arrows).

Internal waves within the water column

(other than the interfacial waves described in

Section 8.4.1) are primarily generated by winds

that generate disturbances in the surface mixed

layer and by the tides sloshing over bottom

topography. Internal waves then propagate

energy from the disturbances into the ocean

interior (e.g., Polton, Smith, MacKinnon, &

Tejada-Martinez, 2008). Nonlinear interactions

between the internal waves generated at many

different sites then spread the energy to internal

waves at other frequencies.

Observed waves are usually analyzed by

spectral analysis, including filters to remove

frequencies that are not characteristic of internal

waves (Section 6.5). Observed internal wave

spectra are so similar from one place to another

that it took several decades of work to begin to

delineate variations in the spectrum due to local

generation. The general form of the internal

wave spectrum was introduced by Garrett and

Munk (1972, 1975); their later modification is

referred to as the Garrett-Munk 79 spectrum

(Munk, 1981), and remains widely used. Much

of what is now known about internal wave

distributions and generation has arisen from

understanding the reasons for the nearly

universal (empirical) spectral shape and from

describing differences from this shape.

The energetic tides, which have very specific

frequencies dictated by the moon and sun orbits

(Section 8.6), produce internal waves when they

sweep over topography, if their frequency falls

between f and N. This means that the propagation

direction relative to the vertical of tidally

generated internal waves can be precisely predicted

because the direction is set, exactly, by

the wave frequency. Tidally generated internal

waves that propagate energy upward and

outward have been observed from the Hawaiian

ridge (Figure 8.11b).

Energy can pile up in internal waves, usually

as the waves propagate toward shallow water

near the coast, creating large localized disturbances

called solitary waves or solitons. Internal

solitons are associated with tides moving over

banks or straits. Internal solitary waves have

been observed in a number of locations; acoustic

backscattering from wave-generated turbulence

was used to produce the extraordinary images

from the continental shelf off Oregon in

Figure 8.11c (Moum et al., 2003).

236


(b)

0

100

0.06

0.05

Depth (m)

200

300

400

0.04

0.03

0.02

(u¢) 2 +(v¢) 2 (m 2 s −2 )

500

0.01

600

−150 −100 −50 0 50 100 150

Distance (km)

0

FIGURE 8.11 Internal wave observations. (a) Rotary spectra from a current meter at 55 m depth in the Mid-Atlantic bight:

bold is clockwise and thin is counterclockwise; the dashed curve is the modified Garrett-Munk spectrum. Source: From Levine

(2002). (b) Velocity variance (variability) observed along a section crossing the Hawaiian Ridge, which is located just below

the bottom of the figure at 0 km; the black rays are the (group velocity) paths expected for an internal wave with frequency

equal to the M 2 tide; distance (m) is from the center of the ridge. Source: From Cole, Rudnick, Hodges, & Martin (2009). This

figure can also be found in the color insert. (c) Breaking internal solitary wave, over the continental shelf off Oregon. The

image shows acoustic backscatter: reds indicate more scatter and are related to higher turbulence levels. Figure 8.11 can also

be found in the color insert. ÓAmerican Meteorological Society. Reprinted with permission. Source: From Moum et al. (2003).

TIDES 237

8.5. LARGE-SCALE CONTINENTAL

SHELF AND COASTAL-TRAPPED

WAVES

The physical boundary in the coastal ocean

permits a particular class of large-scale surface

gravity wave for which rotation is important.

These coastal-trapped waves have their highest

amplitude at or near the coast and decay away

toward the open ocean (see review in Brink,

1991). They have large length scales (tens to

hundreds of kilometers) and are subinertial

(frequencies lower than the inertial frequency).

The purest such coastal-trapped wave, for an

ideal ocean with a flat bottom and vertical sides,

is the Kelvin wave (Section 7.7.6).

The more general coastal-trapped (or topographic)

waves behave like Kelvin waves, but

are strongly modified by the side slope (continental

slope), and are more like Rossby waves

(Section 7.7.3). These topographic Rossby waves

always propagate with shallow water to the right

in the Northern Hemisphere (and to the left in

the Southern Hemisphere). Continental shelf waves

are similar to topographic Rossby waves, but

solved with a bottom configuration that includes

the continental shelf and slope and a flat deep

ocean bottom offshore of the slope.

8.6. TIDES

The once or twice daily rise and fall of the

tides and their long-term variations are the

most predictable of all oceanographic

phenomena. Water piles up against the coast

during the flood tide, and falls away during the

ebb tide. As the waves associated with internal

tides in the deep ocean run into seamounts,

ridges, or the ocean sides, they break and become

turbulent, becoming the major source of dissipation

in the deep ocean (Section 8.4). In this section

we present only a brief introduction to tides.

More complete pedagogical discussions of this

important topic are available in many sources

such as Pugh (1987), Komar (1998), Open

University (1999), Stewart (2008), and Garrison

(2001). Hendershott’s introductory lecture in

Balmforth, Llewellyn-Smith, Hendershott, and

Garrett (2005) is especially helpful.

8.6.1. The Equilibrium Tide

The moon and sun exert gravitational forces

on Earth, including its thin shell of ocean. In

1687, Sir Isaac Newton published the expression

for the gravitational attractive force between

two bodies:

F ¼ G mM

r 2

(8.9a)

Here F is the gravitational attractive force

directed along the line separating the two bodies

in Newtons; r is the distance between them in

meters; m is the mass of one body (e.g., the

moon); M is the mass of the other body (e.g.,

the Earth), and G is Newton’s universal gravitational

constant (6.67 10 11 Nm 2 kg 2 ).

The equilibrium tide is the shape that the ocean

would take due to the gravitational attraction of

the moon or sun on the water if Earth were

a pure water-covered planet, with no continents

and no topography. The tide-generating force on

the ocean is the difference between (1) the gravitational

attraction of the moon (or sun) at

Earth’s center of mass and (2) the gravitational

attraction of the moon (and sun) on the ocean.

As shown in Fig 8.12a, this is the difference

between the force (F C ) between the moon and

Earth centers, and the force between the moon

and either the far side of Earth (F A at the “antipodal

point”) or the near side of Earth (F S at the

“sublunar point”). (These statements can also

refer to the sun.) In Figure 8.12a, the force differences

are T A ¼ F A F C or T S ¼ F S F C . We see

right away, without even writing down the

expressions for the forces, that T A and T S are

the same size and pointed in opposite directions

from each other. This results in a bulge of ocean

toward the moon on the sublunar (near) side

238


(a)

T A

(b)

Antipodal point

F A F c

T s

F s

a

Earth (radius a)

Sublunar point

R A = r+a

Equilibrium tide force balances

R s = r-a

r (center to center)

(c)

moon

moon or sun

T moon

T moon

T sun

T sun

Spring tide alignment

moon

sun

Neap tide alignment

FIGURE 8.12 The equilibrium tide. (a) Tide-generating force due to the moon or sun. (b) Earth-moon-sun alignment

during spring tide, which also includes the case when the moon is opposite the sun. Source: After NOAA (2008).

(c) Alignment during neap tide. In (a), the F’s are the net gravitational acceleration at the antipodal, center, and sublunar

points, and the T’s are the net tidal gravitational accelerations.

sun

and a bulge away from the moon on the antipodal

(far) side. As Earth rotates, there are therefore

two bulges and hence two high tides per day.

Derivation of the shape of the equilibrium

tide for every point on Earth is complicated

(Komar, 1998; Open University, 1999; Stewart,

2008), beyond our scope, and is not included

here. But we can derive the simpler expressions

for the maximum tidal amplitudes at the

sublunar and antipodal points. Writing down

the expressions for the forces using Eq. (8.9a),

the lunar gravitational acceleration (force per

unit mass) at a point in the ocean that lies

a distance R from the center of mass of the

moon is toward the moon and of size Gm/R 2

where m is the mass of the moon. Meanwhile,

at the center of Earth, the acceleration of the

center of mass of Earth (in its orbit about the

center of mass of the Earth-moon system) is

toward the moon and of size Gm/r 2 where r is

the distance from Earth’s center to the moon’s

center. The distance of the moon to the sublunar

point is R ¼ R s ¼ r a, where a is Earth’s radius.

The tidal acceleration of a fluid parcel at the

sublunar point is toward the moon:

T S ¼ Gm

R S

2

Gm

r 2 w2Gma r 3

(8.9b)

(A Taylor series expansion assuming a << r

yields this approximate result.) The distance

of the moon to the antipodal point is R ¼ R A ¼

r þ a. The tidal acceleration of a fluid parcel at

the antipodal point is

T A ¼ Gm

R A

2

Gm

r 2 w

2Gma

r 3

(8.9c)

which is directed away from the moon. These

accelerations are illustrated in Figure 8.12a.

TIDES 239

Thus, the tide-generating force at Earth’s

surface has a component toward the moon on

the side of Earth facing the moon, and also

away from the moon on the other side of Earth.

This is simply because the force at the ocean’s

surface on the side facing the moon is greater

than at the center of Earth, while the force on

the ocean on the side opposite the moon is less

than at the center of Earth. 1

Moreover, the tide-generating force decreases

as the inverse third power of the distance to

the tide-generating body, even though the

Newtonian attraction decreases like the inverse

square of the distance (Eq. 8.9a). This is because

the differences taken in Eq. (8.9b, c) are between

two large terms that nearly cancel.

Earth, and observers fixed to it, rotate under

the equilibrium tidal potential so that an Earthbound

observer sees a high equilibrium tide

when the tide-generating body is at its highest

elevation above the horizon and another of equal

magnitude when the tide-generating body is at

its lowest elevation below the horizon. When

the sun is the tide-generating body, the two

maximum equilibrium tides are 12 hours apart.

However, when the moon is the tide-generating

body, the interval between them is about 12

hours, 25 minutes, because the moon orbits Earth

in the same direction as Earth’s rotation. Therefore,

high and low tides due to the moon occur

slightly less often than twice per day, and the

time of high and low tide shifts with each day.

In wave language, these hypothetical tides have

frequencies of two cycles per solar and lunar

day, respectively, and so are called semidiurnal.

When the tide-generating body is in Earth’s

equatorial plane, then the two high tides are of

equal size. However, the sun is in Earth’s equatorial

plane only twice per year and the moon is

in this plane only twice per month. As a result,

the size of the two high tides each day at a given

point on Earth differs. This is called the daily

inequality (also called diurnal inequality). The

solar daily inequality is greatest twice per year,

at the solstices, when the sun is at its greatest

distance from Earth’s equatorial plane; the solar

daily inequality vanishes at the intervening

equinoxes. The lunar daily inequality varies

similarly over the tropical month, which is

defined by successive northward passages of

the moon across Earth’s equatorial plane. In

wave language, the occurrence of a daily

inequality may be viewed as the constructive

and destructive interference of a semidiurnal

tide (two cycles per day) with a diurnal tide

(one cycle per day). Because the solar daily

inequality vanishes twice per year, there are

two solar diurnal tidal components that interfere

destructively twice per year.

The lunar equilibrium tide amplitude is

about 20 cm, which is much smaller than the

actual tides observed along many coasts and

harbors; the difference is due to the influence

of coastal boundaries (Section 8.6.2). The sun is

much farther away from Earth than the moon,

so even though it has much greater mass than

the moon, the solar tidal forcing is only about

half that of the lunar tidal forcing. (However,

the solar tide response for given locations can

be larger than the lunar tide response.) When

1 An equivalent derivation of the equilibrium tide-generating force is in terms of the centrifugal force associated with

the rotation of Earth around the center of mass of the moon-Earth system (the barycenter, which is located about

4670 km from Earth’s center hence inside the Earth). The gravitational acceleration between the moon and Earth centers,

F C , is balanced by the centrifugal acceleration of this rotation around the barycenter (F cf = F C ). The centrifugal acceleration

around the barycenter is the same at every point on Earth because Earth is a rigid body (e.g. M. Hendershott’s

Lecture 1 in Balmforth et al., 2005). In an Earth-centered coordinate system, the tide-generating acceleration is then the sum

of this invariant centrifugal acceleration and the gravitational acceleration of the ocean towards the moon, which depends

on nearness to the moon. That is, T S =F cf þ F S = F C þ F S and T A =F cf þ F A = F C þ F A , which are identical to the

expressions given above.

240


the Earth, moon, and sun are aligned

(Figure 8.12b), and also when the moon is

exactly opposite the sun, the lunar and solar

tides reinforce each other, producing very large

high tides. (This alignment is called syzygy.)

These are called spring tides (two per month).

When the moon is perpendicular to the Earthsun

axis, the lunar and solar tides do not reinforce

each other, and the two periods of smallest

high tides of the month occur; these are called

neap tides. This aspect of the semi-monthly variation

in tidal amplitude is sometimes called the

fortnightly tide (one fortnight equals two

weeks).

The orbit of the moon around Earth is elliptical

rather than circular. Therefore, once a month the

moon is closest to Earth (perigee) and the lunar

tidal range is highest. Once a month, the moon

is farthest from Earth (apogee) and the lunar tidal

range is smallest. Similarly, Earth’s orbit around

the sun is elliptical. When the sun is closest to

Earth (perihelion, which occurs around January

2), the spring tide is largest (perigean spring

tide). When the sun is farthest from Earth

(aphelion, which occurs around July 2), the tidal

range is reduced.

The plane of Earth’s orbit around the sun is

called the ecliptic. Earth’s equatorial plane is

tilted about 23 26’ to the ecliptic. The plane of

the moon’s orbit is tilted about 5 degrees to

the ecliptic; this tilt is referred to as the moon’s

declination. Thus the maximum declination of

the moon is about 28 26’ and its minimum

declination is about 18 26’. The moon’s orbit

precesses with a period of 18.6 years; during

this period, the moon’s declination shifts from

its minimum to its maximum. Therefore the

size of the lunar daily inequality varies with

a period of 18.6 years.

8.6.2. Dynamic Tides

Because the Earth, sun, and moon motions

are well known, the tide-generating force is

very precisely known. Given the regularity

and predictability of the forcing, why do tides

at any given location on the coastline differ

from those farther along the coast, why are

actual tides at coastlines sometimes much larger

than the equilibrium tide, and why are some

locations dominated by semidiurnal tides while

others are dominated by diurnal tides? The

continents block the free propagation of the

equilibrium tide westward as the Earth turns.

The result is a complex pattern of tides that

move around each of the ocean basins. Depending

on how each basin responds to each particular

frequency in the tide-generating force, the

tide that results at any given location is unique,

being a function of the lunar and solar tidal

forcing and the basin and coastline geometry.

The frequency of each component is determined

astronomically. The relative amplitudes of the

components depend on location.

The primary tidal frequencies are semidiurnal

(twice a day due mainly to the lunar

tide) and diurnal (once a day). In some locations

there is almost no semidiurnal component,

while in other locations there may be almost

no diurnal. The tide is usually expressed in

terms of tidal constituents. The principal constituents,

in order of amplitude, are M 2 (lunar semidiurnal),

K 1 (luni-solar declinational, diurnal),

S 2 (solar semi-diurnal), O 1 (lunar diurnal,

accounting for the moon’s declination), N 2

(lunar elliptical, semidiurnal), P 1 (solar diurnal,

accounting for the sun’s declination), K 2 (lunisolar

declinational, semidiurnal), and a number

of other semidiurnal, diurnal, fortnightly, and

longer period frequencies. Tables of the constituents,

their equilibrium tide amplitude, and their

periods are given in numerous textbooks (e.g.,

Defant, 1961; Komar, 1998; Stewart, 2008).

Computer software is readily available for

prediction of tides based on the tidal constituents

and an observed tidal record. National

agencies provide such predictions as a service

(e.g., NOAA CO-OPS, 2010).

The spring and neap tides are illustrated

in Figure 8.13 using a two-month record at

TIDES 241

Tidal Height (m relative to MLLW)

2

1

0

NOAA/NOS/CO−OPS La Jolla Water Level

New moon

Full moon

FIGURE 8.13 Tides at La Jolla, California.

Data from NOAA CO-OPS (2010).

5 10 15 20 25 30

June, 2010

5 10 15 20 25 30

July, 2010

Los Angeles. The small circles show the times of

the full and new moons, which coincide with the

spring (highest) tides. The neap tides are the lulls

in between. The mixture of semidiurnal and

diurnal tides produces the two separate envelopes

of tides e the lower high tides near the

center and the higher high tides that show that

spring-neap cycles. Plots from other locations

can look quite different depending on the relative

amplitudes of the diurnal and semidiurnal

components.

A global map of the M 2 tide is shown in

Figure 8.14. The curves in Figure 8.14a are cotidal

lines, which indicate the time of passage of high

tide (measured in terms of phase from 0 to 360 ).

Where the cotidal lines intersect, the amplitude

is zero (Figure 8.14b). These special points are

called amphidromes.

In addition to possibly large changes once or

twice daily in the volume of water at the coast,

tides can also promote vertical mixing and

break down the stratification of the water. Water

moving in and out over a subsurface bottom

feature is referred to as “tidal flushing.”

Georges Bank provides a nice example of tidal

mixing effects. The M 2 tide impinging on the

bank creates clockwise circulation, especially

a jet along its northern side that can reach

100 cm/sec (Chen & Beardsley, 2002 and

Figure 8.15a). A “tidal mixing front” appears

at the edge of Georges Bank when the water

column is stratified, separating well-mixed

water over the shallow bank from stratified

water offshore (Figure 8.15b). Tidal mixing

moves colder, nutrient-rich water onto the

bank from greater depths off the bank. The

result is very high productivity over Georges

Bank, as observed by satellite surface color

images indicating high chlorophyll content

(Figure 8.15c).

In ice-forming regions at high latitudes, there

is often a subsurface temperature maximum

beneath the much fresher, colder (freezing)

surface layer. Tidal mixing over banks in such

places can create polynyas (open water) by mixing

the subsurface warm water to the surface

where it can melt the sea ice (Section 3.9; Figures

3.12, 10.29, and 12.23).

In marginal seas, gulfs, and estuaries, the

source of tidal forcing differs from the open

242


FIGURE 8.14 Maps of (a) cotidal (phase) lines ( ) and (b) tidal amplitude (cm) for the M 2 tide (lunar semidiurnal). Source:

From Ray (1999).

ocean. Tidal currents from the open ocean

impinge on the coastal region, forcing motions

in the coastal ocean and estuaries. These are

called co-oscillation tides (Bowden, 1983). Cooscillation

tides can have non-trivial amplitudes

if there is a resonance between the open ocean

tide and the natural frequency of the basin. If

there is a resonance, the highest amplitude of

ESTUARIES 243

(a)

(c)

(b)

Depth (m)

0

50

100

150

11

8

6

7

9

10

10

T ( C)

9

10

11

12

13

FIGURE 8.15 Tidal effects on Georges Bank. (a) Schematic circulation and (b) summer temperature ( C) structure. TMF ¼

Tidal Mixing Front. SBF ¼ Shelf Break Zone. Source: From Hu et al. (2008). (c) Chlorophyll a concentration (mg/m 3 )on

October 8, 1997, from the SeaWiFS satellite. Figure 8.15c can also be found in the color insert. Source: From Sosik (2003).

the tide is at the head of the estuary or gulf. The

Gulf of Maine and Bay of Fundy (location in

Figure 8.15a) have a maximum tidal amplitude

of 15 m; this is a strong co-oscillation tide, in

resonance with the M 2 tide at a period of 13.3

hours (Garrett, 1972).

8.7. WATER PROPERTIES IN

COASTAL REGIONS: RIVER

RUNOFF

In this brief section, which appears only in

the online supplementary materials for the

textbook (Section S8.7), we discuss and illustrate

the impact of river runoff on coastal conditions,

and the importance of river runoff in global

freshwater budgets for the ocean.

8.8. ESTUARIES

An estuary, in the strictest definition, is

formed at the mouth of a river, where the river

meets the sea (Dyer, 1997). The defining characteristic

of estuarine circulation is that inflow is

denser than outflow, which is diluted relative

to the inflow. Estuaries are classified in terms

244


of their shape and their stratification. Discussion

of estuarine stratification, circulation, and flushing

times appears only in the online supplementary

materials for the textbook in Section S8.8.

8.9. CORAL REEFS

The physical oceanography of coral reefs was

of particular interest to George Pickard, the original

author of this text. He published several

papers and a book on the Great Barrier Reef in

1977 (Pickard, Donguy, Henin, & Rougerie,

1977). We retain this section of the coastal oceanography

chapter, but it has not been updated

from the 5th edition. The material appears

only in the online supplementary materials for

the textbook in Section S8.9.

8.10. ADJACENT SEAS

This is a lengthy description of the circulation

and water properties in a number of adjacent

(marginal) seas of the Atlantic Ocean (Mediterranean,

Black, Baltic, and North Seas), the Pacific

Ocean (South China, East China, Yellow, and

Japan or East Seas; Okhotsk, and Bering Seas),

and the Indian Ocean (Red Sea and Persian

Gulf). This material appears only in the online

supplementary materials for the textbook in

Section S8.10.

C H A P T E R

9

Atlantic Ocean

9.1. INTRODUCTION AND

OVERVIEW

The Atlantic Ocean is a long, narrow ocean

basin bisected by the Mid-Atlantic Ridge

(MAR) (Figure 2.9). Wind-driven gyres and the

wind-driven tropical circulation dominate

transports in the upper ocean (Figure 9.1). The

gyres and their western boundary currents

include the anticyclonic subtropical gyres of

the North Atlantic (Gulf Stream and North

Atlantic Current) and South Atlantic (Brazil

Current), and the cyclonic subpolar gyre of the

northern North Atlantic (East Greenland

Current and Labrador Current). The subtropical

gyres include eastern boundary current

upwelling systems: the Canary Current system

in the North Atlantic and Benguela Current

System (BCS) in the South Atlantic. The tropical

circulation is predominantly zonal (east-west),

including the North Equatorial Countercurrent

and the South Equatorial Current, and has

a low-latitude western boundary current (North

Brazil Current; NBC).

Conversion of upper ocean waters to denser

intermediate and deep waters (meridional overturning

circulation or thermohaline circulation) in

the northern North Atlantic is associated with

a deep circulation, including Deep Western

Boundary Currents (DWBCs; Section 7.10.3).

Most of the final conversion from the surface

to the deeper layers occurs within the Labrador

Sea and Nordic Seas (Chapter 12). This conversion

also affects the Atlantic’s upper ocean

circulation: it increases the northward transport

in the North Atlantic’s Gulf Stream and North

Atlantic Current by approximately 10% and

provides a connection of tropical and subtropical

waters to the subpolar North Atlantic. This

overturning circulation results in net northward

heat transport through all latitudes of the

Atlantic, as it draws warm, saline surface waters

north and sends dense, cold, fresher waters

south at depth. In the South Atlantic, this

reverses the usual poleward direction of

subtropical heat transport found in all other

subtropical regions (Section 5.6).

In the south, the Atlantic connects with the

other oceans through the Southern Ocean

(Chapter 13). As it enters the South Atlantic

from the Drake Passage, the Subantarctic Front

(SAF) of the Antarctic Circumpolar Current

(ACC) makes an important northward excursion

along the coast of South America as the

Malvinas (or Falkland) Current and then loops

partially back southward to begin a long, slow,

southward drift as it moves eastward to the

Indian Ocean and beyond to the Pacific. Warm

surface water from the Indian Ocean enters the

South Atlantic where the Agulhas Current

rounds the southern tip of Africa. Most of the

Agulhas retroflects back to the Indian Ocean


245



246

9. ATLANTIC OCEAN

(a)

100˚W 90˚ 80˚ 70˚ 60˚ 50˚ 40˚ 30˚ 20˚ 10˚ 0˚ 10˚E

70˚

70˚

Nordic Seas

60˚

Hudson Bay

Davis

Strait

Labrador C.

WGC

Labrador

Sea

East Greenland C.

Irminger

Sea

Irminger C.

Charlie Gibbs FZ

Denmark

Strait

Ridge

Reykjanes

Iceland Basin

EGC

NIC

Rockall Plateau

Iceland-Faroe

Front

Rockall Trough

Norwegian

Atlantic C.

North Sea

60˚

50˚

40˚

30˚

20˚

10˚

0˚

G. Mexico

LC

Yucatan C.

Cape

Hatteras

Florida C.

Gulf Stream

Antilles C.

Slopewater C.

Caribbean C.

N. Recirc. Gyre

Bermuda

NBC

Rings

Grand

Banks

North Atlantic C.

NAC

North Equatorial Current

Vema FZ

North Brazil C.

Mann

Eddy

Azores

Azores Current

North Equatorial Countercurrent

Canary Isl.

Guinea

Dome

N. South Equatorial Current

Canary

Current

System

Bay of

Biscay

Portugal C.

System

Strait of

Gibraltar

Guinea Current

C. South Equatorial Current EUC

50˚

40˚

30˚

20˚

10˚

0˚

-10˚ -10˚

100˚W 90˚ 80˚ 70˚ 60˚ 50˚ 40˚ 30˚ 20˚ 10˚ 0˚ 10˚E

Angola

Dome

-5000 -4000 -3000 -2000 -1000 0

FIGURE 9.1 Atlantic Ocean surface circulation schematics. (a) North Atlantic and (b) South Atlantic; the eastward EUC

along the equator just below the surface layer is also shown (gray dashed).

INTRODUCTION AND OVERVIEW 247

(b) 80˚W 70˚ 60˚ 50˚ 40˚ 30˚ 20˚ 10˚ 0˚ 10˚ 20˚ 30˚E

10˚N

10˚N

Guinea Current

0˚

0˚

10˚

20˚

30˚

40˚

50˚

60˚

Drake Passage

North Brazil C.


N. South Equatorial Current

C. South Equatorial Current

EUC

Subantarctic

Front

Polar Front

ACC

Malvinas Current

Malvinas

Brazil Current

Falkland

Plateau

Zapiola Rise

Brazil

Basin

Vitoria-Trindade

Seamounts

Vema Rio Grande

Channel Rise

Argentine Basin

S. Georgia

Scotia Sea

Southern Boundary

Weddell Gyre

Weddell Sea

South Equatorial Countercurrent

Angola Basin

South Equatorial Current

South Atlantic Current

Subantarctic Front

Polar Front

Southern ACC

Front

Namib Col

Walvis

Ridge

Angola

Dome

Cape

Basin

Angola

Current

Benguela

Current

System.

Cape

Agulhas

Agulhas

Rings

Agulhas C.

Antarctic

Circumpolar

Current

10˚

20˚

30˚

40˚

50˚

60˚

70˚S

80˚W 70˚ 60˚ 50˚ 40˚ 30˚ 20˚ 10˚ 0˚ 10˚

20˚

70˚S

30˚E

FIGURE 9.1

(Continued).

-5000 -4000 -3000 -2000 -1000 0

248

9. ATLANTIC OCEAN

(Chapter 11), but the process sheds large eddies

of Indian Ocean water that move northwestward

into the Atlantic. A small portion of the

Agulhas waters also enters the South Atlantic’s

Benguela Current. Dense bottom waters from

the Antarctic enter the Atlantic from the

Weddell Sea.

In the north, the Atlantic connects with the

Nordic Seas and Arctic Ocean (Chapter 12),

which are separated topographically from the

North Atlantic by the ridge running from

Greenland to Iceland and then from Iceland to

the Faroe and Shetland Islands. Northward

flow from the Atlantic to the Nordic Seas feeds

into the Norwegian Atlantic Current along the

coast of Norway. Southward flow back into

the Atlantic occurs in the fresh surface layer

in the East Greenland Current (EGC) and also

through Davis Strait into the Labrador Sea,

and as dense subsurface overflows over each

of the three channels in the Greenland-Shetland

ridge. These overflows form the dense deep

waters of the North Atlantic and the deep

part of the Atlantic’s branch of the global overturning

circulation. (The other branch is associated

with dense water production in the

Antarctic.)

The marginal seas of the North Atlantic also

include important sites for water mass mixing

and conversion. The subtropical western

boundary current flows through the Intra-

American Seas (Caribbean Sea and Gulf of

Mexico) before emerging back into the North

Atlantic. The Mediterranean Sea has a series of

nearly separated sub-basins (each with characteristic

water mass formation and circulation)

and its own marginal sea, the Black Sea. Net

evaporation in the Mediterranean contributes

about one-third of the observed salinity difference

between the Atlantic and Pacific Oceans.

The Mediterranean’s dense water re-enters the

North Atlantic from the Strait of Gibraltar. In

the northwest, the Labrador Sea (which is

more of a large embayment than a marginal

sea) is the site of intermediate water formation

that contributes to the meridional overturning

circulation. Baffin Bay, to its north, connects

the Atlantic to the Arctic Ocean west of Greenland,

and has its own internal water mass

formation process. In the northeast are the

shallow intra-European shelf seas, the North

Sea, and the Baltic Sea. These marginal seas

are described in detail in Chapter S8 of the

online supplement located at http://booksite.

academicpress.com/DPO/; S denotes supplemental

material.

Water masses of the upper ocean in the

Atlantic are similar to those found in the

wind-driven gyres of the other oceans,

including those associated with thermocline

ventilation (Central Waters and Subtropical

Underwater; STUW), and those associated

with the strong currents (Subtropical Mode

Water; STMW). The North Atlantic Current of

the subpolar North Atlantic has its own mode

water d the Subpolar Mode Water (SPMW).

The northern North Atlantic and its adjacent

seas produce new deep water (North Atlantic

Deep Water; NADW) for the global ocean.

The local, convective sources are the Labrador,

Mediterranean, and Nordic Seas. Because of

the local deep water sources, the deep waters

of the northern North Atlantic are relatively

young, measured in decades, in contrast to

the deep and bottom waters of the North

Pacific, which are hundreds of years old

(Chapter 10).

In this text, the Atlantic is presented before

the Pacific and Indian Oceans because it has

been historically central for development of

ideas about the general circulation and water

mass formation. From a pedagogical point of

view, it might be more advantageous to present

the North Pacific circulation first (Chapter 10),

since the wind-driven subtropical and subpolar

circulations dominate in the North Pacific, while

the North Atlantic’s upper ocean circulation

also includes a significant inflow to the deep

overturn of the meridional overturning circulation

(MOC). In both oceans the wind-driven

FORCING 249

subtropical circulation transports 30 to more

than 140 Sv (depending on location). Whereas

15 to 20 Sv of water weave through the upper

North Atlantic’s circulation as part of the

MOC to form NADW, less than 2 Sv traverse

the North Pacific, ultimately forming the analogous

North Pacific Intermediate Water (Chapters

10 and 14).

Climate variability in the Atlantic is vigorous

(Section 9.9 and online supplementary Chapter

S15). Much of the quasi-decadal variability in

the North Atlantic is associated with the North

Atlantic Oscillation (NAO), which is linked to

the Arctic Oscillation. Effects of the NAO at

the sea surface include variability in the westerly

winds, airesea buoyancy fluxes, and

surface ocean properties. The NAO affects the

subtropical and subpolar circulations and water

mass formation rates and properties. Longer

term variation in Atlantic Ocean circulation at

centennial to millennial scales, called the

Atlantic Multidecadal Oscillation (AMO), has

been described, and is important in understanding

possible anthropogenic climate change

in the Atlantic Ocean. Tropical climate modes

intrinsic to the Atlantic Ocean are also observed,

and are separate from the El Niño-Southern

Oscillation (ENSO) of the Pacific Ocean, whose

effects also intrude into the Atlantic sector. In

the south, the Southern Annular Mode (SAM)

has a major center of action in the Weddell Sea

and South Atlantic sector of the Southern Ocean.

9.2. FORCING

The long-term mean external forcing for the

Atlantic’s general circulation is described in

this section. Seasonal effects are mostly not

covered. Some of the interannual to decadal

variations in forcing are discussed in the online

supplementary chapter on climate variability

(Chapter S15).

9.2.1. Wind Forcing

Wind stress drives ocean circulation via frictional

Ekman transport in the surface layer. 1

The surface layer’s convergences and divergences

then drive interior ocean circulation

(Chapter 7). The annual mean and seasonal

mean wind stress is shown in Figure 5.16

(global) with supplementary Atlantic-only

maps available online (Figure S9.3). The eastwest

(zonal) part of the wind field includes

mainly westerly winds north of 30 N and south

of 30 S, and easterly trade winds in the region

between. Meridional components notably

include equatorward winds along the eastern

boundaries: along northern Africa from the

Strait of Gibraltar to about 10 N, and along

southern Africa up to about 10 S. These largescale,

longshore winds force the Canary and

Benguela Current Systems.

The Ekman transport divergence (upwelling)

and convergence (downwelling) are represented

by the wind stress curl of Figures 5.16d and S9.3a

through Eqs. (7.21) and (7.44). Downwelling

regions fill the subtropics and upwelling regions

occur in the northern North Atlantic and in the

Southern Ocean south of about 50 S.

The Sverdrup transport (Section 7.8) is shown

in Figures 5.17 and S9.3b. The subtropical gyres

are the regions of equatorward interior flow

closed by poleward western boundary currents.

Based on the location of southward Sverdrup

transport, the North Atlantic’s subtropical gyre

extends northward to 50 Nto52 N, the latitude

of the UK. The northward western boundary

currents for this gyre include the Gulf Stream

System and the North Atlantic Current east of

Newfoundland. The South Atlantic’s subtropical

gyre extends well south of Africa. The

southward western boundary current for this

circulation is the Brazil Current.

The maximum Sverdrup transport predicted

for the Gulf Stream from these National Centers

1 Except at the equator, where the frictional layer transport is directly downwind.

250

9. ATLANTIC OCEAN

for Environmental Prediction (NCEP) mean

winds is about 20 Sv. The NCEP winds are

known to be too weak (Taylor, 2000); the

Sverdrup transport is likely to be more like

30e50 Sv. This is much less than the maximum

Gulf Stream transport of more than 140 Sv,

arising from its recirculation gyres. In the South

Atlantic, the Brazil Current Sverdrup transport

at 30 S (where Africa forms an eastern boundary)

is a comparable 25 Sv. Just south of the

tip of Africa (Cape Agulhas), the Sverdrup transport

at the South American coast jumps to more

than 85 Sv because the west-east integration

includes the full width of the Indian Ocean,

from the coast of Australia/Tasmania westward

to South America. The actual circulation does

not include an interior zonal jet at 35 Sacross

the South Atlantic; instead, the Agulhas, which

would feed such a jet, retroflects (turns abruptly

eastward) and creates eddies that propagate into

the South Atlantic. The observed Brazil Current

transport does jump to higher values south of

34 S, but this appears to be associated with local

recirculation, as in the Gulf Stream (Section 9.5).

9.2.2. Buoyancy Forcing

Buoyancy forcing is the sum of heat and

freshwater airesea fluxes (Figures 5.4a, 5.12,

5.15 and online supplementary Figure S9.4).

The Atlantic Ocean has two of the largest annual

mean heat/buoyancy loss regions on the globe:

in the Gulf Stream where it separates from the

North American coast at 35e38 N(>200 W/m 2 )

and in the Nordic Seas (>100 W/m 2 ). Both

are associated with poleward transport of

warm water that is cooled by the atmosphere,

including large latent (evaporative) heat loss.

Similarly, in the South Atlantic, the Brazil

Current and Agulhas retroflection are regions

of heat loss (>100 W/m 2 ). Net heat gain occurs

in the tropics, with the highest gain (greater

than 100 W/m 2 ) along the equator. Heat is also

gained in narrow ribbons along the coasts, associated

with the upwelling systems.

Net evaporation minus precipitation (E P)

minus runoff for the Atlantic shows the typical

large subtropical net evaporation regions

centered at 10e20 latitude on both sides of

the equator, flanking the tropical net precipitation

region associated with the Intertropical

Convergence Zone (ITZC). Net precipitation/

runoff is found in the subpolar North Atlantic,

especially around the continental margins and

in the adjacent seas (as runoff). E P for the

Atlantic is tipped toward net evaporation

compared with the Pacific Ocean, so its mean

salinity is higher than in the Pacific. The higher

overall salinity of the Atlantic is due to larger

evaporation throughout the subtropics.

Airesea buoyancy flux is dominated by heat

flux with a smaller contribution from freshwater

flux (online supplementary Figures S5.8 and

S9.4). Net evaporation in the subtropics enlarges

the subtropical buoyancy loss regions to cover

the full gyre in both the North and South

Atlantic. Freshwater input from the Amazon,

Congo, and Orinoco Rivers is greater than

0.4 Sv, on the order of the largest components

of the global freshwater budget (Dai & Trenberth,

2002; Talley, 2008). Freshwater input in subpolar

coastal regions is also evident (Newfoundland

region, British Isles). Buoyancy loss, even within

the Mediterranean Sea where evaporation greatly

increases salinity, is nevertheless controlled

mainly by heat loss. (Evaporation is accompanied

by latent heat loss from the ocean.)

9.3. NORTH ATLANTIC

CIRCULATION

The surface circulation of the North Atlantic

(Figures 9.1 and 9.2a; Figure S9.1 and Tables

S9.1 and S9.2 in the online supplement) includes

an anticyclonic subtropical gyre and a cyclonic

subpolar gyre that stretches northward into the

Nordic Seas. Basics of the surface circulation

have been well known since the nineteenth

century (e.g., review in Peterson, Stramma, &

FIGURE 9.2

(1994).

Steric height (10 m 2 s 2 ) at (a) 0 dbar and (b) 500 dbar, adjusted to estimate the absolute geostrophic circulation. Source: From Reid

NORTH ATLANTIC CIRCULATION 251

252

9. ATLANTIC OCEAN

Kortum, 1996). By the mid-twentieth century,

volume transports had been estimated

(Sverdrup, Johnson, & Fleming, 1942), and the

modern picture of the surface circulation began

to emerge, with depiction of intense, narrow

western boundary currents and recirculations

(e.g., Iselin, 1936; Defant, 1961; Dietrich, 1963).

The North Atlantic’s subtropical gyre, like all

subtropical gyres, is asymmetric, with strong,

narrow western boundary currents and broad

southward flow throughout the central and

eastern subtropics. The subtropical western

boundary current is composed of two connected

portions: the Gulf Stream System south of about

40 N, and part of the North Atlantic Current

System east of Newfoundland and north of

40 N. The eastern boundary upwelling system

is called the Canary and Portugal Current System.

The westward flow on the equatorward side of

the gyre is the North Equatorial Current.

The cyclonic subpolar gyre is less asymmetric

and more strongly controlled by topography. 2

It has swift, narrow western boundary currents

along Greenland and Labrador (EGC and Labrador

Current) that are connected by the West

Greenland Current (WGC), which is on an

eastern boundary. The North Atlantic Current

(NAC) is the eastward flow on the southern

side of the subpolar region; branches of the

NAC flow northeastward toward the Nordic

Seas. At the sea surface, the cyclonic subpolar

gyre encompasses both the subpolar North

Atlantic and the Nordic Seas (Chapter 12).

Southward return flow from the Nordic Seas

occurs in the EGC.

The subtropical and subpolar surface circulations

are connected through the NAC, with net

northward transfer of upper ocean water

required by the MOC.

With increasing depth, the anticyclonic

subtropical gyre shrinks westward and northward

toward the Gulf Stream System. The

cyclonic subpolar gyre becomes closed south

of the Greenland-Faroe ridges. At depths below

about 1500 m, the “abyssal” circulation becomes

evident, with emergence of a DWBC that carries

the newly formed intermediate and deep waters

from the subpolar North Atlantic southward

toward the equator (Section 9.6). Below the

depth of the MAR, the circulation is confined

to various abyssal basins, but on average transports

the northern North Atlantic waters southward

and the bottom waters from the Southern

Ocean northward.

9.3.1. Subtropical Circulation

We start our detailed description of circulation

with the subtropical gyre (Figure 9.1 and

Figure S9.1 and Table S9.1, which are found

in the online supplement). The subtropical

western boundary current system consists of

both the Gulf Stream System (Section 9.3.2) and

the more northern NAC (Section 9.3.4). The

Canary and Portugal Current Systems are the

eastern boundary current system (Section 9.3.3).

9.3.2. Gulf Stream System

The Gulf Stream System consists of multiple

segments with different names depending on

location (and author). Nomenclature, therefore,

can be confusing (Stommel, 1965). 3 We will

follow Stommel’s definition, in which the Florida

Current refers to the western boundary current

through the constriction between Florida and

the Bahamas, and Gulf Stream refers to the

continuation of this boundary current north of

2 Topographic control is a greater factor in the subpolar region than in the subtropics, due to deep penetration of the currents

resulting from weaker vertical stratification and a larger Coriolis parameter.

3 “I often use the term Gulf Stream in a more general sense than that proposed by Iselin; and I do not speak of the Florida

Current as extending to Cape Hatteras, but restrict the use of this term to mean the current actually within the Florida

Straits. Unfortunately, the naming of things is more a matter of common usage than of good sense” (Stommel, 1965).


the Florida Straits and after it separates from the

western boundary at Cape Hatteras and flows

eastward out to sea. The phrase Gulf Stream

Extension may also be used to describe the

separated current, especially east of the New

England Seamounts.

The subtropical Gulf Stream System begins

where the North Equatorial Current, joined by

the northward low latitude western boundary

current, enters the Caribbean Sea through the

complex of the Antilles islands (Figures 9.1 and

9.3 and Figure S9.5 in the online supplementary

material). The maximum sill depth for currents

entering the Caribbean Sea is 1815 m at Anegada

Passage (Fratantoni, Zantopp, Johns, & Miller,

1997), reflected in nearly uniform properties

below sill depth (see Figure 9.7). The exit

sill depth through the Straits of Florida, described

in the following text, is much shallower at 640 m,

and limits the maximum density of waters that

can flow completely through the Intra-American

Seas. (Denser waters can flow northward east of

the Antilles.) Within the Caribbean, the upper

ocean circulation consists of the westward Caribbean

Current and a local wind-driven cyclonic

circulation in the Colombia Basin.

Net transport into the Caribbean is estimated

at 28.4 Sv (Johns, Townsend, Fratantoni, &

Wilson, 2002). Below sill depth, there is a

vigorous cyclonic circulation of about 15 Sv

that simply moves the deep waters around in

this isolated deep basin (Joyce, Hernandez-

Guerra, & Smethie, 2001). The Caribbean

Current forms into a western boundary current

along the Honduran coast, called the Cayman

Current, and then exits northward into the

Gulf of Mexico through the Yucatan Channel

as the Yucatan Current. Moored observations of

the Yucatan Current from 1999e2001 showed

a mean transport of 23 Sv and maximum surface

velocity in excess of 130 cm/sec, sometimes

reaching 300 cm/sec (Candela et al., 2003;

100°W 90°W 80°W 70°W 60°W

30˚N

30˚N

31.5 5.0

25˚

Gulf of Mexico

Yucatan Ch.

Staits of Florida

28.4

1.2

NW Providence Ch.

Atlantic Ocean

25˚

20˚

15˚

10˚N

Yucatan Basin

28.4

Cayman Basin

1.9

Windward

Psg.

Mona

Psg.

Greater Antilles

3.0 2.5

Venezuela Basin

21.4 18.4

Caribbean Sea

Colombia Basin

Anegada

Psg.

3.1

1.1

1.6

1.5

2.9

5.7

Lesser

Antilles

Grenada

Psg.

20˚

15˚

10˚N

100°W 90°W 80°W 70°W 60°W

FIGURE 9.3 Gulf Stream System formation region. Volume transports (Sv) through the Caribbean and Gulf of Mexico.

After Johns et al. (2002).

254

9. ATLANTIC OCEAN

Cetina et al., 2006) (online supplementary

Figure S9.6). The velocity structure is typical of

strong currents restricted to a narrow channel,

with a central core of flow and weak, flanking

countercurrents (opposite direction).

After entering the Gulf of Mexico, the

western boundary current, now named the

Loop Current, flows northward to the middle of

the Gulf and turns east toward the Straits of

Florida. Loops, characterized by high sea

surface temperature (SST), frequently pinch

off, forming anticyclonic eddies that propagate

westward, often ending their existence on the

shelf of the eastern Texas coast (Figure 9.4a).

From the Gulf of Mexico, the western

boundary current escapes into the North

Atlantic. It turns northward along the coast of

Florida and forms the Florida Current and the

Gulf Stream. A small part of the Gulf Stream

originates in the Antilles Current, which is

a highly variable, weak western boundary

current in the open ocean east of the Antilles,

Puerto Rico, Cuba, and the Bahamas (Rowe

et al., 2010).

The Florida Current/Gulf Stream is a narrow,

intense, northward flow. The Florida Current is

well monitored in the confined strait between

Florida and the Bahamas (Figure 9.5a and online

supplementary Figures S9.7 and S9.8). Maximum

surface velocities exceed 180 cm/sec,

concentrated in a 20 km band in the western

part of the channel over the continental slope.

The mean transport at 27 N is 32 Sv with

seasonal and interannual variability each of the

order 2 to 3 Sv; maximum seasonal transport

occurs in summer (Baringer & Larsen, 2001).

FIGURE 9.4 Sea surface temperature from the GOES satellite. (a) Gulf of Mexico showing the Loop Current beginning to

form an eddy. (b) Gulf Stream, showing meander at the Charleston Bump and downstream shingling. Black contours are

isobaths (100, 500, 700, 1000 m). This figure can also be found in the color insert. Source: From Legeckis, Brown and Chang (2002).


(a)

(b)

(c)

FIGURE 9.5 Gulf Stream velocity sections and transports. (a) Mean velocity of the Florida Current at the Straits of Florida

at 27 N. Source: From Leaman, Johns, and Rossby (1989). (b) Smoothed geostrophic velocity at Cape Hatteras. Source: From

Pickart and Smethie (1993). (c) Gulf Stream transport (Sv) at different longitudes; Cape Hatteras and the New England

Seamounts are indicated by hatching. Barotropic and baroclinic transports are indicated. Source: From Johns et al. (1995).

After the Gulf Stream emerges from the

Florida Straits, it remains a western boundary

current until leaving the coast at Cape Hatteras

(about 35 N, 75 30’W). This location is

referred to as the separation point. An SST

image (Figure 9.4b) shows the narrow

boundary current, with a quasi-permanent

meander at 32 N due to topography (“Charleston

Bump”; Bane & Dewar, 1988) and time dependent

“shingle” structures in which meanders

peel backward on the inshore side of the

current.

256

9. ATLANTIC OCEAN

This segment of the Gulf Stream System is the

prototype for western boundary currents,

informing simple theoretical models of the

“Gulf Stream” and other subtropical western

boundary currents (e.g., Section 7.8). The current

extends to the ocean bottom over the continental

slope while its typical width remains <100 km;

its volume transport increases to more than 90

Sv at the separation point (Leaman, Johns, &

Rossby, 1989), fed by westward flow inflow

from the Sargasso Sea, including the vigorous

recirculation gyre. The mean velocity section at

Cape Hatteras (Figure 9.5b) shows the concentrated

Gulf Stream and the southward flow

inshore and beneath it in the DWBC (Pickart &

Smethie, 1993).

East of separation at Cape Hatteras, the Gulf

Stream is one of the most powerful currents in

the world’s oceans in terms of volume transport

(up to 140 Sv), maximum velocity (up to 250 cm/

sec), average velocity (about 150 cm/sec), and

eddy variability. It reaches to the ocean bottom

with bottom velocities exceeding 2 cm/sec. It

remains a narrow (<120 km wide), but strongly

meandering current for hundreds of kilometers,

carrying a warm, saline core of surface water far

eastward into the North Atlantic (Figure 1.1a). Its

structure is asymmetric, with strongest surface

flow on the northern (western) side of the

current, shifting southward (eastward) with

depth. The current decays quickly to the north

of this core; temperature and salinity also change

rapidly here (Figure 9.7). This sharp transition is

often called the “cold wall” of the Gulf Stream.

The instantaneous Gulf Stream is far from

steady. Its meanders often become large enough

to pinch off into rings on both sides (Section

9.3.6). The envelope of its meandering paths is

illustrated by the positions of its cold wall

(Figure 9.6): it is narrowest at Cape Hatteras

and then spreads to about 300 km in width

downstream, which is 3 times wider than its

instantaneous width. Between Cape Hatteras

and about 69 W, the Gulf Stream envelope

widens but follows sloping bottom topography,

FIGURE 9.6 Gulf Stream northern edges every two days

from infrared surface temperature for (top) April to

December 1982, (middle) all of 1983, and (bottom) April

1982 to September 1984. The faint white curves are the mean

tracks. Source: From Cornillon (1986).

which perhaps constrains its meandering. East

of the New England Seamounts and 69 W,

large-scale meandering sets in. The envelope

width compares remarkably well with the historical

Franklin and Folger map of the Gulf Stream

location (Figure 1.1b from Richardson, 1980a).

The Gulf Stream transport increases rapidly

downstream, from about 60 Sv at separation to

more than 140 Sv at 65 W(Figure 9.5c). It then

loses water to the south, much of it prior to

reaching 50 W. Between its separation point

and to at least 55 W, there is mean westward

surface flow just south of the Gulf Stream,

referred to as the Gulf Stream recirculation.

With the Gulf Stream, this forms the recirculation

gyre (sometimes called the “Worthington

Gyre”). The total transport of the Gulf Stream

is many times larger in this recirculation region

than predicted by Sverdrup transport theory

(Section 7.8). The recirculation is likely driven

by the Gulf Stream’s instability, which forces

westward flow on its flanks, and inertial overshoot

of the separated current.


North of the Gulf Stream, the westward flow of

the Slope Water Current forms an elongated

cyclonic gyre with the Gulf Stream, called the

Northern Recirculation Gyre (Hogg, Pickart,

Hendry, & Smethie, Jr., 1986). Here the wind stress

curl drives upwelling. The westward current is

partly supplied from the Labrador Current.

At the sea surface, the Worthington Gyre

extends all along the Gulf Stream (Figure 9.1).

Its southward flow offshore of the Florida

Current turns eastward into the central western

North Atlantic at about 22e25 N. This is called

the Subtropical Countercurrent, and has an exact

analog in the North Pacific’s Kuroshio gyre

circulation (Section 10.3.1). The eastward flow

then bends back to join the westward flow

North Equatorial Current. The entire recirculation

and Subtropical Countercurrent form the

so-called “C-shape” of the surface gyre.

Even though it is losing water to the south,

part of the Gulf Stream continues eastward to

the Grand Banks of Newfoundland at 50 W.

Here a portion turns northward and re-forms

as a western boundary current east of Flemish

Cap, where it is called the North Atlantic Current

(Section 9.3.4). The remainder of the Gulf Stream

continues eastward and southward, splitting

into two branches, one at 42e43 N and one

farther south at 35 N, called the Azores Current.

The branch at 42 N passes north of the Azores

and weakens considerably to the east. The

Azores Current extends eastward toward the

Strait of Gibraltar where a small amount of

surface water flows into the Mediterranean

Sea. Other than this remarkably zonal jet, the

subtropical gyre primarily turns southward in

the central and eastern North Atlantic. These

flows turn westward and feed the North Equatorial

Current, 4 completing our anticyclonic circuit

of the subtropical gyre.

The separated Gulf Stream is mostly in

geostrophic balance and is vertically sheared, so

its isopycnals and isotherms slope up toward

the north, with a 300 to 500 m depth change across

the 150 km wide current (Figure 9.7 and online

supplementary Figure S9.9). It contains a warm

and salty core close to the surface, due to advection

from lower latitudes. The cold wall on the

north side of the Gulf Stream (which was tracked

in Figure 9.6) is a front that is less than 20 km wide.

9.3.3. Canary and Portugal

Current Systems

The subtropical gyre has a classic eastern

boundary upwelling regime: the Canary

Current System south of the Strait of Gibraltar

and the Portugal Current System north of the

Strait. These are separated by the eastward

Azores Current, which is associated with the

Mediterranean inflow (New, Jia, Coulibaly, &

Dengg, 2001). The eastern boundary currents

are associated with large-scale alongshore

winds that create offshore Ekman transport

(Figure 5.16 and online supplementary

Figure S9.3a; Sections 7.9 and 10.3.1).

The Canary Current System (Figure 9.8), along

the North African coast, is the more energetic,

better developed, and better studied of the two

systems (Mittelstaedt, 1991). The Canary Current

is the equatorward (southward) near-coastal

current. It is present year-round, but its termination

in the south is seasonally dependent.

Between 20 N and 23 N, the Canary Current

turns offshore to join the North Equatorial

Current. The upwelling-favorable winds from

Gibraltar to Cape Blanc are strongest in summer.

The equatorward winds are strongest offshore,

leading to positive wind stress curl in the Canary

Current region, which augments the upwelling

forcing. A poleward undercurrent flows along

the continental shelf beneath the Canary Current,

north of 25 N, centered at about 600 m depth,

with a mean speed of about 5 cm/sec.

4 The NEC is not actually “Equatorial,” because it is separated from the equator by the vigorous eastward flow of the North

Equatorial Countercurrent. Instead, the NEC is the equatorward side of the subtropical gyre.

258

Depth [m]

0

500

1000

1500

2000

2500

3000

3500

4000

4500

5000

5500

6000

6500

South America

4

8

2.

3.

Puerto Rico

1.8

.8 3 3.2 3.4

1.6

16

15

6 7 10

5

3.8 4

3.6

3.4

3

2.4

2.2

2

20

18

North America

34.76

South America

34.93 34.91

34.96

5

Puerto Rico

34.84

34.92

36.9

36.5 36.6

35.08 35.1

35.06

35.04

35.02

35

34.98

34.9

34.86

34.94

34.88

1.4

Pot.

6000

Temp.

Salinity

Oxygen

6500

15°N 20°N 25°N 30°N 35°N 40°N 15°N 20°N 25°N 30°N 35°N 40°N 15°N 20°N 25°N 30°N 35°N 40°N

FIGURE 9.7 Subtropical North Atlantic at 66 W in August 1997. (a) Potential temperature ( C), (b) salinity, and (c) oxygen (mmol/kg). This figure can

also be found in the color insert. (World Ocean Circulation Experiment section A22.)

North America

South America

120 140

160

190

200

210

220

Puerto Rico

230 250

235

240 245

260

265

255

255

260

220

200

260

265

260

200

160

210

255

260

270

0

1500

65

4

14 67890123

North America

0

500

1000

2000

2500

3000

3500

4000

4500

5000

5500

9. ATLANTIC OCEAN


(a)

(b)

38

C. S. Vicente

36

34

Latitude (°N)

32

30

28

26

Madeira

Canarias

C. Jubi

C. Bojador

C. Beddouza

C. Sim

C. Ghir

24

22

20

4000

1000

3000 2000

C. Blanco

22 20 18 16 14 12 10 8 6

Longitude (°W)

FIGURE 9.8 Canary Current System. (a) SST (satellite AVHRR image) on August 27, 1998. This figure can also be found

in the color insert. (b) Schematic of upwelling (horizontal bars), eddy fields (dots), and preferred filaments (arrows). Source:

From Pelegrí et al. (2005).

Like other eastern boundary current systems,

the Canary Current is synoptically complex,

containing large offshore jets of upwelled water

associated with capes in the coastline

(Figure 9.8). The Canary Islands at 28e29 N

create especially vigorous filaments and an

eddy field south of the islands.

The Portugal Current is part of the generally

southward mean flow along the eastern

boundary north of the Strait of Gibraltar. Inflow

is from the branch of the NAC that lies about

45 N. Seasonality is marked, with southward

winds in spring and summer that reverse to

northward in fall and winter. This creates

a reversal in the coastal surface flow from equatorward

in summer (Portugal Current), to poleward

in autumn and winter (Portugal Coastal

Countercurrent; Ambar & Fiuza, 1994). Spring

and summer are thus the upwelling seasons.

During upwelling season, there is a poleward

undercurrent called the Portugal Coastal Undercurrent.

The deep extension of the poleward

undercurrent is an important conduit for northward

flux of Mediterranean Water (MW) exiting

from the Strait of Gibraltar (Section 9.8.3).

9.3.4. North Atlantic Current

The NAC begins as a northward western

boundary current at about 40 N, 46 W, east of

the Grand Banks of Newfoundland, fed by

a branch of the Gulf Stream. At 51 N, the

NAC separates from the boundary and turns

abruptly eastward, in a feature referred to as

260

9. ATLANTIC OCEAN

the Northwest Corner (Rossby, 1996, 1999; Zhang &

Hunke, 2001). The NAC then flows eastward

as a free jet, steered by the Charlie Gibbs Fracture

Zone at 52 N, and splits into multiple

branches. The southward branches become

part of the North Atlantic’s anticyclonic subtropical

circulation. The northward branches,

which retain locally intense frontal structures,

feed into the subpolar circulation and on

northward into the Nordic Seas.

Where the NAC is a western boundary

current, it has two roles. It is dynamically part

of the wind-driven subtropical gyre circulation,

responding to the Sverdrup forcing across the

width of the North Atlantic. It also carries the

10 to 20 Sv of northward flow of the MOC.

The northward flow eventually enters the

Norwegian Sea and the Irminger Sea, where it

is a source of the dense waters formed in the

Nordic and Labrador Seas, respectively.

The eastward flow of the NAC has some

aspects in common with the simpler North

Pacific Current (Section 10.3.1), but connectivity

of the subtropical and subpolar circulations

differs. As the pathway for the upper ocean

part of the Atlantic’s MOC, the NAC includes

more net northward transport than the North

Pacific Current since the North Pacific’s MOC

is much weaker. As a western boundary current

of the subtropical gyre circulation, the NAC has

no counterpart in the North Pacific.

Formation of the NAC is complicated, illustrated

schematically in the online supplementary

Figure S9.10. A branch of the separated

Gulf Stream turns north roughly along the

4000 m isobath east of the Grand Banks and is

joined by colder water from the inshore Slopewater

Jet and from a northward turn of the Labrador

Current, which also lies inshore of the

NAC (Figure 9.9a). By the time it reaches the

southern flank of the Flemish Cap, the NAC

can be considered a true western boundary

current, extending well inshore of the 4000 m

isobath (Figure 9.9b). Here the NAC’s velocity

structure is similar to the Gulf Stream’s

velocity structure, with maximum mean surface

velocity >60 cm/sec and northward flow

extending to the ocean bottom (Meinen & Watts,

2000). The core of the current shifts offshore

with increasing depth. The southward flow

inshore of the NAC, intensified at the ocean

bottom, is the DWBC (Section 9.6). The NAC

transport at 42 30’N has been observed to

exceed 140 Sv, of which about 50 Sv recirculates

in a local, permanent eddy (Mann Eddy). The

NAC’s net northward transport is thus about

90 Sv (Meinen & Watts, 2000). Of this, 15 to 20

Sv can be considered to be part of the MOC.

As the NAC follows the deep isobaths along

the western boundary east of Newfoundland,

it reaches a latitude where both the integrated

Sverdrup transport becomes zero and the isobaths

turn offshore in the Northwest Corner,

as mentioned previously. Waters on the

subpolar side of the NAC are cold, fresh, and

highly oxygenated to great depth. Waters on

its warm side are nearly subtropical. The sharp

front across the NAC is called the Subarctic

Front. There is some subduction of fresher

surface waters along this front, resulting in

a shallow salinity minimum on the warm side

of the front, called the Subarctic Intermediate

Water. Discussion of the NAC further downstream

is in the next section.

9.3.5. Subpolar Circulation

The subpolar gyre in the North Atlantic is the

quasi-cyclonic circulation north of 50 N(Figures

9.1 and 9.2, and Figure S9.1 and Table S9.2

located in the online supplement). It is divided

into western and eastern regimes on either side

of the Reykjanes Ridge. The western part is

a cyclonic gyre in the Labrador and Irminger

Seas. The eastern part is northeastward surface

flow in several topographically-controlled

branches of the NAC that continue northward

into the Nordic Seas. If we consider the subpolar

North Atlantic together with the Nordic Seas, the

surface flow makes a complete cyclonic gyre.


Below the depth of the Greenland-Iceland-Faroe

Ridge, the subpolar North Atlantic flow is

cyclonic throughout the region (Figure 9.2b).

The eastward NAC in the western North

Atlantic forms the southern side of the subpolar

circulation, as well as the northern side of the

subtropical circulation. Its Subarctic Front is

steered through the Charlie Gibbs Fracture

Zone in the MAR. The NAC then splits into

a part that turns southward to the subtropics,

FIGURE 9.9 North Atlantic Current and Labrador Current at the Grand Banks. (a) SST (AVHRR) on October 12, 2008,

showing cold Labrador Current moving southward along the edge of the Grand Banks. Source: From Johns Hopkins APL

Ocean Remote Sensing (1996). This figure can also be found in the color insert. (b) North Atlantic Current and DWBC velocity

section (solid contours and numbers) with temperature contours, from August 1993 to January 1994, from about 48 Wto

41 W at about 42 N. Velocity contours are 10 cm/sec. Source: From Meinen and Watts (2000).

262

9. ATLANTIC OCEAN

FIGURE 9.9

(Continued).

including the Portugal Current, and two northeastward

branches (Fratantoni, 2001; Flatau,

Talley, & Niiler, 2003; Brambilla & Talley, 2008).

The subtropical branch is associated with

typical subtropical gyre subduction. The northeastward

branches are part of the subpolar

circulation. The first turns northward into the

Iceland Basin east of the Reykjanes Ridge and

the second turns northward into Rockall

Trough, close to the eastern boundary. As they

reach the Iceland-Faroe Ridge, both branches

join the Iceland-Faroe Front and move northward

into the Norwegian Atlantic Current in the

Nordic Seas (Section 12.2).

The western cyclonic gyre begins with

a branch of the NAC that turns northward into

the Irminger Current along the western flank of

the Reykjanes Ridge. This turns west and south,

joining the EGC coming out of the Nordic Seas,

then the northward flow in the WGC and finally

the southward flow in the Labrador Current

along the Labrador coast. The Labrador Current

also sweeps in waters from the Arctic through

Baffin Bay and Davis Strait. The Labrador and

EGCs are western boundary currents. The

WGC is a more unstable eastern boundary

current, with eddies shed at Cape Farewell at

the southern tip of Greenland that move

westward into the Labrador Sea, creating

enhanced eddy kinetic energy (EKE) there.

Transport estimates for the EGC and WGC are

16 and 12 Sv, respectively, with the EGC eddies

absorbing the loss (Holliday et al., 2007).

The subpolar circulation is so strongly steered

by topography that the flow around the Labrador

Sea is sometimes referred to as the “Rim

Current.” Within the Labrador Sea (and probably

also the Irminger Sea) the cyclonic Rim Current

has a weak offshore countercurrent, running

clockwise around the sea (Lavender, Davis, &

Owens, 2000). The countercurrent reflects an

enhanced cyclonic dome near the Rim Current.

This could localize the deep convection involved

in Labrador Sea Water (LSW) production closer

to the offshore side of the current than to the

center of the Labrador Sea (Section 9.8.3; Pickart,

Torres, & Clarke, 2002).

From the Labrador Sea, the Labrador Current

continues southward to the Newfoundland

region; in SST images it is cold (Figure 9.9a).

Most of the current flows through Flemish Pass

between Newfoundland and Flemish Cap and

then southward along the continental shelf break

to the Tail of the Grand Banks. Here the cold water

evident in SST images disappears (Figure 9.9a).

Part of the current turns back northward and joins

the inshore side of the NAC. Part of the current

continues westward following the continental

slope toward Nova Scotia, well north of the

Gulf Stream. At the sea surface this westward

flow is called the Slope Water Current. The deeper

southward and westward boundary flow

(Figure 9.9b, below 1000 m) is the DWBC,

carrying new, dense LSW and Nordic Seas Overflow

Waters southward (Section 9.6).

9.3.6. North Atlantic Eddy Variability

and Gulf Stream Rings

North Atlantic eddy variability is represented

by the global EKE and coherent eddy maps of

Figures 14.16 and 14.21 and in Fratantoni (2001)

(Figure S9.11 in the online supplement). From

TROPICAL ATLANTIC CIRCULATION 263

south to north, the highest EKE is found associated

with the western boundary currents: in the

NorthBrazilCurrent,intheGulfofMexicoin

the Loop Current, along the Gulf Stream, and

in the separated Gulf Stream with its large

meanders and ring creation. The NAC

continues the axis of higher EKE toward the

north and east along 50 N. Within the subtropical

gyre, there is also slightly enhanced EKE in

the Azores Current near 35 N. In the subpolar

region, there is high EKE in the EGC and in

the eddy band in the Labrador Sea spawned

at Cape Farewell.

The overall level of EKE is lower in the

subpolar gyre than at lower latitudes. This is

related to the weaker baroclinicity of the

subpolar gyre; that is, the water column is less

stratified, the energetic currents have less

vertical shear, and isopycnals are less sloped.

Subsurface eddy variability is better observed

in the Gulf Stream region of the North

Atlantic than in any other part of any ocean,

using acoustically tracked subsurface floats

(Owens, 1991). High EKE occurs directly beneath

the high surface EKE of the Gulf Stream,

and decreases in amplitude with depth.

Gulf Stream rings are especially large, energetic,

closed eddies formed when meanders of

the Gulf Stream pinch off, forming anticyclonic

warm-core rings to the north and cyclonic coldcore

rings to the south (Figure 9.10). The Gulf

Stream does not have strongly preferred meandering

sites, unlike the other subtropical western

boundary currents; ring formation occurs all

along the front from 70 W to the Grand Banks.

The surface temperature image in Figure 1.1a

includes two obvious cold-core rings south of

the Gulf Stream and one warm-core ring to the

north. Gulf Stream rings can have surface speeds

exceeding 150 cm/sec, be 150 to 300 km in diameter,

be more than 2000 m deep, and can have

lifetimes of more than a year. At any time, in

the area west of 55 W and north of about 30 N,

there may be 3 (anticyclonic) warm-core rings

north of the Gulf Stream and 10 (cyclonic)

cold-core rings to the south (Richardson, 1983).

Approximately five warm-core, and five to eight

cold-core rings form per year.

In ring formation, a meander forms, closes up,

and then separates from the Gulf Stream (Parker,

1971) (online supplementary Figure S9.12). The

ring is in nearly solid-body rotation to about

60 km from the center, which differs from the

form it would have if it were simply a closed

loop of the Gulf Stream. Once formed, both

cold- and warm-core rings propagate westward.

Cold-core rings also move southward in the

recirculation and are often found offshore of

the Gulf Stream as far south as 28 S (Richardson,

1980c, 1983). Rings exchange biologically productive

water from north of the Gulf Stream

with much less productive Sargasso Sea water.

Therefore, warm-core rings appear in ocean

color images as areas of low chlorophyll while

cold-core rings have high chlorophyll.

9.4. TROPICAL ATLANTIC

CIRCULATION

We describe the circulation in the tropical

Atlantic briefly, reserving a more complete

description of typical features of equatorial

circulation for the Pacific (Section 10.7). The

principal near-surface currents for both the

tropics and the South Atlantic are shown in

Figures 9.1 and 9.11 and listed in Table S9.3 in

the online supplement. At the equator, the

Atlantic extends from 45 Wto10 E, a distance

of about 6000 km. Because the equatorial Pacific

is more than twice this wide, the wind-driven

equatorial current systems differ in some

respects, especially in strength. The tropical

Atlantic is bisected by the MAR, which has a

major east-west fracture zone d the Romanche

Fracture Zone d close to the equator. In the

east, the tropical region is limited to the north

by the curve of the African coastline.

The tropical circulation responds strongly to

the trade wind forcing, which has large seasonal

264

9. ATLANTIC OCEAN

FIGURE 9.10 Gulf Stream rings. (a) Locations of Gulf Stream and of warm- and cold-core rings in March to July 1975.

(b, c) Vertical temperature sections along lines A and B in (a) showing Gulf Stream and cold- and warm-core ring structures.

After Richardson et al. (1978).

changes (Stramma & Schott, 1999) as well as

interannual variability (Section 9.9 and online

supplementary Chapter S15). Seasonal wind

changes are related to shifts in strength and

location of the ITCZ, which is most strongly

developed in the summer hemisphere. The

freshwater from the Amazon and Orinoco rivers

empties into the western boundary region and

spreads northwest into the Caribbean Sea and

Gulf Stream System. The Congo River freshwater

spreads southward in the Angola Current

along the African boundary.

Circulation within 10 of the equator is nearly

zonal at depths above the strong topography. The


(a)

depth in meters

0

–100

–200

–300

–400

–500

–60

(b)

Depth (m)

0

500

1000

1500

2000

2500

3000

3500

4000

4500

5000

NBC

- 6.6 Sv

- 14.4 Sv

-15

-10

–40

SEC

- 7.0 Sv

–20

longitude

Min = –0.83 m/s Max = 0.97

–1 0 1

SEUC

2.8 Sv

- 11.2 Sv SICC

6.1 Sv

0

2.2 Sv

LSW

1.6 Sv

0.3 Sv

-5 0 -40

0

0

5

uNADW

10.9 Sv

0

EUC

8.6 Sv

35

12.3 Sv 25 30 15 20 10

0

26.8

mNADW

7.1 Sv

0

5

0

Parnaiba

Ridge

0

5

-5

32.15

- 9.8 Sv

0

INADW

6.5 Sv

5

AABW

- 2.0 Sv

-5

EIC

- 10.2 Sv

+ 3.3 Sv

- 2.3 Sv

0

+ 5.3 Sv

0 - 1.3 Sv

45.9

45.83

0

Equatorial Channel

37.0

0

- 1.4 Sv

NICC

7.7 Sv

4.1 Sv

SEC

- 2.0 Sv

0

–10

–25

–50

–75

–100

–150

–200

–250

–300

–400

–500

Contour 0.1 m/s

24.5

- 7.7 Sv

0 -5

Mid - Atlantic Ridge

5°S 4°S 3°S 2°S 1°S Eq 1°N 2°N 3°N 4°N 5°N

FIGURE 9.11 Tropical current structures. (a) Eastward

velocity along the equator, from a data assimilation. This

figure can also be found in the color insert. Source: From

0

0

0

0

5

5

0

0

0

0

0

-10 -5

0

0

0

dominant surface flow is westward, in the South

Equatorial Current (SEC). The “South SEC” is the

westward flow in the northern part of the South

Atlantic’s subtropical gyre. When it reaches the

South American coast, it splits into the southward

Brazil Current and the northward North Brazil

Current. The “Central SEC” and “North SEC”

straddle the equator to about 5e7 latitude;

directly on the equator there is also a weak westward

flow, driven by the trade winds. This equatorial

part of the SEC is bounded to the north by

the vigorous eastward flow of the North Equatorial

Countercurrent (NECC), which is associated with

the ITCZ wind forcing. The NECC bounds the

tropical circulation, separating it from the North

Equatorial Current of the North Atlantic’s

subtropical gyre.

As the NECC flows eastward, it encounters

Africa and splits into a northward flow toward

Dakar and the eastward Guinea Current along

the coast, with surface speeds in excess of

100 cm/sec (Richardson & Reverdin, 1987).

The Guinea Current follows the coast and eventually

turns south and joins the westward North

SEC. The northward flow turns westward to join

the North Equatorial Current (NEC). The

eastern tropical region between the NEC and

NECC, which forms a cyclone, is an upwelling

region called the Guinea Dome (Siedler,

Zanbenberg, Onken, & Morlière, 1992).

At about 7e8 S between the South SEC and the

Central SEC, there is a quasi-permanent (seasonal)

South Equatorial Countercurrent (SECC),

associated with the southern hemisphere ITCZ.

The SECC terminates at the coast of Africa, where

it is joined by upwelling flow from the Equatorial

Undercurrent (EUC). This turns southward along

the coast, forming the Angola Current, and then

westward into the South SEC, forming a cyclonic

upwelling region called the Angola Dome

(Wacongne & Piton, 1992). With increasing depth,

Bourlès et al. (2008). (b) Mean zonal transports (Sv) (gray

eastward) and water masses at 35 W. Source: From Schott

et al. (2003).

266

9. ATLANTIC OCEAN

the cyclonic gyre enlarges and is more “gyrelike,”

as the eastward flow beneath the surface

SECC is more pronounced and permanent

(Gordon & Bosley, 1991).

Both the Angola and Guinea Domes are

regions of upwelling and great biological

productivity. This results in a large subsurface

tropical oxygen minimum layer, with the lowest

oxygen centers in each of the domes, hence on

either side of the equator (Stramma, Johnson,

Sprintall, & Mohrholz, 2008; Karstensen,

Stramma, & Visbeck, 2008). This signal, centered

at about 500 m depth, is an obvious feature of

any vertical oxygen section in the equatorial

region (e.g., Figure 4.11d).

Along the equator, just below the sea surface

at 60 to 120 m depth, the EUC flows eastward,

similar to the EUC in the Pacific (Figures 9.11a

and 10.23c). An eastward pressure gradient force

created by the easterly trade winds, which pile

surface water up in the west, drives the EUC.

These create a weak version of the equatorial

Pacific’s warm pool and cold tongue. The EUC

core shoals from deeper than 100 m near the

western boundary to about 30 m at the eastern

boundary. Eastward currents in the EUC

core can exceed 80 cm/sec and, occasionally,

100 cm/sec, but do not reach the much larger

mean velocities of the Pacific’s EUC (Wacongne,

1990; Giarolla, Nobre, Malaguti, & Pezzi, 2005).

The full suite of subsurface flows in the equatorial

region (Figure 9.11b) continues the strong

zonal character of the surface currents. The

correspondence with the Pacific equatorial

currents is remarkable (Figures 10.20ae10.21).

On the equator beneath the EUC are found the

westward Equatorial Intermediate Current and

the “stacked jets” of alternating flow down to

about 2000 m. On either side of the equator,

at 2e4 latitude, there are eastward flows

centered around 500 to 1000 m; in the Atlantic

these are referred to as the South and North

Equatorial Undercurrents and the deeper parts

are referred to as the South and North Intermediate

Countercurrents (SICC and NICC). The

transports of each of these currents are substantial,

exceeding 5e10 Sv.

The strong zonal flows in the tropical Atlantic

are unstable, and routinely form regular trains

of planetary waves and eddies (Legeckis &

Reverdin, 1987; Steger & Carton, 1991). These

are the Atlantic Tropical Instability Waves

(TIWs); they correspond with the TIWs in the

tropical Pacific where they were discovered first

(Section 10.7.6). TIWs form on the northern and

southern edges of the equatorial cold tongue.

They have a wavelength of about 900 km, which

means there are typically about 4 to 5 waves

across the width of the Atlantic, and they propagate

westward (phase) at about 25 cm/sec. The

cold tongue and hence the TIWs are seasonal

features, appearing each summer. TIWs form

within several weeks of the cold tongue’s

appearance, grow, and begin to break, similar

to the Pacific TIWs. The energy source is mostly

barotropic instability (Jochum et al., 2004). The

TIWs on the northern and southern flanks

appear to be independent of each other. The rolling

up of the breaking waves is a dramatic

feature in satellite images, and forms large anticyclonic

eddies with diameters of about 500 km

that last for more than a month before decaying

away.

The tropical Atlantic’s low latitude western

boundary current is the NBC, which flows

northward starting from the bifurcation of the

SEC at the South American coast at around

10e15 S. It extends to intermediate depth

(~800 m), carrying surface water down through

the Antarctic Intermediate Water (AAIW) northward

into the North Atlantic. The surface water

also includes much of the 0.2 Sv of fresh water

discharge from the Amazon. Part of the NBC

turns east near the equator, joining the EUC.

The remainder crosses the equator, and then

splits into a portion that joins the eastward

NECC and a portion that continues northward

along the western boundary.

The NBC is surface-intensified (Figure 9.12b),

with velocities exceeding 90 cm/sec at the


(c)

66°W 60°W

54°W

48°W 42°W

32°S

36°S

40°S

44°S

48°S

Jul 14 1994

5 10 15 20

FIGURE 9.12 North Brazil, Brazil, and Malvinas Currents. (a) Satellite ocean color image (CZCS) of the NBC retroflection

prior to ring formation. Source: From Johns et al. (1990). (b) Mean velocity (cm/sec) from current meters in the NBC at about 4 Nin

1990. Source: From Johns et al. (1998). (c) Infrared satellite image of the Brazil-Malvinas confluence. Black lines are current vectors at

moorings, at approximately 200 m depth. Light curve is the 1000 m isobath. This figure can also be found in the color insert.

Source: From Vivier and Provost (1999). (d) Malvinas current mean velocities (cm/sec) at about 41 S, based on current meters

(crosses and diamonds) and satellite altimetry. Positive velocities are northward. Source: From Spadone and Provost (2009).

268

(d)

(a)

(b)

FIGURE 9.12

(Continued).

9. ATLANTIC OCEAN

The NBC is a site of notable mesoscale variability.

It does not easily cross the equator. At

about 5 to 7 N, it retroflects and spawns enormous

anticyclonic rings (400 km diameter) called

“North Brazil Current rings” (Figure 9.12a). Each

carries about 1 Sv of NBC water northward.

Three or more rings are formed each year. Surface

velocities in the rings are 30 to 80 cm/sec, and the

translational speed toward the northwest is

about 10 cm/sec. The rings are deep-reaching,

easily trapping floats at 900 m (Richardson,

Hufford, Limeburner, & Brown, 1994).

The continuation of a northward current past

the retroflection of the NBC is called the Guiana

Current, but the region is highly variable as

a result of the large rings. The flow and rings

impinge on the southern islands of the

Caribbean Sea, and join the westward flow into

the sea as the Caribbean Current (Section 9.3).

Below the NBC, there are opposing DWBCs

(Section 9.6.2). One approaches the equator

from the north carrying NADW centered at

2000e3000 m depth. A deeper one comes from

the south carrying Antarctic Bottom Water

(AABW). Near the equator, there is a tendency

for all layers of the western boundary current

system to detrain or leak water eastward along

the equator. The equatorial Romanche Fracture

Zone channels the deepest equatorial flows

into the deep eastern basins.

surface, decreasing to about 20 cm/sec at 200 m.

The NBC’s mean transport at 4 N is 26 Sv, based

on moored observations (Johns et al., 1998).

The transport has two sources: the wind-driven

circulation and the Atlantic’s MOC. Fratantoni,

Johns, Townsend, and Hurlburt (2000) found

that of the 14 Sv of MOC transport that travel

northward through the system, 7 Sv are carried

in the NBC and its continuation to the Guiana

Current, 3 Sv are carried by NBC rings (see

the previous section), and the remainder are

carried into the upper ocean’s interior

circulation.

9.5. SOUTH ATLANTIC

CIRCULATION

The South Atlantic surface circulation

consists of the eastward Antarctic Circumpolar

Current (ACC) in the south, an anticyclonic

subtropical gyre that is partially contiguous

with the Indian Ocean’s subtropical gyre,

and a cyclonic tropical circulation gyre (Figures

9.1, 9.2, and online supplementary Figure S9.1).

The subtropical gyre’s western boundary

current is the Brazil Current, flowing southward

along the coast of South America. The eastward

SOUTH ATLANTIC CIRCULATION 269

flow on the south side of the gyre is the South

Atlantic Current (SAC). The eastern boundary

upwelling system is the Benguela Current

System (BCS). The broad westward flow on

the north side of the subtropical gyre is the South

Equatorial Current (SEC), which splits at the

western boundary into the Brazil Current and

the NBC.

South of the subtropical Brazil Current gyre,

we enter the domain of the ACC, whose northernmost

front is the SAF (Chapter 13). The

SAF enters the South Atlantic from the Pacific

along the northern boundary of the Drake

Passage and immediately turns northward

along the coast of South America as the

Malvinas Current (or Falkland Current). The

Malvinas can be thought of as the western

boundary current of a cyclonic (subpolar-type)

circulation that is forced by positive wind stress

curl north of the latitude of the Drake Passage.

The Malvinas and Brazil Currents encounter

each other almost head-on at 36e38 S, in the

Brazil-Malvinas confluence. Both separate from

the coast and loop southward. They retain their

identities as separate fronts and move eastward

into the South Atlantic along separate paths east

of 50 W. The Malvinas Current front is then

again called the SAF, and heads eastward and

slightly southward across the South Atlantic

and Indian Oceans at about 50 S.

The South Atlantic’s subtropical gyre is

a conduit for northward flow of upper ocean

waters to the North Atlantic, where they are

ultimately transformed to dense deep waters

in the Labrador and Nordic Seas as part of the

global overturning circulation. Much of the

South Atlantic’s net northward flow originates

from the Indian Ocean via the Agulhas. This

enters the South Atlantic in the northwestward

flow in the BCS and as large Agulhas rings.

Somewhat denser near-surface flow enters

from the Drake Passage and moves northward

through the South Atlantic’s subtropical gyre

as Subantarctic Mode Water (SAMW) and

AAIW. The net northward flow of surface

through intermediate waters moves westward

in the subtropical gyre with the SEC, reaches

the western boundary, and flows northward as

part of the transport of the NBC.

In the deep layers of the South Atlantic, the

circulation is strongly modified by topography.

The DWBCs carry NADW southward and

abyssal AABW northward (Section 9.6.2). But

even in the NADW-dominated layers, there is

northward flow of Circumpolar Deep Water

(CDW), in regions other than the DWBC.

9.5.1. Subtropical Gyre

The South Atlantic’s anticyclonic subtropical

gyre is forced by anticyclonic wind stress curl,

which causes Ekman pumping and equatorward

Sverdrup transport across the South

Atlantic (Figures 5.16d, 5.17 and online supplementary

S9.3). The Brazil Current at the western

boundary is its narrow poleward return flow.

We emphasize two unique aspects of the

South Atlantic’s subtropical gyre compared

with other oceans, both partly associated with

the Agulhas, which is the western boundary

current of the Indian Ocean’s subtropical gyre

(Chapter 11). The first is the throughput of upper

ocean water as part of the Atlantic’s MOC

(Section 9.7). Low latitude Pacific and Indian

Ocean waters enter the Atlantic via the Agulhas

retroflection region and ultimately return southward

in the deep ocean as NADW. The South

Atlantic also imports water from the Pacific

through the Drake Passage that joins the northward

upper ocean flow to the NADW formation

sites. Both of these warm water sources to the

South Atlantic are part of the global overturning

circulation (Section 14.3; Figure 14.11).

A second special aspect is the connection of

the wind-driven subtropical gyre circulations

of the South Atlantic and Indian Oceans. The

subtropical wind forcing (Ekman convergence)

extends south of the African continent to around

50 S, all the way from the South American coast

eastward to Australia and New Zealand

270

9. ATLANTIC OCEAN

(Figure 5.16d). Thus the South Atlantic and

Indian subtropical gyres are connected from 34

to 50 S, with mostly eastward flow in the SAC

plus westward flow in the Agulhas.

The eastward flow of the SAC is organized

into two nearly zonal fronts at about 35 S and

40 S (see online supplementary Figure S9.13a

from Juliano & Alvés, 2007). The SAC transport,

including both fronts, is about 30 Sv. Maximum

surface speeds in the fronts are around 20 cm/

sec. The cores of the currents extend down to

about 800 m. These are both part of the subtropical

front first identified by Krummel (1882) and

called the “subtropical convergence” by Deacon

(1933). We suggest calling the two fronts the

North and South Subtropical Fronts following

Belkin and Gordon (1996) and Provost et al.

(1999). (At the western boundary, the two fronts

originate in the separated Brazil Current and are

therefore called the Brazil Current Front and the

Subtropical Front by Peterson, 1992 and

Tsuchiya, Talley, & McCartney, 1994.)

The North Subtropical Front terminates in the

east where it meets the Agulhas retroflection

and Benguela Current just west of Africa. The

front may turn continuously northward where

it becomes the outer front of the BCS. The South

Subtropical Front continues on eastward into

the Indian Ocean, clearly separated from, and

south of, the Agulhas Return Current (Chapter

11; Belkin & Gordon, 1996).

The SEC originates in the Benguela Current at

34 S at the tip of Africa and moves generally

northwestward toward the broad western boundary

region between 15 and 30 S(Figure 9.1).

Agulhas rings generated at the retroflection

move westward more zonally than the overall

SEC, following the more zonal depth-integrated

streamfunction (Biastoch, Böning, & Lutjeharms,

2008).

9.5.2. Brazil Current

The Brazil Current begins its southward

flow in the surface layer at about 10e15 S

along the coast of Brazil. Deeper parts of the

Brazil Current begin at successively higher

latitudes (toward the south). An anticyclonic

recirculation gyre on the offshore side of the

Brazil Current, south of about 30 S, increases

the Brazil Current’s transport toward the

south. The Brazil Current begins to separate

from the coast at the Brazil-Malvinas confluence

at about 36 S, where the cold Malvinas

water intrudes inshore of the warm Brazil

Current (Figure 9.12c). The main transport

of the Brazil Current finally leaves the

continental shelf somewhat farther south,

around 38 S.

A quantitative overview of Brazil Current

structure and transport appears to be lacking.

Maamaatuaiahutapu, Garçon, Provost, and

Mercier (1998) provided references and

a partial summary of transport estimates.

The Brazil-Malvinas confluence has been

well-measured with moored velocity observations,

but upstream locations have not.

Large-scale inverse circulation models have

generally focused on 32 S. Many studies

have used shallow zero velocity reference

levels for geostrophic velocities, thus underestimating

the current strength. Considering

only estimates that move beyond zero reference

velocity assumptions, the southward Brazil

Current transport is estimated at 2.5 Sv at

its very beginnings at 12 S, with increasing

southward transport of 4 Sv, 11 Sv, 17 Sv, 22

Sv, and 41 Sv at 15 S, 27 S, 31 S, 34 S, and

36 S, respectively (online supplementary

Figure S9.14b from Zemba, 1991; Sloyan &

Rintoul, 2001; Stramma, Ikeda, & Peterson,

1990). The Brazil Current has a large recirculation

gyre that begins south of 30 S and greatly

increases in transport south of 35 S. By 36 S,

the total Brazil Current transport is 70 to 80

Sv, with about half participating in the recirculation

gyre (Zemba, 1991; Peterson, 1992).

About 30 Sv actually exits to the east into the

South Atlantic gyre (Stramma & Peterson,

1990).

SOUTH ATLANTIC CIRCULATION 271

9.5.3. Malvinas Current and

Subantarctic Front

The Malvinas Current originates from the

SAF in the Drake Passage (Figure 9.1 and

supplementary Figure S9.14a). It flows northward

as a western boundary current after

leaving the Drake Passage, following the South

American shelf roughly along the 1000 m isobath,

up to about 38 S. The Malvinas meets the

Brazil Current head-on at this point, as seen in

satellite SST images of the Brazil-Malvinas

confluence (Figure 9.12c). Both currents separate

from the coast and move offshore and

southward. The expected location of the Brazil

Current separation (based on wind stress curl

and Sverdrup transport) is about 10 latitude

south of the actual separation point (Figure 5.17

and supplementary Figure S9.3b); the powerful

Malvinas appears to push the Brazil Current

separation northward (Spadone & Provost,

2009).

Mean surface velocities in the Malvinas

Current are about 40 cm/sec. The Malvinas is

most intense inshore of the 1500 m isobath, and

extends to the ocean bottom (Figure 9.12d). Transport

from the moored array of Figure 9.12disestimated

to be 42 Sv with a variability of 12 Sv

(Spadone & Provost, 2009), which is weaker

than the maximum transport estimate of 70 Sv

from Peterson (1992).

Where the Malvinas separates from the coast,

it turns sharply southward in a feature referred

to as the Malvinas, or Falkland, Loop. Eddies

are generated, mixing with the much warmer

and more saline Brazil Current water. Subsurface

waters in the Malvinas may not turn as

strongly southward, which injects some of the

denser waters into the subtropical gyre. This is

the primary mechanism by which the AAIW

enters the SAC (Talley, 1996a).

After looping back to the south, the Malvinas

reaches the northern escarpment of the Falkland

Plateau and turns to the east along the bathymetry.

Here it is again referred to as the

Subantarctic Front. The SAF is distinct from

the two subtropical fronts. It crosses the South

Atlantic at around 50 S, which is approximately

the latitude of zero wind stress curl

(Figure 5.16d).

In the ACC, the Polar Front lies south of the

SAF. After exiting the Drake Passage to

the South Atlantic, the Polar Front remains in

the Scotia Sea while the SAF executes the large

northward Malvinas Current loop. The two

fronts converge along the north side of the Falkland

Plateau and then remain relatively close

to each other, but distinct in water mass properties,

across the rest of the South Atlantic

(Figure 13.1).

9.5.4. Benguela Current System

The BCS is the eastern boundary current

system for the subtropical South Atlantic. The

BCS extends from 34 S at Cape Agulhas at the

southern tip of Africa northward to 14 S. Details

can be found in Shillington (1998) and Field and

Shillington (2006). The BCS is unique among

eastern boundary currents because of its role in

the northward transport of warm waters in the

global overturning circulation. Some portion of

the Benguela originates in the warm, saline Agulhas

Retroflection waters that round the southern

tip of Africa. The warmth and higher salinity of

waters on its poleward end distinguishes the

BCS from the other eastern boundary current

systems.

The BCS has the characteristics of a classic

upwelling system (Section 7.9) with upwellingfavorable

equatorward winds, offshore Ekman

transport, an equatorward surface current (the

Benguela Current), and a poleward undercurrent.

The equatorward winds that force the

BCS are strongest offshore, leading to positive

wind stress curl close to shore (Figure 5.16d).

As in the other eastern boundary current

upwelling systems, the positive wind curl is

associated with local upwelling in a wider

band than just within the near-coastal strip.

272

9. ATLANTIC OCEAN

Upwelling season in the southern Benguela is

in the austral summer (December through

February), but continues year-round in the

central and northern Benguela, based on

appearance of cold eastern boundary waters in

satellite SST images (Figure 9.13). As in other

eastern boundary current systems, the BCS is

marked by offshore jets of cold water associated

with the coastline. Preferred locations are northwest

of Lüderitz (24e26 S), and at 28e30 S.

At the northern boundary of the BCS, the

northward, cold Benguela Current meets

the southward, warm Angola Current. (The

Angola Current is part of the mid-ocean

cyclonic gyre that occupies the tropical South

Atlantic, described in Section 9.4.) The resulting

Angola-Benguela Front is located near Cape

Frio at about 16 S (Shillington, 1998). The front

is evident in the SST in Figure 9.13a andb.

9.5.5. South Atlantic Eddy Variability

and Agulhas Rings

Eddy variability in the South Atlantic is illustrated

in the global maps of surface EKE and

coherent eddies in Figures 14.16 and 14.21. The

(c)

20°S

Poleward Flow

30°S

40°S

Ben g u el a

Cu

r r en t

Sout h At lant ic

Current

Ag u

W a l v

C ap e

lh as Ext ension

i d g e

i s R

B asi n

SOUTHERN

AFRICA

Cape

Town

A g u l h a s Cu

Agulhas Return Current

r r e n t

So u t h

A t l a n t i c

Cu r r e n t

10°W 0° 10°E 20°E 30°E 40°E

FIGURE 9.13 Benguela Current and Agulhas retroflection. (a, b) AVHRR SST monthly composite for July (winter) and

December (summer) 2005. Source: From UCT Oceanography Department (2009). (c) Schematic of Agulhas retroflection and

eddies, with flow directions in the intermediate water layer. Gray-shaded rings are the Agulhas anticyclones. Dashed rings

are cyclones that are generated in the Agulhas. This figure can be found in the color insert. Source: From Richardson (2007).

DEPTH DEPENDENCE OF THE ATLANTIC OCEAN CIRCULATION 273

highest EKE occurs in the equatorial region and

North Brazil Current, and in the Malvinas/Brazil

Current confluence and Agulhas retroflection.

Each of these regions spawns anticyclonic rings.

Brazil Current warm-core rings form at the

large southward meander of the Brazil Current

Front after it separates from the coast, reaching

45 S near 50 W (supplementary Figure S9.14c

from Lentini, Goni, & Olson, 2006). Approximately

6 rings with a diameter of about 100

km are shed at this meander per year. They drift

southward at a mean speed of 10 km/day and

have a lifetime on the order of 40 days. This

meander and ring formation area encircle

a topographic feature, the Zapiola Rise, above

which there is much lower EKE (Figure 14.16),

and around which there is a permanent anticyclonic

flow, the “Zapiola Eddy,” which is more

well defined with increasing depth (Section 9.6).

Agulhas rings are anticyclonic, warm-core rings

that form when the Agulhas protrudes westward

south of Africa and retroflects back to the east,

between 15 Eand20 E(Figure 9.13c and Section

11.4.2). The centers of the rings are warm and

saline in contrast with the local South Atlantic

waters. The rings are 100e400 km in diameter,

with maximum speeds of more than 100 cm/sec

at the surface, and up to 10 cm/sec even at 4000

m depth (supplementary Figure S9.15 from van

Aken et al., 2003). Like Gulf Stream rings, Agulhas

rings are in solid body rotation out to the

locus of maximum surface speed.

About six Agulhas rings are produced each

year and propagate westward into the South

Atlantic, with about three reaching the South

American coast and entering the North Brazil

Current (Gordon, 2003) (supplementary Figure

S9.13b). Each ring contributes a volume transport

into the South Atlantic of 0.5 to 1.5 Sv (Richardson,

Lutjeharms, & Boebel, 2003). The 6 rings

per year with 3 to 9 Sv thus represent a significant

fraction of the exchange from the Indian to the

Atlantic, with the rest carried by connection to

the Benguela Current. Part of this transport is

simply part of the extended Atlantic/Indian

anticyclonic circulation north of the ACC, and

part of it contributes to transport of warm water

in the global overturning circulation.

9.6. DEPTH DEPENDENCE OF THE

ATLANTIC OCEAN CIRCULATION

The circulation in the upper 1000 to 1500 m of

the Atlantic is mostly associated with wind

forcing through Ekman pumping and subduction/obduction

(Section 7.8). This circulation’s

depth dependence depends on regime (tropical,

subtropical, subpolar). The vigorous subtropical

western boundary currents and equatorial

current systems associated with wind forcing

extend weakly to the bottom, but their lateral

extent is very limited. Outside the energetic

wind-driven western boundary regimes, circulation

below the subtropical and tropical pycnocline

may be mostly associated with buoyancy

forcing and overturning circulation. This

includes weak interior ocean flows that are

easily masked by the eddy field and the slightly

more vigorous DWBCs (Section 7.10.3), which

are observed at all latitudes in the Atlantic. In

contrast, the wind-forced subpolar North

Atlantic circulation, while most vigorous at the

sea surface, extends to the ocean bottom where

it is merged with the buoyancy-driven circulation;

the whole complex mostly follows topographic

contours.

9.6.1. Depth Dependence of the

Wind-Driven Circulation

The subtropical and subpolar gyre circulations

change with depth. The energetic winddriven

circulation of the upper ocean decreases

in energy with depth and also changes shape

laterally. The main points about depth dependence

of the wind-driven, anticyclonic

subtropical circulation apply to all subtropical

gyres, including those of the North and South

Atlantic:

274

9. ATLANTIC OCEAN

1. The western boundary currents and their

extensions penetrate to the ocean bottom,

but are vertically sheared so that the

highest velocities are in the upper ocean.

The recirculation gyres, which are directly

adjacent to and result from these strong

currents, also penetrate to the ocean

bottom.

2. The subtropical gyres shrink westward and

poleward with increasing depth, becoming

compressed into their western boundary

currents and separated extensions.

3. The subtropical gyre circulation can be

conceptualized as multiple layers in which

streamlines begin at the sea surface and move

downward along isopycnals into the interior

ocean (ventilation through the process of

subduction, Section 7.8.5), and deeper layers

of anticyclonic circulation that are not

connected to the sea surface (locally

unventilated). Ventilated layers contain

unventilated regions where the

streamfunctions do not connect to the sea

surface. The flow in each of the layers is

rotated relative to that in the overlying and

underlying layer, so that at any given location

(latitude-longitude), the waters on different

isopycnals forming the local vertical profile

will have come from different geographic

locations at the sea surface. This creates the

subtropical pycnocline structure (Central

Water).

The main points about the depth dependence

of the wind-driven, cyclonic subpolar circulation

are

1. The circulation is nearly “equivalent

barotropic,” meaning the surface current

structure extends to the ocean bottom

(barotropic), even though it diminishes in

strength with depth (equivalent).

2. Ekman divergence in the surface layer drives

upwelling, so there are no regions of

subduction, hence interior ventilation, via

wind-driven flow along streamlines.

(Ventilation in this region is due to the

buoyancy-driven circulation, through

convection or brine rejection.)

3. In the subpolar North Atlantic, the

Greenland-Iceland-Shetland ridge strongly

constrains the subpolar circulation. The

flow above the sill depth of the ridge

extends northward into the Nordic Seas,

and is part of a much larger regional

cyclonic circulation. Below sill depth, the

subpolar circulation is constrained to

follow the complicated isobaths. Thus in

the North Atlantic, there is a shift in the

shape of subpolar circulation above and

below this ridge.

4. The coexisting overturning circulation also

has deep currents that follow isobaths. It

is not straightforward to distinguish winddriven

and thermohaline features in the

North Atlantic’s subpolar gyre. Bottom

intensification is an indication that a given

flow has a significant thermohaline

component. (Examples are DWBCs and

the plunging plumes that overflow from

the Nordic Seas and Mediterranean,

neither of which are wind-driven

features.)

9.6.1.1. Depth Dependence of the

Subtropical Gyre Circulation

The North Atlantic’s subtropical circulation

at the sea surface is anticyclonic, with its intense

western boundary current, but the subtropical

“gyre” is not quite closed (Figure 9.2a). On the

equatorward side, the gyre’s streamlines merge

broadly with the equatorial circulation, into the

eastward NECC. The subtropical Brazil Current

gyre also smoothly merges into the SECC. In the

North Atlantic, the region of highest steric

height at the sea surface parallels the Florida

Current and Gulf Stream, with a maximum

offshore of the Antilles Current. In the South

Atlantic, the region of highest steric height

stretches from 15 to 40 S.


However, just 250 m below the sea surface

and well represented by the 500 dbar map

(Figure 9.2b), the subtropical gyres in both

hemispheres are considerably more “gyrelike,”

with large areas of closed streamlines.

For the Gulf Stream, the region of closed streamlines

is shifted away from the eastern boundary

and toward the separated Gulf Stream. The

region of highest steric height shifts to north of

about 30 N. The Brazil Current gyre likewise

tightens toward the pole and the west, shifting

to south of 30 S. At 500 dbar, both subtropical

gyres have tightened further into their western

and poleward corners.

At 1000 dbar in the North Atlantic (Reid,

1994), the Gulf Stream System is greatly reduced

spatially, to the separated eastward flow of the

Gulf Stream and its two recirculation gyres to

the north and south. At 1000 dbar in the South

Atlantic, the strongest part of the gyre shrinks

toward the Brazil Current’s southwest corner/

separation point.

The Gulf Stream and NAC and their recirculation

gyres penetrate to the ocean bottom

(Figure 9.14 and supplementary Figure S9.16).

Acoustically tracked floats at 2000 m show this

penetration, and also the vanishing of statistically

important mean flow in other regions

(Owens, 1991). At greater depth, current meter

observations along 55 W show the Gulf Stream

and its flanking recirculations (Hogg, 1983). (See

also Figure 9.5b.)

In the South Atlantic, the poleward shrinkage

of the subtropical gyre has been observed with

acoustically tracked floats at the bottom of the

pycnocline and in the AAIW layer (Boebel

et al., 1999; supplementary Figure S9.17). The

westward return flow of the subtropical gyre is

much more zonal and intense in these direct

observations than in the Reid (1994) streamfunctions.

The anticyclonic Zapiola Eddy, embedded

in the eastward flow of the ACC and SAC in the

central Argentine Basin (Section 9.5.5), is more

of a closed circulation at depth than at the sea

surface.

9.6.1.2. Depth Dependence of the North

Atlantic’s Subpolar Gyre

The subpolar gyre is divided at the sea

surface into western and eastern domains

(Section 9.3.5). The western subpolar domain,

west of the Reykjanes Ridge, has an almost

closed cyclonic surface circulation (the Rim

Current introduced in Section 9.3.5). The eastern

domain is the NAC, which flows eastward at

about 50 N and then turns northeastward and

crosses the Iceland-Faroe-Shetland ridge into

the Norwegian Sea.

With depth, the Rim Current extends to the

ocean bottom (Figures 9.2 and 9.14). By middepth,

and down to the bottom, this circulation

is filled with newly formed intermediate and

deep waters (Section 9.8). By 700 m depth, the

mean flow also includes a counterflow offshore

of the boundary current, creating a cyclonic

dome in the shape of a “donut” around the Labrador

Sea and western Irminger Sea (Lavender

et al., 2000; supplementary Figure S9.18). This

donut is the preferred locale of the deeper

convection that creates the densest SPMW and

LSW (Section 9.8.2).

In the eastern subpolar domain, the Iceland-

Faroe-Shetland ridge alters the northeastward

NAC. Below sill depth, the flow must be closed

to the north, and it becomes continuously

cyclonic, nearly following bathymetric contours

(Figures 9.2, 9.14 and supplementary

Figure S9.19 from Bower et al., 2002). Vertical

shear is required for this configuration and

most likely occurs mainly on the eastern flanks

of the Reykjanes Ridge and Rockall Plateau.

9.6.2. Deep Circulation and Deep

Western Boundary Currents

This is a brief overview of the part of the

weak lateral circulation below the main pycnocline

that is mostly associated with density

changes. The deep circulation is often

described in terms of water masses (Section

9.8), since the direction of flow is often inferred

276

9. ATLANTIC OCEAN

FIGURE 9.14

(1994).

Steric height (10 m 2 s 2 ) at (a) 2500 dbar and (b) 4000 dbar, adjusted to estimate the absolute geostrophic circulation. Source: From Reid


from property distributions (because there are

few direct velocity observations). The associated

overturning circulation is described in

Section 9.7.

9.6.2.1. Lateral Circulation and Basin

Connections

Looking at the lateral circulation at 2500 and

4000 dbar (Figure 9.14), the Gulf Stream and its

recirculation features are still present, as is the

subpolar North Atlantic circulation and a

residual of the Brazil-Malvinas confluence.

Circulation in the South Atlantic may be broadly

cyclonic, circling the MAR. Along the coast of

Africa, beneath the Benguela Current, a deepreaching

poleward boundary current occurs at

2500 dbar, analogous to deep poleward flow in

the South Pacific at about the same latitudes.

This transports NADW out of the Atlantic and

into the Indian Ocean. At 4000 dbar, flow in

this Cape Basin, south of the Walvis Ridge, is

likely cyclonic.

Abyssal flows are affected by the topography.

Deep flows often follow topographic contours

and mixing can be related to the structure of

the topography. The mid-ocean ridges confine

deep waters to the abyssal basins. Fracture

zones in the ridges allow for limited exchange

through sometimes vigorous, turbulent flow of

waters from one deep basin to another. Bottom

waters in the downstream basin tend to be relatively

uniform with properties set where the

basin was filled at the fracture zone. Principal

fracture zones affecting Atlantic deep and

bottom waters include the following, each of

which has been studied locally: the Vema and

Hunter Channels (northward flow of AABW

from the Argentine to the Brazil Basin), the

Namib Col in the Walvis Ridge (southeastward

flow of AABW and NADW into the Cape Basin),

the Romanche Fracture Zone in the MAR at the

equator (eastward flow of AABW and NADW),

the Vema Fracture Zone at 11 N in the MAR

(eastward AABW flow into the eastern North

Atlantic), and the Charlie Gibbs Fracture Zone

at 52 N in the MAR (eastward flow of the

Denmark Strait Overflow Water and Labrador

Sea Water).

9.6.2.2. Deep Western Boundary Currents

The dense water masses formed in the

northern North Atlantic must, on average,

spread southward while the dense waters

formed in the Southern Ocean must, on average,

spread northward. DWBCs that respond to

spatially limited sources of dense water

and net upwelling in the ocean interior are

part of the circulation of these newly formed

dense waters. (Dynamically, it is important

to recall from Section 7.10.3 that the DWBCs

do not necessarily flow away from their deep

sources, but in the case of the Atlantic, they

mostly do.)

Historically, Wüst (1935) showed preferential

southward spreading of the North Atlantic’s

oxygenated, saline deep waters along the western

boundary, foreshadowing later discovery

of the DWBC there. In the 1950s, following

H. Stommel’s advice, Swallow and Worthington

(1961) measured the southward DWBC beneath

the Gulf Stream off the coast of South Carolina

(Section 7.10.3). Through the 1960s and 1970s,

DWBCs were traced worldwide (Warren,

1981). Work since then has refined estimates of

transports, described exchange between

DWBCs and the interior, considered the continuity

of DWBCs, and studied local aspects of

interaction of DWBCs with other strong circulation

systems.

Along the western boundary of the Atlantic,

DWBCs associated with both NADW and

AABW are found. The northern DWBC originates

in overflows from the Nordic Seas joined

by mid-depth waters from the Labrador Sea

(Figure 9.15a); direct velocity measurements

east of Greenland show the plume of dense overflow

water moving to the bottom of the northern

North Atlantic (supplementary Figure S9.20

from Dickson and Brown, 1994). The AABW

DWBC lies beneath and offshore of the NADW

278

9. ATLANTIC OCEAN

DWBC (Figures 9.14, 9.15b and Figure 9.25). Existence

of the DWBCs is illustrated by a series of

velocity sections in the textbook Web site (Figures

S9.21, S9.22, S9.23). Oxygen and chlorofluorocarbons

(CFCs) are heightened in the DWBCs

because they carry recently formed NADW

(LSW and Nordic Seas Overflow; NSOW), which

has elevated atmospheric gas concentrations,

into the subtropical North Atlantic (Figures 9.22

and 9.7).

The DWBC that carries NADW southward is

centered around 2500 m, but it extends up to at

least 1500 m in the North Atlantic and tropics,

and down to the bottom in the North Atlantic

(Figures 9.5, 9.9, 9.11). The NADW’s DWBC

begins to form as soon as the NSOWs spill

across the Greenland-Iceland-Shetland ridges

into the deep subpolar North Atlantic, forming

the cyclonic Rim Current at depth (Figure 9.15a).

This abyssal current follows the boundary

(a)

FIGURE 9.15 Schematics of deep circulation. (a) NSOW (blue), LSW (white dashed), and upper ocean (red, orange, and

yellow) in the northern North Atlantic. Source: From Schott and Brandt (2007). (b) Deep circulation pathways emphasizing

DWBCs (solid) and their recirculations (dashed). Red: NSOW. Brown: NADW. Blue: AABW. This figure can also be found in

the color insert. (M.S. McCartney, personal communication, 2009.)


FIGURE 9.15

(Continued).

around Greenland and into the Labrador Sea,

where it picks up LSW along the western

boundary. The whole layered complex flows

out of the Labrador Sea into the Northwest

Corner of the NAC. Part of it joins the eastward

NAC flowing through the Charlie Gibbs Fracture

Zone, and part continues southward along

the western boundary to the east of Flemish

Cap, and then along the Grand Banks of

Newfoundland (Figure 9.9b). This part of the

DWBC moves southward along the western

boundary under and inshore of the Gulf Stream,

where it is seen at Cape Hatteras (Figure 9.5b).

Interaction with the Gulf Stream is complex

(Pickart & Smethie, 1993), as suggested by the

brown zigzag in Figure 9.15b. The NADW’s

DWBC moves on to the equator, where part of

the flow turns eastward along the equator

(Figure 9.11). Part continues into the South

Atlantic, and leaves the western boundary at

25 to 40 S(Figure 9.14a).

Velocities in the NADW’s DWBC are on the

order of 5 to 20 cm/sec and more. Transports

areontheorderof10to35Sv,anddepend

on latitude since the DWBC has significant

recirculations, indicated schematically in

Figure 9.15b. The recirculations increase the

local transport and also mix the DWBC waters

with the interior waters, greatly increasing the

transit time of water parcels down the western

boundary as measured by transient tracers

such as CFCs. At 26.5 N, for instance, the

southward DWBC throughput is as much as

22 Sv, while the net southward transport is

35 Sv, of which 13 Sv are due to the deep recirculation

gyre (Bryden, Johns, & Saunders,

2005a).

Now consider the northward flow of the

AABW’s DWBC, which becomes organized in

the southwestern South Atlantic. The AABW

DWBC moves northward along the coast of

South America offshore of and deeper than

the southward DWBC carrying NADW

(Figure 9.25). Its northward transport is in the

order of 7 Sv in the South Atlantic and into

the North Atlantic. As the AABW approaches

the equator, part of its transport turns eastward

with the NADW (Figure 9.11), and the rest

crosses the equator. At this point it crosses to

the eastern boundary of the basin, riding along

the western flank of the MAR, rather than

remaining a DWBC (Figures 9.14b and 9.15b).

Some of it passes into the deep eastern North

Atlantic through the Vema Fracture Zone. The

AABW loses its transport through upwelling

into the overlying isopycnals, which contain

NADW, and disappears.

9.6.2.3. Recirculations and Time

Dependence

The DWBCs are the most energetic part of the

deep circulation with velocities up to tens of

centimeters per second. The simple, laminar

280

9. ATLANTIC OCEAN

boundary currents predicted by theory are

unlikely to be either simple or laminar. One

analogy might be the extent to which eastern

boundary currents are modeled as simple

laminar flows that arise in response to offshore

Ekman transport whereas in actuality they are

full of local jets and eddies. Another is the extent

to which the actual Gulf Stream is predicted by

simple Sverdrup balance/western boundary

current theory. Both systems have much greater

spatial and temporal variability than simple

theories suggest, although the simple theories

provide the most basic understanding of the

existence of these systems.

DWBCs lie entirely in the deep ocean, so

their spatial and temporal variability have

been difficult to observe. Differences between

the observed DWBCs and simplified theory

include geographically localized detrainment

of water from the DWBCs, and large-scale,

permanent recirculation gyres (e.g., Figure

9.15b). High-resolution numerical modeling

suggests ongoing creation of DWBC eddies

along the South American coast (Dengler

et al., 2004), while deep Lagrangian float

observations show considerable eddy activity

(Hogg & Owens, 1999).

Several detrainment locations and recirculation

gyres for the DWBCs have been described.

At each, the DWBC properties change significantly

as water is exchanged with the interior.

Starting from the north and with the NADW

DWBC, there is detrainment at the exit to the

Labrador Sea and along the Newfoundland/

Flemish Cap region (Bower, Lozier, Gary, &

Böning, 2009). A second is at the Gulf Stream

separation point at Cape Hatteras (Pickart &

Smethie, 1993). There is a tropical set, at about

20 and 5 N, and a large detrainment at the

equator, all as part of the Guiana abyssal gyre

(Kanzow, Send, & McCartney, 2008). In the

South Atlantic, there is a change in DWBC

character as it passes the easternmost point of

South America (around 8 S) with either a recirculation

gyre between 20 and 8 S (Reid, 1994;

Friedrichs, McCartney, & Hall, 1994) or a

change to a more eddy-like character (Dengler

et al., 2004). A major detrainment occurs at

about 20 S upon encountering the Vitória-

Trindade Seamounts, forming the southern

boundary of a recirculation gyre (Tsuchiya

et al., 1994; Hogg & Owens, 1999). The final

detrainment is where the NADW DWBC, along

with the Brazil Current, encounters the Malvinas

Current/SAF.

The AABW DWBC, flowing northward

through the South Atlantic, also has several

major transitions. Large temporal variability is

found where it has been directly observed. The

first transformation occurs where this DWBC

leaves the Argentine Basin and enters the Brazil

Basin, at the Rio Grande Rise around 32 S,

where its deepest flow is confined to the narrow

Vema and Hunter Channels (Hogg, Siedler, &

Zenk, 1999). The second is at the Vitória-

Trindade Seamounts which interrupt and deflect

the DWBC eastward at 20 S (Hogg &

Owens, 1999). A third large change occurs at

the equator, where the northward flow of

AABW shifts over to the eastern side of the

Guiana Basin (McCartney & Curry, 1993).

9.7. MERIDIONAL OVERTURNING

CIRCULATION IN THE ATLANTIC

The MOC of the Atlantic, which is part of the

global overturning circulation (Chapter 14), is

a double cell consisting of (1) northward flow

of upper ocean waters that become denser in

the northern North Atlantic and flow out

southward at depth, eventually becoming

NADW, and (2) northward flow of dense

AABW that upwells into the lower part of the

NADW, disappearing by the mid-latitude

North Atlantic.

The upper ocean waters that flow northward

from the South Atlantic to feed the overturn in

the North Atlantic originate as: (a) upper

Indian Ocean waters from the Agulhas

ATLANTIC OCEAN WATER MASSES 281

retroflection, and (b) the slightly more dense

AAIW and Upper Circumpolar Deep Water

(UCDW: Section 9.8.3). These are transformed

into NADW components in the Labrador,

Mediterranean, and Nordic Seas. The water

masses are described in Section 9.8.

Superimposed above this in the surface

layers are shallow overturning cells that move

the warm, light tropical surface waters poleward

in the subtropics and return them as

cooler, denser subducted pycnocline waters.

While these shallower cells might not grab our

attention because they do not have global scale,

they are responsible for most of the ocean’s

poleward heat redistribution (Chapter 5).

Returning to the full-depth MOC, meridional

transports are computed in layers from zonal

coast-to-coast sections (Section 14.2; see example

in Figure 9.16 from Talley, 2008; compare with

Bryden et al., 2005b and Ganachaud, 2003, whose

results are included in supplementary Figures

S9.24 and S9.25). The net southward transport

of NADW is typically 15e25 Sv (depending on

latitude), which is almost all carried by the

DWBC (Section 9.6.2). Northward transports

that feed this are divided into 3 to 7 Sv of bottom

water with the remainder in the upper ocean

layers (AAIW and pycnocline).

An overturning transport streamfunction

can then be calculated for display; the method

is described in Section 14.2.3. An Atlantic overturning

streamfunction from a high-resolution

global ocean model is shown in Figure 14.8

(Maltrud & McClean, 2005). This particular

calculation shows a maximum NADW cell

of 22 Sv centered at 40 N, with a typical

NADW transport of 16 Sv for the length of

the Atlantic. This particular model also has

almost no AABW bottom cell, which is a

common problem with ocean circulation

models at this time; data-based transport estimates

show a much more robust AABW influx

(Figure 9.16).

The Atlantic’s MOC is included in all schematics

of global overturn (Figures 14.10 and

14.11 and the original sources upon which these

were based). The NADW layer that exits the

Atlantic upwells to the surface waters and downwells

to the bottom waters that then feed back

into the Atlantic. The upwelling occurs broadly

through the Indian and Pacific Oceans and in

the Southern Ocean within the ACC latitudes.

The “downwelling” is formation of dense waters

around Antarctica from upwelled surface waters

that include NADW (Chapter 14).

The meridional heat and freshwater transports

accompanying the overturning circulation

are discussed in Section 14.3, in the context of

the transports for all oceans. Briefly, the heat

transport is northward throughout the length of

the Atlantic, with a maximum in the subtropical

North Atlantic just south of the Gulf Stream

separation, where ocean heat loss is maximum.

The northward sign found even in the South

Atlantic is due to the additional heat loss region

of the Nordic Seas, and is thus associated with

the full-depth MOC. Freshwater transport is

more complicated to discuss, but the most

important result for the Atlantic’s overturning

circulation is that the NADW cell transports

freshwater southward, because the northwardflowing

upper ocean waters are saline, and the

new NADW is fresher because it incorporates

much of the Arctic and subpolar net precipitation

and runoff. (It is dense enough to sink because it

is cooler than the inflowing surface waters.)

9.8. ATLANTIC OCEAN WATER

MASSES

The hydrographic structure and basic water

masses of the Atlantic Ocean were introduced

in Chapter 4 in terms of four layers in the

vertical: surface through pycnocline, intermediate,

deep, and abyssal. Upper ocean water

mass structures and processes are similar in all

oceans. However, the North Atlantic forms

deep waters locally, unlike the North Pacific

and Indian, resulting in a complete asymmetry

282

9. ATLANTIC OCEAN

(a)

0

1000

2000

3000

4000

5000

North Atlantic 24°N

Subducted thermocline

27.3

Antarctic Intermediate Water and Mediterranean Water

27.74

Labrador Sea Water

36.96

North Atlantic Deep Water

45.91

AABW

(b)

Ekman

surface

27.3

27.74

36.96 2

45.91 4

bottom

NADW

34.93

LSW

35.07

(b) North Atlantic 24°N

Ekman

35.88

Upper

36.18

AAIW/MW

35.13

AABW

34.86

–20 –15 –10 –5 0 5 10 15

6000

0 1000 2000 3000 4000 5000 6000

–80 –75 –70 –65 –60 –55 –50 –45 –40 –35 –30 –25 –20 –15

(c)

0

1000

2000

3000

4000

5000

32.0 34.0 34.5 34.7 34.8 34.9 35.0 35.5 36.0 36.5 37.0

South Atlantic 32°S


26.2

Lower thermocline

26.9

Antarctic Intermediate Water

27.4

North Atlantic Deep Water/Circumpolar Deep Water

45.86 45.86

45.88

AABW

(c) South Atlantic 32°S

6000

0 1000 2000 3000 4000 5000 6000

–50 –45 –40 –35 –30 –25 –20 –15 –10 –5 0 5 10 15

(d)

Ekman

surface

26.4

26.9

27.4

45.86 4

bottom

NADW

34.77

(d) South Atlantic 32°S

Ekman

35.87

Upper

35.68

Upper

35.09

AAIW

34.40

AABW

34.82

–20 –15 –10 –5 0 5 10 15

Volume Transport (Sv)

FIGURE 9.16 Salinity and meridional transport in isopycnal layers at (a, b) 24 N in 1981 and (c, d) 32 S in 1959/1972.

The inset map shows section locations. The isopycnals (s q , s 2 , s 4 ) that define the layers are contoured in black on the salinity

sections. Figures 9.16a, c can also be found in the color insert. See also online supplementary Figures S9.24 and S9.25 for

examples from Bryden, Longworth, and Cunningham (2005b) and Ganachaud (2003). After Talley (2008), based on Reid (1994)

velocities.


between the three oceans in their deep water

ages and age-related properties (oxygen, nutrients,

CFCs, etc.; Sections 4.5 and 4.6; Figures

4.11 and 4.22). Together with the supply of

younger waters in the Antarctic, which is

common to all three oceans, much of the interior

Atlantic is affected by ventilation changes

within decades. Therefore care must be taken

in combining Atlantic data sets from different

decades. 5

Most of the main water masses of the Atlantic

Ocean are shown in Figures 9.17 and 9.18 and

listed in Table S9.4 in the online supplement.

We start with summary potential temperaturesalinity

diagrams (Section 9.8.1) and follow

with details about the water masses from

shallow to deep, illustrated with vertical

sections and maps of properties on isopycnals.

9.8.1. Potential Temperature vs.

Salinity and Oxygen

Many water masses are identified by vertical

maxima or minima of salinity or oxygen. The

principal water masses are therefore first illustrated

with salinity and oxygen versus potential

temperature. Figure 9.18 includes thousands of

bottle samples collected in the WOCE. A much

older but useful schematic potential temperature-salinity

(T-S) diagram from the 5th edition

of this text, based on Sverdrup et al. (1942), is

included in the online materials along with

a display of T-S diagrams in each 5 latitudelongitude

square (Figures S9.26 and S9.27). It is

useful to consider these diagrams together

with the surface property maps and vertical

sections of Chapter 4. Each water mass introduced

here is considered in more detail in

subsequent sections.

Looking at the overall ranges of properties,

the highest and lowest temperatures are the

29e30 C of tropical surface water and the negative

temperatures of the bottom water from the

Antarctic. (Water at the freezing point is found

on coastal shelves in the North Atlantic, but is

too fresh to appear here.) Highest salinities at

highest temperatures are in the subtropical

surface waters at 11e24 S and 20e30 N. In

oxygen, the ridge that tilts from 200 mmol/kg

at high temperature to 350 mmol/kg at low

temperature is the locus of 100% saturation,

hence surface water. Lowest oxygen occurs

in the low latitude, upper ocean oxygen minimum

zones, resulting from high biological

productivity.

Below the warm tropical surface water, the

nearly linear T-S relation in both the North

Atlantic and South Atlantic is called the Central

Water. This is the main pycnocline of the

subtropical gyre of each ocean. Central Waters

originate from surface waters that subduct

from different locations and have a range of

densities (Sections 4.2.3 and 7.8.5). The North

Atlantic Central Water is saltier than the South

Atlantic Central Water, and is, in fact, the saltiest

Central Water of all five oceans (Figure 4.7).

5 Water characteristics of the Atlantic have been surveyed numerous times, which is advantageous for in situ study of

climate variability. The first basin-wide survey with temperature, salinity, and oxygen measurements from top to bottom

was carried out on the German Meteor from 1925e1927 (Wüst, 1935). A second major survey was carried out in 1957e1958

as part of the International Geophysical Year, intentionally repeating many of the Meteor sections to obtain a direct

comparison of the distribution of water properties after the interval of 30 years. Throughout the 1970s and 1980s, much of

the Atlantic was surveyed for chemical tracers, as well as basic hydrographic properties, along with a number of newly

eddy-resolving sections; all vertical sampling included conductivity, temperature, depth (CTD) profiling as well as bottle

samples. In the 1990s, all of the Atlantic was surveyed again as part of the World Ocean Circulation Experiment (WOCE).

These various experiments are summarized in numerous papers in Siedler, Church, and Gould (2001). Post-WOCE,

hydrographic sampling continues at a high pace to observe the clear changes in deep water properties associated with

surface changes and to follow anthropogenic carbon signals into the ocean.

284

9. ATLANTIC OCEAN

(a)

0

34

SACW

STUW STUW

NACW

37 36

SPMW

1000

34.3

AAIW

34.5

MW

35.5

2000

WSDW

34.7

34.9

UNADW

34.9

LSW

LSW

Depth (m)

3000

34.7

NADW

34.95

35

NEADW

ISOW

4000

34.65

34.9

5000

AABW

(b)

34.7

34.7 34.7

6000

0 2000 4000 6000 8000 10000 12000 14000 km

25.0

Salinity

60°S 40°S 20°S 0° 20°N 40°N 60°N

25.5

36.5

Neutral density (kg/m 3 - 1000)

26.0

26.5

27.0

27.5

28.0

28.5

WSDW

Sea surface

34

34.3

34.7

35

SACW

AAIW

35.5

34.5

AABW

34.7 34.9

0 2000 4000 6000 8000 10000 12000 14000 km

36

NADW

Ocean bottom

36

37

NACW

36.5

MW

35.5

Sea surface

LSW

SPMW

35

ISOW

FIGURE 9.17 Location of most major Atlantic water masses using a meridional salinity section at 20e25 W, as a function

of (a) depth and (b) neutral density (g N ). (White areas at high density are the ocean bottom. White areas at low density (top

of figure) are above the sea surface.) Inset map in (a) shows station locations. Acronyms are within the text and in Table S9.4

in the online supplement. (See also Figure 4.11b.) (World Ocean Circulation Experiment sections A16 and A23)


(a)


(b)

30

25

20

15

10

5

0

10

8

24

22

25

AAIW

23

26

Coastal

AABW

S. Atlantic Central Water

ISOW

DSOW

LSW

34.0 34.5 35.0 35.5 36.0 36.5 37.0 37.5

> 50°N

20°N-50°N

0°-20°N

20°S-0°

50°S-20°S

Salinity

MW

27

N. Atlantic Central Water

NADW

NEADW

28

> 50°N

20°N-50°N

0°-20°N

20°S-0°

50°S-20°S

29

STUW

(c)


30

25

20

15

10

5

0

(d)

> 50°N

20°N-50°N

0°-20°N

20°S-0°

50°S-20°S

Tropical

oxygen minimum

UCDW

MW

EDW (STMW)

NAC SPMW

ISOW

NEADW DSOW

AABW

NE SPMW

NW SPMW

LSW

0 50 100 150 200 250 300 350

Oxygen (mmol/kg)


6

4

2

27

AAIW

Coastal

DSOW

ISOW

28

NADW

LSW

0

AABW

34.0 34.2 34.4 34.6 34.8 35.0 35.2

Salinity

FIGURE 9.18 Potential temperature ( C) versus salinity for (a) full water column, and (b) water colder than 10 C. (c)

Potential temperature versus oxygen for full water column. (d) Station location map. Colors indicate latitude range.

Contours are potential density referenced to 0 dbar. Data are from the World Ocean Circulation Experiment (1988e1997).

This figure can also be found in the color insert.

286

9. ATLANTIC OCEAN

Within the Central Water, subducted high

salinity surface water creates a near-surface

salinity minimum called Subtropical Underwater

(STUW). This is a noisy presence at

around 25 C in this sample-based T-S relation,

but is much clearer in the vertical sections

shown in the following text. STMWs are also

subducted, but they cannot be seen in T-S

because they fit the Central Water T-S relation;

they are more easily seen in oxygen (described

later).

In the North Atlantic, proceeding downward,

the high salinity MW is the salty protrusion in

T-S at mid-latitudes (20e50 N band). Although

warm, around 12 C, MW is so salty that it is

almost as dense as the much colder NSOW.

The denser North Atlantic waters are best

represented in T-S north of 50 N (red points in

Figure 9.18a, b). The LSW is the main salinity

minimum. The set of colder, fresher, less dense

points that extend toward low salinity away

from the LSW are Greenland and Labrador

coastal waters. Northeast Atlantic Deep Water

(NEADW) is the cluster of higher salinity waters

colder than about 3e4 C. Protruding toward

lower salinity from the NEADW are the two

NSOWs: DSOW and Iceland-Scotland Overflow

Water (ISOW). DSOW is fresher and more

oxygenated than ISOW.

The low salinity AAIW (around 4 C and 34.1

psu) and the high salinity NADW (at 2e3 C) are

found in the tropical and South Atlantic. At the

bottom, we see the narrow tail of the AABW.

Oxygen (Figure 9.18c) provides independent

water mass information. The tropical oxygen

minimum was already mentioned. Clusters of

points around 18 C, 11e14 C and 8e11 C indicate

the high-volume mode waters (Hanawa &

Talley, 2001). They are the Eighteen Degree Water

(EDW) (identical to STMW), NAC SPMW,

northeastern Atlantic SPMW (NE SPMW), and

northwestern Atlantic SPMW (NW SPMW).

All have high oxygen saturation.

Farther down, at lower temperature, the high

oxygen LSW and DSOW appear. ISOW does not

have an obvious oxygen extremum due to

entrainment of lower oxygen water as it overflows

into the North Atlantic. At the bottom

(lowest temperature), the higher oxygen

AABW is seen. In these deeper waters, two

features that appear in oxygen but not well in

salinity are the NEADW, which is an axis

toward lower oxygen in the northern latitudes

(red points) at about 3 C, and the low oxygen

UCDW at about 2 C in the eastern South

Atlantic.

9.8.2. Atlantic Ocean Upper Waters

9.8.2.1. Surface Water and Mixed Layer

Surface temperature (Figure 4.1) is highest,

up to 30 C, in a band north of the equator, in

the ITCZ. The equatorial cold tongue is evident

in the east, separated from the South Atlantic’s

colder subtropics by a band of warmer water at

10 S. The Gulf Stream and Brazil Current

subtropical gyres both include poleward intrusion

of warmer waters along their western

boundaries and equatorward swoops of cooler

waters in the east. In the satellite SST image in

Figure 1.1a, the separated Gulf Stream is

evident as a narrow band of warm water,

with the cooler recirculation just to its south.

The NAC and its northwest corner are evident

east of Newfoundland, and the warm waters

of the NAC spread northward toward Iceland

and Scotland. The coldest waters are in the

EGC, and, in the Labrador Sea, in the Davis

Strait and southward in the Labrador Current.

In the southern South Atlantic, the cold Malvinas

Current loops northward and the strongly

eddying Subantarctic and Polar Fronts stretch

eastward.

Atlantic sea-surface salinity (SSS), as in all

oceans, is dominated by the alternation of

regions of net precipitation/runoff with those

of net evaporation (Figure 4.15, Section 4.3).

Lower surface salinity is found in the tropics,

especially beneath the ITZC; low surface salinity

is evident due to runoff from the Amazon and


Congo Rivers as well. Net evaporation in the

subtropics results in maximum surface salinity

there. The northward swing of the fresh Malvinas

loop in the Southern Ocean is apparent. In

the North Atlantic’s subpolar gyre, lowest salinities

are along the coasts of Greenland and

Labrador, and in the slope water region north

of the Gulf Stream. On a global scale, the high

Atlantic SSS, extending far into the Nordic

Seas, reflects the overall higher salinity of the

Atlantic, which facilitates deep water formation

here and not in the much fresher North Pacific.

The basin-wide pattern of sea surface

density (Figure 4.19, Section 4.4) mainly follows

SST, with lowest density in the tropics and

higher density at higher latitudes. Salinity

affects the surface density in the subpolar

North Atlantic compared with the North

Pacific: at a given latitude, the saltier North

Atlantic is significantly denser than in the

North Pacific. Large river outflows (Amazon,

Congo, and Orinoco) also cause the lowest

surface densities in the tropics.

The winter surface mixed layer in the

Atlantic (Figure 4.4a and c, Section 4.2.2) is

markedly thick throughout the subpolar gyre

and into the Labrador Sea; this thick layer is

called Subpolar Mode Water (Figure 9.19b)

(McCartney & Talley, 1982; Hanawa & Talley,

2001). A band of thick mixed layers is also found

just south of the Gulf Stream (EDW or STMW;

Figure 9.19a) (Worthington, 1959). In the South

Atlantic, winter mixed layers are also thicker

farther south near the ACC (SAMW), but

FIGURE 9.19 Mode waters. (a) EDW thickness from all

Argo profiles from 1998e2008. The EDW is defined here by

17 C T 19 C and dT/dZ 0.006 C/m. The small gray

dots in the background indicate profiles without EDW.

(Young-Oh Kwon, personal communication, 2009.) (b) March

mixed layer depth from a data-assimilating model (SODA).

Source: From Schott et al. (2009). (c) Potential temperature

( C) of the late winter mixed layer, shown only where the

mixed layer is more than 200 m thick. This is the SPMW.

Source: From McCartney and Talley (1982).

(c)

70˚

60˚

50˚

40˚

30˚

60˚W 50˚ 40˚ 30˚ 20˚ 10˚ 0˚ 10˚E

70˚

4

3.5

> 17

10

4

5

5

6

7

12 12

13

14

8

9

10

30˚

60˚W 50˚ 40˚ 30˚ 20˚ 10˚ 0˚ 10˚E

11

8

7

60˚

50˚

40˚

288

9. ATLANTIC OCEAN

much thicker winter mixed layers are found at

similar southern latitudes in the Pacific and

Indian Oceans (McCartney, 1977).

9.8.2.2. Central Water and Subtropical

Underwater

Central Water includes the water in the main

pycnocline in each subtropical gyre (Section

9.8.1 and Section 4.2.3). The North and South

Atlantic Central Waters extend to depths of

300 m on either side of the equator, deepen to

600 to 900 m at mid-latitudes, and are somewhat

shallower at the poleward side of their gyres.

The density range of Central Water is set by

the winter surface density within the Ekman

pumping region of the subtropical gyre. The

highest such density in the North Atlantic is

around s q ¼ 27.2 kg/m 3 , which outcrops in

winter around 52 N. The southern edge of

North Atlantic Central Water is the southern

edge of the subtropical gyre, which is nicely

marked by the onset of very low oxygen in the

tropics at about 20 N (Figure 4.11d).

In the South Atlantic, defining the maximum

density of subduction based on the region of

Ekman downwelling is problematic because the

subtropical gyre is connected to the Indian

Ocean’s subtropical gyre over a broad latitude

range between Africa and the ACC. The

maximum outcrop density in the combined South

Atlantic-Indian subtropical gyre is s q ~26.9kg/

m 3 , south of Australia. Within the South Atlantic

proper, west of Africa, the maximum density of

gyre-wide winter outcropping could be as low

as s q ~ 26.2 kg/m 3 . As in the North Atlantic, the

tropical oxygen minimum marks the northern

edge of the subtropical gyre and Central Water.

The STUWs are shallow vertical salinity

maximum, within the upper 100 m. STUWs are

relatively minor water masses in terms of areal

extent, volume, and formation rates (~1e2 Sv;

O’Connor, Fine, & Olson, 2005), but they nicely

illustrate the subtropical subduction process

(Figure S9.28 in the online supplementary material).

They are embedded in Central Water

arising from equatorward subduction from the

subtropical SSS maxima. In the South Atlantic,

STUWs occur between about 13 and 6 S; in the

North Atlantic, the range is 12 Nto20e25 N

depending on longitude. North Atlantic STUW

potential density is s q ~ 25.5 kg/m 3 . South

Atlantic STUW has a larger density range,

centered at about s q ¼ 24 to 24.5 kg/m 3 in the

western South Atlantic but denser in the eastern

South Atlantic.

The salinity contrast between the STUW

salinity maximum and the underlying fresher

water can be large, leading to favorable conditions

for salt fingering (Section 7.4.3.2). Schmitt,

Perkins, Boyd, and Stalcup (1987) observed

multiple stepped layers of 5e30 m thickness,

indicative of salt fingering, beneath the salinity

maximum east of Barbados. The layers had

remarkable coherence over hundreds of kilometers

in the horizontal and remarkable persistence

in time.

9.8.2.3. Mode Waters

Mode Waters are layers that, in terms of isopycnal

spacing, are relatively thick compared

with surrounding waters on the same isopycnals

and in the vertical (Section 4.2). The North

Atlantic has several STMWs and its SPMW.

The North Atlantic’s principal STMW is

found south of the Gulf Stream; it is also called

Eighteen Degree Water because of its typical

temperature. EDW is the archetype of all

STMW (Worthington, 1959; Masuzawa, 1969).

It can be seen on any vertical section crossing

the Gulf Stream (Figure 9.7). EDW is a permanent

feature, with observations dating back to

the Challenger expedition in 1873 (Worthington,

1976). It has relatively homogeneous properties

centered at about 18 C, 36.5 psu, and s q ¼

26.5 kg/m 3 , with some spatial and temporal

variability. EDW originates in thick winter

mixed layers adjacent to the Gulf Stream and

within the tight recirculation gyre (Section

9.3.2). The mixed layer thickness can reach to

more than 500 m (Figure 9.19a). EDW subducts


southward into the western subtropical gyre,

creating a low stability, subsurface layer far

from the Gulf Stream, throughout most of the

Sargasso Sea. The estimated EDW formation

rate is 2e5 Sv (Kwon & Riser, 2004). This formation

is a conversion from warmer, lighter Gulf

Stream waters to the characteristic 18 C of EDW.

Madeira Mode Water, on the southern flank of

the Azores Current front, is another STMW and

is clearly separate from, and weaker than, the

EDW. It is somewhat cooler (16e18 C), saltier

(36.5e36.8 psu), and denser (s q ¼ 26.5e26.8

kg/m 3 ) than EDW (Siedler, Kuhl, & Zenk, 1987;

New et al., 2001). Its formation rate and volume

are much smaller than those of EDW. Whereas

EDW is a year-round water mass, Madeira

Mode Water is eliminated every year. This difference

can be expressed in terms of their residence

times: the EDW has a residence time of 3e5years

(which results in a permanent reservoir), whereas

the Madeira Mode Water’s residence time is

about 6e9months.

The South Atlantic’s subtropical gyre has

a number of different mode waters related to the

complex frontal system associated with the Brazil

and Malvinas Currents and the SAF (Tsuchiya

et al., 1994). Provost et al. (1999) documented

three STMWs in the western South Atlantic


The coldest and densest (12e14 C, 35.1

psu, s q ¼ 26.7 kg/m 3 ) is on the north side of the

SAF. It is actually the warmest form of SAMW

(Chapter 13), but it subducts into the South Atlantic’s

subtropical gyre like a typical STMW. The

second mode water, at ~13.5 C, 35.3 psu, s q ¼

26.6 kg/m 3 , is the principal STMW associated

with the separated Brazil Current Front. The third

STMW is lighter, warmer, and less extensive.

Returning to the North Atlantic, the most

significant mode water in terms of volume and

impact on internal ocean properties is the

SPMW, which is found throughout the subpolar

region (Figure 9.19b, c and Figure S9.30 in the

online supplementary material). SPMW is an

important part of the upper ocean water that

feeds into the NADW, in both the Nordic Seas

and the Labrador Sea. SPMW (as depicted originally

in McCartney & Talley, 1982) is a broad

water mass arrayed around the cyclonic gyre,

essentially identical with the winter surface

mixed layer. It is generally more than 400 m

thick, and is much thicker on the Iceland-Faroe

Ridge and in the Irminger and Labrador Seas.

The warmest, lightest SPMW (14 C, s q ¼ 26.9

kg/m 3 ) is found east and south of the NAC. As

the NAC moves eastward across the North

Atlantic, its SPMW becomes progressively colder

and denser, reaching about 11 C, s q ¼ 27.2 kg/

m 3 near the British Isles where the NAC bifurcates.

Much of this lightest part of the SPMW

subducts southward into the subtropical gyre,

behaving as an STMW of the NAC.

The NAC turns northeastward, split into at

least three permanent meandering fronts, each

with its own progression of SPMWs on its

eastern (warm) side (Brambilla & Talley, 2008).

These SPMWs do not subduct, but instead

continue in the surface layer, becoming progressively

colder, fresher, and denser toward the

north. The branches east of the Reykjanes Ridge

(in the Iceland Basin and Rockall Trough) carry

SPMW that cools to 8 C by the Iceland-Faroe

Ridge. This SPMW enters the Nordic Seas via

the Norwegian Atlantic Current as part of the

Atlantic Water that eventually is transformed

to NADW (Chapter 12).

The third NAC branch is the Irminger Current,

west of the Reykjanes Ridge (Section 9.3.5). Its

SPMW progresses toward even colder, fresher,

and denser properties around the Irminger Sea,

following the East and West Greenland Currents

into the Labrador Sea. This SPMW is a source of

the LSW (and Irminger Sea Water), which at

about 3e3.5 C, form the upper part of the

NADW (see the next section).

9.8.3. Intermediate Waters

Below the surface layer and pycnocline, at

intermediate depths of about 500e2000 m, the

290

9. ATLANTIC OCEAN

Atlantic Ocean includes three intermediate

water masses, usually identified by vertical

salinity extrema. These are the low salinity

LSW in the north, the high salinity MW in the

subtropical North Atlantic, and the low salinity

AAIW in the South Atlantic and tropical

Atlantic (summary map in Figure 14.13).

These intermediate water masses are characterized

by geographically limited source

regions, unlike the upper ocean water masses.

LSW forms by deep convection in the central

western Labrador Sea, one of the few sites in

the global ocean of such convection (Marshall &

Schott, 1999). AAIW enters the South Atlantic at

the Brazil-Malvinas Current confluence; its low

salinity source is the freshest SAMW of the

southeastern Pacific (Section 13.5). MW enters

the North Atlantic as a dense overflow through

the Strait of Gibraltar (Section S8.10.2 in the

online supplemental material).

9.8.3.1. Labrador Sea Water

LSW is the intermediate depth water mass of

the subpolar and western subtropical North

Atlantic. LSW is characterized by (1) a lateral

and vertical salinity minimum in the subpolar

North Atlantic; (2) a lateral and vertical

minimum in potential vorticity (maximum in

layer thickness) in the subpolar North Atlantic

and subtropical western boundary region; and

(3) a lateral and vertical extremum in dissolved

gases that mark recent ventilation, such as

oxygen and CFCs.

These LSW characteristics result from its convective

formation process and young age relative

to other waters at the same depth. LSW

forms in the Labrador Sea, between Labrador

and Greenland, where winter mixed layers

exceed 800 m and can reach to 1,500 m depth

(Figure 9.20a and supplementary Figure S9.31).

(The deep and bottom waters of the Labrador

Sea are denser NSOW and NADW, which are

never penetrated by the Labrador Sea’s deep

convection.) The LSW source water is mostly

SPMW entering the Labrador Sea from the

Irminger Sea, and includes fresh surface water

from Baffin Bay through the Davis Strait. The

deep winter mixed layers within the Labrador

Sea are capped by lower density in spring, and

the thick layers collapse somewhat thereafter

forming the relatively uniform and thick layer

of LSW. The resulting thick layer of cold, fresh,

dense, oxygenated LSW appears in the leftmost

panels of each property in Figure 9.20. During

the year (1997) of these observations, the new

LSW had properties of 2.9e3.0 C, 34.84 psu,

and s q ¼ 27.78 kg/m 3 . LSW properties are variable

(Chapter S15 in the online supplemental

material); the temperature minimum (<2.8 C)

at about 2000 m in the figure is remnant LSW

from more vigorous convection at a historically

low temperature several years prior.

The new layer of LSW moves southward out

of the Labrador Sea following the Labrador

Current (Figure 9.15a), as evident in salinity,

oxygen, CFCs, and potential vorticity, all of

which have extrema in the LSW (Figure 9.21

and also supplementary Figure S15.4). Upon

reaching the northwest corner of the NAC,

part of the LSW turns eastward with the NAC

and part continues on southward past Flemish

Cap. The LSW that moves eastward mostly

turns northward into the Irminger Sea, while

part moves on eastward through the Charlie

Gibbs Fracture Zone into the eastern subpolar

gyre and northward into the Iceland Basin and

into the Rockall Trough. Because of the shorter

path to the Irminger Sea, LSW is fresher and

more oxygenated there than in the Iceland Basin

and Rockall Trough (Figures 9.20 and 9.21).

The LSW in the Labrador Current turns westward

as part of the DWBC into the slope water

region north of the Gulf Stream. It then moves

southward inshore of and beneath the Gulf

Stream, which can be seen in the high CFCs at

1000e1700 m at the western boundary at 24 N;

elevated oxygen and reduced salinity are

also found here (Figure 9.22). Neither is straightforward:

Lagrangian floats that should track the

southward progression of LSW do not make the


(a)

0

500

1000

3

1500

2000

2500

1.8

3000 2

3500

0 500 km

(c)

0

500

1000

1500

2000

2500

3000

2.2

2

3.6

2.4

2.2

2

300

290

1.6

295

295

3.2

3

2.9

2.8

4

3.8

2.8

2.6

290

3.4

300

295

295

285

290

295

9

8

9

8

9

8

295

29

3500

0 500 km

3

4

3.2

2.9

2.8

58°W 55° 51°W

2

4

5

5

2

2.2

2.4

0 500 1000 1500 km

00

295

3.2

35°W 30° 25° 20° 15° 10°W

285

2.8

2

1.6 1.8

280

3

2.8

2.6

295

285

4

290

3.8

3.6

275

3

2.

270

240

260

280

5

7

3.4

3

6

260

9

8

230

265

8

(b)

230 1000

240

245

255

1500

250

265

5

275

2000

270

24 23

22 21 20 19

18

17 16 14 12 10

8

275

2

265 2500 24

260

0

500

1000

1500

2000

2500

Labrador Sea

3500

0 500 km

0

500

3000

3

3

Labrador

34.9

3000

34.82

34.84

34.84

27.9

27.92

Oxygen s

285

255

3500

q

0 500 1000 1500 km 0 500 km 0 500 1000 1500 km

34.86

34.88

34.9

34.84

34.84

34.91

58°W 55° 51°W

8

Greenland

34.88

34.9

34.88

34.88

Irminger Sea

34.9

34.86

34.9

34.95

34.91

Reykjanes Ridge

35.2

35.1

34.92

34.92

34.93

34.94

34.96

34.9

34.91

34.92

96

34.94

34.95

34.9134.95

Potential temperature Salinity

(d)

27.9

27.

27.74

27.76

27.78

27.5

27.6

27.7

27.72

27.8

27.82

27.84

27.88

0 500 1000 1500 km

27.86

35

34.9

Iceland

Basin

35.3

Rockall Plateau

35.4

35.2

Rockall Trough

35°W 30° 25° 20° 15° 10°W

27.6

27.8

27.84 27.82

7.9

27.88

27.76

27.8

27.5

27.82

27.84

27.9

27.7

27.72

27.74

27.9

27.4

27.3

27.88

0

0

500

1000

1500

2000

2500

3000

3500

500

1000

1500

2000

2500

3000

3500

58°W 55° 51°W

35°W 30° 25° 20° 15° 10°W

58°W 55° 51°W

35°W 30° 25° 20° 15° 10°W

FIGURE 9.20 Subpolar North Atlantic at about 55 N from May to June, 1997. (a) Potential temperature ( C), (b) salinity,

(c) oxygen, and (d) potential density (s q ) in the Labrador Sea (left side) and from Greenland to Ireland (right side). This

figure can also be found in the color insert. (World Ocean Circulation Experiment sections AR7W and A24.)

292

9. ATLANTIC OCEAN

FIGURE 9.21 LSW. (a) Salinity at the LSW potential vorticity minimum. Dark curve is the limit of the PV minimum;

salinity on an intersecting isopycnal is shown south and east of this limit. Source: From Talley and McCartney (1982). (b)

Chlorofluorocarbon-11 (pmol/kg) in the upper LSW layer, at s q ~ 27.71 kg/m 3 . Figure 9.21b can be found in the color insert.

Source: From Schott et al. (2009) and from Kieke et al. (2006).

turn (Bower et al., 2009), while the interaction

with the Gulf Stream is complex, resulting in

significant entrainment of Gulf Stream waters

(Pickart & Smethie, 1993).

Estimated LSW production rates vary from

2 to 11 Sv; the most recent is 3 to 9 Sv based on

the CFC inventory in the subpolar North

Atlantic (Kieke et al., 2006). The southward

export rate in the DWBC, as estimated at 24 N

(Figures 9.16 and 9.22), is around 6e8 Sv, which

is a significant fraction of the total NADWexport

of 15e20 Sv.

9.8.3.2. Mediterranean Water

The North Atlantic also contains a high

salinity water mass, the MW (also called Mediterranean

Overflow Water), at about the same

depth and density range as LSW. Salinity maps

representing the MW were shown in Figure 6.4,

at constant depth, on an isopycnal, and at the

core of maximum salinity. MW enters the

Atlantic as dense water at the Strait of Gibraltar

(Figure 9.23a,b). The total outflow is about 0.7

Sv at 38.4 psu and s q ¼ 28.95 kg/m 3 (Section

S8.10.2 in the online supplemental material).

The overflow plunges downward, entraining

ambient water that reduces its salinity and

density. It follows the topography to the right,

turning northward into the Gulf of Cadiz where

it splits into two cores (Figure 9.23c). It reaches

its neutral buoyancy and depth of 1000e1500

m by about Cape St. Vincent (Candela, 2001).

As the overflow encounters the sharp northward

bend in topography at Cape St. Vincent

and other topographic features along the Iberian

peninsula, anticyclonic eddies of nearly pure

MW are spun off (Bower, Armi, & Ambar, 1997;

Richardson, Bower, & Zenk, 2000; Candela,

2001). These “Meddies” propagate southwestward

and westward into the North Atlantic,

retaining their coherence and high salinity for

enormously long distances and over 2e3 years


They are entirely subsurface. At formation,

the Meddies are small d about 9 km diameter.

After aging and propagation, their radii become

20e100km, with a thickness of about 650 m, and

centered at about 1000 m depth. Approximately


(a)

Depth [m]

80°W 75°W 70°W 65°W 60°W 55°W 50°W 45°W 40°W 35°W 30°W 25°W 20°W

0

25

20

20

15

15

500

10

10

1000

1500

2000

2500

3000

3500

Florida Strait

5

4.4

5

4.4

4

4

3.4

3.4

3

3

2.4 2.4

(b)

0

500

1000

1500

2000

2500

3000

3500

80°W 75°W 70°W 65°W 60°W 55°W 50°W 45°W 40°W 35°W 30°W 25°W 20°W

36.50 36.00

35.70 35.60

35.50

35.30

35.20

Florida Strait

35.20

35.08

35.04

35.00

34.98

34.96

34.95

34.94

34.92

34.92

34.91

4000

4500

5000

5500

(c)

Depth [m]

2

1.6

Distance [km]

2

6000

6000

0 500 1000 1500 2000 2500 3000 3500 4000 4500 5000 5500 6000 6500 0 500 1000 1500 2000 2500 3000 3500 4000 4500 5000 5500 6000 6500

80°W 75°W 70°W 65°W 60°W 55°W 50°W 45°W 40°W 35°W 30°W 25°W 20°W

0

500

1000

1500

2000

2500

3000

3500

4000

4500

5000

5500

6000

Bahamas

Florida Strait

Bahamas

265

240

260

265

250

250

190

180

Mid-Atlantic

Ridge

160

0 500 1000 1500 2000 2500 3000 3500 4000 4500 5000 5500 6000 6500

Distance [km]

160

Mid-Atlantic

Ridge

200

240

220

230

235

140

>

245

245 245

>

Africa

Pot.

temp.

Africa

Oxygen

4000

4500

5000

5500

(d)

0

500

1000

1500

2000

2500

3000

3500

4000

4500

5000

5500

6000

Bahamas

34.90

34.88

34.86

Distance [km]

80°W 75°W 70°W 65°W 60°W 55°W 50°W 45°W 40°W 35°W 30°W 25°W 20°W

Florida Strait

Bahamas

>

>

0.5

0.2

0.6

0.4

2.0

0.5

0.05

<

0.2

0.005

0.02

0.01

2.0

1.0

0.1

Mid-Atlantic

Ridge

Mid-Atlantic

Ridge

0 500 1000 1500 2000 2500 3000 3500 4000 4500 5000 5500 6000 6500

Distance [km]

0.05

0.01

0.005

>

>

0.005

>

0.005

>

0.01

34.90

0.01

0.005

Africa

Salinity

Africa

CFC-11

FIGURE 9.22 Subtropical North Atlantic at 24 N from July to August 1992. (a) Potential temperature ( C), (b) salinity,

(c) oxygen (mmol/kg), and (d) CFC-11 (pmol/kg) at 24 N. This figure can be found in the color insert. (World Ocean

Circulation Experiment section A05). Adapted: From WOCE Atlantic Ocean Atlas; Jancke, Gouretski, and Koltermann (2011).

15e20 may be formed each year (Figure S9.33 in

the online supplementary material). They may

carry up to 50% of the MWinto the North Atlantic.

Advection of high salinity MW into

the subtropical North Atlantic forms the

characteristic “Mediterranean salt tongue”

(Figure 6.4). Because the feature is so striking,

it is tempting to jump to simple conclusions

about the associated circulation. However, the

salt tongue does not mirror the circulation,

294

9. ATLANTIC OCEAN

FIGURE 9.23 (a) Temperature, salinity, and potential density profiles near the strait sill in spring. Source: From Bray,

Ochoa, and Kinder (1995). (b) Potential density at 6 30’W just west of the Strait of Gibraltar, in spring. Source: From Ochoa and

Bray (1991). (c) Outflow pathways of the MW. Source: From Zenk (1975).

which is very weak, but instead is associated

with eddy diffusivity related to planetary

wave and eddy-like motions (Richardson &

Mooney, 1975; Spall, Richardson, & Price,

1993). The associated horizontal eddy

diffusivities have reasonable magnitudes of 8

to 21 10 6 cm 2 /sec.

MW injects high salinity down to intermediate

depth in the North Atlantic, contributing to the

characteristic high salinity of the NADW (Figure


9.17). On the other hand, open ocean subtropical

evaporation in the North Atlantic dominates in

establishing the North Atlantic as the saltiest

ocean, with the Mediterranean contributing

about 30% of this enhancement (Talley, 1996b).

9.8.3.3. Antarctic Intermediate Water

AAIW is the third major intermediate water of

the Atlantic Ocean. It is the low salinity layer at

about 1000 m in the South Atlantic and tropical

Atlantic (Figures 9.17, 9.18, and 14.13). The

AAIW’s salinity minimum originates near Drake

Passage where it is related to the densest, coldest,

freshest SAMW (Chapter 13). The AAIW’s

northern boundary mostly coincides with the

southern boundary of MW, at about 20 N

(Figure 14.13 and Figure S9.34 and S9.35 in the

online supplementary material). AAIWis fresher

than LSW since its Southern Ocean source waters

are fresher than the LSW’s subpolar source

waters in the more evaporative North Atlantic.

In the southern South Atlantic, the AAIW

salinity minimum is slightly denser than s q ¼

27.1 kg/m 3 (4 C, 34.2 psu). In the tropics, the

AAIW layer is eroded from above and is subject

to diapycnal diffusion that increase its “core”

density, potential temperature, and salinity to

about s q ¼ 27.3 kg/m 3 , 5 C, and 34.5 psu

(Talley, 1996a).

The net northward transport of AAIW into

the Atlantic is estimated at 5e7 Sv(Figure 9.16).

In the South Atlantic, AAIW is advected eastward

away from the Malvinas-Brazil Current

confluence and then northward and westward

around the anticyclonic subtropical gyre. It

returns to the South American coast and enters

the North Brazil Current System. It is advected

eastward near the equator as part of the zonally

elongated equatorial current system. In the

tropics, the AAIW joins vertically with UCDW

(Tsuchiya, Talley, & McCartney, 1994). This

complex moves northward into the Gulf Stream

System and NAC, where the remnants of

AAIW/UCDW are marked by elevated nutrients

rather than low salinity (Tsuchiya, 1989).

9.8.4. Deep and Bottom Waters

The deep and bottom waters of the North

Atlantic consist of the NADW and its precursor

components formed in the North Atlantic, and

AABW and CDW formed in the Southern

Ocean. These were introduced in Section 4.3.4

and figure prominently in the global overturning

circulation of Chapter 14. The intermediate

water components of NADW were already

introduced (Section 9.8.3). Here we discuss the

NSOW component, then AABW, and end by

describing the NADW as a whole, especially in

the tropical and South Atlantic where the individual

North Atlantic source waters meld into

the single water mass that is exported from the

Atlantic to the other oceans.

9.8.4.1. Nordic Seas Overflow Waters

The densest new bottom and deep waters in

the North Atlantic originate in the Nordic

Seas. These are discussed in Section 12.6;

reviews can be found in Dickson and Brown

(1994) and Hansen and Østerhus (2000). There

are three sills in the Greenland-Scotland ridge,

all with important dense overflows: (1) the

Denmark Strait between Greenland and Iceland

with flow into the Irminger Sea, (2) the Iceland-

Faroe Ridge with flow into the Iceland Basin,

and (3) the Faroe-Shetland Channel. The last

includes two routes: the Faroe Bank Channel

into the Iceland Basin and the Wyville-Thomson

Ridge to the northern Rockall Channel. The

dense waters flowing through the three channels

are referred to as Denmark Strait Overflow

and Iceland-Scotland Overflow Water, the latter

containing waters from both sills east of Iceland.

We can refer to the overflows collectively as the

Nordic Seas Overflows, although this nomenclature

is not universal.

The mean transports through the three

straits are 3 Sv (DSOW), 0.5e1 Sv (Iceland-

Faroe), and 2e2.5 Sv (Faroe-Shetland Channel),

for a total of 6 Sv (Figures 9.15a and 12.20).

Most of the Faroe-Shetland overflow goes

296

9. ATLANTIC OCEAN

through the Faroe Bank Channel, with just

a few tenths of Sverdrups through the

Wyville-Thomson Ridge. The overflow waters

are separated from the northward-flowing

surface waters by the isopycnal s q ~

27.8 kg/m 3 (Figure 9.24a). The overflow layers

originate in several water masses within the

Nordic Seas, making a single T-S characterization

of their properties impossible; temporal

variability in the mix of waters in the straits

creates variability in the overflow properties

(Macrander et al., 2005). DSOW properties are

(a)

Pressure / dbar

0

100

200

300

400

Poseidon P262 (Jul. 2000): Section across Denmark Strait sill

27.7

27.9

27.6

27.7

27.8

Greenland

500

66

C

28

600

B

65

Iceland

64

700

A

63

35 30 25 20

800

Contours at [26 27 27.2:0.1:28.1] kg/m³

Longitude / °W

−60 −40 −20 0 20 40 60 80

Distance from sill / km

27.6

26 26.2 26.4 26.6 26.8 27 27.2 27.4 27.6 27.8 28

Sigma Theta / kg/m , p ref = 0 dbar

FIGURE 9.24 (a) Potential density in Denmark Strait.

The heavy contour marks the upper bound on the overflow

layer in the strait. Source: From Macrander et al. (2005). (b)

Potential temperature ( C) crossing the Iceland-Faroe Ridge.

Source: From Hansen and Østerhus (2000).

KG5

68

67

Latitude / °N

0.18 C, 34.88 psu, s q ¼ 28.02 kg/m 3 to

0.17 C, 34.66 psu, s q ¼ 27.82 kg/m 3 (Tanhua,

Olsson, & Jeansson, 2005). ISOW properties in

the Faroe Bank Channel are 0.5e3 C, 34.87e

34.90 psu, and s q ¼ 28.02 to 27.8 kg/m 3 (Hansen

& Østerhus, 2000). The upper ocean water

masses south of the ridge are separated from

those north of the ridge by a strong front; east

of Iceland this Iceland-Faroe Front is associated

with concentrated eastward flow that feeds the

northward Norwegian Atlantic Current.

The overflows at each of the sills are dense

and plunge down their respective slopes toward

the deep North Atlantic (Figure 9.24). The overflows

form eddies as they plunge and turbulently

entrain the water masses they pass

through. The entrained waters include SPMW,

LSW, and ambient deep waters. Therefore the

overflow properties change rapidly as their

transport increases. Along the zonal section

south of the sills (Figure 9.20), the Irminger

Sea and Iceland Basin overflows are obvious in

the dense layers banked to the west; in Rockall

Trough, overflow water is much weaker but still

present on the western side near the bottom.

DSOW in the Irminger Basin is markedly fresher

and more oxygenated than ISOW (see also

Figure 9.18), mainly because DSOW entrains

newer LSW. Once the DSOW and ISOW plumes

equilibrate and begin to move further into the

North Atlantic, their maximum densities are

reduced to about 27.92 s q (46.1 s 4 ; Figure 9.20).

ISOW circulates westward through the

Charlie Gibbs Fracture Zone in the Reykjanes

Ridge and joins the DSOW in the southern

Irminger Sea (Figure 9.15a); the combined

NSOW flows cyclonically around Greenland

into the Labrador Sea and then out to the south

beneath the Labrador Current. At this point,

the NSOW is the denser part of the newly forming

DWBC (Section 9.6.2). This dense layer

crosses under the Gulf Stream relatively easily

compared with the LSW (Pickart & Smethie,

1993) and is marked by high oxygen and

high CFCs at the western boundary at 24 N

(Figure 9.22). Here the DWBC waters are usually

referred to as NADW and the portion associated

with NSOW is referred to as Lower North

Atlantic Deep Water (LNADW; Section 9.8.4.3).


9.8.4.2. Antarctic Bottom Water

The densest water in most of the Atlantic originates

in the Southern Ocean south of the ACC. 6

The very densest Antarctic waters in the Atlantic

sector, created by brine rejection in the Weddell

Sea, cannot escape northward past the complex

topography (Chapter 13; Mantyla & Reid,

1983; Reid, 1994). Nevertheless the water that

does escape is often referred to as Antarctic

Bottom Water, and we follow this convention.

AABW is the water colder than about 2 C and

fresher than about 34.8 psu along the full-

Atlantic meridional section (Figures 4.1a and

b and 9.17). Potential temperature is <0 Cin

the south. In the T-S relation (Figure 9.18),

AABW is the coldest tail, stretching down to

less than 0 C. The northward progress and

modification of AABW are severely constrained

by deep topography (Figure 14.14). The coldest

AABW fills the Argentine Basin in the southwestern

South Atlantic. It moves northward

through the constricted Vema Channel into the

Brazil Basin, where AABW colder than 0 Cis

present only in the DWBC along the coast up

to about 15 S(Figure 9.25). The AABW temperature

and salinity increase northward, due to

downward diffusion from the overlying

warmer, saltier NADW. AABW oxygen also

paradoxically increases northward, which is

further evidence of downward diffusion from

the highly oxygenated NADW.

At the equator, the DWBC carrying AABW

splits into an eastward flow that crosses the

MAR through the Romanche Fracture Zone,

and a northward flow that crosses the equator.

The eastward branch is joined by NADW, turns

back southward in the eastern tropical Atlantic,

FIGURE 9.25 Salinity at about 28 S in the western South

Atlantic, with water masses labeled. Source: From Hogg,

Siedler, and Zenk (1999).

and fills the abyssal northeastern South

Atlantic from the north (Figures 14.14 and

14.15). (The Walvis Ridge blocks direct northward

flow into the eastern South Atlantic

from the south.) The northward branch of

AABW shifts to the western flank of the MAR

(Section 9.6). Part of it crosses the ridge through

the Vema Fracture Zone at 11 N, where it is one

source of the northeastern North Atlantic’s

abyssal water (van Aken, 2000). Most of the

AABW continues northward up to the latitude

of Bermuda. At 66 W and 24 N, the AABW is

still apparent as water colder than 1.8 C

mounded to the east toward the MAR (Figures

9.15 and 9.22).

6 While the NSOWs are denser than the bottom waters formed in the Antarctic, intense entrainment of lighter waters as the

NSOW plunges over the sills into the North Atlantic reduces the density of the equilibrated NSOW.

298

9. ATLANTIC OCEAN

9.8.4.3. North Atlantic Deep Water

NADW is the prominent layer of high

salinity, high oxygen, and low nutrients

between about 1500 and 3500 m depth found

through the length of the Atlantic (Figures

4.11, 4.22a, b, and 9.17). We have already examined

the North Atlantic sources of NADW in the

Nordic Seas, the Labrador Sea, and the Mediterranean

Sea. In the subpolar and subtropical

North Atlantic, these waters are easily distinguishable

(NSOW, LSW, and MW). Most

narrowly, the term “NADW” is used where

these source water masses become less easily

distinguished, beginning in the subtropical

North Atlantic’s DWBC and in the tropical

Atlantic. However, it is also appropriate to refer

to the whole complex as NADW in more generalized

water mass studies (e.g., in global overturning

schematics such as Figure 14.11); in

paleoceanography, the balance of source waters

changes dramatically over millennial timescales,

so it is useful to refer to the NADW as

a whole rather than focus on its individual

parts.

NADW is not the only water mass in its depth

range. The fresher CDW moves northward

into the South Atlantic from the Southern

Ocean, as seen in salinity on an isopycnal that

lies at about 2500 m (Figure 9.26) and the circulation

at 2500 m (Figure 9.14a). However,

FIGURE 9.26 Salinity on the isopycnal

s 3 ¼ 41.44 kg/m 3 (referenced to 3000 dbar),

which lies at approximately 2500 m depth.

Source: From Reid (1994).


NADW dominates in terms of net volume transport

(Figure 9.16). In the North Atlantic, high

salinity in the eastern subtropics in Figure 9.26

is due to downward diffusion of salt from the

overlying Mediterranean salt tongue. In the

subpolar North Atlantic, the saltier ISOW

(>34.98 psu) is seen in the east and the fresher

DSOW (~34.94 psu) in the west. The fresher

DSOW spreads southward along the western

boundary, and also spreads eastward toward

the Mid-Atlantic Ridge (MAR) at about 47 N.

In the northeastern North Atlantic, between

the high salinity ISOWand high salinity Mediterannean

tongue, the lower salinity on the isopycnal

in Figure 9.26 is the NEADW (van Aken,

2000). While it is a lateral salinity minimum,

NEADW is also a vertical salinity maximum

(Figure 9.17) and a vertical oxygen minimum

(Figure 9.18c). NEADW is a mixture of local

abyssal and intermediate water masses, with

high salinity from both the ISOW and Mediterranean

salt tongue, and low oxygen from both age/

respiration and northward advection of the

Mediterranean tongue. This contrasts with the

high oxygen ISOW. The underlying deep water

(containing modified AABW) and overlying

LSW also contribute (van Aken, 2000).

In the subtropical North Atlantic, the southward

spread of NADW in the DWBC is marked

by high oxygen and high CFCs (Figures 9.22 and

9.15). At the western boundary at 24 N, the

NADW includes high oxygen at 2000e5000 m

and two striking maxima in CFCs, at 1500 m

and 3500 m (Figure 9.22). The upper

CFC maximum derives from the Labrador Sea

(Bryden et al., 1996; Rhein, Stramma, & Send,

1995). The deeper CFC maximum is coincident

with the deep high oxygen layer of the Lower

NADW (LNADW), which is mostly NSOW.

NADW continues southward in the DWBC,

splitting at the equator into eastward and southward

flow. In the tropics, it is traditional to

distinguish between Upper (UNADW), Middle

(MNADW), and LNADW (Wüst, 1935), which

can be seen at the equator on the 25 W section

(Figure 4.11): UNADW is a salinity maximum

(about 1700 m), MNADW the upper oxygen

maximum (2500 m), and LNADW a separate,

deeper oxygen maximum (3500 m). Only the

LNADW has a simple correspondence with

the upstream sources, with the NSOW (Figures

9.22 and 9.27).

The equatorial UNADW salinity maximum

results from low salinity AAIW cutting into

the top of the NADW, leaving a salinity

maximum at its top, which is deeper and denser

than MW. The MNADW oxygen maximum is

much deeper than the original high oxygen

LSW, which can be seen by comparing the tropical

oxygen and CFC maxima (Figures 4.11 and

9.27; Weiss, Bullister, Gammon, & Warner,

1985; Rhein et al., 1995; Andrié et al., 1999;

Chapter 3). As at 24 N, equatorial CFCs (in

observations taken from 2003e2005) contain

two maxima that directly reflect northern North

Atlantic sources. The deeper CFC maximum is

the same as the Lower NADW oxygen

maximum, deriving from NSOW. The upper

CFC maximum (1000e1500 m) derives from

LSW, and is shallower than the MNADW

oxygen maximum, which is depressed to

greater depth (>1500 m) because of high

consumption of oxygen in the upper layer of

the tropical North Atlantic (Weiss et al., 1985)

whereas CFCs are biologically inert.

Because CFCs have time-dependent surface

sources, they are useful markers of the invasion

of high latitude waters (Figure 9.27). The first

equatorial CFC observations in the 1980s

showed the arrival of LSW as a blob of nonzero

CFCs at 1500 m. By the time of the second

full set of CFC observations in 2003e2005, this

LSW maximum was greatly enlarged and the

LNADW (NSOW) was also marked by a CFC

maximum. The CFC minimum at about

2700 m is mostly associated with the oxygen

minimum between the MNADW and LNADW.

The oxygen/CFC minimum results from

upwelling of AABW and older LNADW in the

eastern tropics (Friedrichs et al., 1994).

300

9. ATLANTIC OCEAN

0

500

1000

1500

2000

2500

3000

3500

4000

4500

5000

5500

6000

4

1.8

0.2

0.1

0.03

0.02

0.05

0.1

3

0.03

0.05

1.6

Rio Grande

Plateau

CFC-11 1988-89

1

2 2

0.5

0.1

0.04

Mid-Atlantic

Ridge

50°S 40° 30° 20° 10°S 0° 10°N 20° 30° 40° 50° 60°N

0.03

0.02

0.05

0.04

0.2

0.1

0.03

1

3

1.6

0.6

0.5

0.3

1.

1

Rockall Plateau

Iceland Basin

1.6

1.8 2

500

1

4 3

1000

.

1500 0.2

2000

0.1

2500

3000

3500 0.2

0.5

4000

0.3

4500

0.1

0.15

5000

5500

6000

CFC-11 2003-2005

3.5

3

2.5 1 0.

2.5

0.2

2

2

0.1

0.1

1

0.1

0.05

0.2

2.5

2

2.

0.05

0.02

1

0.5

0.1

0.02

2

0.02

0.1

0.05

0.03

0.050.04

0.05

50°S 40° 30° 20° 10°S 0° 10°N 20° 30° 40° 50° 60°N

FIGURE 9.27 Chlorofluorocarbon-11 (pmol/kg) along 20e25 W from (a) 1988e1989 and (b) 2003e2005. Section location

is shown in the inset map. Sample locations are indicated by small dots on the plots.

2.5

CLIMATE AND THE ATLANTIC OCEAN 301

In the South Atlantic, the NADW moves

southward in the DWBC to 25 S in the Brazil

Basin. Here there is an eastward breach of

NADW marked at 25 W by higher salinity and

higher oxygen, and even non-zero CFC-11 in

2005 (Figures 4.11b, d and 9.27; Tsuchiya et al.,

1994). These three maxima are nearly coincident

in depth, without the complicated equatorial

layering: the NADW here is becoming the

more homogenized single layer of high salinity

and oxygen that exits the Atlantic. The

NADW, whose southernmost boundary is at

35 S, moves eastward across the MAR, which

is a formidable mixing barrier. East of the ridge,

NADW has markedly lower oxygen and

salinity. It gathers in a broad band of about

1000e1500 km width around the southern end

of Africa. Part of it moves northward into the

Indian Ocean in a broad DWBC underneath

and offshore of the Agulhas (Chapter 11). The

remainder joins the eastward ACC, where it

provides the high salinity core for the LCDW

(Chapter 13).

9.9. CLIMATE AND THE

ATLANTIC OCEAN

Atlantic climate research tends to be focused

on decadal and longer term variability centered

on the northern North Atlantic’s deep-water

formation processes and on sea ice processes

in the Nordic Seas and Arctic (Chapter 12).

This is because the mean ventilation age of

northern North Atlantic deep waters is on the

order of decades or less, with associated

measurable variability. However, climate variability

at all timescales from interannual to

decadal, centennial, and millennial affects all

regions of the Atlantic. Trends related to climate

change (anthropogenic forcing) have also been

documented.

All of the text, figures, and tables relating to

Atlantic Ocean climate variability are located

in Chapter S15 (Climate Variability and the

Oceans) on the supplemental Web site for

this textbook. The chapter describes tropical

Atlantic climate variability: (1) the Atlantic

Meridional Mode (AMM), which is a crossequatorial

mode; (2) Atlantic Niño, which is

a zonal equatorial mode dynamically similar

to El-Niño-Southern Oscillation (ENSO) with

its tropical Bjerknes feedback (Section 7.9.2);

and (3) remote forcing from the Pacific

ENSO. Chapter S15 then describes modes of

decadal and multidecadal variability in the

Atlantic: (1) North Atlantic Oscillation (NAO),

(2) East Atlantic Pattern (EAP), and (3) Atlantic

Multidecadal Oscillation (AMO). The considerable

variability in ocean properties, with an

emphasis on salinity variations is described.

The section ends with a description of the

imprint of climate change on the Atlantic

Ocean including difficulties with detection

given the large natural variability at all depths

in the northern North Atlantic.

C H A P T E R

10

Pacific Ocean


OVERVIEW

The Pacific Ocean is the largest of the three

major oceans. It has well-developed wind-driven

circulation systems in the subtropics, subpolar

North Pacific, and tropics (Sections 10.1e10.7).

In the south, the Pacific circulation transitions

to the Southern Ocean, which connects it to the

other oceans (Chapter 13). The Pacific is also connected

at low latitudes to the Indian Ocean

through passages in the Indonesian archipelago.

It is connected to the Arctic (Chapter 12) through

the very shallow Bering Strait.

The Pacific is the freshest of the three major

oceans because of small differences in net evaporation/precipitation

between the oceans

(Chapter 5). Compared with the North Atlantic,

this freshness completely inhibits formation of

deep waters and weakens formation of intermediate

water in the northern North Pacific

(Section 10.9). At this global scale, the Pacific is

one of the broad regions of deep upwelling

that returns deep waters formed elsewhere

back to mid-depths or even the surface. Because

of its weak thermohaline circulation, the North

Pacific upper ocean circulation is mostly associated

with wind forcing. Therefore, it can be

useful to study the wind-driven circulation first

in the context of the North Pacific and equatorial

Pacific, followed by study of the other oceans.

The tropical Pacific is the center of action for

the interannual climate mode, El Niño-Southern

Oscillation (ENSO: Section 10.8), which impacts

much of the globe through atmospheric “teleconnections.”

Important natural climate variability

of quasi-decadal timescale is also

observed in the Pacific (Section 10.10; Chapter

S15 located in the supplemental material found

on the textbook Web site http://booksite.


material).

The Pacific Ocean has numerous marginal

seas, particularly along its western side; these

are described briefly in the online supplement

Section S8.10. The complicated passages

through the Indonesian archipelago shunt

water from the tropical Pacific to the tropical

Indian Ocean. The Bering Strait at the northern

end of the Bering Sea allows a small leakage of

North Pacific water into the Arctic and hence

into the Atlantic Ocean. The Okhotsk Sea in

the northwestern Pacific is the site for the

densest water formation in the North Pacific;

this densest Pacific water is only of intermediate

depth and is less dense and much smaller in

impact than dense water formation in the North

Atlantic and Antarctic, which supply the deepest

waters of the global ocean, including the

Pacific.

The Pacific Ocean’s surface circulation

(Figures 10.1, 10.2a and Figure S10.1 in the


303



304

10. PACIFIC OCEAN

120°E 180° 120°W 60°W

80°N 80°N

60°N 60°N

East

Kamchatka C.

40°N 40°N

20°N 20°N


Mindanao

Current

Taiwan

WC

Kuroshio Extension

Subtropical Countercurrent

Alaskan

Stream

0° 0°

Indonesian

Throughflow

EKWC

Kuroshio

Kuroshio

ME

HE

TWC

NGCUC

ESC

Oyashio

Primorye C.

WSAG

SE Countercurrent

Anadyr C.

BSC

Beaufort G.

Alaskan G.

North Pacific Current

Alaska

Current

California Current

System



EUC


Mexican

Coastal C.

CRD

Costa Rica

Coastal C.

EUC

Colombia C.

Peru-Chile

Current System

SEC

N. Queensland

Current


20°S 20°S

East

Australian C.

Leeuwin

Current

Tasman Front

EAUC

40°S 40°S

Flinders C.

Subantarctic Front

60°S 60°S

Polar Front

South Pacific Current

Antarctic Circumpolar

Current

SACCF

Ross Sea

Gyre

80°S 80°S

120°E 180° 120°W 60°W

-5000 -4000 -3000 -2000 -1000 0

FIGURE 10.1 PacificOcean: surface circulation scheme. Major near-surface undercurrents at the equator and along the eastern

boundary are also shown (dashed). The South China Sea circulation represents the winter monsoon. Acronyms: SACCF, Southern

ACC Front; EAUC, East Auckland Current; NGCUC, New Guinea Coastal Undercurrent; EUC, Equatorial Undercurrent; CRD,

Costa Rica Dome; ME, Mindanao Eddy; HE, Halmahera Eddy; TWC, Tsushima Warm Current; EKWC, East Korean Warm

Current; WSAG, Western Subarctic Gyre; ESC, East Sakhalin Current; and BSC, Bering Slope Current.

supplementary Web site) includes subtropical

gyres in both hemispheres, a subpolar gyre in

the North Pacific, and the Antarctic Circumpolar

Current (ACC; Chapter 13) in the far

south. The western boundary currents for the

North and South Pacific’s subtropical gyres are

the Kuroshio and East Australian Current (EAC),

respectively. The eastern boundary currents for

these subtropical gyres are the California Current

and the Peru Current, respectively. The western

boundary current for the North Pacific’s

subpolar gyre is the Oyashio/East Kamchatka

Current (EKC). The strongly zonal (east-west)

circulation in the equatorial Pacific is described

INTRODUCTION AND OVERVIEW 305

separately (Section 10.7) because of its

complexity and dynamics, which differ from

mid-latitude wind-driven circulation processes

(Section 7.8). The tropical circulation also

includes low latitude western boundary

currents: the Mindanao Current and the New

Guinea Coastal Undercurrent (NGCUC). The

Pacific Ocean’s deep circulation (Section 10.6)

consists of inflow from the Southern Ocean in

a Deep Western Boundary Current (DWBC) along

the deep plateaus and island chains from New

Zealand northward. Much of the deep flow funnels

through the Samoan Passage in the South

Pacific and then enters the deep tropical ocean.

Deep flow crosses the equator in the west and

then follows the western boundary’s deep

trenches northward, filling in the deep North

Pacific. The “end” of the deep circulation is

reached in the northeastern Pacific, which has

the oldest deep waters of the world, as supported

by carbon-14 content (Chapter 4,

Figure 4.24b).

The inflowing bottom waters upwell through

the length of the Pacific, although most of the

upward transport occurs in the South Pacific

and tropics. Downward diffusion of heat and

FIGURE 10.2

Reid (1997).

Adjusted geostrophic streamfunction (steric height, 10 m 2 /sec 2 ) at (a) 0 dbar and (b) 500 dbar. Source: From

306

10. PACIFIC OCEAN

FIGURE 10.2

(Continued).

freshwater modify the water density, and the

upwelling deep waters create a relatively

homogenous and volumetrically large water

mass called the Pacific Deep Water (PDW; or

Common Water). This returns back to the

Southern Ocean where it joins the Indian Deep

Water (which is formed similarly) and the North

Atlantic Deep Water (which has an entirely

different formation mechanism). Within the

Pacific, there is also upwelling from the deep

waters to shallower layers, including intermediate

and upper ocean layers, with outflow in

different parts of these layers in all directions:

through the Indonesian passages to the Indian

Ocean, southwestward around Australia, northward

through the Bering Strait, and eastward

through the Drake Passage (see Chapter 14).

10.2. WIND AND BUOYANCY

FORCING

The Pacific’s upper ocean gyres and tropical

circulation are mainly wind-driven. The mean

surface winds are dominated by the westerlies

at latitudes poleward of about 30 (north and

NORTH PACIFIC CIRCULATION 307

south) and the easterly trade winds at low latitudes

(Figure 5.16aec and supplementary

Figure S10.2a). The resulting Ekman transport

convergences and divergences drive Sverdrup

transport (Figure 5.17 and supplementary

Figure S10.2b) and hence the gyres. The anticyclonic

subtropical gyres in Figures 10.1 and

10.2a correspond to the Ekman downwelling

regions and equatorward Sverdrup transport.

The cyclonic gyres in the subpolar North Pacific

and south of the ACC in the Ross Sea correspond

to the Ekman upwelling regions and

poleward Sverdrup transport. A narrow tropical

cyclonic cell, centered at about 5 N,

stretches across the width of the Pacific; it

includes the Mindanao Current and the North

Equatorial Countercurrent (NECC). It is associated

with Ekman upwelling beneath the Intertropical

Convergence Zone (ITCZ).

The alongshore wind stress component at

the eastern boundaries creates Ekman transport

divergence that is not represented in wind

stress curl maps. There is also non-zero wind

stress curl in bands along the boundaries, for instance,

upwelling-favorable along the California-

Oregon coast. Both mechanisms drive the

California and Peru-Chile Current Systems

(PCCS).

The Pacific’s annual mean buoyancy forcing

(Figure 5.15) is dominated by heating/cooling

(Figure 5.12). The tropical Pacific has the largest

mean heating of any region on the globe, over

the upwelling cold tongue in the eastern equatorial

region. Bands of ocean heat gain are found

along the west coasts of North and South America,

in the California Current and Peru-Chile

Current upwelling systems. The Kuroshio

region in the North Pacific is one of the strongest

global airesea heat loss regions (>125 W/m 2 ).

The equivalent region along the Australian

coast, in the EAC, also has significant heat loss

(>100 W/m 2 ).

Net evaporation-precipitation (Figure 5.4a) is

directly related to the Pacific’s surface salinity

pattern. There is net precipitation in the ITCZ

(5e10 N). A broader region of net precipitation

occurs in the western tropical Pacific under the

ascending branch of the Walker circulation. Net

precipitation is also found throughout the higher

latitudes in both the North and South Pacific. Net

evaporation is found in the subtropical gyres

under the descending branches of the Hadley

circulation.

The airesea flux maps in Chapter 5 do not

represent the brine rejection process that creates

dense water when sea ice forms. This process is

active in the North Pacific in the Okhotsk and

Bering Seas, and in the northern Japan (East)

Sea. Okhotsk Sea brine rejection is the densest

source of North Pacific Intermediate Water

(Section 10.9.2)

10.3. NORTH PACIFIC

CIRCULATION

The mid-latitude North Pacific surface circulation

(Figures 10.1 and 10.2a; Table S10.1 in the

online supplement), with its subtropical and

subpolar gyres, is the clearest example seen in

all of the oceans of the two-gyre circulation

driven by the westerly and trade winds. This

is because the North Pacific is almost completely

closed to the north and has only a weak thermohaline

circulation. The gyres have the familiar

east-west asymmetry (with strong western

boundary currents and weak meridional flow

spread over much of the remainder of the

ocean), which is understood in terms of meridional

Sverdrup transport (Figure 5.17 and

Section 10.2). With increasing depth, the North

Pacific gyres weaken and shrink, and the

subtropical gyre centers (highest pressure) shift

westward and poleward (Section 10.6).


10.3.1.1. General Description

The North Pacific’s subtropical gyre, like all

subtropical gyres, is anticyclonic (clockwise in

308

10. PACIFIC OCEAN

the Northern Hemisphere), associated with

Ekman downwelling and equatorward Sverdrup

transport (Figure 5.17). Its strong, narrow,

northward western boundary current is the

Kuroshio. After the Kuroshio separates from

the western boundary and flows eastward into

the North Pacific, the current is referred to as

the Kuroshio Extension. The broad eastward

flow on the northern side of this gyre is called

the North Pacific Current or the “West Wind

Drift.” The North Pacific Current also includes

the eastward flow of the subpolar gyre; it is

also called the Subarctic Current (Sverdrup,

Johnson, & Fleming, 1942). The westward flow

on the south side of the subtropical gyre is the

North Equatorial Current, which also includes

the westward flow in the elongated tropical

cyclonic circulation. The concentrated flow

near the eastern boundary is the California

Current System (CCS), which includes a locally

forced eastern boundary current and a poleward

undercurrent (Davidson Current); both are

forced by coastal upwelling (Section 7.9).

The surface subtropical gyre in the western

North Pacific has an overall “C-shape” (Wyrtki,

1975; Hasunuma & Yoshida, 1978). The “C”

looks like a large-scale overshoot of the Kuroshio

as it becomes the Kuroshio Extension,

with a swing back to the west in the recirculation,

followed by southward flow parallel to the Kuroshio,

a turn to the east in the Subtropical Countercurrent

(STCC) at 20e25 N, and then the

westward flow of the North Equatorial Current

(NEC) south of 20 N. This C-shape is common

to surface flow in all subtropical gyres, but the

STCC portion is very shallow; the circulation

just 250 dbar below the surface is a simpler,

closed anticyclonic gyre.

The broad eastward and westward flows

crossing the Pacific include narrow, nearly

zonal (east-west) fronts or frontal zones that

are narrow (less than 100 km wide). Nomenclature

is confusing and contradictory. We adopt

Roden’s (1975, 1991) terms for the central

North Pacific. The Subarctic Frontal Zone

(SAFZ; or Subarctic Boundary), centered at

about 42 N, is embedded in the North Pacific

Current; it roughly separates the subtropical

and subpolar gyres, being slightly south of

the maximum westerly winds. The Subtropical

Frontal Zone (or convergence), at about 32 N

in the central and eastern Pacific, separates

the eastward North Pacific Current from the

westward NEC.

With increasing depth, the subtropical gyre

shrinks toward the west and toward Japan,

and decreases in strength. It disappears around

1500 m depth except in the Kuroshio region

(Figures 10.10 and 10.14).

10.3.1.2. The Kuroshio and Kuroshio

Extension

The Kuroshio (black stream in Japanese,

where shio means current) arises at the western

boundary where the westward flow of the

NEC splits at about 15 N into northward and

southward boundary currents: the Kuroshio

and Mindanao Current, respectively (Figures

10.1 and 10.3). The Kuroshio continues northward,

turns to follow the south coast of Japan,

then separates and flows out to the midsubtropical

gyre. Maximum surface current

speeds in the Kuroshio range between 75 and

250 cm/sec. The width of the current is 80 to

100 km. It has major variability at timescales of

weeks to decades.

The Kuroshio velocity decreases with depth

(Figure 10.4b). The northward velocity core of

the Kuroshio is sometimes flanked on both sides

by weak countercurrents (flowing in the opposite

direction). Where the Kuroshio begins to

leave the western boundary, it passes eastward

through Tokara Strait (Figure 10.4a,b), tracks

eastward roughly parallel to the south coast of

Japan, then passes through gaps in the Izu-

Ogasawara (Izu) Ridge, and finally enters the

open Pacific at Boso Peninsula (Figure 10.3a).

Between Tokara Strait and the Izu Ridge, the

Kuroshio exists in one of two (or three) semistable

states: flowing either nearly directly


(a)

115˚

120˚

125˚

130˚

135˚

140˚

145˚

150˚

155˚

35˚

30˚

25˚

1

26

East China Sea

Tsushima

St.

2

Tokara St.

Rykukyu Islands

C. Shionomisaki

57

42

15

non LM

onLM

Boso

Pen.

Izu-Ogasawara Ridge

Large

meander

Philippine Basin

160

recirculation

N. Pacific Ocean

35˚

30˚

25˚

20˚

115˚

Taiwan

St.

120˚

22

125˚

Subtropical Countercurrent

130˚ 135˚ 140˚

145˚

150˚

20˚

155˚

-5000 -4000 -3000 -2000 -1000 0

FIGURE 10.3 Kuroshio system in the western North Pacific. (a) Schematic of the large meander (LM), straight (near shore

non-large meander) and offshore non-large meander paths (after Kawabe, 1995), and recirculation gyre schematics, with

transports in Sv (after Hasunuma and Yoshida, 1978; Qiu and Chen, 2005). (b) Index of the Kuroshio meander state: distance

offshore of the 16 C isotherm at 200 m averaged between 132 and 140 E. ÓAmerican Meteorological Society. Reprinted with

permission. Source: From Qiu and Miao (2000).

310

10. PACIFIC OCEAN

FIGURE 10.4 Kuroshio velocity structure. Vertical sections of (b) northward velocity of the Kuroshio where it is

a western boundary current, at 24 N(Source: From Bingham & Talley, 1991), and (d) eastward velocity of the Kuroshio

Extension at 152 30’E [red (blue) indicates eastward (westward) flow]. Source: From Yoshikawa et al. (2004). Section

positions are shown in (a) and (c). Mean temperature at 1000 m is contoured in (c). Figure 10.4d can also be found in the

color insert.


along the coast (straight path), or looping far

to the south in a meander (large meander

path). The Kuroshio remains in one of these

states for several years and then switches to

the other state (index in Figure 10.3b). The

mean eastward-flowing Kuroshio Extension

splits close to Shatsky Rise into a southward

branch that feeds a westward flow that creates

a recirculation gyre (Kuroshio Countercurrent)

and an eastward flow that becomes the North

Pacific Current. The recirculation gyre is often

split in two by the Izu Ridge, with one gyre

west of the ridge and south of Japan, and

the other gyre east of the ridge, downstream

of the Kuroshio separation point (Figure

10.3a).

Once the Kuroshio crosses the Izu Ridge and

enters deep water, its upper ocean structure is

like that of the Gulf Stream, with a strong eastward

velocity core and marked, but weaker,

westward recirculation just to the south. The

Kuroshio Extension extends to the ocean bottom

in the deepest water downstream of the separation

point, with 10 cm/sec velocities even at

the bottom (Figure 10.4d). Westward recirculations

flank the deep Kuroshio Extension to the

bottom.

The volume transport of the Kuroshio

increases downstream (Figure 10.3a) from 20

to 25 Sv, where it is a western boundary current

east of Taiwan (Johns et al., 2001; Bingham &

Talley, 1991), to about 57 Sv east of Tokara Strait

but still prior to separation to a maximum of

140 to 160 Sv at 145 E, just to the east of separation.

Considerable recirculation causes much of

these increases (Imawaki et al., 2001). The

transport decreases east of this point, with

water lost southward to the recirculation gyre

and into the Kuroshio Extension bifurcation

fronts (Yoshikawa, Church, Uchida, & White,

2004).

The Kuroshio Extension is highly unstable.

It meanders and produces rings when the

meanders pinch off. The meanders have somewhat

preferred locations, which differs from

the Gulf Stream. The first northward meander

occurs just downstream of the separation

point. This often creates an anticyclonic

warm-core ring of about 200 km diameter

that moves northward. A second preferred

location for northward meandering is at

150 E. Southward meanders, between the

northward meanders, form cyclonic cold-core

rings south of the Kuroshio Extension. The

envelope of paths is several hundred kilometers

wide from the separation point out to

near 160 E (Shatsky Rise), widening to about

500e600 km with the paths becoming considerably

more random (Mizuno & White, 1983; Qiu

& Chen, 2005).

10.3.1.3. North Pacific Current and

Mid-Latitude Fronts

The North Pacific Current is the broad eastward

flow of the central and eastern subtropical

gyre. The mean speed of the North Pacific

Current is small, less than 10 cm/sec. However,

synoptic meridional crossings of the North

Pacific Current reveal larger geostrophic flows

of 20 to 50 cm/sec that reverse direction on the

order of every 100 km (at the eddy scale) and

are deep-reaching. The difficulty of distinguishing

between eddies and permanent flow

features obscured observation of the deep penetration

of the Kuroshio Extension Front until

recently (Figure 10.3d).

The northern and southern “boundaries” of

the subtropical gyre can be considered to be

the SAFZ (40e44 N) and the Subtropical Front

(25e32 N, depending on longitude). In both

frontal zones d which are synoptically about

100e200 km wide and often contain at least

two sharp fronts d temperature, salinity, and

density change rapidly with latitude (Figure

S10.3 in the supplementary Web site). The

frontal zones are relatively zonal over much of

the North Pacific; they veer southward into the

CCS in the east.

The SAFZ arises in the western North Pacific

from both a branch of the Kuroshio Extension

312

10. PACIFIC OCEAN

50°

130° 120°W

CC

SUMMER

40°

SCE

30°

N

50°

CC

DC

WINTER

40°

30°

N

SCC

50°

40°

DC

EARLY

SPRING

CC

30°

N

CC = California Current

DC = Davidson Current

SCC = So. California

Countercurrent

SCE = So. California Eddy

130° 120°W

FIGURE 10.5 (a) Schematic of the surface currents in the CCS in different seasons. Source: From Hickey (1998). (b) Mean

seasonal cycle of satellite-derived surface temperature (color) and altimetric height, showing the geostrophic surface

circulation. Source: From Strub and James (2000, 2009). This figure can also be found in the color insert.


Front and the STCC. It coincides in the open

Pacific with maximum Ekman convergence in

the center of the subtropical gyre.

The SAFZ might be partly associated with the

separated Oyashio front (Section 10.3.2.2). It

coincides approximately with the maximum

westerly wind, marking the transition from the

Ekman downwelling of the subtropical gyre to

the Ekman upwelling of the subpolar gyre.

The northern front in the SAFZ is the southernmost

limit of the very strong halocline of the

subpolar gyre, and the southernmost limit of

the shallow temperature minimum in the

western subpolar gyre. There is a jump in nutrients

across the frontal zone to higher values in

the subpolar surface waters (Figure S10.3c,d in

the supplementary Web site; surface nitrate

map in Figure 4.23).

10.3.1.4. California Current System

The CCS stretches from the Strait of Juan de

Fuca to the tip of Baja California (Figures 10.1

and 10.5). We describe the CCS in some detail,

because it is the principal example in this text

of an eastern boundary current system. In-depth

overviews of the CCS and its variability can

be found in Wooster and Reid (1963), Huyer

(1983), Lynn and Simpson (1987), Hickey

(1998), and Marchesiello, McWilliams, and

Shchepetkin (2003).

The CCS has two regimes: (1) the southward,

shallow, narrow, meandering California Current

Front, with upwelling zones along the coast,

offshore-advecting jets of upwelled water, and

a northward undercurrent or inshore surface

countercurrent and (2) the broad southward

flow of the subtropical gyre. Dynamically, these

two components have entirely different origins:

(1) southward flow due to locally wind-driven

coastal upwelling with a poleward undercurrent

and (2) southward flow that is part of the

large-scale subtropical circulation resulting

from Ekman downwelling and associated equatorward

Sverdrup transport (Figure 5.17). We

discuss only the upwelling system here.

A simplified approach to the dynamics of

subtropical eastern boundary systems, based on

Ekman transport and upwelling, was provided

in Section 7.9. This framework is useful for initial

broad understanding, but these systems tend to

be far more complex than this, which is evident

as soon as we look at satellite images of seasurface

temperature (SST) and ocean color in the

CCS (Figure 10.6). The CCS upwelling is forced

by the alongshore component of the prevailing

westerly winds, which results from their southward

deflection as they encounter the North

American continent (Figure 5.16). The upwelling

is apparent off the North American coast from

British Columbia to California (50e30 N), as

a patchy band of cool surface water within

a region 80 to 300 km from shore, strongest

from April to August (Figure 10.6a). The upwelled

waters are highly productive, which can

be observed with satellite ocean color sensors

(Figure 10.6b). The upwelled water does not originate

from great depth because the ocean is stratified.

Its source is around 150e200 m depth, but

this is deep enough to access the nutrientenriched

waters below the euphotic zone.

The maximum surface velocity of the mean

southward California Current is 40e80 cm/sec

and its width is 50e100 km. The California

Current is in geostrophic balance with the

cross-shore pressure gradient force. It decays

rapidly with depth and is essentially confined

to the top 300 m (Lynn & Simpson, 1987). Thus

the CC is much shallower and carries much

less transport, on the order of only a few Sverdrups,

than a western boundary current such

as the Kuroshio. The decrease in geostrophic

velocity from the surface to 200 m is evidenced

in the upward tilt of isotherms toward the coast

(Figure 10.7). The upwelling, surface PGF and

upward tilt of the isotherms result from offshore

Ekman transport, which has been observed

directly by Chereskin (1995) (Figure 7.7).

An idealized steady state requires warming of

the upwelled water as it moves offshore. Since

the right amount of warming does not generally

314

(a)

10. PACIFIC OCEAN

(b)

44N

Cloud

126W

Cape Blanco

122W

15 15 June 881

Chl

(mg/m 3 )

25.

10.

Cape Blanco

3.

1.0

Cape Mendocino

40N

Cape Mendecino

Point Arena

Point Reyes

0.3

0.1

0.05

Point Arena

Point Reyes

36N

Monterey Bay

Point

Conception

Monterey

Bay

32N

FIGURE 10.6 (a) Satellite SST (July 16, 1988), with subjectively determined flow vectors based on successive images.

(b) Surface pigment concentration from the CZCS satellite on June 15, 1981. Source: From Strub et al. (1991).

occur at exactly the right time, the actual state is

more complicated. The seasonal offshore Ekman

transport creates an upwelling front that moves

offshore. The California Current’s southward

core is located at the upwelling front, as seen in

Figure 10.7, and moves offshore with the front

as it progresses through the upwelling season.

The mean location of the California Current is

therefore offshore, by about 200e300 km, and

not at the coast. This is also evident in the tighter


FIGURE 10.7 Sections of (top) velocity (m/sec), (middle) salinity, (bottom) potential temperature ( C) across the CCS at

41.5 N (left) and 40.0 N (right) in June, 1987. The coast is to the right. Source: From Kosro et al. (1991).

6

7

8

7

10

316

10. PACIFIC OCEAN

dynamic height contours in Figure 10.5 and the

strong fronts in Figure 10.7.

The mean offshore location of the California

Current is also apparent in enhanced dynamic

height/sea-surface height variability due to

a vigorous eddy field, and in low salinity that

reflects the northern source of the surface water

(Figure 10.8). Underneath and inshore of the

California Current, the mean flow is northward

(poleward), centered at the continental shelf

break. This is the California Undercurrent

(CUC). The CUC is approximately 20 km wide

and its core lies at about 250 m, although it

can extend to more than 1000 m depth. Its

maximum speed is more than 10 cm/sec, and

its water originates in the warm, saline, low

oxygen tropical Pacific. The mean CUC is in

geostrophic balance with the offshore pressure

gradient force at this depth. The reversal of

alongshore geostrophic flow from the southward

CC at the surface to the northward CUC

requires sloping isopycnals between the two

currents. The CUC then weakens below its

core. The CUC is thus recognized by a spreading

of the isotherms and isopycnals, upward above

the undercurrent and downward below it.

During winter, upwelling is weak or inactive.

The California Current is far offshore and relatively

weak and the coastal flow is northward

(the Inshore Countercurrent or Davidson Current).

This poleward flow could be the Sverdrup

transport response to the Ekman suction driven

by positive wind stress curl in the CCS region;

it is overwhelmed by the response to coastal

upwelling during the upwelling season

(Marchesiello et al., 2003). When upwelling starts

up again, an upwelling front appears near the

coast as the offshore edge of the Ekman transport.

A strong southward California Current jet

is associated with the front and moves progressively

offshore with time (Figure 10.5; Strub&

James, 2000).

The strong seasonal cycle of the wind forcing is

quantified with upwelling indices. In Figure 10.9,

one index is based on Ekman transport and the

(b)

46

44

42

40

38

36

34

5

5

3

7

3

32

4

Altimeter SSH Std Deviation [cm]

3

5

4

6

9

5

6

7

8

8

7

5

3

9

4

4

5

6

8

7

6

9

8

7

C. Blanco

5

7

Heceta Bank

Pt. St. George

C. Mendocino

4

Pt. Arena

Pt. Reyes

7

7

5

8

6

6

7

6

Pt. Sur

7

432

30

−130 −125 −120

FIGURE 10.8 (a) Mean salinity at 10 m in July (contoured)

with dynamic height standard deviations greater

than 4 dyn cm in gray. Source: From Lynn and Simpson (1987).

(b) Sea surface height standard deviation (cm) from satellite

altimetry. ÓAmerican Meteorological Society. Reprinted with

permission. Source: From Marchesiello et al. (2003).

6

3

6


FIGURE 10.9 (a) Offshore

Ekman transport based on long-term

mean wind stress. Source: From

Huyer (1983). (b) Upwelling index

based on atmospheric pressure

distribution (from Bakun, 1973),

averaged over 1946e1995. Lower

shaded region (positive values or

blues in the original figure) is

upwelling; upper shaded region

(negative values or reds in the original

figure) is downwelling. Source:

From Schwing, O’Farrell, Steger, and

Baltz (1996).

(b)

AVERAGE MONTHLY UPWELLING INDEX

60N 149W

60N 146W

57N 137W

54N 134W

51N 131W

48N 125W

45N 125W

42N 125W

39N 125W

36N 122W

33N 119W

30N 119W

27N 116W

24N 113W

21N 107W

J F M A M J J A S O N D

350

300

250

200

150

100

50

0

–50

–100

–150

–200

other on the strength of alongshore wind component.

1 Maximum upwelling occurs in late spring

and summer (April through July), as evident

from enhanced surface chlorophyll content in

summer (Figure 10.6b), and is highest near Point

Conception (34 N). North of 40 N, the longshore

winds actually cause downwelling in winter as

the Aleutian Low expands southward; north of

45 N, there is downwelling in the annual mean

(Venegas et al., 2008).

1 Neither index includes the wind stress curl component of the upwelling, although we have already noted that it can be

important (Bakun & Nelson, 1991; Pickett & Paduan, 2003).

318

10. PACIFIC OCEAN

The quasi-continuous, alongshore mean circulation

described in the previous paragraphs is

the simplest view of the CCS. However, as seen

in satellite images (Figure 10.6), the upwelled

water does not move offshore in a “sheet,” but

rather in jets at recurring locations associated

with capes or points in the coastline. The circulation

can be either “squirt-like,” in which the jet

goes out to sea and dies, or meandering, in which

the jet goes out and returns. The high mesoscale

eddy activity in the CCS (Figure 10.8) maybe

created by baroclinic instability of the coastal

upwelling current. The eddies spawned by this

instability move the upwelled cold water

offshore and thus maintain the mean balance,

which includes Ekman upwelling (Marchesiello

et al., 2003). Recent studies of the California

Current are beginning to focus on even smaller

spatial scales, called the submesoscale (order of

1 to 10 km). These are associated with the actual

fronts and their instabilities within the mesoscale

eddy field (Capet, McWilliams, Molemaker, &

Shchepetkin, 2008).

10.3.1.5. North Equatorial Current

The NEC is the broad westward flow on the

southern side of the subtropical gyre. It is

between about 8 and 20 N depending on longitude.

The NEC forms gradually in the eastern

Pacific from southward flow of the subtropical

gyre, including the CCS. At the eastern

boundary, it has input from the tropical current

system (Costa Rica Dome and NECC).

As the NEC flows westward, some of it moves

southward and joins the strong eastward flow of

the NECC. When the NEC reaches the western

boundary, it bifurcates at about 14 N into a northward

portion that becomes the Kuroshio and

a southward portion that becomes the Mindanao

Current (Section 10.7.4). In the western Pacific,

the NEC includes a strong zonal surface salinity

front that separates saline water that originates in

the subtropical gyre from fresher NECC surface

water. The location of this front is similar to the

latitude of the NEC bifurcation, and it is also an

ecological front that is important for fisheries

(Kimura & Tsukamato, 2006). These suggest

that the front is a boundary between Ekman

upwelling in the tropical NEC/NECC cyclonic

gyre and downwelling in the anticyclonic

subtropical gyre.

Volume transport of the NEC in the western

Pacific is up to 50 Sv in the top 500 m and 80 Sv

top to bottom (Kaneko, Takatsuki, Kamiya, &

Kawae, 1998; Toole, Millard, Wang, & Pu, 1990).

10.3.1.6. Depth Dependence of the

Subtropical Circulation

The subtropical gyre shrinks spatially with

depth. Like all subtropical gyres, it shrinks

toward the most energetic part of its surface

flow: westward toward the western boundary,

and northward toward the Kuroshio Extension.

The Kuroshio Extension extends to the ocean

bottom as previously noted.

The gyre shrinkage from the sea surface to

about 200 m depth is dramatic (Reid, 1997; represented

by Figure 10.2). The boundary between

eastward and westward flows shifts from south

of 20 N at the sea surface to 25e30 Nat200m.

The C-shape of the western gyre, which includes

the STCC, disappears by 200 m. On the other

hand, the Kuroshio and Kuroshio Extension do

not shift (Figure 10.3d). At 1000e1500 m depth,

the anticyclonic subtropical gyre is found

entirely in the western North Pacific near the

Kuroshio and Kuroshio Extension (Figure 10.10).

Flow in the subtropical regions vacated by

the subtropical gyre is very weak. Steric height

differences over 1000 km distances are on the

order of 1 cm rather than the 10 cm differences

within the gyre proper. Dynamically, on isopycnal

surfaces that are still within the gyre in the

western region, the vacated region is called the

shadow zone (Section 7.8.5). Within these

regions where there is little direct ventilation

from the sea surface, on the eastern and

southern flanks of the subtropical gyres, oxygen

is depleted to the point where denitrification

sets in (Section 10.9.1).


FIGURE 10.10 Steric height (10 m 2 /sec 2 ) at 1000 dbar based on hydrographic data and reference geostrophic velocities

adjusted to provide absolute circulation at all depths. Source: From Reid (1997).

10.3.2. Subpolar Circulation

10.3.2.1. General Description

The cyclonic (counterclockwise) subpolar

gyre in the North Pacific stretches across the

width of the basin and is compressed in the

north-south direction between about 42 N

(Subarctic Front) and the Aleutian Islands/

Alaskan coast (Figure 10.1). It has a southward

western boundary current, the Oyashio/EKC.

A geographic constriction at the southernmost

location of the Aleutian Islands (near the

date line) separates the subpolar gyre into two

portions. The Western Subarctic Gyre is centered

east of the Kuril Islands, and the Alaskan Gyre

is centered in the Gulf of Alaska. They are

320

10. PACIFIC OCEAN

connected through eastward flow along the

southern side of the gyre (Subarctic Current,

which is part of the North Pacific Current, Section

10.3.1.3) and westward flow along the Aleutian

Islands (Alaskan Stream). Completing the nomenclature

for the cyclonic gyre, the Alaska Current is

the northward eastern boundary current along

the coast of Canada and Alaska. An older but exhaustive

treatment of this circulation is found

in Favorite, Dodimead, and Nasu (1976; Figure

S10.4 in the supplementary Web site).

Parts of the subpolar gyre circulation loop

through the Bering and Okhotsk Seas (S8.10

located in the Web site supplementary text).

Transport of 0.8 Sv from the North Pacific to

the Arctic and onward to the Atlantic occurs

through the Bering Strait at the northern end

of the Bering Sea. Both the Bering and Okhotsk

Seas have significant ice cover in winter. As

a result, important water mass transformation

and modification occur in both seas. The

Okhotsk Sea produces the densest water in the

subpolar North Pacific, mainly through sea ice

processes (Section 10.9.2.1).

The subpolar gyre circulation is forced by

Ekman upwelling (suction; Figure 5.16d). The

winds throughout the region are westerlies,

producing southward Ekman transport. The

strongest westerly winds are at about 40 N.

Southward Ekman transport is largest there

and decreases with higher latitude to smaller

southward transports. This requires upwelling

into the Ekman layer, which creates northward

mean Sverdrup transport and the cyclonic

gyre (Figure 5.17).

The upwelled water in the subpolar gyre

comes from just below the Ekman layer. (It

cannot come from greater depth because of the

strong pycnocline, mainly due to the low

salinity surface layer, hence halocline.) The

heightened surface nitrate in Figures 4.22 and

4.23 is a result of this upwelling, which greatly

enhances biological productivity. Major fisheries

including salmon, halibut, saury, and walleyed

pollock are found in the subpolar gyre.

Clearly, the Subarctic Front, marking the

southern boundary of the subpolar gyre’s

upwelling, is an important ecosystem boundary.

With increasing depth, the North Pacific’s

subpolar circulation does not shift location unlike

that of the subtropical gyre. It weakens, but its

boundary currents reach far down into the water

column, even to the bottom. The subpolar gyre is

therefore “quasi-barotropic”: its surface currents

extend to the bottom (barotropic) but weaken

(quasi). Near the bottom there are also additional

currents associated with the topography and

global thermohaline forcing (weak upwelling).

The barotropic nature of the gyre is possibly due

to geographic restriction, with the Alaskan coast

cutting through the region that would be spanned

by the gyre if there were no land. On the other

hand, similar structure is found in other high latitude

cyclonic circulations (North Atlantic

subpolar gyre and the Weddell and Ross Sea

gyres), suggesting a more general dynamical

underpinning.

10.3.2.2. Subpolar Western Boundary

Currents

The southward flow in the subpolar western

boundary current system includes: (1) the EKC

along the Kamchatka peninsula and the northern

Kuril Islands and (2) the Oyashio along the

southern Kuril Islands and Hokkaido. The division

between the two is at Bussol’ Strait, which

is the deepest strait in the Kuril Island chain.

The distinction is drawn because about half of

the EKC loops through the Okhotsk Sea, where

water properties are greatly modified. This

creates a discontinuity in water properties at

Bussol’ Strait where the Okhotsk Sea waters

exit and join the Oyashio.

About 200 km offshore of the Oyashio,

there is a northeastward flow called the Subarctic

Current (see also Figure S10.4 in the

textbook Web site). The Oyashio-Subarctic

Current region is very dynamic and includes

large (100e200 km diameter), deep-reaching,

long-lived anticyclonic eddies with cold, fresh


cores (<3 C, <33.5 psu) that are usually found

between Hokkaido and Bussol’ Strait (online

Figure S10.5). The eddies have two different

origins: either locally at Bussol’ Strait, from

water exiting the Okhotsk Sea, or as warm water

from the Oyashio intrusions (see next paragraph)

that then propagates northeastward

between the Oyashio and the Subarctic Current

and is modified by the local cold, fresh subpolar

water (Yasuda et al., 2001).

The Oyashio separates from the western

boundary at the southernmost cape of Hokkaido.

After separation, it usually makes two large meanders

called the first (coastal) and second (offshore)

Oyashio intrusions (Figure S10.6 from the online

supplemental material). These are unrelated to

the Kuroshio Extension meanders, which are

farther south. Water from the coastal Oyashio

intrusion can penetrate southward along the

Honshu coast, sometimes as far south as the

Kuroshio separation point at around 36 N;

this cold coastal water is visible in the SST

image of Figure 10.11. The location of southernmost

penetration is of great interest to Japanese

fisheries since the nutrient-rich Oyashio

waters support a more biologically productive

ecosystem than the nutrient-depleted Kuroshio

waters. Therefore the Oyashio penetration latitude

is used as a regional climate index.

The Oyashio/EKC is a relatively weak

western boundary current. Maximum surface

velocities are 20e50 cm/sec. Total Oyashio

transport, based on combined direct current

observations and hydrographic data east of

FIGURE 10.11 Oyashio, Kuroshio, and Mixed Water Region east of Japan. Sea surface temperature (NOAA AVHRR

satellite infrared image) with temperature scale from 0 to 25 C; E1, E2, and E3 denote anticyclonic eddies. Source: From

Yasuda et al. (2001).

322

10. PACIFIC OCEAN

Hokkaido, ranges from 5 to 20 Sv, with large

variability (Kono & Kawasaki, 1997; Yasuda

et al., 2001). The EKC transports range from 10

to 25 Sv, relative to various levels of no motion

(Talley & Nagata, 1995). 2

The separated Kuroshio and Oyashio are

about 5 degrees of latitude apart (Figure 10.11).

The region between them is referred to as the

“Transition Region,” “Mixed Water Region,” or,

in older literature, the “Perturbed Area.” Water

properties in this region are transitional between

the Oyashio and Kuroshio properties. Both

currents spawn major mesoscale eddy variability,

some in the form of “rings,” which participate

in water mass modification. Sometimes the

eddies re-merge with their parent currents,

bringing the modified waters back with them.

10.3.2.3. Circulation in the Gulf of Alaska

The North Pacific Current splits as it

approaches the North American continent and

part turns south into the CCS. The remainder

turns north into the Alaska Current, forming

the eastern and northern side of the cyclonic

Alaskan Gyre in the Gulf of Alaska. Where the

coast of Alaska swings southward, at about

143 W, it forms a slanted western boundary

along which the swift southwestward Alaskan

Stream forms as a western boundary current.

The wind field that drives the cyclonic circulation

includes intensified Ekman upwelling in

the Gulf of Alaska.

Details of the North Pacific Current bifurcation

depend on the large-scale wind forcing,

which has seasonal variability, and also interannual

and decadal variability associated mainly

with ENSO and the Pacific Decadal Oscillation

(PDO; Sections 10.8 and 10.10; Chapter S15

from the online supplemental material). The

position of the North Pacific Current bifurcation

is at about 45 N in winter and 50 N in summer

(Figure 10.1). The subpolar gyre, including the

Alaskan Gyre, intensifies during periods when

the atmosphere’s Aleutian Low is especially

strong such as El Niño years and years of low

PDO. When the Aleutian Low and the subpolar

gyre are weak, more subpolar water enters the

CCS (Van Scoy & Druffel, 1993).

The Alaska Current contains dramatic,

large anticyclonic eddies that are permanent,

time-dependent components of the circulation.

“Sitka Eddies” form west of Sitka, Alaska, at

about 57 N and have a diameter of 150e300 km

and surface amplitude of 10e20 cm (Tabata,

1982). “Haida Eddies” or “Queen Charlotte

Eddies” form west of the Queen Charlotte Islands

(Figure S10.7 located in the online supplementary

material). The formation sites are related to

bottom topography. After formation, these eddies

propagate mainly westward into the Gulf of

Alaska and are an important means of transporting

coastal properties into the interior. Large

eddies also populate the Alaskan Stream on the

northwest side of the Gulf of Alaska (Crawford,

Cherniawsky, & Foreman, 2000).

10.4. SOUTH PACIFIC

CIRCULATION


The South Pacific is dominated by its anticyclonic

subtropical gyre, extending from the

ACC at about 50 Stotheequator(Figures 10.1

and 10.2a; Table S10.2 in the online supplementary

material). The gyre is well defined, but its

western boundary current is complicated

because the western boundary is composed of

islands. (Oceanographically, Australia is a large

island since it sits entirely within the subtropical

gyre latitudes.) Connections with the other

Southern Hemisphere oceans occur through the

2 These transport estimates could be low because (1) velocities are often underestimated due to the use of inappropriately

shallow levels of no motion and (2) large anticyclonic eddies can pull much of the Oyashio transport offshore, resulting in

a weak coastal Oyashio and a stronger offshore component.

SOUTH PACIFIC CIRCULATION 323

complex passages of the Indonesian archipelago

and through the Southern Ocean south of

Australia and South America.

The main western boundary current is the

EAC, which flows southward along the coast of

Australia until reaching the northernmost latitude

of New Zealand. The EAC then separates

and flows eastward to New Zealand, where it

re-attaches to the east coast (as a western

boundary current called the East Auckland

Current) and continues a little farther southward.

The EAC is very time-dependent and dominated

by a series of cyclonic and anticyclonic eddies.

The broad eastward flow on the south side

of the subtropical gyre can be called the South

Pacific Current (SPC), following Stramma, Peterson,

and Tomczak (1995), and consistent with

usage of “North Pacific Current” and “North

Atlantic Current” for the West Wind Drifts in

the Northern Hemisphere. The circulation is

bounded to the south by the Subantarctic Front,

which is the northernmost front of the ACC

(Chapter 13).

The northward flow along the coast of South

America is the Peru-Chile Current. Like the

California Current, the Peru-Chile Current is

both the northward flow of the subtropical

gyre and a full coastal upwelling system

(PCCS) with separate eastern boundary current

dynamics driven by alongshore winds. The

westward flow of the subtropical gyre is the

South Equatorial Current (SEC). At the sea

surface, the SEC is located from about 20 S

northward all the way to and across the equator;

its structure at low latitudes is described with

the tropical circulation in Section 10.7.3.

10.4.1.1. East Australian Current

The EAC is the southward western boundary

current along the coast of Australia (Figure 10.12).

A thorough description is found in Ridgway and

Dunn (2003). The EAC forms from the westward

flow of the SEC as it crosses the Coral Sea and

reaches the Australian coast. At the sea surface,

the SEC bifurcates at about 15 S into the

(b)

5

Papua New Guinea

1850

10

1800

15

1750

20

1700

25

Australia

1650

30

1600

35

Tasman Sea

1550

40

Tasmania New Zealand

1500

45

145 150 155 160 165 170 175 180 175 170

Coral Sea

FIGURE 10.12 (a) Schematic of circulation in the

western South Pacific (SEC: South Equatorial Current; EAC:

East Australian Current; TF: Tasman Front). Eddy shedding

from the EAC is depicted in light gray. Source: From Mata

et al. (2006). (b) Mass transport streamfunction relative to

2000 dbar; contour interval is 25 m 2 . Source: From Ridgway

and Dunn (2003).

324

10. PACIFIC OCEAN

southward EAC and northward flow along

Queensland. This bifurcation point moves

toward the south with increasing depth, reaching

500 m at about 22 S (Figure 10.12b and

Figure S10.8 in the supplementary Web site).

The EAC transport intensifies as it flows along

the Australian coast, reaching a maximum

velocity of around 90 cm/sec at 30 S. It begins

to separate from the coast around 31 to 32 S. It

reaches its maximum transport of about 35 Sv

shortly after separation, at 33 S, where it

undergoes a southward meander and retroflection

with part of the transport returning northward

in a tight recirculation. A mean

northward recirculation exists offshore of the

EAC between latitudes 33 S and about 24 S,

and likely has two separate lobes (Figure 10.12).

Most of the EAC flow that does not recirculate

turns eastward into the zonal Tasman Front

and crosses the Tasman Sea to the northern cape

of New Zealand. Transport in the Tasman Front

is estimated at 13 Sv. The EAC flow in the Tasman

Front re-attaches to the coastline at New

Zealand and forms the East Auckland Current

(Roemmich & Sutton, 1998). The East Auckland

Current continues southward and finally separates

from New Zealand at about 43 S

(Figure 10.12), where it meets a northward

loop of the Subantarctic Front (ACC).

The remainder of the EAC reaches southward

through the Tasman Sea to Tasmania. The location

of the southernmost penetration of EAC

waters along Tasmania is used as a regional

climate index, much like the southward penetration

latitude of Oyashio waters along Japan

(Section 10.3.2.2). A small portion continues

southward past Tasmania and turns westward

into the Indian Ocean, connecting the westward

flow of the South Pacific and Indian subtropical

gyres (Speich et al., 2002; Ridgway & Dunn,

2007).

The EAC separates from the coast at about

32 S and meanders strongly southward and

then northward. The meander regularly pinches

off into a ring. The EAC undergoes major

retraction and deformation after such eddy

shedding, which occurs about every 100 days

(Mata, Wijffels, Church, & Tomczak, 2006).

The EAC has long been understood to

be particularly rich in eddies (Hamon, 1965;

Godfrey et al., 1980). EAC eddies sometimes

appear to dominate the mean circulation. Eddy

diameters are 200e300 km, and surface speeds

are up to 180e200 cm/sec, with lifetimes of up

to a year (Boland & Church, 1981). The eddy

centers are well mixed to as much as 300 m depth

(Nilsson & Cresswell, 1981). In austral winter, the

surface water in an eddy may be as much as 2 C

warmer than the surrounding water.

Eddy formation sites in the EAC tend to be

recurrent, so the eddies appear in the mean

dynamic topographies and altimetric height

maps (Figure 10.12 and Figure S10.9 from the

online supplementary material). Two are found

within the recirculation of the EAC along

the Australian coast, and three within the

Tasman Front and East Auckland Current. The

permanence of these eddy sites suggests topographic

control (Ridgway & Dunn, 2003).

10.4.1.2. South Pacific Current and

Subtropical Front

The eastward flow of the South Pacific

subtropical gyre is the SPC (Stramma et al.,

1995; Wijffels, Toole, & Davis, 2001). The broad,

weak eastward flow of the SPC was long identified

with the ACC, but the SPC is dynamically

distinct from the ACC. As an analog of the North

Pacific Current, we consider the SPC to be all of

the eastward flow of the South Pacific’s subtropical

gyre north of the Subantarctic Front. The

SPC flows into the broad, open-ocean part of the

northward Peru-Chile Current, and from there

to the westward SEC. These three currents constitute

the open ocean part of the South Pacific’s

subtropical gyre. Maximum Sverdrup transport

for the subtropical gyre occurs around 30 Sand

is about 35 Sv (Figure 5.17, Figure S10.2b in the

online supplementary materials, and Wijffels

et al., 2001).

SOUTH PACIFIC CIRCULATION 325

The SPC forms as eastward outflow from the

East Australian and East Auckland Currents. In

mid-ocean, it has a somewhat bowed structure,

with a slight northward excursion from offshore

of the EAC to mid-gyre, around 170 W, then

southward to about 140 W and finally, northward

in the main Peru-Chile gyre flow. This

structure appears to be permanent.

The eastward flow of the SPC bifurcates at

the eastern boundary between 40 S and 45 S.

The northward flow joins the Peru-Chile

Current and the southward flow joins the ACC

through Drake Passage.

Within the SPC there is a marked, nearly

zonal Subtropical Front, called the Subtropical

Convergence in earlier works, including earlier

editions of this text. The Subtropical Front is

identified by large meridional gradients in

temperature and salinity in the upper ocean,

with a northward increase of 4 C and 0.5 psu,

sometimes over just a few kilometers (Deacon,

1982; Orsi, Whitworth, & Nowlin, 1995). North

of the Subtropical Front lies the saline, warm

water of the central subtropical gyre; salinities

are greater than 34.9 psu just north of the front.

South of the Subtropical Front is the fresher,

cooler water of the poleward part of the gyre.

Transport of the SPC has not been estimated

as such. An estimate for the Subtropical Front

alone is less than 5 Sv (Stramma et al., 1995).

Otherwise the transport of the broad subtropical

gyre has been mainly estimated from the meridional

(north-south) component through eastwest

sections across the gyre, which are

described in the next subsection.

10.4.1.3. Northward Flow of the

Subtropical Gyre and the Peru-Chile

Current System

Northward flow in the subtropical South

Pacific consists of the broad subtropical gyre

and the swifter, narrow eastern boundary current

system along the coast of South America,

referred to as the PCCS (Figures 10.1 and

10.13). The northward transport is estimated to

be 15 Sv between 180 and the eastern boundary

(Wijffels et al., 2001). Within the broad gyre,

denser surface waters from the south subduct

northward under lighter low latitude waters.

This creates the stratified structure of the central

South Pacific pycnocline (Section 10.9.1), and

the salinity/oxygen layering in the vertical

that facilitates identification of various water

masses.

At the eastern boundary, the PCCS

(Figure 10.13) is a typical eastern boundary

current upwelling system (Sections 7.9 and

10.3.1.4), forced by the alongshore component

of the large-scale winds and an offshore band

of positive wind stress curl. It includes the

northward Peru-Chile Current (also called the

Peru Current and formerly called the Humboldt

Current). The Poleward Undercurrent (also

called the Gunther Current) is found along the

coast beneath the surface layer, as expected for

a typical eastern boundary current system. The

PCCS also contains other currents: a poleward

Peru-Chile Countercurrent 100e300 km offshore,

and an equatorward Peru Coastal Current on

the inshore side. The Peru-Chile Current and

Peru Coastal Current connect to the equatorial

SEC and the cold tongue in the eastern equatorial

Pacific (Figure 10.13). The Equatorial

Undercurrent (EUC) feeds into the Poleward

Undercurrent and Peru-Chile Countercurrent

(Strub et al., 1998).

Maximum upwelling, extending southward

along the Chilean coast to 45 S, occurs in

austral summer. The PCCS upwelling is well

known because of the rich fisheries there. Satellite

ocean color images (Figure S10.10 in the

online supplemental text) vividly show the

effects of coastal upwelling, which lifts nutrients

to the euphotic zone, resulting in high biological

productivity. The permanent upwelling

region extends from about 32 S northward to

the equator; seasonal upwelling occurs south

of this to about 40 S.

Vertical sections across the PCCS at 33 S

(Figure 10.13) show the isotherm structure

326

10. PACIFIC OCEAN

FIGURE 10.13 Peru-Chile Current System. (a) Maps in austral winter and summer. Acronyms: WWD, West Wind Drift;

PC, Peru Current; PCCC, Peru-Chile Countercurrent; PUC, Poleward Undercurrent; PCC, Peru Coastal Current; CCC, Chile

Coastal Current; and CHC, Cape Horn Current. Also, near the equator: CC, Colombia Current; AENC, Annual El Niño

Current; NECC, North Equatorial Countercurrent; SEC, South Equatorial Current; EUC, Equatorial Undercurrent. Source:

From Strub et al. (1998). (b, c) Eastern South Pacific zonal vertical sections at 33 S: temperature ( C) with meridional current

directions and dissolved oxygen (ml/L); companion salinity and phosphate sections appear in Figure S10.11 on the textbook

Web site.

typical of a geostrophic eastern boundary current

system, including the equatorward Peru-Chile

Current above about 500 m and the poleward

subsurface Peru-Chile Undercurrent (PCUC)

near the coast. The undercurrent is characterized

by low oxygen which comes from the tropics and

from local high productivity that traps high

nutrients and low oxygen just beneath the

surface layer (Montecino et al., 2006). High

nutrient content, associated with the low oxygen,

helps to create the characteristic high biological

productivity of this eastern boundary region.

The PCCS is strongly affected by ENSO

(Section 10.8). Collapses of the PCCS fisheries

resulting from changing upwelling conditions

were among the earliest dramatic evidences

for ENSO, which is now known to encompass

the entire equatorial Pacific. During normal

conditions, the Peru-Chile Current extends to

a few degrees south of the equator before

PACIFIC OCEAN MESOSCALE EDDY VARIABILITY 327

turning west into the SEC. The low temperature

of the Peru-Chile Current surface waters

contrasts with higher equatorial temperatures

to the north. During an El Niño (warm phase),

the high temperatures extend 5 to 10 degrees

farther south than usual and the thermocline

deepens by 100 m or so. Upwelling either

weakens or simply draws on warmer water

from this thicker warm layer, thus causing the

surface temperatures to increase. The increase

in temperature was thought to kill fish, but

recent studies have shown that the fish merely

descend below the abnormally warm surface

layer. In every austral summer there is a slight

warming of the sea surface along with an

increase in precipitation. During El Niño years,

however, the warming and the rainfall far

exceed the norm.

10.4.1.4. South Equatorial Current

The SEC is the broad westward geostrophic

flow in the northern limb of the South Pacific’s

subtropical gyre (Figures 10.1 and 10.12). The

SEC forms in the eastern Pacific as the northward

flow of the subtropical gyre turns westward.

The narrow eastern boundary current

(Peru-Chile Current) also feeds into the SEC

close to the equator.

As it reaches the western South Pacific, the

SEC carries water into the Coral Sea off northeastern

Australia. The many islands in the

region complicate the SEC, including intense

zonal jets with large east-west extent (Webb,

2000; Qu & Lindstrom, 2002; Ganachaud, Gourdeau,

& Kessler, 2008). When the SEC reaches

the Australian coast, it bifurcates into the

southward EAC and the northward North

Queensland Current. The latter feeds the

NGCUC, bringing South Pacific water to

the western equatorial Pacific and feeding the

EUC (Section 10.7.4).

The SEC also includes the frictional equatorial

surface flow (Section 10.7), which is bounded to

the north by the powerful eastward NECC.

Because the SEC extends across the equator,

whereas the NEC is separated from the equator

by the NECC, the South Pacific subtropical gyre

is much more directly connected to the equator

than is the North Pacific gyre. Subtropical anomalies

in heat or salinity can more easily reach the

equator from the South Pacific than from the

North Pacific because of this direct SEC connection

(Johnson & McPhaden, 1999).

10.5. PACIFIC OCEAN MESOSCALE

EDDY VARIABILITY

The ocean circulation focused on in this

text is the mean of a highly time-dependent,

turbulent flow. Mesoscale eddy variability at

timescales of weeks to months is easily detected

with instruments such as satellite altimeters,

which measure the surface height variability.

At depth, eddy variability is measured with

moored observations at point locations and

using Lagrangian floats that are usually

deployed at a single depth.

Surface EKE and horizontal eddy diffusivity

in the Pacific are shown in Figures 14.16 and

14.17 and also in Figure S10.12 on the textbook

Web site. High EKE is mostly associated with

strong mean flows: the Kuroshio Extension

(30e40 N), the EAC (25e40 S), the ACC (south

of 50 S), and the NECC (5e10 N). Two zonally

elongated regions of high eddy energy, at 20 N

and 25 S, are associated instead with weak eastward

surface flows. These are the STCCs in both

hemispheres; the flow just below the surface,

even at 200 dbar, is westward (Figure 10.2).

The energy in these unstable mean flows is

mainly released through baroclinic instability,

creating the high EKE (Stammer, 1998; Qiu,

Scott, & Chen, 2008; Section 7.7.5).

High eddy variability in the Pacific in Figures

14.16 and S10.14 also occurs at the locations of

recurrent rings, including the Tehuantepec

eddies in the eastern tropical Pacific (Section

10.7.6), the Kuroshio rings, the EAC rings, and

the rings along the boundaries of the subpolar

328

10. PACIFIC OCEAN

gyre (Haida & Sitka eddies; eddies in the

Oyashio).

10.6. DEPTH DEPENDENCE OF THE

PACIFIC OCEAN CIRCULATION

AND MERIDIONAL OVERTURN

Below the wind-driven subtropical gyres,

and coexisting with the deep-reaching North

Pacific subpolar gyre, the Pacific circulation is

weak, mostly less than several centimeters per

second except in the tropics. Faster currents

(>10 cm/sec) occur in the deeper parts of the

upper ocean western boundary currents and in

the DWBCs, but transports are nevertheless

relatively small, of the order of 10 Sv or less.

As we leave the sea surface, the subtropical

gyres shrink away from the equator, away from

the eastern boundary, and toward the energetic

western boundary currents. The Kuroshio gyre

shrinkage was described in Section 10.3.1.6. In

FIGURE 10.14

From Reid (1997).

Adjusted geostrophic streamfunction (steric height, 10 m 2 /sec 2 ) at (a) 2000 dbar, (b) 4000 dbar. Source:

DEPTH DEPENDENCE OF THE PACIFIC OCEAN CIRCULATION AND MERIDIONAL OVERTURN 329

FIGURE 10.14

(Continued).

the South Pacific, the subtropical gyre shrinks

into the Southwest Pacific Basin, east of New

Zealand and the Tonga-Kermadec Ridge.

On the tropical side vacated by these shrinking

gyres, the flows are nearly zonal except close to

the western and eastern boundaries (Figures

10.2b, 10.10, 10.14 and Figure S10.13 on the textbook

Web site). This zonal flow pattern persists

down to the tops of the major mid-ocean ridges,

roughly between latitudes 20 N and 20 S.

Outside the tropics, the deep flow patterns are

influenced by the overlying gyres, the underlying

topography, and the DWBCs (Figure 10.14). In the

southwest Pacific below 2000 dbar, the circulation

is a combination of a northward DWBC and an

anticyclonic flow that fills the rest of the basin to

the east and north. In the southeast Pacific, in

the Bellingshausen Basin, the flow is weak and

cyclonic from about 800 dbar to the ocean bottom,

with a southward eastern boundary current that

carries the thick, low oxygen layer of PDW southward

to the Southern Ocean (Shaffer et al., 1995;

120˚

150˚

180˚

210˚

240˚

120˚ 150˚ 180˚ 210˚ 240˚ 270˚ 300˚

270˚

300˚

330

10. PACIFIC OCEAN

(a)

0

500

15

10

1000

5

4.0

1500

3.0

2000

2.0

2500

Depth (m)

3000

3500

4000

4500

5000

5500

6000

Tonga-Kermadec Ridge

0.6

1.0

East Pacific Rise

6500

0 1000 2000 3000 4000

PCM9

5000 6000 7000 8000 9000 10000 11000 12000

Distance (km)

(b)

0

500

210

240

1000

1500

165

210

180

180

110

2000

150

150

2500

Depth (m)

3000

3500

180

165

165

4000

4500

205

120˚

150˚

180˚

210˚

240˚

270˚ 300˚

5000

5500

6000

6500

0 1000 2000 3000 4000

120˚150˚180˚210˚240˚270˚

300˚

5000 6000 7000 8000 9000 10000 11000 12000

Distance (km)

FIGURE 10.15 South Pacific sections at 32 S and DWBC. (a) Potential temperature and (b) oxygen (mmol/kg). Neutral

densities 28.00 and 28.10 kg/m 3 are superimposed in (a). Source: From the WOCE Pacific Ocean Atlas, Talley (2007). (c) Mean

northward velocities (cm/sec) from current meters at 32 30’S northeast of New Zealand in 1991e1992. The array location is

within the white box in (a). Source: From Whitworth et al. (1999).

DEPTH DEPENDENCE OF THE PACIFIC OCEAN CIRCULATION AND MERIDIONAL OVERTURN 331

FIGURE 10.15

(Continued).

Figure 10.15b; Section 10.9.3). In the deep North

Pacific north of about 10 N, the abyssal flow

consists of two anticyclonic circulations, one

centered south of the Hawaiian Islands, and the

other centered at about 45 N (Figure 10.14b).

These two gyres are also evident in silica distributions

on deep isopycnals (Talley & Joyce, 1992).

The deep flows include well-delineated

DWBCs (Section 7.10.3). In the southwest Pacific,

the DWBC carries deep and bottom waters

from the Southern Ocean northward into

the Pacific, as seen in observations at 32 S

(Whitworth et al., 1999). Large upward slopes

in isotherms within several stations just east

of the Tonga-Kermadec Ridge indicate the

narrow DWBC, from the bottom up to 1.8 C

(~2500 m; Figure 10.15). Northward transport

of 16 Sv was measured in a narrow, banked

band at the ocean bottom, mostly colder than

1 C. This DWBC continues northward to the

tropics. Its most constricted location is at the

Samoan Passage at 10 S, 169 W(Figure 10.16).

Observed transport of all waters colder than

1.1 C, including those within the passage and

banked against the Manahiki Plateau, was 11.7

Sv (Roemmich, Hautala, & Rudnick, 1996). The

mean northward transport below 4000 m, within

the Samoan Passage, was 6.0 Sv; velocities were

shown in Figure 6.7 (Rudnick, 1997).

The DWBC proceeds northward from the

Samoan Passage region and crosses the equator

at the deep western boundary (Figure 10.17 and

Figure S10.14 on the textbook Web site). Here it

splits into two branches, one following the

western boundary and the other heading

toward the Wake Island Passage (168 30’E,

18 20’N). The western boundary branch is

observed to carry both Lower Circumpolar

Deep Water (LCDW; 1 Sv) and Upper Circumpolar

Deep Water (UCDW; 11 Sv). The flow in

the Wake Island Passage is up to 10 cm/sec

within several hundred meters of the bottom,

with a transport of 4 Sv of LCDW (Kawabe,

Yanagimoto, Kitagawa, & Kuroda, 2005;

Kawabe, Yanagimoto, & Kitagawa, 2006).

North of the Wake Island Passage, the deep

flow moves westward to the boundary and then

northward to an encounter with the Kuroshio

Extension. Further north along the subpolar

boundary, abyssal circulation theory indicates

that the DWBC should flow southward (even

though there is no local source of deep water;

Figure 7.16). The western and northern boundaries

are complicated by a very deep trench, in

332

10. PACIFIC OCEAN

Southern Ocean in the abyssal layers and southward

outflow in the deep to intermediate layers

(e.g., Figures 10.18, 14.6, and Figure S10.15 on

the textbook Web site). Estimates of the northward

transport of the deepest water into the

South Pacific (LCDW or Antarctic Bottom Water;

AABW) range from 7 to 20 Sv, but the large range

might simply be due to layer choices. Most of this

water upwells into the PDW and returns southward.

Most of the upwelling occurs in the South

Pacific and tropics; at 24 N in the North Pacific

the bottom upwelling cell is much weaker and

much more confined to the bottom layers.

10.7. TROPICAL PACIFIC

CIRCULATION AND WATER

PROPERTIES

FIGURE 10.16 DWBC in the Samoan Passage. (a)

Potential temperature ( C) on WOCE P31 across the

passages. (b) Mean northward velocity (cm/sec) through

the passage measured by current meters (1992e1994).

Source: From Roemmich et al. (1996).

which the observed flow is southward/westward

at the continental boundary and northward/eastward

along the offshore side of the

trench (Figure 10.17; Owens & Warren, 2001).

The net DWBC transport is small (order 3 Sv)

and southward/westward, matching theory.

The net meridional overturning in the Pacific

consists of northward transport from the

10.7.1. Introduction

The Pacific equatorial current system is dominated

by strong zonal (east-west) flows with

weak meridional (north-south) currents in the

ocean interior (Figure 10.2a; Table S10.3 and

Figure S10.1 located on the textbook Web site).

At the sea surface there are three major zonal

currents. Below the surface there is a complex

set of reversing zonal flows. At the western

boundary, strong meridional currents connect

the zonal flows.

The three major zonal surface currents are (1)

the westward-flowing NEC between about 8 N

and 20 N, (2) the westward SEC from about 3 N

to 10 S, and (3) the narrow NECC flowing to the

east between them, centered at about 5 N. These

were well-known components of the Pacific

surface circulation before 1940. The other major

equatorial current lies just below the thin surface

layer of the SEC and it is the eastward-flowing

EUC, which is one of the fastest permanent

currents in the world. The eastward South Equatorial

Countercurrent (SECC) in the western South

Pacific between 10 and 12 S is much weaker

and more time-dependent than these. Then

TROPICAL PACIFIC CIRCULATION AND WATER PROPERTIES 333

FIGURE 10.17 Abyssal circulation schematics. After: Owens and Warren (2001), Johnsonand Toole (1993), Kato and Kawabe (2009),

Komaki and Kawabe (2009), Yanigomoto, Kawabe, and Fujio (2010), Whitworth et al. (1999), and Roemmich, Hautala, and Rudnick (1996).

a complicated set of subsurface eastward and

westward mean flows (Section 10.7.3) is evident.

The low latitude western boundary currents

(Section 10.7.4) collect water from the westward

SEC and NEC and feed it into the eastward

subsurface equatorial flows and into the eastward

NECC. The Mindanao Current is the primary

equatorward Northern Hemisphere boundary

current, connecting the westward flow of the

NEC to the eastward flow of the NECC. The

NGCUC is the primary equatorward Southern

Hemisphere boundary current, connecting the

westward flow of the SEC to the eastward flow

of the subsurface equatorial currents (EUC, North

334

10. PACIFIC OCEAN

(a)

(b)

0

1000

2000

3000

4000

5000

North Pacific 24°N


NPIW

NPIW/AAIW/UCDW

PDW 1

PDW 2

PDW 3

45.88

LCDW

26.2

26.9

27.6

36.96

45.84

45.88

Ekman

surface

26.2

26.9

27.6

36.96 2

45.84 4

45.88 4

bottom

North Pacific 24°N

Upper

NPIW

PDW3

Ekman

Upper with Ekman

compensation removed

AAIW/

UCDW

PDW1

PDW2

LCDW

–20 –15 –10 –5 0 5 10 15

80˚

80˚

60˚

60˚

6000

0 2000 4000 6000 8000 10000 12000

40˚

40˚

130 140 150 160 170 180 –170 –160 –150 –140 –130 –120

20˚

20˚

0˚

0˚

(c)

0

1000

2000

3000

4000

5000

30.00 31.00

32.00 33.00 34.00 34.30 34.50 34.60 34.66 34.69 35.00 35.50 36.00 37.00

South Pacific 28°S


27.1

Antarctic Intermediate Water/PDW1

27.6

PDW 2, Upper Circumpolar Deep Water

36.96

PDW3, UCDW

45.84

45.88LCDW1

LCDW2

45.92

LCDW3

36.96

6000

0 2000 4000 6000 8000 10000 12000

160 170 180 –170 –160 –150 –140 –130 –120 –110 –100 –90 –80

(d)

Ekman

surface

27.1

27.6

36.96 2

45.84 4

45.88 4

45.92 4

bottom

–20˚ –20˚

–40˚ –40˚

–60˚ –60˚

South Pacific 28°S

(Upper with Ekman

compensation and

10 Sv ITF removed)

–80˚ –80˚

PDW/

UCDW2

Ekman

Upper

AAIW/PDW1

PDW/

UCDW3

LCDW1

LCDW2

LCDW3

–20 –15 –10 –5 0 5 10 15

Volume Transport (Sv)

FIGURE 10.18 Salinity and meridional transport in isopycnal layers at 24 N (a, b) and at 28 S (c, d). Inset map shows

section locations. The isopycnals (s q , s 2 , s 4 ) that define the layers are contoured on the salinity sections. After Talley (2008).

Overturning transports from Ganachaud (2003) are shown in Figure S10.15 on the textbook Web site.

Subsurface Countercurrent, NSCC; and South

Subsurface Countercurrent, SSCC), and also

crossing the equator to meet the southward flow

of the Mindanao Current and feed the NECC.

Dynamics of the wind-driven equatorial

surface currents and the EUC were presented

briefly in Section 7.9.2, and directly on the

equator, flow is in the direction of wind stress

(in the frictional surface layer) and pressure

gradient force. Moving slightly away from the

equator, the Coriolis force quickly becomes

important; the currents are almost geostrophic


and the upper ocean circulation can be considered

in terms of the usual Sverdrup dynamics

driven by convergence of the wind-driven

Ekman layer (Section 7.5).

10.7.2. Tropical Wind and Buoyancy

Forcing

The tropical surface current system is driven

by the easterly trade winds at the ocean’s surface

(Figure 5.16, Figure S10.16 located on the textbook

Web site, and also the stick plot in

Figure 10.20b). The trade winds are part of the

atmosphere’s Walker and Hadley cells (Section

7.9.2). The trade winds are not uniformly westward;

these surface winds converge at the ITCZ

north of the equator. (A weak, secondary ITCZ

is found in the western South Pacific.) The

wind stress curl associated with the ITCZ is positive,

creating Ekman suction (Figure 5.16d). This

drives cyclonic circulation that is very zonally

elongated. This includes westward flow on the

northern side (part of the NEC), and eastward

flow on the southern side, which is the NECC

(Yu, McCreary, Kessler, & Kelly, 2000). The

SECC, which appears in the western South

Pacific tropics, is driven by a similar mechanism

associated with the Southern Hemisphere ITCZ.

Seasonally, the trade winds are stronger in

the winter hemisphere (Figure 5.16). The

Northern Hemisphere ITCZ lies closer to the

equator, at about 5 N in the east, in February

than in August. In August, the northern ITCZ

shifts northward to 10 N across the whole

Pacific. In the western tropical Pacific, there is

a seasonal monsoon, which is a reversal in

winds in the Northern Hemisphere and equatorial

regions. This especially impacts the surface

equatorial circulation (Section 10.7.3.1).

Airesea fluxes of heat and freshwater in the

tropical Pacific are important for the global

balances of both of these quantities (Figures

5.4, 5.12). The tropical oceans warm due to

high solar radiation (Figure 5.11a). The greatest

warming is in the equatorial cold tongue in the

eastern Pacific (see next section), where lower

surface temperatures result in reduced latent

and longwave heat losses, hence higher net

heating.

The tropical Pacific is also a region of net

precipitation. The precipitation is not uniform

(Figure 5.4). Beneath the ITCZ of the Northern

Hemisphere is a band of net precipitation. The

western Pacific is also a region of net precipitation,

concentrated in two bands centered at the

northern and Southern Hemisphere ITCZs.

The eastern tropical Pacific is a region of net

evaporation. These patterns are directly related

to the Hadley and Walker circulations, with

more precipitation where air rises along the

ITCZ and in the western tropical Pacific.

The net precipitation in the western tropics

creates a low salinity surface layer, with a strong

halocline beneath. The mean stratification here

consists of a so-called barrier layer,inwhichwarm

surface temperature extends to greater depth

than the fresh surface water. The mixed layer

stratification, therefore, is dominated by salinity.

10.7.3. Equatorial Pacific Current

Structure

10.7.3.1. Zonal Currents and Associated

Mid-Ocean Meridional Flows

Zonal flows dominate meridional flows in the

tropics, except at the western boundary. Average

upper ocean zonal velocity, temperature, and

salinity structure in the central Pacific (154 W)

is shown in Figures 10.19 and 10.20 (Wyrtki &

Kilonsky, 1984; WK). The deep equatorial

currents are described for a nearby longitude

(Figure 10.21). These zonal currents are geostrophic

except directly on the equator, 3 and are

therefore reflected in sloping surface dynamic

3 Geostrophy is valid to within about one-quarter degree of the equator with sufficient temporal averaging. A 12-month set

of 43 sections was used for the mean structure in WK.

336

10. PACIFIC OCEAN

FIGURE 10.19 Mean distributions

of surface dynamic height

(6D dyn cm) relative to 1000 db

(dyn cm) and vertical meridional

sections of zonal geostrophic flow

(U in cm/sec), temperature (T

in C), and salinity (S) between

Hawaii and Tahiti, for 12 months

from April 1979. ÓAmerican Meteorological

Society. Reprinted with

permission. Source: From Wyrtki and

Kilonsky (1984).


FIGURE 10.20 (a) Schematic of mean areas occupied by zonal currents between Hawaii and Tahiti for 12 months from April

1979. Dark shading indicates westward flow, light shading indicates eastward flow, blank areas have zonal speeds less than 2 cm/

sec. Acronyms: NEC, North Equatorial Current; NECC, North Equatorial Countercurrent; SEC, South Equatorial Current (three

sections); SECC, South Equatorial Countercurrent; UC, Equatorial Undercurrent (EUC in our notation); EIC, Equatorial Intermediate

Current; and NSCC/SSCC, Northern/Southern Subsurface Countercurrents (Tsuchiya jets). (b) Schematic meridional

section across the equator showing (top) the mean trade winds, (middle) surface circulation, and (bottom) schematic surface

dynamic topography, temperature structure, and meridional circulation below the surface. (“Countercurrent” ¼ “NECC” in our

notation.) ÓAmerican Meteorological Society. Reprinted with permission. Source: From Wyrtki and Kilonsky (1984).

338

10. PACIFIC OCEAN

FIGURE 10.21 Zonal velocity (cm/sec) in the equatorial

Pacific, averaged from 41 sections of direct current

measurements collected in 1982e1983. White is eastward

flow, gray is westward. Source: From Firing, Wijffels, and

Hacker (1998).

height (6D; Figure 10.19), and in isopycnal

slopes, which produce the vertical shear of the

geostrophic currents. The westward NEC and

the southernmost part of the westward SEC

(SEC-3) are the primary westward flows of the

North and South Pacific’s subtropical gyres

(Figure 10.1) and extend down through the

thermocline. Their dynamic heights slope downward

and isotherms tilt upward toward the

equator.

The eastward NECC is a strong, permanent

current that stretches across the whole width

of the Pacific with associated large dynamic

height and isotherm slopes in the opposite

direction to those of the NEC/SEC. In contrast,

the weak, eastward SECC is mostly restricted

to the western Pacific with only a weak expression

in the central Pacific seen as a slight

reversal in surface dynamic height slope

(Figure 10.19).

Atthesurfaceontheequator,thesurface

flowiswestward(SEC-1).ThisequatorialSEC

is in just a thin layer above the EUC. The equatorial

SEC’s flow is the downwind, frictional

equatorial response to the westward trade

winds, in the absence of the Coriolis force and

hence an Ekman layer (Section 7.9.2). It can

disappear at times since it is driven directly

by the wind, and in any case responds quickly

to changes in winds; a reversal to eastward

occurs regularly during westerly wind bursts

at the onset of El Niño (Section 10.8; Hisard &

Hénin, 1984).

At the equator, the EUC lies just beneath the

SEC. Its maximum velocity core at this central

Pacific location lies at 130 m, with average

speeds greater than 90 cm/sec. 4 The EUC was

considered to be weak during the WK measurement

period; it can regularly reach speeds of

120 cm/sec. Despite its thinness in the vertical,

its large speeds are reflected in large transport

(32.3 3.5 Sv in WK’s annual average). The

EUC is easily identified in isotherm structure

at the equator: the 13e26 C isotherms spread

upward above it and downward below it. It

has no expression in surface dynamic height

since it is not a surface current. This creates

the necessary geostrophic vertical shear on

4 The EUC was first discovered in 1951 when researchers from the U.S. Fish and Wildlife Research Service in Honolulu

found that their “long-line” deep fishing equipment drifted strongly eastward in spite of the westward surface currents.

Their gear traveled eastward at speeds of about 1.5 m/sec, which was about three times that of the westward surface

current. A subsequent cruise to investigate this phenomenon was led by Townsend Cromwell; the EUC is also called the

“Cromwell Current.” Unfortunately Dr. Cromwell died the next year in a plane crash on the way to an oceanographic

expedition. See Knauss (1960).


both sides of the equator to create a subsurface

eastward flow with westward flows both above

it (SEC) and below it (Equatorial Intermediate

Current; EIC).

The eastward NSCC and SSCC are just to the

north and south of the equator and slightly

deeper than the EUC. The NSCC is not always

easily distinguishable from the deeper part of

the surface-intensified NECC. In the isotherms

(Figure 10.19), the SCCs are apparent in the

strong upward slopes of the 10 and 11 C

isotherms away from the equator. The SCCs

were first identified from maps of properties

on isopycnals by Tsuchiya (1975). They transport

salinity, oxygen, and nutrients characteristic

of the western Pacific toward the east. In

honor of this first description, the SCCs are often

referred to as “Tsuchiya jets”.

The westward EIC is a weak but persistent

flow along the equator beneath the EUC. Beneath

the EIC, the reversing eastward and westward

flows between 1000 and 2000 m are referred to

as the equatorial stacked jets (Figure 10.21). Off

the equator, around 700e900 m depth, there are

also reversing zonal flows, but in thicker layers

with speeds around 15e20 cm/sec. The deepest

equatorial flows in Figure 10.21 have small

mean speeds, <5 cm/sec, but might be permanent

features. 5 Given the local topography at

159 W, which rises to 3000 m in the north, the

robust currents are south of the equator. Transports

of each of these flows is on the order of

several Sverdrups. Farther from the equator,

within 15e20 of the equator and above the

topography of the mid-ocean ridges (above

3000 m), the intermediate and deep circulation

remains dominantly zonal compared with flow

at higher latitudes. The zonal nature of the flows

is clear in float trajectories at 900 m (Davis, 2005;

Figure S10.13 on the textbook Web site), in steric

height maps for these mid-depths (Figure 10.2b),

and in ocean properties on isopycnals. At 2500 m,

flow includes a narrow eastward tongue at

about 2 S and broad flanking westward flows

centered at 5e8 N and at 10e15 S (Talley &

Johnson, 1994). At the bottom, the westward

equatorial flow is possibly fed by broader eastward

flow north of the equator (Johnson & Toole,

1993). These complex, zonal deep flows are likely

wind-forced (Nakano & Suginohara, 2002).

Returning to the upper ocean, meridional

flows in the equatorial Pacific (Figure 10.20b)

are associated with the major zonal currents.

At the sea surface, the easterly trade winds

cause Ekman transport to the north in the

Northern Hemisphere and to the south in the

Southern Hemisphere. This results in equatorial

divergence, which creates equatorial upwelling.

(There is equatorial downwelling if the winds

shift to westerly, as in the western equatorial

Pacific at the beginning of an El Niño event.)

The equatorial upwelling is fed by equatorward

subsurface flow. The inflow is in the thermocline,

based on water properties, including

salinity (Figure 10.19 “S” panel). The equatorward

inflow can be geostrophic, due to the

west-to-east pressure gradient force set up by

the westward flow of surface water along the

equator to the western boundary. This creates

high pressure in the west and low pressure in

the east.

10.7.3.2. Zonal Structure of the

Equatorial Currents

The equatorial current system extends from

at least 143 E (north of Papua, New Guinea) to

the Galapagos Islands (90 E) and then eastward

to the coast of Ecuador, a distance of approximately

15,000 km. The sea surface is high in

the west and slopes down to the east in the

equatorial band (Figure 10.2 and Figure S10.1

on the textbook Web site). The west-east difference

in surface height is 40e60 cm, with significant

interannual variability associated with

ENSO; the largest slopes occur during La Niña

(Figure 10.22c). Surface dynamic height shows

5 According to Firing (1989), “a 10 year time-series would be ideal for studying annual and interannual variations.”

340

10. PACIFIC OCEAN

(a)

(b)

SST (°C) and Wind

(c)

20°C isotherm depth and Wind

Dynamic height (dyn cm rel. to 500 dbar) and Wind

FIGURE 10.22 (a) SST; (b) depth of the 20 C isotherm, which is an indicator of thermocline depth; and (c) dynamic

height (dyn cm), with superimposed wind velocity vectors, during a period of a well-developed cold tongue (La Niña;

August 2007). Source: From TAO Project Office (2009a). (d) Primary productivity (mg C m 2 day 1 ) based on ocean color,

during a La Niña (July 1998). Source: From McClain et al. (2002).

the same west-east contrast of about 40 dyn cm

(Figure S10.17 on the textbook Web site). The

equatorial sea-surface height slope is due to

the wind-driven westward flow of surface water

in the SEC along the equator. This piles warm

water up in the west, in the region called the

warm pool. The westward equatorial flow is

also associated with equatorial upwelling in

the east. The cold, upwelled surface water in

the east is called the cold tongue. These structures

are obvious in mean SST (Figures 4.1 and 10.22).

Along-equatorial sections of potential temperature,

salinity, and potential density show the

warmer, lighter water to the west and colder,

denser surface water to the east. Surface nutrients

have a similar structure, with higher nutrients

in the cold tongue and nearly complete

depletion in the warm pool (Figure 4.22).


Cold water along the equator has two sources:

upwelling in the eastern Pacific due to the

westward surface flow (SEC) driven along the

equator by the trade winds, and upwelling

due to divergent Ekman transport just off the

equator, also due to the trade winds, which

can occur at all longitudes. Because the warm

pool in the western Pacific is so thick, the Ekman

divergence component of the upwelling does

not bring cold water to the sea surface there.

The pileup of water in the west causes an

eastward pressure gradient force along the

equator. This pressure gradient force drives

the eastward flow of the EUC. The west-to-east

pressure gradient force also creates equatorward

geostrophic flow that feeds the equatorial

upwelling.

The equatorial pycnocline is deep in the west

and tilts upward toward the east (Figure 10.23).

This upward tilt compensates the downward

sea-surface tilt such that the pressure gradient

force along the equator beneath the pycnocline

is very weak. In fact the equatorial flow beneath

the EUC is weakly westward (EIC). The EUC is

located within the pycnocline (Figure 10.23c). It

shoals toward the east along with the pycnocline

(Figure 10.23). It is weak in the western

equatorial Pacific, with speeds less than 40 cm/

sec. It speeds up east of the date line, and reaches

maximum strength around 140 W. This corresponds

to longitudes of greater eastward pressure

gradient force, evident in surface height

and dynamic height. Its transport peaks at

about 2.5 Sv in the central Pacific (Leetmaa &

Spain, 1981).

At the western boundary, the EUC is fed by

the saline NGCUC (Section 10.7.4). As the EUC

flows eastward it encounters the Galapagos

Islands, located on the equator at 91e89 W.

The EUC splits upstream of the islands at about

92 W and flows north and south around the

islands. The southern part is stronger; the

main core of the EUC core is actually slightly

south of the equator from 98 W. East of the Galapagos,

part of the EUC penetrates southeast to

FIGURE 10.23 Mean equatorial (a) potential temperature

( C), (b) salinity, and (c) zonal velocity (cm/sec).

Eastward velocities are shaded. Source: From Johnson et al.

(2002).

5 S and joins the Peru Countercurrent at the

surface and the PCUC at the South American

coast (Section 10.4.1.3; Lukas, 1986).

Other zonal “asymmetries” are apparent in

the other major tropical currents. The eastward

NECC shifts northward toward the east. The

342

10. PACIFIC OCEAN

eastward NSCC and SSCC both shift poleward

as well. The SECC is present permanently only

in the western Pacific and disappears by the

mid-Pacific.

10.7.3.3. Equatorial Upwelling and

Biological Productivity

The Pacific equatorial SST structure is

strongly influenced by upwelling of cold water

from the pycnocline/thermocline. Where the

thermocline is shallow, upwelling creates cold

surface temperature; where and when the thermocline

is deep, upwelling is not as effective

in cooling the surface. The cold tongue and

warm pool are evident in satellite images in

non-El Niño years (Figures 4.1 and 10.24).

Coastal upwelling along Ecuador is also

evident, joining with the equatorial cold tongue.

Upwelled water is often richer in nutrients

than the displaced surface water. Global maps of

surface nutrients show a maximum in the Pacific

cold tongue (nitrate in Figure 4.24), because of the

eastward shoaling of the pycnocline, which is also

the nutricline. This nutrient maximum promotes

biological production. Biological productivity,

measured in amount of carbon produced per

area per day, is high in the upwelled water of

the cold tongue (Figure 10.22d, from a La Niña

period of enhanced upwelling). This calculation

of productivity was based on ocean color from

the SeaWIFs satellite (Figure S10.318 on the textbook

Web site).

10.7.4. Low Latitude Western

Boundary Currents

The Mindanao Current is a 200 km wide

western boundary current that flows southward

along the western boundary of the tropical

North Pacific. Dynamically, it is the western

boundary current associated with the Sverdrup

transport of the elongated tropical cyclonic gyre.

The Mindanao Current carries subtropical

North Pacific waters toward the equator,

including saline water from the subtropical

thermocline and traces of North Pacific Intermediate

Water (Bingham & Lukas, 1994).

The Mindanao Current forms near 14 N

where the westward-flowing NEC splits, with

the northward flow forming the Kuroshio

(Figure 10.1 and Figures S10.1 and S10.19 on

the textbook Web site). It turns eastward at

about 5 N and feeds the NECC. Mindanao

Current speeds are typical of western boundary

currents, reaching a maximum of 100 cm/sec.

Volume transport estimates range from 20 to

40 Sv, consistent with the calculated Sverdrup

transport (Wijffels, Firing, & Toole, 1995).

The Mindanao Eddy (ME in Figure 10.1) is

a recirculating cyclonic feature at the western

boundary of the cyclonic tropical gyre. It forms

between the westward NEC and the eastward

NECC. Its western side is the Mindanao

Current. The Halmahera Eddy (HE in

Figure 10.1) is an anticyclonic feature at the

western boundary just north of the equator

between the eastward NECC and the westward

SEC. The Halmahera Eddy mixes waters from

the North and South Pacific. The properties of

waters that enter the Indonesian Throughflow

(ITF) may therefore depend on the activity of

this eddy (Kashino et al., 1999). Both eddies

are highly dependent on wind forcing.

The NGCUC is the northward western

boundary current of the tropical South Pacific.

The NGCUC is the northernmost part of the

western boundary current that forms from the

westward flow of the SEC (Qu & Lindstrom,

2002), which splits at the Australian coast,

with the southward flow forming the EAC

(Section 10.4.1.1). The split is at 15 S at the sea

surface and shifts poleward to 23 S at 800 m.

The northward boundary current north of 15 S

is referred to as the North Queensland Current

(NQC). (The northward subsurface flow

between 23 S and 15 S is called the Great Barrier

Reef Undercurrent (GBRUC).) The NQC flows

through the Coral Sea, through the Solomon

Sea, and then through the Vitiaz Strait between

New Guinea and New Britain. Beyond that


point it is referred to as the NGCUC. The

NGCUC turns north and then east along

the equator at about 143 E to feed the EUC.

The NGCUC has speeds of 50 cm/sec centered

at 200 m depth and a transport of 7 Sv at 2 S,

which are equivalent to those of the EUC at

the equator.

Lastly, the tropical Pacific and Indian Oceans

are connected via the Indonesian Throughflow,

through the complex passages of the Indonesian

archipelago (Figure 11.11; Section 11.5). Approximately

10e15 Sv flow through the passages,

with significant variability, much of it due to

ENSO. The Pacific’s low latitude western

boundary currents are the source of the ITF.

The flow through the Makassar Strait originates

in the Mindanao Current. South Pacific waters

from the NGCUC enter the Halmahera Sea;

deeper South Pacific waters from the same

source enter through Lifamatola Strait (Hautala,

Reid, & Bray, 1996).

10.7.5. Equatorial Property

Distributions

Although most of the Pacific water mass

description is in Section 10.9, we briefly review

the tropical upper ocean distributions in this

section because they are so clearly linked to

the equatorial current system.

The temperature structure is highly symmetric

about the equator (Figure 10.19). The

thermocline is most intense a few degrees north

and south of the equator, with the isotherms

spreading apart north and south of the 10 N

and 10 S parallels. At the equator, the spreading

of the isotherms marks the core of the EUC.

Below the thermocline, between about 5 S and

12 N, there is a marked thermostad (low

vertical gradient).

For salinity, there is little symmetry across the

equator (Figure 10.19), because the South Pacific

is more saline, because the SEC reaches the

equator and the NEC does not, and because of

the Northern Hemisphere location of the ITCZ.

Salinity maximum layers are subducted equatorward

from both the South and North Pacific

subtropical evaporation maxima; their core

salinities are 36.2 psu and 35.0 psu, respectively.

(These are the Subtropical Underwaters, also

called Tropical Waters in Johnson & McPhaden,

1999.) Because the SEC extends to the equator,

the salinity maximum at the equator comes

directly from the South Pacific subtropical

gyre. The South and North Pacific salinity

maxima are separated laterally by lower salinity

arising from the California Current and downward

diffusion beneath the rainy ITCZ (Johnson

& McPhaden, 1999). The lowest surface salinity

is in the NECC, which lies directly below the

ITCZ. The subsurface low salinity water

entering at about 20 N at 300 m is the North

Pacific Intermediate Water (Section 10.9.2.1).

10.7.6. Intraseasonal and Seasonal

Variability

The equatorial Pacific includes temporal variability

at intraseasonal (20e30 days), seasonal,

monthly-to-interannual, interannual (3e7

years), and interdecadal (10e30 years) timescales.

The most energetic intraseasonal variations

are the Tropical Instability Waves (TIWs).

Seasonal variability includes response to

changes in location and strength of the ITCZs

in both hemispheres. Other variability at weekly

to interannual periods is associated with Rossby

and Kelvin waves (Section 7.7) and at interannual

and longer periods, with ENSO and other

climate modes (Section 10.8 and Chapter S15

on the textbook Web site).

TIWs are large cusp-like spatial oscillations

in SST along the northern edge of the cold

tongue (Figure 10.24) (Legeckis, 1977). The

oscillations are also apparent in ocean color/

chlorophyll (McClain et al., 2002). TIWs have

wavelengths of about 1000 km. The TIW pattern

propagates westward at an average phase

speed of 30 to 50 cm/sec, resulting in a period

of about 20 to 30 days. The TIWs are principally

344

10. PACIFIC OCEAN

FIGURE 10.24 Tropical instability waves. SST from the Tropical Rainfall Mapping Mission (TRMM) Microwave Imager

(TMI) for two successive 10-day periods in August 1998, after establishment of the cold tongue during a La Niña. A more

complete time series (June 1eAugust 30, 1998) is reproduced in Figure S10.20 on the textbook Web site. This figure can also

be seen in the color insert. TMI data are produced by Remote Sensing Systems and sponsored by the NASA Earth Science

MEASURES DISCOVER Project. Data are available at www.remss.com. Source: From Remote Sensing Systems (2004).

due to (barotropic) instability arising from the

horizontal shear between the SEC and the

NECC (Philander, 1978). TIWs are shallow

(100e200 m thick) because the high velocities

of the currents that create them are surfaceintensified.

TIWs appear in summer (June) when the

ITCZ migrates northward and the trade winds

accelerate the portion of the SEC that lies north

of the equator (Vialard, Menkes, Anderson, &

Balmaseda, 2003). In the time series leading to

Figure 10.24 (Figure S10.20 on the textbook

Web site), the equatorial cold tongue emerges

in early June; by June 10 the tongue shows

north-south oscillations due to TIWs. Closed

anticyclonic vortices are found in the troughs

of the waves. Seasonal wind forcing in the tropical

Pacific directly affects SST and the surface

and upper ocean currents (Figure 10.25). The

cold tongue is strongest in August-September,

during the period of strongest trade winds,

accompanied by warmest temperatures in the

warm pool; the west-east contrast is as much

as 10 C. By March, both temperature features

are much weaker and the west-east contrast is

reduced to about 5 C. (The large interannual

variability superimposed on the annual cycle

in Figure 10.25 is due to ENSO.)

The equatorial part of the SEC, which

responds frictionally to the wind stress, varies

mostly in phase with the seasonal winds. The

EUC, which responds to the west-east pressure

gradient set up by the SEC, has a more complicated

response that lags the winds. Johnson,

Sloyan, Kessler, and McTaggert (2002) provided

detailed discussion of the seasonal variability

of each of the upper ocean currents, phasing

with the winds, and spatial structure. Dramatic

seasonal variability occurs just offshore of the

Central American mountain chain (Figure

10.26). Trade winds from the Atlantic funnel

through three major gaps in the mountains,

with wintertime winds reaching 20 m/sec

during several 5- to 7-day-long events. The

wind jets (Tehuantepec, Papagayo, and Panama)

that emerge over the Pacific force dramatic

local circulations and upper layer mixing,

resulting in cool SST (Chelton, Freilich, &

Esbensen, 2000) and ocean color anomalies;

the effects are visible even in the global mean

wind stress curl map from Chelton et al.

(2004; Figure 5.16d).


FIGURE 10.25 Zonal wind speed and SST in the equatorial Pacific to illustrate the annual cycle. Positive wind speed is

toward the east. Climatological means in February and August and an expanded time series for 2000e2007 are shown in

Figure S10.21 on the textbook Web site, to emphasize the seasonal cycle. This figure can also be found in the color insert.

Source: From TAO Project Office (2009a).

FIGURE 10.26 Tehuantepec eddies

evident in sea surface height anomalies from

satellite altimetry in February, 1994. Source:

From Palacios and Bograd (2005).

Anticyclonic eddies are produced by the

wind jets, as a combination of eddy shedding

from the coastal circulation system (coastally

trapped waves) and the strong wind stress

curl in the jets. The eddies propagate offshore.

The best known are the Tehuantepec eddies

(Figure 10.26 and Figure S10.22 on the textbook

Web site). Three to four Tehuantepec eddies

and two to three Papagayo eddies form each

year between October and July with greater

346

10. PACIFIC OCEAN

frequency and intensity during El Niño years

(Palacios & Bograd, 2005).

10.8. EL NIÑO/ LA NIÑA AND THE

SOUTHERN OSCILLATION (ENSO)

El Niño/La Niña is a natural climate variation

that is dynamically centered in the tropical

Pacific. Its “interannual” timescale is 3 to 7 years

for quasi-periodic alternation between the El

Niño and La Niña states. The Southern Oscillation

is an index based on the pressure difference

between two tropical South Pacific locations,

and is closely related to the El Niño state.

Because this index is so closely related to El

Niño events, the full climate phenomenon is

often referred to as El Niño-Southern Oscillation

(ENSO). The ocean and atmosphere are fully

coupled in this climate “cycle.” The coupling

is referred to as the Bjerknes feedback (Section

7.9.2; Bjerknes, 1969).

An El Niño event is marked by an unusual

excursion of warm water (>28 C) to the east in

the equatorial zone, associated with weakened

southeast Trade Winds in the east and stronger

westerlies in the west. La Niña is the opposite d

stronger southeast Trades in the east (and

weak westerlies in the far west) with resulting

cool water (<25 C) extending much further

westward along the equator than usual. The

alternation between states is not regular since

there are many different oceanic and atmospheric

phenomena linked in the full system

plus random, short-term forcing. Therefore,

ENSO predictability is not like that of, say, the

tides, which are forced by very regular, predictable

progressions in the orbits of the earth,

moon, and sun.

El Niño/La Niña events have large and

sometime devastating impacts on ocean

ecosystems, particularly along the South American

coast, but also as far north as the CCS.

ENSO impacts air temperature and precipitation

on global scales (Figures S10.24 and

S10.25 on the textbook Web site), via propagation

of large-scale waves through the atmosphere

and propagation of Kelvin waves

(Section 7.7.6) along the eastern boundary of

the Pacific. Precipitation anomalies during a

composite El Niño include regions of anomalously

low precipitation that are susceptible to

drought and fire and high precipitation that

are susceptible to flooding. Although it is not

located in the tropics, U.S. air temperatures

are affected by ENSO; El Niño signatures

include anomalous warmth over the northwest

and high plains; cool temperatures in the south

and Florida; anomalously dry conditions in the

northwest, east, and Appalachians; and wet

conditions from California through the southeastern

U.S.

Early ideas about the cause of El Niño

centered on local mechanisms along the South

American coast, for instance that the alongshore

winds off Peru changed to lessen or stop the

coastal upwelling. More intensive studies in

the early 1970s, motivated by a major El Niño

event in 1972 that resulted in the collapse of

the Peru/Ecuador anchovy fishery, showed

that El Niño has a much larger geographic scale.

Rasmusson & Carpenter’s (1982) canonical description

of ENSO, based on El Niño’s from

1949 to 1980, was the underpinning for an international

project (Tropical Ocean Global Atmosphere;

TOGA) to study ENSO (1985e1995).

TOGA planning was underway when the strong

El Niño event of 1982/83 provided additional

impetus for the experiment. The importance of

ENSO analysis and prediction is such that

a massive permanent observing system has

been deployed in the tropical Pacific since the

1980s (TAO and TRITON; Section S6.5.6 on the

textbook Web site).

Excellent, regularly updated information, including

background information on dynamics

and impacts, forecasts, and links to many different

ENSO products, is available from several

different Web sites administered by the National

Oceanic and Atmospheric Administration.

EL NIÑO/ LA NIÑA AND THE SOUTHERN OSCILLATION (ENSO) 347

10.8.1. ENSO Description

We first recall the “normal” ocean and atmosphere

conditions in the tropical Pacific (Section

7.9.2; Figure 10.27b). The easterly trade winds

pile up warm equatorial water in the western

tropical Pacific and cause upwelling along the

equator. This causes the cold tongue in SST in

the eastern tropics, and causes the thermocline

to be inclined upward from west to east (Section

10.7.3). The warm-to-cold SST difference along

the equator maintains the Walker circulation in

the atmosphere, thus sustaining this component

of the trade winds. This is an equilibrium state

of the simple coupled ocean-atmosphere, and

the system would remain in this state if it did

not include large-scale propagating waves

such as Kelvin and Rossby waves.

The exaggerated, strong version of the normal

state is the La Niña state (Figure 10.27a). In La

Niña, the warm SST shifts slightly more to the

west, the thermocline is a bit deeper in the

west, the sea surface is higher in the west and

lower in the east, and the Walker circulation in

the atmosphere is stronger.

In an El Niño state, the trade winds are

weaker, because the Walker circulation is

weak or reversed, and the thermocline is more

level (Figure 10.27c). The cold tongue in the

east weakens and disappears, due to both relaxation

of the thermocline and eastward movement

of warm water from the central and

western tropical Pacific. This does not indicate

an absence of upwelling but rather that warm

water is now occupying the eastern tropical

Pacific; the schematic shows easterly trades in

the eastern Pacific, but these upwell only

warm water from the now thicker and warmer

surface layer.

In the time series of SST along the equator

(Figure 10.25), warm SST and weaker trade

winds mark several El Niño events, with the

opposite markers for La Niña events. Time

series indicating the occurrence of El Niño and

La Niña are constructed in various ways. The

first index, the Southern Oscillation Index (SOI)

is the difference in atmospheric pressure

between the western and eastern tropical South

Pacific; meteorological stations at Darwin, Australia,

and Tahiti are used in the SOI because

observations have been made in these locations

for a very long time. Every El Niño event is associated

with low SOI. However, not every SOI

low corresponds to an El Niño. Several indices

are based on SSTs averaged spatially over

portions of the eastern tropical Pacific because

this reflects conditions in the cold tongue (e.g.,

Oceanic Niño Index in Figure 10.28). A multivariate

index based on SST, sea level pressure,

surface air temperature, surface wind, and

cloudiness is also useful (Wolter & Timlin,

1993; Wolter, 2009). Very long time series have

been reconstructed from proxies of temperature

measured in coral heads (Cobb, Charles, Cheng,

(a)

La Niña Conditions

(b)

Normal Conditions

(c)

El Niño Conditions

Convective

Circulation

Equator

Equator

Equator

Thermocline

Thermocline

Thermocline

120°E 80°W

120°E 80°W

120°E 80°W

FIGURE 10.27 (a) La Niña, (b) normal, and (c) El Niño conditions. This figure can also be found in the color insert.

Source: From NOAA PMEL (2009b).

348

10. PACIFIC OCEAN

FIGURE 10.28 (a) Correlation of monthly SST anomalies with the ENSO Nino3.4 index, averaged from 1948 to 2007. The

index is positive during the El Niño phase, so the signs shown are representative of this phase. (Data and graphical interface

from NOAA ESRL, 2009b.) This figure can be found in the color insert. (b) “Oceanic Nino Index” based on SST in the region

5 Nto5 S and 170 W to 120 W. (Data from Climate Prediction Center Internet Team, 2009). Gray and black correspond to El

Niño and La Niña, respectively. Additional indices representing ENSO and the correlation of monthly sea level pressure

anomalies with the ENSO Nino3.4 index are shown in Figure S10.23 on the textbook Web site.

EL NIÑO/ LA NIÑA AND THE SOUTHERN OSCILLATION (ENSO) 349

& Edwards, 2003). Long-term reconstructions of

tropical Pacific SST show that El Niño events at

2e7 years are ubiquitous, although intensity

and duration have varied. Well-documented El

Niño events took place in 1941e1942,

1957e1958, 1965e1966, 1972e1973, 1977e1978,

1982e1983, 1997e1998, and 2002e2003. The

events of 1982e1983 and 1997e1998 were the

largest recorded since the 1880s.

The global reach of ENSO is apparent in

correlations of SST and sea level pressure with

an ENSO index (Figure 10.28b and Figures

S10.23d on the textbook Web site). SST in the

equatorial Pacific shows the pattern previously

described of anomalously warm eastern equatorial

waters during the El Niño phase. The even

simpler sea level pressure pattern extends well

into the ACC region in an alternating zonal

pattern that is similar to that of the Southern

Annular Mode (Section 10.10 and Chapter S15


10.8.2. ENSO Mechanisms

The Bjerknes (1969) feedback is at the heart of

ENSO (Section 7.9.2), but it does not describe

how each stage of ENSO develops or why there

is a transition from one state to another with an

“oscillation” timescale of 3 to 7 years. Bjerknes

speculated that the transition results from

ocean dynamics but could go no further. An

oscillation with a period of several years can

be produced with a model that includes an

eastward-propagating equatorial Kelvin wave

that reflects at the eastern boundary, producing

westward-propagating Rossby waves (Cane,

Münnich, & Zebiak, 1990; Jin, 1996; Van der

Vaart, Dijkstra, & Jin, 2000).

The ENSO cycle, based on Rasmusson &

Carpenter (1982), Jin (1996), and Van der Waart

et al. (2000), is very briefly summarized here.

Moving from normal conditions toward a fullblown

El Niño, the steps are (1) changes of the

trade winds to westerly winds in the western

Pacific, often associated with the atmosphere’s

30e60 day Madden-Julian oscillation; (2) an

oceanic Kelvin wave shooting eastward along

the equator in response; (3) resultant warm

SST anomalies in the eastern and central equatorial

Pacific; and (4) disruption of the Walker

circulation through SST feedback on the atmosphere.

The “recharge oscillator” that transitions

this back toward a La Niña occurs when:

(1) the Kelvin wave reflects at the eastern

boundary and produces westward-propagating

Rossby waves, (2) the Rossby waves move

warm water away from the equator which

weakens the equatorial SST warm anomaly, (3)

the trade winds strengthen a little in response

to the somewhat cooler SST, (4) the strengthened

trades begin pushing the thermocline

back toward a normal state, and (5) Bjerknes

feedback then creates a La Niña state.

The adjustment timescale of this nearly free

oscillation yields the 3e7 year ENSO timescale.

An important property of this system is a delay

between the change in thermocline depth in the

western Pacific and the SST warming in the

eastern Pacific, which can be explained

partially by the Kelvin wave propagation (Jin,

1996).

The actual ENSO system is nonlinear and

messy. Fedorov et al. (2003) described it as

a “slightly damped, swinging pendulum sustained

by modest blows at random times.” The

switch from one state to another, and the intensity

and duration of the resulting state, depend

on many factors. These include phasing of

the shifts relative to the seasonal cycle and also

to the occurrence, timing, and intensity of westerly

wind bursts in the western tropical Pacific

that are associated with the intraseasonal

(30e60 day) Madden-Julian Oscillation in the

atmosphere. Predictability of onset, intensity,

and duration of events is therefore limited.

Because of the widespread economic impacts

of ENSO, skillful prediction several months

ahead has been a goal for many decades. Two

approaches are dynamical modeling and statistical

modeling. Dynamical models use a coupled

350

10. PACIFIC OCEAN

ocean-atmosphere model with initial conditions

based on observations. Statistical models use

observed parameters such as SST or heat content

and winds with a regression method to forecast

ENSO several months ahead. Forecasts are

generally probabilistic, meaning that an

ensemble (large number) of model runs is

made with slightly varying initial conditions.

Given the randomness of “triggering” mechanisms,

Philander and Fedorov (2003) and

Fedorov et al. (2003) highly recommended this

approach. The International Research Institute

for Climate and Society (IRI) at Columbia

University currently monitors 15 dynamical

and 8 statistical model forecasts (http://iri.

columbia.edu/climate/ENSO/currentinfo/SST_

table.html).

10.9. PACIFIC OCEAN WATER

MASSES

Pacific Ocean water properties, like those of

the other oceans, can be considered in four layers

(Section 4.1). The upper ocean layer contains the

mixed layer and main pycnocline (thermocline/

halocline), and is in broad contact with the

atmosphere. The intermediate layer contains

two low salinity water masses that originate at

the sea surface of the subpolar/subantarctic latitudes.

The deep layer contains two deep water

masses, one from the North Pacific and one

from the Southern Ocean. The North Pacific

deep water “source” is entirely internal mixing

and upwelling of waters from the Southern

Ocean, with no contact with the atmosphere.

The Southern Ocean deep water source contains

a mixture of deep waters from all three oceans

(Atlantic, Indian, and Pacific) as well as waters

that are locally ventilated in the Southern Ocean.

The bottom layer contains the densest water that

escapes northward from the Southern Ocean.

The distinction between the deep and bottom

layers is not sharp, and is usually based on the

direction of net meridional transport in the two

layers, with net southward transport in the

deep layer and net northward in the bottom

layer.

The Pacific Ocean is the freshest of the three

main ocean basins. The Atlantic and Indian

Oceans are both net evaporative basins, and

therefore have high overall salinity. The Pacific

evaporation-precipitation balance is nearly

neutral which makes the Pacific fresher than

the Atlantic and Indian.

The most important distinguishing process

for Pacific Ocean water properties is the lack

of a surface source of very dense water in the

North Pacific. This differs entirely from the

Atlantic Ocean. The densest water formed

locally is the relatively light North Pacific Intermediate

Water. On a global scale, the Pacific

Ocean is the low density end-member of the

overturning circulation. Its bottom waters,

which originate in other oceans, are salty and

its upper waters are relatively fresh; cooling to

the freezing point, which occurs in the Bering

and Okhotsk Seas in the northwest Pacific,

cannot increase the surface water density to

a high enough value to punch through to the

deep and bottom layers.

A potential temperature-salinity (T-S) diagram

that represents the major water masses is shown

in Figure 10.29. Table S10.4 on the textbook Web

site lists the principal water masses and an abbreviated

description of the process that initially

forms each water mass. The Pacific World Ocean

Circulation Experiment (WOCE) Hydrographic

Programme Atlas (Talley, 2007) is a comprehensive

source of sections, maps, and property plots.

10.9.1. Pacific Ocean Upper Waters

Pacific surface temperature (Figure 4.1) shows

the usual tropical maximum with poleward

decrease in temperature in both hemispheres.

The highest temperatures (>29 C) are in the

equatorial warm pool. The lower temperatures

of the equatorial cold tongue are also evident.

Isotherms in the PCCS and CCS are deformed,

PACIFIC OCEAN WATER MASSES 351

FIGURE 10.29 Potential T-S

curves for selected stations (inset

map). Acronyms: NPCW, North

Pacific Central Water; SPCW, South

Pacific Central Water; NPSTUW,

North Pacific Subtropical Underwater;

SPSTUW, South Pacific

Subtropical Underwater; NPSTMW,

North Pacific Subtropical Mode

Water; SPSTMW, South Pacific

Subtropical Mode Water; NPIW,

North Pacific Intermediate Water;

AAIW, Antarctic Intermediate

Water; DtW, Dichothermal Water;

MtW, Mesothermal Water; CCS,

California Current System waters;

and PCCS, Peru-Chile Current

System Waters. Mean T-S curves are

shown for every 10 degrees square

in Figure S10.45 on the textbook

Web site. This figure can also be


with colder water near the coasts due to equatorward

advection and upwelling. The coldest

temperatures are in the sea ice areas of the

Okhotsk and Bering Seas, and in the Antarctic.

Pacific surface salinity shows the typical

maxima in the subtropics, in the major subtropical

evaporation centers (Figures 4.14, 5.4).

There is a north-south minimum in the tropics

beneath the ITCZ at 5e10 N, due to excess

precipitation. Salinity is also low at high latitudes

due to excess precipitation. The surface

salinity in the North Pacific is considerably

less than in the North Atlantic, because of the

greater runoff and precipitation. In the South

Pacific the average surface salinity is higher

than in the North Pacific but is lower than in

the South Atlantic.

In the subtropics, there are two important

processes for creating upper ocean waters:

subduction of surface waters equatorward and

downward beneath less dense, lower latitude

surface waters, and production of thick, wellmixed

layers on the warm side of strong current

fronts such as the Kuroshio. These result in

several recognized subtropical water masses

(Table S10.4 on the textbook Web site) as

follows.

The waters that make up the thermocline/

pycnocline in the subtropics are called Central

Waters (Figure 10.29), as also found in the

Atlantic and Indian Oceans. The pycnocline, or

Central Water, is created by subduction and diapycnal

mixing (Section 9.8.1). “Central Water” is

a T-S relation with a large range of temperatures

352

10. PACIFIC OCEAN

and salinities, rather than an extremum of some

property.

North Pacific Central Water (NPCW) extends

from the NECC to about 40 N and is the freshest

of the Central Waters of the world’s oceans

(Figure 4.7). It is separated from the eastern

boundary by another, yet fresher water mass

that characterizes the CCS. This fresher CCS

water is advected southward from the eastern

subpolar gyre.

South Pacific Central Water (SPCW) is saltier

than NPCW since the South Pacific is saltier

overall. SPCW extends from about 10 S southward

to the Subantarctic Front at about 55 S.

Similar to NPCW, SPCW is separated from the

eastern boundary by another, fresher water

mass within the PCCS, advected northward

from fresher high latitude surface waters.

A second water mass associated with subtropical

subduction in both hemispheres is the

Subtropical Underwater (STUW), or subtropical

salinity maximum water. This is identified as

a shallow salinity maximum on the equatorward

part of the subtropical gyre (Figure 10.30). STUW

results from subduction of the very high salinity

surface water in the center of each subtropical

gyre. STUWis found on every meridional section

in the Pacific between 25 S and 25 N. It is very

shallow, with its salinity extremum no more

than 200 m deep, because the isopycnals that

FIGURE 10.30 Salinity: (a) along 165 W (WOCE P15); (b) at neutral density 24.0 kg/m 3 , characteristic of STUW; and

(c) at neutral density 26.00 kg/m 3 , characteristic of SPSTMW. The isopycnals intersect the surface along the dashed contours

Gray contours in (c) indicate winter outcrops. Source: From WOCE Pacific Ocean Atlas, Talley (2007).


outcrop in the surface salinity maximum water

are warm (~26 and 24 C in the South and North

Pacific, respectively) and low density (s q ~ 24.0

and 23.5 kg/m 3 in the South and North Pacific,

respectively).

The third subtropical water mass that we

single out is Subtropical Mode Water (STMW;

Masuzawa, 1969). “Mode” means relatively

large volume on a volumetric potential T-S

diagram. Mode Water is a pycnostad embedded

in the main pycnocline; it results from subduction

of the especially thick winter mixed layers

on the warm side of the separated western

boundary currents (Kuroshio and EAC;

Hanawa & Talley, 2001; Figure S10.26 on the

textbook Web site). The STMW in the North

Pacific (NPSTMW) is in the temperature range

16e19 C and centered at potential density

s q ¼ 25.2 kg/m 3 (Figure 10.29 and Figure

S10.26b on the textbook Web site). It originates

in winter as a thick mixed layer just south of

the Kuroshio. The thick layers subduct into the

general region of the western subtropical gyre

and are evident within the thermocline (Figures

10.31a and S10.26c on the textbook Web site).

The temperature of the STMW is highest

(>18 C) just south of Japan and decreases

toward the east.

In the South Pacific, the South Pacific STMW

(SPSTMW) is present north of the Tasman Front

and East Auckland Current (Figure 10.31b

and Figure S10.26c on the textbook Web site;

Roemmich & Cornuelle, 1992). Its core temperature,

salinity, and density are 15e17 C (just north

of New Zealand) and 17e19 C (region north of

29 S), 35.5 psu, and s q ¼ 26.0 kg/m 3 (SPSTMW

in Figures 10.29 and 10.30). Thus it has the same

temperature range as NPSTMW. It is denser

because it is somewhat more saline, because the

South Pacific is saltier than the North Pacific.

SPSTMW is the weakest of the global STMWs;

without a supplementary vertical density

gradient calculation, the widening of isopycnals

and isotherms on intersecting vertical sections is

somewhat difficult to discern (Figure 10.31b).

The North Pacific’s subpolar gyre is a region of

Ekman upwelling rather than downwelling.

Therefore there is no wind-driven subduction.

Surface densities increase along the cyclonic

path around the gyre; they are higher in the

west than in the east, and are highest in the

Okhotsk Sea, along Hokkaido and just south of

Hokkaido. In the regions of highest surface

density, the densest (intermediate) North Pacific

waters are formed (Section 10.9.2).

The combination of low surface salinity and

upwelling in the subpolar gyre creates a strong

halocline. This supports a temperature minimum

where the surface water becomes very

cold in winter. The temperature minimum is

called Dichothermal Water and is found in the

western subpolar gyre and the adjacent Okhotsk

and Bering Seas. Associated with the temperature

minimum is very high oxygen saturation

in the summertime, due to capping by warm

surface water and slight warming of the subsurface

T min layer. Below the Dichothermal Water,

temperature increases to a maximum and then

decreases to the ocean bottom. The temperature

maximum layer is called Mesothermal Water. The

maximum indicates a substantial advective

component from the east or the south since

otherwise it would acquire the low temperature

of the surface layer.

Tropical Pacific water properties were described

in Section 10.7.5. Complex vertical structure

is created by interleaving of North and

South Pacific waters (Figure 10.19). Nearly zonal

fronts in salinity occur along the equator

(Figure 10.30c). In temperature and density, the

equatorial thermocline/pycnocline ascends

and intensifies from 150e200 m in the west to

less than 50 m in the east (Figure 10.23). The pycnocline

inhibits vertical transfer of water properties.

In the west, a halocline lies within the

upper (warm pool) layer, above the thermocline,

so the pycnocline is determined by salinity

rather than temperature.

One tropical water mass that is distinguished

by a name is the Equatorial 13 C Water

354

10. PACIFIC OCEAN

(a)

Potential Temperature (°C) P10 149°E

0

29

200

10

25

20

Depth (m)

400

600

7

15

800

1000

4.6

0 500 1000 1500 2000 2500 3000 3500 4000 4500

Distance (km)

5

4.0

3

(b)

Depth (m)

0

200

400

600

800

1000

Potential Temperature (°C)

P14C 176°E

35°S 30° 25° 20°S

14

12

11

10

9

8

7

13

6

15

20

19

17

16

0 500 1000 1500

Distance (km)

5

24

4.2

(c)

120˚

40˚

30˚

20˚

10˚

0˚

−10˚

−20˚

−30˚

−40˚

120˚

130˚

130˚

140˚

20

140˚

150˚

150˚

160˚

P10

40

160˚

170˚

170˚

180˚

P14C

FIGURE 10.31 (a) Potential temperature ( C) along 149 E in the North Pacific. (b) Potential temperature along 170 Ein

the South Pacific. Source: From WOCE Pacific Ocean Atlas, Talley (2007). (c) Station locations superimposed on surface

streamfunction. (Data from Niiler, Maximenko, & McWilliams, 2003.)

100

90

60

40

80

70

20

60

50

180˚

190˚

40˚

30˚

20˚

10˚

0˚

−10˚

−20˚

−30˚

−40˚

190˚


(Montgomery & Stroup, 1962; Tsuchiya, 1981).

This is a mode water d a conspicuous thickening

of the equatorial layer centered at 13 C,

at about 75e300 m depth (Figure 10.19). Water

at this temperature is advected eastward across

the Pacific from the low latitude western

boundary currents. The thickening is possibly

linked to the local dynamics of the equatorial

currents, as the water mass is associated with

the North and South Subsurface Countercurrents

(Figure 10.20b).

Finally, two large regions of remarkably low

oxygen (<1 mmol/kg) are found in the eastern

tropical Pacific, centered at 10 N and 7 S, and

most intense near the eastern boundary (Figures

10.32 and 4.20). The most extreme oxygen

minima here coincide with well-developed

subsurface maxima in nitrite (NO 2 ; Figure

10.32b). Nitrite normally occurs within or at the

base of the euphotic zone (widespread band in

the upper 200 m in the figure), as part of the usual

nitrification process. The strongly developed

subsurface nitrite maxima are a unique feature

of denitrification. Remarkably, chlorofluorocarbons

(CFCs) are non-zero in the oxygen minima

(WOCE Pacific Ocean Atlas, Talley, 2007), which

means that these waters are ventilated and that

oxygen is low because of high biological productivity

rather than extreme age.


The intermediate layer of the Pacific is occupied

by two low salinity water masses, the North

Pacific Intermediate Water (NPIW) and the

Antarctic Intermediate Water (AAIW) (e.g.,

Figures 4.12b, 14.13 and S10.27 on the textbook

Web site). The source waters of both are fresh,

cool surface waters at subpolar latitudes. In

the subtropics and equatorial Pacific, the overlying

water is the higher salinity Central Water,

which originates in the high salinity midlatitude

surface waters. Underlying the intermediate

waters is higher salinity Circumpolar

Deep Water, which obtains its higher salinity

from the North Atlantic. Thus the NPIW and

AAIW both appear as vertical salinity minima

in the subtropics and tropics.

The NPIW salinity minimum is confined to

the subtropical North Pacific. The AAIW salinity

minimum, in contrast, is found throughout the

subtropical South Pacific, the tropical Pacific,

and similar regions of the Atlantic and Indian

Oceans. Both NPIW and AAIW are within the

ventilated, higher oxygen part of the water

column. However, neither have particularly

high oxygen content in the Pacific, indicating

that residence time is longer than for the overlying

Central Waters.

Salinity and oxygen content on isopycnals

that represent NPIW and AAIW (Figure 10.33)

reflect the low salinity/high oxygen influx

from (1) the Okhotsk Sea for the NPIW and (2)

the southeast Pacific for the AAIW. These are

the source regions of these water masses.

(Salinity at neutral density 27.30 kg/m 3

provides a straightforward example of the

importance of diapycnal mixing. Throughout

the tropics, salinity is higher and the water is

warmer as this is an isopycnal. There is no

warm, salty surface outcrop for this isopycnal,

so the tropical properties must result from diapycnal

mixing.)

10.9.2.1. North Pacific Intermediate Water

NPIW is the densest water that is directly

ventilated on a regular basis in the North

Pacific. The full NPIW density range is s q ¼

26.7 kg/m 3 to 27.2 kg/m 3 (directly ventilated),

to 27.6 kg/m 3 (ventilated through vigorous

diapycnal mixing in the Kuril Island straits).

The subtropical NPIW salinity minimum has

potential density s q ¼ 26.7 to 26.8 kg/m 3 .On

an NPIW isopycnal, the lowest salinity (hence

coldest) and highest oxygen, indicating the

most recently ventilated water, occur in the

Okhotsk Sea and adjacent subpolar gyre

(Figure 10.33). The main direct ventilation

process for NPIW is brine rejection during sea

ice formation in a coastal (latent heat) polynya

356

10. PACIFIC OCEAN

FIGURE 10.32 Tropical oxygen minima and denitrification regions. Eastern Pacific vertical sections of (a) oxygen

(mmol/kg) and (b) nitrite (mmol/kg) at 88 W (WOCE P19). (c) Oxygen (mmol/kg) at 300 m depth. (d) P19 station locations.

Source: From WOCE Pacific Ocean Atlas, Talley (2007).


FIGURE 10.33 (a, c) Salinity and (b, d) oxygen (mmol/kg) at neutral densities 26.75 kg/m 3 and 27.3 kg/m 3 , characteristic

of NPIW and AAIW, respectively. In the Southern Ocean, white at 26.75 kg/m 3 shows the isopycnal outcrops; the gray curve

in (c) and (d) is the winter outcrop. Depth of the surfaces is shown in the WOCE Pacific Ocean Atlas. This figure can also be

seen in the color insert. Source: From WOCE Pacific Ocean Atlas, Talley (2007).

in the northwestern corner of the Okhotsk Sea

(“NWP” in Figure 10.34b). Polynyas all along

the shelf create brine rejection; the NWP is at

the end of the cyclonic circulation, so the water

has accumulated the most brine. Historical

data suggest that brine rejection can affect

densities up to about s q ¼ 27.1 kg/m 3 . See

also the online supplement Section S8.10.6 on

the Okhotsk Sea.

A sensible heat polynya maintained by tidal

mixing (Figure 3.12b) almost always occurs

over Kashevarov Bank (“KBP” in Figure 10.34b).

The subsurface temperature maximum is mixed

upward, melting the sea ice and fluxing

358

10. PACIFIC OCEAN

into the Oyashio do not have a subsurface

salinity minimum; instead, salinity is lowest at

the sea surface. The NPIW salinity minimum

forms as the renewed Oyashio waters encounter

the warmer, saltier, lighter surface waters of the

Kuroshio in the transition region between the

separated Oyashio and Kuroshio.

The NPIW formation rate based on meridional

overturn across 24 N is 2 Sv, which is small

compared with the other low salinity intermediate

waters. If measured locally, within the

subpolar gyre, where most of the newly ventilated

water remains, the recycling rate could

be higher.

Export of the low salinity NPIW southward

into the subtropics balances the net precipitation

in the subpolar region and net evaporation in

the subtropics. Part of the subpolar freshwater

input also exits northward through the Bering

Strait, where it eventually becomes part of the

North Atlantic Deep Water and is exported to

the low latitude North Atlantic (Talley, 2008).

FIGURE 10.34 Dense water formation in the Okhotsk

Sea. (a) Bottom potential temperature in September, 1999, and

mean velocity vectors at the two moorings. (b) Ice distribution

on February 1, 2000, from the SSM/I microwave imager.

“NWP” is the northwest polynya where the densest water is

formed. Figure 10.34a can also be found in the color insert.

Source: From Shcherbina, Talley, and Rudnick (2003, 2004).

nutrients to the surface layer; this is a highly

productive region biologically. The Okhotsk

Sea waters exit back to the northwest Pacific

through a deep strait in the Kuril Islands

(depth ~ 1500 m). Vigorous tides complete the

process of mixing the high oxygen down to the

maximum density at the sill, s q ~ 27.6 kg/m 3

(Talley, 1991). The renewed waters that exit

10.9.2.2. Antarctic Intermediate Water

AAIW is the low salinity intermediate layer

in all of the Southern Hemisphere oceans north

of the ACC (Figure 14.13; Section 13.4.2).

The Pacific AAIW salinity minimum is at

a depth of about 700e1000 m through most of

the South Pacific. Its potential density is between

s q ¼ 27.05 and 27.15 kg/m 3 in the southeast

Pacific, where it originates in the thick surface

layer (Subantarctic Mode Water) just north of

the Subantarctic Front. Its potential temperature

and salinity in this region are 4e6 C and 34.1e

34.5 psu. The salinity minimum is just the top

of the AAIW layer. We generally identify the

layer down to approximately s q ¼ 27.5 kg/m 3

as AAIW, based on properties that indicate an

identifiable water mass separate from Circumpolar

Deep Water (Section 13.5.2).

AAIW circulates anticyclonically around the

South Pacific’s subtropical gyre. Tongues of low

salinity, high oxygen water on the neutral

density surface 27.30 kg/m 3 (s q ¼ 27.15 kg/m 3 )


originate in the southeast Pacific and stretch

northwestward across the South Pacific

(Figure 10.32c, d). The AAIW salinity minimum

becomes slightly warmer, saltier, and denser

along its path. It enters the tropics in the western

Pacific, where its density becomes distinctly

higher due to higher salinity (mean values of

5.4 C, 34.52 psu, 27.25 kg/m 3 between 15 Sand

the equator).

The northern boundary of the AAIW is at the

Northern Hemisphere tropical-subtropical transition

at about 15 N (Figure 14.13); that is,

AAIW does not enter the North Pacific subtropical

gyre as a salinity minimum. AAIW does

extend northward along the eastern boundary

to about 35 N, in the “shadow zone” outside

the subtropical gyre.

The formation rate of Pacific AAIW is

approximately 5e6 Sv based on airesea fluxes

(Cerovecki, Talley, & Mazloff, 2011). A slightly

smaller rate of 4 Sv was obtained by Schmitz

(1995a), with an additional 10 Sv of AAIW

formation for the Atlantic/Indian.

10.9.3. Deep Waters

Two deep waters, distinct from the bottom

waters, are identified in the Pacific: Pacific

Deep Water and Circumpolar Deep Water.

Historically, Sverdrup thought (essentially by

analogy with the Atlantic) that a slow southward

movement of deep water must occur in

the South Pacific. This is the case, but for

a different reason than in the Atlantic, which

has active deep water formation in the north.

PDW, also known as Common Water, originates

within the Pacific from upwelled bottom waters

and modified UCDW. UCDW originates in the

Southern Ocean as a mixture of PDWand Indian

Deep Water (IDW; both marked by low oxygen)

and deep waters that are formed locally in the

Southern Ocean. PDW and UCDW occupy

approximately the same density (and depth)

range in the Pacific, with UCDW flowing into

the Pacific and PDW flowing out. The net

transport is southward, hence dominated by

PDW (Figure 10.18).

UCDW is described in Section 13.5.3 so is

only referred to here where it interacts with

PDW.

PDW is one of the major deep waters of the

global ocean, with many similarities to IDW

(Chapter 11). PDW has no surface sources,

unlike North Atlantic Deep Water. PDW is

formed entirely internally from upwelling and

diffusion. Because PDW is formed internally

from waters that flow in from the Southern

Ocean, the waters in the PDW are the oldest of

the global ocean. PDW is marked by low

oxygen, high nutrients, no CFCs, and large

D 14 C age (Figures 10.35, 4.12, 4.22, 4.24). The

vertical extrema indicating greatest age are

centered at 2000e2500 dbar, with the most

extreme values in the mid- to high-latitude

North Pacific. These signals of age extend southward

down the length of the Pacific toward the

Southern Ocean. Because the PDW mixes with

the younger surrounding waters as it moves

south, its age appears to decrease toward the

south. These age tracers, especially the low

oxygen, mark the presence of PDW in the

Southern Ocean. Because it is very old, PDW is

well mixed in T-S properties. It includes the

highest peak by far in the global volumetric

T-S diagram (Figure 4.17), at 1.1e1.2 C,

34.68e34.69 psu (corresponding to s 4 ¼ 45.87

kg/m 3 ). (PDW encompasses a wider range of

T-S than this.) For this reason, Montgomery

(1958) named it the (Oceanic) Common Water.

In sections of Figure 4.12, these T-S properties

are found in the North Pacific north of 20 N

from about 3500 m to the bottom.

In the North Pacific, north of 40 N, the most

extreme PDW is found on the isopycnal in

Figure 10.35, as indicated by highest silica and

lowest salinity (and also the most negative

D 14 C in Figure 4.24b). This is the “new” PDW,

which is formed of very old waters. The low

salinity is acquired through downward diffusion

from above. The high silica in the northern

360

10. PACIFIC OCEAN

FIGURE 10.35 (a, c) Salinity and (b, d) silicate for PDW/UCDW (g N ¼ 28.01 kg/m 3 ; s 2 ~ 36.96 kg/m 3 ) and LCDW (g N ¼

28.10 kg/m 3 ; s 4 ~ 45.88 kg/m 3 ). Depths of the two surfaces are approximately 2600e2800 m and 3500e5200 m, respectively,

north of the ACC. Maps of D 14 C (/mille) and d 3 He (%) at g N ¼ 28.01 kg/m 3 and depth and potential temperature at g N ¼

28.10 are found in Figures S10.31 and S10.32 on the textbook Web site. Source: From WOCE Pacific Ocean Atlas, Talley (2007).

North Pacific, which is also a marker of PDW,

comes from both aging of the waters and dissolution

from the underlying silica-rich sediments

(Talley & Joyce, 1992).

PDW and UCDW are horizontally juxtaposed,

especially in the South Pacific. Salinity

and silicate on an isopycnal (Figure 10.35a, b)

show the higher salinity/lower silicate UCDW

entering in the southeast, and the contrasting

low salinity/high silicate PDW moving southward

in the west.

There is also southward flow of PDW along

the South American boundary, evidenced by

the higher silica in Figure 10.35b, but much


more obvious in the vertical section of oxygen at

32 S (Figure 10.15b). Salinity in Figure 10.35a

does not reflect this southward flow because of

the small but noticeable impact of geothermal

heating from the East Pacific Rise. The geothermally

affected waters are beautifully marked by

d 3 He plumes (Talley, 2007). These match the two

westward-extending plumes of higher salinity

in the tropics in Figure 10.35a. (On an isopycnal,

warmer water must be more saline.) The higher

salinity at the eastern boundary in the South

Pacific is consistent with East Pacific Rise heating,

which masks the salinity signature of southward

flow.

When it leaves the Pacific and enters the

Southern Ocean, PDW joins the IDW, which

has a similar density range and is also marked

by low oxygen and high nutrients. The layer is

then referred to as UCDW, which upwells to

the sea surface in the ACC. This upwelled

UCDW is the most likely source of the surface

waters that are transported northward out of

the Southern Ocean (Chapter 14).

10.9.4. Bottom Water (LCDW)

The densest water in the Pacific comes from

the Southern Ocean. Its source is a mixture of

the deep waters of all three oceans (Atlantic,

Indian, and Pacific) that is modified by production

of dense waters around the Antarctic continent

(Section 13.5.3). In the Pacific and Indian

Oceans, it is common to refer to this dense

bottom water mass as Lower Circumpolar

Deep Water (LCDW). The similar layer in the

Atlantic is usually called Antarctic Bottom

Water (AABW), which is the nomenclature we

use when we discuss this bottom layer globally

(Chapter 14).

LCDW is recognized in the Pacific by low

temperatures and higher salinity than the overlying

PDW (vertical section in Figure 4.12). Its

higher oxygen and lower nutrients reflect its

somewhat younger age than the very old PDW

(Figures 4.12 and 4.22).

At the southern end of the Pacific sections,

LCDW is marked by the vertical salinity maximum

within the ACC. The higher salinity is

a long-distance tracer of North Atlantic Deep

Water (Reid & Lynn, 1971). The salinity maximum,

which approximately follows an isopycnal,

extends northward into the deep Pacific;

eventually the maximum salinity is at the ocean

bottom. On the 165 W section, this grounding

occurs at about 5 S, but in the far eastern Pacific

at 88 W, it has already occurred by 45 S (section

P19 in the WOCE Pacific Ocean Atlas, Talley,

2007).

LCDW enters the Pacific in the DWBC in the

southwest, east of New Zealand (Section 10.6).

This inflow is apparent in northward extension

of high salinity and low silica in the southwestern

Pacific on a deep isopycnal surface

characterizing LCDW (Figure 10.35). Some of

this signal succeeds in passing through the

Samoan Passage at 10 S and crosses into the

Northern Hemisphere hugging the western

boundary. Silica in particular shows evidence

of northward flow all the way along the western

boundary to the northern North Pacific.

LCDW properties change to the north as the

layer erodes and upwells across isopycnals

into the PDW, with downward diapycnal diffusion

of heat and freshwater as the source of

buoyancy. The upwelling transports were

described in Section 10.6, with the budgets suggesting

most of the upwelling occurs in the

South Pacific and tropics. Evidence of diapycnal

diffusion is abundant in the property changes

along the LCDW pathway. Salinity on the characteristic

LCDW isopycnal decreases to the

north, and is lowest in the central North Pacific

near the Hawaiian Ridge and in the northwestern

North Pacific (with temperature, of

course, the mirror image). Similar patterns are

apparent on constant depth surfaces and in

bottom properties (Figure 14.14).

The bottom water is subject to low levels of

geothermal heating that increase its temperature

gently, by about 0.05 C from the tropics to the

362

10. PACIFIC OCEAN

northern North Pacific. This change is consistent

with geothermal heating, and affects a bottom

layer of about 1000 m thickness (Joyce, Warren,

& Talley, 1986). This buoyancy source could be

important for the deepest upward flux in the

northern North Pacific, where overturn does

not extend much higher above the bottom than

this (Section 5.6; Figure 10.18).

10.10. DECADAL CLIMATE

VARIABILITY AND CLIMATE

CHANGE

The Pacific Ocean represents a large fraction

of the global ocean’s surface and therefore

a large potential for coupled atmosphereeocean

feedbacks. The interannual ENSO (Section

10.8), which has maximum amplitude in the

tropics, is an excellent example of efficient

coupling. The decadal and longer timescale

climate modes are characterized by much larger

north-south spatial patterns, with extratropical

amplitudes that are similar to tropical amplitudes.

Outside the tropics, coupling of the

ocean and atmosphere is much weaker and so

feedbacks are much weaker and harder to

discern.

All of the text, figures, and tables relating to

climate variability other than ENSO are located

in Chapter S15 (Climate Variability and the

Oceans) on the textbook Web site. Chapter S15

covers the following modes of decadal climate

variability that most directly affect the Pacific:

Pacific Decadal Oscillation (PDO), North Pacific

Gyre Oscillation (NPGO), Pacific North American

teleconnection pattern (PNA), North Pacific Index

(NPI), Southern Annular Mode. It concludes with

a discussion of climate change (trends in

temperature, salinity, and oxygen).

C H A P T E R

11

Indian Ocean


OVERVIEW

The Indian Ocean is the smallest of the three

major oceans. It differs from the Atlantic and

Pacific Oceans in having no high northern latitudes,

extending to only 25 N. The southern

boundary of the circulation is the Antarctic

Circumpolar Current (ACC), within and north

of which the Indian is connected to the Atlantic

and Pacific Oceans. The Indian Ocean also has

an important low latitude connection to the

Pacific Ocean through the Indonesian archipelago.

In the north, the Indian Ocean has two

large embayments west and east of India: the

Arabian Sea and the Bay of Bengal. The deep

Indian Ocean is geographically much more

complex than the deep Atlantic and Pacific

due to its tectonic history (Figure 2.10). Many

deep ridges divide the deep circulation that is

connected with the Southern Ocean into

numerous, complicated pathways.

The Indian Ocean was explored later than

the Atlantic and Pacific, with the first truly

extensive observations during the International

Indian Ocean Expedition (1962 to 1965), whose

results are gathered in the Oceanographic

Atlas of the International Indian Ocean Expedition

(Wyrtki, 1971). During the 1980s and

1990s, international exploration of the Indian

Ocean circulation as part of the World Ocean

Circulation Experiment (WOCE), major

programs in the Arabian and Red Seas, Indonesian

Throughflow, Leeuwin Current, and Agulhas

Current/Retroflection, along with many

national programs, vastly increased the amount

of information about all aspects of the circulation

and water masses in all regions of the

Indian Ocean. While the Indian Ocean remains

relatively less explored than, say, the northern

North Atlantic, it is now fully integrated in the

global observing systems and there are a number

of ongoing regional programs.

The principal upper ocean flow regimes of the

Indian Ocean are the subtropical gyre of the south

Indian Ocean and the monsoonally forced circulation

of the tropics and Northern Hemisphere

(Figure 11.1). These are separated oceanographically

around 10e12 S by a nearly zonal current

(South Equatorial Current: SEC) carrying fresher

Pacific waters westward across the Indian Ocean.

The anticyclonic subtropical gyre is similar to

those of the other four ocean basins. Differences

are that its western boundary current (Agulhas

Current) overshoots the African coast, hence has

a different type of separation from the western

boundary, and that its eastern boundary current

(Leeuwin Current) flows the “wrong way,” toward

the south. In the tropics and northern Indian

Ocean, the circulations are strongly seasonal,

forced by the reversing Southwest and Northeast

Monsoons. In addition, the Arabian Sea and the

Bay of Bengal are thoroughly contrasting oceanographic

regimes, with the saline Arabian Sea and


363



364

11. INDIAN OCEAN

its marginal seas (Red Sea and Persian Gulf)

dominated by evaporation while the fresher

Bay of Bengal is dominated by runoff from all of

the major rivers of India, Bangladesh, and Burma

(Section S8.8 on the textbook Web site http://


supplementary material). The surface waters of

the tropical Indian Ocean are the warmest of

the global open ocean, often exceeding 29 C.

The intermediate and deep flow regimes of

the Indian Ocean include a connection to the

Southern Ocean that is similar to that of the South

Pacific Ocean, with differences largely due to

accidents of topography. The main difference

(a)

20°E 40°E 60°E 80°E 100°E 120°E 140°E

40°N 40°N

Africa

Red

Sea

Arabian

Peninsula

Bab el

Mandeb Gulf of

Aden

Persian

Gulf

Strait of

Hormuz Gulf of

Oman

20°N 20°N

Southwest

Arabian

Sea

Bay of

Monsoon

Bengal

(July-Aug.)

Asia

Andaman

Sea

Sulawesi

Sea

Southwest Monsoon C.

Southern

0° 0°

Gyre

Somali C.

EACC

(East Arabian C.)

Great

Whirl

WICC

LL

EICC


South Java

Current

Java Sea

Banda

Sea

Indonesian

Throughflow

Mozambique

Channel


20°S 20°S

Benguela C.

Agulhas C.

Mozambique C.

EMC

NEMC

Kerguelen

Eastern Gyral Current

Australia

EMC

Rings

South Indian

40°S

Current

Agulhas Agulhas

40°S

Rings Return C.

Antarctic Circumpolar

Current

60°S Weddell

60°S

Gyre

Leeuwin C.

(S. Indian subtropical gyre at 200 m)

Southern ACC Front

Leeuwin C.

Subantarctic Front

Polar Front

Flinders C.

S. Australian C.

Antarctica

80°S 80°S

10°E 40°E 80°E 120°E 150°E

-5000 -4000 -3000 -2000 -1000 0

FIGURE 11.1 Indian Ocean schematic surface circulation. Black: mean flows without seasonal reversals. Gray: monsoonally

reversing circulation (after Schott & McCreary, 2001): (a) Southwest Monsoon (July-August) (b) Northeast Monsoon (January-

February). The ACC fronts are taken directly from Orsi, Whitworth, and Nowlin (1995). The subtropical gyre in the Southern

Hemisphere just 200 m below the sea surface differs significantly from the surface circulation, as indicated by the dashed curve.

Acronyms: EACC, East African Coastal Current; EICC, East Indian Coastal Current; EMC, East Madagascar Current; LH and

LL, Lakshadweep high and low; NEC, North Equatorial Current; NEMC, Northeast Madagascar Current; and WICC, West

Indian Coastal Current. See also Figure S11.1 from the textbook Web site, which is a surface height map based on Niiler,

Maximenko, and McWilliams (2003), with labeled currents, and Figure S11.2, reproduced from Schott and McCreary (2001).

WIND AND BUOYANCY FORCING 365

(b)

20°E 40°E 60°E 80°E 100°E 120°E 140°E

20°N 20°N

Northeast

Monsoon

(Jan.-Feb.)

0° 0°

Mozambique

Current

EACC

(East Arabian C.)

Somali C.

EMC

WICC

LH

EICC

Northwest Monsoon C. (NEC)


South Java

Current


Mindanao C.

Indonesian

Throughflow

20°S 20°S

20°E 40°E 60°E 80°E 100°E 120°E 140°E

NGCUC

FIGURE 11.1

(Continued).

-5000 -4000 -3000 -2000 -1000 0

for the Indian Ocean is a limited source of intermediate

(deep) water in the Red Sea in the northwest

Indian Ocean. This water mass is similar to

the Mediterranean Overflow Water of the

Atlantic. Both are highly saline, hence “dying”

the intermediate and deep waters with high

salinity, but both have low transport and hence

limited impact on deep ventilation rates.

The role of the Indian Ocean in the global

overturning circulation is that of an upwelling

region, like the Pacific Ocean. Near-bottom

waters from the North Atlantic and Antarctic

enter from the south and participate in a complicated

upwelling pattern that likely includes

return of Indian Deep Water to the Southern

Ocean, as well as upwelling to near the surface.

Upper ocean waters from the Pacific Ocean that

participate in the global circulation also traverse

the Indian Ocean (Indonesian Throughflow or

ITF), enter the Agulhas Current, and finally

enter the Atlantic Ocean.

The principal currents of the Indian Ocean are

shown in Figure 11.1 and also in Figure S11.1 in

the supplement on the textbook Web site, where

they are also listed in Tables S11.1 and S11.2. The

wind forcing, including monsoons, is described

in Section 11.2, followed by the monsoonal and

tropical circulation in Section 11.3. The subtropical

circulation, ITF, and Red Sea/Persian Gulf

regimes are described in Sections 11.4e11.6 and

Section S8.10 on the textbook Web site. The intermediate

and deep circulations are presented in

Section 11.7. Water masses are described in

Section 11.8 and summarized in Table S11.3 on

the textbook Web site. A few aspects of climate

variability are included in Chapter S15 on the

textbook Web site.

11.2. WIND AND BUOYANCY

FORCING

The wind forcing of the Indian Ocean is one

of its most unique features. The mean wind

pattern (Figure 5.16a) of the south Indian Ocean

is like that of the Atlantic and Pacific, with westerly

winds at high latitude (Southern Ocean)

and trade winds at low latitudes. The northern

366

11. INDIAN OCEAN

Indian Ocean, however, is dominated by the

seasonally reversing monsoons (Figures 5.16b, c

and Figures S11.3 and S11.4 on the textbook

Web site), which change the ocean circulation

seasonally.

11.2.1. Mean Wind Forcing

The mean winds in the Southern Hemisphere

result in Ekman downwelling over the broad

latitude region from 50 Sto10 S (Figure 5.16d

and Figure S11.3a on the textbook Web site).

This produces Sverdrup forcing for a standard,

anticyclonic subtropical gyre (Figure 5.17 and

Figure S11.3b). The gyre forcing is different

from that for the South Pacific and South

Atlantic, because the southern cape of Africa, at

about 35 S, lies well within the major subtropical

gyre forcing. The subtropical gyre “runs out” of

western boundary before it “runs out” of wind

forcing for the gyre. Consequently, the western

boundary current, the Agulhas, overshoots the

tip of Africa, making it different from the

other four subtropical gyre western boundary

currents. The wind forcing then continues the

subtropical circulation far to the west to the coast

of South America, where there is a southward

western boundary current (Brazil Current). The

actual circulation is much more complex as the

Agulhas turns back to the east after it separates

from the African coast, shedding large eddies

at the retroflection that propagate westward

into the South Atlantic rather than continuing

westward as a smooth flow to the coast of South

America. In any case, the wind forcing ensures

that the subtropical gyres of the Indian and South

Atlantic are connected.

At the eastern side of the Indian Ocean’s

subtropical gyre region, there is some connection

with the South Pacific’s subtropical circulation.

East of Tasmania, the subtropical circulation is

more part of the South Pacific’s circulation,

although part of the East Australian Current

(EAC) leaks into the Indian Ocean circulation

(Section 10.4.1).

The mean winds in the tropical and northern

Indian Ocean produce a net upwelling region

between the equator and 15e20 S. This is associated

with a cyclonic gyre consisting of the

westward SEC on the south side, the eastward

South Equatorial Countercurrent (SECC) on

the north side, and a northward western

boundary current (East African Coastal Current;

EACC).

The Southwest Monsoon, producing net

downwelling and Sverdrup transport forcing

for a mean anticyclonic circulation, dominates

in the mean winds in the Arabian Sea. The

Northeast Monsoon regime, though, is quite

different (see next section).

11.2.2. Monsoonal Wind Forcing

The northern and tropical Indian Ocean is

subject to monsoonal wind forcing. The word

monsoon is derived from the Arabic word

“mausim,” which means seasons. There is

a nearly complete reversal of winds from

summer to winter, and the ocean circulation

responds accordingly.

Monsoons are the seasonal changes of the

large-scale winds (Figure 5.16b, c and online supplementary

Figure S11.4 from Schott, Dengler, &

Schoenefeldt, 2002, which also includes seasurface

temperature; SST). These arise in

response to the change in sign of the large-scale

temperature difference between the ocean and

land mass. In summertime, the land mass is

warm and in winter it is cold. The ocean surface

temperature varies a little with seasons, but not

nearly as much as the land. So during the

summer in the tropics, the large-scale winds

blow toward the warm continent, and in winter

they blow toward the ocean. A thorough explanation

is much more complex and well beyond

our oceanographic scope.

Monsoons are named for the prevailing wind

direction. In the northern Indian Ocean in

summer, the Southwest Monsoon blows from

southwest to northeast, from the western Indian

MONSOONAL AND TROPICAL OCEAN CIRCULATION 367

Ocean and Arabian Sea onto India. (The Southwest

Monsoon winds are a continuation across

the equator of the southeast trade winds, which

continue throughout the year.) The southwesterly

winds are concentrated in a narrow jet,

called the Somali (or Findlater) Jet, which is

apparent in the July winds described in

Figure S11.4 on the textbook Web site. This is

the “wet-season” in India and most of Southeast

Asia. In winter (November-March), the Northeast

Monsoon blows from northeast to southwest,

from the continental landmass to the

ocean. This is the dry season, with relatively

cool conditions.

The Southwest Monsoon winds are stronger

than the Northeast Monsoons, so the annual

mean wind pattern looks like a weak version

of the Southwest Monsoon.

The transitions between Southwest and

Northeast Monsoons are relatively quick, taking

place in 4e6 week periods in April-June and

October-November. During the transitions, the

equatorial winds are eastward across the full

width of the Indian Ocean.

11.2.3. Buoyancy Forcing

Airesea fluxes of heat and freshwater are

shown in the global maps in Chapter 5

(Figure 5.15). The Indian Ocean has no high

northern latitudes that could result in substantial

heat loss. Its northernmost reaches, in the Red

Sea and Persian Gulf, do experience net cooling

and evaporation, and form dense waters. The

Red Sea outflow is dense enough to penetrate

deep into the water column, but the volume

transport of the overturn is small and the saline

overflow water mainly results in a salty “dye”

for the deep northern Indian waters.

The tropical Indian Ocean is a region of net

heating, with largest heating along the coast of

Africa, in the Somali Current, associated with

upwelling and large permanent eddies. There

is net precipitation in the east due to rising air

above the Indian Ocean’s very warm pool of

surface water. These features are reversed in

the east-west direction compared with the

regions of highest heating and net precipitation

in the Pacific and Atlantic Oceans, because the

warmest region of the Indian Ocean is the eastern

tropics; there is no equatorial cold tongue. This

results from the tropical Indian Ocean’s strongly

seasonal winds as opposed to the prevailing easterly

trade winds of the Pacific and Atlantic.

In the subtropics, the Indian Ocean’s surface

forcing also differs from that of other subtropical

oceans because the eastern boundary

regime is dominated by the southward Leeuwin

Current rather than an equatorward eastern

boundary current. Both the Agulhas and Leeuwin

Current regions thus experience net heat

loss. The Agulhas region has the highest heat

loss of all regions of the Indian Ocean. The

high heat loss extends far to the east along

the Agulhas Return Current (Section 11.4.2).

The subtropics are also a region of net evaporation,

although the contribution to total buoyancy

flux is small.

11.3. MONSOONAL AND

TROPICAL OCEAN CIRCULATION

The ocean circulation in the tropics and

northern Indian Ocean is dominated by the

reversing monsoonal wind forcing. Thorough

overview and discussion of these circulations

is provided by Tomczak and Godfrey (1994)

and by Schott and McCreary (2001). Ocean

adjustment to strongly variable winds includes

generation of large-scale waves such as Rossby

and Kelvin waves (Section 7.7.3). Dynamical

understanding of the current reversals, production

of undercurrents and eddies, and so forth,

requires incorporation of these wave processes.

We do not describe these mechanisms here.

The monsoonally forced circulation is north of

the SEC front (north of 10e15 S). The SEC flows

westward in all seasons and splits at the coast of

Madagascar into the Northeast Madagascar

368

11. INDIAN OCEAN

Current (NEMC) and the East Madagascar

Current. The latter feeds the Agulhas. The

NEMC flows northwestward and reaches the

African coast where it splits again, into southward

flow through Mozambique Channel and

northward flow in the EACC. The northward

flow along Madagascar and Africa is expected

from the Indian-wide cyclonic forcing south of

the equator (Figure 5.17).

The behavior of the EACC as it reaches the

equator depends on the monsoon. During the

Southwest Monsoon and the buildup to it,

the EACC feeds the northward Somali Current,

which crosses the equator. This current is

notable for its high speeds, measured up to

360 cm/sec. Its transport is about 65 Sv, most

of it in the upper 200 m. The continuation of

northward flow along the Arabian Peninsula

during the Southwest Monsoon is not generally

given a name, but Tomczak and Godfrey (1994)

and Böhm et al. (1999) referred to it as the East

Arabian Current, which we have adopted for

Figure 11.1. (At the northeastern termination of

the Arabian Peninsula, there is a persistent eastward

jet during the Southwest Monsoon called

the Ras al Hadd Jet.) The SEC, the Somali

Current, and the Southwest Monsoon Current

comprise a strong seasonal wind-driven gyre

in the northern Indian Ocean.

During the Southwest Monsoon, the midocean

circulation from south of the equator to

the northern boundary is eastward. The eastward

flow between 7 S and Sri Lanka/southern

India is called the Southwest Monsoon Current.

Within both the Arabian Sea and the Bay of Bengal,

the circulation is eastward with a tendency

to be anticyclonic (Figures 11.1 and 11.2). Both

the West Indian and East Indian Coastal Currents

flow eastward.

The western boundary currents during the

Southwest Monsoon have remarkably large,

recurrent eddy structures (Figure 11.3). As the

Somali Current crosses the equator, part of it

turns out to the east at 4 N, into the Southern

Gyre. Another large eddy, the Great Whirl, forms

at 10 N. Formation of the Great Whirl precedes

formation of the Southern Gyre during the transition

to the Southwest Monsoon (Schott &

McCreary, 2001). There is another smaller recurrent

eddy, the Socotra Gyre (or Eddy), at about

12 N. Northeastward flow continues on along

the Arabian Peninsula, associated with a major

upwelling region off the Oman coast.

During the Northeast Monsoon (November

to March), the equatorial current (SECC)

continues to flow eastward, but a westward

flow, the Northwest Monsoon Current, appears

from the equator to 8 N, along the south side

of Sri Lanka and India. From 8 S to the equator,

the South Equatorial Countercurrent flows eastward.

The surface circulations in the Arabian

Sea and Bay of Bengal reverse. The Somali

Current flows southward. The West Indian

and East Indian Coastal Currents flow westward.

The overall circulation is weaker and

more disorganized than during the Southwest

Monsoon. The Southwest Monsoon is stronger

than the Northeast Monsoon (Figure 11.2), and

thus the ocean responds more consistently to

the Southwest Monsoon.

During the monsoon transitions in spring

and fall, when the equatorial winds are westerlies

rather than trades, the equatorial surface

circulation reverses. The normal SEC, which is

driven by trade winds, is a westward flow.

The westerly winds cause the surface currents

to flow eastward (Figure 11.4). These flows are

called the Wyrtki Jets. Surface speeds exceed

100 cm/sec. The Wyrtki Jets are much stronger

than the intervening westward flows driven by

the trade winds, so the annual mean surface

current is also eastward.

The Pacific and Atlantic equatorial circulations

have well-defined permanent subsurface Equatorial

Undercurrents (EUCs) that flow eastward.

Because the Indian Ocean equatorial winds

reverse and the equatorial trade winds are rather

weak, there is only a weak EUC and only during

part of the year. The EUC is found in the thermocline

east of 60 E during February-June.

MONSOONAL AND TROPICAL OCEAN CIRCULATION 369

40 E 50 E 60 E 70 E 80 E

30 N 30 N

MGSVA

JAS

20 N 20 N

0.2 m/s

ship drift

20°N 0.2 m/s

JUL

10 N 10 N

15°N

0 0

10°N

10 S 10 S

30 N 30 N

5°N

80°E 90°E 100°E

MGSVA

DJF

20 N 20 N

0.2 m/s

ship drift

20°N 0.2 m/s

NOV

10 N 10 N

15°N

0 0

10°N

10 S 10 S

40 E 50 E 60 E 70 E 80 E

5°N

80°E 90°E 100°E

FIGURE 11.2 Surface circulation. Left: Arabian Sea (surface drifters). Right: Bay of Bengal (ship drift). Top: Southwest

Monsoon. Bottom: Northeast Monsoon. Source: From Schott and McCreary (2001).

During the Southwest Monsoon, the winds

are northeastward along the coasts of Somalia

and the Arabian Peninsula (Figure 11.2). This

causes offshore Ekman transport and upwelling

at the coast. The upwelled water along the coasts

is cold (~20e24 C) compared with the tropical

surface waters (>27 C; Figure 11.5). In addition

to the direct coastal upwelling, there is broader

scale upwelling since the axis of the Somali Jet

lies offshore so Ekman transport increases

offshore. The resulting wind stress curl creates

upwelling from the coast to the center of the jet.

Farther offshore of the Somali Jet, the wind stress

curl creates downwelling. (These regions are

apparent in the mean wind-stress curl map of

Figure 5.16d, since the Southwest Monsoon

dominates the Arabian Sea winds.)

The upwelled water during the Southwest

Monsoon is rich in nutrients. Ocean productivity

changes dramatically in this area with

370

11. INDIAN OCEAN

show minimum temperature in September,

coinciding with maximum biomass, at the

height of the Southwest Monsoon upwelling

season. Maximum temperature and minimum

biomass occur in January-March during the

Northeast Monsoon.

11.4. SOUTH INDIAN OCEAN

SUBTROPICAL CIRCULATION

FIGURE 11.3 Somali Current regime during the Southwest

Monsoon (August/September, 1995). This figure can

also be seen in the color insert. Source: From Schott and

McCreary (2001).

the reversal of the monsoon. High productivity

along the southwest coast of Oman, in the

Persian Gulf, along the west coast of India,

and the east coast of Somalia is apparent during

the monsoon (global maps in Figure 4.29;

Arabian Sea maps in Figure S11.5 on the textbook

Web site).

Vertical sections across the upwelling system

in the Somali Current during the Southwest

Monsoon show the uplifted isotherms at the

coast, and high nutrients and low oxygen that

accompany them (Figure 11.6). Monthly time

series of surface temperature and biomass

The subtropical gyre of the Southern Hemisphere

Indian Ocean differs from the other

ocean basins’ subtropical gyres in its connections

to the South Atlantic and South Pacific

circulations and location relative to the continents.

Australia, Tasmania, and New Zealand

form an eastern boundary for an Indian Ocean

subtropical gyre. The African coast creates

only a partial western boundary, allowing an

Indian Ocean subtropical gyre only north of

the Cape of Good Hope. However, the wind

stress pattern for the Southern Hemisphere

dictates that the Indian Ocean’s gyre extends

westward all the way to the western boundary

of the South Atlantic. Thus the Indian and South

Atlantic gyres are inextricably linked, but with

major complications due to the Agulhas Current

along the coast of Africa, which is one of the

most powerful western boundary currents in

the world. The far southeastern boundary of

the Indian’s subtropical gyre is also not quite

complete and there is leakage from the South

Pacific’s subtropical circulation via a small

branch of the EAC. Meanwhile, the eastward

flow of the southernmost part of the Indian’s

subtropical gyre partially continues into the

South Pacific. Thus the Southern Ocean

connects all three Southern Hemisphere

subtropical gyres.

11.4.1. Subtropical Gyre

The wind-forced anticyclonic subtropical

gyre of the south Indian Ocean includes the

SOUTH INDIAN OCEAN SUBTROPICAL CIRCULATION 371

FIGURE 11.4 Mean zonal surface currents at the equator, based on ship drift data. Left: monthly means. Right: annual

mean. ÓAmerican Meteorological Society. Reprinted with permission. Source: From Han, McCreary, Anderson, and Mariano (1999).

FIGURE 11.5 SST in July 2003 (Southwest Monsoon), from the MODIS satellite. This color can also be seen in the color

insert. Source: From NASA Goddard Earth Sciences (2007a).

372

11. INDIAN OCEAN

FIGURE 11.6 Sections in

the Somali Current upwelling

regime at 12 N, 29 Auguste

1 September 1964. Source: From

Schott and McCreary (2001); after

Swallow and Bruce (1966).

westward flow of the SEC in the north, the eastward

flow of the South Indian Current in the

south, and the southward flow of the East Madagascar

Current (EMC) and Agulhas Current along

its western boundary. The northward flow in the

eastern part of the gyre is broad in longitude

and is sometimes called the West Australian

Current. The south Indian Ocean does not have

a narrow, northward eastern boundary current,

unlike the other world ocean subtropical gyres.

Instead, the Leeuwin Current is a narrow, southward

flow along the coast of Australia.


The northern boundary of the subtropical

gyre is well defined in the upper ocean by

a strong property front within the SEC at about

10e15 S. The SEC carries fresh water from the

Indonesian passages westward, resulting in

the property front.

The surface geostrophic flow in the subtropical

gyre does not seem gyre-like in any depictions

(e.g., Figures 11.1, 11.8a, and 14.1; also

Stramma & Lutjeharms, 1997). The broad northward

flow heads eastward and feeds into the

southward Leeuwin Current rather than

turning westward to complete the anticyclonic

flow. This eastward flow at the sea surface,

centered around 17 S, is called the Eastern Gyral

Current (Wijffels et al., 1996; Domingues et al.,

2007); it is analogous to the Subtropical Countercurrent

of the North Pacific. A portion of the

anticyclonic gyre does reappear south of Australia,

in the form of the westward Flinders

Current (Bye, 1972; Hufford, McCartney, &

Donohue, 1997; Middleton & Cirano, 2002); its

surface expression turns back to join the Leeuwin

Current, but just 200 m below the surface

(dashed in Figure 11.1a), the westward flow

south of Australia continues on to the northwest

across the whole expanse of the south Indian

Ocean.

By 200 m depth, the anticyclonic gyre is well

formed and conforms to the shape of the net

gyre transport and also the wind-forced

Sverdrup transport (Figure 5.17). The anticyclonic

gyre extends all the way eastward to Tasmania.

The center of the gyre at this depth is at

35e36 S, which also, coincidentally, corresponds

to the southern tip of Africa. The western

boundary current here is the Agulhas. In the

northern part of the gyre at this depth, the westward

flow of the SEC reaches the coast of Madagascar

and splits into southward and northward

flows. The southward flow is the EMC, which

flows along the coast of Madagascar and then

shoots west to the African coast where it forms

the Agulhas. The northward flow, the NEMC,

also continues west to the coast of Africa where

it splits. A southward portion flows through

Mozambique Channel and joins the Agulhas

Current. A northward portion joins the EACC

to become part of the tropical circulation.

The subtropical gyre shrinks poleward and

toward the western boundary with increasing

depth, typical of all subtropical gyres (Figures

11.7 and 11.8). A useful measure of the poleward

shift is the bifurcation point of the EMC. At

800e900 m, it shifts to about the center of Madagascar.

By 1000 m, it shifts to the southern end of

Madagascar. By 2000 dbar (Figure 11.14), the

anticyclonic circulation is entirely in the western

basin, retaining the Agulhas as a strong western

boundary current that reaches to the ocean

bottom.

11.4.2. Agulhas Current and

Retroflection

The Agulhas Current is one of the strongest

currents in the global ocean. It is narrow and swift,

with synoptic speeds that exceed 250 cm/sec

(Figure 11.8). It reaches to the ocean bottom and

is narrowest at about 33 S where it is most

strongly pinned to the western boundary. The

Agulhas location is well indicated by the subtropical

gyre Sverdrup transport (Figure 5.17). It forms

mainly from southward flow in the EMC and

from weaker southward advection through the

Mozambique Channel, mainly in the form of

large anticyclonic eddies.

The Agulhas Current transport is approximately

70 Sv. On the inshore side of the

Agulhas, there is a well-defined, narrow countercurrent/undercurrent

that reaches to the

ocean bottom. Its surface velocity is also large,

exceeding 50 cm/sec (Figure 11.8d). It has

a transport of about 15 Sv. The undercurrent is

one pathway for transport of North Atlantic

Deep Water (NADW) into the Indian Ocean

(Bryden & Beal, 2001).

The Agulhas follows the continental shelf to

where it ends at about 36 S and then separates

from the boundary. It overshoots into the South

374

11. INDIAN OCEAN

FIGURE 11.7 Adjusted steric height (10 m 2 /s 2 ) at (a) 0 dbar, (b) 200 dbar, and (c) 800 dbar. Source: From Reid (2003). The

closely related geostrophic streamfunction based on subsurface float observations from Davis (2005) is shown in the

supplement Figure S11.6 on the textbook Web site.


FIGURE 11.7

(Continued).

Atlantic and then retroflects back into the Indian

Ocean. It sheds large rings at the retroflection

that propagate westward into the South

Atlantic. The Agulhas Retroflection and Agulhas

rings are presented in Chapter 9 because

of their impact on South Atlantic circulation

and water properties. An infrared satellite

image showing the Agulhas Current, its eddies,

and retroflection is presented in Figure 9.13

along with a schematic depiction of the Agulhas

retroflection and eddy shedding.

Eddy kinetic energy (EKE) is high in the

Agulhas retroflection (Figure 11.8c), including

a long, narrow band to the east of the retroflection,

which follows the Subtropical Front (South

Indian Current), also called the Agulhas Return

Current in the figure. The peak of EKE variability

is west of 27 E, which is the location of

the Agulhas Plateau. This band of high EKE is

apparent in global maps of EKE (Figure 14.16).

Enhanced EKE is also found southwest of

Madagascar, in the location where the EMC

separates and flows westward to the African

coast.

The location of peak EKE variability varies

seasonally. The area of variability is further

west in the austral winter and further south

in summer. The location of the Agulhas front

can be detected from satellite SST data

(AVHRR). The Agulhas fronts in SST are farther

west in austral winter as well (Quartly &

Srokosz, 1993; Figure S11.7 on the textbook

Web site).

Quasi-permanent meander sites in the

Agulhas Return Current (Subtropical Front)

are apparent in the Agulhas front locations,

and are included in the Agulhas system schematic

(Figure 11.8a). There is a northward

meander at 26 E and a second at 32 E. The

first one goes around the north side of the

Agulhas Plateau. These permanent meander

sites resemble those of the Kuroshio (Chapter

10), with similar zonal spacing between

meanders.

376

11. INDIAN OCEAN

(a)

(c)

28°S

30°S

South Africa

32°S

34°S

offshore cyclone

36°S

100 cms −1

24°E 26°E 28°E

30°E 32°E 34°E 36°E

(b)

cm

(d)

0

500

1000

50

3

4

56

7

50

25

10

11

8

9

−200

−150

−100

−50

12

13

14

15

25.5

26.4

27

16

22

18

14

10

Depth (m)

1500

2000

2500

10

0

−10

−25

0

27.92

10 20 30 40 50

longitude (°E)

6

3000

3500

0

28.08

4000

4500

0 50 100 150 200

Distance (km)

−200 −100 0 100 200

Velocity (cms −1 )

FIGURE 11.8 (a) Schematic of the Agulhas Current system and local topography. Source: From Schmitz (1995b). (b) RMS

sea surface height variability for eight years from the Topex/Poseidon altimeter. Source: From Quartly and Srokosz (2003,

Satellite observations of the Agulhas Current system, Phil. Trans. Roy. Soc. A., 361, p. 52, Fig. 1b). (c) Average velocity from 0e75 m

depth and (d) velocity section at 36 S, with neutral density surfaces overlaid, from ADCP observations in FebruaryeMarch,

2003. Source: From Beal, Chereskin, Lenn, and Elipot (2006). Figures c and d are from the ÓAmerican Meteorological Society. Reprinted

with permission.


11.4.3. Leeuwin Current

The Leeuwin Current is the eastern boundary

current for the south Indian Ocean, and

is located off Western Australia. The current

was identified prior to 1969, but named only

in 1980 following the start of intensive observations

(Cresswell & Golding, 1980). It differs

from other subtropical eastern boundary

currents since it flows poleward rather than

equatorward. It is about 50e100 km wide and

2000 km long. It follows along the continental

shelf break within 100 km of the coast, from

about 22 S, off the Northwest Cape, to the

southwestern tip of Australia (Cape Leeuwin)

at 35 S. The current then turns eastward

toward the Great Australian Bight (Figure 11.9)

FIGURE 11.9 Leeuwin Current (LC) and Leeuwin

Undercurrent (LUC). Other acronyms: SEC, South Equatorial

Current; LCS, Leeuwin Current source region; and GI,

geostrophic inflow. Source: From Pearce (1991); from Schott

and McCreary (2001).

where it continues along the shelf break,

bringing warm, saline waters into the region.

It continues eastward as the South Australian

Current, veers southward past Bass Strait

between Australia and Tasmania, and then

flows southward along the western coast of

Tasmania, where the boundary current is

historically referred to as the Zeehan Current

(Ridgway & Condie, 2004).

The Leeuwin Current has a mean southward

speed of 25 cm/sec at the surface, peaking at

greater than 50 cm/sec. It opposes the mean

wind stress, which is northward (Figures 5.16a,

11.2). Its maximum poleward transport at 33 S

is 5 Sv in the upper 250 m (Smith, Huyer, Godfrey,

& Church, 1991; Feng, Meyers, Pearce, &

Wijffels, 2003). Below it there is an equatorward

undercurrent (Leeuwin Undercurrent), with

speeds of up to 40 cm/sec, a depth range of

300e800 m, and a transport of 1e2 Sv.

The Leeuwin Current has marked seasonal

variability, peaking in surface speed in April-

May and in transport in June-July, when the

northward alongshore wind stress that opposes

the current is weakest (Feng et al., 2003). The

Leeuwin Current has elevated levels of mesoscale

eddy variability compared with eastern

boundary current regions in the other oceans

(see the map of surface EKE in Figure 14.16).

Both anticyclonic (warm-core) and cyclonic

(cold-core) rings are shed by the Leeuwin

Current from preferential locations associated

with coastline shape (Morrow & Birol, 1998;

Fang & Morrow, 2003). The eddies move westward

a long distance into the Indian Ocean, preferentially

along the band of high eddy energy

associated with the Eastern Gyral Current.

In addition to its poleward flow, the Leeuwin

Current differs from other eastern boundary

currents in that upwelling does not occur on to

the shelf. The isotherms off western Australia

slope strongly downward from about 200 km

offshore to the continental slope (Figure 11.10), in

contrast to the situation in the subtropical eastern

boundary current regions off the western United

378

11. INDIAN OCEAN

FIGURE 11.10 Mean potential temperature ( C), salinity, potential density, and velocity (cm/sec) at 32 S in the Leeuwin

Current. Shading indicates salinity of 35.5e35.7 psu. Source: From Feng et al. (2003).

States, South Africa, and South America, where

the isotherms slope upward toward the shore

and upwelling of cool water occurs.

The Leeuwin Current (in the upper 150 m) in

the north is warm and relatively fresh (35.0

psu), and has low dissolved oxygen and high

phosphate content. Its northern source waters

are both the tropical Indian Ocean and the ITF.

As it flows southward, it retains a high temperature

core and therefore transports a significant

amount of heat to the south. As it moves southward,

the Leeuwin Current becomes saltier,

INDONESIAN THROUGHFLOW 379

reaching about 35.7 psu at 33 S(Figure 11.10), due

to inflow of waters from the subtropical gyre

(Smith et al., 1991; Domingues et al., 2007). Taken

together with its high eddy activity, the overall

picture is of inflow of south Indian waters that

feed the southward Leeuwin Current, which

generates eddies that transport the water back

to the west. Thus the Leeuwin Current does not

continuously transport ITF waters from the

North West Shelf of Australia all the way down

to Cape Leeuwin (Domingues et al., 2007).

The Leeuwin Undercurrent, as well as the

broad northward West Australia Current, transports

South Indian Central Water, Subantarctic

Mode Water (SAMW), and Antarctic Intermediate

Water (AAIW) northward. These flows

are part of the broader anticyclonic subtropical

gyre, driven by Sverdrup transport.

The poleward flow of the Leeuwin Current is

driven by a southward pressure gradient force

associated with the flow from the Pacific through

the Indonesian archipelago (Godfrey and

Weaver, 1991; Feng, Wijffels, Godfrey, & Meyers,

2005). The downward slope of about 0.3 m in the

sea surface along the western coast of Australia

from 20 to 32 S is readily apparent in

Figure 11.7a and Figure S11.1 in the supplement

on the textbook Web site. This overwhelms the

local eastern boundary forcing by equatorward

winds. Wind variability, however, dominates

the seasonality of the Leeuwin Current transport

(Smith et al., 1991). Interannual variability of the

Leeuwin Current is dominated by El Niño-

Southern Oscillation (ENSO) signals that propagate

southward along the coast of Australia from

the Indonesian archipelago (Feng et al., 2003).

11.5. INDONESIAN

THROUGHFLOW

The Indonesian Archipelago is the low latitude

connection between the Pacific and Indian

Oceans. Flow through the archipelago is referred

to as the ITF. The ITF is unidirectional, from the

Pacific to the Indian Ocean, since sea-surface

pressure (sea level) is higher on the Pacific

side. The Indonesian Archipelago has exceedingly

complicated geography (Figure 11.11).

More than 10 Sv of fresher, high nutrient Pacific

waters thread through this complex. The global

overturning circulation has transports of this

order. The ITF is one of the major upper ocean

elements of this global circulation, being part of

the movement of 10e15 Sv from the Pacific

Ocean, through the Indian Ocean, and back to

the Atlantic Ocean (Chapter 14).

Pacific water enters the Indonesian Archipelago

mainly through Makassar and Lifamatola

Straits. The sources of this Pacific water are discussed

briefly in Section 10.7.4. The Makassar

Strait is shallower (680 m at Dewakang Sill),

but carries most of the transport, at least 9 Sv,

which is of North Pacific origin (Gordon, Susanto,

& Ffield, 1999). The deeper Lifamatola

Strait (1940 m) is the pathway for South Pacific

water into the Indonesian Archipelago and for

the deeper part of the throughflow into the

Indian Ocean. Transport through this strait is at

least 2e3 Sv (Gordon, Giulivi, & Ilahude, 2003;

Talley & Sprintall, 2005). Some upper layer South

Pacific water also passes through the Halmahera

Sea. Within the Indonesian Archipelago, the

waters are mixed horizontally and vertically.

There is also some internal modification through

local heating and slight freshening.

The throughflow waters exit the Indonesian

Archipelago through three principal routes:

Lombok Strait, Ombai Strait (connecting to

Savu and Sumba Straits), and Leti Strait (connecting

to Timor Passage). The deepest sill for

the outflow is 1250 m northeast of Timor, at Leti

Strait. All of these outflow straits have been

instrumented at some time or another, and the

transports through each are 2e5Sv(Figure 11.11).

In addition to various current meter deployments

in the principal straits, all of the straits

were instrumented with a pair of shallow pressure

gauges across the straits in 1995e1999,

allowing simultaneous observation of flows

380

11. INDIAN OCEAN

FIGURE 11.11 Indonesian Archipelago and Throughflow with transports (Sv). Lower panel summarizes transport

above and below 680 m (Makassar Strait sill depth). This figure can also be seen in the color insert. Source: From Gordon

(2005).

through the exit straits. Variability is large, and

includes an ENSO signal (Hautala et al., 2001).

An international array of current meters and

pressure gauges is now in place to monitor

both the inflows and outflows (Figure 11.11).

After the ITF waters exit the archipelago,

they form into a narrow westward flow

centered at 12 S, within the SEC. The fresh

upper ocean waters are easy to see on any

meridional salinity section in the eastern

RED SEA AND PERSIAN GULF OUTFLOWS 381

Indian Ocean (Figure 4.13b). The deeper part of

the throughflow is also observable as a salinity

minimum at about the same depth and density

as the low salinity AAIW that reaches from the

south to nearly this latitude. The deeper

expression of the ITF is unambiguously of

Indonesian Archipelago origin based on its

higher nutrient levels, especially in silica

(Talley & Sprintall, 2005). The salinity

minimum is called the Indonesian Intermediate

Water (or Banda Sea Intermediate Water in

early treatments).

The SEC is a zonal current that carries the

throughflow waters westward across the Indian

Ocean. Mass balances within the Agulhas

Current indicate that the Indonesian waters

must join this current and then exit the Indian

Ocean. The baroclinic structure of the Agulhas

suggests that the excess transport from the

throughflow is in the upper ocean. The waters

that enter the Agulhas that match the transport

through the Indonesian Archipelago are greatly

modified within the Indian Ocean and are

unlikely to be the same water parcels.

Model studies suggest that dramatic global

changes in upper ocean circulation, temperature

and salinity, winds, and precipitation

would occur if the ITF were cut off for a period

(e.g., Schneider, 1998; Song, Vecchi, &

Rosati, 2007). Song et al. (2007) show that the

eastern tropical Pacific warms and the tropical

Indian Ocean cools, reducing the strength of

the trade winds and reinforcing the SST

changes, and also shifting the precipitation

(Figure S11.8 in the online supplementary

material). ENSO variability would change.

The Pacific Ocean becomes fresher and the

Indian Ocean saltier as the fresher Pacific water

would no longer be exported to the Indian

Ocean. The flow that would normally go

through the Archipelago would instead go

south along the coast of Australia, leading to

marked surface warming southeast of

Australia.

11.6. RED SEA AND PERSIAN GULF

OUTFLOWS

The Red Sea is one of the two global sources

of high salinity intermediate water; the other is

the Mediterranean Sea (Chapter 9 and Section

S8.10 on the textbook Web site). Despite its relatively

low latitude, the Red Sea achieves this

distinction because of huge evaporation leading

to high salinities even with its relatively high

temperatures. Circulation, formation, and properties

of the very saline waters within the Red

Sea are described in Section S8.10.7 on adjacent

seas in Chapter S8 (Figure S8.25).

The pure, newly formed Red Sea Water spills

out over the Bab el Mandeb and into the Gulf of

Aden (Figure 11.12). Intensive hydrographic and

current observations of the Red Sea outflow in

the Gulf of Aden document the progress of the

highly saline, dense overflow water (Bower,

Johns, Fratantoni, & Peters, 2005). The total

outflow transport is no greater than 0.4 Sv, but

it is extremely saline and dense: 39.7 psu and

s q ¼ 27.5e27.6 kg/m 3 . It also has elevated chlorofluorocarbon

(CFC) content as a result of its

renewal in the Red Sea (Mecking &Warner,

1999). The plume of dense, saline water mixes

vigorously as it plunges over the sill. It follows

two paths with different mixing characteristics,

which are visible in Figure 11.12. The saline

water mass, as a whole, turns to the right

because of Coriolis force and hugs the southern

boundary of the Gulf of Aden where it continues

to mix and be diluted. The equilibrated water

mass is referred to as either Red Sea Overflow

Water (RSOW) or Red Sea Water, with properties

of 38.8e39.2 psu, s q ¼ 27.0e27.48 kg/m 3 and

depth 400e800 m. As it settles into the intermediate

layer of the Arabian Sea, the RSOW affects

the layer s q ¼ 27.0e27.6 kg/m 3 at depths of

400e1400m. Its vertical salinity maximum core,

at about s q ¼ 27.3, is visible on all sections in

the western tropical and northern Indian Ocean

(Figures 11.13 and 11.19). It is greatly diluted as

382

11. INDIAN OCEAN

FIGURE 11.12 (a, b) Red Sea Overflow Water: salinity with potential density contours overlaid on sections in the Gulf of

Aden in FebruaryeMarch, 2001. North is on the left. Source: From Bower et al. (2005). ÓAmerican Meteorological Society.

Reprinted with permission. (c) Red Sea outflow in the Gulf of Aden: climatological salinity on the isopycnal s q ¼ 27.20 kg/m 3 .

Source: From Bower, Hunt,and Price (2000). This figure can also be seen in the color insert.

RED SEA AND PERSIAN GULF OUTFLOWS 383

Station

No.

0

200

900

905

910

915

920

925

930

935

940

35.4

36.6

CTD Salinity for I1 9°N

26.0

26.6

945

950

955

970

975

980

986

990

35.0

995

1000

1005

33

34.7 34.3

34.9

26.0

1010

1014

Depth (m)

400

600

27.2

35.2

27.0

35.0

800

27.4

1000

0

500

1000

55°E

35.4

60° 65° 70° 75° 80° 85° 90° 95°

36.0

34.3

27.0

35.0

35.2

27.4

1500

35.0

34.9

27.6

2000

34.8

Depth (m)

2500

3000

3500

37.00

34.75

45.90

34.73

4000

4500

34.72

45.90

5000

5500

I1

6000

910

920

930

940

950

970

980

995

1005

1014

6500

0 1000 2000 3000 4000 5000

Distance (km)

FIGURE 11.13 Salinity along 9 N with selected potential density s q , s 2, and s 4 contours and station track overlaid. After

the WOCE Indian Ocean Atlas, Talley 2011).

384

11. INDIAN OCEAN

it spreads southward, but can be detected in the

Mozambique Channel and on into the Agulhas

Current (Beal, Ffield, & Gordon, 2000).

In contrast to the Red Sea, the much shallower

Persian Gulf contributes its highly saline water to

the Arabian Sea at a lower density and hence shallower

in the water column. Circulation is into the

Gulf on the northern side of the Straits of Hormuz

and out on the southern side (Figure S8.25b in the

online supplementary materials). Evaporation is

in excess of 1.6 m/yr and there is a small annual

mean heat loss. Temperature is between 15 and

35 C and salinity is up to 42 psu. Dense water

formation (s q > 29.5 kg/m 3 )occursinlatewinter

in the southern Persian Gulf where winter

temperatures are low and salinities high

(<19 C, >41 psu; Swift & Bower, 2003; Johns

et al., 2003). Outflow at the Straits of Hormuz in

winter has been observed at up to 41 psu at

21 C (averaging 39.5 psu). The potential density

of 29 kg/m 3 is much denser than bottom water

in the Indian Ocean. However, the outflow transport

at about 0.15 Sv is small (Johns et al., 2003).

The water is so significantly diluted during the

outflow that it contributes only to the upper

200e350 m (s q ~26.4e26.8 kg/m 3 ) of the Arabian

Sea. In Figure 11.13 this is the downward bulge in

high salinity from the surface layer. (The high

salinity surface layer in the Arabian Sea is of local

evaporative origin.)

11.7. INTERMEDIATE AND DEEP

CIRCULATION

The intermediate depth circulation at about

1000 m is dominated by zonal flows in the tropics

and the anticyclonic gyre in the south Indian

Ocean (Figure 11.7). By 2000 dbar (Figure 11.14)

the anticyclonic gyre is restricted to the western

Indian Ocean. There is a remnant of the SEC

and SECC structure in the tropics, where flows

remain basically zonal. Circulation in the

Arabian Sea and Bay of Bengal is weak. The

tops of the mid-ocean ridges begin to intrude.

By 3500 dbar, the circulation is strongly guided

by the topography (see details in Figure 2.10).


carry Circumpolar Deep Water (CDW) northward

into the Indian Ocean, along the deep

western boundaries in each of the Indian

Ocean’s basins. The principal western pathway

is through the Crozet and Madagascar Basins,

northward along the Madagascar coast, through

Amirante Passage, and into the Somali and

Arabian Basins. The eastern deep pathway is

through the Southeast Indian Ridge at 120 E

into the South Australia Basin, and then through

gaps east and west of Broken Plateau into the

Central Indian and West Australia Basins. The

deepest flow entering the Central Indian Basin

comes from the West Australia Basin through

several fractures in the Ninetyeast Ridge.

The northward deep flows can be recognized

by the deep and abyssal water masses that are

seen on various vertical sections (Section 11.8

below). NADWand CDW, which has high salinity

from the NADW mixed into it, can be seen against

each of the (five) deep western boundaries

formed by various ridges and undersea plateaus

on the 33 S crossing. Cold, dense, fresher Lower

Circumpolar Deep Water (LCDW), originating

as dense deep water in the Antarctic, is most

evident flowing northward against the Mozambique

Plateau and Southwest Indian Ridge.

The Indian Ocean’s net meridional overturn is

obtained from the total transport in isopycnal

layers from a complete east-west crossing of the

Indian Ocean (Figure 11.15). The direction of

transport at any given level is difficult to discern

from the circulation maps. The Indian Ocean has

no northern deepwater source. It has a small input

at intermediate depth from the Red Sea. Therefore

there should be net northward inflow in the deep

water and outflow in the upper ocean. The zonally

integrated transport in isopycnal layers at 33 Sis

shown in Figure 11.15, based on two independent

analyses of the same data set; another analysis,

from Talley (2008), is shown in Figure S11.10 on

the textbook Web site. (Both analyses shown

INTERMEDIATE AND DEEP CIRCULATION 385

FIGURE 11.14 Adjusted steric height (10 m 2 /s 2 ) at (a) 2000 dbar and (b) 3500 dbar. Source: From Reid (2003).

386

11. INDIAN OCEAN

(c)

(a)

0

1000

2000

3000

4000

5000

0

I5 (32S)

−16.3± 5.1× 10 9 kg/s

−20 −10 0 10

Overturning streamfunction (10 9 kg/s)

25

26.2

26.9

27.36

27.7

27.96

28.07

28.11

28.15

28.23

35° 40° 45° 50° 55° 60° 65° 70° 75° 80° 85° 90° 95° 100° 105° 110°E

26.0

(b)

–10

0

10

20

23

26.5

26.9

27.36

27.6

27.96

28.11

28.23

28.3

Neutral density (kg m –3 )

Overturning streamfunction (10 9 kg/s)

FIGURE 11.15 (a)

and (b) Net northward

(meridional) transport

(Sv) for the Indian Ocean

at 33 S, integrated from

the bottom to the top. See

also Figure S11.9 in the

online supplementary

materials. Source of (a):

From Ganachaud, Wunsch,

Marotzke, and Toole (2000).

Source of (b): From Robbins

and Toole (1997). The righthand

vertical coordinate

is neutral density. (c)

Neutral density (kg/m 3 )

at 33 S. Heavy contour is

isoneutral surface 27.95

kg/m 3 ,markingthedivision

between net northward

flow below and

southward above. After

WOCE Indian Ocean Atlas,

Talley (2011).

500

1000

27.3

1500

27.6

2000

27.95

27.9

28.00

2500

Depth (m)

3000

3500

28.10

28.06

4000

28.20

28.14

4500

28.26

28.20

5000

28.28

5500

6000

6500

0 1000 2000 3000 4000

Distance (km)

5000 6000 7000 8000

WATER MASSES 387

here are also integrated from bottom to top, so the

actual direction of flow at a given level is the

change from below to above the level.) The net

meridional overturn from the deep to the intermediate/upper

ocean north of 33 Sis11e12 Sv based

on the two analyses shown. The transition from

northward to southward transport occurs around

2100 dbar, at neutral density 27.96 kg/m 3 .(There

is also an additional 5 to 10 Sv of southward flow

in the upper layers due to the ITF waters moving

southward across 33 S.) The required upwelling

rate north of the vertical section is approximately

3to5 10 5 cm/sec. This upwelling is an important

part of the return of global deep waters to the

upper ocean, and is mirrored by similar

upwelling in the Pacific Ocean. This upwelling

requires a diapycnal diffusivity of 2 to 10 cm 2 /

sec, which is within the range expected for the

global ocean.

11.8. WATER MASSES

We describe water masses from top to bottom

in four general layers: upper ocean and thermocline/pycnocline,

intermediate layer, deep layer,

and bottom layer. These are illustrated in the

vertical sections at 33 S, 9 N, 60 E, and 95 Ein

Figures 11.16, 11.13, 11.19 and 4.13 and 4.22,

respectively. Principal water masses of the Indian

Ocean are given in Table S11.3 in the online textbook

supplement; these generally follow the

nomenclature for the Atlantic and Pacific Oceans.

A schematic potential temperature-salinity (T-S)

diagram with water masses labeled is shown in

Figure 11.17. Full water column T-S and potential

temperature-oxygen diagrams based on WOCE

data are also shown, in Figure 11.18.

11.8.1. Upper Ocean

Surface waters of the Indian Ocean have

the nearly zonal distribution of temperature

that is typical of all oceans (Figures 4.1 through

4.6). The tropical Indian Ocean represents a

westward extension of the Pacific’s warm pool.

Cooler SSTs are found in the western Indian

Ocean, likely due to northward advection in

the NEMC and EACC (Section 11.3). There is

a notable absence of a cold tongue in the eastern

tropics; this differs from the Atlantic and Pacific

and is due to the lack of persistent trade winds

at the equator in the Indian Ocean.

The western tropical Pacific and the northern

Indian Ocean have the warmest SSTs on the

globe, together constituting the tropical ocean’s

warmest pool, remaining between 26 and 30 C.

However, the tropical Pacific and Indian Ocean

surface heat budgets are entirely different. As

argued by Loschnigg and Webster (2000),

because the tropical Indian Ocean is nearly

cloud-free while the western tropical Pacific is

shielded by clouds, the tropical Indian experiences

large net heating into northern summer

compared with the Pacific (75e100 W m 2

compared with 10e20 W m 2 ). Pacific SST regulation

is likely a combination of local balances

including cloud feedbacks and atmospheric

circulation (Ramanathan & Collins, 1991;

Wallace, 1992). The Indian Ocean equilibrium,

on the other hand, must be maintained by

cross-equatorial ocean heat transports, accomplished

by very shallow meridional overturn

dominated by the summer monsoon; however,

it is unclear whether this overturning cell has

significant heat transport (Schott et al., 2002).

Surface salinity in the Southern Hemisphere

Indian Ocean includes the usual subtropical

salinity maximum due to net evaporation

(Figure 4.15). The maximum surface salinity is

not as high as in the South Pacific or South

Atlantic, and is centered somewhat farther

south. In the north, the Arabian Sea and Bay of

Bengal have opposite surface salinity characteristics.

The Arabian Sea has high surface salinity,

up to 36.5 psu, due to evaporation, while in the

Bay of Bengal the salinity decreases from about

34 psu at about 5 N to 31 psu or less in the north.

The low values in the Bay of Bengal are due to

the considerable river runoff. The band of low

388

11. INDIAN OCEAN

salinity at about 10 S in the SEC is due to both

net precipitation of the Intertropical Convergence

Zone and to the ITF, carrying Pacific

waters westward (Section 11.5).

The contrast between highly saline Arabian

Sea surface waters and much fresher Bay of Bengal

surface waters is apparent in the vertical section

of salinity at 9 N(Figure 11.13). Both are best

developed in the upper 150 m. Surface

temperatures are high in both regions, so the

surface density in the Bay of Bengal is much

lower than in the Arabian Sea because of the

fresh water. Some of this low salinity water is

carried past India, especially when the Northeast

Monsoon Current is flowing westward.

A hint of this low salinity is apparent

in Figure 11.13 west of the Indian landmass,

at 75 E.

(a)

0

26.0

15

500

10

26.8

1000

31.9

5

31.9

4.0

1500

3.0

2000

2.0

2500

Depth (m)

3000

3500

1.6

45.90

1.0

37.00

45.90

1.0

4000

4500

5000

0.2

0.2

0.6

5500

6000

6500

0 1000 2000 3000 4000

Distance (km)

5000 6000 7000 8000

FIGURE 11.16 Sections at 33 S in 1987. (a) Potential temperature ( C) and (b) salinity. See Figure S11.10 on the textbook

Web site for the corresponding oxygen section. Selected isopycnals used for maps in other figures are overlain (bold). Station

locations are on the inset maps. Source: From WOCE Indian Ocean Atlas, Talley (2011); see also Toole & Warren (1993).

WATER MASSES 389

(b)

0

35.6

26.0

35.4

500

1000

1500

2000

34.68

31.9

35.0

34.4

34.70

34.9

34.80

34.73

34.70 34.50

34.50

34.60 34.66

34.69

34.71

31.9

34.55

26.8

Depth (m)

2500

3000

3500

34.80

34.80

34.75

34.74

34.73

37.00

34.72

45.90

34.74

34.73

45.90

34.73

4000

34.75

34.70

4500

34.69

34.71

5000

34.69

5500

6000

6500

0 1000 2000 3000 4000

Distance (km)

FIGURE 11.16 (Continued).

5000 6000 7000 8000

High salinity water (34.9e 35.5 psu) is found

in the tropics below the surface layer and north

of the SEC front at 10 S. The deeper part of this

high salinity layer is referred to as Red Sea Overflow

Water because of its high salinity source,

while the shallower part attains its high salinity

from the Persian Gulf (Beal et al., 2000). The

high salinity layer is also found in the Bay of

Bengal beneath the thin, fresh surface layer.

Although the highest salinity waters are

confined north of the SEC, a deep, diluted

high salinity layer extends farther south beneath

the subtropical gyre in the south Indian Ocean

(Figures 11.19 and 4.13b). At 95 E, its core as

marked by the 34.73 psu contour is at 2000 m,

although it still is not found much farther south

than about 15 S. This deep layer is discussed

further in the following paragraph.

The upper ocean water mass that marks the

division between the tropics and the subtropics,

390

11. INDIAN OCEAN

FIGURE 11.17 Mean T-S

curves for the Indian Ocean.

(a) (b) (c)

N

N

N

N

S

S

S

S

S

S

S

669

714

69 9

68 4

75 9

744

729

8 04

774

7 89

E E E E

Latitude

20°N

9°N

2°S

1 3°S

2 4°S

3 5°S


25

20

15

10

5

0

22

23

24

25

26

27

AAIW

ITF

Central Water

LCDW

Equatorial Water

28

SAMW

STUW

RSOW

Arabian Sea Surface

PGW

34. 5 35. 0 35. 5 36. 0 36. 5

Salinity


25

20

15

10

5

0

0 40 80 120 160 200 240

Oxygen ( mol/kg)

FIGURE 11.18 (a) Station locations, (b) potential temperature ( C) d salinity and (c) potential temperature ( C) d

oxygen (mmol/kg) for the Indian Ocean along 60 E. This figure can also be seen in the color insert. After the WOCE Indian

Ocean Atlas, Talley (2011).

at the SEC front at ~10 S, is the Indonesian

Throughflow Water (Section 11.5). At intermediate

depths, the extension of this is the Indonesian

Intermediate Water (IIW). Both are salinity

minima in the north-south direction. The IIW

is also a vertical salinity minimum, distinct

from the AAIW of the subtropical gyre to its

south.

20 N

10 N

0

10 S

20 S

30 S

40 S

40 E 50 E 60 E 70 E

WATER MASSES 391

0

500

CTD Salinity for I7 60°E

30° S 25° 20° 15° 10° 5° 0° 5° 10° 15° N

36.0

26.0

35.4

1000

34.4

34.50

31.9

35.4

1500

34.9

35.0

34.70

2000

34.73

34.80

2500

34.74

37.00

34.75

Depth (m)

3000

3500

34.75

34.74

34.73

4000

34.71

45.90

34.72

4500

5000

5500

6000

i07 I07

744

734 739

729

.724

673

.719

.714

.709

.703

.698

.693

.688

.683

678

809

804

799

.794

.789

784

779

774

769

.764

.759

754

749

6500

0 1000 2000 3000 4000

Distance (km)

5000 6000 7000

FIGURE 11.19 Salinity along 60 E, with potential density contours used for salinity maps. Station track is overlaid. After

the WOCE Indian Ocean Atlas, Talley (2011).

In the south Indian subtropical gyre, which

extends from the SEC front at 10e12 S southward

to the Subantarctic Front (SAF) at about

45 S, the thermocline/pycnocline is typical of

all subtropical gyres. Subtropical gyres are

ventilated primarily by subduction from the

sea surface. The two water masses associated

with the subduction, and that have distinctive

T-S signatures, are the Central Water, which

is the main T-S relation of the pycnocline, and

the Subtropical Underwater (STUW), which is

the shallow salinity maximum layer at about

s q ¼ 26.0 kg/m 3 in the upper part of the Central

Water. These water masses are marked in the T-S

diagrams (Figures 11.17 and 11.18). The surface

water in the center of the gyre at the top of the

Central Water is saline, as previously described.

The base of the pycnocline (Central Water) can

be taken roughly to be the salinity minimum

of the AAIW, which lies a little deeper than

1000 m at 40 S and rises up to about 500 m at

the SEC front. The STUW is the shallow (<300 m

deep), subsurface high salinity layer that

extends toward the equator from the center of

392

11. INDIAN OCEAN

FIGURE 11.20 Salinity at

s q ¼ 26.0 kg/m 3 , at a depth of

150e200 m through most of the

Indian Ocean. Source: From Reid

(2003).

the gyre. It is found only north of 25 S, that is,

north of the highest surface salinity.

Salinity on the isopycnal s q ¼ 26.0 kg/m 3

(Figure 11.20) illustrates the high salinity

STUW, particularly its high salinity source in

the eastern Indian Ocean near Australia. The

isopycnal also illustrates the low salinity of the

ITF as it crosses the Indian Ocean around 10 S,

and the huge difference in salinity between the

Arabian Sea and Bay of Bengal.

There are two major mode waters in the

upper ocean in the south Indian subtropical

gyre. These have no signature in salinity, but

are easily identifiable in potential vorticity

(inverse isopycnal layer thickness), since they

result from thick surface mixed layers in

winter. A potential vorticity section at 33 S

(Figure 11.21a) illustrates these two water masses,

the Indian Ocean Subtropical Mode Water

(STMW) of the Agulhas and SAMW. The

STMW is the weak potential vorticity minimum

in the far west around 26.0 kg/m 3 (Toole &

Warren, 1993; Fine, 1993). SAMW is the major

potential vorticity minimum layer across the

whole section, at s q ¼ 26.5 to 26.8 kg/m 3 . The

Indian Ocean STMW forms as a thick layer

north of the Agulhas Return Current (Figures

11.1 and 11.8). Its potential temperature, salinity

and potential density are 17e18 C, 35.6 psu, and

s q ¼ 26.0 kg/m 3 . Like the STMW of the EAC,

the Indian Ocean STMW is weak compared

with the STMWs of the Gulf Stream and

Kuroshio, possibly because the airesea heat

loss in the Agulhas region is much lower than

in either of the Northern Hemisphere western

boundary currents. On the isopycnal map in

Figure 11.20, STMW has no signature d it is

not a water mass that is identified by a salinity

extremum.

Indian Ocean SAMW is a much stronger and

much more pervasive mode water than the

Indian Ocean’s STMW. SAMW forms all along

the SAF from the western to the eastern Indian

Ocean (McCartney, 1982). In the west (west of

60 E), it has a potential density of about s q ¼

26.5 kg/m 3 (14 C, 35.4 psu) and occupies the

WATER MASSES 393

recirculation region of the Agulhas Return

Current and SAF. SAMW is even stronger in

the southeast Indian Ocean. This Southeast

Indian Subantarctic Mode Water (SEISAMW) is

the strongest of all of the global SAMWs: mixed

layers in the southeast Indian sector of the

Southern Ocean are thicker than anywhere

else, reaching 700 m in winter (Figure 4.4). The

potential temperature, salinity, and potential

density of new SEISAMWs are 8e9 C, 34.55

psu, s q ¼ 26.8e 26.9 kg/m 3 (Hanawa & Talley,

2001).

All of the Indian Ocean SAMWs subduct

northward into the subtropical gyre. SEISAMW

is the densest water that is directly ventilated in

the subtropical gyre and forms the base of the

pycnocline and Central Water. It has no

extremum in salinity; the underlying AAIW,

originating from the Malvinas-Brazil Current

confluence in the southwest Atlantic, has lower

(a)

SEISAMW

0

STMW

500

26.0

240

26.8

240

Potential

Vorticity

1000

AAIW

1500

2000

210

31.9

31.9

180 160

160 155

150

175

180

140

120

2500

210

37.00

180

100

3000

220

195

80

3500

45.90

190

45.90

4000

210

60

4500

40

5000

220

210

5500

Oxygen ( mol/kg) (contours)

20

6000

Potential vorticity (x10 -14 cm -1 sec -1 ) (color)

0

6500

0 1000 2000 3000 4000

Distance (km)

5000 6000 7000 8000

FIGURE 11.21 (a) Potential vorticity [10 14 (cm s) 1 ] (shading), oxygen (light contours), selected isopycnals (dark

contours) at 33 S in the Indian Ocean (see supplemental Figures S11.9 and S11.10 on the textbook Web site for additional

oxygen and isopycnal contouring). The STMW and SEISAMW potential vorticity minima are labeled, as is the AAIW

(salinity minimum). (b) Potential vorticity [10 14 (cm s) 1 ] on the neutral density surface g n ¼ 26.88 kg/m 3 , equivalent to

s q ¼ 26.8 kg/m 3 , representative of SEISAMW. Source: From McCarthy and Talley (1999).

394

11. INDIAN OCEAN

FIGURE 11.21

(Continued).

salinity. Potential vorticity on an isopycnal

surface representative of SEISAMW shows the

formation and gyre subduction regions

(Figure 11.21b): low potential vorticity indicates

thick layers in the south and central Indian

Ocean, with an extension of these low values

northward into the subtropical gyre, ending at

the southern side of the SEC around 18 S.

As a thick, well-ventilated layer, the SEI-

SAMW carries high oxygen waters as well as

its thickness. In Figures 4.13d and 11.21a (also

supplementary Figure S11.10), it is visible as

the high oxygen layer centered at about 500 m,

coincident with low potential vorticity. This

high oxygen extends all the way northward to

the SEC at about 12 S. In the western Indian

Ocean, a slight oxygen maximum associated

with SEISAMW can be traced along the western

boundary all the way into the Arabian Sea.


The two low salinity intermediate waters in

the Indian Ocean are AAIW and IIW. IIW has

already been discussed in reference to the ITF

in Section 11.5 and is mentioned here just for

completeness. RSOW is a high salinity intermediate

water with its salinity maximum core in

the same density range as AAIW, and it was

partially described above in Section 11.6.

Salinity on an isopycnal at the AAIW and

RSOW cores is shown to illustrate the spread

WATER MASSES 395

FIGURE 11.22 Salinity at

s 1 ¼ 31.87 kg/m 3 (equivalent to

s q ¼ 27.3 to 27.4 kg/m 3 ), at

a depth of 900e1200 m in most

of the Indian Ocean. Source:

From Reid (2003).

of both water masses (Figure 11.22). IIW also

affects the same isopycnal.

AAIWis a global, Southern Hemisphere water

mass characterized by a salinity minimum in the

vertical at densities of s q ¼ 27.0e27.3 kg/m 3 and

at about 500e1000 m depth (Figure 14.13).

Within the Indian Ocean, AAIW can be readily

recognized as the low salinity layer (salinity

minimum) below the thermocline throughout

the subtropical Indian Ocean south of the SEC

at about 12 S. The greatly eroded salinity

minimum extends into the tropics along the

western boundary (in the EACC and Somali

Current) and is found along the equator and

into the western Arabian Sea. The main part of

the AAIW, in the subtropical gyre, is at about

1100 m just north of the SAF and shoals with

the subtropical gyre’s isopycnals to about 500

m at about 15 S. Its salinity minimum core south

of 25 S has a mean potential temperature,

salinity, and potential density of 4.7 C, 34.39

psu, s q ¼ 27.2 kg/m 3 (s 1 ¼ 31.8 kg/m 3 ).

AAIW in the Indian Ocean comes from the

southwestern Atlantic Ocean, where the cold,

fresh waters of the Malvinas (Falkland) Current

loop far to the north and encounter the subtropical

waters of the South Atlantic. The southeast

Pacific’s SAMW and AAIW carried in this

current are then submerged beneath the new

South Atlantic SAMWs, producing a different

type of AAIW that is denser and of higher

potential vorticity and lower oxygen than the

southeast Pacific AAIW. This Atlantic AAIW

fills the Atlantic and Indian Ocean subtropical

gyres. That AAIW in the Indian Ocean does

not originate there, which is apparent from

global salinity, oxygen, and potential vorticity

maps on the AAIW isopycnals; the lowest

salinity, highest oxygen, and lowest potential

vorticity are from the Malvinas Current region.

The southeastern Indian Ocean AAIW has

much higher potential vorticity than the Pacific

and Atlantic AAIWs, indicating that the Indian

Ocean AAIW is the most eroded and hence

396

11. INDIAN OCEAN

most distant from its surface source

(Figure 11.21a; Talley, 1996).

AAIW from the south freely circulates up to

about 20 S. North of this there is a large gradient

in depth and properties with the salinity

minimum shoaling and eroded to lower density

and higher salinity to the north, displaced by

IIW and RSOW. This shift is clear in the vertical

section at 60 E(Figure 11.19). This latitude is the

northern boundary of the subtropical gyre at

this depth and density, and is readily apparent

in the circulation maps at 800 and 900 m

(Figure 11.7). A potential vorticity map for

AAIW shows an especially striking subtropical-tropical

boundary with well-mixed potential

vorticity within the subtropical gyre and

nearly zonal contours north of the boundary

(McCarthy & Talley, 1999).

In terms of global meridional overturn, the

AAIW layer in the Indian Ocean paradoxically

has southward transport, although its low salinities

are advected northward around the

subtropical gyre. However, there is more volume

transport upwelling from the deep water into

the AAIW layer and moving south than there

is actual AAIW moving north (Figure 11.15 and

Figure S11.9 on the textbook Web site).

RSOW (or Red Sea Water, depending on the

author) is the salinity maximum core at s q ¼

27.2e27.4 kg/m 3 in the Arabian Sea and

western Indian Ocean (Figures 11.13, 11.19,

and map in 11.22). RSOW results from overflow

of 0.4 Sv of highly saline Red Sea Water that has

a density of s q ¼ 27.6 kg/m 3 as it flows over the

sill at Bab el Mandeb into the Gulf of Aden

(Section 11.6). High salinity fills the Arabian

Sea on the RSOW isopycnal, spreading eastward

to the eastern boundary at 5 N and southward

along the western boundary toward the Agulhas

(Beal et al., 2000).

The high salinity within the Arabian Sea

extends downward across isopycnals to much

greater depths than the RSOW salinity

maximum. CFCs are present in the RSOW depth

range, but are essentially absent below 1500 m

in the Arabian Sea, indicating that whatever

process diffuses high salinity downward is

slow (Mecking & Warner, 1999). This deeper

high salinity is described in the next section.

11.8.3. Deep and Bottom Waters

There are no surface sources of deep or

bottom water in the Indian Ocean, even though

the densities of new Red Sea Water and Persian

Gulf Water are high enough to match the bottom

density. Both overflows have small volumes and

mix and settle out at intermediate and shallow

depths, respectively. Based on mass budgets,

the deepest Indian Ocean waters upwell to the

deep, intermediate, and thermocline layers.

Therefore, water parcels in the deepest layers

come from the ocean surface in the Atlantic

and Southern Oceans. The water mass entering

the Indian Ocean from the south is the CDW.

In the western Indian Ocean, NADW also enters

directly from the South Atlantic without passing

through the ACC.

Despite the lack of Indian Ocean surface

ventilation for the deep and bottom waters, we

distinguish a deep water of Indian Ocean origin

(the Indian Deep Water; IDW). This is deep water

that is “formed” within the Indian Ocean by

diffusion and upwelling rather than by surface

ventilation. Its low oxygen and high nutrient

content reflect high age as it advects back to

the Southern Ocean. Here IDW joins the fresher

Pacific Deep Water, which is also marked by low

oxygen and high nutrients, and together they

upwell to the surface in the Southern Ocean as

Upper Circumpolar Deep Water. The circuit of

the bottom and deep waters through the Indian

(and Pacific) Ocean is thus an important part of

the global overturning circulation.

On any given isopycnal, or at any given

depth in the deep Indian Ocean, we might find

both CDW and IDW, so distinguishing between

them is a highly regional exercise. One way to

distinguish between the deep and bottom layers

is in terms of net meridional transport (Section

WATER MASSES 397

11.7; Figure 11.15 and Figure S11.9 on the textbook

Web site). Waters below about 2000 m

depth (s 2 ~ 37.0 kg/m 3 or neutral density

27.96 kg/m 3 ) have net northward transport,

and waters above have net southward transport.

We could, for instance, consider the southward

and northward layers to be the deep and bottom

layers, respectively. However, this masks important

modification in the “bottom” layer, and

much of what we define as “Indian Deep Water”

would then occur in the bottom layer.

In terms of water masses, we will consider

the deep waters to be the layer containing the

high salinity core of CDW/NADW and a high

salinity core of IDW, and the bottom waters to

be the colder, fresher bottom layer. This latter

is also CDW, and is referred to as such in

most water mass descriptions of the Indian

Ocean. Here, as for the Pacific and Southern

Ocean descriptions (Chapters 10 and 13), we

call these deepest waters Lower Circumpolar

Deep Water (also known as Antarctic Bottom

Water).

In the deep water layer, there are both

southern and northern source salinity maxima

(e.g., 2500e3000 m depth in the salinity section

at 60 E in Figure 11.19). These are (1) CDW,

with the high salinity of NADW, found south

of 25 S and (2) IDW, from the north, in which

the elevated salinity is created by downward

diffusion that accompanies the deep upwelling

in the northwest Indian Ocean (Arabian Sea).

These two saline deep waters affect a representative

isopycnal (s 2 ¼ 37.0 kg/m 3 in Figure 11.23).

The Arabian Sea’s high salinity is clearly separated

from the CDW/NADW. The southern

CDW/NADW salinity maximum has high

oxygen and low silica as well (Reid, 2003;

WOCE Indian Ocean Atlas in Talley, 2010),

and potential vorticity also transitions abruptly

at about 25 S (McCarthy & Talley, 1999). The

separation between the southern CDW and

northern IDW high salinity layers is even more

marked in the eastern Indian Ocean

(Figure 4.13b).

The bottom layer of the Indian Ocean has net

northward transport. The bottom waters are

greatly modified as they circulate northward

into the Indian Ocean as a result of diapycnal

mixing, acquisition of silica from the bottom

sediments, and aging that reduces oxygen and

increases nutrients. The amount of activity of

each of these depends strongly on the deep basin.

As a result, it is not useful to distinguish between

CDWand IDW by depth or density ranges unless

looking carefully at a specific region.

The principal bottom water mass is LCDW,

also called Antarctic Bottom Water in Southern

Ocean and global contexts (Section 13.5).

LCDW is formed as dense water around Antarctica,

although the variety that extends northward

into the Indian Ocean is not the densest

Antarctic water. The northward circulation

pathways of LCDW, including DWBCs, are

described in Section 11.7.

At 33 S, this deep, cold, fresh, dense, high

oxygen water mass (<1 C, <34.71 psu, s 4 >

45.96 kg/m 3 , >210 mmol/kg) is found in the

deep basins that connect to the Southern Ocean

(Agulhas region, and the Mozambique, Crozet,

and Perth Basins). The densest, coldest waters

are not present in the Madagascar and Central

Indian Basins since they are not open to the

south. Bottom waters that make it to the Arabian

Basin in the northwest and Bay of Bengal in the

northeast have densities of s 4 >45.88 kg/m 3

and 45.94 kg/m 3 , respectively, and their potential

temperatures are 1.4 C and 0.8 C, respectively.

The Central Indian Basin is connected to

the southern source waters via the West Australia

Basin through several gaps in the Ninetyeast

Ridge; therefore its bottom waters are

warmer and less dense (1.0 C, 45.92 kg/m 3 )

than in the West Australia Basin.

LCDW upwells into the IDW. Observations of

its transformation and overturning transport

calculations using WOCE data can be found in

several sources (Johnson et al., 1998; Warren &

Johnson, 2002). Of the 12 Sv or so that upwell

out of the bottom layer (Section 11.7), about 4

398

11. INDIAN OCEAN

FIGURE 11.23 Salinity maps. (a) At s 2 ¼ 37.0 kg/m 3 , at about 2600 m depth, representative of the deep waters. (b) At

s 4 ¼ 45.89 kg/m 3 , at about 3500 m depth, representative of the bottom waters. Source: From Reid (2003).

CLIMATE AND THE INDIAN OCEAN 399

Sv progress northward in the westernmost

Indian Ocean basin (Mascarene Basin) and less

than 2 Sv make it through Amirante Passage

into the Somali Basin. All of this upwells. By

implication, the remaining ~8 Sv proceeds into

the central and eastern Indian Ocean. Of this, 2

Sv crosses into the Central Indian Basin from

the West Australian Basin and upwells. In

contrast, southward transport in the western

Indian Ocean of deep waters, including IDW,

appears to account for almost all of the upwelled

water from all of the Indian Ocean.

11.9. CLIMATE AND THE INDIAN

OCEAN

Climate variability at interannual to

decadal timescales has been documented in

the Indian Ocean. Because of its importance

to agriculture, interannual and longer term

variability in the monsoon has been of special

interest. Although the airesea coupling

process that creates ENSO is centered in the

tropical Pacific, ENSO dominates interannual

climate variability in the Indian Ocean.

Beyond its response to ENSO, the tropical

Indian Ocean has internal interannual variability.

A tropical Indian Ocean dipole mode

has been described, whose simplest index is

the east-west difference in tropical SST. In

the Southern Hemisphere, the Indian Ocean

is affected by the decadal Southern Annular

Mode (Antarctic Oscillation).

The text, figures, and tables relating to

climate variability are included in Chapter

S15 (Climate Variability and the Oceans) on

the textbook Web site. It covers the following

modes of climate variability that most directly

affect the Indian Ocean: ENSO effects in the

Indian Ocean, the Indian Ocean dipole mode,

the Southern Annular Mode, and climate

change (trends in temperature, salinity and

circulation).

C H A P T E R

12

Arctic Ocean and Nordic Seas

12.1. INTRODUCTION

The Arctic Ocean is a mediterranean sea surrounded

by the North American, European, and

Asian continents (Figures 2.11 and 12.1). It is

connected to the Atlantic Ocean on both sides

of Greenland and to the Pacific Ocean through

the shallow Bering Strait. The Nordic Seas is the

region south of Svalbard and north of Iceland.

This region is central for transformation and

production of some of the densest waters in

the global ocean, creating the densest part of

the North Atlantic Deep Water (Chapter 9),

and is a high latitude connection of the fresher

North Pacific waters to the saltier North Atlantic

waters. The Arctic’s sea ice cover is a vital

component of global climate because of its

high albedo (high solar reflectivity; Section

5.4). The Arctic’s sea ice cover is sensitive to

climate change. Because of important climate

changes and initiation of difficult hydrographic

time series in this ice-covered region beginning

in the 1990s, there is a large and growing body

of information about circulation, water masses,

and ice cover in the Arctic. In addition to

numerous journal publications, we note the

volume edited by Hurdle (1986) in which the

useful term “Nordic Seas” was first introduced,

a recent compendium from the Arctic-Subarctic

Ocean Fluxes study (Dickson, Meincke, &

Rhines, 2008), and an upcoming volume from

the Arctic Climate System Study (ACSYS;

Lemke, Fichefet, & Dick, in preparation).

Rudels’ (2001) review is a good overview of

the materials presented in this chapter.

The Arctic Ocean is divided into the Canadian

Basin (depth about 3800 m), and the

Eurasian Basin (depth about 4200 m; Section

2.11 and Figure 2.11). These basins are separated

by the Lomonosov Ridge, which extends from

Greenland past the North Pole to Siberia. The

maximum sill depth is about 1870 m (Björk

et al., 2007). The Eurasian Basin is subdivided

into the Nansen and Amundsen Basins; the

Canadian Basin is subdivided into the Makarov

and Canada Basins. Broad continental shelves of

50 to 100 m depth characterize the Arctic margin

north of Eurasia and the Alaskan coast, occupying

about 53% of the area of the Arctic Ocean

(north of Fram Strait) but containing less than

2% of the total volume of water (Jakobsson,

2002).

The deepest connection of the Arctic Ocean

with the other oceans is to the Nordic Seas

through Fram Strait, which lies between Greenland

and Spitsbergen with a sill depth of 2600 m

(Section 12.2). The sill depth north and east of

Svalbard, separating it from Franz Josef Land

and Novaya Zemlya, is only about 200 m

(Coachman & Aagaard, 1974). The Bering Strait

connection to the Bering Sea and the Pacific

Ocean is narrow and has a sill depth of only

45 m, but the transport, especially its freshwater

content, from the Pacific into the Arctic is


401



402

12. ARCTIC OCEAN AND NORDIC SEAS

significant, on the order of 1 Sv. There are also

connections from the Arctic to the North

Atlantic through the Canadian Archipelago by

several channels, principally Nares Strait (sill

depth 250 m) and Lancaster Sound (sill depth

130 m), which lead to Baffin Bay and then to

the Atlantic.

The Nordic Seas, between Fram Strait and

the Greenland-Scotland ridge, include the

Norwegian, Greenland, and Iceland Seas. These

commonly used oceanographic names are

loosely linked to the formal topographic names

(e.g., Perry, 1986). The Greenland-Scotland ridge

is comprised of three main sections (Hansen &

Østerhus, 2000): Denmark Strait between

Greenland and Iceland (sill depth 620 m), the

Iceland-Faroe Ridge (sill depth 480 m), and the

Faroe-Shetland ridge (sill depth of 840 m in

Faroe Bank Channel). Within the Nordic Seas,

the Greenland Sea is separated from the Norwegian

Sea by Mohns Ridge, and from the Iceland

Sea by the Jan Mayen fracture zone. The Norwegian

and Iceland Seas are separated by Aegir

Ridge. Each of these seas has a somewhat

separate circulation and water mass structure,

which is discussed in Section 12.2. West of

Greenland, the Arctic and Atlantic connect

through Baffin Bay and Davis Strait and through

(or past) Hudson Bay. These regions are discussed

in Section 12.3. The remainder of this

chapter is devoted to the Arctic Ocean circulation

(Section 12.4), water mass structure (Section

12.5), sea ice (Section 12.7), and climate variability

(Section S15.4 in Chapter S15 on the

textbook Web site http://booksite.academic

press.com/DPO/; “S” denotes supplemental

material).

The surface circulation is shown schematically

in Figure 12.1, which is referred to

throughout this chapter. An overall schematic

of the surface circulation and water mass formation

in the Arctic and Nordic Seas is shown in

Figure 12.2, relevant to conditions in previous

decades when deep water was still actively

forming in the Greenland Sea. The schematic is

still useful even though Nordic Seas convection

is currently to intermediate depth only. Both

Figures 12.1 and 12.2 show inflows from the

Atlantic and Pacific and surface outflow back

to the Atlantic. The Arctic surface circulation is

divided into principally cyclonic circulation in

the Nordic Seas and Eurasian Basin, and principally

anticyclonic circulation in the Canadian

Basin (Beaufort Gyre). The Transpolar Drift

(TPD) is the major cross-polar circulation

between these two systems. Figure 12.2 also

shows the overturn by open ocean convection

in the Nordic Seas and by shelf brine rejection

in the Arctic, and denser outflow back into the

North Atlantic.

12.2. THE NORDIC SEAS

The Nordic Seas are comprised of the Greenland

and Norwegian Seas, the Iceland Basin

between Iceland and Jan Mayen, and the Boreas

Basin between Greenland and Svalbard. The

densest water renewal in the Northern Hemisphere

is in the Greenland Sea. The Greenland

Sea produces denser waters than the Arctic

because it is closer to the high salinity inflow

from the Atlantic Ocean; winter cooling of this

more saline water produces denser waters

than in the fresher Arctic. Dense Arctic waters

also flow into the Greenland Sea and are an

important part of the mixture of waters that ultimately

overflows the sills into the North

Atlantic (Aagaard, Swift, & Carmack, 1985).

The Nordic Seas waters that overflow the

Greenland-Scotland ridge to become the dense

core of North Atlantic Deep Water (NADW)

are not the deep waters of the Nordic Seas,

which lie below the sill depth. Therefore, the

issue of whether Nordic Seas deep water

renewal extends to the ocean bottom, which

has not occurred since the 1980s, is not as important

for NADW formation as the processes that

determine properties at sill depth. These also

include the properties of the deepest water,

THE NORDIC SEAS 403

180˚

150˚W

150˚E

Bering Strait

inflow

“Rim current”

Siberian Coastal Current

90˚W

120˚W

Cape Bathurst

polynya

“Rim current”

Alaskan Coastal Current

Beaufort

Gyre

Chukchi Sea

polynya

Transpolar Drift

Laptev Sea

polynya

Kara Sea

polynyas

120˚E

90˚E

60˚W

Labrador

Current

Baffin Current

30˚W

EGC

West Greenland Current

North Water

polynya

NIC

EGC

IC

North Atlantic

Current

Northeast Water

polynya

Nordbukta

EIC

JMC

IFF

WSC

deep

convection

NAC

Storfjorden

polynya

Norwegian Coastal Current

“Rim current”

“Rim current”

30˚E

60˚E

0˚

FIGURE 12.1 Schematic surface circulation of the Arctic and Nordic Seas, including some of the major polynyas (gray

shading) and the Greenland Sea and Iceland Sea deep convection sites (dark gray). Topography as in Figure 2.11, where

place names can be found. Heavy lines indicate the principal circulation components, generally with larger transports than

those depicted with finer lines. Acronyms: EGC, East Greenland Current; EIC, East Iceland Current; IC, Irminger Current;

IFF, Iceland-Faroe Front; JMC, Jan Mayen Current; NAC, Norwegian Atlantic Current; and NIC, North Irminger Current.

(After Rudels, 2001; Loeng et al., 2005; Rudels et al., 2010; Østerhus & Gammelsrød, 1999; and Straneo & Saucier, 2008. & Polynya

locations from IAPP (2010) and Martin (2001)).

404


(a)

90°

Russia

Chukchi Sea

Barents

Sea

Norway

180° 0°

WSC

EGC

(b)

(c)

depth (m)

Bering

Sea

0

2000

4000

Bering Strait

Canada

Brine Formation

-0.4

-0.8

-1.2

DW

Canadian Basin

Eurasian

Basin

Norwegian

Sea

Greenland

gyre

0.90 0.94

90°

σ 1 = 32.785

deep waters

Canadian

Basin

Baffin

Bay

Greenland

Mid-Gyre Convection

surface waters

σ 0 = 27.9

intermediate waters

Lomonosov Ridge

Arctic Ocean

σ 2 = 37.457

Eurasian

Basin

Fram Strait

PW

Greenland

Sea

ASW

Iceland

Sea

upper

AIW

PIW

Atlantic Ocean

Mid-Gyre Convection

Dennmark Strait

Norwegian

Sea

lower

AIW

AW

DW:

see insert

6

4

2

0

potential temperature (°C)

32 33 34 35

salinity

FIGURE 12.2 Overall schematic of (a) circulation, (b) water mass layers and transformation sites, and (c) water masses in

potential temperature-salinity. Deep convection in the Greenland Sea in (b) has been replaced by mid-depth convection since the

1980s. Acronyms in (a): EGC, East Greenland Current; WSC, West Spitsbergen Current. Acronyms in (c): AW, Atlantic Water;

AIW, Arctic Intermediate Water; ASW, Arctic Surface Water; DW, Deep Water; PIW, Polar Intermediate Water; PW, Polar Water.

Source: From Aagaard, Swift, & Carmack (1985); amended by Schlichtholz and Houssais (2002).

THE NORDIC SEAS 405

since they affect the overall stratification of the

Nordic Seas.

In the next subsections, circulation, water

masses, and deep-water formation are briefly

described.

12.2.1. Nordic Seas Circulation

The overall circulation of the Nordic Seas is

cyclonic (Figure 12.1 and Figure S12.1 on the

textbook Web site). Exchange with the North

Atlantic is in the upper ocean, above the ridges

that stretch between Greenland and Scotland.

Warm, saline waters from the North Atlantic

enter in the east, in the Norwegian Atlantic

Current, which is a continuation of part of the

North Atlantic Current (Chapter 9). The North

Atlantic Current enters the Norwegian Sea in

two branches: an eastern (near-coastal) branch

along the coast of Ireland that reaches and

passes over the Wyville-Thompson Ridge

between the Shetland and Faroe Islands

(“Faroe-Shetland Ridge” in Figure 2.11), and

a western (mid-ocean) branch that reaches the

east coast of Iceland, then turns eastward along

the Iceland-Faroe Ridge where it forms a strong

current/front, and finally joins the Norwegian

Atlantic Current.

The southward-flowing western boundary

current of the Nordic Seas is the East Greenland

Current (EGC). The EGC enters the Nordic

Seas from the Arctic through Fram Strait. (This

is the main export route for sea ice from the

Arctic Ocean.) At about 72 N, part of the EGC

continues southward along the coast of Greenland

and part splits off to the east into the Jan

Mayen Current. This bifurcation is likely due to

the bathymetry, which steers circulation

throughout the water column. The Jan Mayen

Current is important for dense water formation

in the Greenland Sea. The speeds in the Norwegian

Atlantic Current and EGC are up to 30 cm/

sec, but the average is more like 20 cm/sec.

The Norwegian Atlantic Current flows northward

along the coast of Norway to the Arctic

Ocean. It includes a separate coastal current,

called the Norwegian Atlantic Coastal Current.

As it rounds the northern side of Norway,

a branch of the Norwegian Atlantic Current

splits off to the east into the Barents Sea,

following the coast. The rest of the Norwegian

Atlantic Current continues toward Spitsbergen/Svalbard

and splits again, a portion flowing

northward through Fram Strait as the West

Spitsbergen Current and the remainder turning

southward and joining the EGC. Upper ocean

flow from the Arctic into the Nordic Seas occurs

through Fram Strait as the EGC. Within the

Nordic Seas, there are several gyral circulations,

each associated with topographic features that

split the boundary currents.

Subsurface waters exit the Nordic Seas southward

as overflows over each of the three sills

between Greenland and Scotland. The water

masses that dominate in the outflows depend

on the sill depths, with intermediate water exiting

at Denmark Strait and the Iceland-Faroe

Ridge, and the densest overflow waters (but still

at intermediate depth) through the deeper Faroe

Bank Channel. Deep water from the Arctic

Ocean also enters the Nordic Seas through

Fram Strait (2500 m depth).

12.2.2. Nordic Seas Water Masses

The water masses of the Nordic Seas are

complicated and changing in time (Section

S15.4 on the textbook Web site) because of

the local nature of intermediate to deep

convection responding to variations in local

air-sea fluxes, and because of the Nordic

Seas’ location between the northern North

Atlantic and Arctic Oceans, both with variable

surface waters. Here we describe the dominant

Nordic Seas water masses, following

Aagaard et al. (1985), Rudels (2001), and Jones

(2001). These include two surface waters,

three intermediate waters, and three deep

waters (listed in Table S12.1 in the textbook

Web site).

406


The two major surface waters are the warm,

saline Atlantic Water (AW) and the cold, fresh

Polar Surface Water. AW inflow enters in the

Norwegian Atlantic Current (Figure 12.1), at

7to9 C and about 35.2 psu. There is a strong pycnocline

at about 400 m, which separates the

upper layer from the underlying Norwegian

Sea Deep Water (described at the end of this

section). The AW cools and freshens as it moves

northward in the Norwegian Atlantic Current.

By the time it reaches Spitsbergen, the surface

layer is 1 to 3 C with a salinity of about 35.0

psu. Because the upper layer of the Norwegian

Atlantic Current is so warm, the eastern Norwegian

Sea is usually ice-free in winter

(Figure 12.20a). The warmth of this current is

critical for the relatively mild climate of

Scandinavia.

The Polar Surface Water is relatively fresh

(<34 psu) and close to freezing (< 1.5 C). Polar

Surface Water enters the Nordic Seas from the

Arctic through Fram Strait in the EGC. In

Figure 12.3, this is the very cold, fresh surface

layer in the top 200 m on the west side of the

section. By the time this water reaches the

middle of the Greenland Sea (Figure 12.4 at

73.5 N), the layer is thinner (top 100 m) and

warmer. The presence of very cold, relatively

fresh surface water, with a strong halocline, is

typical of ice-covered regions, which include

both the upstream Arctic and also the EGC

region locally. The presence of Polar Surface

Water results in much colder upper waters in

the Greenland Sea than in the Norwegian Sea

(Figure 12.3). Within the Greenland gyre,

offshore of the EGC, upper ocean temperature

and salinity are less stratified, resulting from

local convection that mixes the water column

to intermediate depths.

We describe three intermediate waters in the

Nordic Seas. One is the shallow, subsurface,

warm, and saline layer (~150 m, >2 C, 35 psu)

in the EGC (on the western side of both panels

in Figure 12.4). This is a remnant of AW that

has been cooled, densified, and capped

FIGURE 12.3 (a) Potential temperature ( C) and (b)

salinity in the Fram Strait in 1980. See Figure 2.11 for location

of the strait. Source: From Mauritzen (1996).

(covered) at the top by the Polar Surface Water;

its sources are both modified AW from the

Arctic and recirculation with modification

within the Nordic Seas. This remnant is sometimes

called the recirculating AW.

A second intermediate water, Arctic Intermediate

Water (AIW), is the cold, fresher layer

( 1.2 C, 34.88 psu), centered at truly intermediate

depths (~800 m). Through much of the

Nordic Seas, AIW is a salinity minimum layer,

lying below the salinity maximum AW. AIW is

supplied from the Arctic Ocean through Fram

THE NORDIC SEAS 407

FIGURE 12.4 (a) Potential temperature ( C) and (b)

salinity across the southern Greenland Sea at 73.5 N in 1985.

Source: From Mauritzen (1996).

Strait and is modified by deep convection in

the Greenland Sea; production (transformation)

of AIW has continued to the present

although production of the densest Greenland

Sea Deep Water ceased in the early 1990s. In

the Greenland gyre where AIW is formed, it

is a salinity extremum only during non-winter

months when it is capped by a warmer surface

layer.

The third intermediate layer in the Nordic

Seas is called upper Polar Deep Water (uPDW).

uPDW enters the Nordic Seas through Fram

Strait from the Arctic. In the Nordic Seas,

uPDW is found in the EGC, more prominently

at Fram Strait than farther south. It is

characterized by cold temperatures (0 C

declining to 0.5 C) and salinities of 34.85 to

34.9 psu.

At least three distinct deep waters are found

in the Nordic Seas: Greenland Sea Deep Water,

Norwegian Sea Deep Water, and Arctic Ocean

Deep Water. Greenland Sea Deep Water is the

bottom layer colder than 1.2 C and fresher

than 34.896 psu in Figure 12.4. It is formed by

very intermittent deep convection within the

Greenland gyre. Convection also occurs in the

Boreas Basin in the northern Greenland Sea,

creating dense water similar to Greenland Sea

Deep Water. This densest layer has not been

formed in recent decades and is shrinking. In

the past several decades, convection has been

limited to 1700e2000 m depth; the water formed

there primarily encompasses AIW.

Arctic Ocean Deep Water is the saline deep

water in the Nordic Seas (S > 34.92 psu); its

high salinity comes from brine rejection in the

shelf seas of the Arctic (Section 12.5). It is

composed of deep waters from both the

Eurasian and Canadian Basins. It flows southward

through Fram Strait into the Nordic Seas

as a deep western boundary current. Its core of

high salinity hugs the Greenland coast between

1500 and 2000 m (Figure 12.3).

Norwegian Sea Deep Water is a mixture of

Arctic Ocean Deep Water and Greenland Sea

Deep Water. It does not have a separate convective

or brine rejection source. Norwegian Sea

Deep Water is also found in the eastern and

northern Greenland Sea, where it forms a barrier

to the passage of the colder Greenland Sea Deep

Water into the Arctic.

12.2.3. Vertical Convection in the

Nordic Seas and Dense Water Formation

Historically, the deep-water renewal with

highest density in the Northern Hemisphere

has been in the Greenland Sea (and its neighboring

Boreas Basin). This and the intermediate

waters of the Nordic Seas contribute to the

408


densest part of the NADW after they flow over

the Greenland-Scotland ridge complex and

plunge to the bottom layer of the northern

North Atlantic (Chapter 9). (Because of the sill

depth, the densest Nordic Seas waters do not

cross the ridge.) Dense water renewal is

apparent in the high oxygen content of the

deep waters of both the Norwegian and Greenland

Seas (260e325 mmol/kg or 6e7.5 ml/L),

reflecting a short residence time of about

40 years. Formation of deep waters in the Nordic

Seas occurs as open ocean convection, which

can be either simple mixed layer deepening

of the existing waters in winter, or deep penetrative

plume convection that pushes through

the existing, mid-depth stratification (Ronski

& Budéus, 2005a). Dense water formation

through brine rejection is not an important

factor in the Nordic Seas, unlike the Arctic,

likely due to the lack of extensive shallow

continental shelves that would allow the water

column to become brine-enriched.

From data collected in the first half of the

twentieth century, winter cooling resulted in

overturning from the surface to the ocean

bottom. Deepest convection occurs in the

Greenland Sea. However, top-to-bottom convection

became very rare after the mid-1980s,

so much so that the vertical stratification of

the Greenland Sea has changed from a onelayer

to a two-layer structure (Ronski &

Budéus, 2005b; see Figure S12.2 on the textbook

Web site).

Deep vertical convection cells or chimneys

(Section 7.10.1) renew the dense Nordic Seas

waters in the northern Greenland Sea. (In

convection regions, chimneys have scales on

the order of 50 km, while convective plumes

within the chimneys have scales on the order

of 1 km.) There are at least two other convection

regions as well, in the Boreas Basin, which is just

north of the Greenland Sea, closer to Fram Strait,

and in the Iceland Sea (Swift & Aagaard, 1981),

which both contribute to the important dense

intermediate waters. We concentrate here on

the Greenland Sea chimney as it is well defined

and observed.

The chimney-formation region of the Greenland

Sea is well defined east of the EGC, north

of the Jan Mayen Current, and west of Spitsbergen

(Clarke, Swift, Reid, & Koltermann, 1990).

The deep circulation here is cyclonic and topographically

steered, which then partially steers

the upper ocean circulation and chimney location.

There is often an ice tongue, called the

Odden, stretching around the southern part of

this cyclonic circulation, along the Jan Mayen

Current (Figures 12.1 and 12.5a). The open

water inshore of the ice tongue is called the

Nordbukta. This general region is referred to

as Odden-Nordbukta. The Odden is a region

of active ice formation. The Nordbukta is

a partial polynya (Section 3.9.6) kept open by

deep mixing that brings warmer subsurface

water to the surface and by offshore winds; it

has characteristics of both latent and sensible

heat polynyas, and is not always open (Comiso,

Wadhams, Pedersen, & Gersten, 2001). The relationship

between the presence of sea ice in the

Odden and deep convection is unclear;

although one might expect brine rejection to

contribute to buoyancy loss and convection,

the presence of sea ice might inhibit deep

convection, which is the deep water renewal

mechanism here.

Formation of deep vertical convection cells

or chimneys has been directly observed near

the Odden-Nordbukta (Figure 12.5; Morawitz

et al., 1996; Wadhams, Holfort, Hansen, &

Wilkinson, 2002; Wadhams et al., 2004). Using

acoustic tomography (Section S6.6.1 on the

textbook Web site) and moored measurements

in winter 1988e1989, the development of the

winter mixed layer and its temperature were

observed. Truly well mixed layers were not

seen, most likely because the horizontal resolution

of tomography is chimney-scale and not

plume-scale, but the deepening of the chimney

was clear. The column of near-freezing water

extended to almost 1500 m in late March (vernal

THE NORDIC SEAS 409

(b)

(c)

(d)

0

Ice Cover

( )

100

50

0

0

-0.8

-0.6

-0.8

-1.6

-0.6

-1.6

-0.8

-1.4

-1.2

-1.2 -1.4

-1.2

-0.8

-1.0

-1.2

Depth (km)

1

Depth (km)

1

-1.2

-1.2

-1.0

-1.2

-1.2

-1.2

-1.0

2

6 5 4 3 2

West Longitude

74 75 North Latitude

76

2

Oct

1988

Nov Dec Jan Feb Mar Apr

1989

May Jun

(C)

FIGURE 12.5 (a) The Odden ice tongue off the east coast of Greenland, February 12, 1993. Source: From Wadhams et al.

(1996). (b) Greenland Sea chimney region with 1988e1989 tomographic array location. (c) Mixed layer depth (with contours

on bottom plane). Source: From Morawitz et al. (1996). (d) Potential temperature ( C, contour intervals of 0.2 C) time series

at the array. Source: From Morawitz, Cornuelle, and Worcester (1996). Figures b, c, and d are Ó by the American Meteorological

Society. Reprinted with permission. See also Figure S12.3 in the online supplement.

410


equinox). More traditional wintertime, shipbased

observations in 2001 also showed

“deep” convection in the Greenland Sea, to

1800 m, which extended through the temperature

minimum layer (1000e1500 m) and into

the underlying temperature maximum layer

(Wadhams et al., 2002; Figure S12.3 on the textbook

Web site).

In both of these experiments, the convection

had the two-layer vertical structure of recent

decades, without penetration to the ocean

bottom, hence not renewing the cold, bottom

layer of (now older) Greenland Sea Deep Water

(e.g., Ronski & Budéus, 2005b; Figure S12.2 seen


What mechanisms other than deep convection

might ventilate the deepest waters in the

Greenland Sea? Other possibilities include

double diffusion (Section 7.4.3.2; Carmack &

Aagaard, 1973) and thermobaricity (Section

3.5.5) during deep plume convection (Clarke

et al., 1990; Ronski & Budéus, 2005a). Double

diffusion in the Greenland Sea is of the diffusive

variety, with cold, fresh water overlying

warmer, saltier water. The thermobaric effect

resulting from simply shifting the colder upper

ocean water parcels down by several hundred

meters into the warmer underlying water could

cause overturn sufficient to extend plumes to

the bottom because the equation of state is

nonlinear (cold water being more compressible

than warm).

Dense water production in the northern

Nordic Seas is also due to ice formation and

brine rejection, specifically in a recurrent,

wind-forced (latent heat) polynya in the

StorfjordenonthesouthernsideofSvalbard

(Haarpaintner, Gascard, & Haugan, 2001).

The polynya occurs between fast ice attached

to the coast and the offshore pack ice. Brine

rejection there enhances the shelf salinity by

more than 1 psu. The resulting plume of

dense water cascading off the shelf contributes

to the deep waters of the Norwegian

Sea, and also modifies the AW flowing

northward into the Arctic through the adjacent

Fram Strait.

12.3. BAFFIN BAY AND

HUDSON BAY

The Labrador Sea, lying west of Greenland

and within the geographic North Atlantic, is

an important source of intermediate depth

ventilation that feeds into the NADW. Since

the Labrador Sea is part of the subpolar North

Atlantic, its processes are considered in Chapter

9. However, the Labrador Sea has important

Arctic sources from the Canadian Archipelago,

through Hudson and Baffin Bays, which

connect to the Labrador Sea through Hudson

and Davis Straits, respectively. Most of the

North Pacific input to the Arctic through the

Bering Strait reaches the North Atlantic through

these bays. Surface flow is in only one direction,

from the Arctic to the Labrador Sea; however,

there is flow into Baffin Bay and Hudson Bay

from the Labrador Sea. The freshwater export

through Davis Strait, which includes considerable

sea ice, is an important factor in conditions

for deep convection in the Labrador Sea; with

greater freshwater flux, the Labrador Sea is

more likely to be “capped,” and not convecting

as efficiently.

Hudson Bay (Figure 12.6) is an extensive

shallow body of water, averaging only about

90 m in depth, with maximum depths of about

200 m. Hudson Bay is ice-covered in winter

and ice-free in summer. Hudson Bay contributes

a significant amount, 50%, of the freshwater

transport of the Labrador Current, based on

observations in Hudson Strait (Straneo &

Saucier, 2008). Hudson Bay has substantial river

freshwater input, from many (42) rivers, each of

moderate flow (Déry, Stieglitz, McKenna, &

Wood, 2005). There is considerable seasonal

river runoff from the south and east sides,

giving rise to a marked horizontal stratification

and an estuarine-type circulation. In summer,

BAFFIN BAY AND HUDSON BAY 411

FIGURE 12.6 Schematic

circulation in Hudson Bay and,

peripherally, Baffin Bay. Source:

From Straneo and Saucier (2008).

the upper water properties range from 1 to 9 C

and S ¼ 25 to 32 psu while the deeper water

properties range from 1.6 to 0 C and 32 to

33.4 psu. The low salinities are generally in the

south and east, near the main sources of runoff

and consistent with a general anticlockwise

circulation in the upper layer. A few observations

taken in winter through the ice indicate

upper salinities from 28 psu in the southeast to

33 psu in the north, with temperatures everywhere

at the freezing point appropriate to the

salinity. The implication is that the waters are

vertically mixed each year; the high dissolved

oxygen values of 200 to 350 mmol/kg in the deepest

water are consistent with this condition.

Baffin Bay, with a maximum depth of 2400 m,

is separated from the Labrador Sea (and hence

from the Atlantic) by the sill in the Davis Strait,

which is about 640 m deep (Rudels, 1986). Sill

depths between the Arctic and Baffin Bay are

120e150 m (Jones et al., 2003). Baffin Bay’s

temperature and salinity structure include

a cold, fresh surface layer to about 200 m,

a temperature and salinity maximum at about

700 m (>0.5 C, 34.5 psu), and cold, fresher

bottom waters (< 0.4 C, 34.25e34.5 psu;

Rudels, 1986). Winter convection within Baffin

Bay is likely limited to 200 m, and therefore

does not produce either the temperature

maximum or cold bottom waters. The temperature

maximum signature comes from the Labrador

Sea, via the West Greenland Current

(Chapter 9). However, much of the water in

the temperature maximum layer, and most of

the deep and bottom water, come from the

Arctic through Nares Strait (Bailey, 1957;

Rudels, 1986). As the annual inflow to Baffin

Bay is relatively small, the bay is a deep hole

compared with the inlet sills, and deep water

formation is minimal. Its deep water has

412


a long residence time, reflected in depleted

oxygen content and elevated nutrients, and

denitrification occurs in the deep waters (Jones

et al., 2003).

12.4. ARCTIC OCEAN:

CIRCULATION AND ICE DRIFT

The Arctic Ocean’s surface circulation is

dominantly cyclonic (counterclockwise) on the

Eurasian side and anticyclonic (clockwise) in

the Beaufort Gyre in the Canadian Basin

(Figure 12.1 and Section 12.4.2). A major current,

the TPD, flows directly across the Arctic

between these two circulations, from the Bering

Strait side to the Fram Strait. Inflows to the

Arctic are from the Nordic Seas, via the Norwegian

Atlantic Current that splits into the West

Spitsbergen Current (on the west side of Spitsbergen)

and flows into the Barents Sea, and from

the Pacific, via Bering Strait. There is some

flow from the Labrador Sea into Baffin and

Hudson Bays, but this does not continue

onward into the Arctic proper. The intermediate

and deep circulations (Figure 12.10 and Section

12.4.3) resemble each other and are cyclonic

throughout. They are strongly topographically

controlled.

Much of what is known about surface circulation

is based on ice drift, but there are some

differences between the two. Geostrophic calculations

and water mass tracking also provide

information on the surface flows. Ice drift is

important since the large amounts of ice that

exit the Arctic into the Nordic Seas affect the

salinity structure of the region and the albedo

(surface reflectivity) of the high northern latitudes,

which in turn affect Earth’s climate.

12.4.1. Ice Drift and Wind Forcing

The oldest records of Arctic ice movement

were based on ships held in the ice, such as the

Fram (Figure 12.7) and the Sedov, and from

movements of camps on the ice. Modern ice drift

is obtained from microwave satellite imagery

and from buoys deployed on the ice (International

Arctic Buoy Program; Figures 12.8 and

12.9). These various sources yield a consistent

picture of the surface-layer movement. Some of

the ice drift features oppose local upper ocean

circulation (Section 12.4.2). The mean ice drift

includes an anticyclonic (clockwise) circulation

in the Canadian Basin (Beaufort Gyre) leading

out to the TPD, with westward drift along the

Alaskan sector as part of the Beaufort Gyre. Ice

drifts southward from Baffin Bay through Davis

Strait into the Labrador Sea. Except in summer,

there is mean ice drift away from the Eurasian

coast toward the TPD. In the Eurasian Basin,

ice flows from the Laptev Sea into the TPD and

subsequently the Fram Strait (coincidentally the

track of the ship Fram, Figure 12.7). The TPD

feeds into strong southward flow (ice export)

through Fram Strait and the anticyclonic Beaufort

Gyre. In the Eurasian Basin, ice also flows

from the Kara Sea around the northern tip of

Novaya Zemlya into the Barents Sea and then

into the Norwegian Sea.

Ice drift speeds are of the order of 1 to 4 cm/

sec, equivalent to 300 to 1200 km/yr; in comparison,

the Arctic Ocean is about 4000 km across.

The speed and distance may be compared to

the 3 years taken by the Fram to drift from the

Laptev Sea to Spitsbergen, and the 2.5 years

for the Sedov to drift about 3000 km. The movement

is not steady, but has frequent variations of

speed and direction. There is a definite seasonal

change in the ice movements. The weakest ice

drift is in summer. Large variations in ice drift

are associated with the phase of the Arctic Oscillation

and the Atlantic Multidecadal Oscillation

(Chapter S15 on the textbook Web site).

The ice motion is related to both wind driving

(e.g., Ekman response) and advection by non-

Ekman surface currents, including the

geostrophic flow. Ice buoy vectors and the

mean sea level pressure (SLP) associated with

the wind forcing are shown in Figure 12.9, and

ARCTIC OCEAN: CIRCULATION AND ICE DRIFT 413

FIGURE 12.7 (a) Track of the Fram (1893e1896). (b) The ship was intentionally frozen into the ice in 1893 and drifted with

the pack until 1896. Ó www.frammuseum.no. Source: From Frammuseet (2003).

414


the atmospheric low is centered over the North

Pole. The Beaufort high is pushed much closer

to the Canadian/Siberian sides.

FIGURE 12.8 Annual mean Arctic sea ice motion from

1979e2003 from Special Sensor Microwave Imager (SSM/I)

passive microwave satellite data (extended from Emery,

Fowler, & Maslanik, 1997; data from NSIDC, 2008a). Monthly

means are shown in Figure S12.4 seen on the textbook Web

site.

also in the overlying contours in Figure 12.8.

SLP is dominated in the Siberian/Canadian

sector by the Beaufort high, which is an extension

of the Siberian High. This high-pressure

zone forces the anticyclonic Beaufort Gyre.

Mean geostrophic winds over the pole are

from the Eurasian to the Canadian/Greenland

side, roughly in the direction of the TPD. The

SLP ridge over Greenland in winter creates

strong northerly winds through Fram Strait

and southward along the coast of Greenland,

roughly paralleling the major ice export path

(see also supplementary Figure S12.5 from

Bitz, Fyfe, & Flato, 2002 on the textbook Web

site). The low pressure over the Nordic and

Barents Seas is a northward extension of the Iceland

Low, and forces cyclonic circulation in

these seas. In summer, the SLP contrasts are

much smaller, the winds much weaker, and

12.4.2. Upper Layer Circulation

The upper ocean circulation pattern

(Figure 12.1 and Table S12.2 on the textbook

Web site) is cyclonic in the Eurasian Basin and

around the rim of the Arctic above the shelves.

A large-scale anticyclonic circulation (Beaufort

Gyre) occurs in the Canadian Basin. Inflows

come from the Nordic Seas and from the Bering

Sea (Pacific). Outflows occur through Fram

Strait to the Nordic Seas in the EGC and through

the Canadian Archipelago to Baffin Bay and the

Labrador Sea.

The major currents (heavy curves in

Figure 12.1) are the:

1. Inflowing Norwegian Atlantic Current, which

splits into the northward-flowing West

Spitsbergen Current and eastward flow into

the shallow Barents Sea. The latter joins near

coastal inflow from the Norwegian Coastal

Current (Figure 12.1).

2. TPD that flows across the pole from the

Alaskan and eastern Asian coasts toward

Greenland and the Fram Strait, forming the

EGC.

3. Anticyclonic Beaufort Gyre, which is driven by

the mean high-pressure system above the

Beaufort Sea. It is a superficial feature. The

intermediate and deep circulations are

cyclonic (Figure 12.10).

Shown in thinner curves in Figure 12.1 are

weaker, but nevertheless critical, flows. These

include the Bering Strait inflow from the Pacific,

which has much smaller transport than from

the Nordic Seas. A cyclonic rim current connects

the shelf seas and feeds dense water formation

on the shelves (Rudels, Friedrich, & Quadfasel,

1999). Each portion of this current has a separate

name (see Rudels, 2001; Rudels et al., 2010). The

rim current is found from the Norwegian Sea to

ARCTIC OCEAN: CIRCULATION AND ICE DRIFT 415

(a)

Winter

(b)

Summer

H

1020.9

H

1012.9

L

1008.6

Scale: 2 cm/s =

Scale: 2 cm/s =

(c)

10.0

83-88

FIGURE 12.9 Mean sea level pressure (1979e1998) with mean ice buoy velocities for (a) winter (January-March) and (b)

summer (July-September). ÓAmerican Meteorological Society. Reprinted with permission. Source: From Rigor, Wallace, and Colony

(2002). (c) Mean wind vectors from ECMWF for 1983e1988. Source: From Zhang and Hunke (2001). Mean sea level pressure

maps from Bitz et al. (2002) are also shown in Figure S12.5 seen on the textbook Web site.

the Barents Sea, and around the Arctic to the

Kara and Laptev Seas. It branches off into the

interior Arctic at each of the major island groups

in each of these seas, joining the TPD toward

Greenland. Each of the island groups in the

Barents and Kara Seas includes cyclonic flow

between the island and the coastal rim current

and anticyclonic flow around the island group.

416


(a)

X

X

X

FIGURE 12.10 Circulation

schematics. (a) Subsurface Atlantic

and intermediate layers of the

Arctic Ocean and the Nordic Seas.

Convection sites in the Greenland

and Iceland Seas, and in the

Irminger and Labrador Seas are

also shown (light blue), as is

a collection point for brine-rejected

waters from the Barents Sea. Source:

From Rudels et al. (2010). This figure

can also be found in the color

insert. (b) Deep circulation; circled

crosses indicate entry sites from

dense shelf waters, and the Lomonosov

Ridge overflow site. Source:

From Rudels (2001).

X

X

(b)

The rim current continues into the Canadian

Basin, picks up the Bering Strait inflow, and

onward as the Alaskan Coastal Current, transporting

Bering Strait water eastward to the

Canadian archipelago (Jones, Anderson, &

Swift, 1998; Rudels, 2001).

Water enters the Canadian archipelago along

several different routes. The most important are

ARCTIC OCEAN WATER MASSES 417

the western routes feeding through Lancaster

Sound, a central route through Jones Sound, and

an eastern route through Nares Strait (Figure 12.1).

The circulation differs somewhat from ice

drift, especially in the Makarov Basin. The TPD

and anticyclonic Beaufort Gyre (Canadian Basin)

are evident in both circulation and ice drift.

However, the rim current is not apparent in ice

drift. Similarly, cyclonic flow in Baffin Bay and

the Labrador Sea is not apparent in ice drift,

which is dominated by southward export.

12.4.3. Intermediate and Deep

Circulation

Circulation in the intermediate layer, including

the subsurface, warm AW layer (Sections

12.3 and 12.5) and the intermediate layer of the

Arctic is shown in Figure 12.10a, representing

flows between 200 and 900 m depth. The largescale

circulation is cyclonic. Cyclonic cells are

embedded in this overall cyclonic circulation,

with separate cyclonic cells in each of the major

basins (Nansen, Amundsen, Makarov, and Canada).

This circulation has many similarities to

the surface flow and ice drift (Figures 12.1, 12.8,

and 12.9), but the anticyclonic Beaufort Gyre

has completely disappeared, replaced by

cyclonic flow throughout the Canadian Basin.

Major sources of water masses at this level are

also indicated in Figure 12.10, including brinerejected

waters from the Siberian shelves that

flow out into the deeper Arctic, and deep convection

sites in the Nordic Seas (Section 12.2), and in

the Irminger and Labrador Seas.

The deep circulation patterns (Figure 12.10b)

are nearly identical to the intermediate circulation;

that is, the Arctic circulation is nearly barotropic.

Because of topography, the deep flow

cannot connect across the Lomonosov Ridge,

so the continuous cyclonic rim current at middepths

is absent at depth.

Deep water enters and exits the Arctic from

the Nordic Seas through Fram Strait, with the

boundaries to the right of the flows (northward

flow on the east side and southward flow on the

west side). The overall flow in both the Eurasian

and Canadian Basins is cyclonic, with the

Lomonosov Ridge acting as a barrier. The sources

of deep water within the Arctic are the

brine-rejected waters from the continental

shelves; injection points from the shelves to

the deep ocean are denoted in Figure 12.10b

by crossed circles. Also indicated in this figure

is the saddle in the Lomonosov Ridge, where

an intensive experiment in 2005 showed incursions

of waters from Makarov Basin over to

the Eurasian side of the ridge (Björk et al.,

2007). (Figure 12.10b implies the opposite direction,

which was the generally accepted concept

prior to this experiment.)

12.5. ARCTIC OCEAN WATER

MASSES

The Arctic Ocean can be described in terms of

three main layers (Figures 12.11 and 12.12; Table

S12.3 on the textbook Web site): (1) Polar Surface

Water from the sea surface to about 200 m

depth, (2) intermediate waters, including AW,

from about 200 to 800 m (0 C isotherm), and

(3) various deep/bottom waters below this to

the bottom. Within the main water mass classifications,

the details can be complex. We mainly

follow Swift and Aagaard (1981), Aagaard

et al. (1985), Rudels (2001), and Loeng et al.

(2005); the latter two are reviews.

There are two external oceanic sources for the

Arctic Ocean waters: the Atlantic via the Nordic

Seas in the Norwegian Atlantic Current, and the

Pacific via Bering Strait. These inflow waters can

be identified far into the Arctic. 1 In addition,

there is significant freshwater input, mainly

1 Water mass properties in the Arctic persist a long distance from their sources, reflecting lower turbulence and hence lower

mixing than in other major ocean basins, due to the ice cover that isolates the ocean from direct wind forcing and waves.

418


from river runoff. Because of their low density,

the Bering Strait and river inputs enter the

near surface layer (Section 12.5.1), while the

AW enters an intermediate layer (Section 12.5.2).

Sea ice formation is the mechanism for Arctic

water mass transformation. Through brine rejection

over the broad continental shelves, dense

shelf waters are created. High production occurs

in recurrent latent heat polynyas in the Laptev,

Barents, and Kara Seas (Figure 12.1). As these

brine-rejected waters leave the shelves, they

mix mostly into the pycnocline, but they are

also the source of the deeper waters, depending

on their initial density and vigor of mixing.

(a)

0

1000

2000


Oden 2005 Oden 1991

7

−1.7

< −1.5

0

0.8

0.8

0.4

0.2

0

−0.5

−0.2

−0.4

0.6

−0.5

−0. 8

−0.5

2

1

2

9

8

2

0

R/V Knorr 2002

−0.9

0

−0.7

−0.9 −0.8

<

2

(b)

0

32

34

1000

34.92

2000

Salinity

Oden 2005 Oden 1991

34.8 34.85

34.93

34.9

34.91

34.93

34.94

35

34.91

34.92

R/V Knorr 2002

34.8

35 34.85

>

34.9

34.9

34.91

3000

4000

(c)

0

1000

2000

4000

ALASKA

8

28.08

3000 8

28

28.1

ALASKA

CANADIAN

BASIN

Potential density σθ

Oden 2005 Oden 1991

26

27

28.09

27.6

CANADIAN

BASIN

27.9

LOMONOSOV RIDGE

6

−0.7

28

LOMONOSOV RIDGE

28.09

−0.9

28.05

28.09

−0.9

EURASIAN BASIN

0 1000 2000 3000 4000 5000

75°N 80°N 85°N 90°N 85°N 80°N 75°N

28.08

28.1 1

1 28.1

EURASIAN BASIN

0 1000 2000 3000 4000 5000

75°N 80°N 85°N 90°N 85°N 80°N 75°N

SVALBARD

SVALBARD

−1

−1.1

GREENLAND

SEA

R/V Knorr 2002

28.05

28.08

GREENLAND

SEA

28

ICELAND

ICELAND

SEA

ICELAND

27

ICELAND

SEA27

2

3000 4

4

4000

(d)

0

1000

2000

3000

4000

ALASKA

34.9 34.93 CANADIAN

34.94 BASIN

0 1000 2000 3000 4000 5000

7

37

ALASKA

34.96

Potential density σ 2

Oden 2005 Oden 1991

36

37

37.44

34.955

37.2

CANADIAN

BASIN

37.1

34.95

37.4

LOMONOSOV RIDGE

37.3

LOMONOSOV RIDGE

34.94

37.42

37.43

37.44

37.45

37.46

37.47

37.35

34.93

EURASIAN BASIN

EURASIAN BASIN

37.45 37.46

37.45

37.3

37.35

37.4

37.42

37.43

37.44

37.45

0 1000 2000 3000 4000 5000

75°N 80°N 85°N 90°N 85°N 80°N 75°N

SVALBARD

34.91

GREENLAND

SEA

75°N 80°N 85°N 90°N 85°N 80°N 75°N

SVALBARD

R/V Knorr 2002

GREENLAND

SEA

34.9

34.

ICELAND

34

ICELAND

SEA

34

ICELAND

37

37

ICELAND

SEA 3

34

FIGURE 12.11 Arctic Ocean and Nordic Seas: (a) potential temperature ( C), (b) salinity, (c) potential density referenced

to the sea surface, (d) potential density referenced to 2000 dbar. (e) station locations. Oxygen and CFC-11 are shown in Figure

12.16. Data sets were collected between 2000 and 2005. After Aagaard et al. (1985).

BERING

STRAIT


(e)

180˚

150˚W

150˚E

120˚ 120˚

120˚ 120˚

CANADIAN

BASIN

Oden 2005

19 Aug - 25 Sep 2005

90˚

North Pole

LOMONOSOV

RIDGE

EURASIAN

BASIN

90˚

Oden 1991

17 Aug - 3 Oct 1991

FRAM

STRAIT

60˚ 60˚

Knorr

30 May - 1 July 2002

GREENLAND

SEA

BARENTS SEA

60˚

ICELAND

SEA

NORWEGIAN

SEA

30˚W

30˚E

FIGURE 12.11

(Continued).

0˚

Also because of brine rejection and the freshness

of sea ice, ice formation and melt over

the open Arctic freshen the surface layer,

contributing to a strong underlying halocline.

River outflows also contribute to this fresh

surface layer. This salinity structure can then

stably support vertical temperature inversions,

just as in the Southern Ocean (Chapter 13) and

the northern North Pacific (Chapter 10).

Laterally, there is an important demarcation

between the Eurasian and Canadian Basins. In

the upper ocean, this arises from the different

properties of the separate Atlantic and Pacific

inflows. In the deep water, the Lomonosov

Ridge blocks communication.

12.5.1. Surface and Near-Surface Waters

The surface layer, down to about 200 m, is

comprised of the Polar Mixed Layer (PML), a

shallow temperature maximum layer in some

regions (Canadian Basin), and the halocline. It

includes significant inputs from Bering Strait

(summer and winter Bering Strait Waters), from

river runoff, and from brine-rejected shelf waters.

Following Rudels (2001), this whole complex is

called the Polar Surface Water (Figure 12.12;Table

S12.3 seen on the textbook Web site).

The PML exists across the whole Arctic; it

extends from the surface to between 25 and

50 m depth. Its salinity is strongly influenced

420

(a)

Pressure

200

400

600

800

1000

2000

−2 −1 0 1

0

Freezing point

Canadian

Basin

Eurasian

Basin

Polar Surface

Water


(b)

30 31 32 33 34 35

Halocline

Atlantic

Water

Deep

Water

(c)

Salinity

29 30 31 32 33 34

Summer Pacific

halocline water

200

Canadian Basin

250

−1.5 −1.0 −0.5


S

θ

0

50

100

150

3000

(Bottom Water:

uniform

potential

temperature)

4000

−2 −1 0 1 30 31 32 33 34 35

Potential temperature (°C) Salinity

FIGURE 12.12 Arctic Ocean: (a) Potential temperature and (b) salinity profiles for the Canadian (dashed) and Eurasian

Basins (solid). Station locations are shown in Figure 12.17a: dashed profiles are stations CaB and MaB and the solid profile is

NaB. (c) Expanded potential temperature and salinity profile in the Canadian Basin (CaB in Figure 12.17a). After Steele et al.

(2004).

by the freezing or melting of ice and has a wide

range from 28 to 33.5 psu. The temperature is

also controlled by melting and freezing, which

involves considerable heat transfer at constant

temperature (the freezing point). As a consequence,

the temperature remains close to the

freezing point, from 1.5 C at a salinity of 28

psu to 1.8 C at a salinity of 33.5 psu. Seasonal

variations in water properties are largely limited

to this layer and range up to 2 psu in salinity and

0.2 C in temperature.

In the Eurasian Basin, temperature is nearly

constant (isothermal), near the freezing point,

through the shallow halocline (solid in

Figure 12.12, which includes warmer water at

the surface since this is a summer observation).

The halocline depth is 25e100 m. Because it is

nearly isothermal, the halocline cannot be

a simple vertical mixture of the PML and AW.

Rather, it includes shelf waters from the

Eurasian Shelf (Coachman & Aagaard, 1974;

Aagaard, Coachman, & Carmack, 1981). The


considerable Siberian river runoff flows into the

cold, low salinity surface layer. Ice formation

creates saline shelf waters at the freezing point.

These mix together and continue out into the

Arctic Ocean in the 25 to 100 m layer, creating

the isothermal halocline. Major canyons along

the shelf feed the saline AW onto the shelf; the

vertical mixing process is similar to an estuary

in which fresh river water flows over saline

seawater (Section 8.8).

Below 100 m in the Eurasian Basin, there is

a thermocline with temperature increasing

downward to the temperature maximum of

the intermediate Atlantic layer (AW) that enters

from the Nordic Seas.

The brine-rejected shelf waters in the

Eurasian sector are relatively saline compared

with other brine-rejected waters in the Arctic

because the saline, warm AW (Section 12.5.2)

is a source. These shelf waters can reach a sufficiently

high density to ventilate the deep water

in the Eurasian sector. Shelf waters from the

Barents and Kara Seas are especially implicated

(Aagaard et al., 1981).

In the Canadian Basin, the Polar Surface Water

below the mixed layer includes summer and

winter Bering Strait waters and Alaskan Coastal

Water (ACW), as well as brine-rejected shelf water

components (Figure 12.12c). These multiple sources

create more complicated vertical and horizontal

structures than in the Eurasian Basin. The

ACW and summer Bering Strait Water (sBSW)

are warm and create a temperature maximum at

50 to 100 m depth beneath the PML (labeled

“summer Pacific halocline water” in Figure

12.12c). The temperature maximum is supported

by a strong halocline. Below this, there isa temperature

minimum at about 150 m depth, due to

winter Bering Sea Water. Below this, the temperature

increases downward to the maximum in the

AW (see next section).

Circulation and temperature in the upper

temperature maximum layer (ACW and sBSW)

are shown in Figure 12.13. The warmest temperature

maxima are in the Beaufort Gyre, and are

FIGURE 12.13 (a) Schematic circulation of summer

Bering Strait Water (blue) and Alaskan Coastal Water (red)

during the positive phase of the Arctic Oscillation

(Chapter S15 on the textbook Web site). (b) Temperature

( C) of the shallow temperature maximum layer, which

lies between 50 and 100 m depth, in the Canadian Basin.

This figure can also be seen in the color insert. Source: From

Steele et al. (2004).

422


FIGURE 12.14 Salinity along a section in the Chukchi Sea (March 1982), including a high salinity bottom layer created by

brine rejection. Source: From Aagaard et al. (1985).

due to ACW. The cooler temperature maxima

are in the sBSW. ACW enters the Arctic from

the eastern coastal side of the Chuckchi Sea,

and Bering Strait Water enters from the center

and western side. ACW joins an eastward

coastal circulation and also forms eddies that

move into the central Beaufort Sea (loops in

Figure 12.13a). Bering Strait Water stays more

in the center of the Arctic and joins the TPD.

Brine rejection on the shelves in the Canadian

Basin produces waters that enter the halocline

(Polar Surface Water) in the Canadian Basin.

An example of late winter salinity distribution

with brine-rejected waters in the Chukchi Sea is

shown in Figure 12.14. Because the ambient

water is not saline, these new brine-rejected

waters are not salty (dense) enough to penetrate

through the Atlantic layer and do not contribute

to the Canadian Basin Deep Water (CBDW;

Section 12.5.3).

12.5.2. Atlantic Water

Below the cold Polar Surface Water, the Arctic

Ocean is characterized throughout by the AW

temperature maximum at a depth of 200 to 900

m (Figures 12.11, 12.12, 12.15, and 12.17). In

the Nordic Seas, the AW is a surface water

mass, with maximum temperature at the sea

surface. Where it enters the West Spitsbergen

Current in Fram Strait, it becomes a subsurface

maximum with cold, fresh Polar Surface Water

riding over the top. Some of the AW branches

back into the Nordic Seas in the EGC. The

remainder flows around the Arctic cyclonically,

mostly as a “rim” current along the continental

shelf break (Figure 12.1; Rudels et al., 1999).

This circulation is not in the same direction as

the surface circulation or ice drift.

Along its cyclonic path, both the temperature

and salinity of the AW decrease (Figures 12.12

and 12.15 and Figure S15.12 in Chapter S15

seen on the textbook Web site). At Fram Strait,

the AW temperature is around 3 C and its

salinity is greater than 35.0 psu. In the Arctic

Ocean, AW temperature decreases gradually to

0.4 C and its salinity to 34.80e34.9 psu. Its

core shifts downward from the surface (from

200 m in the Fram Strait to 500 m in the Canadian

Basin) and becomes more dense. These

changes are due to mixing with waters above

and below, and with cold shelf waters that

advect in from the side (Aagaard et al., 1981;

Rudels et al., 1999).

12.5.3. Deep and Bottom Water

Deep Water extends downward from the

lower 0 C isotherm, at about 800 m depth, to

the bottom (Figures 12.11 and 12.12; Table

S12.3 on the textbook Web site). Deep Water

comprises about 60% of the total water volume

of the Arctic Ocean (Aagaard et al., 1985). The

densest water in the Arctic is produced within


FIGURE 12.15 Atlantic Water in the Arctic. (a) Temperature maximum part of T-S diagram for core method analysis of

flow direction for Atlantic Water, (b) circulation inferred from successive erosion of core shown in (a) stations 1 to 6,

(c) depth, and (d) potential temperature ( C) of the Atlantic Water temperature maximum in the 1970s. (c, d) are ÓAmerican

Meteorological Society. Reprinted with permission; Polyakov et al. (2004) and Polyakov et al. (2010).

the Arctic. Because relatively shallow sills separate

the Arctic from the Atlantic and Pacific,

most of the deep water cannot flow out into

either the Atlantic or Pacific, nor can deep

waters from either of these regions enter.

Deep-water production that fills this isolated

deep layer must therefore be balanced by

upwelling. Consequently, Arctic deep waters

are relatively uniform in temperature and

salinity, including in the vertical.

424


We recognize three deep waters, following

Jones (2001). First is uPDW, which is found

throughout the Arctic and is exported into the

Nordic Seas through the Fram Strait. This layer

lies below the AW and above the Lomonosov

Ridge at about 1700 m, so there is open

communication with all regions of the Arctic.

uPDW is not obviously marked in potential

temperature and salinity: with increasing depth

through the UPDW, potential temperature

decreases and salinity increases. However,

uPDW can be readily distinguished in oxygen,

silicate, and chlorofluorocarbons (CFCs). The

Arctic Ocean is relatively well ventilated above

the ridge depth of the Lomonosov Ridge. On

the Canadian Basin side, the waters below

ridge depth have lower oxygen, higher silicate,

and low CFCs (Figure 12.16). Above this deep

layer and below the warm AW, the water

column is well ventilated, including an oxygen

maximum.

The other two major deep waters are separated

by the Lomonosov Ridge, between the

Canadian and Eurasian Basins. Deep water in

the Eurasian Basin is called Eurasian Basin Deep

Water (EBDW). Deep water on the Canadian

side of the Lomonosov Ridge is called Canadian

Basin Deep Water (CBDW) and has different properties

from EBDW (Figures 12.12 and 12.17).

Worthington (1953) deduced the existence of

the Lomonosov Ridge from the difference in

deep water properties in the Canadian and

Eurasian sectors. Deep waters from the Nordic

Seas also enter the Arctic through the Fram Strait,

so both the relatively cold, fresh, dense Greenland

Sea Deep Water and slightly warmer, saltier

Norwegian Sea Deep Water are found in the

Eurasian Basin (e.g., Aagaard et al., 1985).

The EBDW and CBDW can be split vertically

into deep and bottom waters. The bottom water

layer is recognized by uniform properties in the

vertical, hence it is nearly adiabatic (see the end

of this section).

In potential temperature, the progression is

from coldest deep waters in the Nordic Seas,

to slightly warmer (w 0.95 C) in the Eurasian

Basin. (In Figure 12.17, stations in the Makarov

and Amundsen Basins are included; these are

sub-basins of the Canadian and Eurasian Basins,

respectively, as can be seen from similarities

of the Makarov-Canadian and Amundsen-

Eurasian properties.) Potential temperature is

much more uniform in the deep water than in

situ temperature, with the deep temperature

minimum erased when the effect of adiabatic

compression is taken into account. The bottommost

layer, which can be more than 1000 m

thick, is adiabatic (see the end of this section).

On the other hand, there is a small but remarkable

potential temperature minimum in both the

EBDW and CBDW, associated with a characteristic

smooth upward curve in potential temperature-salinity

(T-S) space (Figures 12.17b and

12.18b). This minimum does not result from

choice of reference pressure. This “hook” in

the T-S relation, which is due to geothermal

heating, is even more apparent in Figure 12.18

(Timmermans, Garrett, & Carmack, 2003).

In salinity, the freshest bottom waters are

found in the Nordic Seas, with higher salinity

in the Eurasian Basin and highest in the Canadian

Basin. In any given region, the vertical variation

in the deep water is smaller than the

overall difference in salinity between these

regions.

Potential density variation between the

regions is dominated by potential temperature.

Thus the cold, fresher Nordic Seas Deep Waters

are denser than the EBDW, and the CBDW is the

least dense. To compare the density of the

bottom waters, it is important to use a deep

pressure reference level. Relative to 4000 dbar,

the potential density progression is from

densest in the Nordic Seas to least dense in the

Canadian Basin. However, relative to 0 dbar,

the Eurasian Basin waters are the densest.

EBDW is ventilated from the Eurasian continental

shelves around the Arctic through brine

rejection, which contributes salt. The densest

shelf water is formed in the Barents and Kara


(a)

0

1000

2000

3000

4000

(b)

0

1000

2000

3000

4000

280

290

ALASKA

ALASKA

3

0.1

6

Oxygen (μmol/kg)

380 390

286

0

1000 2000 3000 4000 5000

CFC-11 (pmol/kg)

0.5

0.05

0.5

295

290

288

CANADIAN

BASIN

CANADIAN

BASIN

Oden 2005 Oden 1991

300

Oden 2005

7

5

4 5

2

3

1

3

1.5

3

2

0.2

0.1

0.05

290

300

305 310 320

LOMONOSOV RIDGE

0.02 0.2

LOMONOSOV RIDGE

0.5

310

0.2

0.1

310

0.1

0.20.1

0

0.1

0.2 0.050.02

0.05

0.2 0.1

0.1

0

0

1000 2000 3000 4000 5000

0.5

305

305

75°N 80°N 85°N 90°N 85°N 80°N 75°N

2

1

1

310 305

300

EURASIAN BASIN

1.5

0.5

0.5

380

305

305

1

EURASIAN BASIN

SVALBARD

300

4.

310

R/V Knorr 2002

5

4

1

300

75°N 80°N 85°N 90°N 85°N 80°N 75°N

SVALBARD

<

R/V Knorr 2002

330

GREENLAND

SEA

> 0.5

360

>

>

GREENLAND

SEA

310

300

<

ICELAND

SEA

28 27

29 26 25

23 22 20 19 18 17 16 14 12 10 2468

4

300

6

1

3

2

295

ICELAND

295

295

ICELAND

ICELAND

SEA

00

0

1

FIGURE 12.16 Vertical section

across the Arctic and Nordic Seas. The

section extends from the Chukchi Sea

north of Bering Strait to the North Pole

to Svalbard and Iceland (on the right).

Corresponding sections of potential

temperature, salinity and potential

density were shown in Figure 12.11,

along with a station location map.

(a) Oxygen (mmol/kg), and (b) CFC-11

(pmol/kg). Station locations are shown

in Figure 12.11e. Vertical sections from

the Canadian Basin (Swift et al., 1997)

and the Eurasian Basin (Schauer et al.,

2002) are shown in Figure S12.6 on the

textbook Web site.

426


(a)

210˚

180˚

150˚

(b)

0

500

MaB NP

CaB

AmB

NaB GrS IcS

WSC

270˚

240˚

NP

CaB

MaB

AmB

120˚

90˚

Pressure

1000

1500

2000

2500

2000

2500

3000

WSC

IcS

AmB

GrS

NP

CaB

MaB

NaB

NaB

3000

3500

300˚

330˚

GrS

IcS

NAC

0˚

WSC

30˚

60˚

3500

4000

4500

4000

4500

34.89 34.90 34.91 34.92 34.93 34.94 34.95 34.96

30 31 32 33 34 35

Salinity

Pressure

(c)

0

500

1000

1500

2000

2500

3000

3500

4000

4500

GrS

CaB

MaB

NP

GrS

NaB

AmB

2000

2500

3000

3500

4000

NaB

AmB

WSC

GrS

WSC

IcS

AmB

NaB

IcS

CaB

MaB

NP

4500

–1.2 –1.1 –1.0 –0.9 –0.8 –0.7 –0.6 –0.5


–2 –1 0 1 2 3 4 5 6 7


(d)


2.5

2.0

1.5

1.0

0.5

0.0

–0.5

–0.5

–0.6

–0.7

–0.8

–0.9

–1.0

–1.1

46.38

IcS

WSC NaB

46.4

–1.0

IcS

CaB GrS

–1.5

MaB NP NaB

AmB

–2.0

30 31 32 33 34 35

Salinity

25

GrS

CaB

26

AmB

46.44

46.48

NP

MaB

–1.2

34.89 34.90 34.91 34.92 34.93 34.94 34.95 34.96

27

IcS WSC

FIGURE 12.17 (a) Station map (1994 and 2001), (b) salinity, (c) potential temperature ( C), and (d) potential temperaturesalinity.

Acronyms: CaB, Canada Basin; MaB, Makarov Basin; NP, North Pole; AmB, Amundsen Basin; NaB, Nansen Basin;

WSC, West Spitsbergen Current; GrS, Greenland Sea; IcS, Iceland Sea; and NAC, Norwegian Atlantic Current. This figure

can also be found in the color insert. Expanded from Timmermans and Garrett (2006).

28

Seas. About 10% of the EBDW can be accounted

for by brine rejection, with the remainder being

the original Nordic Seas Deep Waters that enter

through the Fram Strait (Östlund, Possnert, &

Swift, 1987).

In the Canada Basin (a sub-basin of the

Canadian Basin d see Figure 2.11), the

brine-rejected shelf waters are not dense

enough to renew the deep water. The bottom

waters of the Canada Basin have a mean age


FIGURE 12.18 (a) Schematic of bottom water connections, including approximate sill depths. Potential density is relative

to 2000 dbar. Arrows indicate mass fluxes overflowing into the basins and bottom arrows indicate geothermal heat flux. (b)

Potential temperature ( C) and (c) salinity at stations in the Makarov and Canada Basins (MB and CB, respectively). The

Makarov station “MB” is MaB in Figure 12.17a. ÓAmerican Meteorological Society. Reprinted with permission. Source: From

Timmermans and Garrett (2006).

of about 450 years compared with 250 years in

the Eurasian Basin, based on 14 Candthelow

levels of anthropogenic tracers (Macdonald,

Carmack, & Wallace, 1993; Schlosser et al.,

1997).

For the deep waters just above the Lomonosov

sill depth, the connection is mainly from

the Canadian Basin side to the Amundsen

Basin side. That is, the Amundsen Basin

bottom water is identical to the Eurasian Basin

bottom water, but there is a remarkable transition

to the warmer, more saline Canadian Basin

water above about 2000 m (Figure 12.17; Björk

et al., 2007).

428


Adiabatic (vertically uniform) bottom layers

of about 500 m thickness are apparent in the

Amundsen and Nansen Basin potential temperature

and salinity profiles in Figure 12.17, and

also at the North Pole station, which is on the

Makarov Basin side of the Lomonosov Ridge.

The Canada Basin has an even thicker adiabatic

bottom layer, from 2600 m to the bottom

(Figure 12.18). (The “Canada Basin” profile in

Figure 12.17 is not deep enough to capture this

layer.) The existence of adiabatic bottom layers

indicates that water crossed a sill to fill the

deep basin, which is then uniform in properties

below sill depth. The sill depth can be inferred

from the “break” at the top of the adiabatic

bottom layer.

In the Amundsen and Canada Basins, the adiabatic

bottom layer is warmer than the overlying

water, with a remarkably smooth curve connecting

the temperature minimum with the adiabatic

bottom layer. This temperature structure is due to

geothermal heating from below (Timmermans

et al., 2003). The absence of this structure in the

Makarov Basin indicates ongoing replenishment

(spillover) of cool water from the Amundsen

Basin. Based on the deep properties in each of

the basins, there is likely some small flow of

Amundsen Basin deep water to the Makarov

Basin, while any significant deep flow from the

Makarov Basin to the Canada Basin can be ruled

out (Timmermans & Garrett, 2006).

12.6. ARCTIC OCEAN

TRANSPORTS AND BUDGETS

The Arctic Ocean/Nordic Seas is a globally

important region of heat loss and dense water

mass production. Cooling of the AW flowing

through the Nordic Seas is the only major aire

sea heat exchange that occurs at high latitudes.

This Nordic Seas heat loss is responsible for

the existence of northward heat transport

through the full length of the Atlantic Ocean,

including the South Atlantic, compared with

the equatorially symmetric Pacific Ocean,

assuming that the Gulf Stream and Kuroshio

regions can be considered equivalent in terms

of their roles in heat loss in the Atlantic and

Pacific (Sections 5.5, 9.7, and 14.3).

The Arctic/Nordic Seas region is also important

for the global freshwater budget because of

its connection between the Pacific and Atlantic

Oceans, its net runoff and precipitation, and its

ice export to the Atlantic Ocean. The Arctic

freshwater export to the North Atlantic becomes

part of the newly formed NADW, and is an

important control on the salinity of that water

mass. Both as part of natural climate cycles

and as a response to anthropogenic change,

the Arctic’s ice cover varies. This changes the

albedo of the Northern Hemisphere and can

be an important part of climate feedback.

And not least of all, the Nordic Seas/Arctic

Ocean, together with the Labrador Sea,

comprise the main Northern Hemisphere region

where upper ocean waters are converted to

dense waters, thus providing the downward

“limb” of the part of the global overturning

circulation associated with the North Atlantic.

This transformation, from upper ocean to deep

ocean water, results from the large heat loss.

The transformation process shifts freshwater

from the surface layer down to the deep water

layer, so the dense water is the primary means

of exporting freshwater southward out of the

Arctic and into the mid-latitude Atlantic.

What is the current picture of the production

of dense waters in the Nordic Seas/Arctic

Ocean? Warm AW enters the Nordic Seas via

severalroutesandgathersintheNorwegian

Atlantic Current (Figures 12.1 and 12.19).

Part recirculates in the Nordic Seas and part

proceeds northward into the Arctic. It is

joined there by Bering Strait Water and

surface water from rain and runoff. This

whole upper layer cools further in the Arctic

and is a source of the dense EBDW; part of

the AW simply becomes more dense. Most of

this modified AW and Arctic Ocean Deep

ARCTIC OCEAN TRANSPORTS AND BUDGETS 429

210˚

150˚

240˚

0. 8 Sv

120˚

0

Net runoff/precip 0.2 Sv

0.5 Sv

1 Sv

1 Sv

-

2.3 Sv

1.8 Sv

60˚

300˚

2.5 Sv

3 Sv

1 Sv

330˚

0. 8 Sv

3. 8 Sv 3. 9 Sv

2 Sv

30˚

FIGURE 12.19 Volume transport budget. Red and orange are upper ocean inflows. Green is upper ocean outflow. Blue is

intermediate/deep outflow. Transports are listed in Sverdrups. See Figure S12.7 on the textbook Web site for the color

version.

Water returns to the Nordic Seas via the EGC,

and a smaller part returns southward west of

Greenland.

In the Nordic Seas, this returned, modified

Arctic water joins the locally circulating AW

and surface waters. A further densification

step occurs, mostly through deep convection

in the Greenland Sea (and Boreas Basin, which

adjoins it to the north). Buoyancy loss and

deep mixing also occur in the Iceland Sea

(Figures 12.1 and 12.10), contributing overall to

the new AIW layer. Brine rejection, specifically

430


in Storfjorden on the south side of Svalbard

(Figure 12.1), is also a densification process in

the Nordic Seas. The net result is production

of AIW (in the current decades), and, in earlier

decades, Greenland Sea Deep Water. The

portions of these that can overflow the relatively

shallow sills into the North Atlantic then

become part of the NADW.

Transports within this overturning system,

consisting of both the Nordic Seas and Arctic,

are as follows (Figure 12.19). The Norwegian

Atlantic Current transports 8.5 Sv of AW northward

into the Nordic Seas. The Bering Strait

funnels 0.8 Sv from the Pacific Ocean into the

Arctic Ocean (Roach et al., 1995). There is

approximately 0.2 Sv of runoff and precipitation

within the Arctic and Nordic Seas. The

net input is therefore 9.5 Sv. Outflows across

the Greenland-Scotland ridge include 6 Sv of

denser water beneath the Atlantic inflows and

3.5 Sv of lighter water from the Arctic Ocean

west and east of Greenland (via Davis Strait

and the EGC, respectively). Of the dense overflows,

3 Sv is in the Denmark Strait, 1 Sv is

over the Iceland-Faroe Ridge, and about 2 Sv

is through the Faroe-Shetland Channel. Therefore,

within the overall system, 6 Sv is converted

to denser water from the 9.5 Sv of

lighter inflow (Figure 12.19; following Jones,

2001 and Rudels et al., 1999).

For the Arctic portion alone, the inflow

consists of 1.8 Sv into the Barents Sea, 1e1.5 Sv

in the West Spitsbergen Current into the Arctic,

0.8 Sv through the Bering Strait, and 0.1e0.2 Sv

of runoff. The net input to the Arctic is thus

3.7e4.3 Sv. Outflow from the Arctic includes 1

Sv west of Greenland to the Labrador Sea, and

2.8e3.3 Sv through the Fram Strait into the

Nordic Seas. Of this Fram Strait transport, 0.5

Sv is Polar Surface Water and the remainder is

denser water e modified AW (~1 Sv) and

uPDW/EBDW (~1.3 Sv). Here “modified

Atlantic Water” is the AW core that has been

modified within the Arctic Ocean, becoming

denser (s q > 27.97), colder, and fresher, and

returning southward. This is the northernmost

transformation pathway leading to NADW

production, with a net conversion to 2.2e2.8 Sv

of denser water.

To reach the total conversion of 6 Sv of dense

overflow waters, this already denser water from

the Arctic joins the Nordic Seas water and all are

further transformed to the net 6 Sv of waters

denser than the Greenland-Scotland AW

inflows. Thus about half of the transformation

that feeds NADW is from waters that remain

within the Nordic Seas and do not circulate

through the Arctic; the other half is initially

transformed to denser water during a circuit

through the Arctic.

Residence times are estimated from volume

transports and layer volumes (Section 4.7).

Using their complete volume budget, Aagaard

and Greisman (1975) estimated that the surface

water is substantially replaced in 3 to 10 years,

the deep water in 20 to 25 years, and the

bottom water in the Eurasian Basin in about

150 years.

12.7. SEA ICE IN THE ARCTIC

The properties of sea ice were introduced in

Section 3.9, with a discussion of how salt water

freezes and the accompanying brine-rejection

process. We also introduced the concept of

polynyas, which are regions of open water

within ice-covered regions. Here we specifically

describe Arctic sea ice and its seasonal cycle.

Photographs of sea ice in the Beaufort Sea are

shown in Figure S12.8 seen on the textbook

Web site.

12.7.1. Distribution of Arctic Sea Ice

Sea ice covers most of the Arctic. Year-round

(multi-year) sea ice is found in some parts of the

Arctic, although the coverage is declining

(Chapter S15). Even in late winter, there are

regions that are almost always ice-free

SEA ICE IN THE ARCTIC 431

FIGURE 12.20

(2009a).

Ice concentration in 1979 in: (a) late winter (March) and (b) late summer (September). Source: From NSIDC

(Figure 12.20a). These include the eastern

Nordic Seas and part of the Barents Sea shelf,

where warm Atlantic waters flow northward

in the Norwegian Atlantic Current. Multi-year

ice is found throughout the Canadian Basin

and Greenland side of the Arctic (Figure 12.21).

First-year ice, by definition, is the ice in regions

of open water in late summer. Comparison of

the late winter and late summer panels in

Figure 12.20 gives an idea of where first-year

ice occurs: in the Barents and Kara Seas on the

Eurasian side, and periphery of the Chukchi

and Beaufort Seas on the Canadian side. Ice in

the Labrador Sea and Hudson Bay is also firstyear

ice.

Multi-year ice is found in the central Arctic,

particularly in the Canadian Basin (e.g., late

summer 1979 coverage in Figure 12.20b). The

oldest ice (>4 years) borders the Canadian

Archipelago (Figure 12.21). As sea ice cover

has been declining, all of the shelf regions and

Canadian archipelago areas have become more

ice free in late summer, and at some point there

will no longer be multi-year ice in the Arctic

(Section S15.4 on the textbook Web site).

In addition to categorizing sea ice by its age,

Arctic ice may be divided into three categories

that are closely related to age: Polar Cap Ice,

Pack Ice, and Fast Ice. The most extensive is

the Polar Cap Ice. It is always present and covers

about 70% of the Arctic Ocean, extending from

the pole to approximately the 1000 m isobath.

Cap Ice is very hummocky and is, on average,

several years old. In winter, the average ice

432


FIGURE 12.21 Arctic ice ages: (a) 2004 and (b) cross-section of ice age classes (right) as a function of time (Hovmöller

diagram), extending along the transect across the Arctic from the Canadian Archipelago to the Kara Sea shown in (a). This

figure can also be seen in the color insert. Source: Extended from Fowler et al. (2004).

thickness is 3 to 3.5 m but hummocks increase

the height locally up to 10 m above sea level.

(In the ridging process, two ice floes meet and

deform vertically to form a ridge, with one-third

of the ridge going up and two-thirds of the ridge

going down. In rafting, two ice floes also meet,

but one floe rises up and over the other.) Some

of this Cap Ice melts in the summer and the

average thickness decreases to about 2.5 m.

Leads and polynyas, which are open water

spaces, may form. In the autumn these freeze

over and the ice in them is squeezed into ridges

or is rafted. Polar Cap Ice is only penetrable by

the heaviest icebreakers.

The occasional ice islands, which have fairly

uniform ice thickness that is considerably

greater than the regular Cap Ice, originate

from glaciers on northern Ellesmere Island.

Pack Ice lies outside the Polar Cap. It consists

of a smaller fraction of multi-year and more

first-year ice than Cap Ice. It is lighter than

Cap Ice and up to a few meters thick. It covers

about 25% of the Arctic area, extending inshore

of the 1000 m isobath. Its area varies somewhat

from year to year. Seasonally, its areal extent is

least in September and greatest in May. Some

of it melts in summer and some is added to

the Cap by rafting. Pack Ice is advected southward

in the EGC and the Baffin and Labrador

Currents. While icebreakers can penetrate Pack

Ice, it impedes navigation in the northern parts

of the Canadian Archipelago, along the east

coast of Greenland, in Baffin Bay and the Labrador

Sea, and in the Bering Sea.

The edge of the Pack Ice is the marginal ice

zone. In this region, which can be tens to

hundreds of kilometers wide, the sea ice is loose

and broken. Surface waves provide energy to

break up the ice. As the waves enter the

marginal ice zone, they are scattered by the ice

SEA ICE IN THE ARCTIC 433

floes and their energy is attenuated. Upwelling,

eddies, and jets occur along the ice edge. Higher

levels of biological productivity are found in the

marginal ice zone compared with surrounding

waters.

Lastly, Fast Ice forms from the shore out to the

Pack and consists of first-year ice that forms

each winter. This ice is “fast” or anchored to

the shore and extends to about the 20 m isobath.

In the winter it develops to a thickness of 1e2m,

but it breaks up and melts completely in

summer. When it breaks away from the shore,

it may have beach material frozen into it and

this may be carried some distance before being

dropped as the ice melts, giving rise to “erratic”

material in the bottom deposits.

The general circulation of the Cap and Pack

Ice is similar to that of the Polar Surface Water

(Section 12.4). This moves the ice around and

exports it from the Arctic. Although Polar Cap

Ice is always present, it is not always the same

ice in a given location. Up to one-third of the

total Cap and Pack Ice is carried away through

Fram Strait in the EGC each year, while other

ice is added from the Pack Ice. Ice export

through Fram Strait and down the coast of

Greenland is at a rate of about 3 km/day. The

ice exports through Fram and Davis Straits are

major factors in the Arctic freshwater budget.

The volume of freshwater exported as ice is

approximately equal to the total continental

runoff into the Arctic basin.

12.7.2. Build-Up and Break-Up of

Arctic Sea Ice; Polynyas

To give some idea of the variation in ice

conditions with latitude, we present brief

accounts of the build-up and break-up of sea

ice from about 48 N to about 80 N in the Canadian

north. In the Gulf of St. Lawrence

(46e51 N), there is only first-year ice. Ice forms

first in the inner area (river), then along the

north shore, and becomes a hazard to shipping

in the main Gulf by January by covering most

of the area by the end of February with ice to

0.6 m thickness. Break-up starts in mid-March

and ships can move freely by mid-April along

mid-Gulf over the deep Laurentian Channel.

All the ice melts by summer. In severe winters,

build-up and break-up can be two months

earlier or later.

In Baffin Bay/Davis Strait (63e78 N), there

is mostly first-year ice of 1.5e2 m thickness,

with some older ice of up to 3 m thickness

entering from the north through Smith Sound.

In this region, ice cover is more common than

open water. Baffin Bay and the west side of

Davis Strait are largely covered by ice until

mid-May, mostly clear by mid-August except

off Baffin Island; ice then starts to develop

from the north by late October. Interannual

variations are considerable, with some areas

clearing by mid-June in a good year but

freeze-up starting as early as the end of August

in a “bad” year. A large, recurrent polynya,

called the “North Water Polynya,” is found in

the north end of Baffin Bay (Figures 12.1,

12.22); it is generally clear of persistent ice

throughout the winter and is a source of dense

water for Baffin Bay.

In the Canadian Archipelago, ice cover may

break up but floes remain present throughout

the year, and icebreakers are needed for surface

supply to northern outposts there. First-year ice

develops to 2.4 m thickness and multi-year ice to

4.5 m. Some clearing does take place in Lancaster

Sound (74 N, leading west from Baffin Bay)

and in passages further west by July, but floes

continue to be present.

The western Arctic (120 W to the Bering Strait)

is largely an open sea area north from the Canada/Alaska

coast, which is at about 70 N, with

a slow eastward current (Alaska Coastal Current).

Multi-year ice (Arctic Pack) of up to 4.5 m thickness

is general over the open sea south to 72 N,

while fast ice develops to 2 m thickness along

the coast. Open water is usually found near the

coast from mid-August to mid-September and

can even extend to 73 N, but in extreme years

434


1 Cape Bathurst 9 Franklin Strait 17 Hell Gate – Cardigan Strait

2 Lambert Channel 10 Bellot Strait 18 Lady Ann Strait

3 Roes Welcome Sound 11 Prince Regent Inlet 19 Bylot Island

4 Committee Bay 12 Lancaster Sound 20 Coburg Island

5 Foxe Basin 13 Viscount Melville Sound 21 North Water (NOW)

6 Frobisher Bay 14 Karluk Brooman 22 Flagler Bay

7 Cumberland Sound 15 Queens Channel and Penny Strait 23 Lincoln Sea

8 Fury and Hecla Strait 16 Dundas Island

FIGURE 12.22 Polynyas in the Canadian Archipelago. Predominantly latent heat polynyas: North Water, Cape Bathurst.

Tidally mixed polynyas: Committee Bay, Dundas Island, Lambert Channel and possibly Queens Channel, Bellot Strait, Fury,

and Hecla Strait. Polynyas in the Barents, Kara, and Laptev Seas are illustrated in Figures S12.9eS12.11 on the textbook Web

site. Source: From Hannah et al. (2009).

the Arctic Pack may extend to the coast in August.

Ship movements along the coast are generally

limited to September.

In the open North Pacific, ice does not occur,

but it is formed in the adjacent seas to the north

and west, that is, the Bering Sea, the Sea of

Okhotsk, and the northern Sea of Japan. In the

Bering Sea, pack ice extends in winter to about

58 N but clears completely in the summer,

retreating north through the Bering Strait to

CLIMATE VARIATIONS AND THE ARCTIC 435

70e72 N. Likewise in the Okhotsk and Japan

Seas, sea ice is seasonal, disappearing completely

each summer.

Polynyas (Section 3.9.6) are found all along the

Arctic margins and throughout the Canadian

archipelago (Figures 12.1, 12.22). Several satellite

images of Arctic polynyas are included in

Figures S12.9eS12.11 in the textbook Web site).

The wind-forced latent heat polynyas are especially

important for dense water formation

because of the continual production of sea ice

and hence brine within them; much of the AIW

is formed in the Siberian shelf polynyas (Martin

& Cavalieri, 1989; Smith et al., 1990). Latent

heat polynyas depicted in Figure 12.1 include

the North Water and Northeast Water around

northern Greenland, the Laptev Sea and Cape

Bathurst polynyas, and the polynyas around

Svalbard, Franz Josef Land, Nova Zemlya, and

Severnaya Zemlya. The Storfjorden polynya

(Svalbard) is the source of very dense water for

the Nordic Seas. The Laptev Sea flaw polynya

is the region of highest ice production in the

Arctic. The many polynyas of the Canadian

archipelago are shown in Figure 12.22; of these,

several are kept open through tidal forcing that

mixes warmer subsurface waters upward

(sensible heat polynyas) while others are windforced

(Hannah, DuPont, & Dunphy, 2009).

12.7.3. Arctic Icebergs

Icebergs differ from sea ice in that they originate

on land, have no salt content, have a density

of about 900 kg/m 3 (which is less than that for

pure ice because there are gas bubbles in

icebergs), and have much greater vertical dimensions.

They are a more serious hazard to shipping

than sea ice because of their large mass. In the

North Atlantic, the chief source of icebergs is

calving from the glaciers of west Greenland,

with a much smaller number from the western

side of Baffin Bay. The total number formed

each year is estimated at as many as 40,000.

Icebergs vary considerably in dimensions (height

above sea level/length), from 1.5 m/5 m for

“growlers,” 1e5 m/10mfor“bergybits,”5e15

m/15e60 m for small bergs, and 50e100 m/

120e220 m for large bergs. The ratio of volume

below sea level to that above is close to 7 to 1,

but the ratio of maximum depth below sea level

to height above it is less than this, depending on

the shape of the iceberg.

Icebergs have a large draft, so their movements

are chiefly determined by ocean currents.

(Pack ice motions are much more determined by

wind stress.) Icebergs have an average life of

2e3 years. They may travel up to 4000 km

from their origin in west Greenland. From there

they move northward in the WGC, across Baffin

Bay, and then south in the Baffin Island and Labrador

Currents at about 15 km/day, many

becoming grounded on the shelf. A small

proportion passes into the North Atlantic off

Newfoundland where they are usually a few

tens of meters high. The tallest recorded in this

region was 80 m high and the longest was about

500 m. The main season for icebergs in the

Grand Banks region is from March to July. Since

its inception in 1914 following the Titanic

disaster in 1912, the U.S. Coast Guard International

Ice Patrol has provided information about

icebergs coming south in the Labrador Current

to the Grand Banks region. The regular annual

surveys of both ice and oceanographic conditions,

as well as basic descriptions and understanding

of conditions in this region, provide

a century of information about the ice, circulation,

and water masses of the Labrador Sea

and Newfoundland regions (Chapter 9).

12.8. CLIMATE VARIATIONS

AND THE ARCTIC

The Arctic Ocean and Nordic Seas are central

to Northern Hemisphere climate variability. Four

modes of climate variability/change are

frequently used for describing variability in this

region: the Arctic Oscillation (also called the

436


Northern Annular Mode), the North Atlantic Oscillation,

theAtlantic Multidecadal Oscillation, and

global change driven by anthropogenic forcing.

Anthropogenic climate change scenarios show

the largest temperature changes in the Arctic.

Arctic sea ice extent and volume have been

decreasing and the ice has been becoming

younger and thinner since the late 1970s. Climate

feedbacks involving Arctic sea ice cover are

central to understanding and forecasting climate

change. Upper ocean temperature structure,

which is affected by sea ice and salinity as well

as by circulation and airesea fluxes, is an important

factor for understanding sea ice.

All of the remaining text, figures, and tables

relating to Arctic climate variability are found

in Section S15.4 in Chapter S15 (Climate Variability

and the Oceans) on the textbook Web

site. The following Arctic Ocean topics are

covered in Section S15.4: (1) Arctic Oscillation

or Northern Annular Mode, (2) Atlantic Multidecadal

Oscillation, (3) variations in Arctic sea

ice cover, and (4) variations in Nordic Seas and

AW properties, including discussion of longterm

trends that might reflect anthropogenically

forced climate change. The North

Atlantic Oscillation is discussed in Section

S15.1.

C H A P T E R

13

Southern Ocean

13.1. INTRODUCTION

The “Southern Ocean” is the broad ocean

region surrounding Antarctica (Figures 13.1

and 2.12). It is not a formal geographic region

in the sense of the Pacific, Atlantic, or Indian

Oceans or the many marginal seas, as it is not

surrounded by continental land masses.

However, the concept of a Southern Ocean is

important because the latitude range of the

Drake Passage between South America and the

Antarctic Peninsula has no north-south boundaries

(except in the deep water). As a result,

the strong Antarctic Circumpolar Current

(ACC) flows continuously eastward, encircling

Antarctica without wrapping back to the west;

it dominates the Southern Ocean’s large-scale

circulation. There is no western boundary at

the Drake Passage latitudes to support western

boundary currents and wind-driven gyres in

the upper ocean, although deep topography

does provide barriers for western boundary

currents in the deep and abyssal waters (Section

7.10.3). The ACC is the ocean’s closest analog to

the major wind systems, the westerlies and easterlies,

since the atmosphere also has no boundaries.

However, adding to the complexity of

the ACC, its strongest currents lie mostly north

or south of the Drake Passage, where there are

western boundaries (South America to the north

and the Antarctic Peninsula to the south), with

just a brief sojourn within the actual Drake

Passage. The coastline of Antarctica, south of

the ACC, includes two major indentations: the

Weddell and Ross Seas. These do have western

boundaries and thus support regional winddriven

gyres with western boundary currents.

The Southern Ocean is bounded to the

south by the Antarctic continent. Its northern

“boundary” is not well defined. The Antarctic

Treaty Limit at 60 S could be taken as a political

northern limit of the Southern Ocean. However,

the Southern Ocean oceanographic regime

extends well north of 60 S. If the presence of

the ACC is used to define the Southern Ocean,

then its northernmost boundary is at about

38 S, which is the northernmost excursion of

the ACC (Figure 13.1; Section 13.3). The most

inclusive definition in recent use extends the

region up to 30 S to fully encompass all

Southern Ocean phenomena northward to the

Subtropical Front in each ocean (Chapters

9e11). We do not insist on one definition of the

Southern Ocean. The processes described in

this chapter are associated mainly with the

ACC and regions to its south, and also include

the connections between the ACC and the ocean

basins to its north.

The narrowest constriction north of Antarctica

is the Drake Passage, between South

America and the Antarctic Peninsula. The

complicated bathymetry here and to the east,

in the Scotia Arc, presents the greatest latitudinal

blockage for the flow of the ACC. Two wider


437



438

13. SOUTHERN OCEAN

90 W

120 W

60 W

180

150 W

Subtropical Zone

Subantarctic Zone

Polar Frontal Zone

Antarctic Zone

STF

Ross S. Gyre

SAF

Southern Zone

SB

PF

SACCF

ASF

Ross

polynya

Mertz polynya

Adelie

Land

Wilkes

Land

Weddell

polynya

Cosmonaut

polynya

Darnley

polynya

Maud Rise

polynya

Antarctic Zone

Malvinas C.

Weddell

Gyre

Subpolar Region

SACCF

Brazil C.

SB

SAF

STF

Subantarctic Zone

30 W

0

60°S

SAF

PF

Polar Frontal Zone

150 E

30 E

STF

120 E

30°S

60 E

90 E

FIGURE 13.1 The Southern Ocean geography, principal fronts, and oceanographic zones (see Table 13.1). The Subtropical

Front (STF) is the oceanographic northern boundary for the region. The eastward Antarctic Circumpolar Current (ACC)

includes these fronts: Subantarctic Front (SAF), Polar Front (PF), Southern ACC Front (SACCF), Southern Boundary (SB).

Front locations from Orsi et al. (1995). The westward Antarctic Slope Front (ASF) (thin) follows the continental slope.

Circulation of the ocean basins north of the SAF is not represented; see the maps in Chapters 9, 10 and 11. Major polynyas

(dark gray patches) are labeled; all polynyas are shown in Figure 13.20.

FORCING 439

constrictions are set by southern Africa and

Australia. In all three constrictive regions, the

southward-flowing subtropical western boundary

currents (Brazil Current, Agulhas, and East

Australian Current) interact with the Southern

Ocean circulation. Mid-ocean ridges cross

through the Southern Ocean, resulting in strong

steering of the ACC through gaps in the ridges.

Several large undersea plateaus (Kerguelen,

Campbell, and Falkland) deflect the ACC. In

the latitudes of Drake Passage, deep topography

that can allow meridional geostrophic flow

occurs in the Drake Passage-Scotia Arc region,

Kerguelen Plateau, and Macquarie Ridge south

of New Zealand (Warren, 1990; Section 13.5).

Because the ACC connects the three major

ocean basins and because it is a deep-reaching

current, it is the vehicle for most flow between

the oceans. (There is a small transport of about

1 Sv from the Pacific to the Atlantic through

the Arctic Ocean, and a transport of 10e15 Sv

between the Pacific and Indian Oceans through

the Indonesian passages, but these are weaker

than the more than 100 Sv in the ACC.) The

unique character of deep-water masses originating

in each of the oceans is present in the

ACC. The waters mingle, upwell, are transformed

into both denser and lighter waters,

and then re-emerge to enter the ocean basins

to the north of the ACC.

Because of its high southern latitude and sea

ice formation, the Southern Ocean produces its

own very dense deep and bottom waters,

mostly along the coast of Antarctica. These

dense waters fill the deepest part of the oceans

to the north.

13.2. FORCING

The annual mean wind forcing for the

Southern Ocean is dominated by westerlies in

the latitude band 40e60 S and easterlies closer

to Antarctica, south of 60 S(Figure 13.2a). The

westerlies are not zonally uniform. They are

maximum in the Indian Ocean sector, centered

at about 50 S. The westerlies also have a significant

southward component, especially in the

eastern Indian Ocean, south of Australia. The

westerlies drive northward Ekman transport in

the latitudes of the Subantarctic Front (SAF)

and Polar Front (PF), which are part of the

ACC (Section 13.3). The net transport northward

across the (circumpolar) SAF is significant,

on the order of 30 Sv, which must be fed by

upwelled water from the south.

The wind stress curl is associated with

Ekman upwelling and downwelling (Section

7.5.4). The zero wind stress curl, associated

with zero Ekman upwelling, occurs at the

maximum wind stress, which is around 50 S.

Upwelling (positive values in Figure 13.2a)

occurs south of this, with highest Ekman

upwelling rates closer to the continent. Ekman

downwelling occurs north of the westerly

wind maximum and is strongest in the eastern

Atlantic and throughout the Indian Ocean

sector.

Close to the Antarctic continent, the winds are

easterlies, and can be very strong as a result of the

continental forcing (katabatic winds, which also

include a northward component). The easterlies

drive Ekman transport toward the continent,

inducing downwelling at the boundary. This

results in a mounding of the sea surface and

deepening of the pycnocline next to the continent,

which results in the westward geostrophic

flow found near the continent at most locations.

Surface buoyancy forcing is the sum of aire

sea heat and freshwater fluxes. These two

separate components are shown globally in

Figures 5.4 and 5.12 and S5.8, which can be

seen on the textbook Web site http://booksite.


material. The net buoyancy flux, converted

to equivalent heat flux units of W/m 2

as in Figure 5.15, is shown for the Southern

Ocean in Figure 13.2b. The net buoyancy flux

south of about 45 S is positive, meaning that

the surface waters become less dense. (This

440

13. SOUTHERN OCEAN

(a)

0˚ 30˚E 60˚ 90˚ 120˚ 150˚E 180˚ 150˚W 120˚ 90˚ 60˚ 30˚W 0˚

–1.0

40˚S

–1.5

–1.5

–0.5

-1

0.5

40˚S

2

0.5

60˚

2.5

1

60˚

4

5

1

6

Wind stress 0.1 N/m 2

80˚S

80˚S

0˚ 30˚E 60˚ 90˚ 120˚ 150˚E 180˚ 150˚W 120˚ 90˚ 60˚ 30˚W 0˚

(b)

40˚S

–2.0 –1.5 –1.0 –0.5 0 0.5 1.0 1.5 2.0 2.5

50

–25

–100

0

-100

50

0

–25

25

50

–100

–25

60˚

50

60˚

0

0

25

25

0

0

0 0

0

Buoyancy 0 flux (equiv. W/m 2 )

0

0

80˚S

80˚S

0˚ 30˚E 60˚ 90˚ 120˚ 150˚E 180˚ 150˚W 120˚ 90˚ 60˚ 30˚W 0˚

0

25

0

0

5

–25

0

Wind stress curl

x sgn(lat)

(x 10 –7 N/m 3 )

FIGURE 13.2 (a) Annual average wind stress (N/m 2 ) (vectors) and wind stress curl ( 10 7 N/m 3 ) (shading) multiplied

by 1 in the Southern Hemisphere so that positive values (dark grays) indicate Ekman upwelling, from the NCEP reanalysis

1968e1996 (Kalnay et al., 1996). (b) Annual mean airesea buoyancy flux, converted to equivalent heat flux (W/m 2 ), based on

Large and Yeager (2009) airesea fluxes. Positive values indicate that the ocean is becoming less dense. Contour interval is

25 W/m 2 (grid-scale contouring along the Antarctic coast has been removed). Dashed contours are the Subantarctic and

Polar Fronts from Orsi et al. (1995).

40˚S

map is clearly missing the buoyancy loss and

hence density gain in the coastal polynyas,

which are not represented in these products,

and which produce the deep and bottom waters

of the Antarctic.) This is the only large region of

the world ocean where freshwater fluxes are

a significant contribution to the net airesea

flux, but heat fluxes here are of similar magnitude

and warm the ocean. How can such

a cold, high latitude region be warming on

average? The upwelling of very cold water

and its subsequent northward Ekman transport

appear to control the airesea fluxes such that

the slightly warmer maritime air equilibrates

the cooler water. The highest buoyancy/heat

gain occurs along the SAF and PF of the ACC,

where westerly winds and hence northward

Ekman transport are high.

In the Southern Ocean, the regions of highest

buoyancy loss, due almost entirely to heat loss

(Figures 5.15 and S5.8 in the supplemental material

on the textbook Web site), are in the western

boundary current regions (Agulhas, East

Australian Current, and Brazil Current), and in

the Leeuwin Current (west coast of Australia).

Annual mean heat losses exceed 100 W/m 2 in

SOUTHERN OCEAN FRONTS AND ZONES 441

these regions. A zonal band of buoyancy (heat)

loss extends along the Agulhas Return Current,

stretching southeastward from Africa more than

half way across the Indian Ocean, with values

exceeding 25 to 50 W/m 2 . In the Pacific Ocean,

there is also a quasi-zonal band of buoyancy

loss in a similar position north of the SAF. The

highest buoyancy loss regions are associated

with southward mean flows, which bring

warmer waters into cooler regimes.

13.3. SOUTHERN OCEAN FRONTS

AND ZONES

Because of the open zonal passage and the

nearly zonal ACC (Section 13.4), isopleths of

all properties in the Southern Ocean are nearly

zonal (east-west) to great depth. Near-surface

potential temperature, salinity, and geopotential

anomaly (Figures 13.3 and 13.7) illustrate this

zonal nature, particularly in the latitude

range of Drake Passage. South of the ACC,

the surface properties and circulation are organized

by the cyclonic gyres (clockwise in the

Southern Hemisphere) in the Weddell and

Ross Seas, and are not as zonal as in the ACC

latitude band.

The nearly zonal isopleths of properties in the

ACC are organized into three major fronts separating

four broad zones in which isopleths are

more widely spaced (Figure 13.1). Within the

fronts, the currents are strong and eastward. In

the zones between the fronts, the flow is dominated

by eddies and can be in any direction.

The fronts encircling Antarctica as part of the

ACC are the SAF, the PF, and the Southern ACC

Front (SACCF; Figure 13.1 and Table 13.1). South

of the SACCF, Orsi, Whitworth, and Nowlin

(1995) define the Southern Boundary (SB), which

is the southern edge of the low oxygen layer of

the Upper Circumpolar Deep Water (UCDW); it

is not a dynamical front (Section 13.5.3 below).

Separate from, and south of the ACC, the

Antarctic Slope Front (ASF) is found at most

locations along the continental slope, with westward

flow and separating very dense shelf

water from offshore water (Jacobs, 1991;

Whitworth, Orsi, Kim, & Nowlin, 1998). On

the shelves, especially where they are broad,

and close to the coast, the westward flow of

the Antarctic Coastal Current (ACoC) is found.

The fronts separate the zones: Subantarctic

Zone (SAZ; north of the SAF), Polar Frontal

Zone (PFZ; between the SAF and the PF),

Antarctic Zone (AZ; between the PF and SACCF),

the Southern Zone (SZ; between the SACCF and

SB), and the Subpolar Region (south of the SB;

Orsi et al., 1995). The Subpolar Region (or

Subpolar Zone; SPZ) includes the Weddell and

Ross Sea gyres. On the continental shelf south

of the ASF, dense shelf water is found; this can

be considered the Continental Zone.

This classification scheme, from Orsi et al.

(1995), supersedes a commonly used older

scheme that did not include the SACCF and

SB, but instead identified a Continental Water

Boundary and a Continental Zone (CZ). The

older schemes were only appropriate for the

Drake Passage region.

The fronts and zones, as well as the typical

meridional (north-south) circulation and water

masses, are summarized schematically in

Figure 13.4. The water masses are discussed in

Section 13.5.

13.3.1. Fronts

Mean positions of the ACC fronts from Orsi

et al. (1995) as well as the ASF are shown in

Figure 13.1. InFigure 13.5 the ACC fronts are

shown in two regions that have been very well

mapped and described. The strong impact of

topography on the paths of the fronts is

apparent. In the Drake Passage and southwest

Atlantic (Figure 13.5a), the SAF follows the

boundary closely; it is an actual western

boundary current along the coast of South

America, where it is called the Malvinas (or

Falkland) Current. The other ACC fronts flow

442

13. SOUTHERN OCEAN

FIGURE 13.3 Properties at 50 m depth. (a) Potential temperature ( C), (b) salinity. This figure can also be seen in the

color insert. Source: From WOCE Southern Ocean Atlas, Orsi and Whitworth (2005).


FIGURE 13.3

(Continued).

444

13. SOUTHERN OCEAN

TABLE 13.1

Fronts and Zones of the Antarctic

Circumpolar Current and Southern Ocean

Feature Acronym Short Description

Subantarctic

Front

SAF

Northernmost ACC front

Polar Front PF Central ACC front

Southern ACC

Front

Southern

Boundary

Antarctic Slope

Front

Antarctic

Coastal Current

Subantarctic

Zone

Polar Frontal

Zone

SACCF

SB

ASF

ACoC

SAZ

PFZ

Southernmost dynamical

ACC front

Mostly along the continental

shelf, but also including the

Weddell Sea front

Continental slope front,

south of the ACC

Westward coastal flow

North of the SAF

Between the SAF and PF

Antarctic Zone AZ Between the PF and SACCF

Southern Zone SZ Between the SACCF and SB

Subpolar Region SPZ Between the ASF and SB

Continental

Zone

CZ

South of the Antarctic Slope

Front

Following Orsi et al., (1995) and Whitworth et al., (1998).

through passages in the many island chains and

loop around following the topography. The SAF

and PF merge together along the northern edge

of the Falkland Plateau. This merger of ACC

fronts is not uncommon; the distinction between

the SACCF and SB is also not always strong.

Likewise, in some regions the fronts are split

into multiple stable fronts, as observed between

Tasmania and Antarctica (Figure 13.5b). Again

these fronts are strongly influenced by topography,

in this case the Southeast Indian Ridge

and Macquarie Ridge.

The ACC fronts are sharpest in or just below

the surface layer. They are associated with

strongly sloping isopycnals in the water column

below, over much wider latitude ranges than the

surface fronts (Figure 13.6). These underlying

zones of steeply sloping isopycnals are referred

to using the surface front nomenclature (SAF,

PF, etc.). Most of the eastward flow of the ACC

is carried in these fronts.

The ACC fronts and the ASF are also associated

with transitions in water properties, as

depicted in potential temperature-salinity (T-S)

profiles crossing the ACC (Figure 13.7). These

transitions are often the practical means of identifying

the ACC fronts, especially when using

large data sets for which detailed examination

of each crossing of the ACC is impractical, or

for which velocity measurements are not available.

Such ”proxy“ markers of the fronts

include: (1) the existence of a particular water

property (e.g., temperature, temperature

gradient, salinity, oxygen) at a particular depth

and (2) the transition between water property

regimes typical of the zones between the fronts.

These markers are based on observations that

link the strongest eastward currents with

subsurface temperature and salinity structure.

The markers are not completely robust; they

may vary from region to region and the fronts

split into many different time-dependent fronts

in some regions, or merge in others (e.g.,

Figure 13.5). But the markers are a useful starting

point for finding the fronts in many regions.

A vertical section south of Tasmania that

crosses all the ACC fronts is used to illustrate

the indicators of the fronts (Figure 13.6).

The SAF is the northern edge of the ACC. It

was first identified in the region south of Australia,

and it exists in all other sectors of the

Southern Ocean (Emery, 1977; Orsi et al., 1995).

The SAF has large eastward flow, reflected in

steeply sloping isopycnals at all depths. At

most locations the SAF is the southernmost limit

of the low salinity intermediate layer, Antarctic

Intermediate Water (AAIW), and of the thick

layer of surface water, Subantarctic Mode Water

(SAMW; Section 13.5). Numerous other indicators

of the SAF have also been used, many


Continental

Zone

Subpolar

Region

ASF

Southern

Zone

SB SACCF

Antarctic

Zone

Polar Frontal

Zone

PF

SAF

Subantarctic

Zone

STF

CSW

2000 m

AASW

SAMW SASW STSW

SAMW

Antarctic Intermediate

Water

Upper Circumpolar Deep Water

Lower Circumpolar Deep Water

4000 m

Antarctic Bottom Water

Antarctic Circumpolar Current

to east (toward reader)

over most of depth

FIGURE 13.4 A schematic meridional section in the Southern Ocean showing the water masses, meridional circulation,

fronts, and most zones. Acronyms: Continental Shelf Water (CSW), Antarctic Surface Water (AASW), Subantarctic Mode

Water (SAMW), Subantarctic Surface Water (SASW), Subtropical Surface Water (STSW), Antarctic Slope Front (ASF),

Southern Boundary (SB), Southern ACC Front (SACCF), Polar Front (PF), Subantarctic Front (SAF), and Subtropical Front

(STF). After Speer, Rintoul, and Sloyan (2000).

(a)

70 W

60 W

Brazil

Current

50 W

40 W

30 W

(b)

80 W

40 S

50 S

6000

5000

4000

3000

2000

60 S

Malvinas/Falkland

Current

Falkland

Islands

Drake

Passage

Argentine

Basin

Falkland Plateau

Scotia South

Sea Georgia

South Scotia Ridge

Weddell

Sea

SB

60 S

PF

50 S

SACCF

Mid-Atlantic

Ridge

SAF

STF

40 S

20 W

1000

0

70 S

80 W

70 W

60 W

50 W

40 W

30 W

70 S

20 W

FIGURE 13.5 (a) Drake Passage and southwest Atlantic fronts. (Fronts from Orsi et al., 1995; bathymetry (m) from

Smith & Sandwell, 1997.) (b) Fronts south of Australia (Tasmania). N, M, and S refer to northern, middle, and southern

branches of the given fronts. Source: From Sokolov and Rintoul (2002).

446

13. SOUTHERN OCEAN

FIGURE 13.6 (a) Potential

temperature ( C), (b) salinity,

(c) neutral density (kg m 3 ), and

(d) oxygen (mmol/kg) along 140 E

from Antarctica to Tasmania

(WOCE Hydrographic Programme

Atlas section S3, from Talley, 2007).

Fronts: Subantarctic Front (SAF),

Polar Front (PF), Southern ACC

Front (SACCF), Southern Boundary

(SB), and Antarctic Slope Front

(ASF). Location of section is shown

by station dots in Figure 13.5b.


(a)

12

(b)

12

10

26

10

26


8

SAZ

8

SAZ

6

PFZ

SAF

6

SAF

4

4

PFZ

PF

2

2

PF

AZ

AZ

SACCF

0

0

Weddell Sea

SACCF

Atlantic 0°W

–2

–2

Ross Sea Pacific 130°W

33.5 34.0 34.5 35.0 33.5 34.0 34.5 35.0

Salinity

Salinity

26.5

27

27.5

FIGURE 13.7 Potential temperature-salinity relations: (a) Atlantic Ocean (Greenwich meridian) and (b) Pacific Ocean

(130 W), encompassing the fronts and zones of the ACC (Table 13.1). Contours are potential density s q (kg/m 3 ). Line near

bottom is the freezing point. Acronyms as in Table 13.1: SAZ (Subantarctic Zone), SAF (Subantarctic Front), PFZ (Polar

Frontal Zone), PF (Polar Front), AZ (Antarctic Zone), SACCF (Southern ACC Front).

27

26.5

27.5

associated with large horizontal (north-south)

changes in properties in the upper ocean (see

list in Belkin & Gordon, 1996). The SAF can be

identified at many locations by the occurrence

of the 4 or 5 C isotherm at 200 m depth, or

with a maximum horizontal gradient between

the 3 and 5 C isotherms (Sievers & Emery,

1978). All of these indicators are present in

Figure 13.6.

The SAF is farthest north (39e40 S) in the

western South Atlantic, just off the coast of

Argentina (Figures 13.1 and 13.5a). It shifts

southward as it progresses toward the east

and is farthest south (about 58 S) when it reaches

the eastern South Pacific and the Drake

Passage. At the eastern end of Drake Passage,

the SAF is close to the northern boundary at

55 S. As it leaves Drake Passage, the SAF

hugs the western boundary, shooting northward

to about 39 S to regain its northernmost

position. Along this coast, the SAF is a true

western boundary current d the Malvinas

Current.

The PF is within the ACC and is also a strong

eastward flow. The PF is identified by water

properties as the northern edge of the shallow

temperature minimum (Section 13.5). Again,

there are numerous indicators (see summary

in Belkin & Gordon, 1996). For instance, in

most regions it can be identified as the northernmost

location of the 2 C isotherm surrounding

the temperature minimum layer (Botnikov,

1963; Joyce, Zenk, & Toole, 1978; Orsi et al.,

1995), or as the location where the shallow

temperature minimum begins to steeply

descend toward the north. The PF is found, on

average, at about 50 S in the Atlantic and Indian

Oceans and at about 60 S in the Pacific, reaching

its southernmost location of about 63 S west of

the Drake Passage (Figures 13.1 and 13.5).

448

13. SOUTHERN OCEAN

The SACCF, introduced by Orsi et al. (1995),

is a major front with a large current near the

southern side of the ACC. Practical indicators

of the SACCF, at least in the southwest Atlantic,

include potential temperature less than 0 C

in the temperature minimum at depths shallower

than 150 m, or potential temperature

greater than 1.8 C in the potential temperature

maximum at depths greater than 500 m

(Meredith et al., 2003). It is distinct from the

SB, described next, in that the SACCF is a strong

dynamical feature, whereas the SB marks the

southern edge of the ACC in terms of water

properties.

The SB is the southern boundary of the oxygen

minimum that characterizes the UCDW

(Section 13.5.3). Of the major ACC water

masses, only Lower Circumpolar Deep Water

(LCDW) is found south of this. The SB is also

the northern boundary of the very cold, nearly

isothermal water mass found near Antarctica.

The SB is circumpolar in extent. It was first

observed and defined as the Continental Water

Boundary from observations in Drake Passage

(Sievers & Emery, 1978). Since this circumpolar

boundary is not close to the Antarctic continent

in large regions such as the Weddell Sea, the

more general “Southern Boundary” was

proposed (Orsi et al., 1995) and is used here. It

is located at the continental shelf only along

the west side of the Antarctic Peninsula.

The ASF lies along the continental slope at

many locations around Antarctica (Whitworth

et al., 1998). Flow in the ASF is westward. It is

mainly characterized by a pycnocline that

angles downward toward the continental slope,

due both to Ekman downwelling driven by the

easterly winds, and to downward penetration

of dense shelf waters. The ASF separates very

cold, dense waters on the continental shelf

from the offshore waters of the SZ, which

include Antarctic Surface Water (ASW) and upwelled

Lower Circumpolar Deep Water (LCDW).

Figure 13.17 shows a good example of the front,

which is “V-shaped.” The ASF is absent along

the western side of the Antarctic Peninsula,

where the ACC comes close to the continental

slope and isopycnals slope upward rather than

downward.

The ACoC is a westward coastal current that

lies on top of the continental shelf, within the

dense shelf water. It is not a water mass

boundary. It is sometimes nearly identical to

the ASF, especially where the continental shelf

is narrow. On the other hand, in some places,

notably along the western side of the Antarctic

Peninsula, the only westward flow is in the

ACoC (Klinck et al., 2004).

13.3.2. Zones

The SAZ is the region north of the SAF. At

most longitudes in the SAZ, salinity decreases

downward to a minimum value at 500 m or

deeper, and then increases below this. This

salinity minimum is known as Antarctic Intermediate

Water (AAIW; Section 13.5.2). The

higher salinity surface waters above the salinity

minimum are characteristic of the evaporative

subtropical gyres. Close to the SAF, the SAZ is

also characterized by a thick, near-surface layer

of nearly uniform properties, known as Subantarctic

Mode Water (SAMW; Section 13.5). The

northern boundary of the SAZ can be taken to

be the Subtropical Front, located at about 30 S

in each ocean (Figure 13.1).

Despite the name “Subantarctic,” the SAZ is

the poleward part of the subtropical circulation

regime in the Pacific, Atlantic, and Indian

Oceans, in which surface flow is dominantly

eastward. A difference between the Southern

Ocean subtropical regimes and the two

Northern Hemisphere subtropical gyres is that

part of the eastward flow in the SAZ in the

Pacific leaks through the Drake Passage into

the South Atlantic’s SAZ. Also, the SAZ is

continuous from the Atlantic to the Indian

Ocean, because the eastward part of the

subtropical circulation connects these two

oceans.

SOUTHERN OCEAN CIRCULATION AND TRANSPORTS 449

The PFZ (Gordon, Georgi, & Taylor, 1977) is

between the PF and the SAF. This zone varies

dramatically in both width and shape. There

are places where the SAF and PF merge, particularly

in the southwestern Atlantic, and there is

no PFZ at all. Within the PFZ, there is a dramatic

transition in T-S characteristics from the almost

isothermal T-S curve of the ASW that is south

of the PF to the much warmer and more saline

conditions north of the SAF (Figure 13.7). The

T-S relation in the PFZ is complicated due to

interleaving. (Interleaving is apparent as zigzagging

between the T-S profiles of the AZ and the

T-S profiles of the SAZ.)

The PFZ is occupied by strong eddies that

form as northward meanders of the PF

(Savchenko, Emery, & Vladimirov, 1978) or

southward meanders of the SAF (Figure 13.18).

The cold PF eddies can move northward to

become linked with the SAF, and thus carry

water from south of the PF across to the north

part of the PFZ, and vice versa, contributing to

the meridional exchange of heat between the

north and the south (Section 13.6).

The AZ is south of the PF and north of the

SACCF. It is characterized by a thin surface

layer of cold ASW (Section 13.5.1) with low

salinity from summer melting of sea ice. In

non-winter profiles, there is a subsurface

temperature minimum in the upper 200 m,

with temperatures from 1.5 to 2 C.

The SPZ lies between the SACCF and SB.

Within the SPZ, the low oxygen UCDW upwells

to the surface and is converted to very cold

ASW.

The SZ lies between the SB and the Antarctic

Shelf Front. In some sectors, this is a very broad

region, encompassing most of the Weddell and

Ross Sea gyres. In other regions, it is extremely

narrow, such as where the SB impinges on the

continental shelf west of the Antarctic

Peninsula.

In the CZ, south of the Antarctic Shelf Front,

there is a very cold water mass (<0 C). In winter

the layer is nearly isothermal and extends to

great depth (>500 m). Its density is controlled

by salinity. In some locations, these continental

shelf waters are the source of the very dense

deep and bottom waters known as Antarctic

Bottom Water (AABW; Section 13.5.4).

13.4. SOUTHERN OCEAN

CIRCULATION AND TRANSPORTS

The Southern Ocean circulation is dominated

by the strong, deep, eastward-flowing current

known as the Antarctic Circumpolar Current

(ACC), which runs completely around the globe

(Figures 13.1, 13.8, 14.2). The ACC was once

known as the “West Wind Drift” because it is

partially driven by the strong westerly winds

in the region, that is, winds from the west

causing the ocean to flow to the east. The westerly

wind in the Southern Ocean was notorious

in sailing ship days and, together with the eastward

current, made it difficult for such vessels

to round Cape Horn from the Atlantic to the

Pacific. The wind stress, combined with the

Coriolis force, also contributes a northward

Ekman component to the surface current. This

affects the formation of sharp fronts (Section

13.3) and convergences. The northward Ekman

transport is an important part of the meridional

overturn of the Southern Ocean. Below this

wind-driven surface layer, the density structure

appears to be in geostrophic balance with the

circulation.

The ACC is not purely zonal. The ACC as

a whole is farthest north just off the coast of

Argentina in the southwest Atlantic (northern

edge at 38 S), and farthest south just west of the

Drake Passage in the southeast Pacific (northern

edge at 58 S). The 2000 km southward spiral of

the ACC from the western Atlantic to the eastern

Pacific has important consequences for the water

masses of the Southern Ocean (Section 13.5).

South of the ACC are two cyclonic

“subpolar” gyres, one in the Weddell Sea and

the other in the Ross Sea. These gyres result in

450

13. SOUTHERN OCEAN

FIGURE 13.8 Geopotential height

anomaly at 50 dbar relative to

1000 dbar, in dynamic meters (10 J kg 1 ).

Source: From Orsi et al. (1995).

westward flow along the Antarctic coast, as seen

in the surface steric height maps for the Atlantic,

Pacific, and Indian Oceans (Figure 13.8 and

Figures 9.2, 10.2, 11.7). A nearly continuous

circumpolar westward flow driven by easterly

winds was hypothesized by Deacon (1937);

these two gyres and westward flow along the

continental shelf break in the Indian Ocean

result in such a picture, but with no apparent

westward flow in Drake Passage (Figure 13.9).

13.4.1. Antarctic Circumpolar Current

Early concepts of the ACC were that it is

a broad current of uniform velocity. It is now

clear that the ACC is composed of a series of

narrow jets that provide the overall large eastward

transport of the ACC (Section 13.3). The

narrow jets are confined within the broader

envelope of the ACC defined by the southernand

northernmost streamlines that are continuous

all the way around Antarctica (Figure 13.8;

see also Figure 14.2). In its circuit around the

continent, the ACC is severely obstructed in

the narrow Drake Passage (Figure 13.9), followed

downstream by a major northward

excursion along the western boundary of South

America (Malvinas/Falkland Current). In the

Australasian sector, the bottom topography of

Campbell Plateau (New Zealand) also constricts

the ACC, again accompanied by a northward

excursion of the ACC with the plateau acting

as a western boundary. The ACC path is also

affected by mid-ocean ridges as mentioned in

Section 13.1.

The ACC jets extend to the ocean bottom,

with bottom velocities in the same direction as

the jets (Section 13.4.3). This means that


54°S

56°S

58°S

60°S

62°S

64°S

66°W

68°W

25 cm/s

56°W

58°W

60°W

62°W

64°W

120

100

streamfunction, cm

FIGURE 13.9 Mean currents in the Drake Passage,

averaged over 30e300 m depth, from 128 ADCP crossings

over 5 years. Strong currents from north to south are the

Subantarctic Front (56 S), the Polar Front (59 S), and the

Southern ACC Front (62 S). This figure can also be found in

the color insert. After Lenn, Chereskin, and Sprintall (2008).

transport estimates based on temperature and

salinity measurements and the geostrophic

method with a “depth of no motion” are too

low. Direct current measurements are required

for total transports, at a minimum to provide

a reference velocity for geostrophic velocity

calculations.

Velocity and transport measurements of the

ACC have been made at a number of locations.

Because the Drake Passage is constricted and

relatively easy to access, the most comprehensive

observations have been made there, starting

in 1933 and continuing to the present (see

summary in Peterson, 1988). Recent monitoring

programs include annual hydrographic sections

and monthly sections of acoustic Doppler

80

60

40

20

0

current profiler (ADCP) velocities and expendable

bathythermograph (XBT) temperature

profiles (Figure 13.9). A long time series is also

continuing between Tasmania and Antarctica,

and other estimates have been made in the

central South Atlantic and Indian Oceans.

Mean surface current speeds for the whole of

the ACC are about 20 cm/sec. However, as

noted previously, most of the flow is carried in

the fronts. The three jets in the Drake Passage

were first clearly identified from data collected

in 1976 (Nowlin, Whitworth, & Pillsbury, 1977;

Figure 13.5). From surface drifters throughout

the Southern Ocean, the highest speeds are in

the SAF, with means from 30 to 70 cm/sec; the

PF is nearly as energetic with mean speeds of

30 to 50 cm/sec (Hofmann, 1985). Within the

Drake Passage, near-surface SAF and PF speeds

range up to about 50 cm/sec; the SACCF speeds

are somewhat lower (Figure 13.9).

Most ACC transport measurements have

been made in the Drake Passage because here

the ACC is clearly limited to the north and

south. The frontal structure observed in the

Drake Passage is the canonical structure

described in 13.3.1. Even just east of the Passage,

the PF splits in two where the ACC encounters

the Falkland Plateau (Arhan, Naveira Garabato,

Heywood, & Stevens, 2002). In the other

intensely observed region of the ACC, south of

Tasmania, the SAF, PF, and SACCF are each normally

two or more separate fronts (Sokolov &

Rintoul, 2002).

Transport estimates for the ACC in the Drake

Passage from 1933 through 1988 were published

by Peterson (1988). An early credible estimate

of 110 Sv (see Sverdrup, Johnson, & Fleming,

1942) is in the range of present estimates. The

first modern observations were made in the

1970s during the International Southern Ocean

Study, using a combination of current meters,

geostrophic calculations, and pressure gauges.

Means of 124 Sv (range 110 to 138 Sv), 139 Sv

(range 28 to 290 Sv), and 134 Sv (range 98 to

154 Sv) were estimated using different sets of

452

13. SOUTHERN OCEAN

measurements covering different lengths of

time (respectively, Nowlin et al., 1977; Bryden &

Pillsbury, 1977; Whitworth & Peterson, 1985).

Using six repeated hydrographic sections

from 1993 to 2000 in the Drake Passage,

Cunningham, Alderson, King, and Brandon

(2003) reported a baroclinic eastward transport

of 107.3 Sv 10.4 Sv relative to no motion at

3000 m. Most of this transport is in the SAF (53

10 Sv) and PF (57.5 5.7 Sv). Along a vertical

section just east of the Drake Passage, between

the Falklands and South Georgia Island, Arhan

et al. (2002) found a mean eastward transport

of 129 21 Sv concentrated in the SAF (52 6

Sv), and in two branches of the PF d one located

over the sill of the Falkland Plateau (44 9 Sv)

and the other in the northwestern Georgia Basin

(45 9 Sv).

In the Australian sector, between Tasmania

and Antarctica, six repeated hydrographic

sections yielded a mean transport of 147 Sv relative

to the ocean bottom (Rintoul, Hughes, &

Olbers, 2001). This is larger than the Drake

Passage transport. Eastward transport south of

Australia includes a contribution on the order

of 10 Sv that enters the Pacific and flows back

into the Indian Ocean north of Australia through

the Indonesian Passages (Chapters 10, 11).

The assumption of zero velocity at the ocean

bottom for referencing geostrophic transports

in the ACC is a useful starting point since

flow in the fronts is in the same direction

from top to bottom. However, since bottom

velocities are on the order of 4 to 10 cm/sec

based on direct current measurements, such

zero-at-the-bottom-referenced transports can

have large errors (Donohue, Firing, & Chen,

2001). Global and Southern Ocean inverse

models, which use geostrophic velocities calculated

from hydrographic data constrained so

that transports through closed sections must

balance, provide independent estimates of net

transport of the ACC. Macdonald and Wunsch

(1996) obtained 142 Sv through the Drake

Passage and 153 Sv between Australia and

Antarctica. Sloyan and Rintoul (2001) obtained

135 Sv for the Drake Passage and the section

between Africa and Antarctica, and 147 Sv

between Tasmania and Antarctica, similar to

the bottom reference level results.

13.4.2. Weddell and Ross Sea Gyres

The cyclonic circulations south of the ACC in

the Weddell and Ross Sea gyres are important

sites for formation of the densest waters in the

Antarctic and hence the global ocean. The

Weddell gyre is separated from the ACC by

the Weddell gyre front, which is identical with

the SACCF and is nearly co-located with the

SB (Section 13.3.1; Orsi et al., 1995). Within the

Weddell Sea, the flow is cyclonic (Figure 13.8).

The track of the Endurance from 1914 to 1916,

led by Sir Ernest Shackleton, illustrates the

cyclonic flow as the ship became frozen into

the pack ice (Figure 13.10).

The Weddell Gyre extends far to the east, to

the longitude of Africa, to 20 E at about 54 S

(Orsi, Nowlin, & Whitworth, 1993). Its northern

boundary is the Scotia Ridge in the west and

FIGURE 13.10 Track of the Endurance (1914d1916).

Source: From Stone (1914); Ó Royal Geographical Society.


then it approximately follows the 4000 and 5000

m bottom contour. There may be two separate

cyclonic gyres contained within the full Weddell

Gyre, centered at 30 W and 10 E. Southward

flow into the Weddell Gyre carries water from

the ACC and, therefore, the oceans north of

the ACC.

The Weddell Gyre has a western boundary

current that flows northward along the

Antarctic Peninsula. It carries new dense waters

from the Weddell shelves.

The net transport of the Weddell Gyre had

been estimated to be greater than 20 Sv based

on an absolute velocity analysis (Reid, 1994) or

15 Sv relative to 3000 dbar (Orsi et al., 1995). Direct

current measurements in the early 1990s suggest

30 to 50 Sv (Schröder & Fahrbach, 1999).

The southern Weddell Sea is occupied by the

Filchner-Ronne ice shelf (Figure 13.11a). The

eastern portion is the Filchner and the western

portion is the Ronne, separated by Berkner

Island. Beneath the ice shelf there is a sub-ice

shelf cavity of seawater. This is mixed vigorously

by tides, and is a factor in modifying

water masses of the southern Weddell Sea

(Makinson & Nicholls, 1999). Although the

Weddell gyre is cyclonic, flow in the cavity

appears to be anticyclonic, with ocean waters

(new dense shelf water, Section 13.5.4) entering

in the west, modified under the ice shelf, and

emerging colder and fresher in the east. The

outflow in the east from under the Filchner Ice

Shelf is a major source of the dense shelf water

that becomes Weddell Sea Bottom Water

(Jenkins & Holland, 2002).

The Ross Sea gyre is in the Pacific sector of the

Southern Ocean. Its northern edge is strongly

associated with topography (like that of the

Weddell gyre), following the Pacific-Antarctic

Ridge. Its transport is on the order of 20 Sv

based on absolute geostrophic velocities (Reid,

1997), or 10 Sv relative to 3000 m (Orsi et al.,

1995), and it, too, has a northward western

boundary current along Victoria Land, carrying

dense shelf waters.

The Ross Sea ice shelf is the largest ice shelf in

the world (Figure 13.11b). The sub-ice cavity

beneath the ice shelf is an important site for

(a)

(b)

Transantarctic Mountains

Weddell Sea

Filchner Ice

Shelf

Ronne Ice

Shelf

Ross Ice

Shelf

Ross Island

Ross Sea

FIGURE 13.11 (a) The Filchner-Ronne Ice Shelf in the southern Weddell Sea. (b) The Ross Ice Shelf in the southern Ross

Sea. Source: From Scambos et al. (2007) database.

454

13. SOUTHERN OCEAN

dense shelf water formation and modification.

Inflow is from the east and outflow is to the

west and north.

There are numerous other ice shelves around

Antarctica. The NSIDC (2009c) Web site is an

excellent source of information.

13.4.3. Mid-Depth to Bottom

Circulation

The eastward ACC extends from the surface

through mid-depths to near the ocean bottom,

as seen in the global circulation maps in

0

500

26.0

1000

27.3

1500

27.6

2000

27.9

2500

28.0

Depth (m)

3000

3500

28.36

28.20

28.10

28.04

4000

4500

0

5000

5500

30 S

6000

60 S

6500

0 1000 2000 3000 4000

Distance (km)

120 E 150 E 180

5000 6000

FIGURE 13.12 Neutral density section in the western Pacific into the Tasman Sea (WOCE section P11, location on insert).

Source: From WOCE Pacific Ocean Atlas, Talley (2007).

SOUTHERN OCEAN WATER MASSES 455

Chapter 14 (Figures 14.1 through 14.4), and in

the Pacific and Indian Oceans at 900 m from

floats (Figures S10.13 and S11.6 from Davis,

2005 as seen on the textbook Web site). Whether

specific portions of the currents reach the bottom

depends on bottom depth and topography, but

the ACC and the gyres are continuous to at least

3000 dbar (Reid, 1994, 1997, 2003). Below this

depth, mid-ocean ridges begin to impede the

continuous eastward progress of the ACC. At

3500 dbar there are no continuous streamlines

through the Drake Passage, thus separating the

Pacific and Atlantic below this depth.

By 4000 dbar, the circulation is broken into

regional deep gyres confined within the deep

basins. The gyres are cyclonic. Deep Western

Boundary Currents (DWBCs) are evident as

part of these deep gyres, especially along the

eastern coasts of South America (into the South

Atlantic) and New Zealand (into the South

Pacific). The DWBCs carry dense AABW northward

away from the continent. Because of the

generally cyclonic deep flow, currents along

the coast of Antarctica below about 3000 dbar

are westward. This is an important route connecting

dense Antarctic shelf waters from one

formation region to another.

The Weddell and Ross Sea gyres also extend

to the ocean bottom (Figure 14.4b). The Weddell

gyre is evident to at least 5000 dbar, and the Ross

Sea gyre down to at least 4000 dbar, both within

the confines of their deep basins.

The geostrophic shear of the ACC is large, as

reflected in the downward slope of isopycnals

toward the north across the current (Figure

13.12). Surface currents decrease from about

50 cm/sec at the sea surface to 4e10 cm/sec at

the bottom (direct current observations in the

Pacific by Donohue et al., 2001). The ACC fronts,

which are identified using potential temperature

and salinity (not shown), are evident to

the ocean bottom embedded within the general

slope of the ACC isopycnals. The rise of middepth

and abyssal isopycnals to the upper ocean

south of the ACC is an important factor in

creation of very dense AABWs from the deep

waters that enter the ACC from the oceans to

the north.

The top-to-bottom extent of the ACC is

important to note since its dynamical balance

is presumed to be between surface westerly

wind stress and bottom stress associated with

the topography. This differs from the dynamics

of the wind-driven gyres of the rest of the oceans

because there is no meridional boundary to

support a western boundary current at the latitude

of Drake Passage.

13.5. SOUTHERN OCEAN WATER

MASSES

Water masses in the Southern Ocean can be

considered in four layers: surface/upper ocean

waters, intermediate waters, deep waters, and

bottom waters (Figure 13.4; Table S13.2 in the

online supplement on the textbook Web site).

There are differing conventions for naming

water masses in the Southern Ocean; we follow

Whitworth et al. (1998) and Orsi, Johnson, and

Bullister (1999). These are mostly identified by

salinity, potential temperature, and potential

density, although one of the deep-water masses

(UCDW) is often identified by an oxygen

extremum. The surface waters are of local

origin. The one intermediate water of the

Southern Ocean, AAIW, originates as a fresh,

relatively dense surface layer in the Drake

Passage region. The deep waters mainly originate

from the Atlantic, Pacific, and Indian

Oceans, and mingle and mix in the ACC to

become CDW. It upwells south of the ACC

where a portion becomes the source of the

bottom waters around Antarctica. Some of these

dense Antarctic waters also modify the CDW.

A potential T-S diagram (Figure 13.13) shows

a typical station from the Atlantic from each of

the zones of the ACC. The zones can be categorized

by the water masses within them. The AZ,

south of the PF, contains ASW, Upper and

456

13. SOUTHERN OCEAN


12

10

8

6

4

2

0

Atlantic Ocean

October, 1983

Subantarctic Zone station: 39°S, 0°E

Polar Frontal Zone station: 48°S, 1°E

Antarctic Zone (Weddell Sea) station: 61°S, 1°E

26.5

27

0

26

Antarctic Surface Water

Subantarctic Surface Water

Upper

Circumpolar

Deep Water


freezing point

Shelf Water

–2

33.5 34.0 34.5 35.0

Salinity

1000

500

Winter Water

0

200

3000

5000

200

Subantarctic

Mode Water

27.5

2000

Lower

Circumpolar

Deep

Water

4000

5000

500

1000

Antarctic

Bottom Water

FIGURE 13.13 Potential

temperature-salinity curve of

Southern Ocean waters in the

Atlantic sector showing the

different water masses.

Lower CDW, Weddell Sea Deep Water, Ross Sea

Deep Water, and the Bottom Waters. The PFZ,

between the PZ and SAFs, contains the same

water masses, but at greater depth. The SAZ

contains Subantarctic Surface Water, SAMW,

AAIW, and the deep and bottom waters.

13.5.1. Surface Waters

13.5.1.1. Subantarctic Surface Water

and Subantarctic Mode Water

The Subantarctic Surface Water occupies up

to 500 m of the upper water column north of

the SAF. It has a temperature of 4 to 10 Cin

the winter and up to 14 C in summer, and

a salinity from 33.9 to 34 psu in winter and as

low as 33 psu in summer as ice melts. The lowest

temperatures and salinities are found in the

Pacific sector and the highest in the Atlantic

sector. Temperature and salinity of the surface

water increase toward the north. A high salinity

surface layer, to a depth of 150 to 450 m, is

present in all sectors. This is the surface layer

of the three (Atlantic, Indian, and Pacific)

subtropical gyres, which are dominated by

evaporation.

Within the Subantarctic Surface Water, just

north of the SAF, there are very thick mixed

layers in wintertime. These are known as

SAMW (McCartney, 1977, 1982; Hanawa &

Talley, 2001). In the central and eastern Indian

Ocean, these mixed layers can reach to more

than 500 m depth over a large region (Figure 4.4

and Figure S13.1 found on the textbook Web

site). In the South Pacific, the winter mixed layer

depths are not quite as extreme, but they are,

nevertheless, greater than about 300 m thick at

most longitudes. In the South Atlantic, the thick


winter mixed layers are more modest, on the

order of 200 m thick. These thick mixed layers

are capped in summertime by surface warming,

and are advected either eastward along the SAF

or subducted northward into the subtropical

gyres of the three oceans. Where they subduct,

the thick surface mixed layers become thick

layers within the permanent pycnocline. Given

that these thick layers move around the circulation

at speeds typical of the gyres, the volume

transport of SAMW within the gyres is higher

than that of thinner layers that are also subducted

into the gyres. Therefore, the SAMW

layer supplies a relatively large amount of

surface water to the subtropical pycnocline.

This may explain why the SAMW can be identified

within the subtropical gyres by high oxygen

content.

The meridional (south-north) sections of

neutral density in Figures 13.6c and 13.12

show the local type of SAMW just north of the

SAF. SAMW temperatures are warmest

(>14 C) east of South America, where the SAF

is farthest north. They decrease toward the

east as the SAF moves southward. At the longitude

of Australia, SAMW temperatures are

8e9 C. The very thick winter mixed layers in

this region are the densest waters that outcrop

in the Indian Ocean subtropical gyre, and they

become the primary source of high oxygen to

the base of the Indian Ocean pycnocline. For

this reason, it is useful to apply a special name

to the SAMW of this region: the Southeast Indian

SAMW or SEISAMW.

In the South Pacific, the SAF continues to

shift southward and the SAMW temperatures

continue to decline toward the east to

a minimum of about 4 C just west of Drake

Passage. This is the coldest, densest (and also

freshest) SAMW. This is nearly identical with

the salinity minimum of the AAIW (Section

13.5.2). This southeast Pacific SAMW and the

AAIW are the densest waters that outcrop in

the South Pacific’s subtropical gyre. Therefore

the portion of SAMW and AAIW that subducts

northward forms the base of the permanent pycnocline

in the subtropical gyre.

13.5.1.2. Antarctic Surface Water

The surface layer south of the SAF is referred

to as Antarctic Surface Water (ASW). ASW is

very cold and fresh because of cooling and

freezing in winter and ice melt in summer.

ASW extends to the base of the mixed layer in

winter. In summer, the ASW consists of a warm,

fresh surface layer of less than 50 m thickness,

overlying a cold, fresh layer (temperature

minimum) that is the remnant of the cold winter

surface layer. The temperature minimum is

sometimes referred to as “Winter Water.” The

warm surface layer is cooled to freezing in winter,

erasing this vertical temperature structure.

The subsurface temperature maximum layer

below the ASW lies below the influence of

winter freezing. This warmer water is the CDW.

Because there is easy exchange of surface

waters across the SB and the SACCF, Whitworth

et al. (1998) argued that Continental Shelf Waters

that are less dense than the CDW should be

considered as part of the ASW. Over the continental

shelf there is sometimes no CDW temperature

maximum underlying the ASW. They

suggest using the density of the nearby CDW

to define the ASW on the shelf.

The open ocean ASW layer is 100- to 250-m

thick. Its salinity ranges from 33 to 34.5 psu.

The ASW temperature is between 1.9 and

1 C in winter and between 1 and 4 C in

summer. The seasonal cycle of sea ice formation

and melting limits the range of winter-summer

temperature variation. A considerable proportion

of the heat inflow during summer is necessary

to melt the ice, leaving only a small part to

raise the temperature of the water.

South of the SACCF, the ASW is a true surface

layer, with the temperature minimum in summer

located at about 50 m depth at all longitudes.

Because it is tightly associated with

winter sea ice formation, the temperature of

the ASW temperature minimum (e.g., Winter

458

13. SOUTHERN OCEAN

Water) is nearly uniform throughout the

Southern Ocean south of the SACCF. ASW water

mass variations are therefore controlled by

salinity variations. As reviewed in Whitworth

et al. (1998), on the continental shelves ASW

can sometimes reach to the bottom, or even as

deep as 600 m on the slope in the Weddell Sea.

Between the SACCF and the PF, the temperature

minimum of the ASW increases in depth

and the temperature rises toward the north,

likely due to greater absorption of heat during

summer. The bottom of the ASW layer is no

more than 250 m deep. The PF is identified in

most places as the northernmost location of

the ASW’s temperature minimum.

13.5.1.3. Continental Shelf Water

South of the ASF, sitting on the continental

shelf, is a thick, very cold, nearly isothermal

layer. This layer is very close to the freezing

point in winter. It can be characterized in places

by an increase of salinity with depth. The

salinity stratification is most likely due to brine

rejection from the ice formation, which creates

denser, saltier water that settles at the bottom

on the shelf. Antarctic continental shelves are

quite deep (400e500 m) because of the large

mass of ice on Antarctica that depresses the

entire continent and its shelves.

Shelf Water is defined by Whitworth et al.

(1998) to be water as dense as AABW (neutral

density greater than 28.27 kg m 3 ) but near the

freezing point, hence colder than 1.7 C.

Waters above shelf water but denser than

ASW are called Modified CDW (Section 13.5.3).

Because shelf water is close to the freezing

point, variations in properties depend on

salinity. The saltiest continental shelf water is

the source of the dense bottom waters (Section

13.5.4).

13.5.2. Antarctic Intermediate Water

Throughout the subtropical gyres of the

Southern Hemisphere and the tropics of the

Pacific and Atlantic, there is a low salinity layer

at about 500 to 1500 m depth (Figures 13.6, 13.7,

13.13, 13.14). This is known as AAIW. It is found

north of the SAF, which is identified by the presence

of AAIW on its northern side at almost all

longitudes (Section 13.3.2). In the Pacific, the

AAIW spreads north to about 10e20 N where

it meets the North Pacific Intermediate Water

(NPIW) with lower density and salinity

(Chapter 10). In the Atlantic, the AAIW also

spreads to about 15e20 N where it meets the

Mediterranean Water (MW) with its much higher

salinity (Chapter 9). A weak signature of AAIW

can also be found in the Gulf Stream (Tsuchiya,

1989). In the Indian Ocean, AAIW is found to

about 10 S where it meets the fresh intermediate

water originating from the Indonesian Throughflow

(ITF; Banda Sea Intermediate Water) (e.g. Talley &

Sprintall, 2005).

In the T-S diagram for the Atlantic between

the SAF in the south and Iceland in the north

(Figure 13.14), the salinity minimum of the

AAIW can be identified easily (see also Chapter

9). Its temperature is 4e5 C and its potential

density is about s q ¼ 27.3 kg/m 3 , which characterizes

AAIW in the Atlantic and Indian Oceans.

Throughout the Pacific Ocean, its potential

density is lower, around s q ¼ 27.1 kg/m 3 .

The AAIW temperature and salinity change

toward the north in the Atlantic, although its

density remains relatively constant. The

AAIW salinity minimum is coldest and freshest

at the SAF. With increasing latitude, its

salinity and temperature increase. In the

subtropical North Atlantic (most of the profiles

north of 15 N in Figure 13.14), the AAIW

salinity minimum disappears, replaced by the

salinity maximum of the Mediterranean Overflow

Water.

AAIW has relatively high oxygen content in

the southeast Pacific and southwest Atlantic of

250e300 mmol/kg since it has only recently left

the surface in those regions. Oxygen on the

AAIW isopycnal indicates that it is in the surface

layer just west of the Drake Passage and


25

20

Atlantic Ocean 20° to 25°W

Blue: south of 51°S

Purple: 51°S to 32°S

Red: 32°S to 1°N

Orange: 1°N to 63°N

23

23.5

FIGURE 13.14 Potential

temperature-salinity diagram in

the Weddell Sea and Atlantic

Ocean. This figure can also be


24

24.5


15

10

25

25.5

26

60 N

27

28

5

26.5

30 N

28.5

29

0

30 S

0

60 S

90 W 60 W 30 W 0 30 E

33 34 35 36 37

Salinity

southern coast of Chile (Talley, 1999). In the

southeast Pacific, part of this low salinity

surface layer subducts northward and becomes

the AAIW of the Pacific Ocean.

AAIW in the Atlantic and Indian Oceans is

a modified version of this Pacific AAIW. New

Pacific AAIW is advected by the SAF through

the Drake Passage and into the Malvinas loop

east of South America. During the transit from

the Pacific, the AAIW properties change somewhat

to higher density and lower temperature.

As the AAIW rounds the loop, it plunges downward

to just beneath the thermocline in the

subtropical South Atlantic. From here it spreads

eastward along the SAF, and then northward

into the South Atlantic’s subtropical gyre. Part

of this Atlantic AAIW continues eastward into

the Indian Ocean and is advected northward

into the Indian subtropical gyre. In the Indian

Ocean, AAIW is a long way from its surface

origin, and it does not have especially large

oxygen content.

There is a long-standing controversy about

the origin of AAIW. The traditional view is

that it is formed by the sinking of ASW across

the SAF, at all longitudes around Antarctica, as

a natural result of northward Ekman transport

of the ASW. The opposing view of a more localized

source of the salinity minimum in the

southeast Pacific and the Drake Passage, as

described previously, may be supported by the

distribution of oxygen, salinity, and potential

vorticity (inverse layer thickness) on the AAIW

isopycnals (Talley, 1999).

460

13. SOUTHERN OCEAN

The traditional view of circumpolar formation

is possibly appropriate for the waters

directly beneath the AAIW salinity minimum.

From this perspective, the AAIW should be

defined as the salinity minimum and the layer

below the minimum that differs from CDW

(see the next section). In the T-S diagram

(Figure 13.14), the AAIW would then be defined

to include the salinity minimum and the part of

the nearly isothermal layer beneath it that lies

above the oxygen minimum of the UCDW (see

the next section). The division between AAIW

and UCDW occurs at about s q ¼ 27.5 kg/m 3 .

The densest outcrop on the north side of the

SAF sets the salinity minimum that defines the

top of the layer. The remainder of the so-defined

AAIW layer then comes from surface waters in

the PFZ that cross the SAF.

13.5.3. Circumpolar Deep Water

CDW is the very thick layer that extends

from just below the ASW (south of the SAF) or

the AAIW (north of the SAF) to just above the

dense bottom waters that are created on the

Antarctic shelves. CDW is partially derived

from the Deep Water of each of the ocean basins:

North Atlantic Deep Water (NADW), Pacific

Deep Water (PDW), and Indian Deep Water

(IDW). These northern deep waters enter the

ACC where they mix together. CDW upwells

across the ACC into the upper ocean in the

AZs and PFZs where it is transformed into the

Antarctic water masses (Figure 13.4). Shelf

water formed around Antarctica that is not

dense enough to become bottom water becomes

part of the CDW. Weddell Sea Deep Water is

a major source of such renewal of CDW. CDW

thus has an important component of locally

formed Antarctic waters.

CDW is usually divided into Upper CDW

and Lower CDW (UCDW and LCDW). There

are differing conventions on how to make this

division. We identify UCDW as an oxygen

minimum layer and LCDW as the salinity

maximum layer, following Whitworth et al.

(1998), Orsi et al. (1999), and Rintoul et al.

(2001). We also define the bottom of the CDW

as the isopycnal that is completely circumpolar

in the Southern Ocean, connecting through

Drake Passage, following Whitworth et al.

(1998) and Orsi et al. (1999). These definitions

differ from previous editions of this text.

In the AZ (south of the PF), UCDW includes

the temperature maximum layer at 1.5 to 2.5 C

that lies at 200e600 m, below the ASW. The

oxygen minimum layer (oxygen <180 mmol/kg)

in the AZ is nearly coincident with the temperature

maximum. The oxygen minimum is

a very large-scale feature that comes from the

deep waters north of the ACC, whereas the

temperature maximum is found only south of

the PF where the sea surface is near the freezing

point. Therefore, the oxygen minimum is the

most useful way to identify UCDW. Of the three

deep waters that form CDW, the PDW and IDW

have low oxygen (Chapters 10 and 11). Their

contribution to CDW creates the UCDW oxygen

minimum. North of the SAF, the UCDW oxygen

minimum lies at about 1500 m, centered at

a potential density of about s q ¼ 27.6 kg/m 3

and potential temperature of about 2.5 C. The

oxygen minimum slopes upward across the

SAF, following the upward slope of the isopycnals.

Oxygen is higher and potential temperature

is lower in the UCDW within and south of

the ACC, due to mixing with colder, newer

surface waters in this region.

UCDW also has high nutrient concentrations.

Where the UCDW upwells to just below the

surface layer in the AZ, it supplies nutrients to

the surface layer. This is one of the reasons for

prolific phytoplankton (plant) growth and

consequently, zooplankton in this region.

Zooplankton is a food source for larger animals

in the sea, which drew the major whaling

industry to the Southern Ocean.

LCDWincludes the vertical salinity maximum

that comes from NADW (Chapter 9; Reid &

Lynn, 1971; Reid, 1994). The lower boundary of


LCDW is the neutral density 28.27 kg m 3

(approximately s 4 ¼ 46.06 kg m 3 ), which

roughly corresponds to older definitions of 0 C

as the top of the AABW (Section 13.5). In the

AZ, LCDW lies at 400e700 m. In the SAZ, north

of the SAF, LCDW is found at 2500e3000 m in

the Atlantic but reaches to the ocean bottom in

the Pacific and most of the Indian Oceans.

The definition of the boundary between

LCDW and AABW is somewhat arbitrary. As

a result, LCDW so defined is the bottom water

for most of the world ocean outside the

Southern Ocean, except in the northern North

Atlantic where the densest water originates in

the Nordic Seas (Chapter 9). In many contexts,

this LCDW is referred to as AABW, but we

retain the more restrictive definition here. In

Chapters 9 and 14 we refer to the whole

LCDW complex as AABW.

Maps of properties in the LCDW are shown in

Figure 13.15. Potential temperature at the core of

LCDW is 1.3e1.8 C and potential density is

around s q ¼ 27.8 kg/m 3 .SalinityintheLCDW

salinity maximum is highest in the Atlantic

sector, around 34.8 to 34.9 psu. In the Indian

Ocean its maximum salinity is around 34.75

psu and in the Pacific around 34.72 psu. This

eastward salinity decrease comes from the lower

salinity IDW and PDW that join the ACC in their

sectors. Lower salinity deep waters south of the

ACC also reduce its salinity. (The salinity of

LCDW is lower in the AZ than in the PFZ.)

The NADW salinity maximum in the Atlantic

that yields the LCDW salinity maximum was

first observed in 1821, but was only later recognized

as originating in the North Atlantic by

Merz and Wüst (1922). In the western South

Atlantic, the NADW salinity maximum even

has a slight potential temperature maximum at

about 3 C, just below the slightly colder AAIW

(vertical section in Figure 4.11a and T-S diagram

of Figure 13.14). This slight temperature

maximum completely disappears in the SAZ

and ACC and is not a characteristic of LCDW.

LCDW flows northward from the SAZ into the

eastern South Atlantic; its salinity maximum is

less extreme than that of the NADW in the

west and includes no potential temperature

maximum.

LCDW also flows northward into the Indian

and Pacific Oceans, where its presence is indicated

by high salinity. The high salinity core

remains above the bottom in the Indian Ocean

but lies on the bottom in the Pacific north of

about 10e20 S, depending on longitude.

Some recent authors refer to the LCDW

salinity maximum core as NADW throughout

the Southern Ocean and well northward into

the Indian and Pacific basins. This ignores the

important inputs from the Antarctic, Pacific,

and Indian regions, so we prefer the CDW

nomenclature as used by Southern Ocean

specialists.

Specifically, the densest LCDW fills a much

greater region of the world oceans than the

densest NADW. The global impact of AABW/

LCDW is shown in Chapter 14 (Figures 14.14

and 14.15).

13.5.4. Antarctic Bottom Water

AABW is water in the Southern Ocean that is

denser than CDW and warmer than the freezing

point (Orsi et al., 1999; Whitworth et al., 1998).

As described in Section 13.5.3, CDW is defined

as being truly circumpolar, hence extending

through Drake Passage. The isopycnal that

divides AABW and CDW is therefore neutral

density 28.27 kg m 3 . Potential temperature

and salinity on this neutral surface are shown

in Figure 13.16. This neutral surface covers the

entire ACC region and extends northward in

the western South Atlantic, and into two basins

in the western Indian Ocean. Otherwise it is

confined to the southern regions by the major

ridges of the Southern Ocean. The coldest water

at this and higher densities is at the freezing

point on the continental shelves of Antarctica;

this water is Continental Shelf Water and is

considered separate from AABW.

462

(a)

13. SOUTHERN OCEAN

(b)

(c)

(d)

FIGURE 13.15 Properties along a Lower Circumpolar Deep Water isopycnal (neutral density 28.05 kg m 3 ), corresponding

roughly to the salinity maximum core. (a) Potential temperature ( C), (b) salinity, (c) depth (m), (d) oxygen (mmol/kg).

This figure can also be found in the color insert. Source: From WOCE Southern Ocean Atlas, Orsi and Whitworth (2005).


(a)

(b)

(c)

(d)

FIGURE 13.16 Properties on an Antarctic Bottom Water isopycnal (neutral density 28.27 kg m 3 ). (a) Potential

temperature and (b) salinity. Bottom properties (depths greater than 3500 m): (c) potential temperature ( C) and (d) salinity.

This figure can also be found in the color insert. Source: From WOCE Southern Ocean Atlas, Orsi and Whitworth (2005).

464

13. SOUTHERN OCEAN

An older definition of AABW is all southern

deep water that is colder than 0 C. The bottom

potential temperature map in Figure 13.16

shows that this region is more restricted in the

South Pacific than that of the neutral density

28.27 kg m 3 , and does not quite reach to the

Drake Passage. We therefore adopt the neutral

density definition.

The rather arbitrary neutral density distinction

between AABW and CDW means that the

southern-origin bottom waters of the global

ocean are AABW only in the Southern Ocean

and a small distance into the Southern Hemisphere

basins. North of this, the bottom waters

are LCDW (Figure 13.16 compared with

Figure 13.15). The restrictive neutral density

definition of AABW includes all of the regional

bottom waters in the Southern Ocean, including

Weddell Sea, Adélie, and Ross Sea Bottom

Waters (Whitworth et al., 1998), as well as Weddell

Sea Deep Water, which is colder than 0 C.

AABW is formed in polynyas along the continental

margins of the Weddell Sea, the Ross

Sea, the Adélie coast of Antarctica south of

Australia, and possibly also in Prydz Bay

(Tamura, Ohshima, & Nihashi, 2008). AABW is

a mixture of the near-freezing, dense Continental

Shelf Water (Section 13.5.1.3) and the

offshore CDW, which are separated by the

ASF. As the very dense shelf water spills down

the slope, it mixes with CDW to produce

AABW. In an example from the Weddell Sea

(Figure 13.17 from Whitworth et al., 1998),

both Continental Shelf Water close to the

freezing point (on the shelf and down the slope)

and AABW above the freezing point are

apparent. Both have neutral density greater

than 28.27 kg/m 3 . The CDW temperature and

salinity maxima are also observed offshore in

the figure. The V-shaped ASF is also evident,

reflecting geostrophic shear with westward

flow along the shelf break.

FIGURE 13.17 Vertical sections of (left) potential temperature and (center) salinity at about 35 W in the western Weddell

Sea. (Right) Potential temperature versus salinity. Dashed contours in (left) and (center) are neutral density. Near-horizontal

dashed line in right panel is the freezing point at 0 dbar. Source: From Whitworth et al. (1998).


AABW formed in the Weddell Sea is freshest

and coldest (34.53e34.67 psu, 0.9 to 0 C), along

the Adélie coast is intermediate in properties

(34.45e34.69 psu, 0.5 to 0 C), and in the Ross

Sea is warmest and saltiest (34.7e34.72 psu,

0.3 to 0 C; Rintoul, 1998). Volumetrically,

most AABW is of Weddell Sea origin (66%),

with Adélie Land contributing an intermediate

amount (25%), and the Ross Sea the smallest

amount (7%; also Rintoul, 1998). The fresh

Weddell Sea AABW and the salty Ross Sea

AABW are clear on the neutral surface in

Figure 13.16. Because the Adélie Land AABW

is intermediate in properties, it is not as obvious.

Within the Weddell Sea, the water masses

involved in dense water formation are the

ASW, CDW (also called “Warm Deep Water”),

Shelf Water, Weddell Sea Deep Water, and Weddell

Sea Bottom Water. Weddell Sea Deep Water

and Weddell Sea Bottom Water are defined by

potential temperature between 0 and 0.7 C

and potential temperature less than 0.7 C,

respectively. Weddell Sea Deep Water is a very

thick water mass, occupying depths of about

1500 to 4000 m. It has no particular property

extremum. It is formed within the Weddell Sea

in a manner similar to Weddell Sea Bottom

Water.

Weddell Sea Bottom Water is formed through

two processes: (1) mixing of ASW, UCDW

(known in the Weddell Sea literature as Warm

Deep Water), and Shelf Water formed on the

western shelf of the Weddell Sea and (2) mixing

of Ice Shelf Water (western Shelf Water that is

modified under the ice shelves) with Weddell

Sea Deep Water and UCDW. Western Shelf

Water is at the freezing point of almost 2.0 C,

which is possible because its pressure is about

400 dbar (station 742 in Figure 13.17). Its salinity

increases from 34.4 to 34.8 psu from east to west

along the shelf, enriched by brine rejection

during sea ice formation along its cyclonic circulation.

Potential density reaches s q ¼ 27.96 kg/m 3

(neutral density of 28.75 kg/m 3 ), among the

highest in the Southern Ocean. (Values to

s q ¼ 28.1 kg/m 3 are found in the Ross Sea

where the shelf waters are saltier than in the

Weddell Sea.)

Only AABW from the Weddell Sea can escape

northward from the Antarctic region through

a deep gap in the South Scotia Ridge. This

AABW enters the Scotia Sea, flows westward

to Drake Passage, and then eastward with the

ACC. As it crosses the ACC, it spreads northward

into the western South Atlantic, reaching

northward to the Brazil Basin. It also spreads

northward in the Indian Ocean into the Mozambique

and Crozet basins (Figure 13.16).

13.5.5. Overturning Budgets

The meridional overturning cell of the

Southern Ocean is shown schematically in

Figure 13.4. Ekman transport in the surface layer

is northward. UCDW and LCDW move southward

into the Southern Ocean and upwell. Buoyancy

loss due to cooling and salinification

through brine rejection create the dense Continental

Shelf Waters. These mix with LCDW and

create Modified CDW and AABW, which are

the dense waters that move northward out of

the Southern Ocean to fill the basins to the north.

UCDW experiences a buoyancy gain

(becoming lighter) through freshwater and

some heat input (Speer, Rintoul, & Sloyan,

2000). UCDW is incorporated in ASW and

moves northward along with the northward

Ekman transport. This northward transport is

incorporated in the denser part of AAIW.

Estimates of the overturning rates vary. The

northward Ekman transport across the SAF is

between 20 and 30 Sv based on various wind

products. An AABW formation rate of about

10 Sv is estimated by Orsi et al. (1999) based

on transient tracers. Various estimates of the

net northward transport of the denser part of

LCDW and of AABW northward out of the

Antarctic are 22e27 Sv, 32 Sv, 48 Sv, and 50 Sv

(Talley et al., 2003; Macdonald & Wunsch,

1996; Schmitz, 1995a; Sloyan & Rintoul, 2001,

466

13. SOUTHERN OCEAN

respectively). Taking these together with the

Orsi et al. (1999) estimate for AABW formation,

the LCDW formation rate in the Antarctic is at

least equal to the formation rate of AABW, and

may be much larger. Southward transport in

the UCDW and possibly LCDW must balance

the sum of northward Ekman and dense water

transports.

The dynamics of the Southern Ocean overturning

are beyond the scope of this text.

However, we do note that net southward

geostrophic transport in the upper ocean is not

possible across the latitude band of the Drake

Passage, since such transport requires a westeast

pressure gradient that must be supported

by a meridional boundary (Warren, 1990).

Above the depth of the undersea topography,

there is no such boundary. 1 Yet it is in precisely

this depth range that UCDW must cross to the

south. Speer et al. (2000) and others proposed

that this occurs through eddies. The eddy field

of the Southern Ocean is described next.

13.6. EDDIES IN THE SOUTHERN

OCEAN

Eddies are present in all regions of the global

ocean (Section 14.4), but have a special role in

the Southern Ocean because of the lack of an

upper ocean north-south boundary at the latitude

of the Drake Passage. By “eddies,” we

mean features with horizontal scales of at least

several kilometers, up to about 200 km, which

are departures from the time mean velocity or

properties such as temperature. (We do not

mean purely closed elliptical features in the total

flow or property contours, although the departure

from the mean sometimes has this sort of

shape.) In some Southern Ocean literature, there

is also reference to “standing eddies,” which are

departures from the zonal (west-east) mean, but

which have no time dependence; these can have

much larger spatial scale than the temporal

eddies. Most eddies arise from instabilities of

the ocean currents. Strong currents, such as the

fronts of the ACC, are especially unstable and

therefore have highly energetic eddy fields.

The wind-driven gyres in all other ocean

basins transport properties like heat, freshwater,

and chemicals. These gyres consist of largely

upper ocean currents forced by Ekman convergence

and divergence, closed by a western

boundary current (Section 7.8). No similar

wind-driven gyre can be present across the latitude

band of the Drake Passage since there is no

meridional boundary. We know from property

distributions that major exchange does occur.

One mechanism for exchange is the eddy field.

Consequently, evaluation of the eddy field is

central to understanding the ACC. In this

respect, the Southern Ocean in this latitude

range is analogous to the mid-latitude atmosphere,

where eddies play a dominant role in

the dynamics.

Heat transport at the latitude of the Drake

Passage is southward and is carried by eddies

rather than the mean flow (deSzoeke & Levine,

1981). This result was originally based on

inference from the mean heat transport and estimated

airesea heat flux in the Southern Ocean,

but has been substantiated by eddy-resolving

studies in recent years.

There have been few in situ regional studies

of eddy variability in the ACC due to its remoteness.

Long time series of velocity and temperature

have been collected only in the Drake

Passage and south of Australia, with results

extrapolated to other regions. Wide geographic

information but at limited depths is available

from subsurface floats, surface drifters, and

altimetry.

1 More precisely, there is no meridional boundary above the density that occurs at the sill in the Drake Passage latitude range;

this sill is actually located at Macquarie Ridge south of New Zealand and not in the Drake Passage. The other shallow

region in the Drake Passage latitude range is at Kerguelen Plateau in the central Indian Ocean.

EDDIES IN THE SOUTHERN OCEAN 467

Most of the eddy variability of the ACC is at

the mesoscale, with space and timescales of

about 90 km and 1 month (Gille, 1996). This

mesoscale variability is largely associated with

meanders of the SAF and PF, presumably due

to their instabilities. A snapshot of the eddy

field in the southeast Pacific, from altimetry, is

shown in Figure 13.18. The climatological positions

of the SAF and PF are superimposed.

The largest anomalies are the order of 20 to 30

cm; these are either meanders of the fronts or

cutoff eddies from the fronts. (Superposition of

the mean field is required to determine which

it is.)

Eddy activity is often depicted using eddy

kinetic energy (EKE), which is proportional to

the mean squared velocity anomaly (e.g., total

velocity with the time mean subtracted). Global

maps of EKE have been based in recent years on

Lagrangian surface drifters and subsurface

floats and on geostrophic velocity anomalies

calculated from sea surface height measured

by satellite altimeters (Figure 14.16).

A circumpolar band of high EKE follows the

ACC, mostly due to eddies of the SAF and PF,

which are vigorous, unstable eastward currents.

The EKE band for the ACC is most easily

defined in the Pacific Ocean, where it jumps

northward as the ACC passes New Zealand

(Campbell Plateau) and then shifts smoothly

southward toward the Drake Passage. In the

Atlantic Ocean, the band of high EKE along

the western boundary also includes eddies of

the Brazil Current extending eastward from

South America, and eddies of the Agulhas

Retroflection (Chapters 9 and 11). In the Indian

Ocean, the Agulhas front extending eastward

and shifting southward merges with the ACC

so that it is difficult from EKE alone to determine

where the high EKE of the ACC begins.

Cyclonic eddies in the Australian sector of the

ACC have been studied in situ. The cyclonic

FIGURE 13.18 Snapshot of eddies in the southeast Pacific and Drake Passage: sea surface height anomalies (cm) for the

week of October 1, 2005 from Topex/Poseidon altimetry (Aviso product). The climatological Subantarctic Front (SAF) and

Polar Front (PF) are marked.

468

13. SOUTHERN OCEAN

eddies are spawned by meanders of the PF and

SAF. The single eddy surveyed in Savchenko

et al. (1978) originated south of the PF and had

a cold core (Figure S13.2 on the textbook Web

site). Morrow, Donguy, Chaigneau, and Rintoul

(2004) paired in situ observations and satellite

altimetry to study a large ensemble of long-lived

cyclonic eddies generated by meanders of the

SAF. They concluded that these eddies play

an important role in cooling and freshening the

region north of the SAF where mode waters are

formed, equivalent to that of Ekman transport.

13.7. SEA ICE IN THE SOUTHERN

OCEAN

13.7.1. Sea Ice Cover

Sea ice in the Southern Ocean has a major

impact on Southern Hemisphere albedo and

on water properties, including deep and bottom

water formation in the Southern Hemisphere.

Southern Ocean sea ice covers an enormous

area at its maximum extent in late winter, but,

unlike the Arctic Ocean, almost all of the sea

ice is lost each year (Figure 13.19). Therefore

much of the sea ice in the Southern Ocean is

“first-year ice.” The exceptions are in the

western Weddell Sea and along the Ross Sea

Ice Shelf where ice cover usually persists

throughout the year.

In winter, the pack ice extends out 65 to 60 S.

Icebergs may be found between 50 and 40 S.

The relatively zonal distribution of the sea ice

edge is probably due to the zonal character of

the currents in the Southern Ocean.

Tabular icebergs in the Southern Ocean originate

from the ever-evolving ice shelves,

described in part in Section 13.4.2. A map of

all of the shelves is shown in Figure S13.3 on

the textbook Web site. Shelf ice is very thick

FIGURE 13.19 Annual

progression of sea ice

concentration in 1991,

computed from the Special

Sensor Microwave Imager

(SSM/I) carried on the

Defense Meteorological

Satellite Program satellites.

Source: From Cavalieri, Parkinson,

Gloersen, and Zwally

(1996, 2008).

SEA ICE IN THE SOUTHERN OCEAN 469

and extensive: the Ross Ice Shelf is 35 to 90 m

above sea level with corresponding depth

below, extending 700 km out to the Pacific. Shelf

ice is the extension of glaciers from the Antarctic

continent out on to the sea where the ice floats

until bergs break off. These tabular bergs may

be 80 to 100 km long and tens of kilometers

wide. In late 1987, the biggest berg recorded

broke off from the Ross Ice Shelf. It was 208 km

long, 53 km wide, and 250 m thick e it was

claimed to provide enough fresh water if

melted to satisfy the needs of Los Angeles or

New Zealand for 1000 years.

In March 2002, the northernmost of the ice

shelves, the Larsen B, with an area similar to

that of the state of Rhode Island, broke up due

to warming of the Antarctic Peninsula. The

unexpectedly high speed of the break-up has

been attributed to the presence of meltwater

ponds on top of the ice shelf; these filled the

crevasses with water and allowed them to

extend to the bottom of the ice shelf, thus

creating faster break-up than if the crevasses

had been filled with air (Scambos, Hulbe,

Fahnestock, & Bohlander, 2000).

When floating ice shelves break off and melt,

there is no change in sea level because the ice is

already displacing the water before it melts.

However, when the break-up includes continental

ice, or if the break-up contributes to

increased flow of land-fast glacial ice to the

sea, then it does cause sea level rise.

Regions of low ice cover, or polynyas, occur

in the Antarctic as well as in the Arctic (Section

12.7.2; Section 3.9). Much information about

them has been obtained from satellite observations

as well as from ships (Comiso & Gordon,

1987). Latent heat polynyas are found in many

locations around the coastline and ice shelf

edge. The resulting brine rejection produces

dense shelf water, some of which is dense

enough to create AABW. Three polynya regions

are most productive of AABW: the southern

Weddell Sea (68%), the Ross Sea (8%), and Adélie

Land (24%; Rintoul, 1998; Barber & Massom,

2007). Sea ice production is large in latent heat

polynyas, so a map of this production

(Figure 13.20) is a good indication of the location

of the polynyas and hence of dense water

formation, although the relationship between

sea ice production and dense water formation

is not one-to-one. Of the many polynyas displayed

in the East Antarctic region, the productive

Mertz glacier region is the main Adélie

Land source of dense water (Williams et al.,

2010). The Darnley polynya on the west side of

Prydz Bay is another potential source of dense

water that is just beginning to be explored

(Tamura et al., 2008).

Sensible heat polynyas in the Weddell sector

of the Antarctic have been observed in the

Cosmonaut Sea area (43 E, 66 S) and over

Maud Rise (2 E, 64 S; Comiso & Gordon,

1987). While these are not “ice factories” in the

sense of latent heat polynyas, they may be locations

of open ocean deep convection. The Maud

Rise polynya (“Weddell polynya”) was very

large in 1974 and persisted through three

winters. This was an unusual event, having

not recurred as of 2008; it has been linked to

feedbacks with the Southern Annular Mode,

which amplified the existing forcing due to

upwelling over the rise (Gordon, Visbeck, &

Comiso, 2007).

Significant year-to-year changes in sea ice

cover occur in the Southern Ocean. These are

linked to climate change at interannual to

decadal timescales, including El Niño-Southern

Oscillation, the various circumpolar modes of

decadal variability that have been determined,

and variations in the Southern Annular Mode

(Section 13.8).

13.7.2. Sea Ice Motion

The motion of the Southern Ocean ice cover

is related to the winds and, less importantly, to

the general circulation. The sea ice has been

tracked with the passive microwave satellite

SSM/I sensor; daily through long-term average

470

13. SOUTHERN OCEAN

FIGURE 13.20 Antarctic latent heat polynyas: sea ice production, averaged over 1992e2001. This figure can also be

found in the color insert. Source: From Tamura et al. (2008).

data sets are available from the National Snow

and Ice Data Center (Fowler, 2003). The mean

annual ice motion is shown in Figure 13.21.

The ice drifts generally westward next to the

continent; this matches the cyclonic general

circulation in the Weddell and Ross Sea gyres.

Eastward ice motion in the ACC matches both

the wind forcing and mean circulation there.

Northward ice motion occurs in wide regions

of the Ross and Weddell Sea gyres, as well

as in a wide region between 90 and 150 E,

north of Prydz Bay; katabatic winds blowing

off the Antarctic continent are a factor in this

northward motion. These large-scale patterns

of ice motion in the annual mean persist

throughout the year, based on monthly mean

maps using the same data set.

13.8. CLIMATE VARIABILITY IN

THE SOUTHERN OCEAN

Climate variability in the Southern Ocean is

still being characterized because of the shortness

of good time series. It is dominated by

the circumpolar Southern Annular Mode. El

CLIMATE VARIABILITY IN THE SOUTHERN OCEAN 471

FIGURE 13.21 Mean ice motion for 1988e1994 with the mean atmospheric pressure superimposed. Source: From Emery,

Fowler, and Maslanik (1997).

Niño-Southern Oscillation (Chapter 10) has an

impact on Southern Ocean climate modes,

especially at interannual timescales. Longer

timescales may be partially linked to anthropogenic

change.


relating to these Southern Ocean climate variability

topics are located in Chapter S15

(Climate Variability and the Oceans) on the

textbook Web site.

C H A P T E R

14

Global Circulation and

Water Properties

In this chapter we summarize the circulation

and water properties at a global scale, synthesizing

the regional elements from the individual

ocean basins (Chapters 9 through 13), and

present some evolving views of the global overturning

circulation. For courses providing just

a limited introduction to the ocean’s circulation

and water properties, it might suffice to use

materials from Chapter 4 and this chapter,

with highlights from the forcing fields in

Chapter 5 and introductory materials in the

basin Chapters 9 through 13.

The surface circulation systems (Section

14.1.1) have been observed for centuries in all

of their complexity, and are the best mapped

part of the circulation because of ease of access.

These circulations impact navigation, pollutant

dispersal, the upper ocean’s productive

euphotic zone, and continental shelves and

coastal zones. As the interface with the atmosphere,

the surface layer and circulation are

directly involved in ocean-atmosphere feedbacks

that affect both the mean states of the

ocean and atmosphere and also seasonal to

climate scale variability.

Just a few hundred meters below the sea

surface, some parts of the circulation change

dramatically as the wind-driven gyres contract

and weaken. At intermediate and abyssal

depths (Section 14.1.2), the circulation is

dominated by the deep penetration of the most

vigorous surface currents, and by circulation

associated with large-scale buoyancy forcing

and weak diapycnal processes that can change

the density of the water internally (Section 14.5).

The large-scale circulations include very

weak vertical velocities that connect these

deeper layers with each other and with the

upper ocean, referred to as the overturning

circulation (Section 14.2). The overturning circulation

includes shallow cells that cycle water

within the warmest, lowest density parts of the

ocean, which can be important for poleward

heat export from the tropics and subtropics.

The deeper overturning circulations, connecting

intermediate and deep waters to the sea surface,

are generally much more global in scope than

the wind-driven, upper ocean circulation

systems. The grandest scale overturning circulations

are those associated with North Atlantic

Deep Water (NADW) formation in the northern

North Atlantic and Nordic Seas, and with

dense water formation in the Southern Ocean.

A weaker, smaller scale overturning circulation

is associated with North Pacific Intermediate

Water (NPIW) formation in the North Pacific.

The drivers for the ocean circulation, its

variability, and its mixing are the winds and

airesea-ice buoyancy fluxes; the tides are an

additional source of energy for turbulent


473



474

14. GLOBAL CIRCULATION AND WATER PROPERTIES

dissipation, which is central to the overturning

circulation. In Chapter 5 we presented all of

these forcing fields. In Section 14.3 we revisit

the ocean’s heat and freshwater transports

with an emphasis on their relation to elements

of the ocean circulation.

Time dependence characterizes fluid flows at

all timescales. While this text mostly emphasizes

the large-scale, time-averaged circulation,

each basin chapter also introduces regions of

persistent local eddy variability. Here we

summarize the global distribution of eddy variability

and associated eddy diffusivity (Section

14.5). A brief overview of climate variability

and climate change in the global ocean is

provided in Section 14.6, but the main materials

are presented in the supplemental material as

Chapter S15 on the text Web site http://


supplemental material.

14.1. GLOBAL CIRCULATION

14.1.1. Upper Ocean Circulation

Systems

The global surface circulation is shown schematically

in Figure 14.1. This surface circulation

derives most of its characteristic shape and

strength from the basin-scale wind forcing in

each ocean (Section 5.8, Figures 5.16 and 5.17).

The anticyclonic subtropical gyres in each of

the five ocean basins are evident, with their

poleward western boundary currents: the Gulf

Stream and North Atlantic Current (NAC),

Kuroshio, Brazil Current, East Australian

Current (EAC), and Agulhas Current systems.

Each anticyclonic gyre has its eastern boundary

current regime: Canary, California, Benguela,

Peru-Chile, and Leeuwin Current systems,

respectively. The eastern boundary currents

Gulf

Stream

System

Equator

Labrador

Current

North Atlantic

Current

NECC

NEUC

North Brazil

Current System

Brazil

Current

System

East Greenland

Current

NEC

Norwegian

Atlantic Current

SEC

Canary

Current

System

Subtropical

Gyre

Atlantic Equatorial

Current System

Subtropical

Gyre

Benguela

Current

System

40

N

Somali

Current

System

Indonesian

Throughflow

Subtropical

Gyre

SEC

Kuroshio

System

40S

East Kamchatka

Current

Oyashio

Leeuwin

Current

East Australian

Current System

Bering

Strait

Subpolar Gyre

Subtropical Gyre

NEC

Subtropical Gyre

Beaufort

Gyre

Alaska

Gyre

Pacific Equatorial/Tropical Current System

North Pacific

Current

California

Current

System

SEC

NEUC

NECC

Peru-Chile

Current

System

Equator

Malvinas

Current

Agulhas

Current

System

Weddell

Sea

Gyre

Antarctic

Circumpolar

Current System

Subtropical Gyres

Equatorial and Tropical Circulations

Intergyre and/or Interbasin Exchanges

Polar & Subpolar Current Systems

Ross Sea

Gyre

FIGURE 14.1

(1996b).

Surface circulation schematic. This figure can also be found in the color insert. Modified from Schmitz

GLOBAL CIRCULATION 475

flow equatorward with the exception of the

Leeuwin Current, which flows poleward.

The higher latitude cyclonic circulations with

their equatorward western boundary currents

are evident in the Arctic and Nordic Seas, North

Atlantic, North Pacific including the marginal

seas, and the Weddell and Ross Seas. The

respective boundary currents are the East

Greenland (EGC) and Labrador Currents, the

East Kamchatka Current (EKC) and Oyashio,

and the boundary currents of the Weddell and

Ross Sea gyres.

In the tropics, the quasi-zonal tropical circulation

systems are apparent, including equatorial

countercurrents, equatorial currents, and lowlatitude

western boundary currents. Large-scale

tropical cyclonic circulation systems include the

zonally elongated North Equatorial Current and

Countercurrent “gyres” at 5e10 N in the Pacific

and Atlantic, the Angola Dome (South Atlantic),

and Costa Rica Dome (North Pacific).

While all circulation is time dependent to

some extent, tropical circulation variability is

particularly strong relative to the mean, with

fast responses to changing winds yielding

strong seasonal and interannual variability. Of

the major western boundary currents, only the

Somali Current system in the northwestern

Indian Ocean and the circulation in the Bay of

Bengal change direction completely (seasonally),

responding quickly to the reversing

monsoonal winds because of the narrow width

of their basins, which reduces the response

time to the changing winds.

The ocean circulations are connected to each

other. The North Pacific is connected to the

North Atlantic with a small transport (<1 Sv)

through the Bering Strait, through the Arctic,

and then southward both west and east of

Greenland. The tropical Pacific feeds water

into the Indian Oceans through the Indonesian

passages with a modest transport (~10 Sv).

The three major oceans south of South America

(Drake Passage), Africa, and Australia/New

Zealand are connected through the Antarctic

Circumpolar Current (ACC), with a large transport

(>100 Sv). (The Bering Strait and Indonesian

Throughflow (ITF) outflows from the Pacific

are supplied from the Southern Ocean as well.)

There are also many connections with

marginal seas that affect water properties within

the open ocean. Many of these connections are

shown in Figure 14.1; they are mostly discussed

in the ocean basin chapters.

Surface circulation mapped directly from

data has improved dramatically in recent years

because of more complete surface drifter data

sets and satellite altimetry (see Chapter S16 on

the textbook Web site). A drifter-based surface

dynamic topography map, which reflects the

surface geostrophic circulation, is shown in

Figure 14.2a (Maximenko et al., 2009). Globally,

the highest dynamic height is in the subtropical

North Pacific, which is around 70 cm higher

than the highest dynamic topography of the

subtropical Atlantic. The lowest dynamic height

surrounds Antarctica, south of the ACC. Relatively

low dynamic height is found in the

subpolar North Atlantic and North Pacific.

Surface velocity (Figure 14.2b) highlights

include the high velocities and narrow

boundary currents, the zonal tropical circulation

systems, and the ACC. This total surface

velocity field includes both geostrophic and

Ekman components. In the geostrophic flow

field, depicted by contours of surface dynamic

height (Figure 14.2a), complete closed subtropical

gyres are missing or distorted in some

regions, especially the North Atlantic. But

when the total including the Ekman component

is considered, the surface circulation appears

more gyre-like, and more similar to the 200 m

geostrophic flow (Figure 14.3). The streamlines

for total surface velocity also indicate regions

of convergence, where the streamlines terminate

in mid-gyre, and divergence where the streamlines

originate in mid-gyre. Convergence and

divergence are associated only with the Ekman

velocity, since geostrophic flow is non-divergent

by definition (Section 7.6).

476


FIGURE 14.2 (a) Surface dynamic topography (dyn cm), with 10 cm contour intervals, and (b) surface velocity

streamlines, including both geostrophic and Ekman components; color is the mean speed in cm/sec. This figure can also be

found in the color insert. Source: From Maximenko et al. (2009).

Just below the surface layer, even at 200 dbar,

all five wind-driven subtropical gyres are

tighter (more localized) than at the surface

(Figure 14.3). At this depth there is no Ekman

flow, so the total mean velocity is represented

by the absolute dynamic topographies in the

basin chapters, which the relative dynamic

topography in Figure 14.3 strongly resembles.

Compared with the sea surface, the subtropical

gyres are shifted toward their strong western

boundary currents and extensions, that is,

toward the west and the poles.

At 1000 dbar relative to 2000 dbar, the anticyclonic

gyres retreat even more toward their


FIGURE 14.3 Steric height (dyn cm) relative to 2000 dbar at (a) 200 dbar and (b) 1000 dbar, using mean temperature and

salinity from five years of float profiles (2004e2008). Source: From Roemmich and Gilson (2009).

western boundary current extensions (Figure

14.3b). 1 The contrast between maximum dynamic

height of the Pacific and Atlantic remains,

with the Pacific higher than the Atlantic. The

Southern Hemisphere gyres are much more

exaggerated at 1000 dbar than in the upper

ocean, while the Kuroshio and Gulf Stream gyres

are weaker. In Figure 14.3b, the Gulf Stream gyre

even appears to have disappeared in this relative

velocity calculation, in favor of general

northeastward flow into the subpolar gyre, but

the absolute geostrophic streamfunction even as

deep as 2500 dbar in Chapter 9 (Figure 9.14a)

retains a closed anticyclonic Gulf Stream gyre.

The subpolar circulations and ACC are more

barotropic than the subtropical circulations,

with little change in position from the sea

surface to the ocean bottom. This marked shift

in behavior from the subtropics to the subpolar

regions is most likely due to a reduction in

1 The absolute geostrophic streamfunctions at 1000 dbar in the basin chapters differ somewhat from the relative geostrophic

flow at 1000 dbar relative to 2000 dbar in Figure 14.3b, because the 1000 and 2000 dbar circulations are both weak. Therefore the

non-zero flow field at 2000 dbar is important to include when computing the 1000 dbar flow relative to 2000 dbar.

478


stratification, which then allows much deeper

penetration of surface signals.

14.1.2. Intermediate and Deep

Circulation

At intermediate depth (Figure 14.4a), the

geostrophic circulation, represented by steric

height, retains the western boundary currents,

their recirculations, and the ACC of the upper

ocean. It also retains a strongly zonal character

in the tropics (where flows are not well resolved

in the set of studies used in the figure). Importantly,

Deep Western Boundary Currents

(DWBCs; Section 7.10) appear by this depth,

and there is a transition to the structure of the

open-ocean deep flows that are affected by topography,

especially the mid-ocean ridges.

FIGURE 14.4 Streamlines for the (a) mid-depth circulation at 2000 dbar and (b) deep circulation at 4000 dbar. (Adjusted

steric height, representing the absolute geostrophic flow.) Source: From Reid (1994, 1997, and 2003).


Unlike the surface currents, deep currents

that are not a deep expression of a surface

current have only generic names. They are

mostly identified by location.

In more detail, each of the five subtropical

anticyclonic gyres retains some expression at

2000 dbar, including a western boundary

current that separates from the coast and flows

eastward, and a very compact anticyclonic

circulation on the equatorward, offshore side.

The Gulf Stream, Brazil Current, Agulhas, and

a deep version of the South Pacific’s anticyclonic

gyre (somewhat east of New Zealand at this

depth due to topography) are present. The

Kuroshio gyre is shifted north of its surface

location in Figure 14.4a, but in Chapter 10 we

noted that the Kuroshio Extension does extend,

weakly, to the ocean bottom at the same location

as its surface core (Figure 10.3).

The high latitude cyclonic circulations evident

at the sea surface are also present in the northern

North Atlantic, North Pacific, and south of the

ACC in the Weddell and Ross Seas, continuing

the near-barotropic character previously noted.

Major circulation features that appear at 2000

dbar, but not at the sea surface, include the

DWBCs and poleward flows along the eastern

boundaries. The DWBCs at 2000 dbar are southward

in the Atlantic and mostly northward in

the Pacific and Indian Oceans. The Atlantic

DWBC carries NADW from the northern North

Atlantic southward to about 40 S, where it

enters the ACC system. At this depth, unlike

at 4000 dbar, there is no DWBC in the subtropical

South Pacific, which has a deep-reaching

subtropical gyre. However, a northward

DWBC does form within the tropical Pacific

and can be traced northward along the western

boundary to past the northern boundary of

Japan, where it encounters a southward

DWBC. These flows are more clearly defined

at 4000 dbar. In the Indian Ocean, the northward

DWBCs are also more easily identified at 4000

dbar (next), but include northward flow along

the east coast of Madagascar and a hint of

northward flow in mid-basin that follows the

Central Indian Ridge.

At 2000 dbar, all three oceans export water

southward into the Southern Ocean. Part of

this southward transport is gathered in broad

poleward flows near the eastern boundaries,

and is evident in water properties in the South

Pacific and South Atlantic (see the basin chapters,

9e13). In the Indian Ocean, poleward

flow is evident in water properties west of

Australia, but it does not continue southward

to the ACC. In the South Atlantic, southward

flow of NADW is most vigorous in the DWBC

along the western boundary. For a dynamically

complete description, we note that the 2000 dbar

flows near the eastern boundaries of the North

Atlantic and North Pacific are also poleward,

indicating that basin-scale cyclonic circulation

at this depth is ubiquitous.

The circulation at 4000 dbar (Figure14.4b) is

greatly affected by topography. Here DWBCs

carry deep and bottom water northward from

the Antarctic into each of the three oceans. The

northward DWBC in the South Atlantic,

carrying Antarctic Bottom Water (AABW), shifts

eastward at the equator, becoming an eastern

boundary current along the mid-Atlantic Ridge

(Chapter 9); the DWBC at the continental

western boundary in the North Atlantic is

southward, carrying the deepest components

of NADW. The northward DWBCs in the Indian

Ocean follow the ridge systems, east of

Madagascar and northward into the Arabian

Sea, east of the Southeast Indian Ridge and

Central Indian Ridge, and east of Broken

Plateau into the western Australia Basin. In the

Pacific, the principal northward DWBC is east

of New Zealand, flowing through Samoan

Passage into the tropics, crossing the equator

and then northward along the western

boundary and also through the Wake Island

Passage and along the Izu-Ogasawara Ridge.

In the far northern North Pacific, the DWBC

is southward. This is counterintuitive if one

assumes (incorrectly) that DWBCs must carry

480


water away from their sinking sources. There is

no surface source of deep water in the northern

North Pacific. This DWBC thus best illustrates

the dynamics of the deep circulation, which is

driven by upwelling from the bottom and

deep water layers into the overlying intermediate

and shallow layers. According to the

Stommel and Arons (1960a,b) solution,

upwelling stretches the deep water column

thereby creating poleward flow in the deep

layers in mid-ocean through potential vorticity

conservation (Section 7.10.3). This poleward

flow is counterintuitive, since the main flow in

the basins is toward the high latitude sources

of deep water. In this theoretical framework,

the DWBCs are a consequence of closing the

mass balance for this upwelling. They are not

simply drains of dense water.

14.2. GLOBAL MASS TRANSPORTS

AND OVERTURNING

CIRCULATION

The overturning circulation in each ocean is

described in Chapters 9e13. We present here

a global picture of the volume/mass overturns.

Their role in global heat and freshwater transport

is summarized in Section 14.3.

The global overturning circulation is complex

and three-dimensional, with dominant paths

that we attempt to depict in a simplified manner.

The student is cautioned that these pathways

are not isolated tubes flowing through the ocean.

Many transport pathways depicted schematically

as narrow “ribbons” are broad and thick

flows, covering large regions. There is much

circulation and mixing from one “path” to

another. Throughout, and especially in the

upper ocean, the pathways can be caught up in

the much stronger wind-driven circulation.

Historically, there has been an emphasis on the

“meridional” overturning circulation (MOC),

which is important for the latitudinal redistribution

of heat, freshwater, and other properties.

However, some important elements of the global

overturning circulation and these redistributions

are not meridional. Inter-basin transports between

the Pacific, Indian, and Atlantic Oceans are crucial

for the global mean ocean state. Even at the ocean

gyre scale, the mean state is maintained by some

zonal components, such as between western

boundary air-sea heat loss regions and eastern

boundary air-sea heat gain regions.

Calculation and depiction of the MOC

usually includes computing the meridional

mass transports across each coast-to-coast,

zonal transect in isopycnal (or depth) layers

(Section 14.2.1), computing the upwelling and

downwelling transports between the layers in

closed geographic regions bounded by two transects

(Section 14.2.2), and computing the overturning

streamfunction to visualize the

overturn in two dimensions (Section 14.2.3).

Schematics of the overturn (14.2.4) are often constructed

to assist interpretation, but are obviously

not essential to the calculations.

More generally, this methodology applies to

any closed region, and could be used for zonal

transports across meridional sections, or even

transports into an open ocean region enclosed

by station data.

14.2.1. Mass Transports in Layers

into Closed Regions

The overturning circulation is calculated by

first defining closed geographical regions

within which net mass must be (nearly)

conserved. 2 For example, a closed region can

be defined by two coast-to-coast, top-to-bottom

transects (labeled “N” and “S” in Figure 14.5);

2 The mass balance is not exactly zero because there is a very small exchange of freshwater with the atmosphere. When

time dependence is considered, there can also be a time-dependent storage or deficit of mass within some regions; this also

is proportionally small when considering large regions.

GLOBAL MASS TRANSPORTS AND OVERTURNING CIRCULATION 481

Western

boundary

Layer 4

V S4

Layer 3

Layer 2

South latitude

V S3

V S2

North latitude

V N4

V N3

V N2

W 3

W 2

W 1

Sea surface

Surface layer

Intermediate layer

Deep layer

Interface 3

Interface 2

Interface 1

Eastern

boundary

FIGURE 14.5 Meridional overturning

circulation transport

calculation: example for four

layers. The mass transports for

each layer “i” through the southern

and northern boundaries of each

layer are V Si and V Ni .Thevertical

transport across each interface is

W i .Arrowdirectionsarethosefor

positive sign; the actual transports

can be of any magnitude and sign.

The sum of the four transports (two

horizontal and two vertical) into

a given closed layer must be 0 Sv.

The small amount of transport

across the sea surface due to evaporation

and precipitation is not

depicted.

Layer 1

V S1

Bottom layer

V N1

in the mean, the same amount of water must

move out of the region through one section as

moves in through the other (Section 5.1). For

data analysis, the latitudes are those of ocean

transects along which the data were collected.

For models, any latitudes can be chosen and

often many are used, thinking ahead to the overturning

streamfunction calculation described in

Section 14.2.3.

To look at overturn, the closed region is next

divided vertically into layers (c.f. i ¼ 1, 2, 3, 4 in

Figure 14.5). The boundaries between the layers

can be defined in different ways. Exact choices

depend on the purposes of the calculation. The

usual choices are isopycnals (or neutral density

surfaces), constant depths, and sometimes even

isotherms (of potential temperature). Isopycnal

(isoneutral) surfaces are usually the most informative,

because they are directly related to the

airesea buoyancy fluxes and diapycnal diffusion

that transform waters from one layer to another.

The net mass transports in the layers along the

southern and northern boundaries of the region

are then calculated (V Si and V Ni in Figure 14.5).

For hydrographic sections, the transports are

usually based on geostrophic velocities from

the sections, plus Ekman transports perpendicular

to the sections. Calculating the geostrophic

velocities from hydrographic station data is not

trivial since reference velocities are required

(Section 7.6); the overall mass conservation

constraint is one of the important inputs for

determining the best set of reference velocities.

Three different global calculations are superimposed

in Figure 14.6a (two based on data and

one on a global ocean model) and a fourth is

shown in Figure 14.6b; a fifth is represented in

Figure 14.9b and c by its overturning streamfunction

(Lumpkin & Speer, 2007). The most

robust elements of the overturning circulation

are common to all of these calculations.

For all five analyses in Figures 14.6 and 14.9,

mass transports were first computed for a large

number of isopycnal layers. These were combined

into three or four larger layers representing: the

upper ocean above the main pycnocline; a deep

482


(b)

40˚N

20˚N

0˚

20˚S

40˚S

60˚N

15.3

6.8

7.0

1.1 7.7

2.5

80˚N

19.0

17.6

3.9

80˚W 40˚W 0˚ 40˚E 80˚E 120˚E 160˚E 160˚W 120˚W

1.0

Mass transports (Sv)

Surface

1.0

Intermediate

Deep

0.7

Bottom

0.6

2.8

1.8

2.4

0.4

1.0

5.2

7.6

10.7

10.1

5.6

6.2

3.1

10.3

10.0

13.6

80˚N

3.9 5.9

0.5

60˚N

40˚N

20˚N

0˚

20˚S

3.6

2.5

40˚S

60˚S

60˚S

80˚S

80˚W 40˚W 0˚ 40˚E 80˚E 120˚E 160˚E 160˚W 120˚W

FIGURE 14.6 Net transports (Sv) in isopycnal layers across closed hydrographic sections (1 Sv ¼ 1 10 6 m 3 /sec). (a)

Three calculations from different sources are superimposed, each using three isopycnal layers (see header). Circles between

sections indicate upwelling (arrow head) and downwelling (arrow tail) into and out of the layer defined by the circle color.

This figure can also be found in the color insert. Source: From Maltrud and McClean (2005), combining results from their POP

model run, Ganachaud and Wunsch (2000), and Schmitz (1995). (b) Fourth calculation based on velocities from Reid (1994, 1997,

2003), with ribbons indicating flow direction and oveturn locations schematically. Source: From Talley (2008).

80˚S


layer that includes North Atlantic, Pacific, and

Indian Deep Waters; and a bottom layer that is

mainly dense Antarctic water (Lower Circumpolar

Deep Water, LCDW or AABW).

For the upper ocean layer, robust results are

(a) net northward mass transport through the

entire Atlantic (also including intermediate

water in Figure 14.6b), (b) southward transport

out of the Indian Ocean, (c) northward transport

into the Pacific, (d) westward transport from the

Pacific into the Indian Ocean through the Indonesian

passages (ITF), and (e) northward transport

out of the Pacific into the Atlantic through

the Arctic (Bering Strait). Along these pathways,

there is also weak upwelling into the warm

water path from deeper layers within the Pacific

and Indian Oceans.

Deep water in Figures 14.6 and 14.9 is transported

southward through the length of the

Atlantic and southward out of the Pacific. These

are the NADW and Pacific Deep Water (PDW),

respectively. Deep transport in the Indian Ocean

in Figure 14.6a is small and northward. When

the deep Indian layer is subdivided, as in

Figure 14.6b, the thinner layers have nearly

balancing northward and southward transports

of about 6 Sv; these are NADW moving northward

and Indian Deep Water (IDW) moving

southward at a slightly lower density.

Bottom water moves northward from the

Antarctic into all three oceans. Figure 14.6b

shows the northward penetration of bottom

water into the subtropical North Atlantic as

well, while the thicker layer used in Figure

14.6a subsumes this Antarctic water in the

bottom part of the southward-moving NADW.

The layer transports differ from one section to

another within each map. Therefore there is

convergence or divergence between the sections.

This results in upwelling and downwelling, as

described next.

The weak overturn of the North Pacific is also

depicted in Figure 14.6b. This cell transports

approximately 2 Sv of warm water northward,

and slightly denser NPIW southward.

14.2.2. Upwelling and Downwelling

Returning to the method (Figure 14.5), we

next calculate the vertical (diapycnal) transport

through the interfaces between layers within

the closed regions. The transports and velocities

for each layer i are

M Ti ¼ V Ni V Si þ W i 1 W i ¼ 0 (14.1a)

W i ¼ V Ni V Si þ W i 1 (14.1b)

w i ¼ W i =A i

(14.1c)

in which the vertical transport through the

bottom (“W 0 ”) is zero. A i is the area of each interface

and w i is the average vertical velocity

through the interface. The upwelling or transport

W i , in units of Sverdrups, across the top

interface of each layer is calculated from the

divergence of the horizontal transports in the

closed region plus the upwelling transport

across the bottom interface (Eq. 14.1b). We start

with the bottom layer, which has no flow through

its bottom boundary, labeled i ¼ 1inFigure 14.5.

The total transport M T1 into the closed region

must be 0 by continuity (Eq. 14.1a). This yields

the upwelling or downwelling transport W 1

across the upper interface of the bottom layer

since there is no transport through the ocean

bottom. The average upwelling velocity w 1 ,in

m/sec, across this interface is this transport

divided by the surface area A 1 of that interface.

Next move upward through each of the

layers and find the sum of the transports

through the side boundaries and through the

bottom interface (V Ni ,V Si , and W i 1 ). This yields

the net transport W i through the upper interface

of this box. Continue this for all layers up to

the surface. Because the overall velocity calculation

should have been carried out with mass

conservation (including Ekman transport in

the uppermost layer), there should be no net

upwelling or downwelling across the sea

surface, which is the upper interface of the

topmost box.

484


For example, if there is a net flow of 2 Sv

northward into the southern side of the bottom

box and a net flow of 1 Sv northward out of

the northern side of the box, then there must

be a net loss of 1 Sv within the closed bottom

region. Therefore, 1 Sv must upwell across its

top interface. The average upwelling velocity,

with an interface surface area of, say, 10 13 m 2 ,

would be 10 5 cm/sec.

Now, switching back to results for the actual

global ocean (Figure 14.6), we find: (a) the net

lateral (meridional) transport requires net

downwelling from the surface to the deep water

in the northern North Atlantic, (b) there is also

downwelling from the upper ocean and deep

water to bottom water in the Antarctic, and (c)

there is diapycnal upwelling of bottom water

in all three oceans in the low latitude region

between about 30 S and 24 N (the locations of

the zonal hydrographic sections). While most

of this upwelling is into just the overlying

deep water layer, some reaches the upper ocean

in the Indian and Pacific Oceans.

The “downwelling” process in the North

Atlantic is localized deep convection in the

Greenland and Labrador Seas followed by

entrainment of surrounding waters; this is the

production of NADW (Chapter 9). The

“downwelling” in the Antarctic is localized

brine rejection combined with entrainment that

increases its volume tenfold, mainly along continental

shelves and near ice shelves; this is the

production of AABW (Chapter 13).

“Upwelling” in low latitudes is associated

with eddy diffusion, driven by deep turbulence,

which has large geographical heterogeneity (see

Sections 7.3.2 and 14.5; Figure 14.7). In the

Indian and Pacific Oceans, this upwelling

produces the Indian Deep Waters (IDW) and

the PDW from upwelled AABW. In the Atlantic

Ocean, the upwelled AABW joins the NADW.

In greater detail, in the Indian Ocean there is

upwelling from the bottom to the deep water,

from the deep water to the intermediate layer,

and even a small amount of upwelling to the

thermocline layer. The South Pacific has similar

FIGURE 14.7 Modeled upwelling across the isopycnal 27.625 kg/m 3 , which represents upwelling from the NADW layer.

This figure can also be found in the color insert. Source: From Kuhlbrodt et al. (2007); adapted from Döös and Coward (1997).


processes, with net inflow in the bottom layer

and outflow in all layers above it, hence with

net upwelling into each of the layers, although

the quantities and vertical distribution differ

from the Indian Ocean.

In the Southern Ocean, using a finer division

of layers than in Figure 14.6, there is also diapycnal

upwelling from the deep water to the upper

ocean. The upwelling source waters are largely

the IDW and PDW that enter the Southern

Ocean at a slightly lower density and shallower

depth than the NADW. All three northern

source deep waters (NADW, PDW, and IDW)

physically upwell here to near the sea surface,

mostly adiabatically along the steeply sloped

isopycnals. The actual Southern Ocean diapycnal

“upwelling” can occur mostly near the sea

surface where airesea buoyancy flux can

directly transform the upwelled water. The

airesea buoyancy flux map of Figure 5.15 shows

the requisite (small) net heating and net precipitation

along the ACC that create lighter surface

waters. Part of the adiabatically upwelled water

also experiences cooling and brine rejection,

hence buoyancy loss, and sinks to make the

deep and bottom waters in the Antarctic.

The actual location of diapycnal upwelling

(buoyancy gain) is likely very complex. Observations

and budgets are as yet relatively sparse

so we do not have a detailed picture from

observations. Much more detail is available

from general circulation models than from

data, and suggest very localized processes. For

the isopycnal layer associated with NADW,

much of the diapycnal upwelling occurs in the

Southern Ocean south of the ACC, but there is

also enhancement along the equator and in

other regions associated with the circulation

and with complex topography (Figure 14.7).

14.2.3. Meridional Overturning

Streamfunction

A final quantitative step in depicting the overturning

circulation is to compute the meridional

overturning transport streamfunction for each

ocean and for the globe. The overturning streamfunction

is one of the basic diagnostic outputs for

ocean models used to study climate, as in the

Coupled Model Intercomparison Project. In

Chapter 7, we introduced the concept of a streamfunction

for geostrophic flow in the horizontal

plane (Eq. 7.23f): the velocity is parallel to the

streamfunction and its magnitude is equal to

the derivative of the streamfunction in the

direction perpendicular to the flow. Therefore,

the geostrophic streamfunction is the horizontal

integral of the geostrophic velocity field.

The overturning transport streamfunction is

conceptually similar. It is calculated and plotted

in a vertical plane, with a single horizontal

direction. For the MOC, this horizontal direction

is north-south. At any given latitude, the overturning

streamfunction J is the vertical integral

of the mass transport, summed from the bottom

of the ocean (bottom layer) to the surface:

J i ¼ PN

V i

i ¼ 1

JðzÞ ¼ R z

0

JðrÞ ¼ R r

o

R xeast

x west

vðx 0 ; z 0 Þdx 0 dz 0

R xeast

x west

vðx 0 ; r 0 Þdx 0 dr 0

(14.2a,b,c)

The transport streamfunction J has units of

transport (Sv). The discrete sum form (14.2a) is

the calculation that is actually carried out in N

layers that can be defined in depth or density

(or any other pseudo-vertical coordinate).

Upper case V i is the transport through the

section in each layer, that is, the integral of

velocities in that layer, however the layer is

defined. For the more mathematical reader/

student, two integral forms are given in

Eq. (14.2b,c), to make explicit the difference

between integrating in depth or in density.

Lower case v is the velocity (in m/sec) perpendicular

to the transect, which proceeds from

one coast, at x east , to the other coast, at x west .

486


To calculate the overturning streamfunction

J (14.2a), it is preferable to subdivide the water

column at each latitude into a large number of

layers, many more than the 3 or 4 depicted in

Figures 14.5 and 14.6. Again, the optimal layers

are isopycnal or isoneutral, rather than defined

in depth, although most published overturning

streamfunctions are computed with depth

layers.

The overturning streamfunction is calculated

at each available latitude (very few for hydrographic

data; many for an ocean model). The

transport streamfunctions for all latitudes can

then be contoured as a function of latitude and

vertical coordinate (Figures 14.8 and 14.9). If

the layers are defined by isopycnals or neutral

density surfaces, the streamfunction can be projected

back to depth coordinates by choosing the

average depth of the isopycnals at each latitude.

The depiction of the overturn in each separate

ocean from a global ocean model (Figure 14.8) is

representative of most published calculations,

although actual numerical values of the overturn

(in Sv) differ. These individual ocean overturns

were described in Chapters 9e13.

1. The Atlantic has an NADW cell with sinking

in the north and an AABW cell with inflow

from the south and upwelling into the

NADW layer.

2. The combined Pacific/Indian overturn

includes inflow of bottom waters (mostly

AABW) that upwell into the deep water and

thermocline layers, mostly in the Southern

Hemisphere and tropics.

3. Meridional overturns in the Northern

Hemispheres of the Pacific and Indian are

weak. The 2 Sv overturn of NPIW is apparent,

as is the weak, deeper overturn of Red Sea

Water (RSW) in the Indian Ocean.

The global overturning streamfunction is

constructed by summing the layer transports for

all oceans at each latitude (Figure 14.9). The major

features, found in all recent calculations, are (a)

shallow subtropical-tropical overturning cells

with sinking at about 30 latitude and rising in

the tropics; (b) the large deep cell due to NADW

with sinking in the north, occupying most of

Northern Hemisphere water column and extending

southward to about 35 S; (c) the deep cell

centered in the Southern Hemisphere at 3000 m,

with northward bottom flow (AABW) and southward

deep flow, mainly as PDW and IDW; and

(d) a top-to-bottom overturn in the south next

to Antarctica that forms AABW. These are all

principally connected to diapycnal processes,

with downwelling and upwelling limbs.

When the globally averaged overturning

streamfunction is calculated in depth layers

(Figure 14.9a,c) rather than isopycnal layers

(Figure 14.9b), the Southern Ocean also includes

a strong surface-intensified overturning cell

with sinking around 50 S and upwelling

between 35 and 50 S. This is called the “Deacon

cell.” It mostly disappears when the overturning

is calculated in isopycnal layers, that is, it is not

associated with a large amount of diapycnal

flux. The Deacon cell is mostly due to (adiabatic)

flow along isopycnals that change depth and

latitude: in the Southern Ocean there is a large

component of northward flow that returns

southward at the same density but at greater

depth (Döös & Webb, 1994; Kuhlbrodt et al.,

2007).

There is also well-documented diapycnal

upwelling in the Southern Ocean (Chapter 13).

The part of the Deacon cell that is mostly due

to depth-averaging is the downwelling limb

between 50 and 40 S. The “diabatic” Deacon

cell (Speer, Rintoul, & Sloyan, 2000), which

involves diapycnal transport, includes northward

Ekman transport in the surface layer

across the ACC, fed by upwelling from deep

waters that mostly rise to the surface adiabatically

(along isopycnals) as they move southward

across the ACC. Buoyancy gain at the

sea surface in the ACC vicinity then allows these

waters to move northward, decreasing in

density. They are mostly deposited into the

Subantarctic Mode Water (SAMW) layer just


FIGURE 14.8 Meridional overturning streamfunction (Sv) from a high resolution general circulation model for the (a)

Atlantic, (b) Pacific plus Indian, and (c) Indian north of the ITF. The Southern Ocean is not included. Source: From Maltrud and

McClean (2005).

488


north of the ACC and then move into the

Southern Hemisphere gyre circulations to the

north.

14.2.4. Overturning Circulation

Schematics

Here we use schematics to summarize the

elements of the global overturn, based on the

preceding transport, upwelling, and streamfunction

calculations. All such schematics are

incomplete since they cannot represent the

complexities of the large-scale circulation or

eddying and time-dependent paths. Therefore

they should always be interpreted cautiously.

Richardson (2008) presented a good history of

such overturning schematics from the earliest

in the nineteenth century to the present.

The widely known popularized depiction of

the global NADW cell, often referred to as the

“great ocean conveyor,” is shown in Figure 14.10

(after Broecker, 1987, 1991, which were based on

Gordon, 1986). Although this diagram has deep

deficiencies in terms of representing the actual

global overturn, it is nevertheless useful for

public education: it is simple and it is global. It

nicely illustrates the Atlantic-Pacific/Indian

asymmetry in deep water formation, with

sinking somewhere near the northern North

Atlantic but not in the other two oceans. This

particular simplification captures only a part

of the global overturn because the important

multiple roles of the Southern Ocean were

intentionally excluded; other overly simplified

descriptions do not include the essential roles

of the Indian and Pacific Oceans.

FIGURE 14.9 Global meridional overturning streamfunction (Sv) for (a) a global coupled climate model with high

resolution in latitude. Source: After Kuhlbrodt et al. (2007). (b, c) For hydrographic section data at several latitudes, plotted as

a function of neutral density and pressure; contour intervals are 2 Sv. The white contours are typical winter mixed layer

densities; gray contours indicate bathymetric features (ocean ridge crests). Source: After Lumpkin and Speer (2007).


(b)

24.7

26.99

6

27.6

level γ n

27.88

28

28.06

28.11

18

12

0

-14

-10

-2

28.2

(c)

p (dbar)

1000

2000

3000

4000

5000

80S 62S 32S 24N 48N 60N 80N

0

-2

12

-10

-14

20

10

0

20

10

0

−10

−20

6000

80S 62S 32S 24N 48N 60N 80N

Latitude

FIGURE 14.9 (Continued).

The next three simplified schematics illustrate

the elements that we consider essential

for a comprehensive teaching presentation of

the global overturn. A number of global overturn

schematics capture most of the aspects; in

particular we note Gordon (1991), Schmitz (1995),

Lumpkin and Speer (2007), and Kuhlbrodt

et al. (2007).

The global overturning can be divided into

two major, connected global cells, one with

dense water formation around the North

Atlantic and the other with dense water formation

around Antarctica. These are the NADW

and AABW cells, respectively. These two cells

are interconnected, especially in the Southern

Ocean, complicating any simple representation

of the overturn. A third, weak overturning cell

is found in the North Pacific, forming a small

amount of intermediate water (NPIW); it is

mostly unconnected to the NADW/AABW cells,

but is included because it contrasts the weakness

of dense water formation in this high-latitude

region with that in the high-latitude regions of

the Atlantic and Southern Ocean.

Essential features for the global NADW cell

are as follows. 3 The NADW cell begins with

3 We ignore along-isopycnal exchange of deep waters between oceans. We also have to ignore the finer steps of the

large-scale upwelling process, which could better be likened to hundreds of steps on different staircases in a building

of many floors, rather than a single leap from one very thick layer to another.

490


FIGURE 14.10 Simplified global NADW cell, which retains sinking only somewhere adjacent to the northern North

Atlantic and upwelling only in the Indian and Pacific Oceans. See text for usefulness of, and also issues with, this popularization

of the global circulation, which does not include any Southern Ocean processes. Source: After Broecker (1987).

warm water entering the Atlantic from the

Indian, via the Agulhas, and from the Pacific,

via Drake Passage. This upper ocean water

moves northward through the entire length of

the Atlantic (becoming first lighter and then

denser), and then sinks at several sites in the

northern North Atlantic (Nordic Seas, Labrador

Sea, and Mediterranean Sea). These denser

waters flow south and exit the Atlantic as

NADW. Bottom water (AABW) also enters the

Atlantic from the south. It upwells into the

bottom of the NADW layer in a diffusive

process.

The warm Indian Ocean source water for

NADW includes water from (1) the Pacific via

the ITF, (2) the southeastern Indian Ocean south

of Australia (sourced from the Southern Ocean

and also somewhat from the Pacific), and (3)

upwelling from the underlying deep water layers

within the Indian Ocean. The ITF waters in the

Pacific Ocean originate from upwelling from

deep and intermediate waters within the Pacific

(South Pacific and tropics) and from the upper

ocean in the southeastern Pacific. The upper

ocean source waters from Drake Passage arise

in the southeastern Pacific (SAMW and some

Antarctic Intermediate Water).

Now following the NADW as it leaves the

Atlantic, part enters the Indian Ocean directly

around the southern tip of Africa, contributing

to the IDW. Most enters the ACC, where it

upwells. Here it becomes the source for

deep water formation around Antarctica. This

is the main connection of the NADW and

AABW cells.

The AABW cell begins with this NADW

upwelling to near the sea surface around

Antarctica, where it is subjected to brine rejection

in polynyas (Chapter 13). The densest

waters thus formed sink; the part that escapes

HEAT AND FRESHWATER TRANSPORTS AND OCEAN CIRCULATION 491

northward across topography and into the main

ocean basins is referred to as AABW (although

the densest bottom waters are confined to the

Southern Ocean). This AABW moves northward

at the bottoms of the Atlantic, Indian, and

Pacific Oceans. In all three oceans, AABW

upwells into the local deep water, that is, into

the NADW, IDW, and PDW. Because there are

no volumetrically important surface sources of

dense water in the northern Indian and Pacific

Oceans, this upwelled AABW is the primary

volumetric source of the IDW and PDW,

whereas AABW is only a minor component of

NADW.

The IDW and PDW (which can be traced by

low oxygen because they are composed of old,

upwelled waters) flow southward into the

Southern Ocean above the NADW layer because

they are less dense than NADW. Here, like

NADW, they upwell to the sea surface.

However, they upwell farther north than

NADW because they are less dense. The upwelled

IDW/PDW in the Antarctic feeds two

cells: (1) northward flux of surface water across

the ACC that joins the upper ocean circulation,

accomplished initially by Ekman transport;

and (2) the dense AABW formation, which

then recycles this mass back through the deep

water routes, along with the NADW. The first

of these is a major source of the upper ocean

waters that then feed northward to the NADW

formation region, again connecting the AABW

and NADW cells.

The vertical pathways connecting NADW,

AABW, and also, importantly, IDW and PDW,

are illustrated in Figure 14.11c, which is a

collapsed, two-dimensional version of Figure

14.11a and b. If we tried to sketch the NADW

and AABW cells directly from a global meridional

overturning streamfunction (Figure 14.9a),

they would appear to be completely separate.

This is incorrect as the global average is

missing the important basin-specific roles

of the Indian and Pacific upwelling and diffusive

formation of IDW and PDW, which are

high-volume water masses with large associated

meridional and upwelling transports.

14.3. HEAT AND FRESHWATER

TRANSPORTS AND OCEAN

CIRCULATION

The global circulation redistributes heat and

freshwater within and between the ocean

basins. In Chapter 5, the heat and freshwater

budgets, airesea fluxes, and meridional transports

were described. Here we briefly describe

the components of the circulation that redistribute

heat and freshwater.

Globally averaged, heat is transported meridionally

by the ocean from the tropics to higher

latitudes. The largest heat gains are in the

tropics, with heat gain also in upwelling regions

such as the eastern boundary currents. Individually,

the Pacific and Indian Oceans move heat

poleward. The Atlantic Ocean transports heat

northward throughout its length to balance the

combined Gulf Stream and Nordic Seas heat

loss regions.

The meridional heat transports are mostly

associated with the upper ocean circulations,

which are wind driven. The shallow tropical

cells carry heat from the tropics to the

subtropics. The subtropical gyres then carry

the heat toward the enhanced heat-loss regions

of their western boundary currents. The somewhat

cooled water returns southward, subducted

into the upper part of the subtropical

gyres. This leads to a net poleward heat transport

in all five anticyclonic subtropical gyres

(Figure S14.2 seen on the textbook Web site).

The cyclonic subpolar gyres of the North Pacific

and North Atlantic also transport heat poleward,

with warmer surface inflow in the east

cooling to form the colder, denser waters in

the northern and western parts of both gyres

(NPIW and Labrador Sea Water/NADW).

In the subtropical North and South Pacific

and Indian Oceans, this upper ocean gyre

492


process accounts for almost all of the net poleward

heat transport. However, in the North

Atlantic, the Gulf Stream gyre accounts for

only part of the northward heat transport (about

0.4 PW of 1.2 PW total in the calculation in

Talley, 2003). In the South Atlantic, the net heat

transport is northward (~0.4 PW), toward the

equator, even though the upper ocean gyre

carries heat southward (~0.1 PW). The formation

of NADW, associated with heat loss in the

northern North Atlantic and Nordic Seas,

accounts for the remaining northward heat

transport, due to northward volume transport

of warm upper ocean water and southward

return of the new, cold NADW (Figure 5.12

and Figures S5.9 and S14.3 from the textbook

Web site).

Freshwater is transported by the oceans from

regions of net precipitation and runoff to

regions of net evaporation. The tropical cells

export freshwater poleward from the rainy

Intertropical Convergence Zone toward the net

evaporation centers (Figure 5.4a). On the poleward

side of the evaporation centers, the

subtropical gyres transport freshwater equatorward

(salty water poleward in the western

boundary currents, and freshened subducted

water toward the evaporation centers).

FIGURE 14.11 Global overturning circulation schematics. (a) The NADW and AABW global cells and the NPIW cell.

(b) Overturn from a Southern Ocean perspective. Source: After Gordon (1991), Schmitz (1996b), and Lumpkin and Speer (2007).

(c) Two-dimensional schematic of the interconnected NADW, IDW, PDW, and AABW cells. The schematics do not accurately

depict locations of sinking or the broad geographic scale of upwelling. Colors: surface water (purple), intermediate

and Southern Ocean mode water (red), PDW/IDW/UCDW (orange), NADW (green), AABW (blue). See Figure S14.1 on

the textbook Web site for a complete set of diagrams. This figure can also be found in the color insert. Source: From Talley

(2011).

HEAT AND FRESHWATER TRANSPORTS AND OCEAN CIRCULATION 493

(c)

Southern Ocean

wind-driven upwelling &

surface buoyancy flux

SAMW, AAIW

Low, mid-latitude upper ocean waters

LCDW

UCDW

Pacific-Indian

upwelling &

diffusion

PDW/IDW

Antarctica

AABW

formation

(brine

rejection)

NADW

PDW/IDW

formation

(diffusion)

NADW

formation

(convection)

AABW

FIGURE 14.11

(Continued).

494


Of the deeper overturning cells, only the

NADW and NPIW overturns carry a significant

amount of freshwater equatorward. Both of

these cells consist of saltier poleward flow of

surface water that is freshened and joined

by higher latitude fresh water (Nordic Seas,

Arctic and Bering Strait input for the NADW),

with southward flow of fresher water.

The other three major deep water overturns

in the global circulation d formation of

AABW, IDW, and PDW d have little impact

on either heat or freshwater transport. In the

case of IDW and PDW, this is because they

form by upwelling of AABW and NADW, so

alteration of their properties is due to diapycnal

diffusion, which is a slow, weak means of

change compared with direct airesea fluxes at

surface outcrops. In the case of AABW, even

though there is direct atmospheric forcing, the

heat and freshwater transports are small

because the source water is already cold, so it

can be cooled only slightly more and can only

be freshened by a limited amount and still

remain dense enough to sink. 4

Heat and freshwater are also transported in

the global overturning circulation by the ITF,

moving 10 to 15 Sv from the Pacific to the

Indian Ocean, and by flow through the Bering

Strait, moving less than 1 Sv of low salinity

water (32.5 psu) from the Pacific to the Atlantic.

The ITF loop exports heat and freshwater from

the Pacific because the ITF is warmer and

fresher than the compensating inflow into the

Pacific from the Southern Ocean. In the Indian,

the ITF imports heat and freshwater input

because the ITF is warmer and fresher than

the Agulhas outflow that drains the ITF water.

Bering Strait exports freshwater from the

Pacific to the Atlantic, because the flow through

the strait, at 32.5 psu, is fresher than the

volumetrically compensating inflow from the

Southern Ocean.

14.4. GLOBAL PROPERTY

DISTRIBUTIONS

We return here to the global perspective of

water properties introduced in Chapter 4. We

first describe the global pattern of sea level

height since it is partly related to the temperature/salinity

distribution. We then focus on

global summaries of the water masses that

were introduced in Chapters 9e13, here relating

the property structures to the global circulation

and overturning.

14.4.1. Sea Level

The ocean’s mean surface height distribution

(relative to the global mean surface height) can

be inferred from the global dynamic topography

of Figure 14.2a (Maximenko et al., 2009). Actual

surface height (relative to the geoid) is close to

the dynamic height divided by g ¼ 981 cm s 2

(Eq. 7.28).

The dynamic topography also yields the

surface geostrophic circulation. Using the

global map, we compare the corresponding

large-scale features in each ocean. For instance,

the surface height difference from west to east

across the North Pacific subtropical gyre is

about 70 cm. In contrast, the west-to-east

drop across the North Atlantic subtropical

gyre is about 40 cm. There is a similar contrast

between the South Pacific and South Atlantic

subtropical gyres of about 70 cm versus 40

cm difference. This means that there is more

equatorward volume transport in the Pacific

subtropical gyres than in the Atlantic gyres.

4 Although the upwelled Antarctic surface waters incorporate a large amount of freshwater in the Antarctic, the newly

forming AABW can only be freshened a limited amount and still retain a density that is high enough to allow sinking. The

remaining freshwater stays in the upper ocean and is exported in the Southern Ocean’s upper ocean overturns, contributing

to Antarctic Intermediate Water (Talley, 2008).

GLOBAL PROPERTY DISTRIBUTIONS 495

The simplest reason is that the Sverdrup transport

in the Pacific is proportionally higher than

in the Atlantic because the Pacific is that much

wider and the winds, and hence Ekman pumping,

are similar.

Looking at the global scale, the very low

surface height in the Southern Ocean contrasts

with the rest of the world ocean. The high

gradient between the low Southern Ocean pressure

and high pressure just to its north marks

the eastward geostrophic flow of the ACC,

which is principally wind-driven.

Separate from, and somewhat masked by

these wind-driven gyre differences, a remarkable

global feature is the overall higher surface

height in the Pacific compared with the Atlantic.

This is associated with the relatively lower

density of the Pacific compared with the

Atlantic, which is associated with the lower

mean salinity of the Pacific.

14.4.2. Water Mass Distributions

Water masses in the upper ocean, at intermediate

depth (below the pycnocline), in the deep

ocean (2000e4000 m), and near the bottom are

presented here mostly using schematics; maps

and sections were shown in Chapters 9-13.

Only a subset of the water masses introduced

in previous chapters are included, but these

are representative of most of the processes that

determine the property structures.

The upper ocean water masses are represented

here by mode waters, reviewed in more

detail in Hanawa and Talley (2001; Figure 14.12).

(Unrepresented by this schematic are the

upper ocean water masses associated with

subduction d the Central Water and Subtropical

Underwater of the main pycnocline of

each ocean basin.) All mode waters are associated

with strong fronts, most of which are

80˚N

60˚W 0˚ 60˚E 120˚E

180˚

120˚W

1

60˚

26.9–27.75

40˚

26.5

26.5–26.8

7

25.2

26.2

24–25.425.4

20˚

0˚

–

20˚

25.5

40˚

60˚

27.1

26.2–26. 7

0

26.2–26.3

26.0

26.85

26.0

26.95

25.5

80˚S

FIGURE 14.12 Mode Water distributions, with typical potential densities and schematic subtropical gyre, and ACC

circulations. Source: After Hanawa and Talley (2001). Medium grays are STMWs in each subtropical gyre. Light grays are

eastern STMWs in each subtropical gyre. Dark grays are SPMW (North Atlantic), Central Mode Water (North Pacific), and

SAMW (Southern Ocean).

496


well-known strong currents, such as the Gulf

Stream, Subantarctic Front, and so forth. These

fronts have strongly sloping isopycnals that

favor lower stratification and hence deeper

mixing on the warm side of the front.

Subtropical Mode Waters (STMWs) are associated

with each subtropical western boundary

current. STMWs fill a large portion of the western

subtropical gyres. Each STMW has a temperature

of around 16 to 19 C; the ubiquity of this temperature

is due to the similarity of western boundary

current separation latitudes and the surface

temperature distribution in each subtropical

gyre (Figure 4.1). However, the potential densities

of the STMWs differ greatly because of the

salinity differences between the oceans. The

North Atlantic is the saltiest, so it has the densest

STMW, and so forth.

STMW formation mechanisms include deep

winter mixed layer outcrops close to the strong

fronts, preconditioned by the isopycnal slopes

associated with the fronts, and cross-frontal

transports driven by wind or the dynamics

of the front. Each STMW is subducted into the

interior of its subtropical gyre and becomes

isolated from the sea surface within a few

hundred kilometers of the front.

Eastern STMWs (lighter grays Figure 14.12)

in each ocean basin are less dense than STMWs

in most oceans, and are the least dense class of

mode waters shown in the map. They are found

where the fronts that define the gyres swing

southward, except in the North Atlantic, where

the Azores Current is the relevant front, and

head directly for the Strait of Gibraltar.

Subpolar Mode Water (SPMW) in the North

Atlantic is associated with the northeastward

flow of the NAC and the cyclonic Irminger

and Labrador Sea circulations. Within the

NAC separation region, SPMW functions like

an STMW, and subducts southward into the

subtropical North Atlantic circulation. SPMW

in the northeastern North Atlantic is associated

with the northeastward branches of the NAC

that enter the Norwegian Sea. SPMW in the

northwestern North Atlantic ultimately

becomes the new Labrador Sea Water (LSW)

that sinks and spreads away from the Labrador

Sea (McCartney & Talley, 1982). Formation of

SPMW is like that of STMW: deep winter mixed

layers on the warm side of the strong fronts.

Central Mode Water in the North Pacific is

associated with the eastward flow of the North

Pacific Current’s subarctic front, which is more

or less a continuation of the Oyashio, and lies

north of the Kuroshio front. Again, this front

favors deep mixed layers on its warm side.

SAMW is the series of mode waters along

the northern side of the Subantarctic Front

(McCartney, 1977, 1982). Like STMWs, these

are associated with deep winter mixed layers

within several hundred kilometers or less of

the strong front. The SAMWs subduct northward

into the subtropical gyres where they

become an important part of the pycnocline.

The best-developed (thickest, lowest stratification)

SAMWs are found in the eastern Indian

Ocean and across the Pacific where winter

mixed layers are thickest. After subducting

northward into the pycnocline, these SAMWs

carry tracers associated with large surface ventilation

(high oxygen, chlorofluorocarbons; CFCs)

far into the Indian and South Pacific.

The major intermediate waters of the global

ocean, each identified by a vertical salinity

extremum, are shown in Figure 14.13. The greens

and blues are low salinity intermediate waters

and the oranges are high salinity intermediate

waters. Each of these intermediate waters forms

predominantly in a localized region (locations

marked on the map) and then is advected by

the circulation. Each is associated with the global

overturning circulation, in that formation

involves a conversion of surface waters to densities

that reach to intermediate depths, below the

pycnocline. The overall reach of each intermediate

water is greater than indicated by the location

of its vertical extremum, which is simply an

imperfect marker of the spread of water from

a surface source. For instance, most of the


40˚

60˚

80˚N

Labrador Sea

Water 27.8 σ θ

60˚W 0˚ 60˚E 120˚E 180˚ 120˚W

Mediterranean Water

28.0 σ θ

North Pacific

Intermediate Water

27.0 σ θ

20˚

0˚

Red Sea

Water

27.7 σ θ

20˚

40˚

60˚


Water 27.1 σ θ

80˚S

FIGURE 14.13 Low- and high-salinity intermediate waters. AAIW (dark green), NPIW (light green), LSW (dark blue),

MW (orange in Atlantic), RSW (orange in Indian). Light blue in Pacific: overlap of AAIW and NPIW. Light blue in Indian:

overlap of AAIW and RSW. Cross-hatching: mixing sites that are particularly significant for the water mass. Red dots

indicate the primary formation site of each water mass; fainter dots mark the straits connecting the Mediterranean and Red

Seas to the open ocean. The approximate potential density of formation is listed. This figure can also be found in the color

insert. Source: After Talley (2008).

Okhotsk Sea water that provides the NPIW

salinity minimum in the subtropical North

Pacific resides in the subpolar gyre; however, it

is not a vertical salinity extremum, and is therefore

not as easily identified.

The three major low salinity intermediate

waters d LSW, NPIW, and AAIW d result

from relatively fresh, cold, dense water at

subpolar latitudes that sinks beneath the

warmer, saltier, lighter subtropical waters. The

formation mechanism differs for each of these

water masses: LSW, deep convection and

sinking in the Labrador Sea; NPIW, brine rejection

and sinking in the Okhotsk Sea followed

by strong mixing in the Kuril Island passages;

AAIW, deep mixed layers and underlying fresh

subantarctic water sinking in the Drake Passage

region and subducted smoothly northward into

the Pacific and subducted with large mixing

northward into the Atlantic/Indian.

The temperatures of these three low salinity

intermediate waters are similar: 3 to 5 C. Their

densities differ greatly because the overall

salinity of their respective oceans differs.

NPIW forms in the relatively fresh North Pacific

and is the freshest and least dense of the intermediate

waters, while LSW forms in the salty

North Atlantic and is the saltiest and most

dense.

The high salinity intermediate waters, Mediterranean

Water (MW) and RSW, result from

high evaporation and winter cooling in the Mediterranean/mid-east

region. The resulting deep

convection is strongly localized within the Mediterranean

and Red Seas. The dense waters flow

out over sills through narrow straits to join their

respective ocean circulations, both sinking to

intermediate depth and entraining large amounts

of ambient ocean water as they equilibrate at

depth in the open ocean. Once equilibrated,

498


they are much warmer than the low salinity intermediate

waters, of the order 12 to 15 C, but they

are dense because of their high salinity.

NADW and AABW (or LCDW) are the two

large-scale dense/bottom water masses that

are always included in describing the global

overturning circulation. Both are formed at the

sea surface by buoyancy loss due to airesea

fluxes. The overturn associated with NADW is

estimated at 15 to 20 Sv. The overturning estimates

associated with AABW range from 12 to

more than 25 Sv.

IDW and PDW are not represented in the

maps included here. Both are formed by

upwelling of bottom water (AABW) and admixture

of NADW within their respective oceans.

Downward diffusion from above modifies their

properties relative to their deep source waters.

They have little or no surface source, and therefore

the water mass decompositions shown in

Figures 14.14 and 14.15 focus on NADW and

AABW.

The sources of NADW and AABW are represented

by the map showing the location of

a deep isopycnal in Figure 14.14a. At this density,

these two sources are separated by topography

south of Nova Scotia and Newfoundland

(which thus under-represents the global reach

of NADW as described in the following paragraphs).

The dense source of Nordic Seas

Overflow water is deep convection east of

Greenland. The multiple, distributed sources of

AABW are brine rejection due to sea ice formation

in polynyas within the Weddell and Ross

Seas and at several locations along the Antarctic

coast.

(a) Location of the isopycnal σ 4 = 45.92

80˚N 60˚W 0˚ 60˚E 120˚E 180˚ 120˚W

Greenland Sea Deep Water 28.1 σ θ

40˚

60˚

80˚S

Nordic Seas

Overflow Water

27.9 σ θ 46.1 σ 4

80˚N

20˚

0˚

20˚

40˚

60˚

Antarctic Bottom Water: 27.7-27.9 σ θ 45.9-46.2 σ 4

80˚S

Weddell Sea Bottom Water

Adélie Land Bottom Water

Ross Sea Bottom Water

FIGURE 14.14 Deep and bottom waters. (a) Distribution of waters that are denser than s 4 ¼ 45.92 kg/m 3 . This is

approximately the shallowest isopycnal along which the Nordic Seas dense waters are physically separated from the

Antarctic’s dense waters. At lower densities, both sources are active, but the waters are intermingled. Large dots indicate the

primary formation site of each water mass; fainter dots mark the straits connecting the Nordic Seas to the open ocean. The

approximate potential density of formation is listed. Source: After Talley (1999). (b) Potential temperature ( C), and (c) salinity

at the ocean bottom, for depths greater than 3500 m. Source: After Mantyla and Reid (1983).


FIGURE 14.14

(Continued).

The global maps of bottom potential temperature

and salinity in Figure 14.14b,c show the

contrasting warm, saline Nordic Seas and cold,

fresh AABW properties. The NADW occupies

most of the bottom of the North Atlantic and

eastern South Atlantic, as also seen in the

water mass decomposition considered next

(Figure 14.19). The colder AABW occupies the

Southern Ocean, the western South Atlantic,

and dominates in the Indian and Pacific Oceans.

500


FIGURE 14.15 Fractions of NADW and AABW. (a) Fraction of NADW on the isoneutral surface g N ¼ 28.06 kg/m 3 (s 4 ~

45.84 kg/m 3 , at a depth of 2500e3000 m north of the ACC; G. Johnson, personal communication, 2009). (b) Fraction of

AABW in the bottom water (with the remaining fraction being mostly NADW). Source: From Johnson (2008). Both maps are

from an OMP analysis using as inputs the properties of NADW at a location just south of Greenland, downstream from the

Nordic Seas Overflows, and of AABW in the Weddell Sea. The complete figures are reproduced on the textbook Web site as

Figures S14.4 and S14.5.

EDDY VARIABILITY AND DIFFUSIVITY 501

The effect of vertical mixing is apparent in

these global maps. The western Indian Ocean

has higher bottom salinity than the entering

AABW, which fills the bottommost layer. This

is due to downward mixing from the overlying

higher salinity RSW. In the Pacific Ocean, the

bottom salinity distribution also indicates

downward mixing from overlying waters: in

the southwestern Pacific, a deep vertical salinity

maximum remains from the NADW influence in

the circumpolar deep waters, and this creates

higher salinity at the bottom. Farther north in

the Pacific, the bottom waters are fresher and

warmer, which is partially due to the elimination

of the densest bottom waters through

upwelling (diapycnal mixing) and downward

mixing of overlying fresher water. Full explanation

requires detailed consideration of properties

on deep isopycnal surfaces.

The global reach of NADW is demonstrated

using the fraction of NADW (compared with

AABW) on an isopycnal surface that typifies

the high salinity core in the ACC (Figure 14.15

and Figure S14.4 seen on the textbook Web

site). The maps were computed by G. Johnson

(personal communication, 2009) using his

(Johnson, 2008) application of optimum multiparameter

analysis (OMP) (Section 6.7.3) to

global water masses. Reid and Lynn (1971)

were the first to show and describe the global

pattern of salinity on this NADW isopycnal.

Salinity on nearly the same isopycnal surface

in the Southern Ocean is shown in Figure 13.17.

Water with a large fraction of NADW reaches

southward from its source in the northern North

Atlantic, down along the western side of the

South Atlantic, and then eastward at 20e30 S.

There is a transition to a lower NADW fraction

(i.e., salinity) between 30 to 40 S, which indicates

the onset of much more AABW. A tongue

of higher NADW fraction (i.e., salinity) spreads

eastward south of Africa and then in patches

along the core of the ACC eastward into the

Pacific Ocean. The pattern of somewhat higher

fractions (salinity) then extends northward

into the Pacific Ocean in the DWBC just east of

New Zealand. Along the entire path, the

NADW fraction (higher salinity) decreases

downstream as the AABW fraction increases.

The global reach of AABW is also demonstrated

with OMP analysis applied at the ocean

bottom (Figure 14.15b and S14.5 from the textbook

Web site). The pattern is similar to that

on the deep isopycnal (Figure 14.15a). AABW

dominates the world ocean’s bottom water, but

with a respectable NADW fraction of about

0.3. AABW is mostly blocked from crossing the

mid-Atlantic and Walvis Ridges in the South

Atlantic, so NADW dominates the eastern South

Atlantic as well as the North Atlantic.

In both the deep and bottom water maps,

AABW covers significantly more of the ocean

than NADW. Johnson (2008) estimated that

two-thirds of the deep and bottom water arises

from AABW and one-third from NADW. If the

overturning rates for NADW and AABW of 19

and 28 Sv from Figure 14.5 are used together

with Johnson’s (2008) calculations of the

volumes of NADW and AABW, a residence

time of about 500 years for both water masses

is obtained. If, however, the overturning rates

for the two water masses are more like 17 Sv

each, as summarized in Johnson (2008), the residence

times differ, with NADW around 500

years and AABW around 870 years. Uncertainties

in these values are large. As both water

masses are in similar deep/bottom water

environments, differences in residence times

would rely on differences in how they are

affected by geographically heterogeneous diapycnal

diffusion.

14.5. EDDY VARIABILITY

AND DIFFUSIVITY

This introductory textbook is mostly written

from the point of view of a mean circulation

and simple departures from the mean, including

some seasonal and climate variability, and

502


energetic, recurring time-dependent features

such as Gulf Stream or Agulhas rings. However,

all regions of the ocean have some level of eddy

variability, which is defined as the departure of

instantaneous velocities or sea surface/isopycnal

heights from the mean. Variability can range

from random noise, to wavelike disturbances,

to closed, coherent features. Eddy variability is

responsible for stirring in the nearly horizontal

(along-isopycnal) direction, and is thus critically

important to along-isopycnal eddy diffusivity

(Sections 7.2.4 and 7.3.2).

Warm- and cold-core rings generated by the

meandering of major currents are large, closed

features that one might typically associate with

“eddies.” On the other hand, eddy variability in

the central parts of the gyres may look more like

spectral noise. Some kinds of eddies extend from

the sea surface to the bottom, while others are

concentrated in the surface layer, and others can

be embedded entirely within subsurface layers.

Horizontal, eddy length scales are kilometers

to thousands of kilometers. Timescales are

typically weeks to months, but can sometimes

be many years for coherent vortices such as

Meddies (Chapter 9). This is considered to be

the ocean’s mesoscale. These are the length and

timescales associated with planetary waves

such as Rossby and Kelvin waves. The most

important length scale for this variability is the

Rossby deformation radius, which depends on

latitude and vertical stratification (Section 7.7.4,

Figure S7.30 on the textbook Web site). A new

category of shorter submesoscale variability,

mostly associated with the surface layer, is

now being vigorously analyzed through theory,

modeling, and observations. Because this layer

is so shallow, the horizontal spatial scales are

small, on the order of kilometers. (This can be

thought of as related to an internal Rossby

deformation radius using a vertical length scale

of about 100 m rather than 1000 m.)

Internal waves and tides (Sections 7.5.1, 8.4,

and 8.6) have even shorter timescales. This

high frequency variability is critically important

for ocean turbulence and hence diapycnal diffusivity

and mixing (Sections 7.3 and 7.4). We

therefore present some recent global results for

near-inertial and tidal variability.

14.5.1. Eddy Energy and Lateral Eddy

Diffusivity Distributions

The basic physics concepts of kinetic and

potential energy were described in terms relevant

to the ocean in Section 7.7.5. Eddy kinetic

energy (EKE) is calculated using departures of

the instantaneous (synoptic) velocity from the

mean velocity, regardless of how the mean is

defined (leaving some ambiguity that should

be carefully described in any given study).

EKE maps are almost always based on lateral

currents and not on the vertical velocities, which

are much smaller (but important for the diapycnal

eddy diffusivity described in the following

section). EKE for surface currents was first

derived from ship drift observations, but is

now much more easily constructed from surface

drifter velocities and from surface velocities

derived from altimetric surface heights. EKE

maps for deeper levels are calculated from

Lagrangian float observations; moored current

meter arrays are also used to calculated eddy

energy locally. Eddy potential energy is calculated

using departures of instantaneous sea

surface height and isopycnal heights from their

mean values; currently satellite altimetry data

are valuable for this, and in situ Argo profiling

float data set will also be valuable after many

more years of data are collected.

Surface EKE (Figure 14.16 and Figure S14.6

seen on the textbook Web site) is mostly related

to the mean current speeds (Figure 14.2b,

Section 14.1). Eddy energy has several sources,

including current instabilities (Section 7.7.5).

Mean flows with strong velocity shear in both

the horizontal and vertical tend to be the most

unstable. Horizontal shear generates barotropic

instabilities that draw energy from the shear.

Vertical shear in geostrophic flow is associated


FIGURE 14.16 Eddy kinetic energy (cm 2 s 2 ) from surface drifters. Source: From NOAA AOML PHOD (2009).

A complementary figure based on satellite altimetry (from Ducet, Le Traon, & Reverdin, 2000) is reproduced in Figure S14.6c on

the textbook Web site. This figure can also be found in the color insert.

with sloped isopycnals; the eddy energy is

generated through release of the potential

energy of the sloping isopycnals through baroclinic

instability. Overall, satellite altimetry analysis

has reinforced the importance of flow

instabilities, and especially of baroclinic instability,

in all regions (Stammer, 1998).

On the other hand, mid-ocean eddies away

from strong currents can be generated by mechanisms

other than flow instability, that is,

through direct wind forcing. An easily visible

example in Figure 14.16 is the high EKE band

just west of Central America; these are eddies

in the Gulf of Tehuantepec, forced by very

strong winds through the adjacent mountain

passes (Figures 5.16 and 10.21).

Although it is mostly related to the currents,

the EKE distribution differs in important ways

from the speed distribution (Figure 14.2b and

Figure S14.6 seen on the textbook Web site).

The strongest eddies, such as Agulhas rings,

propagate away from the mean flow that

created them, accounting for broader EKE

maxima compared with the speed (mean kinetic

energy) maxima. The most striking large-scale

difference of EKE from the mean speed distribution

is the high EKE in bands around 20 to 30

latitude, especially in the Pacific and Indian

Oceans, but also with a signature in the Atlantic.

These regions have low mean surface velocities,

yet the enhanced EKE stands out starkly in the

global mean. These regions have shallow eastward

surface flow with underlying westward

flow (the Subtropical Countercurrents in the

Pacific, Eastern Gyral Current in the Indian,

and the Azores Current and Subtropical Countercurrents

in both hemispheres in the Atlantic).

This vertical shear is associated with tilted

isopycnals and enhanced baroclinic instability

(e.g., Palastanga, van Leeuwen, Schouten, &

deRuijter, 2007; Qiu, Scott, & Chen, 2008).

Horizontal eddy diffusivity can be calculated

from the eddy variability measured by Lagrangian

drifters. The highest values of surface

eddy diffusivity might exceed 2 10 4 m 2 /sec

(2 10 8 cm 2 /sec) (Figure 14.17a and Figure

S14.7 on the textbook Web site). The pattern of

surface diffusivity corresponds roughly to the

504


FIGURE 14.17 (a) Horizontal eddy diffusivity (m 2 /sec) at the sea surface (color) with mean velocity vectors, based on

surface drifter observations. Source: From Zhurbas and Oh (2004). (b) Eddy diffusivity ellipses at 900 m based on subsurface

float velocities. Colors indicate different scales (see figure headers). Source: From Davis (2005). The Atlantic surface map and

Indian 900 m map from the same sources are reproduced in Chapter S14 (Figures S14.7 and S14.8) on the textbook Web site.

Both Figures 14.7a and 14.7b can also be found in the color insert.


EKE pattern. The eddy diffusivity is not exactly

proportional to EKE because timescales for

diffusion depend on location and on the underlying

processes that create the variability

(Lumpkin, Treguier, & Speer, 2002; Shuckburgh,

Jones, Marshall, & Hill, 2009). The map in

Figure 14.17a shows a scalar diffusivity, calculated

using a modified version of Davis’ (1991)

method; the full horizontal diffusivity is a tensor,

hence can have different values in the zonal and

meridional directions, since velocity variations

can be in any direction relative to the mean

velocity.

Below the sea surface, out of reach of satellites

and surface drifters, eddy statistics are

more difficult to compile. 5 Horizontal eddy

diffusivities based on subsurface floats at

900 m (Figure 14.17b) have maximum values

of about 0.8 10 4 m 2 /sec, which are robustly

less than those at the sea surface despite differences

in computation methods, including the

use of ellipses that show the directional difference

in diffusivity. As at the sea surface, the

900 m eddy diffusivity is high mostly where

currents are strong. It is also mostly isotropic

(the ellipses are “round”) except in the tropics,

where it is highly directional, with much larger

values in the east-west direction, matching the

strongly zonal direction of the tropical currents

(Davis, 2005).

14.5.2. Observed Scales, Speeds,

and Coherence of Eddy Variability

This is a very brief introduction to the large

amount of work describing the ocean’s eddy

variability. A simple time-space display

(Hovmöller diagram) of sea-surface height

(SSH) anomalies at mid-latitude in each ocean

(Figure 14.18) reveals generally westward propagation

of the dominant features, which is

typical behavior for Rossby waves (Section

7.7.3). Similar patterns are found at almost all

latitudes, except near the equator where eastward-propagating

Kelvin waves are also found,

and in strong eastward flows such as the ACC

that advect the variability to the east (a Doppler

shift).

Phase speeds calculated from SSH plots

like those of Figure 14.18 yield robustly westward

propagation, close to the speeds of firstmode

baroclinic Rossby waves (Figure 14.19;

Chelton & Schlax, 1996; Stammer, 1997). However,

the difference from simple Rossby wave speeds

is important: the observed speeds are almost

twice as fast at mid-latitudes. The non-Rossbywavelike

behavior of the variability is likely

due to nonlinear interactions with other modes

(e.g., Wunsch, 2009), and to the prevalence of

coherent eddies that propagate westward (Figure

14.21). Such coherent eddies are nonlinear by

definition.

Frequency and wavenumber spectra (Section

6.5.3) provide statistical information about variability

observed from satellites, current meters,

and so forth. The directional wavenumber spectrum

from satellite altimetry in Figure 14.20b

again shows that most energy propagates westward

(solid curve) rather than eastward

(dashed). In the frequency spectrum, the annual

cycle is the most energetic signal (peak indicated

by dashed vertical line in the left panel),

because this is the strongest external forcing

frequency, associated with seasonal changes.

Other than this peak, the frequency spectrum

is relatively smooth.

The spectra in Figure 14.20 are nearly flat at

lower frequencies and wavenumbers (longer

5 The best EKE estimates are made at long-term current meter moorings, which tend to be deployed in dynamically

interesting regions such as the Gulf Stream, with little sampling elsewhere. Subsurface floats provide information at their

target depths. Acoustically tracked floats provide the best statistics since their locations are observed nearly continuously,

but they are not global. Profiling floats that are tracked when they surface, approximately every 10 days, can provide

global statistics, although global maps are not yet available.

506


FIGURE 14.18

Surface-height anomalies

at 24 degrees latitude

in each ocean, from

a satellite altimeter. This

figure can also be found

in the color insert.

Source: From Fu and

Chelton (2001).


FIGURE 14.19 (a) Westward phase speeds (cm/sec) in

the Pacific Ocean, calculated from the visually mostdominant

SSH anomalies from satellite altimetry. The

underlying curves are the fastest first-mode baroclinic

Rossby waves speeds at each latitude. (b) The ratio of

observed and theoretical phase speeds, showing that the

observed phase speeds are generally faster than theorized.

Source: From Chelton and Schlax (1996).

time and space scales), and steeply sloped at

higher frequencies and wavenumbers. (The flat

parts are called “white” because white noise

includes all frequencies with roughly the same

energy; the sloped parts are called “red” because

they slope up to higher energy at lower

frequency.) The spectra transition from flat to

sloped at a relatively well-defined frequency

and wavenumber; this can be called a “cutoff”

frequency or wavenumber. The cutoff marks

a shift in the physical processes that dominate

the white versus the red parts of the

spectra. The spectral slopes are consistent

with generation of energy through baroclinic

instability (Section 7.7.5) rather than external

changes in forcing (mainly wind; Stammer,

1997).

One explanation for the lack of detailed correspondence

between observations of westward

propagation and the obvious initial explanation

of Rossby waves is that a great deal of energy

is actually contained in coherent eddies, which

differ from Rossby wave behavior in many

ways, but retain the westward propagation of

Rossby waves. In each ocean basin chapter

(9e13), we described some of the major eddy

(ring) formation and propagation locations.

These phenomena often have names (“Halmahera

eddy,” “Gulf Stream rings,” “Agulhas

rings,” “Queen Charlotte Eddy,” “Brazil

Current Rings,” etc.) because they occur so

frequently in a given location and so greatly

dominate variability and often transport of

properties near those locations. There have

been indications of more widespread coherent

eddies in Lagrangian data sets (e.g. Shoosmith,

Richardson, Bower, & Rossby, 2005).

Coherent vortices (eddies) have now been

shown, from satellite altimetric data, to exist

in large regions of the oceans (Figure 14.21;

Chelton, Schlax, Samelson, & de Szoeke, 2007).

These maps of vortices resemble, to some extent,

the EKE maps of Figure 14.16. The major bands

of eddies are in the western boundary currents/

extensions, in large bands at mid-latitudes

where flow is eastward (subtropical countercurrents,

Azores Current), and in the ACC.

The eddies mostly propagate westward; in the

eastward flow of the ACC, they are advected

eastward. These eddies are strongest (highest

SSH) in the western boundary current extensions,

but their populations are highest in the

ACC and mid-latitude bands (subtropical

508


FIGURE 14.20 (a) Frequency and (b) wavenumber spectra of SSH in the eastern subtropical North Pacific, using 15 years

of satellite altimetry observations. The dashed line in (a) is the annual frequency. In the wavenumber panel, solid is

westward propagating, and dashed is eastward propagating energy. Source: From Wunsch (2009).

countercurrents). The eddy diameters are generally

larger than the Rossby deformation radius

(Figure S7.30 seen on the textbook Web site),

exceeding 200 km in the regions most dominated

by eddies, and dropping off to about

100 km at higher latitudes (Chelton et al., 2007).

14.5.3. Diapycnal Diffusion and

Near-Inertial Motion

In the ocean’s overturning circulation, density

is altered along the circulation path (Section

7.10). Surface water becomes dense enough to

sink to great depth mostly in well-defined small

regions. The dense waters eventually return to

lower density, due to downward diapycnal

eddy diffusion of buoyancy (Figure S7.40 on

the textbook Web site). In the ocean’s interior,

this is a weak and slow process, associated

with turbulence generated by internal wave

breaking (Section 7.3.2). Based on the observed

ocean stratification and simple models, the

globally averaged diapycnal eddy diffusivity is

about 10 4 m 2 /sec (Munk, 1966; see Sections

7.3.2 and 7.10.2). Within the main pycnocline

the diapycnal eddy diffusivity is much smaller,

of the order 10 5 m 2 /sec (e.g., Gregg, 1987;

Ledwell, Watson, & Law, 1993). On the other

hand, diapycnal diffusivity as observed near

the ocean bottom, and up into the water column

over regions of very rough topography, is higher

than the Munk value (Figure 7.2).

If isopycnals are raised up to near the sea

surface through mechanical forcing (e.g., Ekman

suction due to wind stress curl) or due to

sloping associated with strongly sheared

geostrophic currents, then much higher

diffusivities are available to waters on those

isopycnals because of the much higher levels

of near-surface turbulence due to wind and

airesea flux forcing. This uplift occurs in

shallow cells in the tropics and along the eastern

boundaries, where upwelling is strong, and also

in the Southern Ocean and subpolar North

Pacific, where open-ocean isopycnals rise up to

the sea surface.


FIGURE 14.21 Tracks of coherent cyclonic and anticyclonic eddies with lifetimes of more than 4 weeks, based on altimetric

SSH, color coded by a “nonlinearity parameter,” which is the ratio of velocity within the eddy compared with the

eddy propagation speed. White areas indicate no eddies or trajectories within 10 degrees latitude of the equator. This figure

can also be found in the color insert. Source: From Chelton et al. (2007).

510


Wind-forced near-inertial motion (Figure 7.4)

in the surface layer is expected to result in

higher diapycnal diffusivities there; the geographical

distribution of this motion should be

an indication of geographic variations in surface

layer diffusivity. The near-inertial motion has

been mapped globally from the drifter data set

(Figure 14.22a). Mean speeds of the near-inertial

motions are 10 cm/sec, ranging up to much

higher than 20 cm/sec beneath the atmosphere’s

mid-latitude storm tracks and in the

eastern tropical Pacific and western tropical

Atlantic. Observed inertial current radii are

10e30 km (Chaigneau, Pizarro, & Rojas, 2008).

Energy spectra from surface drifter velocities

show the prevalence of inertial energy at all

latitudes (Figure 14.22b). This is important

because it demonstrates that the wind energy

is indeed concentrating in the inertial band,

and therefore energy for the surface layer’s

turbulent mixing is largely near-inertial. Thus

the map of inertial energy in Figure 14.22a

reflects the spatially varying capability of the

surface ocean to mix. The inertial frequency

depends on latitude, going from 0 at the

equator to almost 2 cycles per day at 70 latitude

(solid curve in the figure). There is also

energy at low frequencies at all latitudes, which

is largely due to geostrophic motion, and

energy at the semidiurnal period (2 cycles per

day; vertical yellow bands in the figure) mainly

due to tides.

14.6. CLIMATE AND THE

GLOBAL OCEAN

In present usage, climate variability refers to

natural climate variability and climate change

refers to anthropogenically forced variations in

climate. The latter is also referred to as “global

change.” We include climate in an oceanography

text not necessarily because of oceanatmosphere

feedbacks, which might be weak

in all but the tropical modes, but because

FIGURE 14.22 Near-inertial motion. (a) Average inertial

current speeds (cm/sec), based on surface drifters. Source:

From Chaigneau et al. (2008). (b) Rotary power spectra in

2.5 degree latitude bins in the Pacific Ocean. The solid curve

is the inertial frequency at each latitude; the dashed curve is

twice the inertial frequency. Negative frequencies rotate

counterclockwise and positive frequencies rotate clockwise.

Source: From Elipot and Lumpkin (2008). Both Figures 14.22a

and 14.22b can also be found in the color insert.

climate variability and change affect ocean variability

in properties and circulation.


relating to climate variability have been moved

to Chapter S15 (Climate Variability and the

CLIMATE AND THE GLOBAL OCEAN 511

Oceans) on the supplemental Web site for the textbook,

which also includes climate variability

materials from each of the basin chapters. In the

supplement, we present figures and a table of

the modes of the interannual, decadal, and longer

term climate variability that appear to have the

most imprint on the ocean. We also discuss

changesin oceanproperties (temperature,salinity,

oxygen) and to some extent circulation, as they

relate to climate variability and climate change.

References

Aagaard, K., Coachman, L.K., Carmack, E., 1981. On the

halocline of the Arctic Ocean. Deep-Sea Res. 28,

529e545.

Aagaard, K., Greisman, P., 1975. Toward new mass and heat

budgets for the Arctic Ocean. J. Geophys. Res. 80,

3821e3827.

Aagaard, K., Swift, J.H., Carmack, E.C., 1985. Thermohaline

circulation in the arctic mediterranean seas. J. Geophys.

Res. 90, 4833e4846.

Ambar, I., Fiuza, A.F.G., 1994. Some features of the Portugal

Current System: A poleward slope undercurrent, an

upwelling-related summer southward flow and an

autumn-winter poleward coastal surface current. In:

Katsaros, K.B., Fiuza, A.F.G., Ambar, I. (Eds.), Proceedings

of the Second International Conference on Air-Sea

Interaction and on Meteorology and Oceanography of

the Coastal Zone. American Meteorological Society,

pp. 286e287.

Andrié, C., Ternon, J.-F., Bourlès, B., Gouriou, Y., Oudot, C.,

1999. Tracer distributions and deep circulation in the

western tropical Atlantic during CITHER 1 and ETAM-

BOT cruises, 1993e1996. J. Geophys. Res. 104,

21195e21215.

Ångström, A., 1920. Applications of heat radiation

measurements to the problems of the evaporation from

lakes and the heat convection at their surfaces. Geogr.

Ann. 2, 237e252.

Annamalai, H., Xie, S.P., McCreary, J.P., Murtugudde, R.,

2005. Impact of Indian Ocean sea surface temperature on

developing El Niño. J. Climat. 18, 302e319.

Anselme, B., 1998. Sea ice fields and atmospheric

phenomena in Eurasiatic arctic seas as seen from the

NOAA-12 satellite. Int. J. Rem. Sensing 19, 307e316.

Antonov, J.I., Locarnini, R.A., Boyer, T.P., Mishonov, A.V.,

Garcia, H.E., 2006. World Ocean Atlas 2005, vol 2:

Salinity. In: Levitus, S. (Ed.), NOAA Atlas NESDIS 62.

U.S. Government Printing Office, Washington, D.C.,

182 pp.

Aoki, S., Bindoff, N.L., Church, J.A., 2005. Interdecadal

water mass changes in the Southern Ocean between 30E

and 160E. Geophys. Res. Lett. 32, doi:10.1029/

2004GL022220.

Arhan, M., Naveira Garabato, A., Heywood, K.J.,

Stevens, D.P., 2002. The Antarctic Circumpolar Current

between the Falkland Islands and South Georgia.

J. Phys. Oceanogr. 32, 1914e1931.

Armi, L., 1978. Some evidence for boundary mixing in the

deep ocean. J. Geophys. Res. 83, 1971e1979.

Armi, L., Farmer, D.M., 1988. The flow of Mediterranean

Water through the Strait of Gibraltar. Progr. Oceanogr.

21, 1e105 (also Farmer and Armi, 1988).

Assaf, G., Gerard, R., Gordon, A., 1971. Some mechanisms

of oceanic mixing revealed in aerial photographs.

J. Geophys. Res. 76, 6550e6572.

Bailey, W.D., 1957. Oceanographic features of the Canadian

Archipelago. J. Fish. Res. Bd. Can. 14, 731e769.

Bainbridge, A.E., Broecker, W.S., Spencer, D.W., Craig, H.,

Weiss, R.F., Ostlund, H.G., 1981e1987. The Geochemical

Ocean Sections Study (7 volumes). National Science

Foundation, Washington, D.C.

Bakun, A., 1973. Coastal upwelling indices, west coast of

North America, 1946e71. U.S. Dept. of Commerce,

NOAA Tech. Rep. NMFS SSRF-671, 103 pp.

Bakun, A., Nelson, C.S., 1991. The seasonal cycle of windstress

curl in subtropical eastern boundary current

regions. J. Phys. Oceanogr. 21, 1815e1834.

Balmforth, N.J., Llewellyn-Smith, S., Hendershott, M.,

Garrett, C., 2005. Geophysical Fluid Dynamics/WHOI

2004 Program of Study: Tides. WHOI Technical Report,

WHOI-2005-08, 327 pp. http://hdl.handle.net/1912/98

(accessed 10.01.09).

Bane, J.M., 1994. The Gulf Stream System: An observational

perspective. Chapter 6. In: S.K., Majumdar, S.K.,

Miller, E.W., Forbes, G.S., Schmalz, R.F., Panah, A.A.

(Eds.), The Oceans: Physical-Chemical Dynamics and

Human Impact. The Pennsylvania Academy of Science,

pp. 99e107.

Barber, D.G., Massom, R.A., 2007. Chapter 1: The role of sea

ice in Arctic and Antarctic polynyas. In: Polynyas:

Windows to the World, Elsevier Oceanography Ser, vol.

74. Elsevier, Amsterdam, pp. 1e54.

Baringer, M., Larsen, J., 2001. Sixteen years of Florida

Current transport at 27 N. Geophys. Res. Lett. 28,

3179e3182.

513

514

REFERENCES

Barnett, T.P., Pierce, D.W., AchutaRao, K., Gleckler, P.,

Santer, B., Gregory, J., et al., 2005. Penetration of humaninduced

warming into the World’s Oceans. Science 309,

284e287.

Barnston, A.G., Livezey, R.E., 1987. Classification, seasonality

and persistence of low-frequency atmospheric

circulation patterns. Mon. Weather Rev. 115,

1083e1126.

Beal, L.M., Chereskin, T.K., Lenn, Y.D., Elipot, S., 2006. The

sources and mixing characteristics of the Agulhas

Current. J. Phys. Oceanogr. 36, 2060e2074.

Beal, L.M., Ffield, A., Gordon, A.L., 2000. Spreading of Red

Sea overflow waters in the Indian Ocean. J. Geophys.

Res. 105, 8549e8564.

Beal, L.M., Hummon, J.M., Williams, E., Brown, O.B.,

Baringer, W., Kearns, E.J., 2008. Five years of Florida

Current structure and transport from the Royal Caribbean

Cruise Ship Explorer of the Seas. J. Geophys. Res.

113, C06001. doi:10.1029/2007JC004154.

Beardsley, R.C., Boicourt, W.C., 1981. On estuarine and

continental-shelf circulation in the Middle Atlantic

Bight. In: Warren, B.A., Wunsch, C. (Eds.), Evolution of

Physical Oceanography. MIT Press, Cambridge, MA,

pp. 198e223.

Becker, J.J., Sandwell, D.T., 2008. Global estimates of seafloor

slope from single-beam ship soundings. J. Geophys. Res.

113, C05028. doi:10.1029/2006JC003879.

Becker, J.J., Sandwell, D.T., Smith, W.H.F., Braud, J.,

Binder, B., Depner, J., et al., 2009. Global bathymetry and

elevation data at 30 arc seconds resolution:

SRTM30_PLUS. Mar. Geod. 32, 4355e4371.

Belkin, I.M., 2004. Propagation of the “Great Salinity

Anomaly” of the 1990s around the northern North

Atlantic. Geophys. Res. Lett. 31, L08306. doi:10.1029/

2003GL019334.

Belkin, I.M., Gordon, A.L., 1996. Southern Ocean fronts

from the Greenwich meridian to Tasmania. J. Geophys.

Res. 101, 3675e3696.

Bendat, J.S., Piersol, A.G., 1986. Random Data: Analysis and

Measurement Procedures, second ed. Wiley, New York,

566 pp.

Bennett, A.F., 1976. Poleward heat fluxes in southern

hemisphere oceans. J. Phys. Oceanogr. 8, 785e789.

Bernard, E.N., Robinson, A.R. (Eds.), 2009. The Sea, vol. 15:

Tsunamis. Harvard University Press, Cambridge, MA,

462 pp.

Bevington, P.R., Robinson, D.K., 2003. Data Reduction and

Error Analysis for the Physical Sciences. McGraw Hill,

Dubuque, IA, 320 pp.

Biastoch, A., Böning, C.W., Lutjeharms, J.R.E., 2008.

Agulhas leakage dynamics affects decadal variability in

the Atlantic overturning circulation. Nature 456,

489e492.

Bindoff, N.L., McDougall, T.J., 2000. Decadal changes along

an Indian Ocean section at 32 degrees S and their

interpretation. J. Phys. Oceanogr. 30, 1207e1222.

Bindoff, N.L., Willebrand, J., Artale, V., Cazenave, A.,

Gregory, J., Gulev, S., et al., 2007. Observations: Oceanic

climate change and sea level. In: Solomon, S., Qin, D.,

Manning, M., Chen, Z., Marquis, M., Averyt, K.B. (Eds.),

Climate Change 2007: The Physical Science Basis.

Contribution of Working Group I to the Fourth Assessment

Report of the Intergovernmental Panel on Climate

Change. Cambridge University Press, Cambridge, UK

and New York.

Bingham, F.M., Lukas, R., 1994. The southward intrusion of

North Pacific Intermediate Water along the Mindanao

Coast. J. Phys. Oceanogr. 24, 141e154.

Bingham, F.M., Talley, L.D., 1991. Estimates of Kuroshio

transport using an inverse method. Deep-Sea Res. 38

(Suppl.), S21eS43.

Bishop, J.K.B., 1999. Transmissometer measurement of POC.

Deep-Sea Res. I 46, 353e369.

Bitz, C.M., Fyfe, J.C., Flato, G.M., 2002. Sea ice response

to wind forcing from AMIP models. J. Clim. 15,

522e536.

Bjerknes, J., 1969. Atmospheric teleconnections from the

equatorial Pacific. Mon. Weather Rev. 97, 163e172.

Bjerknes, V., Bjerknes, J., Solberg, H., Bergeron, T., 1933.

Physikalische Hydrodynamik mit Anwendung auf die

Dynamische Meterologie, 3. Springer, Berlin.

Björk, G., Jakobsson, M., Rudels, B., Swift, J.H.,

Anderson, L., Darby, D.A., et al., 2007. Bathymetry and

deep-water exchange across the central Lomonosov

Ridge at 88e89 N. Deep-Sea Res. I 54, 1197e1208.

Boccaletti, G., Ferrari, R., Fox-Kemper, B., 2007. Mixed layer

instabilities and restratification. J. Phys. Oceanogr. 37,

2228e2250.

Boebel, O., Davis, R.E., Ollitrault, M., Peterson, R.G.,

Richardson, P.L., Schmid, C., Zenk, W., 1999. The

intermediate depth circulation of the western South

Atlantic. Geophys. Res. Lett. 26, 3329e3332.

doi:10.1029/1999GL002355.

Böhm, E., Morrison, J.M., Manghnani, V., Kim, H.-S.,

Flagg, C.N., 1999. The Ras al Hadd Jet: Remotely sensed

and acoustic Doppler current profiler observations in

1994e1995. Deep-Sea Res. II 46, 1531e1549.

Boland, F.M., Church, J.A., 1981. The East Australian

Current 1978. Deep-Sea Res. 28A, 937e957.

Botnikov, V.N., 1963. Geographical position of the Antarctic

Convergence zone in the Southern Ocean. Sov. Antarct.

Exped. Inf. Bull., Engl. Transl. 4 (41), 324e327.

Bourlès, B., Lumpkin, R., McPhaden, M.J., Hernandez, F.,

Nobre, P., Campos, E., et al., 2008. The Pirata program:

History, accomplishments, and future directions. B. Am.

Meteorol. Soc. 89, 1111e1125.

REFERENCES 515

Bowden, K.F., 1983. Physical Oceanography of Coastal

Waters. Ellis Horwood Series in Marine Science. Halsted

Press, New York, 302 pp.

Bower, A.S., Armi, L., Ambar, I., 1997. Lagrangian observations

of Meddy formation during a Mediterranean

undercurrent seeding experiment. J. Phys. Oceanogr. 27,

2545e2575.

Bower, A.S., Hunt, H.D., Price, J.F., 2000. Character and

dynamics of the Red Sea and Persian Gulf outflows.

J. Geophys. Res. 105, 6387e6414.

Bower, A.S., Johns, W.E., Fratantoni, D.M., Peters, H., 2005.

Equilibration and circulation of Red Sea Outflow Water

in the western Gulf of Aden. J. Phys. Oceanogr. 35,

1963e1985.

Bower, A.S., LeCann, B., Rossby, T., Zenk, W., Gould, J.,

Speer, K., et al., 2002. Directly measured mid-depth

circulation in the northeastern North Atlantic Ocean.

Nature 410, 603e607.

Bower, A.S., Lozier, M.S., Gary, S.F., Böning, C.W., 2009.

Interior pathways of the North Atlantic meridional

overturning circulation. Nature 459, 243e247.

Boyer, T.P., Antonov, J.I., Levitus, S., Locarnini, R., 2005.

Linear trends of salinity for the world ocean: 1955e1998.

Geophys. Res. Lett. 32, L01604. doi:1029/2004GL021791.

Brambilla, E., Talley, L.D., 2008. Subpolar Mode Water in the

northeastern Atlantic: 1. Averaged properties and mean

circulation. J. Geophys. Res. 113, C04025. doi_10.1029/

2006JC004062.

Bray, N., Ochoa, J., Kinder, T., 1995. The role of the interface

in exchange through the Strait of Gibraltar. J. Geophys.

Res. 100, 10755e10776.

Bretherton, F.P., Davis, R.E., Fandry, C.B., 1976. A technique for

objective analysis and design of oceanographic experiments

applied to MODE-73. Deep-Sea Res. 23, 559e582.

Brink, K.H., 1991. Coastal-trapped waves and wind-driven

currents over the continental shelf. Annu. Rev. Fluid

Mech. 23, 389e412.

Brink, K.H., 2005. Coastal physical processes overview. In:

Robinson, A.F., Brink, K.H. (Eds.), The Sea, vol. 13: The

Global Coastal Ocean: Multiscale Interdisciplinary

Processes. Harvard University Press, Cambridge, MA,

pp. 37e60.

Brink, K.H., Robinson, A.R. (Eds.), 1998. The Sea, vol. 10:

The Global Coastal Ocean: Processes and Methods.

Harvard University Press, Cambridge, MA, 628 pp.

Broecker, W.S., 1974. “NO,” a conservative water-mass

tracer. Earth Planet. Sci. Lett. 23, 100e107.

Broecker, W.S., 1987. The biggest chill. Nat. Hist 97, 74e82.

Broecker, W.S., 1991. The great ocean conveyor. Oceanography

4, 79e89.

Broecker, W.S., 1998. Paleocean circulation during the last

deglaciation: A bipolar seesaw? Paleoceanography 13,

119e121.

Broecker, W.S., Clark, E., Hajdas, I., Bonani, G., 2004. Glacial

ventilation rates for the deep Pacific Ocean. Paleoceanography

19, PA2002. doi: 10.1029/2003PA000974.

Broecker, W.S., Peng, T., 1982. Tracers in the Sea. Lamont-

Doherty Geological Observatory. Columbia University,

690 pp.

Bryan, K., 1963. A numerical investigation of a nonlinear

model of a wind-driven ocean. J. Atm. Sci. 20, 594e606.

Bryden, H.L., Beal, L.M., 2001. Role of the Agulhas Current

in Indian Ocean circulation and associated heat and

freshwater fluxes. Deep-Sea Res. I 48, 1821e1845.

Bryden, H.L., Candela, J., Kinder, T.H., 1994. Exchange

through the Strait of Gibraltar. Progr. Oceanogr. 33,

201e248.

Bryden, H.L., Griffiths, M.J., Lavin, A.M., Millard, R.C.,

Parrilla, G., Smethie, W.M., 1996. Decadal changes in

water mass characteristics at 24 N in the subtropical

North Atlantic Ocean. J. Clim. 9, 3162e3186.

Bryden, H.L., Imawaki, S., 2001. Ocean heat transport. In:

Siedler, G., Church, J. (Eds.), Ocean Circulation and

Climate, International Geophysics Series. Academic

Press, San Diego, CA, pp. 455e474.

Bryden, H.L., Johns, W.E., Saunders, P.M., 2005a. Deep

western boundary current east of Abaco: Mean structure

and transport. J. Mar. Res. 63, 35e57.

Bryden, H.L., Longworth, H.R., Cunningham, S.A., 2005b.

Slowing of the Atlantic meridional overturning circulation

at 26.5 N. Nature 438, 655e657.

Bryden, H.L., Pillsbury, R.D., 1977. Variability of deep flow

in the Drake Passage from year-long current measurements.


Bye, J.A.T., 1972. Ocean circulation south of Australia. In:

Hayes, D.E. (Ed.), Antarctic Oceanology I, The Australian-New

Zealand Sector, Antarctic Research Series, 19.

AGU, Washington, D.C, pp. 95e100.

CalCOFI ADCP, 2008. CalCOFI ADCP. Scripps Institution of

Oceanography. http://adcp.ucsd.edu/calcofi/ (accessed

10.18.18).

Cameron, W.M., Pritchard, D.W., 1963. Estuaries. In:

Hill, M.N. (Ed.), The Sea, vol. 2: Ideas and Observations.

Wiley-Interscience, New York, pp. 306e324.

Candela, J., Tanahara, S., Crepon, M., Barnier, B., Sheinbaum, J.,

2003. Yucatan Channel flow: Observations versus

CLIPPER ATL6 and MERCATOR PAM models. J. Geophys.

Res. 108, 3385. doi:10.1029/2003JC001961.

Candela, J., 2001. Mediterranean Water and global circulation.

In: Siedler, G., Church, J. (Eds.), Ocean Circulation

and Climate, International Geophysics Series. Academic


Cane, M.A., Münnich, M., Zebiak, S.F., 1990. A study of selfexcited

oscillations of the tropical ocean-atmosphere

system. Part I: Linear analysis. J. Atmos. Sci. 47,

1562e1577.

516

REFERENCES

Capet, X., McWilliams, J.C., Molemaker, M.J.,

Shchepetkin, A.F., 2008. Mesoscale to submesoscale transition

in the California Current System. Part III: Energy

balance and flux. J. Phys. Oceanogr. 38, 2256e2269.

Carmack, E., Aagaard, K., 1973. On the deep water of the

Greenland Sea. Deep-Sea Res. 20, 687e715.

Cartwright, D.E., 1999. Tides: A Scientific History. Cambridge

University Press, Cambridge, UK, 292 pp.

Castro, S.L., Wick, G.A., Emery, W.J., 2003. Further refinements

to models for the bulk-skin sea surface temperature

difference. J. Geophys. Res. 108, 3377e3395.

Cavalieri, D., Parkinson, C., Gloersen, P., Zwally, H.J., 1996,

updated 2008. Sea ice concentrations from Nimbus-7

SMMR and DMSP SSM/I passive microwave data, 1991.

Boulder, Colorado USA: National Snow and Ice Data

Center. Digital media. http://nsidc.org/data/nsidc-

0051.html (accessed 11.11.08).

Cayan, D.R., 1992. Latent and sensible heat flux anomalies

over the northern oceans: Driving the sea surface

temperature. J. Phys. Oceanogr. 22, 859e881.

CDIP, 2009. The Coastal Data Information Program, Scripps

Institution of Oceanography. http://cdip.ucsd.edu/

(accessed 5.15.09).

Cerovecki, I., Talley, L.D., Mazloff, M., 2011. Transformation

and formation rates of Subantarctic Mode Water based

on airesea fluxes, in preparation.

Cetina, P., Candela, J., Sheinbaum, J., Ochoa, J., Badan, A.,

2006. Circulation along the Mexican Caribbean coast. J.

Geophys. Res. 111, C08021. doi:10.1029/2005JC003056.

Chaigneau, A., Pizarro, O., Rojas, W., 2008. Global climatology

of near-inertial current characteristics from

Lagrangian observations. Geophys. Res. Lett. 35, L13603.

doi:10.1029/2008GL034060.

Chang, P., Ji, L., Li, H., 1997. A decadal climate variation in

the tropical Atlantic Ocean from thermodynamic air-sea

interactions. Nature 385, 516e518.

Chapman, B., 2004. Initial visions of paradise: Antebellum

U.S. government documents on the South Pacific. J. Gov.

Inform. 30, 727e750.

Chatfield, C., 2004. The Analysis of Time Series: An Introduction,

sixth ed. Chapman and Hall/CRC Press, Boca

Raton, FL, 333 pp.

Chelton, D.B., deSzoeke, R.A., Schlax, M.G., El Naggar, K.,

Siwertz, N., 1998. Geographical variability of the first

baroclinic Rossby radius of deformation. J. Phys. Oceanogr.

28, 433e460.

Chelton, D.B., Freilich, M.H., Esbensen, S.K., 2000. Satellite

observations of the wind jets off the Pacific coast of

Central America. Part II: Regional relationships and

dynamical considerations. Mon. Weather Rev. 128,

2019e2043.

Chelton, D.B., Ries, J.C., Haines, B.J., Fu, L.L., Callahan, P.S.,

Cazenave, A., 2001. Satellite altimetry. In: Fu, L.-L.,

Cazenave, A. (Eds.), Satellite Altimetry and Earth Sciences.

Academic Press, San Diego, CA, 463 pp.

Chelton, D.B., Schlax, M.G., 1996. Global observations of

oceanic Rossby waves. Science 272, 234e238.

Chelton, D.B., Schlax, M.G., Freilich, M.H., Milliff, R.F.,

2004. Satellite measurements reveal persistent smallscale

features in ocean winds. Science 303, 978e983.

Chelton, D.B., Schlax, M.G., Samelson, R.M., de

Szoeke, R.A., 2007. Global observations of large oceanic

eddies. Geophys. Res. Lett. 34, L15606. doi:10.1029/

2007GL030812.

Chen, C., Beardsley, R.C., 2002. Cross-frontal water

exchange on Georges Bank: Some results from an U.S.

GLOBEC/Georges Bank Program Model Study. J. Oceanogr.

58, 403e420.

Chen, C.T., Millero, F.J., 1977. Speed of sound in seawater at

high pressures. J. Acoust. Soc. Am. 62, 1129e1135.

Chereskin, T.K., 1995. Direct evidence for an Ekman balance in

the California Current. J. Geophys. Res. 100, 18261e18269.

Chereskin, T.K., Trunnell, M., 1996. Correlation scales,

objective mapping, and absolute geostrophic flow in the

California Current. J. Geophys. Res. 101, 22619e22629.

Chiang, J.C.H., Vimont, D.J., 2004. Analogous Pacific and

Atlantic meridional modes of tropical atmospheree

ocean variability. J. Clim. 17, 4143e4158.

CIESM, 2001. CIESM Round table session on Mediterranean

water mass acronyms. 36th CIESM Congress, Monte

Carlo, 26 September 2001. https://www.ciesm.org/

catalog/WaterMassAcronyms.pdf (accessed 6.5.09).

Clark, C.O., Webster, P.J., Cole, J.E., 2003. Interdecadal

variability of the relationship between the Indian Ocean

zonal mode and East African coastal rainfall anomalies.

J. Clim. 16, 548e554.

Clarke, R.A., Swift, J.H., Reid, J.L., Koltermann, K.P., 1990. The

formation of Greenland Sea Deep Water: Double diffusion

or deep convection? Deep-Sea Res. Part A 37, 1385e1424.

Climate Prediction Center Internet Team, 2006. AAO, AO,

NAO, PNA. NOAA National Weather Service Climate

Prediction Center. http://www.cpc.noaa.gov/products/

precip/CWlink/daily_ao_index/teleconnections.shtml

(accessed 4.20.09).

Climate Prediction Center Internet Team, 2009. Cold and

warm episodes by season. NOAA National Weather

Service Climate Prediction Center. http://www.cpc.

noaa.gov/products/analysis_monitoring/ensostuff/

ensoyears.shtml (accessed 3.26.09).

Coachman, L.K., Aagaard, K., 1974. Physical oceanography

of arctic and subarctic seas. In: Herman, Y. (Ed.), Marine

Geology and Oceanography of the Arctic Seas. Springer-

Verlag, New York, pp. 1e81.

Cobb, K.M., Charles, C.D., Cheng, H., Edwards, R.L., 2003.

El Niño/Southern Oscillation and tropical Pacific climate

during the last millennium. Nature 424, 271e276.

REFERENCES 517

Cole, S.T., Rudnick, D.L., Hodges, B.A., Martin, J.P., 2009.

Observations of tidal internal wave beams at Kauai

Channel, Hawaii. J. Phys. Oceanogr. 39, 421e436.

Comiso, J., Wadhams, P., Pedersen, L., Gersten, R., 2001.

Seasonal and interannual variability of the Odden ice

tongue and a study of environmental effects. J. Geophys.

Res. 106, 9093e9116.

Comiso, J.C., Gordon, A.L., 1987. Recurring polynyas over

the Cosmonaut Sea and the Maud Rise. J. Geophys. Res.

92, 2819e2833.

Conkright, M.E., Levitus, S., Boyer, T.P., 1994. World Ocean

Atlas 1994 vol 1: Nutrients. NOAA Atlas NESDIS 1. U.S.

Department of Commerce, Washington, D.C., 150 pp.

Cornillon, P., 1986. The effect of the New England

Seamounts on Gulf Stream meandering as observed

from satellite IR imagery. J. Phys. Oceanogr. 16, 386e389.

Cox, R.A., McCartney, M.J., Culkin, F., 1970. The specific

gravity/salinity/temperature relationship in natural

seawater. Deep-Sea Res. 17, 679e689.

Cox, R.A., Smith, N.D., 1959. The specific heat of seawater.

Philos. Trans. Roy. Soc. London A 252, 51e62.

Craik, A.D.D., 2005. George Gabriel Stokes on water wave

theory. Ann. Rev. Fluid Mech. 37, 23e42.

Crawford, W., 2002. Physical characteristics of Haida

Eddies. J. Oceanogr. 58, 703e713.

Crawford, W., Cherniawsky, J., Foreman, M., 2000. Multiyear

meanders and eddies in the Alaskan Stream as

observed by TOPEX/Poseidon altimeter. Geophys. Res.

Lett. 27, 1025e1028.

Cresswell, G.R., Golding, T.J., 1980. Observations of a southflowing

current in the southeastern Indian Ocean. Deep-

Sea Res. 27A, 449e466.

Cunningham, S.A., Alderson, S.G., King, B.A.,

Brandon, M.A., 2003. Transport and variability of the

Antarctic Circumpolar Current in Drake Passage. J.

Geophys. Res. 108 (C5), 8084. doi:10.1029/2001JC001147.

Cunningham, S.A., Kanzow, T., Rayner, D., Baringer, M.O.,

Johns, W.E., Marotzke, J., et al., 2007. Temporal variability

of the Atlantic meridional overturning circulation

at 26.5 N. Science 317, 935e938.

Curray, J.R., Emmel, F.J., Moore, D.G., 2003. The Bengal

Fan: Morphology, geometry, stratigraphy, history and

processes. Mar. Petr. Geol. 19, 1191e1223.

Curry, R., Dickson, B., Yashayaev, I., 2003. A change in the

freshwater balance of the Atlantic Ocean over the past

four decades. Nature 426, 826e829.

Curry, R.G., McCartney, M.S., 2001. Ocean gyre circulation

changes associated with the North Atlantic Oscillation.


Cushman-Roisin, B., 1994. Introduction to Geophysical Fluid

Dynamics. Prentice Hall, Englewood Cliffs, NJ, 320 pp.

da Silva, A.M., Young, A.C., Levitus, S., 1994. Atlas of

surface marine data, vol. 1: Algorithms and Procedures.

NOAA Atlas NESDIS 6. U.S. Department of Commerce,

NOAA, NESDIS.

Dai, A., Trenberth, K.E., 2002. Estimates of freshwater

discharge from continents: Latitudinal and seasonal

variations. J. Hydromet. 3, 660e687.

d’Asaro, E.A., Eriksen, C.C., Levine, M.D., Paulson, C.A.,

Niiler, P., Van Meurs, P., 1995. Upper-ocean inertial currents

forced by a strong storm. Part 1: Data and comparisons

with linear theory. J. Phys. Oceanogr. 25, 2909e2936.

Davis, R.E., 1976. Predictability of sea surface temperature

and sea level pressure anomalies over the North Pacific.


Davis, R.E., 1991. Observing the general circulation with

floats. Deep-Sea Res. 38 (Suppl.), S531eS571.

Davis, R.E., 2005. Intermediate-depth circulation of the

Indian and South Pacific Oceans measured by autonomous

floats. J. Phys. Oceanogr. 35, 683e707.

Davis, R.E., deSzoeke, R., Niiler, P., 1981. Variability in the

upper ocean during MILE. Part II: Modeling the mixed

layer response. Deep-Sea Res. 28A, 1453e1475.

Deacon, G.E.R., 1933. A general account of the hydrology of

the South Atlantic Ocean. Disc. Rep. 7, 171e238.

Deacon, G.E.R., 1937. The hydrology of the Southern Ocean.

Disc. Rep. 15, 3e122.

Deacon, G.E.R., 1982. Physical and biological zonation in the

Southern Ocean. Deep-Sea Res. 29, 1e15.

deBoyer Montégut, C., Madec, G., Fischer, A.S., Lazar, A.,

Iudicone, D., 2004. Mixed layer depth over the global

ocean: An examination of profile data and a profilebased

climatology. J. Geophys. Res. 109, C12003.

doi:10.1029/2004JC002378.

Defant, A., 1936. Die Troposphäre des Atlantischen Ozeans. In

Wissenschaftliche Ergebnisse der Deutschen Atlantischen

Expedition auf dem Forschungs- und Vermessungsschiff

“Meteor” 1925-1927, 6(1), 289-411 (in German).

Defant, A., 1961. Physical Oceanography, vol. 1. Pergamon


Del Grosso, V.A., 1974. New equation for the speed of sound

in natural waters (with comparisons to other equations.

J. Acoust. Soc. Am. 56, 1084e1091.

Delworth, T.L., Mann, M.E., 2000. Observed and simulated

multidecadal variability in the northern hemisphere.

Clim. Dynam. 16, 661e676.

Dengler, M., Quadfasel, D., Schott, F., Fischer, J., 2002.

Abyssal circulation in the Somali Basin. Deep-Sea Res. II

49, 1297e1322.

Dengler, M., Schott, F.A., Eden, C., Brandt, P., Fischer, J.,

Zantopp, R.J., 2004. Break-up of the Atlantic deep

western boundary current into eddies at 8 S. Nature 432,

1018e1020.

Déry, S.J., Stieglitz, M., McKenna, E.C., Wood, E.F., 2005.

Characteristics and trends of river discharge into Hudson,

James, and Ungava Bays, 1964e2000. J. Clim. 18, 2540e2557.

518

REFERENCES

Deser, C., Holland, M., Reverdin, G., Timlin, M., 2002.

Decadal variations in Labrador Sea ice cover and North

Atlantic sea surface temperatures. J. Geophys. Res. 107,

3035. doi:10.1029/2000JC000683.

Deser, C., Phillips, A.S., Hurrell, J.W., 2004. Pacific interdecadal

climate variability: Linkages between the tropics

and the North Pacific during boreal winter since 1900.

J. Clim. 17, 3109e3124.

deSzoeke, R.A., Levine, M.D., 1981. The advective flux of

heat by mean geostrophic motions in the Southern

Ocean. Deep-Sea Res. 28, 1057e1085.

Deutsch, C., Emerson, S., Thompson, L., 2005. Fingerprints

of climate change in North Pacific oxygen. Geophys. Res.

Lett. 32, L16604. doi:10.1029/2005GL023190.

Dickson, R., Brown, J., 1994. The production of North

Atlantic Deep Water: Sources, rates, and pathways.

J. Geophys. Res. 99, 12319e12341.

Dickson, R., Lazier, J., Meincke, J., Rhines, P., Swift, J., 1996.

Long-term coordinated changes in the convective

activity of the North Atlantic. Progr. Oceanogr. 38,

241e295.

Dickson, R., Yashayaev, I., Meincke, J., Turrell, B., Dye, S.,

Holfort, J., 2002. Rapid freshening of the deep North

Atlantic Ocean over the past four decades. Nature 416,

832e837.

Dickson, R.R., Curry, R., Yashayaev, I., 2003. Recent changes

in the North Atlantic. Phil. Trans. Roy. Soc. A 361,

1917e1933.

Dickson, R.R., Meincke, J., Malmberg, S.-A., Lee, A.J., 1988.

The “Great Salinity Anomaly” in the northern North

Atlantic 1968e1982. Progr. Oceanogr. 20, 103e151.

Dickson, R.R., Meincke, J., Rhines, P. (Eds.), 2008. Arctic-

Subarctic Ocean Fluxes: Defining the Role of the Northern

Seas in Climate. Springer, The Netherlands, 736 pp.

Dietrich, G., 1963. Allgemeine Meereskunde. Gebruder

Borntraeger Verlagsbuchhandlung, Berlin-Stuttgart.

(English translation: General Oceanography. Wiley-

Interscience, New York), 492 pp.

DiLorenzo, E., Schneider, N., Cobb, K.M., Franks, P.J.S.,

Chhak, K., Miller, A.J., et al., 2008. North Pacific Gyre

Oscillation links ocean climate and ecosystem change.

Geophys. Res. Lett. 35, L08607. doi:10.1029/

2007GL032838.

Dittmar, W., 1884. Report of researches into the composition

of ocean water collected by HMS Challenger

during the years 1873e76. Voyage of the H.M.S.Challenger:

Physics and chemistry, 1, part 1. Longmans &

Co., London.

Dmitrenko, I.A., Tyshko, K.N., Kirillov, S.A., Eicken, H.,

Hölemann, J.A., Kassens, H., 2005. Impact of flaw

polynyas on the hydrography of the Laptev Sea. Global

Planet. Change 48, 9e27.

Dmitrenko, I.A., Wegner, C., Kassens, H., Kirillov, S.A.,

Krumpen, T., Heinemann, G., et al., 2010. Observations

of supercooling and frazil ice formation in the Laptev

Sea coastal polynya. J. Geophys. Res. 115, C05015.

doi:10.1029/2009JC005798.

Domingues, C.M., Church, J.A., White, N.J., Gleckler, P.J.,

Wijffels, S.E., Barker, P.M., et al., 2008. Improved estimates

of upper-ocean warming and multi-decadal sealevel

rise. Nature 453, 1090e1093.

Domingues, C.M., Maltrud, M.E., Wijffels, S.E., Church, J.A.,

Tomczak, M., 2007. Simulated Lagrangian pathways

between the Leeuwin Current System and the upperocean

circulation of the southeast Indian Ocean. Deep-

Sea Res. II 54, 797e817.

Dong, S., Gille, S.T., Sprintall, J., Talley, L., 2008. Southern

Ocean mixed-layer depth from Argo float profiles.

J. Geophys. Res. 113, C06013. doi:10.1029/2006JC004051.

Donohue, K.A., Firing, E., Chen, S., 2001. Absolute

geostrophic velocity within the Subantarctic Front in the

Pacific Ocean. J. Geophys. Res. 106, 19869e19882.

Döös, K., Coward, A., 1997. The Southern Ocean as the major

upwelling zone of North Atlantic Deep Water. International

WOCE Newsletter, 27, 3e4. http://woce.nodc.

noaa.gov/wdiu/wocedocs/newsltr/ (accessed 09.01.09).

Döös, K., Webb, D.J., 1994. The Deacon cell and the other

meridional cells of the Southern Ocean. J. Phys. Oceanogr.

24, 429e442.

Doron, P., Bertuccioli, L., Katz, J., Osborn, T.R., 2001.

Turbulence characteristics and dissipation estimates in

the coastal ocean bottom boundary layer from PIV data.


Doronin, Y.P., Khesin, D.E., 1975. Sea Ice (Trans., 1977).

Amerind Publishing Company, New Delhi, India, 323 pp.

Ducet,N.,LeTraon,P.Y.,Reverdin,G.,2000.Globalhigh-resolution

mapping of ocean circulation from TOPEX/Poseidon

and ERS-1 and -2. J. Geophys. Res. 105, 19477e19498.

Durack, P.J., Wijffels, S.E., 2010. Fifty-year trends in global

ocean salinities and their relationship to broad-scale

warming. J. Clim. 23, 4342e4362.

Dyer, K.R., 1997. Estuaries: A Physical Introduction, second

ed. Wiley, New York, 195 pp.

Egbert, G.D., Ray, R., 2001. Estimates of M2 tidal energy

dissipation from TOPEX/Poseidon altimeter data.

J. Geophys. Res. 106, 22475e22502.

Ekman, V.W., 1905. On the influence of the Earth’s rotation on

ocean currents. Arch. Math. Astron. Phys. 2 (11), 1e53.

Ekman, V.W., 1923. Über Horizontalzirkulation bei winderzeugten

Meereströmungen. Ark. Math. Astron Fys. 17,

1e74 (in German).

Elipot, S., Lumpkin, R., 2008. Spectral description of oceanic

near-surface variability. Geophys. Res. Lett. 35, L05606.

doi:10.1029/2007GL032874.

REFERENCES 519

Emery, W.J., 1977. Antarctic polar frontal zone from Australia

to the Drake Passage. J. Phys. Oceanogr. 7,

811e822.

Emery, W.J., Fowler, C.W., Maslanik, J.A., 1997. Satellite

derived Arctic and Antarctic sea ice motions: 1988e

1994. Geophys. Res. Lett. 24, 897e900.

Emery, W.J., Meincke, J., 1986. Global water masses:

summary and review. Oceanol. Acta 9, 383e391.

Emery, W.J., Thomson, R.E., 2001. Data Analysis Methods in

Physical Oceanography, second ed. Elsevier, Amsterdam,

638 pp.

Enfield, D., Mestas-Nuñez, A., Trimble, P., 2001. The

Atlantic Multidecadal Oscillation and its relation to

rainfall and river flows in the continental U.S. Geophys.

Res. Lett. 28, 2077e2080.

Eriksen, C.C., 1982. Geostrophic equatorial deep jets. J. Mar.

Res. 40 (Suppl.), 143e157.

Fang, F., Morrow, R., 2003. Evolution, movement and decay

of warm-core Leeuwin Current eddies. Deep-Sea Res. II

50, 2245e2261.

Farmer, D.M., Freeland, H.J., 1983. The physical oceanography

of fjords. Progr. Oceanogr. 12, 147e219.

Favorite, F., Dodimead, A.J., Nasu, K., 1976. Oceanography

of the subarctic Pacific region, 1960e71. International

North Pacific Fisheries Commission, Vancouver,

Canada, 33, 187 pp.

Fedorov, A.V., Harper, S.L., Philander, S.G., Winter, B.,

Wittenberg, A., 2003. How predictable is El Niño? B.

Am. Meteorol. Soc. 84, 911e919.

Feely, R.A., Sabine, C.L., Lee, K., Berelson, W., Kleypas, J.,

Fabry, V.J., et al., 2004. Impact of anthropogenic CO 2 on

the CaCO 3 system in the oceans. Science 305, 362e366.

Feng, M., Meyers, G., Pearce, A., Wijffels, S., 2003. Annual

and interannual variations of the Leeuwin Current

at 32 S. J. Geophys. Res. 108, C11. doi:10.1029/

2002JC001763.

Feng, M., Wijffels, S., Godfrey, S., Meyers, G., 2005. Do

eddies play a role in the momentum balance of the

Leeuwin Current? J. Phys. Oceanogr. 35, 964e975.

Field, J.G., Shillington, F.A., 2006. Variability of the Benguela

Current System. In: Robinson, A.R., Brink, K.H. (Eds.),

The Sea, vol. 14B: The Global Coastal Ocean: Interdisciplinary

Regional Studies and Syntheses. Harvard

University Press, Boston, MA, pp. 835e864.

Fine, R.A., 1993. Circulation of Antarctic Intermediate Water

in the South Indian Ocean. Deep-Sea Res. I 40, 2021e2042.

Fine, R.A., Lukas, R., Bingham, F.M., Warner, M.J.,

Gammon, R.H., 1994. The western equatorial Pacific: A

water mass crossroads. J. Geophys. Res. 99, 25063e25080.

Fine, R.A., Maillet, K.A., Sullivan, K.F., Willey, D., 2001.

Circulation and ventilation flux of the Pacific Ocean.

J. Geophys. Res. 106, 22159e22178.

Firing, E., 1989. Mean zonal currents below 1500 m near the

equator, 159 W. J. Geophys. Res. 94, 2023e2028.

Firing, E., Wijffels, S.E., Hacker, P., 1998. Equatorial subthermocline

currrents across the Pacific. J. Geophys. Res.

103, 21413e21423.

Flatau, M.K., Talley, L.D., Niiler, P.P., 2003. The North

Atlantic Oscillation, surface current velocities, and SST

changes in the subpolar North Atlantic. J. Clim. 16,

2355e2369.

Fofonoff, N.P., 1954. Steady flow in a frictionless homogeneous

ocean. J. Mar. Res. 13, 254e262.

Fofonoff, N.P., 1977. Computation of potential temperature

of seawater for an arbitrary reference pressure. Deep-Sea

Res. 24, 489e491.

Fofonoff, N.P., 1985. Physical properties of seawater: A new

salinity scale and equation of state for seawater. J. Geophys.

Res. 90, 3332e3342.

Forch, C., Knudsen, M., Sorensen, S.P., 1902. Berichte über

die Konstantenbestimmungen zur Aufstellung der

hydrographischen Tabellen. Kgl. Dan. Vidensk. Selsk.

Skr., 6, Raekke, Naturvidensk. Mat., Afel. XII. 1, 151 (in

German).

Forchhammer, G., 1865. On the composition of seawater in

the different parts of the ocean. Philos. Trans. Roy. Soc.

Lond. 155, 203e262.

Foster, T.D., 1972. An analysis of the cabbeling instability in

seawater. J. Phys. Oceanogr. 2, 294e301.

Fowler, C., 2003, updated 2007. Polar Pathfinder daily 25 km

EASE-Grid sea ice motion vectors. National Snow and

Ice Data Center. http://nsidc.org/data/nsidc-0116.html.

ftp://sidads.colorado.edu/pub/DATASETS/ice_

motion/browse (accessed 11.01.08).

Fowler, C., Emery, W.J., Maslanik, J., 2004. Satellite-derived

evolution of Arctic sea ice age: October 1978 to March

2003. IEEE Remote Sensing Lett. 1, 71e74.

Frammuseet, 2003. Picture archive d 1st Fram voyage.

http://www.fram.museum.no (accessed 3.17.09).

Fratantoni, D.M., 2001. North Atlantic surface circulation

during the 1990s observed with satellite-tracked drifters.

J. Geophys. Res. 106, 22067e22093.

Fratantoni, D.M., Johns, W.E., Townsend, T.L.,

Hurlburt, H.E., 2000. Low-latitude circulation and mass

transport pathways in a model of the tropical Atlantic

Ocean. J. Phys. Oceanogr., 301944e301966.

Fratantoni, D.M., Zantopp, R.J., Johns, W.E., Miller, J.L.,

1997. Updated bathymetry of the Anegada-Jungfern

Passage complex and implications for Atlantic inflow

to the abyssal Caribbean Sea. J. Mar. Res. 55,

847e860.

Friedrichs, M., McCartney, M., Hall, M., 1994. Hemispheric

asymmetry of deep water transport modes in the

western Atlantic. J. Geophys. Res. 99, 25165e25179.

520

REFERENCES

Fu, L.-L., Chelton, D.B., 2001. In: Fu, L.-L., Cazenave, A.

(Eds.), International Geophysics Series. Large-scale

ocean circulation. In Satellite Altimetry and Earth

Sciences: A Handbook of Techniques and Applications,

69. Academic Press, San Diego, CA, pp. 133e170.

Fuglister, F.C., 1960. Atlantic Ocean Atlas of temperature

and salinity profiles and data from the IGY of

1957e1958. Woods Hole Oceanographic Institution Atlas

Series 1, 209 pp.

Fyfe, J.C., 2006. Southern Ocean warming due to human

influence. Geophys. Res. Lett. 33, L19701. doi:10.1029/

2006GL027247.

Ganachaud, A., 2003. Large-scale mass transports, water

mass formation, and diffusivities estimated from World

Ocean Circulation Experiment (WOCE) hydrographic

data. J. Geophys. Res. 108, 3213. doi: 10.1029/

2002JC002565.

Ganachaud, A., Gourdeau, L., Kessler, W., 2008. Bifurcation

of the subtropical south equatorial current against

New Caledonia in December 2004 from a hydrographic

inverse box model. J. Phys. Oceanogr. 38, 2072e2084.

Ganachaud, A., Wunsch, C., 2000. Improved estimates of

global ocean circulation, heat transport and mixing from

hydrographic data. Nature 408, 453e457.

Ganachaud, A., Wunsch, C., 2003. Large-scale ocean heat

and freshwater transports during the World Ocean

Circulation Experiment. J. Clim. 16, 696e705.

Ganachaud, A., Wunsch, C., Marotzke, J., Toole, J., 2000.

Meridional overturning and large-scale circulation of the

Indian Ocean. J. Geophys. Res. 105, 26117e26134.

Gardner, W.D., 2009. Visibility in the ocean and the effects of

mixing. Quarterdeck 5(1), Spring 1997. http://oceanz.

tamu.edu/~pdgroup/Qdeck/gardner-5.1.html (accessed

2.18.09).

Gardner, W.D., Mishonov, A.V., Richardson, M.J., 2006.

Global POC concentrations from in-situ and satellite

data. Deep-Sea Res. II 53, 718e740.

Garrett, C., 1972. Tidal resonance in the Bay of Fundy and

Gulf of Maine. Nature 238, 441e443.

Garrett, C.J., Munk, W., 1972. Spaceetime scales of internal

waves. Geophys. Fluid Dyn. 2, 255e264.

Garrett, C.J., Munk, W., 1975. Spaceetime scales of internal

waves: A progress report. J. Geophys. Res. 80, 291e297.

Garrison, T., 2001. Essentials of Oceanography, second ed.

Brooks/Cole, Pacific Grove, CA, 361 pp.

Gent, P.R., McWilliams, J.C., 1990. Isopycnal mixing in ocean

circulation models. J. Phys. Oceanogr. 20, 150e155.

Giarolla, E., Nobre, P., Malaguti, M., Pezzi, L.P., 2005. The

Atlantic Equatorial Undercurrent: PIRATA observations

and simulations with GFDL modular ocean model at

CPTEC. Geophys. Res. Lett. 32, L10617. doi:10.1029/

2004GL022206.

Gill, A.E., 1982. Atmospheric-Ocean Dynamics. Academic


Gill, A.E., Niiler, P., 1973. The theory of seasonal variability

in the ocean. Deep-Sea Res. 20, 141e177.

Gille, S.T., 1996. Scales of spatial and temporal variability in

the Southern Ocean. J. Geophys. Res. 101, 8759e8773.

Gille, S.T., 2002. Warming of the Southern Ocean since the

1950s. Science 295, 1275e1277.

Gille, S.T., 2003. Float observations of the Southern Ocean. Part

I: Estimating mean fields, bottom velocities, and topographic

steering. J. Phys. Oceanogr. 33, 1167e1181.

Gille, S.T., 2005. MAE 127: Statistical methods for environmental

sciences and engineering. http://www-pord.

ucsd.edu/~sgille/mae127/index.html (accessed 4.14.09).

Giosan, L., Filip, F., Constatinescu, S., 2009. Was the Black

Sea catastrophically flooded in the early Holocene?

Quaternary Sci. Rev. 28, 1e6.

Girton, J.B., Pratt, L.J., Sutherland, D.A., Price, J.F., 2006. Is

the Faroe Bank Channel overflow hydraulically

controlled? J. Phys. Oceanogr. 36, 2340e2349.

Gladyshev, S., Talley, L., Kantakov, G., Khen, G.,

Wakatsuchi, M., 2003. Distribution, formation and

seasonal variability of Okhotsk Sea Intermediate Water.

J. Geophys. Res. 108, 3186. doi:10.1029/2001JC000877.

Godfrey, J., Cresswell, G., Golding, T., Pearce, A., Boyd, R.,

1980. The separation of the East Australian Current.


Godfrey, J.S., Weaver, A.J., 1991. Is the Leeuwin Current

driven by Pacific heating and winds? Progr. Oceanogr.

27, 225e272.

Gonzalez, F.I., 1999. Tsunami! Sci. Am., 56e63. May 1999.

Gordon, A., 1991. The role of thermohaline circulation in

global climate change. In: LamonteDoherty Geological

Observatory 1990 & 1991 Report. Lamont-Doherty

Geological Observatory of Columbia University,

Palisades, New York, pp. 44e51.

Gordon, A.L., 1986. Interocean exchange of thermocline

water. J. Geophys. Res. 91, 5037e5046.

Gordon, A.L., 2003. Oceanography: The brawniest retroflection.

Nature 421, 904e905.

Gordon, A.L., 2005. Oceanography of the Indonesian Seas

and their throughflow. Oceanography 18, 14e27.

Gordon, A.L., Bosley, K.T., 1991. Cyclonic gyre in the tropical

South Atlantic. Deep-Sea Res. 38 (Suppl.), S323eS343.

Gordon, A.L., Georgi, D.T., Taylor, H.W., 1977. Antarctic

polar front zone in the western Scotia Sea d Summer

1975. J. Phys. Oceanogr. 7, 309e328.

Gordon, A.L., Giulivi, C.F., Ilahude, A.G., 2003. Deep

topographic barriers within the Indonesian seas. Deep-

Sea Res. II 50, 2205e2228.

Gordon, A.L., Susanto, R.D., Ffield, A., 1999. Throughflow

within Makassar Strait. Geophys. Res. Lett. 26, 3321e3328.

REFERENCES 521

Gordon, A.L., Visbeck, M., Comiso, J.C., 2007. A possible

link between the Weddell polynya and the Southern

Annular Mode. J. Clim. 20, 2558e2571.

Gregg, M.C., 1987. Diapycnal mixing in the thermocline:

A review. J. Geophys. Res. 94, 5249e5286.

Grist, J.P., Josey, S.A., 2003. Inverse analysis adjustment of

the SOC air-sea flux climatology using ocean heat

transport constraints. J. Clim. 20, 3274e3295.

Grötzner, A., Latif, M., Barnett, T.P., 1998. A decadal climate

cycle in the North Atlantic Ocean as simulated by the

ECHO coupled GCM. J. Clim. 11, 831e847.

Gruber, N., Sarmiento, J.L., 1997. Global patterns of marine

nitrogen fixation and denitrification. Glob. Biogeochem.

Cyc. 11, 235e266.

Guza, R., Thornton, E., 1982. Swash oscillations on a natural

beach. J. Geophys. Res. 87, 483e491.

Haarpaintner, J., Gascard, J.-C., Haugan, P.M., 2001. Ice

production and brine formation in Storfjorden, Svalbard.

J. Geophys. Res. 106, 14001e14013.

Häkkinin, S., Rhines, P.B., 2004. Decline of subpolar North

Atlantic circulation during the 1990s. Science 304,

555e559.

Hall, A., Visbeck, M., 2002. Synchronous variability in the

southern hemisphere atmosphere, sea ice and ocean

resulting from the annular mode. J. Clim. 15, 3043e3057.

Hamon, B.V., 1965. The East Australian Current, 1960e1964.

Deep-Sea Res. 12, 899e921.

Han, W., McCreary, J.P., Anderson, D.L.T., Mariano, A.J.,

1999. Dynamics of the eastern surface jets in the equatorial

Indian Ocean. J. Phys. Oceanogr. 29, 2191e2209.

Hanawa, K., Talley, L.D., 2001. Mode Waters. In: Siedler, G.,

Church, J. (Eds.), Ocean Circulation and Climate. International

Geophysics Series. Academic Press, San Diego,

CA, pp. 373e386.

Hannah, C.G., Dupont, F., Dunphy, M., 2009. Polynyas and

tidal currents in the Canadian Arctic Archipelago. Arctic

62, 83e95.

Hansen, B., Østerhus, S., 2000. North Atlantic-Nordic Seas

exchanges. Progr. Oceanogr. 45, 109e208.

Hansen, B., Østerhus, S., Turrell, W.R., Jónsson, S.,

Valdimarsson, H., Hátún, H., et al., 2008. The inflow of

Atlantic water, heat, and salt to the Nordic Seas across

the Greenland-Scotland Ridge. In: Dickson, R.R.,

Meincke, J., Rhines, P. (Eds.), Arctic-Subarctic Ocean

Fluxes: Defining the Role of the Northern Seas in

Climate. Springer, The Netherlands, pp. 15e44.

Hardisty,J.,2007.Estuaries: Monitoring and Modeling the

Physical System. Blackwell Publishing, Maiden, MA,

157 pp.

Hasunuma, K., Yoshida, K., 1978. Splitting of the subtropical

gyre in the western North Pacific. J. Oceanogr. Soc. Japan

34, 160e172.

Hautala, S., Reid, J., Bray, N., 1996. The distribution and

mixing of Pacific water masses in the Indonesian Seas.

J. Geophys. Res. 101, 12375e12389.

Hautala, S.L., Sprintall, J., Potemra, J., Chong, J.C.C.,

Pandoe, W., Bray, N., et al., 2001. Velocity structure and

transport of the Indonesian throughflow in the major

straits restricting flow into the Indian Ocean. J. Geophys.

Res. 106, 19527e19546.

Hecht, M.W., Hasumi, H. (Eds.) 2008. Ocean Modeling in

an Eddying Regime. AGU Geophys. Monogr. Ser. 177,

350 pp.

Heezen, B.C., Ericson, D.B., Ewing, M., 1954. Further

evidence for a turbidity current following the 1929

Grand Banks earthquake. Deep-Sea Res. 1, 193e202.

Helland-Hansen, B., 1916. Nogen hydrografiske metoder.

In: Forhandlinger ved de 16 Skandinaviske Naturforskerermote,

pp. 357e359 (in Norwegian).

Helland-Hansen, B., 1934. The Sognefjord section. Oceanographic

Observations in the northernmost part of the

North Sea and the southern part of the Norwegian Sea.

J. Johnstone Mem. Vol., Liverpool, 257 pp.

Herbers, T.H.C., Elgar, S., Sarap, N.A., Guza, R.T., 2002.

Nonlinear dispersion of surface gravity waves in

shallow water. J. Phys. Oceanogr. 32, 1181e1193.

Hickey, B.M., 1998. Coastal oceanography of western North

America from the tip of Baja California to Vancouver Island.

In:Robinson,A.R.,Brink,K.H.(Eds.),TheSea,vol.11,The

Global Coastal Ocean: Regional Studies and Syntheses.

John Wiley and Sons, New York, pp. 345e394.

Hisard, P., Hénin, C., 1984. Zonal pressure gradient, velocity

and transport in the Atlantic Equatorial Undercurrent

from FOCAL cruises (July 1982eFebruary 1984). Geophys.

Res. Lett. 11, 761e764.

Hofmann, E.E., 1985. The large-scale horizontal structure of

the Antarctic Circumpolar Current from FGGE drifters.

J. Geophys. Res. 90, 7087e7097.

Hogg, N.G., 1983. A note on the deep circulation of the

western North Atlantic: Its nature and causes. Deep-Sea

Res. 30, 945e961.

Hogg, N.G., Owens, W.B., 1999. Direct measurement of the

deep circulation within the Brazil Basin. Deep-Sea Res. II

46, 335e353.

Hogg, N.G., Pickart, R.S., Hendry, R.M., Smethie Jr., W.M.,

1986. The northern recirculation gyre of the Gulf Stream.

Deep-Sea Res. 33, 1139e1165.

Hogg, N.G., Siedler, G., Zenk, W., 1999. Circulation and

variability at the southern boundary of the Brazil Basin.


Holliday, N.P., Meyer, A., Bacon, S., Alderson, S.G., de

Cuevas, B., 2007. Retroflection of part of the east

Greenland current at Cape Farewell. Geophys. Res. Lett.

34, L07609. doi:10.1029/2006GL029085.

522

REFERENCES

Holte, J., Gilson, J., Talley, L., Roemmich, D., 2010. Argo

mixed layers. Scripps Institution of Oceanography,

UCSD. http://mixedlayer.ucsd.edu (accessed 2.24.10).

Holte, J., Talley, L., 2009. A new algorithm for finding mixed

layer depths with applications to Argo data and

Subantarctic Mode Water formation. J. Atmos. Ocean.

Tech. 26, 1920e1939.

Horrillo, J., Knight, W., Kowalik, Z., 2008. Kuril Islands

tsunami of November 2006: 2. Impact at Crescent City by

local enhancement. J. Geophys. Res. 113, C01021.

doi:10.1029/2007JC004404.

Hosoda, S., Suga, T., Shikama, N., Mizuno, K., 2009. Surface

and subsurface layer salinity change in the global ocean

using Argo float data. J. Oceanogr. 65, 579e586.

Hu, J., Kawamura, H., Hong, H., Qi, Y., 2000. A review on

the currents in the South China Sea: seasonal circulation,

South China Sea Warm Current and Kuroshio intrusion.

J. Oceanogr. 56, 607e624.

Hu, S., Townsend, D.W., Chen, C., Cowles, G.,

Beardsley, R.C., Ji, R., et al., 2008. Tidal pumping and

nutrient fluxes on Georges Bank: A process-oriented

modeling study. J. Marine Syst. 74, 528e544.

Hufford, G.E., McCartney, M.S., Donohue, K.A., 1997.

Northern boundary currents and adjacent recirculations

off southwestern Australia. Geophys. Res. Lett. 24 (22),

2797e2800. doi_10.1029/97GL02278.

Hughes, S.L., Holliday, N.P., Beszczynska-Möller, A. (Eds.),

2008. ICES Report on Ocean Climate 2007. ICES Cooperative

Research Report No. 291, p. 64. http://www.noc.

soton.ac.uk/ooc/ICES_WGOH/iroc.php (accessed 7.1.09).

Hurdle, B.G. (Ed.), 1986. The Nordic Seas. Springer-Verlag,

New York, 777 pp.

Hurlburt, H.E., Thompson, J.D., 1973. Coastal upwelling on

a b-plane. J. Phys. Oceanogr. 19, 16e32.

Hurrell, J., 2009. Climate Indices. NAO Index Data provided

by the Climate Analysis Section, NCAR, Boulder,

Colorado (Hurrell, 1995). http://www.cgd.ucar.edu/

cas/jhurrell/nao.stat.winter.html (accessed 6.23.09).

Hurrell, J.W., Kushnir, Y., Ottersen, G., Visbeck, M., 2003. An

overview of the North Atlantic Oscillation. In: The North

Atlantic Oscillation: Climate Significance and Environmental

Impact. Geophys. Monogr. Ser. 134, 1e35.

Huyer, A., 1983. Coastal upwelling in the California Current

System. Progr. Oceanogr. 12, 259e284.

IAPP, 2010. International Arctic Polynya Programme. Arctic

Ocean Sciences Board. http://aosb.arcticportal.org/

iapp/iapp.html (accessed 11/26/10).

Ihara, C., Kushnir, Y., Cane, M.A., 2008. Warming trend of

the Indian Ocean SST and Indian Ocean dipole from

1880 to 2004. J. Clim. 21, 2035e2046.

Imawaki, S., Uchida, H., Ichikawa, H., Fukasawa, M.,

Umatani, S., 2001. Satellite altimeter monitoring the

Kuroshio transport south of Japan. Geophys. Res. Lett.

28, 17e20.

IOC, SCOR, IAPSO, 2010. The international thermodynamic

equation of seawater d 2010: Calculation and use of

thermodynamic properties. Intergovernmental Oceanographic

Commission, Manuals and Guides No. 56.

UNESCO (English), 196 pp.

IPCC, et al., 2001. Climate Change 2001: The Scientific

Basis. In: Houghton, J.T., Ding, Y., Griggs, D.J.,

Noguer, M., van der Linden, P.J., Dai, X. (Eds.),

Contribution of Working Group I to the Third Assessment

Report of the Intergovernmental Panel on Climate

Change. Cambridge University Press, Cambridge, UK

and New York, 881 pp.

IPCC, 2007. Summary for Policymakers. In: Solomon, S.,

Qin, D., Manning, M., Chen, Z., Marquis, M.,

Averyt, K.B., et al. (Eds.), Climate Change 2007: The

Physical Science Basis. Contribution of Working Group I

to the Fourth Assessment Report of the Intergovernmental

Panel on Climate Change. Cambridge University

Press, Cambridge, UK, New York.

ISCCP, 2007. ISCCP and other cloud data, maps, and plots

available on-line. NASA Goddard Institute for Space

Studies. http://isccp.giss.nasa.gov/products/onlineData.

html (accessed 10.16.10).

Iselin, C.O’D. 1936. A study of the circulation of the western

North Atlantic. Papers in Physical Oceanography and

Meteorology, 4(4), 10 pp. MIT and Woods Hole Oceanographic

Institution.

Iselin C.O’D, 1939. The influence of vertical and lateral

turbulence on the characteristics of the waters at middepths.

Trans. Am. Geophys. Union 20, 414e417.

Ishii, M., Kimoto, M., Sakamoto, K., Iwasaki, S.I., 2006.

Steric sea level changes estimated from historical ocean

subsurface temperature and salinity analyses. J. Oceanogr.

62, 155e170.

Ivers, W.D., 1975. The deep circulation in the northern

Atlantic with special reference to the Labrador Sea. Ph.D.

Thesis, University of California at San Diego, 179 pp.

Jackett, D.R., McDougall, T.J., 1997. A neutral density variable

for the world’s oceans. J. Phys. Oceanogr. 27, 237e263.

Jacobs, S.S., 1991. On the nature and significance of the

Antarctic Slope Front. Mar. Chem. 35, 9e24.

Jakobsen, F., 1995. The major inflow to the Baltic Sea during

January 1993. J. Marine Syst. 6, 227e240.

Jakobsson, M., 2002. Hypsometry and volume of the Arctic

Ocean and its constituent seas. Geochem. Geophys.

Geosys. 3 (5), 1028. doi:10.1029/2001GC000302.

Jenkins, A., Holland, D., 2002. A model study of ocean

circulation beneath Filchner-Ronne Ice Shelf, Antarctica:

Implications for bottom water formation. Geophys. Res.

Lett. 29, 8. doi:10.1029/2001GL014589.

REFERENCES 523

Jenkins, W.J., 1998. Studying thermocline ventilation and

circulation using tritium and 3He. J. Geophys. Res. 103,

15817e15831.

Jerlov, N.G., 1976. Marine Optics. Elsevier, Amsterdam,

231 pp.

Jin, F.F., 1996. Tropical ocean-atmosphere interaction, the

Pacific cold tongue, and the El Niño-Southern Oscillation.

Science 274, 76e78.

JISAO, 2004. Arctic Oscillation (AO) time series, 1899 d

June 2002. JISAO. http://www.jisao.washington.edu/

ao/ (accessed 3.18.10).

Jochum, M., Malanotte-Rizzoli, P., Busalacchi, A., 2004.

Tropical instability waves in the Atlantic Ocean. Ocean

Model. 7, 145e163.

Johannessen, O.M., Shalina, E.V., Miles, M.W., 1999. Satellite

evidence for an arctic sea ice cover in transformation.

Science 286, 1937e1939.

Johns Hopkins APL Ocean Remote Sensing, 1996. Sea

surface temperature imagery. http://fermi.jhuapl.edu/

avhrr/sst.html (accessed 6.10.09).

Johns, W., Lee, T., Schott, F., Zantopp, R., Evans, R., 1990. The

North Brazil Current retroflection: seasonal structure and

eddy variability. J. Geophys. Res. 95, 22103e22120.

Johns, W.E., Beal, L.M., Baringer, M.O., Molina, J.R.,

Cunningham, S.A., Kanzow, T., et al., 2008. Variability of

shallow and deep western boundary currents off the

Bahamas during 2004e05: Results from the 26 N RAP-

IDeMOC Array. J. Phys. Oceanogr. 38, 605e623.

Johns, W.E., Jacobs, G.A., Kindle, J.C., Murray, S.P.,

Carron, M., 1999. Arabian Marginal Seas and Gulfs:

Report of a Workshop held at Stennis Space Center,

Miss. 11e13 May, 1999. University of Miami RSMAS.

Technical Report 2000e01.

Johns, W.E., Lee, T.N., Beardsley, R.C., Candela, J.,

Limeburner, R., Castro, B., 1998. Annual cycle and

variability of the North Brazil Current. J. Phys. Oceanogr.

28, 103e128.

Johns, W.E., Lee, T.N., Zhang, D., Zantopp, R., Liu, C.-T.,

Yang, Y., 2001. The Kuroshio east of Taiwan: Moored

transport observations from the WOCE PCM-1 array.


Johns, W.E., Shay, T.J., Bane, J.M., Watts, D.R., 1995. Gulf

Stream structure, transport, and recirculation near 68 W.

J. Geophys. Res. 100, 817e838.

Johns, W.E., Townsend, T.L., Fratantoni, D.M., Wilson, W.D.,

2002. On the Atlantic inflow to the Caribbean Sea. Deep-

Sea Res. I 49, 211e243.

Johns, W.E., Yao, F., Olsen, D.B., Josey, S.A., Grist, J.P.,

Smeed, D.A., 2003. Observations of seasonal exchange

through the Straits of Hormuz and the inferred heat and

freshwater budgets of the Persian Gulf. J. Geophys. Res.

108 (C12), 3391. doi:10.1029/2003JC001881.

Johnson, G.C., 2008. Quantifying Antarctic Bottom Water

and North Atlantic Deep Water volumes. J. Geophys.

Res. 113, C05027. doi:10.1029/2007JC004477.

Johnson, G.C., Gruber, N., 2007. Decadal water mass variations

along 20 W in the northeastern Atlantic Ocean.

Progr. Oceanogr. 73, 277e295.

Johnson, G.C., McPhaden, M.J., 1999. Interior pycnocline

flow from the subtropical to the equatorial Pacific Ocean.


Johnson, G.C., Musgrave, D.L., Warren, B.A., Ffield, A.,

Olson, D.B., 1998. Flow of bottom and deep water in the

Amirante Passage and Mascarene Basin. J. Geophys.

Res. 103, 30973e30984.

Johnson, G.C., Sloyan, B.M., Kessler, W.S., McTaggert, K.E.,

2002. Direct measurements of upper ocean currents and

water properties across the tropical Pacific during the

1990s. Progr. Oceanogr. 52, 31e61.

Johnson, G.C., Toole, J.M., 1993. Flow of deep and bottom

waters in the Pacific at 10 N. Deep-Sea Res. I 40, 371e394.

Jones, E.P., 2001. Circulation in the Arctic Ocean. Polar Res.

20, 139e146.

Jones, E.P., Anderson, L.G., Swift, J.H., 1998. Distribution of

Atlantic and Pacific waters in the upper Arctic Ocean:

Implications for circulation. Geophys. Res. Lett. 25, 765e768.

Jones, E.P., Swift, J.H., Anderson, L.G., Lipizer, M.,

Civitarese, G., Falkner, K.K., et al., 2003. Tracing Pacific

water in the North Atlantic Ocean. J. Geophys. Res. 108,

3116. doi:10.1029/2001JC001141.

Josey, S.A., Kent, E.C., Taylor, P.K., 1999. New insights into the

ocean heat budget closure problem from analysis of the

SOC air-sea flux climatology. J. Clim. 12, 2856e2880.

Josey, S.A., Marsh, R., 2005. Surface freshwater flux variability

and recent freshening of the North Atlantic in the

eastern subpolar gyre. J. Geophys. Res. 110, C05008.

doi:10.1029/2004JC002521.

Joyce, T.M., Hernandez-Guerra, A., Smethie, W.M., 2001.

Zonal circulation in the NW Atlantic and Caribbean

from a meridional World Ocean Circulation Experiment

hydrographic section at 66 W. J. Geophys. Res. 106,

22095e22113.

Joyce, T.M., Warren, B.A., Talley, L.D., 1986. The geothermal

heating of the abyssal subarctic Pacific Ocean. Deep-Sea

Res. 33, 1003e1015.

Joyce, T.M., Zenk, W., Toole, J.M., 1978. The anatomy of the

Antarctic Polar Front in the Drake Passage. J. Geophys.

Res. 83, 6093e6114.

Juliano, M.F., Alvés, M.L.G.R., 2007. The Subtropical Front/

Current systems of Azores and St. Helena. J. Phys.

Oceanogr. 37, 2573e2598.

Kalnay, E., Kanamitsu, M., Kistler, R., Collins, W., Deaven, D.,

Gandin, L., et al., 1996. The NCEP-NCAR 40-year reanalysis

project. Bull. Am. Meteorol. Soc. 77, 437e471.

524

REFERENCES

Kaneko, I., Takatsuki, Y., Kamiya, H., Kawae, S., 1998. Water

property and current distributions along the WHP-P9

section (137 e142 E) in the western North Pacific.

J. Geophys. Res. 103, 12959e12984.

Kanzow, T., Send, U., McCartney, M., 2008. On the variability

of the deep meridional transports in the tropical

North Atlantic. Deep-Sea Res. I 55, 1601e1623.

Kaplan, A., Cane, M., Kushnir, Y., Clement, A.,

Blumenthal, M., Rajagopalan, B., 1998. Analyses of

global sea surface temperature 1856e1991. J. Geophys.

Res. 103, 18567e18589.

Kara, A.B., Rochford, P.A., Hurlburt, H.E., 2003. Mixed layer

depth variability over the global ocean. J. Geophys. Res.

108, 3079. doi:10.1029/2000JC000736.

Karbe, L., 1987. Hot brines and the deep sea environment.

In: Edwards, A.J., Head, S.M. (Eds.), Red Sea. Pergamon

Press, Oxford, UK, 441 pp.

Karstensen, J., 2006. OMP (Optimum Multiparameter) analysis

d USER GROUP. OMP User group. http://www.ldeo.

columbia.edu/~jkarsten/omp_std/ (accessed 4.24.09).

Karstensen, J., Stramma, L., Visbeck, M., 2008. Oxygen

minimum zones in the eastern tropical Atlantic and

Pacific oceans. Progr. Oceanogr. 77, 331e350.

Kashino, Y., Watanabe, H., Herunadi, B., Aoyama, M.,

Hartoyo, D., 1999. Current variability at the Pacific

entrance of the Indonesian Throughflow. J. Geophys.

Res. 104, 11021e11035.

Kato, F., Kawabe, M., 2009. Volume transport and distribution

of deep circulation at 156 W in the North Pacific.


Kawabe, M., 1995. Variations of current path, velocity, and

volume transport of the Kuroshio in relation with the

large meander. J. Phys. Oceanogr. 25, 3103e3117.

Kawabe, M., Yanagimoto, D., Kitagawa, S., 2006. Variations

of deep western boundary currents in the Melanesian

Basin in the western North Pacific. Deep-Sea Res. I 53,

942e959.

Kawabe, M., Yanagimoto, D., Kitagawa, S., Kuroda, Y., 2005.

Variations of the deep circulation currents in the Wake

Island Passage. Deep-Sea Res. I 52, 1121e1137.

Kawai, H., 1972. Hydrography of the Kuroshio Extension.

In: Stommel, H., Yoshida, K. (Eds.), Kuroshio: Physical

Aspects of the Japan Current. University of Washington

Press, Seattle and London, pp. 235e352.

Kawano, T., Doi, T., Uchida, H., Kouketsu, S., Fukasawa, M.,

Kawai, Y., et al., 2010. Heat content change in the Pacific

Ocean between 1990s and 2000s. Deep-Sea Res. II 57,

1141e1151.

Kawano, T., Fukasawa, M., Kouketsu, S., Uchida, H., Doi, T.,

et al., 2006. Bottom water warming along the pathway of

Lower Circumpolar Deep Water in the Pacific Ocean.

Geophys. Res. Lett. 33, L23613. doi:10.1029/2006GL027933.

Kelley, D.E., Fernando, H.J.S., Gargett, A.E., Tanny, J.,

Özsoy, E., 2003. The diffusive regime of double-diffusive

convection. Progr. Oceanogr. 56, 461e481.

Kennett, J.P., 1982. Marine Geology. Prentice Hall, Englewood

Cliffs, NJ, 813 pp.

Kern, S., 2008. Polynya area in the Kara Sea, Arctic, obtained

with microwave radiometry for 1979e2003. IEEE T.

Geosc. Remote Sens. Lett. 5, 171e175.

Kessler, W.S., 2006. The circulation of the eastern tropical

Pacific: A review. Progr. Oceanogr. 69, 181e217.

Kessler, W.S.. The Central American mountain-gap winds

and their effects on the ocean. http://faculty.

washington.edu/kessler/t-peckers/t-peckers.html

(accessed 3.27.09).

Key, R.M., 2001. Ocean Process Tracers: Radiocarbon. In:

Steele, J., Thorpe, S., Turekian, K. (Eds.), Encyclopedia of

Ocean Sciences. Academic Press, Ltd., London,

pp. 2338e2353.

Kieke, D., Rhein, M., Stramma, L., Smethie, W.M.,

LeBel, D.A., Zenk, W., 2006. Changes in the CFC

inventories and formation rates of Upper Labrador Sea

Water, 1997e2001. J. Phys. Oceanogr. 36, 64e86.

Killworth, P.D., 1979. On chimney formation in the ocean.


Killworth, P.D., 1983. Deep convection in the world ocean.

Rev. Geophys. 21, 1e26.

Kimura, S., Tsukamoto, K., 2006. The salinity front in the

North Equatorial Current: A landmark for the spawning

migration of the Japanese eel (Anguilla japonica) related

to the stock recruitment. Deep-Sea Res. II 53, 315e325.

Kinder, T.H., Coachman, L.K., Galt, J.A., 1975. The Bering

Slope Current System. J. Phys. Oceanogr. 5, 231e244.

King, M.D., Menzel, W.P., Kaufman, Y.J., Tanre, D.,

Bo-Cai, G., Platnick, S., et al., 2003. Cloud and aerosol

properties, precipitable water, and profiles of temperature

and water vapor from MODIS. IEEE T. Geosc.

Remote Sens. 41, 442e458.

Klein, B., Roether, W., Civitarese, G., Gacic, M., Manca, B.B.,

d’Alcalá, M.R., 2000. Is the Adriatic returning to dominate

the production of Eastern Mediterranean Deep

Water? Geophys. Res. Lett. 27, 3377e3380.

Klein, B., Roether, W., Manca, B.B., Bregant, D., Beitzel, V.,

Kovacevic, V., et al., 1999. The large deep water transient

in the Eastern Mediterranean. Deep-Sea Res. I 46,

371e414.

Klein, S.A., Soden, B.J., Lau, N.C., 1999. Remote sea surface

temperature variations during ENSO: Evidence for

a tropical atmospheric bridge. J. Clim. 12, 917e932.

Klinck, J.M., Hofmann, E.E., Beardsley, R.C., Salihoglu, B.,

Howard, S., 2004. Water mass properties and circulation on

the west Antarctic Peninsula continental shelf in austral

fall and winter 2001. Deep-Sea Res. II 51, 1925e1946.

REFERENCES 525

Knauss, J.A., 1960. Measurements of the Cromwell Current.


Knauss, J.A., 1997. Introduction to Physical Oceanography,

second ed. Waveland Press, Long Grove, IL, 309 pp.

Knudsen, M. (Ed.), 1901. Hydrographical Tables. G.E.C.

Goad, Copenhagen, 63 pp.

Kobayashi, T., Suga, T., 2006. The Indian Ocean HydroBase:

A high-quality climatological dataset for the Indian

Ocean. Progr. Oceanogr. 68, 75e114.

Koltermann, K.P., Gouretski., V., Jancke, K., 2011. Hydrographic

Atlas of the World Ocean Circulation Experiment

(WOCE). Vol 3: Atlantic Ocean. In: Sparrow, M.,

Chapman, P., Gould, J. (Eds.). International WOCE

Project Office, Southampton, UK in press.

Komaki, K., Kawabe, M., 2009. Deep-circulation current

through the Main Gap of the Emperor Seamounts

Chain in the North Pacific. Deep-Sea Res. I 56,

305e313.

Komar, P.D., 1998. Beach Processes and Sedimentation.

Prentice Hall, Upper Saddle River, NJ, 544 pp.

Komar, P.D., Holman, R.A., 1986. Coastal processes and the

development of shoreline erosion. Annu. Rev. Earth Pl.

Sc. 14, 237e265.

Kono, T., Kawasaki, Y., 1997. Modification of the western

subarctic water by exchange with the Okhotsk Sea.


Kosro, P.M., Huyer, A., Ramp, S.R., Smith, R.L., Chavez, F.P.,

Cowles, T.J., et al., 1991. The structure of the transition

zone between coastal waters and the open ocean off

northern California, winter and spring 1987. J. Geophys.

Res. 96, 14707e14730.

Kossina, E., 1921. Die Tiefen des Weltmeeres, Veröffentil.

des Inst. für Meereskunde, Neue Folge, Heft 9, E.S.

Mittler und Sohn, Berlin (in German).

Kraus, E.B., Turner, J.S., 1967. A one-dimensional model of

the seasonal thermocline, II. The general theory and its

consequences. Tellus 19, 98e105.

Krishnamurthy, V., Kirtman, B.P., 2003. Variability of the

Indian Ocean: relation to monsoon and ENSO.Q. J. Roy.


Krummel, O., 1882. Bemerkungen über die Meeresstromungen

und Temperaturen in der Falklandsee,

Archiv der deutschen Seewarte, V(2), 25 pp. (in German).

Kuhlbrodt, T., Griesel, A., Montoya, M., Levermann, A.,

Hofmann, M., Rahmstorf, S., 2007. On the driving

processes of the Atlantic meridional overturning circulation.

Rev. Geophys. 45, RG2001. doi:10.1029/

2004RG000166.

Kunze, E., Firing, E., Hummon, J.M., Chereskin, T.K.,

Thurnherr, A.M., 2006. Global abyssal mixing inferred

from lowered ADCP shear and CTD strain profiles.


Kuo, H.-H., Veronis, G., 1973. The use of oxygen as a test for

an abyssal circulation model. Deep-Sea Res. 20, 871e888.

Kushnir, Y., Seager, R., Miller, J., Chiang, J.C.H., 2002. A

simple coupled model of tropical Atlantic decadal

climate variability. Geophys. Res. Lett. 29, 2133.

doi:10.1029/2002GL015874.

Kwon, Y.-O., Riser, S.C., 2004. North Atlantic Subtropical

Mode Water: A history of ocean-atmosphere interaction

1961e2000. Geophys. Res. Lett. 31, L19307. doi:10.1029/

2004GL021116.

Langmuir, I., 1938. Surface motion of water induced by

wind. Science 87, 119e123.

Laplace, P.S., 1790. Mémoire sur le flux et reflux de la mer.

Mém. Acad. Sci. Paris, 45e181 (in French).

Large, W.G., McWilliams, J.C., Doney, S.C., 1994. Oceanic

vertical mixing: A review and a model with a non-local

K-profile boundary layer parameterization. Rev. Geophys.

32, 363e403.

Large, W.G., Yeager, S.G., 2009. The global climatology of an

interannually varying air-sea flux data set. Clim.

Dynam. 33, 341e364.

Lavender, K.L., Davis, R.E., Owens, W.B., 2000. Mid-depth

recirculation observed in the interior Labrador and

Irminger seas by direct velocity measurements. Nature

407, 66e69.

Lavín, M.F., Marinone, S.G., 2003. An overview of the

physical oceanography of the Gulf of California. In:

Velasco Fuentes, O.U., Sheinbaum, J., Ochoa de la

Torre, J.L. (Eds.), Nonlinear Processes in Geophysical

Fluid Dynamics. Kluwer Academic Publishers, Dordrecht,

Holland, pp. 173e204.

Le Traon, P.Y., 1990. A method for optimal analysis of fields with

spatially variable mean. J. Geophys. Res. 95, 13543e13547.

Leaman, K., Johns, E., Rossby, T., 1989. The average distribution

of volume transport and potential vorticity with

temperature at three sections across the Gulf Stream.


Ledwell, J.R., Watson, A.J., Law, C.S., 1993. Evidence for

slow mixing across the pycnocline from an openeocean

tracererelease experiment. Nature 364, 701e703.

Ledwell, J.R., Watson, A.J., Law, C.S., 1998. Mixing of a tracer

in the pycnocline. J. Geophys. Res. 103, 21499e21529.

Lee, Z., Weidemann, A., Kindle, J., Arnone, R., Carder, K.L.,

Davis, C., 2007. Euphotic zone depth: Its derivation and

implication to ocean-color remote sensing. J. Geophys.

Res. 112, C03009. doi:10.1029/2006JC003802.

Leetmaa, A., Spain, P.F., 1981. Results from a velocity

transect along the equator from 125 to 159 W. J. Phys.

Oceanogr. 11, 1030e1033.

Legeckis, R., 1977. Long waves in the eastern Equatorial

Pacific Ocean: A view from a geostationary satellite.

Science 197, 1179e1181.

526

REFERENCES

Legeckis, R., Brown, C.W., Chang, P.S., 2002. Geostationary

satellites reveal motions of ocean surface fronts.

J. Marine Syst. 37, 3e15.

Legeckis, R., Reverdin, G., 1987. Long waves in the equatorial

Atlantic Ocean during 1983. J. Geophys. Res. 92,

2835e2842.

Lemke, P., Fichefet, T., Dick, C., 2011. Arctic Climate

Change d The ACSYS decade and beyond. Springer

Atmospheric and Oceanographic Sciences Library, in

preparation since 2005.

Lenn, Y.-D., Chereskin, T.K., Sprintall, J., Firing, E., 2008.

Mean jets, mesoscale variability and eddy momentum

fluxes in the surface layer of the Antarctic Circumpolar

Current in Drake Passage. J. Mar. Res. 65,

27e58.

Lentini, C.A.D., Goni, G.J., Olson, D.B., 2006. Investigation

of Brazil Current rings in the confluence region.

J. Geophys. Res. 111, C06013. doi:10.1029/2005J

C002988.

Lentz, S.J., 1995. Sensitivity of the inner-shelf circulation to

the form of the eddy viscosity profile. J. Phys. Oceanogr.

25, 19e28.

Leppäranta, M., Myrberg, K., 2009. Physical Oceanography

of the Baltic Sea. Springer, Berlin, 378 pp. with online

version.

Lerczak, J.A., 2000. Internal waves on the southern

California shelf. Ph.D. Thesis, University of California,

San Diego, 253 pp.

Levine, M.D., 2002. A modification of the Garrette

Munk internal wave spectrum. J. Phys. Oceanogr. 32,

3166e3181.

Levitus, S., 1982. Climatological Atlas of the World Ocean.

NOAA Professional Paper 13. NOAA, Rockville, MD,

173 pp.

Levitus, S., 1988. Ekman volume fluxes for the world ocean

and individual ocean basins. J. Phys. Oceanogr. 18,

271e279.

Levitus, S., Antonov, J.I., Boyer, T.P., 2005. Warming of the

world Ocean, 1955e2003. Geophys. Res. Lett. 32, L02604.

doi:10.1029/2004GL021592.

Levitus, S., Boyer, T.P., 1994. World Ocean Atlas 1994

Volume 4: Temperature. NOAA Atlas NESDIS 4. U.S.


Levitus, S., Boyer, T.P., Antonov, J., 1994a. World Ocean

Atlas Volume 5: Interannual variability of upper ocean

thermal structure. NOAA/NESDIS. Tech. Rpt. OSTI

ID:137204.

Levitus, S., Burgett, R., Boyer, T.P., 1994b. World Ocean

Atlas 1994 Volume 3: Salinity. NOAA Atlas NESDIS

3. U.S. Department of Commerce, Washington, D.C.,

99 pp.

Lewis, E.L., 1980. The practical salinity scale 1978 and its

antecedents. IEEE J. Oceanic Eng OE-5, 3e8.

Lewis, E.L., Fofonoff, N.P., 1979. A practical salinity scale.

J. Phys. Oceanogr. 9, 446.

Lewis, E.L., Perkin, R.G., 1978. Salinity: Its definition and

calculation. J. Geophys. Res. 83, 466e478.

Lewis, M.R., Kuring, N., Yentsch, C., 1988. Global patterns

of ocean transparency: Implications for the new

production of the open ocean. J. Geophys. Res. 93,

6847e6856.

Libes, S., 2009. Introduction to Marine Biogeochemistry,

Second Edition. Elsevier, Amsterdam, 909 pp.

Lien, R.-C., Gregg, M.C., 2001. Observations of turbulence in

a tidal beam and across a coastal ridge. J Geophys. Res.

106, 4575e4591.

Lighthill, J., 1978. Waves in Fluids. Cambridge University

Press, New York and London, 504 pp.

Liu, W.T., Katsaros, K.B., 2001. Air-sea fluxes from satellite

data. In: Siedler, G., Church, J. (Eds.), Ocean Circulation

and Climate, International Geophysics Series. Academic

Press, pp. 173e180.

Loeng, H., Brander, K., Carmack, E., Denisenko, S.,

Drinkwater, K., et al., 2005. Chapter 9 Marine Systems.

In: Symon, C., Arris, L., Heal, B. (Eds.), Arctic Climate

Impact Assessment d Scientific Report. Cambridge

University Press, UK, 1046 pp.

Lorenz, E., 1956. Empirical orthogonal functions and

statistical weather prediction. Scientific Report No. 1. Air

Force Cambridge Research Center, Air Research and

Development Command, Cambridge, MA, 49 pp.

Loschnigg, J., Webster, P.J., 2000. A coupled oceane

atmosphere system of SST modulation for the Indian

Ocean. J. Clim. 13, 3342e3360.

Lozier, M.S., Owens, W.B., Curry, R.G., 1995. The

climatology of the North Atlantic. Progr. Oceanogr. 36,

1e44.

Lukas, R., 1986. The termination of the Equatorial

Undercurrent in the eastern Pacific. Progr. Oceanogr.

16, 63e90.

Lukas, R., Yamagata, T., McCreary, J.P., 1996. Pacific lowlatitude

western boundary currents and the Indonesian

throughflow. J. Geophys. Res. 101, 12209e12216.

Lumpkin, R., Speer, K., 2007. Global ocean meridional

overturning. J. Phys. Oceanogr. 37, 2550e2562.

Lumpkin, R., Treguier, A.M., Speer, K., 2002. Lagrangian

eddy scales in the northern Atlantic Ocean. J. Phys.

Oceanogr. 32, 2425e2440.

Luyten, J.R., Pedlosky, J., Stommel, H., 1983. The ventilated

thermocline. J. Phys. Oceanogr. 13, 292e309.

Lynn, R.J., Reid, J.L., 1968. Characteristics and circulation

of deep and abyssal waters. Deep-Sea Res. 15,

577e598.

Lynn, R.J., Simpson, J.J., 1987. The California Current

system: The seasonal variability of its physical characteristics.

J. Geophys. Res. 92, 12947e12966.

REFERENCES 527

Maamaatuaiahutapu, K., Garçon,V.,Provost,C.,Boulahdid,M.,

Osiroff, A., 1992. Brazil-Malvinas confluence: Water mass

composition. J. Geophys. Res. 97, 9493e9505.

Maamaatuaiahutapu, K., Garçon, V., Provost, C.,

Mercier, H., 1998. Transports of the Brazil and Malvinas

Currents at their confluence. J. Mar. Res. 56, 417e438.

Macdonald, A.M., Suga, R., Curry, R.G., 2001. An isopycnally

averaged North Pacific climatology. J. Atmos.

Ocean Tech. 18, 394e420.

Macdonald, A.M., Wunsch, C., 1996. An estimate of

global ocean circulation and heat fluxes. Nature 382,

436e439.

Macdonald, R.W., Carmack, E.C., Wallace, D.W.R., 1993.

Tritium and radiocarbon dating of Canada Basin deep

waters. Science 259, 103e104.

Mackas, D.L., Denman, K.L., Bennett, A.F., 1987. Leastsquare

multiple tracer analysis of water mass composition.

J. Geophys. Res. 92, 2907e2918.

Mackas, D.L., Strub, P.T., Thomas, A., Montecino, V., 2006.

Eastern ocean boundaries pan-regional overview. In:

Robinson, A.R., Brink, K.H. (Eds.), The Sea, vol. 14A:

The Global Coastal Ocean: Interdisciplinary Regional

Studies and Syntheses. Harvard University Press,

Boston, MA, pp. 21e60.

Mackenzie, K.V., 1981. Nine-term equation for the sound

speed in the oceans. J. Acoust. Soc. Am. 70, 807e812.

Macrander, A., Send, U., Valdimarsson, H., Jónsson, S.,

Käse, R.H., 2005. Interannual changes in the overflow

from the Nordic Seas into the Atlantic Ocean through

Denmark Strait. Geophys. Res. Lett. 32, L06606.

doi:10.1029/2004GL021463.

Madden, R., Julian, P., 1994. Observations of the 40-50 day

tropical oscillation: A review. Mon. Weather Rev. 122,

814e837.

Makinson, K., Nicholls, K.W., 1999. Modeling tidal currents

beneath Filchner-Ronne Ice Shelf and on the adjacent

continental shelf: their effect on mixing and transport.

J. Geophys. Res. 104, 13449e13466.

Malanotte-Rizzoli, P., Manca, B.B., Salvatore Marullo, Ribera

d’ Alcalá, M., Roether, W., Theocharis, A., et al., 2003.

The Levantine Intermediate Water Experiment (LIWEX)

Group: Levantine basin d A laboratory for multiple

water mass formation processes. J. Geophys. Res. 108

(C9), 8101. doi:10.1029/2002JC001643.

Malmgren, F., 1927. On the properties of sea-ice. Norwegian

North Polar Expedition with the Maud, 1918e1925. Sci.

Res. 1 (5), 67 pp.

Maltrud, M.E., McClean, J.L., 2005. An eddy resolving

global 1/10 ocean simulation. Ocean Model. 8, 31e54.

Mantua, N.J., Hare, S.R., Zhang, Y., Wallace, J.M.,

Francis, R.C., 1997. A Pacific interdecadal climate oscillation

with impacts on salmon production. B. Am.

Meteor. Soc. 78, 1069e1079.

Mantyla, A.W., Reid, J.L., 1983. Abyssal characteristics of

the World Ocean waters. Deep-Sea Res. 30, 805e833.

Marchesiello, P., McWilliams, J.C., Shchepetkin, A., 2003.

Equilibrium structure and dynamics of the California

Current System. J. Phys. Oceanogr. 33, 753e783.

Mariano, A.J., Ryan, E.H., Perkins, B.D., Smithers, S., 1995.

The Mariano Global Surface Velocity Analysis 1.0. USCG

Report CG-D-34e95, p. 55. http://oceancurrents.rsmas.

miami.edu/index.html (accessed 3.4.09).

Marshall, G., 2003. Trends in the Southern Annular Mode

from observations and reanalyses. J. Clim. 16, 4134e4143.

Marshall, J., Schott, F., 1999. Open-ocean convection: observations,

theory, and models. Rev. Geophys. 37, 1e64.

Martin, S., 2001. Polynyas. In: Steele, J.H., Turkeian, K.K.,

Thorpe, S.A. (Eds.), Encyclopedia of Ocean Sciences.

Academic Press, pp. 2243e2247.

Martin, S., Cavalieri, D.J., 1989. Contributions of the Siberian

shelf polynyas to the Arctic Ocean intermediate and

deep water. J. Geophys. Res. 94, 12725e12738.

Martinson, D.G., Steele, M., 2001. Future of the Arctic sea ice

cover: Implications of an Antarctic analog. Geophys.

Res. Lett. 28, 307e310.

Maslanik, J., Serreze, M., Agnew, T., 1999. On the record

reduction in 1998 western Arctic sea-ice cover. Geophys.

Res. Lett. 26, 1905e1908.

Masuzawa, J., 1969. Subtropical Mode Water. Deep-Sea Res.

16, 453e472.

Mata, M.M., Wijffels, S.E., Church, J.A., Tomczak, M., 2006.

Eddy shedding and energy conversions in the East

Australian Current. J. Geophys. Res. 111, C09034.

doi:10.1029/2006JC003592.

Mauritzen, C., 1996. Production of dense overflow waters

feeding the North Atlantic across the Greenland-Scotland

Ridge. Part 1: Evidence for a revised circulation

scheme. Deep-Sea Res. I 43, 769e806.

Maury, M.F., 1855. The Physical Geography of the Sea.

Harper and Brothers, New York, 304 pp.

Maximenko, N., Niiler, P., Rio, M.H., Melnichenko, O.,

Centurioni, L., Chambers, D., et al., 2009. Mean dynamic

topography of the ocean derived from satellite and

drifting buoy data using three different techniques.

J. Atmos. Ocean. Tech. 26, 1910e1919.

McCarthy, M.C., Talley, L.D., 1999. Three-dimensional isoneutral

potential vorticity structure in the Indian Ocean.

J. Geophys. Res. 104, 13251e13268.

McCartney, M., Curry, R., 1993. Transequatorial flow of

Antarctic Bottom Water in the western Atlantic Ocean:

Abyssal geostrophy at the equator. J. Phys. Oceanogr. 23,

1264e127.

McCartney, M.S., 1977. Subantarctic Mode Water. In: Angel,

M.V. (Ed.), A Voyage of Discovery: George Deacon

70th Anniversary Volume, supplement to Deep-Sea Res.,

pp. 103e119.

528

REFERENCES

McCartney, M.S., 1982. The subtropical circulation of Mode

Waters. J. Mar. Res. 40 (Suppl.), 427e464.

McCartney, M.S., Talley, L.D., 1982. The Subpolar Mode

Water of the North Atlantic Ocean. J. Phys. Oceanogr. 12,

1169e1188.

McClain, C., Christian, J.R., Signorini, S.R., Lewis, M.R.,

Asanuma, I., Turk, D., et al., 2002. Satellite ocean-color

observations of the tropical Pacific Ocean. Deep-Sea Res.

II 49, 2533e2560.

McClain, C., Hooker, S., Feldman, G., Bontempi, P., 2006.

Satellite data for ocean biology, biogeochemistry, and

climate research. Eos Trans. AGU 87 (34), 337e343.

McDonagh, E.L., Bryden, H.L., King, B.A., Sanders, R.J.,

Cunningham, S.A., Marsh, R., 2005. Decadal changes in

the south Indian Ocean thermocline. J. Clim. 18,

1575e1590.

McDougall, T.J., 1987a. Neutral surfaces. J. Phys. Oceanogr.

17, 1950e1964.

McDougall, T.J., 1987b. Thermobaricity, cabbeling, and

water-mass conversion. J. Geophys. Res. 92, 5448e5464.

McDougall, T.J., Jackett, D.R., Millero, F.J., 2010. An algorithm

for estimating Absolute Salinity in the global

ocean. Submitted to Ocean Science, a preliminary

version is available at Ocean Sci. Discuss. 6, 215e242.

http://www.ocean-sci-discuss.net/6/215/2009/osd-6-

215-2009-print.pdf and the computer software is available

from http://www.TEOS-10.org.

McPhaden, M.J., Busalacchi, A.J., Cheney, R., Donguy, J.-R.,

Gage, K.S., Halpern, D., et al., 1998. The Tropical Ocean-

Global Atmosphere observing system: A decade of

progress. J. Geophys. Res. 103, 14169e14240.

MEDOC Group, 1970. Observations of formation of deepwater

in the Mediterranean Sea, 1969. Nature 227,

1037e1040.

Mecking, S., Warner, M.J., 1999. Ventilation of Red Sea Water

with respect to chlorofluorocarbons. J. Geophys. Res.

104, 11087e11097.

Meehl, G.A., Stocker, T.F., Collins, W.D., Friedlingstein, P.,

Gaye, A.T., Gregory, J.M., et al., 2007. Global climate

projections. In: Solomon, S., Qin, D., Manning, M.,

Chen, Z., Marquis, M., Averyt, K.B., et al. (Eds.), Climate

Change 2007: The Physical Science Basis. Contribution of

Working Group I to the Fourth Assessment Report of the

Intergovernmental Panel on Climate Change. Cambridge

University Press, Cambridge, UK and New York.

Mei, C.C., Stiassnie, M., Yue, D.K.-P., 2005. Theory and

Applications of Ocean Surface Waves: Part I, Linear

Aspects; Part II, Nonlinear Aspects. World Scientific,

New Jersey and London, 1136 pp.

Meinen, C.S., Watts, D.R., 1997. Further evidence that the

sound-speed algorithm of Del Grosso is more accurate

than that of Chen and Millero. J. Acoust. Soc. Am. 102,

2058e2062.

Meinen, C.S., Watts, D.R., 2000. Vertical structure and

transport on a transect across the North Atlantic Current

near 42 N: Time series and mean. J. Geophys. Res. 105,

21869e21891.

Menard, H.W., Smith, S.M., 1966. Hypsometry of ocean

basin provinces. J. Geophys. Res. 71, 4305e4325.

Meredith, J.P., Watkins, J.L., Murphy, E.J., Ward, P.,

Bone, D.G., Thorpe, S.E., et al., 2003. Southern ACC

Front to the northeast of South Georgia: Pathways,

characteristics and fluxes. J. Geophys. Res. (C5), 108.

doi:10.1029/2001JC001227.

Meredith, M.P., Hogg, A.M., 2006. Circumpolar response of

Southern Ocean eddy activity to a change in the

Southern Annular Mode. Geophys. Res. Lett. 33, L16608.

doi:10.1029/2006GL026499.

Merz, A., Wüst, G., 1922. Die Atlantische Vertikal Zirkulation.

Z. Ges. Erdkunde Berlin 1, 1e34 (in German).


3 Beitrag. Zeitschr. D.G.F.E. Berlin (in German).

Middleton, J.F., Cirano, M., 2002. A northern boundary

current along Australia’s southern shelves: The

Flinders Current. J. Geophys. Res. 107. doi:10.1029/

2000JC000701.

Millero, F.J., 1967. High precision magnetic float densimeter.

Rev. Sci. Instrum. 38, 1441e1444.

Millero, F.J., 1978. Freezing point of seawater, Eighth report

of the Joint Panel of Oceanographic Tables and Standards.

Appendix 6. UNESCO Tech. Papers.

Millero, F.J., Feistel, R., Wright, D.G., McDougall, T.J., 2008.

The composition of Standard Seawater and the definition

of the reference-composition salinity scale. Deep-

Sea Res. I 55, 50e72.

Millero, F.J., Perron, G., Desnoyers, J.E., 1973. The heat

capacity of seawater solutions from 5 to 35 C and from

0.5 to 22% chlorinity. J. Geophys. Res. 78, 4499e4507.

Millero, F.J., Poisson, A., 1980. International one-atmosphere

equation of state of seawater. Deep-Sea Res. 28, 625e629.

Millot, C., 1991. Mesoscale and seasonal variabilities of the

circulation in the western Mediterranean. Dynam.

Atmos. Oceans 15, 179e214.

Millot, C., Taupier-Letage, I., 2005. Circulation in the

Mediterranean Sea. In: Saliot, E.A. (Ed.), The Handbook

of Environmental Chemistry, vol. 5. Part K. Springer-

Verlag, Berlin Heidelberg, pp. 29e66.

Mills, E.L., 1994. Bringing oceanography into the Canadian

university classroom. Scientia Canadensis. Can. J. Hist.

Sci. Tech. Med. 18, 3e21.

Mittelstaedt, E., 1991. The ocean boundary along the northwest

African coast: Circulation and oceanographic properties

at the sea surface. Progr. Oceanogr. 26, 307e355.

Mizuno, K., White, W.B., 1983. Annual and interannual

variability in the Kuroshio current system. J. Phys.

Oceanogr. 13, 1847e1867.

REFERENCES 529

Mobley, C.D., 1995. Optical properties of water. In: Bass, M.,

Van Stryland, E.W., Williams, D.R., Wolfe, W.L. (Eds.),

Handbook of Optics, Vol. 1, Fundamentals, Techniques,

and Design. McGraw-Hill 43.1e43.56.

Molinari, R.L., Fine, R.A., Wilson, W.D., Curry, R.G., Abell, J.,

McCartney, M.S., 1998. The arrival of recently formed

Labrador Sea Water in the Deep Western Boundary

Current at 26.5 N. Geophys. Res. Lett. 25, 2249e2252.

Monismith, S.G., 2007. Hydrodynamics of coral reefs. Annu.

Rev. Fluid Mech. 39, 37e55.

Montecino, V., Strub, P.T., Chavez, F., Thomas, A.,

Tarazona, J., Baumgartner, T., 2006. Biophysical interactions

off western South-America. In: Robinson, A.R.,

Brink, K.H. (Eds.), The Sea, vol. 14A: The Global Coastal

Ocean: Interdisciplinary Regional Studies and Syntheses.

Harvard University Press, Boston, MA, pp. 329e390.

Montgomery, R.B., 1938. Circulation in the upper layers of

the Southern North Atlantic deduced with the use of

isentropic analysis. Papers Phys. Oceanogr. and Met. 6,

MIT and Woods Hole Oceanographic Institution, 55 pp.

Montgomery, R.B., 1958. Water characteristics of Atlantic

Ocean and of world ocean. Deep-Sea Res. 5, 134e148.

Montgomery, R.B., Stroup, E.D., 1962. Equatorial waters and

currents at 150 W in July-August 1952. Johns Hopkins

Oceanographic Study, No.1, 68 pp.

Morawitz, W.M.L., Cornuelle, B.D., Worcester, P.F., 1996. A case

study in three-dimensional inverse methods: combining

hydrographic, acoustic, and moored thermistor data in the

Greenland Sea. J. Atm. Oceanic Tech. 13, 659e679.

Morawitz, W.M.L., Sutton, P.J., Worcester, P.F.,

Cornuelle, B.D., Lynch, J.F., Pawlowicz, R., 1996. Threedimensional

observations of a deep convective chimney

in the Greenland Sea during winter 1988/1989. J. Phys.

Oceanogr. 26, 2316e2343.

Morel, A., Antoine, D., 1994. Heating rate within the upper

ocean in relation to its bio-optical state. J. Phys. Oceanogr.

24, 1652e1665.

Morrow, R., Birol, F., 1998. Variability in the southeast

Indian ocean from altimetry: Forcing mechanisms for the

Leeuwin Current. J. Geophys. Res. 103, 18529e18544.

Morrow, R., Donguy, J.-R., Chaigneau, A., Rintoul, S.R.,

2004. Cold-core anomalies at the subantarctic front,

south of Tasmania. Deep-Sea Res. I, 1417e1440.

Moum, J.N., Farmer, D.M., Smyth, W.D., Armi, L., Vagle, S.,

2003. Structure and generation of turbulence at interfaces

strained by internal solitary waves propagating

shoreward over the continental shelf. J. Phys. Oceanogr.

33, 2093e2112.

Müller, R.D., Sdrolias, M., Gaina, C., Roest, W.R., 2008. Age,

spreading rates and spreading symmetry of the world’s

ocean crust. Geochem. Geophys. Geosyst. 9, Q04006.

http://www.ngdc.noaa.gov/mgg/ocean_age/. doi:10/

1029/2007GC001743(accessed 2.01.09).

Munk, W., 1966. Abyssal recipes. Deep-Sea Res. 13, 707e730.

Munk, W., 1981. Internal waves and small-scale processes. In:

Warren, B.A., Wunsch, C. (Eds.), Evolution of Physical

Oceanography. The MIT Press, Boston, MA, pp. 264e290.

Munk, W., Wunsch, C., 1982. Observing the ocean in the

1990’s: a scheme for large-scale monitoring. Philos.

Trans. Roy. Soc. A 307, 439e464.

Murray, J.W., Jannasch, H.W., Honjo, S., Anderson, R.F.,

Reeburgh, W.S., Top, Z., et al., 1989. Unexpected changes

in the oxic/anoxic interface in the Black Sea. Nature 338,

411e413.

Naimie, C.E., Blain, C.A., Lynch, D.R., 2001. Seasonal mean

circulation in the Yellow Sea d A model-generated

climatology. Continental Shelf Res. 21, 667e695.

Nakano, H., Suginohara, N., 2002. Importance of the eastern

Indian Ocean for the abyssal Pacific. J. Geophys. Res.

107, 3219. doi:10.1029/2001JC001065.

Nansen, F., 1922. In: Nacht und Eis. F.U. Brodhaus, Leipzig,

Germany, 355 pp. (in German).

NASA, 2009a. Ocean color from space: global seasonal

change. NASA Goddard Earth Sciences Data and

Information Services Center. http://daac.gsfc.nasa.gov/

oceancolor/scifocus/space/ocdst_global_seasonal_

change.shtml (accessed 2.18.09).

NASA, 2009b. Ocean Color Web. NASA Goddard Space

Flight Center. http://oceancolor.gsfc.nasa.gov/ (accessed

2.18.09).

NASA Earth Observatory, 2010. Global maps. NASA Goddard

Space Flight Center. http://earthobservatory.nasa.

gov/GlobalMaps/ (accessed 12.13.10).

NASA Goddard Earth Sciences, 2007a. An assessment of the

Indian Ocean, Monsoon, and Somali Current using

NASA’s AIRS, MODIS, and QuikSCAT data. NASA

Goddard Earth Sciences Data Information Services

Center. http://daac.gsfc.nasa.gov/oceancolor/scifocus/

modis/IndianMonsoon.shtml (accessed 7.1.08).

NASA Goddard Earth Sciences, 2007b. Sedimentia. NASA

Goddard Earth Sciences Ocean Color. http://disc.gsfc.

nasa.gov/oceancolor/scifocus/oceanColor/sedimentia.

shtml (accessed 4.3.09).

NASA Goddard Earth Sciences, 2008. Ocean color: classic

CZCS scenes, Chapter 4. NASA Goddard Earth Sciences

Data Information Services Center. http://disc.gsfc.nasa.

gov/oceancolor/scifocus/classic_scenes/04_classics_

arabian.shtml (accessed 1.9.09).

NASA Visible Earth, 2006. Visible Earth: Sun glint in the

Mediterranean Sea. NASA Goddard Space Flight

Center. http://visibleearth.nasa.gov/view_rec.php?id¼732


NASA Visible Earth, 2008. Eddies off the Queen Charlotte

Islands. NASA Goddard Space Flight Center. http://

visibleearth.nasa.gov/view_rec.php?id¼2886 (accessed

3.26.09).

530

REFERENCES

National Data Buoy Center, 2006. How are estimates of

wind-seas and swell made from NDBC wave data?

NOAA/NDBC. http://www.ndbc.noaa.gov/windsea.


National Data Buoy Center, 2009. NDBC Web Site. NOAA/

NDBC. http://www.ndbc.noaa.gov/ (accessed 5.15.09).

National Research Council, 2010. Ocean acidification: A

national strategy to meet the challenges of a changing

ocean. National Academies Press, Washington D.C.,

152 pp.

Naval Postgraduate School, 2003. Basic concepts in

physical oceanography: tides. Navy Operational

Ocean Circulation and Tide Models. Department of

Oceanography, Naval Postgraduate School. http://

www.oc.nps.edu/nom/day1/partc.html (accessed

3.30.09).

Neilson, B.J., Kuo, A., Brubaker, J., 1989. Estuarine Circulation.

Humana Press, Clifton, N.J, 377 pp.

New, A.L., Jia, Y., Coulibaly, M., Dengg, J., 2001. On the role

of the Azores Current in the ventilation of the North

Atlantic Ocean. Progr. Oceanogr. 48, 163e194.

Niiler, P.P., Maximenko, N.A., McWilliams, J.C., 2003.

Dynamically balanced absolute sea level of the

global ocean derived from near-surface velocity observations.

Geophys. Res. Lett. 30, 22. doi:10.1029/

2003GL018628.

Nilsson, C.S., Cresswell, G.R., 1981. The formation and

evolution of East Australian Current warm-core eddies.


NOAA, 2008. Tides and Water Levels. NOAA Ocean Service

Education. http://oceanservice.noaa.gov/education/

kits/tides/welcome.html (accessed 3.29.09).

NOAA, 2009. Arctic Change: Climate indicators d Arctic

Oscillation. NOAA Arctic. http://www.arctic.noaa.gov/

detect/climate-ao.shtml (accessed 3.17.09).

NOAA AOML PHOD, 2009. The Global Drifter Program.

NOAA AOML. http://www.aoml.noaa.gov/phod/

dac/gdp.html (accessed 9.09).

NOAA CO-OPS, 2010. Tides and Currents. NOAA/

National Ocean Service. http://co-ops.nos.noaa.gov/

index.shtml (accessed 10.26.10).

NOAA CPC, 2005. Madden/Julian Oscillation (MJO).

NOAA/National Weather Service. http://www.cpc.

ncep.noaa.gov/products/precip/CWlink/MJO/mjo.shtml


NOAA ESRL, 2009. Linear correlations in atmospheric

seasonal/monthly averages. NOAA Earth System Research

Laboratory Physical Sciences Division. http://www.cdc.

noaa.gov/data/correlation/ (accessed 10.30.09).

NOAA ESRL, 2010. PSD Map Room Climate Products

Outgoing Longwave Radiation (OLR). NOAA ESRL

PSD. http://www.cdc.noaa.gov/map/clim/olr.shtml


NOAA National Weather Service, 2005. Hydrometeorological

Prediction Center (HPC) Home Page. National

Weather Service. http://www.hpc.ncep.noaa.gov/ (accessed

1.3.05).

NOAA NESDIS, 2009. Ocean Products Page. NOAA/NES-

DIS/OSDPD. http://www.osdpd.noaa.gov/PSB/EPS/

SST/SST.html (accessed 2.18.09).

NOAA NGDC, 2008. Global Relief Data d ETOPO. NOAA

National Geophysical Data Center. http://www.ngdc.

noaa.gov/mgg/global/global.html (accessed 9.24.08).

NOAA PMEL TAO Project Office, 2009a. The TAO project.

TAO Project Office, NOAA Pacific Marine Environmental

Laboratory. http://www.pmel.noaa.gov/tao/

(accessed 6.1.09).

NOAA PMEL TAO Project Office, 2009b. El Niño theme page:

access to distributed information on El Niño. NOAA Pacific

Marine Environmental Laboratory. http://www.

pmel.noaa.gov/tao/elnino/nino-home.html (accessed

3.26.09).

NOAA PMEL, 2009c. Global tropical moored array. NOAA

Pacific Marine Environmental Laboratory. http://www.

pmel.noaa.gov/tao/global/global.html (accessed

5.20.09).

NOAA PMEL, 2009d. Impacts of El Niño and benefits of El

Niño prediction. NOAA Pacific Marine Environmental

Laboratory. http://www.pmel.noaa.gov/tao/elnino/

impacts.html (accessed 3.26.09).

NOAAWavewatch III, , 2009. NCEP MMAB operational wave

models. NOAA/NWS Environmental Modeling Center/

Marine Modeling and Analysis Branch. http://polar.ncep.

noaa.gov/waves/index2.shtml (accessed 5.14.09).

NODC, 2005a. World Ocean Atlas 2005 (WOA05). NOAA

National Oceanographic Data Center. http://www.

nodc.noaa.gov/OC5/WOA05/pr_woa05.html (accessed

4.28.09).

NODC, 2005b. World Ocean Database 2005 (WOD05).

NOAA National Oceanographic Data Center. http://

www.nodc.noaa.gov/OC5/WOD05/pr_wod05.html

(accessed 4.28.09).

NODC, 2009. Data sets and products, National Oceanographic

Data Center Ocean Climate Laboratory. http://www.

nodc.noaa.gov/OC5/indprod.html (accessed 12.15.09).

Nowlin, W.D., Whitworth, T., Pillsbury, R.D., 1977. Structure

and transport of the Antarctic Circumpolar Current at

Drake Passage from short-term measurements. J. Phys.

Oceanogr. 7, 788e802.

NSIDC, 2007. Arctic sea ice news fall 2007. National Snow

and Ice Data Center. http://nsidc.org/arcticseaicenews/


NSIDC, 2008a. Polar Pathfinder Daily 25 km EASE-Grid Sea

Ice Motion Vectors. National Snow and Ice Data Center.

http://nsidc.org/data/docs/daac/nsidc0116_icemotion.

gd.html (accessed 02.01.09).

REFERENCES 531

NSIDC, 2008b. Arctic sea ice down to second-lowest extent;

likely record-low volume. National Snow and Ice Data

Center. http://nsidc.org/news/press/20081002_seaice_

pressrelease.html (accessed 3.17.09).

NSIDC, 2009a. Cryospheric climate indicators. National

Snow and Ice Data Center. http://nsidc.org/data/

seaice_index/archives/index.html (accessed 2.25.09).

NSIDC, 2009b. Arctic climatology and meteorology primer.

National Snow and Ice Data Center. http://nsidc.org/

arcticmet/ (accessed 3.1.09).

NSIDC, 2009c. Images of Antarctic Ice Shelves. National


iceshelves_images/index.html (accessed 3.5.09).

NWS Internet Services Team, 2008. ENSO temperature and

precipitation composites. http://www.cpc.noaa.gov/

products/precip/CWlink/ENSO/composites/EC_LNP_


O’Connor, B.M., Fine, R.A., Olson, D.B., 2005. A global

comparison of subtropical underwater formation rates.

Deep-Sea Res. I. 52, 1569e1590.

Ochoa, J., Bray, N.A., 1991. Water mass exchange in the Gulf

of Cadiz. Deep-Sea Res. 38 (Suppl.), S465eS503.

ODV, 2009. Ocean Data View. Alfred Wegener Institute.

http://odv.awi.de/en/home/ (accessed 4.28.09).

Officer, C.B., 1976. Physical Oceanography of Estuaries

(and Associated Coastal Waters). Wiley, New York.

465 pp.

Oguz, T., Tugrul, S., Kideys, A.E., Ediger, V., Kubilay, N.,

2006. Physical and biogeochemical characteristics of the

Black Sea. In: Robinson, A.R., Brink, K.H. (Eds.), The Sea,

Vol., 14A: The Global Coastal Ocean: Interdisciplinary

Regional Studies and Syntheses. Harvard University

Press, Boston, MA, pp. 1333e1372.

Olbers, D.J.M., Wenzel, M., Willebrand, J., 1985. The

inference of North Atlantic circulation patterns from

climatological hydrographic data. Rev. Geophys. 23,

313e356.

Olsen, S.M., Hansen, B., Quadfasel, D., Østerhus, S., 2008.

Observed and modeled stability of overflow across the

Greenland-Scotland ridge. Nature 455, 519e523.

Olson, D., Schmitt, R., Kennelly, M., Joyce, T., 1985. A twolayer

diagnostic model of the long-term physical

evolution of warm-core ring 82B. J. Geophys. Res. 90,

8813e8822.

Oort, A.H., Vonder Haar, T.H., 1976. On the observed

annual cycle in the ocean-atmosphere heat balance over

the northern hemisphere. J. Phys. Oceanogr. 6, 781e800.

Open University, 1999. Waves, Tides and Shallow-

Water Processes, second ed. Butterworth-Heinemann,

Burlington, MA, 228 pp.

Orsi, A.H., Johnson, G.C., Bullister, J.L., 1999. Circulation,

mixing, and production of Antarctic Bottom Water.


Orsi, A.H., Nowlin, W.D., Whitworth, T., 1993. On the

circulation and stratification of the Weddell Gyre. Deep-

Sea Res. I. 40, 169e203.

Orsi, A., Whitworth, T., 2005. Hydrographic Atlas of the

World Ocean Circulation Experiment (WOCE). Volume

1: Southern Ocean. In: Sparrow, M., Chapman, P.,

Gould, J. (Eds.). International WOCE Project Office,

Southampton, UK ISBN 0-904175-49-9. http://

woceatlas.tamu.edu/ (accessed 4.20.09).

Orsi, A.H., Whitworth, T., Nowlin, W.D., 1995. On the

meridional extent and fronts of the Antarctic Circumpolar

Current. Deep-Sea Res. I. 42, 641e673.

Osborn, T.R., Cox, C.S., 1972. Oceanic fine structure. Geophys.

Astrophys. Fluid Dynam. 3, 321e345.

Osborne, J., Swift, J.H., 2009. Java OceanAtlas. http://odf.

ucsd.edu/joa/ (accessed 4.20.09).

Østerhus, S., Gammelsrød, T., 1999. The abyss of the Nordic

Seas is warming. J. Clim. 12, 3297e3304.

Östlund, H., Possnert, G., G., Swift, J., 1987. Ventilation rate

of the deep Arctic Ocean from carbon 14 data. J. Geophys.

Res. 92, 3769e3777.

Owens, W.B., 1991. A statistical description of the mean

circulation and eddy variability in the northwestern

Atlantic using SOFAR floats. Progr. Oceanogr. 28, 257e303.

Owens, W.B., Warren, B.A., 2001. Deep circulation in the

northwest corner of the Pacific Ocean. Deep-Sea Res. I

48, 959e993.

Özsoy, E., Hecht, A., Ünlüata, Ü., Brenner, S., Oguz, T.,

Bishop, J., et al., 1991. A review of the Levantine Basin

circulation and its variability during 1985e1988. Dynam.


Özsoy, E., Ünlüata, U., 1998. The Black Sea. In:

Robinson, A.R., Brink, K.H. (Eds.), The Sea, Vol. 11: The

Global Coastal Ocean: Regional Studies and Syntheses.

Harvard University Press, Boston, MA, pp. 889e914.

Palacios, D.M., Bograd, S.J., 2005. A census of Tehuantepec

and Papagayo eddies in the northeastern tropical

Pacific. Geophys. Res. Lett. 32, L23606. doi: 10.1029/

2005GL024324.

Palastanga, V., van Leeuwen, P.J., Schouten, M.W.,

deRuijter, P.M., 2007. Flow structure and variability in

the subtropical Indian ocean: instability of the South

Indian Ocean Countercurrent. J. Geophys. Res. 112,

C01001. doi:10.1029/2005JC003395.

Parker, C.E., 1971. Gulf Stream rings in the Sargasso Sea.


Paulson, C.A., Simpson, J.J., 1977. Irradiance measurements

in the upper ocean. J. Phys. Oceanogr. 7, 952e956.

Payne, R.E., 1972. Albedo of the sea surface. J. Atmos. Sci.

29, 959e970.

Pearce, A.F., 1991. Eastern boundary currents of the

southern hemisphere. J. Roy. Soc. Western Austral. 74,

35e45.

532

REFERENCES

Pedlosky, J., 1987. Geophysical Fluid Dynamics, second ed.

Springer-Verlag, New York, 732 pp.

Pedlosky, J., 2003. Waves in the Ocean and Atmosphere.

Springer-Verlag, Berlin, 260 pp.

Peeters, F.J.C., Acheson, R., Brummer, G.-J.A.,

de Ruijter, W.P.M., Schneider, R.R., Ganssen, G.M., et al.,

2004. Vigorous exchange between the Indian and

Atlantic oceans at the end of the past five glacial periods.

Nature 430, 661e665.

Pelegrí, J.L., Arístegui, J., Cana, L., González-Dávila, M.,

Hernández-Guerra, A., Hernández-León, S., et al., 2005.

Coupling between the open ocean and the coastal

upwelling region off northwest Africa: Water recirculation

and offshore pumping of organic matter. J. Marine

Syst. 54, 3e37.

Perry, R.K., 1986. Bathymetry. In: Hurdle, B. (Ed.), The

Nordic Seas. Springer-Verlag, New York, pp. 211e236.

Peterson, R.G., 1988. On the transport of the Antarctic

Circumpolar Current through Drake Passage and its

relation to wind. J. Geophys. Res. 93, 13993e14004.

Peterson, R.G., 1992. The boundary currents in the western

Argentine Basin. Deep-Sea Res. 39, 623e644.

Peterson, R.G., Stramma, L., Kortum, G., 1996. Early concepts

and charts of ocean circulation. Progr. Oceanogr. 37, 1e115.

Philander, S.G.H., 1978. Instabilities of zonal equatorial

currents: II. J. Geophys. Res. 83, 3679e3682.

Philander, S.G.H., Fedorov, A., 2003. Is El Niño sporadic or

cyclic? Annu. Rev. Earth Pl. Sci. 31, 579e594.

Phillips, O.M., 1977. The Dynamics of the Upper Ocean.

Cambridge University Press, Cambridge, UK, 336 pp.

Pickard, G.L., 1961. Oceanographic features of inlets in the

British Columbia mainland coast. J. Fish. Res. Bd. Can.

18, 907e999.

Pickard, G.L., Donguy, J.R., Hénin, C., Rougerie, F., 1977. A

review of the physical oceanography of the Great Barrier

Reef and western Coral Sea. Australian Institute of

Marine Science, 2. Australian Government Publishing

Service, Canberra, 134 pp.

Pickard, G.L., Stanton, B.R., 1980. Pacific fjords d A review

of their water characteristics. In: Freeland, H.J.,

Farmer, D.M., Levings, C.D. (Eds.), Fjord Oceanography.

Plenum Press, pp. 1e51.

Pickart, R.S., McKee, T.K., Torres, D.J., Harrington, S.A.,

1999. Mean structure and interannual variability of the

slopewater system south of Newfoundland. J. Phys.

Oceanogr. 29, 2541e2558.

Pickart, R.S., Smethie, W.M., 1993. How does the Deep

Western Boundary Current cross the Gulf Stream?


Pickart, R.S., Torres, D.J., Clarke, R.A., 2002. Hydrography

of the Labrador Sea during active convection. J. Phys.


Pickett, M.H., Paduan, J.D., 2003. Ekman transport and

pumping in the California Current based on the U.S. Navy’s

high-resolution atmospheric model (COAMPS).

J. Geophys. Res. 108, 3327. doi:10.1029/2003JC001902.

Polton, J.A., Smith, J.A., MacKinnon, J.A., Tejada-

Martínez, A.E., 2008. Rapid generation of highfrequency

internal waves beneath a wind and wave

forced oceanic surface mixed layer. Geophys. Res. Lett.

35, L13602. doi:10.1029/2008GL033856.

Polyakov, I.V., 22 co-authors, 2005. One more step toward

a warmer Arctic. Geophys. Res. Lett. 32, L17605.

doi:10.1029/2005GL023740.

Polyakov, I.V., 17 co-authors, 2010. Arctic Ocean warming

contributes to reduced polar ice cap. J. Phys. Oceanogr

40, 2743e2756.

Polyakov, I.V., Alekseev, G.V., Timokhov, L.A., Bhatt, U.S.,

Colony, R.L., Simmons, H.L., et al., 2004. Variability of

the intermediate Atlantic Water of the Arctic Ocean over

the last 100 years. J. Clim. 17, 4485e4497.

Polzin, K.L., Toole, J.M., Ledwell, J.R., Schmitt, R.W., 1997.

Spatial variability of turbulent mixing in the abyssal

ocean. Science 276, 93e96.

Pond, S., Pickard, G.L., 1983. Introductory Dynamical Oceanography,

second ed. Pergamon Press, Oxford, 329 pp.

Poole, R., Tomczak, M., 1999. Optimum multiparameter

analysis of the water mass structure in the Atlantic

Ocean thermocline. Deep-Sea Res. I 46, 1895e1921.

Potter, R.A., Lozier, M.S., 2004. On the warming and salinification

of the Mediterranean outflow waters in the

North Atlantic. Geophys. Res. Lett. 31, L01202.

doi:10.1029/2003GL018161.

Press, W.H., Flannery, B.P., Teukolsky, S.A., Vetterline, W.T.,

1986. Numerical Recipes. Cambridge University Press,

Cambridge, UK, 818 pp.

Price, J.F., Baringer, M.O., 1994. Outflows and deep water

production by marginal seas. Progr. Oceanogr. 33, 161e200.

Price, J.F., Weller, R.A., Pinkel, R., 1986. Diurnal cycling:

Observations and models of the upper ocean response to

diurnal heating, cooling and wind mixing. J. Geophys.

Res. 91, 8411e8427.

Pritchard, D.W., 1989. Estuarine classification d ahelpor

ahindrance.In:Neilson,B.J.,Kuo,A.,Brubaker,J.(Eds.),

Estuarine Circulation. Humana Press, Clifton, N.J, pp. 1e38.

Proshutinsky, A.Y., Johnson, M.A., 1997. Two circulation

regimes of the wind-driven Arctic Ocean. J. Geophys.

Res. 102, 12493e12514.

Provost, C., Escoffier, C., Maamaatuaiahutapu, K.,

Kartavtseff, A., Garçon, V., 1999. Subtropical mode

waters in the South Atlantic Ocean. J. Geophys. Res. 104,

21033e21049.

Pugh, D.T., 1987. Tides, Surges, and Mean Sea-level. J. Wiley,

Chichester, UK, 472 pp.

REFERENCES 533

Purkey, S.G., Johnson, G.C., 2010. Antarctic bottom water

warming between the 1990s and 2000s: Contributions to

global heat and sea level rise budgets. J. Clim. 23,

6336e6351.

Qiu, B., Chen, S., 2005. Variability of the Kuroshio Extension

jet, recirculation gyre, and mesoscale eddies on decadal

time scales. J. Phys. Oceanogr. 35, 2090e2103.

Qiu, B., Huang, R.X., 1995. Ventilation of the North Atlantic

and North Pacific: Subduction versus obduction. J. Phys.

Oceanogr. 25, 2374e2390.

Qiu, B., Miao, W., 2000. Kuroshio path variations south of

Japan: Bimodality as a self-sustained internal oscillation.


Qiu, B., Scott, R.B., Chen, S., 2008. Length scales of eddy

generation and nonlinear evolution of the seasonally

modulated South Pacific Subtropical Countercurrent.


Qu, T., Lindstrom, E., 2002. A climatological interpretation

of the circulation in the western South Pacific. J. Phys.

Oceanogr. 32, 2492e2508.

Quartly, G.D., Srokosz, M.A., 1993. Seasonal variations in

the region of the Agulhas retroflection: Studies with

Geosat and FRAM. J. Phys. Oceanogr. 23, 2107e2124.

Quartly, G.D., Srokosz, M.A., 2003. Satellite observations of

the Agulhas Current system. Philos. Trans. Roy. Soc. A

361, 51e56.

Ralph, E.A., Niiler, P.P., 1999. Wind-driven currents in the

tropical Pacific. J. Phys. Oceanogr. 29, 2121e2129.

Ramanathan, V., Collins, W., 1991. Thermodynamic

regulation of ocean warming by cirrus clouds deduced

from observations of the 1987 El Nino. Nature 351,

27e32.

Rasmusson, E.M., Carpenter, T.H., 1982. Variations in tropical

sea surface temperature and surface wind fields

associated with the outer Oscillation/El Niño. Mon.

Weather Rev. 110, 354e384.

Ray, R.D., 1999. A global ocean tide model from TOPEX/

POSEIDON altimetry: GOT99.s. NASA/TM -1999e209478,

58 pp.

Redfield, A.C., 1934. On the proportion of organic derivatives

in sea water and their relation to the composition of

plankton. James Johnstone Memorial Volume, Liverpool,

UK, pp. 176e192.

Redi, M.H., 1982. Oceanic isopycnal mixing by coordinate

rotation. J. Phys. Oceanogr. 12, 1154e1158.

Reed, R.K., 1995. On geostrophic reference levels in the

Bering Sea basin. J. Oceanogr. 51, 489e498.

Reid, J.L., 1973. The shallow salinity minima of the Pacific

Ocean. Deep-Sea Res. 20, 51e68.

Reid, J.L., 1989. On the total geostrophic circulation of the

South Atlantic Ocean: Flow patterns, tracers and transports.



North Atlantic Ocean: Flow patterns, tracers and transports.



Pacific Ocean: Flow patterns, tracers and transports.



Indian Ocean: Flow patterns, tracers and transports.


Reid, J.L., Lynn, R.J., 1971. On the influence of the Norwegian-Greenland

and Weddell seas upon the bottom

waters of the Indian and Pacific Oceans. Deep-Sea Res.

18, 1063e1088.

Remote Sensing Systems, 2004. TMI sea surface temperatures

(SST). <http://www.ssmi.com/rss_research/tmi_

sst_pacific_equatorial_current.html> (accessed 3.27.09).

Reverdin, G., Durand, F., Mortensen, J., Schott, F.,

Valdimarsson, H., 2002. Recent changes in the surface

salinity of the North Atlantic subpolar gyre. J. Geophys.

Res. 107, 8010. doi:10.1029/2001JC001010.

Rhein, M., Stramma, L., Send, U., 1995. The Atlantic

Deep Western Boundary Current: Water masses and

transports near the equator. J. Geophys. Res. 100,

2441e2457.

Richardson, P.L., 1980a. Benjamin Franklin and Timothy

Folger’s First Printed Chart of the Gulf Stream. Science

207, 643e645.

Richardson, P.L., 1980b. Gulf Stream ring trajectories.


Richardson, P.L., 1983. Gulf Stream rings. In: Robinson, A.R.

(Ed.), Eddies in Marine Science. Springer-Verlag, Berlin,

pp. 19e45.

Richardson, P.L., 2005. Caribbean Current and eddies as

observed by surface drifters. Deep-Sea Res. II 52,

429e463.

Richardson, P.L., 2007. Agulhas leakage into the Atlantic

estimated with subsurface floats and surface drifters.


Richardson, P.L., 2008. On the history of meridional overturning

circulation schematic diagrams. Progr. Oceanogr.

76, 466e486.

Richardson, P.L., Bower, A.S., Zenk, W., 2000. A census of

Meddies tracked by floats. Progr. Oceanogr. 45, 209e250.

Richardson, P.L., Cheney, R.E., Worthington, L.V., 1978. A

census of Gulf Stream rings Spring 1975. J. Geophys.

Res. 83, 6136e6144.

Richardson, P.L., Hufford, G., Limeburner, R., Brown, W.,

1994. North Brazil Current retroflection eddies. J. Geophys.

Res. 99, 5081e5093.

Richardson, P.L., Lutjeharms, J.R.E., Boebel, O., 2003.

Introduction to the “Interocean exchange around

Africa.” Deep-Sea Res. II 50, 1e12.

534

REFERENCES

Richardson, P.L., Mooney, K., 1975. The Mediterranean

outflow d a simple advection-diffusion model. J. Phys.


Richardson, P.L., Reverdin, G., 1987. Seasonal cycle of

velocity in the Atlantic North Equatorial Countercurrent

as measured by surface drifters, current meters, and ship

drifts. J. Geophys. Res. 92, 3691e3708.

Ridgway, K.R., Condie, S.A., 2004. The 5500-km-long

boundary flow off western and southern Australia.

J. Geophys. Res. 109, C04017. doi: 10.1029/2003JC001921.

Ridgway, K.R., Dunn, J.R., 2003. Mesoscale structure of the

mean East Australian Current System and its relationship

with topography. Progr. Oceanogr. 56, 189e222.

Ridgway, K.R., Dunn, J.R., 2007. Observational evidence for

a Southern Hemisphere oceanic supergyre. Geophys.

Res. Lett. 34 doi:10.1029/2007GL030392.

Rigor, I.G., Wallace, J.M., Colony, R.L., 2002. Response of sea

ice to the Arctic Oscillation. J. Clim. 15, 2648e2663.

Rintoul, S.R., 1998. On the origin and influence of Adelie

Land Bottom Water. In: Jacobs, S., Weiss, R. (Eds.), 1998.

Ocean, Ice, and Atmosphere: Interactions at the

Antarctic Continental Margin. Antarctic Research

Series 75, American Geophysical Union, Washington,

pp. 151e171.

Rintoul, S.R., Hughes, C.W., Olbers, D., 2001. The Antarctic

Circumpolar Current System. In: Siedler, G., Church, J.

(Eds.), Ocean Circulation and Climate, International

Geophysics Series. Academic Press, San Diego, CA,

pp. 271e302.

Risien, C.M., Chelton, D.B., 2008. A global climatology of

surface wind and wind stress fields from eight years of

QuikSCAT scatterometer data. J. Phys. Oceanogr. 38,

2379e2413.

Roach, A.T., Aagaard, K., Pease, C.H., Salo, S.A.,

Weingartner, T., Pavlov, V., et al., 1995. Direct measurements

of transport and water properties through the

Bering Strait. J. Geophys. Res. 100, 18443e18458.

Robbins, P.E., Toole, J.M., 1997. The dissolved silica budget

as a constraint on the meridional overturning circulation

in the Indian Ocean. Deep-Sea Res. I 44,

879e906.

Robinson, A.R., Brink, K.H. (Eds.), 1998. The Sea, Vol. 11:

The Global Coastal Ocean: Regional Studies and

Syntheses. Harvard University Press, Cambridge, MA,

1090 pp.

Robinson, A.R., Brink, K.H. (Eds.), 2005. The Sea, Vol. 13:

The Global Coastal Ocean: Multiscale Interdisciplinary

Studies. Harvard University Press, Cambridge, MA,

1033 pp.

Robinson, A.R., Brink, K.H. (Eds.), 2006. The Sea, Vol. 14A:



Cambridge, MA, 840 pp.

Robinson, A.R., Golnaraghi, M., Leslie, W.G., Artegiani, A.,

Hecht, A., Lazzoni, E., et al., 1991. The eastern Mediterranean

general circulation: Features, structure and

variability. Dynam. Atmos. Oceans 15, 215e240.

Robinson, I.S., 2004. Measuring the Oceans from Space: The

Principles and Methods of Satellite Oceanography.

Springer-Verlag, Chichester, UK, 669 pp.

Rochford, D.J., 1961. Hydrology of the Indian Ocean. 1. The

water masses in intermediate depths of the southeast

Indian Ocean. Aust. J. Mar. Fresh. Res. 12, 129e149.

Roden, G.I., 1975. On North Pacific temperature, salinity,

sound velocity and density fronts and their relation to

the wind and energy flux fields. J. Phys. Oceanogr. 5,

557e571.

Roden, G.I., 1991. Subarctic-subtropical transition zone of

the North Pacific: Large-scale aspects and mesoscale

structure. In: Wetherall, J.A. (Ed.), Biology, Oceanography

and Fisheries of the North Pacific Transition Zone

and the Subarctic Frontal Zone. NOAA Technical Report,

105, pp. 1e38.

Rodhe, J., 1998. The Baltic and North Seas: A processoriented

review of the physical oceanography. In:

Robinson, A.R., Brink, K.H. (Eds.), The Sea, Vol. 11:

The Global Coastal Ocean: Regional Studies and

Syntheses. Harvard University Press, Boston, MA,

pp. 699e732.

Rodhe, J., Tett, P., Wulff, F., 2006. The Baltic and North Seas:

A regional review of some important physical-chemicalbiological

interaction processes. In: Robinson, A.R.,

Brink, K.H. (Eds.), The Sea, Vol. 14A: The Global Coastal

Ocean: Interdisciplinary Regional Studies and

Syntheses. Harvard University Press, Boston, MA,

pp. 1033e1076.

Roemmich, D., Cornuelle, B., 1992. The Subtropical Mode

Waters of the South Pacific Ocean. J. Phys. Oceanogr. 22,

1178e1187.

Roemmich, D., Gilson, J., Davis, R., Sutton, P., Wijffels, S.,

Riser, S., 2007. Decadal spin-up of the South Pacific

subtropical gyre. J. Phys. Oceanogr. 37, 162e173.

Roemmich, D., Hautala, S., Rudnick, D., 1996. Northward

abyssal transport through the Samoan passage and

adjacent regions. J. Geophys. Res. 101, 14039e14055.

Roemmich, D., Sutton, P., 1998. The mean and variability of

ocean circulation past northern New Zealand: Determining

the representativeness of hydrographic climatologies.

J. Geophys. Res. 103, 13041e13054.

Roemmich, D.L., 1983. Optimal estimation of hydrographic

station data and derived fields. J. Phys. Oceanogr. 13,

1544e1549.

Roemmich, D.L., Gilson, J., 2009. The 2004-2008 mean and

annual cycle of temperature, salinity, and steric height in

the global ocean from the Argo Program. Progr. Oceanogr.

82, 81e100.

REFERENCES 535

Ronski, S., Budéus, G., 2005a. How to identify winter

convection in the Greenland Sea from hydrographic

summer data. J. Geophys. Res. 110, C11010. doi:10.1029/

2003JC002156.

Ronski, S., Budéus, G., 2005b. Time series of winter

convection in the Greenland Sea. J. Geophys. Res. 110,

C04015. doi:10.1029/2004JC002318.

Ross, D. A., 1983. The Red Sea. In Estuaries and Enclosed

Seas, Ed. B.H. Ketchum. Ecosystems of the World, 26,

Elsevier, 293e307.

Rossby, T., 1996. The North Atlantic Current and

surrounding waters: at the crossroads. Rev. Geophys. 34

(4), 463e481.

Rossby, T., 1999. On gyre interactions. Deep-Sea Res. II 46,

139e164.

Rothrock, D., Yu, Y., Maykut, G., 1999. Thinning of the

Arctic sea-ice cover. Geophys. Res. Lett. 26, 3469e3472.

Rowe, E., Mariano, A.J., Ryan, E.H., 2010. The Antilles

Current. Ocean Surface Currents. University of Miami,

RSMAS, CIMAS. <http://oceancurrents.rsmas.miami.

edu/atlantic/antilles.html> (accessed 1.10.10).

Rudels, B., 1986. The outflow of polar water through the

Arctic archipelago and the oceanographic conditions in

Baffin Bay. Polar Res. 4, 161e180.

Rudels, B., 2001. Arctic Basin circulation. In: Steele, J.H.,

Thorpe, S.A., Turekian, K.K. (Eds.), Encyclopedia of

Ocean Sciences. Elsevier Science Ltd., Oxford, UK,

pp. 177e187.

Rudels, B., Anderson, L.G., Eriksson, P., Fahrbach, E.,

Jakobsson, M., Jones, E.P., et al., 2011. ACSYS Chapter 4:

Observations in the Ocean. In: Lemke, P., Fichefet, T.,

Dick, C. (Eds.), Arctic Climate Change d The ACSYS

Decade and Beyond. Springer-Verlag, Berlin in press.

Rudels, B., Bjork, G., Nilsson, J., Winsor, P., Lake, I.,

Nohr, C., 2005. Interaction between waters from the

Arctic Ocean the Nordic Seas north of Fram Strait and

along the East Greenland Current: results from the

Arctic Ocean-20 Oden expedition. J. Marine Sys. 55,

1e30.

Rudels, B., Friedrich, H.J., Quadfasel, D., 1999. The Arctic

circumpolar boundary current. Deep-Sea Res. II 46,

1023e1062.

Rudnick, D.L., 1996. Intensive surveys of the Azores Front 2.

Inferring the geostrophic and vertical velocity fields.

J. Geophys. Res. 101, 16291e16303.

Rudnick, D.L., 1997. Direct velocity measurements in the

Samoan Passage. J. Geophys. Res. 102, 3293e3302.

Rudnick, D.L., 2008. SIO 221B: Analysis of physical oceanographic

data. <http://chowder.ucsd.edu/Rudnick/

SIO_221B.html> (accessed 4.14.09).

Rudnick, D.L., Boyd, T.J., Brainard, R.E., Carter, G.S.,

Egbert, G.D., Gregg, M.C., et al., 2003. From tides to

mixing along the Hawaiian Ridge. Science 301, 355e357.

Rydevik, D., 2004. A picture of the 2004 tsunami in Ao

Nang, Thailand. Wikipedia OTRS system. <http://en.

wikipedia.org/wiki/File:2004-tsunami.jpg#file> (accessed

9.22.05).

Sabine, C.L., Feely, R.A., Gruber, N., Key, R.M., Lee, K.,

Bullister, J.L., et al., 2004. The oceanic sink for anthropogenic

CO2. Science 305, 367e371.

Saji, N.H., Goswami, B.N., Vinayachandran, P.N.,

Yamagata, T., 1999. A dipole mode in the tropical Indian

Ocean. Nature 401, 360e363.

Salmon, R., 1998. Lectures on Geophysical Fluid Dynamics.

Oxford University Press, New York, 378 pp.

Sandström, J., 1908. Dynamische Versuche mit Meerwasser.

Annalen der Hydrographie und Maritimen Meteorologie,

6e23 (in German).

Sankey, T., 1973. The formation of deep water in the Northwestern

Mediterranean. Progr. Oceanogr. 6, 159e179.

Savchenko, V.G., Emery, W.J., Vladimirov, O.A., 1978. A

cyclonic eddy in the Antarctic Circumpolar Current south of

Australia: results of Soviet-American observations aboard

the R/V Professor Zubov. J. Phys. Oceanogr. 8, 825e837.

Scambos, T., Haran, T., Fahnestock, M., Painter, T.,

Bohlander., J., 2007. MODIS-based Mosaic of Antarctica

(MOA) data sets: Continent-wide surface morphology

and snow grain size. Remote Sens. Environ. 111, 242e257.

Scambos, T., Hulbe, C., Fahnestock, M., Bohlander, J., 2000.

The link between climate warming and break-up of

ice shelves in the Antarctic Peninsula. J. Glaciol. 46,

516e530.

Schauer, U., Rudels, B., Jones, E.P., Anderson, L.G.,

Muench, R.D., Björk, G., et al., 2002. Confluence and

redistribution of Atlantic water in the Nansen, Amundsen

and Makarov basins. Ann. Geophys. 20, 257e273.

Schlichtholz, P., Houssais, M.-H., 2002. An overview of the

theta-S correlations in Fram Strait based on the MIZEX

84 data. Oceanologia 44, 243e272.

Schlitzer, R., Roether, W., Oster, H., Junghans, H.-G.,

Hausmann, M., Johannsen, H., et al., 1991. Chlorofluoromethane

and oxygen in the Eastern Mediterranean.

Deep-Sea Res. 38, 1531e1551.

Schlosser, P., co-authors, 1997. The first trans-Arctic 14C

section: Comparison of the mean ages of the deep waters

in the Eurasian and Canadian basins of the Arctic Ocean.

Nucl. Instrum. Methods 123B, 431e437.

Schmitt, R.W., 1981. Form of the temperature-salinity relationship

in the Central Water: Evidence for doublediffusive

mixing. J. Phys. Oceanogr. 11, 1015e1026.

Schmitt, R.W., Perkins, H., Boyd, J.D., Stalcup, M.C., 1987.

C-SALT: An investigation of the thermohaline staircase

in the western tropical North Atlantic. Deep-Sea Res. 34,

1655e1665.

Schmitz, W.J., 1995. On the interbasin-scale thermohaline

circulation. Rev. Geophys. 33, 151e173.

536

REFERENCES

Schmitz, W.J., 1996a. On the eddy field in the Agulhas

Retroflection, with some global considerations. J. Geophys.

Res. 101, 16259e16271.

Schmitz, W.J., 1996b. On the World Ocean Circulation:

Volume I: Some global features/North Atlantic circulation.

Woods Hole Oceanographic Institution Technical

Report, WHOI-96-03, Woods Hole, MA, 141 pp.

Schneider, N., 1998. The Indonesian throughflow and the

global climate system. J. Clim. 11, 676e689.

Schneider, N., Cornuelle, B.D., 2005. The forcing of the

Pacific Decadal Oscillation. J. Clim. 18, 4355e4373.

Schott, F.A., Brandt, P., 2007. Circulation and deep water

export of the subpolar North Atlantic during the 1990s.

Geophys. Monogr. Ser. 173, 91e118. doi:10.1029/

173GM08.

Schott, F.A., Dengler, M., Brandt, P., Affler, K., Fischer, J.,

Bourlès, B., et al., 2003. The zonal currents and transports

at 35 W in the tropical Atlantic. Geophys. Res. Lett.

30, 1349. doi:10.1029/2002GL016849.

Schott, F.A., Dengler, M., Schoenefeldt, R., 2002. The shallow

overturning circulation of the Indian Ocean. Progr.


Schott, F.A., McCreary Jr., J., 2001. The monsoon circulation

of the Indian Ocean. Progr. Oceanogr. 51, 1e123.

Schott, F.A., McCreary, J.P., Johnson, G.A., 2004. Shallow

overturning circulations of the tropical-subtropical

oceans. In Earth’s Climate: The Ocean-Atmosphere

Interaction. AGU Geophy. Monogr. Ser. 147, 261e304.

Schott, F.A., Stramma, L., Giese, B.S., Zantopp, R., 2009.

Labrador Sea convection and subpolar North Atlantic

Deep Water export in the SODA assimilation model.


Schott, F.A., Zantopp, R., Stramma, L., Dengler, M.,

Fischer, J., Wibaux, M., 2004. Circulation and deep water

export at the western exit of the subpolar North Atlantic.


Schröder, M., Fahrbach, E., 1999. On the structure and the

transport of the eastern Weddell Gyre. Deep-Sea Res. 46,

501e527.

Schwing, F.B., O’Farrell, M., Steger, J., Baltz, K., K., 1996.

Coastal Upwelling Indices, West coast of North America,

1946e1995. U.S. Dept. of Commerce, NOAA Tech.

Memo. NOAA-TM-NMFS-SWFC-231, 207.

Sclater, J., Parsons, B., Jaupart, C., 1981. Oceans and continents:

similarities and differences in the mechanisms of

heat loss. J. Geophys. Res. 86, 11535e11552.

SeaWiFS Project, 2009. SeaWiFS captures El Nino-La Nina

transitions in the equatorial Pacific. NASA Goddard

Space Flight Center. <http://oceancolor.gsfc.nasa.

gov/SeaWiFS/BACKGROUND/Gallery/pac_elnino.jpg>

(accessed 3.26.09).

Seibold, E., Berger, W.H., 1982. The Sea Floor. Springer

Verlag, Berlin, 356 pp.

Sekine, Y., 1999. Anomalous southward intrusions of the

Oyashio east of Japan 2. Two-layer numerical model.

J. Geophys. Res. 104, 3049e3058.

Serreze, M.C., Holland, M.M., Stroeve, J., 2007. Perspectives on

the Arctic’s shrinking sea-ice cover. Science 16, 1533e1536.

Shaffer, G., Salinas, S., Pizarro, O., Vega, A., Hormazabal, S.,

1995. Currents in the deep ocean off Chile (30 S). Deep-

Sea Res. I 42, 425e436.

Shcherbina, A., Talley, L.D., Rudnick, D.L., 2003. Direct

observations of brine rejection at the source of North

Pacific Intermediate Water in the Okhotsk Sea. Science

302, 1952e1955.

Shcherbina, A., Talley, L.D., Rudnick, D.L., 2004. Dense

water formation on the northwestern shelf of the

Okhotsk Sea: 1. Direct observations of brine rejection.

J Geophys. Res. 109, C09S08. doi:10.1029/2003JC002196.

Shell, K.M., Frouin, R., Nakamoto, S., Somerville, R.C.J.,

2003. Atmospheric response to solar radiation absorbed

by phytoplankton. J. Geophys. Res. 108 (D15), 4445.

doi:10.1029/2003JD003440.

Shillington, F.A., 1998. The Benguela upwelling system off

southwestern Africa. In: Robinson, A.R., Brink, K.H.

(Eds.), The Sea, Vol. 11: The Global Coastal Ocean:

Regional Studies and Syntheses. Harvard University

Press, Boston, MA, pp. 583e604.

Shimada, K., Kamoshida, T., Itoh, M., Nishino, S.,

Carmack, E.C., McLaughlin, F., et al., 2006. Pacific Ocean

Inflow: influence on catastrophic reduction of sea ice

cover in the Arctic Ocean. Geophys. Res. Lett. 33,

L08605. doi; 10.1029/2005GL025624.

Shinoda, T., Kiladis, G.N., Roundy, P.E., 2009. Statistical

representation of equatorial waves and tropical instability

waves in the Pacific Ocean. Atmos. Res. 94, 37e44.

Shoosmith, D.R., Richardson, P.L., Bower, A.S., Rossby, H.T.,

2005. Discrete eddies in the northern North Atlantic as

observed by looping RAFOS floats. Deep-Sea Res. II 52,

627e650.

Shuckburgh, E., Jones, H., Marshall, J., Hill, C., 2009.

Understanding the regional variability of eddy diffusivity

in the Pacific sector of the Southern Ocean. J. Phys.

Oceanogr. 39, 2011e2023.

Siedler, G., Church, J., Gould, J., 2001. Ocean Circulation and

Climate: Observing and Modelling the Global Ocean.

AP International Geophysics Series Vol. 77, 715 pp.

Siedler, G., Kuhl, A., Zenk, W., 1987. The Madeira Mode

Water. J. Phys. Oceanogr. 17, 1561e1570.

Siedler, G., Zanbenberg, N., Onken, R., Morlière, A., 1992.

Seasonal changes in the tropical Atlantic circulation:

Observation and simulation of the Guinea Dome.


Sievers, H., Emery, W., 1978. Variability of the Antarctic

Polar Frontal Zone in the Drake Passage d Summer

1976e1977. J. Geophys. Res. 83, 3010e3022.

REFERENCES 537

Simpson, J.H., 1998. Tidal processes in shelf seas. In:

Brink, K.H., Robinson, A.R. (Eds.), The Sea, Vol. 10: The

Global Coastal Ocean: Processes and Methods. Harvard


Simpson, J.J., Koblinsky, C.J., Peláez, J., Haury, L.R.,

Wiesenhahn, D., 1986. Temperature d plant pigment d

optical relations in a recurrent offshore mesoscale eddy

near Point Conception, California. J. Geophys. Res. 91,

12919e12936.

SIO, 2008. SRTM30_plus, Satellite Geodesy, Scripps Institution

of Oceanography. University of California San

Diego. <http://topex.ucsd.edu/WWW_html/srtm30_

plus.html> (accessed 9.24.08).

Sloyan, B.M., Rintoul, S.R., 2001. The Southern Ocean limb

of the global deep overturning circulation. J. Phys.


Smethie, W.M., Fine, R.A., 2001. Rates of North Atlantic

Deep Water formation calculated from chlorofluorocarbon

inventories. Deep-Sea Res. I 48, 189e215.

Smith, J.A., 2001. Observations and theories of Langmuir

circulation: a story of mixing. In: Lumley, J.L. (Ed.), Fluid

Mechanics and the Environment: Dynamical Approaches.

Springer, New York, pp. 295e314.

Smith, R.D., Maltrud, M.E., Bryan, F.O., Hecht, M.W., 2000.

Numerical simulation of the North Atlantic Ocean at

1/10 . J. Phys. Oceanogr. 30, 1532e1561.

Smith, R.L., Huyer, A., Godfrey, J.S., Church, J.A., 1991. The

Leeuwin Current off Western Australia, 1986e1987.


Smith, S.D., 1988. Coefficients for sea surface wind stress,

heat flux, and wind profiles as a function of wind speed

and temperature. J. Geophys. Res. 93, 15467e15472.

Smith, S.D., Muench, R.D., Pease, C.H., 1990. Polynyas and

leads: An overview of physical processes and environment.

J. Geophys. Res. 95, 9461e9479.

Smith, W.H.F., Sandwell, D.T., 1997. Global seafloor topography

from satellite altimetry and ship depth soundings.

Science 277, 1957e1962.

Smith, W.H.F., Scharroo, R., Titov, V.V., Arcas, D.,

Arbic, B.K., 2005. Satellite altimeters measure tsunami.

Oceanography 18, 10e12.

Sofianos, S.S., Johns, W.E., 2003. An Oceanic General

Circulation Model (OGCM) investigation of the Red Sea

circulation: 2. Three-dimensional circulation in the Red

Sea. J. Geoph. Res. 108, 3066. doi: 10,1029/200IJC001185.

Sokolov, S., Rintoul, S.R., 2002. Structure of Southern Ocean

fronts at 140 E. J. Marine Syst. 37, 151e184.

Song, Q., Vecchi, G.A., Rosati, A.J., 2007. The role of the

Indonesian Throughflow in the Indo-Pacific climate

variability in the GFDL coupled climate model. J. Clim.

20, 2434e2451.

Sosik, H., 2003. Patterns and scales of variability in the

optical properties of Georges Bank waters, with special

reference to phytoplankton biomass and production.

H. Sosik, Woods Hole Oceanographic Institution.

<http://www.whoi.edu/science/B/sosiklab/gbgom.htm>

(accessed 3.29.08).

Spadone, A., Provost, C., 2009. Variations in the Malvinas

Current volume transport since October 1992. J. Geophys.

Res. 114, C02002. doi:10.1029/2008JC004882.

Spall, M.A., Richardson, P.L., Price, J., 1993. Advection and

eddy mixing in the Mediterranean salt tongue. J. Marine

Res. 51, 797e818.

Speer, K., Rintoul, S.R., Sloyan, B., 2000. The diabatic

Deacon cell. J. Phys. Oceanogr. 30, 3212e3222.

Speich, S., Blanke, B., de Vries, P., Drijfhout, S., Doos, K.,

Ganachaud, A., et al., 2002. Tasman leakage: A new

route in the global ocean conveyor belt. Geophys. Res.

Lett. 29, 1416. doi:10.1029/2001GL014586.

Spiess, F., 1928. Die Meteor Fahrt: Forschungen und Erlebnisse

der Deutschen Atlantischen Expedition, 1925e1927.

Verlag von Dietrich Reimer, Berlin, 376 pp. (in German,

English translation Emery, W.J., Amerind Publishing Co.

Pvt. Ltd., New Delhi, 1985).

Stabeno, P.J., Reed, R.K., 1995. Circulation in the Bering Sea

basin observed by satellite-tracked drifters: 1986e1993.


Stammer, D., 1997. Global characteristics of ocean variability

estimated from regional TOPEX/POSEIDON altimeter

measurements. J. Phys. Oceanogr. 27, 1743e1769.

Stammer, D., 1998. On eddy characteristics, eddy transports,

and mean flow properties. J. Phys. Oceanogr. 28, 727e739.

Stammer, D., Wunsch, C., 1999. Temporal changes in eddy

energy of the oceans. Deep-Sea Res. 46, 77e108.

Stammer, D., Wunsch, C., Ponte, R.M., 2000. De-aliasing of

global high frequency barotropic motions in altimeter

observations. Geophys. Res. Lett. 27, 1175e1178.

Steele, M., Morison, J., Ermold, W., Rigor, I., Ortmeyer, M.,

Shimada, K., 2004. Circulation of summer Pacific halocline

water in the Arctic Ocean. J. Geophys. Res. 109,

C02027. doi:10.1029/2003JC002009.

Steger, J.M., Carton, J.A., 1991. Long waves and eddies in

the tropical Atlantic Ocean: 1984e1990. J. Geophys. Res.

96, 15161e15171.

Stewart, R.H., 2008. Introduction to Physical Oceanography.

Open-source textbook. <http://oceanworld.tamu.edu/

ocean410/ocng410_text_book.html> (accessed 3.28.09).

Stocker, T.F., Marchal, O., 2000. Abrupt climate change in

the computer: Is it real? Proc. Natl. Acad. Sci. USA 97l,

1362e1365.

Stommel, H., 1948. The westward intensification of winddriven

currents. Trans. Am. Geophys. Union 29, 202e206.

Stommel, H.M., 1958. The abyssal circulation. Deep-Sea Res.

5, 80e82.

Stommel, H.M., 1961. Thermohaline convection with two

stable regimes of flow. Tellus 13, 224e230.

538

REFERENCES

Stommel, H.M., 1965. The Gulf Stream: A Physical and

Dynamical Description, second ed. University of

California Press, Berkeley, and Cambridge University

Press, London, 248 pp.

Stommel, H.M., 1979. Determination of water mass properties

of water pumped down from the Ekman layer to

the geostrophic flow below. Proc. Nat. Acad. Sci. USA

76, 3051e3055.

Stommel, H.M., Arons, A., 1960a. On the abyssal circulation

of the World Ocean d I. Stationary planetary flow

patterns on a sphere. Deep-Sea Res. 6, 140e154.

Stommel, H.M., Arons, A., 1960b. On the abyssal circulation

of the World Ocean d II. An idealized model of the

circulation pattern and amplitude in oceanic basins.


Stommel, H.M., Arons, A., Faller, A., 1958. Some examples

of stationary planetary flow patterns in bounded basins.

Tellus 10, 179e187.

Stommel, H.M., Niiler, P.P., Anati, D., 1978. Dynamic

topography and recirculation of the North Atlantic.

J. Marine Res. 36, 449e468.

Stone, B., 1914. Map of track of the ’Endurance’ in Weddell Sea.

Royal Geographical Society. <http://images.rgs.org/

imageDetails.aspx?barcode¼27820> (accessed 10.15.06).

Stramma, L., Cornillon, P., Woller, R.A., Price, J.F.,

Briscoe, M.G., 1986. Large diurnal sea surface temperature

variability: satellite and in situ measurements.


Stramma, L., Ikeda, Y., Peterson, R.G., 1990. Geostrophic

transport in the Brazil Current region north of 20 S.

Deep-Sea Res. 37, 1875e1886.

Stramma, L., Johnson, G.C., Sprintall, J., Mohrholz, V., 2008.

Expanding oxygen minimum zones in the tropical

oceans. Science 320, 655e658.

Stramma, L., Kieke, D., Rhein, M., Schott, F., Yashayaev, I.,

Koltermann, K.P., 2004. Deep water changes at the

western boundary of the subpolar North Atlantic during

1996 to 2001. Deep-Sea Res. I 51, 1033e1056.

Stramma, L., Lutjeharms, J.R.E., 1997. The flow field of the

subtropical gyre of the South Indian Ocean. J. Geophys.

Res. 102, 5513e5530.

Stramma, L., Peterson, R.G., 1990. The South Atlantic

Current. J. Phys. Oceanogr. 20, 846e859.

Stramma, L., Peterson, R.G., Tomczak, M., 1995. The South

Pacific Current. J. Phys. Oceanogr. 25, 77e91.

Stramma, L., Schott, F., 1999. The mean flow field of the

tropical Atlantic Ocean. Deep-Sea Res. II 46, 279e303.

Stramski, D., Reynolds, R.A., Babin, M., Kaczmarek, S.,

Lewis, M.R., Röttgers, R., et al., 2008. Relationships

between the surface concentration of particulate

organic carbon and optical properties in the eastern South

Pacific and eastern Atlantic Oceans. Biogeosciences 5,

171e201.

Straneo, F., Saucier, F., 2008. The outflow from Hudson Strait

and its contribution to the Labrador Current. Deep-Sea

Res. I 55, 926e946.

Strub, P.T., James, C., 2000. Altimeter-derived variability of

surface velocities in the California Current System: 2.

Seasonal circulation and eddy statistics. Deep-Sea Res. II

47, 831e870.

Strub, P.T., James, C., 2009. Altimeter-derived circulation in

the California Current. College of Oceanic and Atmospheric

Sciences. Oregon State University. <http://

www.coas.oregonstate.edu/research/po/research/strub/

index.html> (accessed 4.2.09).

Strub, P.T., Kosro, P.M., Huyer, A., Brink, K.H.,

Hayward, T.L., Niiler, P.P., et al., 1991. The nature of cold

filaments in the California current system. J. Geophys.

Res. 96, 14743e14769.

Strub, P.T., Mesias, J.M., Montecino, V., Ruttlant, J.,

Salinas, S., 1998. Coastal ocean circulation off

western South America. In: Robinson, A.R., Brink, K.H.

(Eds.), The Sea, Vol. 11: The Global Coastal Ocean d

Regional Studies and Syntheses. Wiley, New York,

pp. 273e313.

Sundby, S., Drinkwater, K., 2007. On the mechanisms

behind salinity anomaly signals of the northern North

Atlantic. Progr. Oceanogr. 73, 190e202.

Sutton, R.T., Jewson, S.P., Rowell, D.P., 2000. The elements of

climate variability in the tropical Atlantic region. J. Clim.

13, 3261e3284.

Sverdrup, H.U., 1947. Wind-driven currents in a baroclinic

ocean. Proc. Nat. Acad. Sci. USA 33, 318e326.

Sverdrup, H.U., Johnson, M.W., Fleming, R.H., 1942. The

Oceans: Their Physics, Chemistry and General Biology.

Prentice Hall Inc., Englewood Cliffs, NJ, 1057 pp.

Swallow, J.C., Bruce, J.C., 1966. Current measurements off

the Somali coast during the southwest monsoon of 1964.


Swallow, J.C., Worthington, L.V., 1961. An observation of

a deep countercurrent in the western North Atlantic.


Swift, J.H., 1986. The Arctic Waters. In: Hurdle, B. (Ed.),

The Nordic Seas. Springer-Verlag, New York,

pp. 129e154.

Swift, J.H., Aagaard, K., 1981. Seasonal transitions and

water mass formation in the Iceland and Greenland seas.

Deep-Sea Res. 28, 1107e1129.

Swift, J.H., Aagaard, K., Timokhov, L., Nikiforov., E.G., 2005.

Long-term variability of Arctic Ocean Waters: Evidence

from a reanalysis of the EWG data set. J. Geophys. Res.

110, C03012. doi:10.1029/2004JC002312.

Swift, J.H., Jones, E.P., Aagaard, K., Carmack, E.C.,

Hingston, M., Macdonald, R.W., et al., 1997. Waters of

the Makarov and Canada basins. Deep-Sea Res. II 44,

1503e1529.

REFERENCES 539

Swift, S.A., Bower, A.S., 2003. Formation and circulation of

dense water in the Persian/Arabian Gulf. J. Geophys.

Res. 108 (C10) doi:10.1029/2002JC001360.

TAO Project Office, 2009a. TAO/TRITON data display and

delivery. NOAA Pacific Marine Environmental Laboratory.

<http://www.pmel.noaa.gov/tao/disdel/disdel.html>

(accessed 3.27.09).

TAO Project Office, 2009b. TAO Climatologies. NOAA Pacific

Marine Environmental Laboratory. <http://www.pmel.

noaa.gov/tao/clim/clim.html> (accessed 7.5.09).

Tabata, S., 1982. The anti-cyclonic, baroclinic eddy off Sitka

Alaska, in the northeast Pacific Ocean. J. Phys. Oceanogr.

12, 1260e1282.

Talley, L.D., 1991. An Okhotsk Sea anomaly: Implication for

ventilation in the North Pacific. Deep-Sea Res. 38

(Suppl.), S171eS190.

Talley, L.D., 1993. Distribution and formation of North Pacific

Intermediate Water. J. Phys. Oceanogr. 23, 517e537.

Talley, L.D., 1996a. Antarctic Intermediate Water in the

South Atlantic. In: Wefer, G., Berger, W.H., Siedler, G.,

Webb, D. (Eds.), The South Atlantic: Present and Past

Circulation. Springer-Verlag, New York, pp. 219e238.

Talley, L.D., 1996b. North Atlantic circulation and variability,

reviewed for the CNLS conference. Physica D 98,

625e646.

Talley, L.D., 1999. Some aspects of ocean heat transport by

the shallow, intermediate and deep overturning circulations.

In: Clark, P.U., Webb, R.S., Keigwin, L.D. (Eds.),

Mechanisms of Global Climate Change at Millennial

Time Scales, Geophys. Mono. Ser. 112, American

Geophysical Union, pp. 1e22.

Talley, L.D., 2003. Shallow, intermediate, and deep overturning

components of the global heat budget. J. Phys.


Talley, L.D., 2007. Hydrographic Atlas of the World Ocean

Circulation Experiment (WOCE). Volume 2: Pacific

Ocean. In: Sparrow, M., Chapman, P., Gould, J. (Eds.).

International WOCE Project Office, Southampton, UK

ISBN 0-904175-54-5.

Talley, L.D., 2008. Freshwater transport estimates and the

global overturning circulation: Shallow, deep and

throughflow components. Progr. Oceanogr. 78, 257e303.

doi:10.1016/j.pocean.2008.05.001.

Talley, L.D., 2011a. Hydrographic Atlas of the World Ocean

Circulation Experiment (WOCE). vol 3: Indian Ocean. In:

Sparrow, M., Chapman, P., Gould, J. (Eds.). International

WOCE Project Office, Southampton, U.K. <http://

www-pord.ucsd.edu/whp_atlas/indian_index.htm>

Online version (accessed 4.20.09).

Talley, L.D., 2011b. Schematics of the global overturning

circulation. In preparation.

Talley, L.D., Johnson, G.C., 1994. Deep, zonal subequatorial

jets. Science 263, 1125e1128.

Talley, L.D., Joyce, T.M., 1992. The double silica maximum

in the North Pacific. J. Geophys. Res. 97, 5465e5480.

Talley, L.D., McCartney, M.S., 1982. Distribution and circulation

of Labrador Sea Water. J. Phys. Oceanogr. 12,

1189e1205.

Talley, L.D., Min, D.-H., Lobanov, V.B., Luchin, V.A.,

Ponomarev, V.I., Salyuk, A.N., et al., 2006. Japan/East

Sea water masses and their relation to the sea’s circulation.


Talley, L.D., Nagata, Y., 1995. The Okhotsk Sea and Oyashio

Region. PICES Scientific Report, 2. North Pacific Marine

Science Organization (PICES), Sidney, B.C., Canada, 227 pp.

Talley, L.D., Sprintall, J., 2005. Deep expression of the

Indonesian Throughflow: Indonesian Intermediate

Water in the South Equatorial Current. J. Geophys. Res.

110, C10009. doi:10.1029/2004JC002826.

Talley, L.D., Tishchenko, P., Luchin, V., Nedashkovskiy, A.,

Sagalaev, S., Kang, D.-J., et al., 2004. Atlas of Japan (East)

Sea hydrographic properties in summer, 1999. Progr.


Talley, L.D., Yun, J.-Y., 2001. The role of cabbeling and

double diffusion in setting the density of the North

Pacific Intermediate Water salinity minimum. J. Phys.

Oceanogr. 31, 1538e1549.

Tamura, T., Ohshima, K.I., Nihashi, S., 2008. Mapping of sea

ice production for Antarctic coastal polynyas. Geophys.

Res. Lett. 35, L07606. do:10.1029/2007GL032903.

Tanhua, T., Olsson, K.A., Jeansson, E., 2005. Formation of

Demark Strait overflow water and its hydro-chemical

composition. J. Marine Sys. 57, 264e288.

Taylor, P.K. (Ed.), 2000. Intercomparison and validation of

ocean-atmosphere energy flux fields d Final report of

the Joint WCRP/SCOR Working Group on Air-Sea

Fluxes. WCRP-112, WMO-TD-1036, 306 pp.

Teague, W.J., Ko, D.S., Jacobs, G.A., Perkins, H.T.,

Book, J.W., Smith, S.R., et al., 2006. Currents through the

Korea/Tsushima Strait. Oceanography 19, 50e63.

Thompson, D.W.J., Solomon, S., 2002. Interpretation of

recent southern hemisphere climate change. Science 296,

895e899.

Thompson, D.W.J., Wallace, J.M., 1998. The Arctic- Oscillation

signature in the wintertime geopotential height and

temperature fields. Geophys. Res. Lett. 25, 1297e1300.

Thompson, D.W.J., Wallace, J.M., 2000. Annular modes in

the extratropical circulation. Part I: Month-to-month

variability. J. Clim. 13, 1000e1016.

Thompson, S.L., Warren, S.G., 1982. Parameterization of

outgoing infrared radiation derived from detailed radiative

calculations. J. Atmos. Sci. 39, 2667e2680.

Thomson, J., Elgar, S., Herbers, T.H.C., 2005. Reflection and

tunneling of ocean waves observed at a submarine

canyon. Geophys. Res. Lett. 32, L10602. doi:10.1029/

2005GL022834.

540

REFERENCES

Thorpe, S.A., 2004. Langmuir circulation. Annu. Rev. Fluid

Mech. 36, 55e79. doi:10.1146/annurev.fluid.36.052203.

071431.

Thoulet, J., Chevallier, A., 1889. Sur la chaleur spécifique de

l’eau de mer a divers degres de dilution et de concentration.

C.R. Acad. Sci. 108, 794e796 (in French).

Thurman, H.V., Trujillo, A.P., 2002. Essentials of Oceanography,

7th ed. Prentice Hall, NJ, 524 pp.

Timmermans, M.L., Garrett, C., 2006. Evolution of the deep

water in the Canadian Basin in the Arctic Ocean. J. Phys.


Timmermans, M.L., Garrett, C., Carmack, E., 2003. The thermohaline

structure and evolution of the deep waters in the

Canada Basin, Arctic Ocean. Deep-Sea Res. I 50, 1305e1321.

Titov, V., Rabinovich, A.B., Mofjeld, H.O., Thomson, R.E.,

Gonzalez, F.I., 2005. The global reach of the 26 December

2004 Sumatra tsunami. Science 309, 2045e2048.

Tomczak, M., 1981. A multiparameter extension of temperature/salinity

diagram techniques for the analysis of

non-isopycnal mixing. Progr. Oceanogr. 10, 147e171.

Tomczak, M., 2000, 2002. Shelf and Coastal Oceanography.

Open-source textbook. <http://www.es.flinders.edu.au/

~mattom/ShelfCoast/index.html> (accessed 3.28.09).

Tomczak, M., Godfrey, J.S., 1994. Regional Oceanography:

An Introduction. Pergamon Press, Oxford, UK, 422 pp.

Tomczak, M., Godfrey, J.S., 2003. Regional Oceanography:

An Introduction, second ed. Daya Publications, Delhi,

390 pp., ISBN: 8170353068. (Online, open source version

at. <http://www.es.flinders.edu.au/~mattom/regoc/

pdfversion.html>.

Tomczak, M., Large, D., 1989. Optimum multiparameter

analysis of mixing in the thermocline of the eastern

Indian Ocean. J. Geophys. Res. 94, 16141e16149.

Toole, J.M., Millard, R.C., Wang, Z., Pu, S., 1990. Observations

of the Pacific North Equatorial Current bifurcation

at the Philippine coast. J. Phys. Oceanogr. 20, 307e318.

Toole, J.M., Warren, B.A., 1993. A hydrographic section

across the subtropical South Indian Ocean. Deep-Sea

Res. I 40, 1973e2019.

Tourre, Y.M., White, W.B., 1995. ENSO Signals in global upperocean


Tourre, Y.M., White, W.B., 1997. Evolution of the ENSO

signal over the Indo-Pacific domain. J. Phys. Oceanogr.

27, 683e696.

Treguier, A.M., 2006. Ocean models. In: Chassignet, E.P.,

Verron, J. (Eds.), Ocean Weather Forecasting: An Integrated

view of Oceanography. Springer, The Netherlands.

Trenberth, K.E., Caron, J.M., 2001. Estimates of meridional

atmosphere and ocean heat transports. J. Clim. 14,

3433e3443.

Trenberth, K.E., Hurrell, J.W., 1994. Decadal atmosphereocean

variations in the Pacific. Clim. Dyn. 9, 303e319.

Trenberth, K.E., Jones, P.D., Ambenje, P., Bojariu, R.,

Easterling, D., Klein Tank, A., et al., 2007. Observations:

Surface and Atmospheric Climate Change. In:

Solomon, S., Qin, D., Manning, M., Chen, Z.,

Marquis, M., Averyt, K.B., et al. (Eds.), Climate Change

2007: The Physical Science Basis. Contribution of


Intergovernmental Panel Climate Change. Cambridge

University Press, Cambridge, UK and New York.

Tsuchiya, M., 1975. Subsurface countercurrents in the

eastern equatorial Pacific Ocean. J. Mar. Res. 33 (Suppl),

145e175.

Tsuchiya, M., 1981. The origin of the Pacific equatorial 13 C

water. J. Phys. Oceanogr. 11, 794e812.

Tsuchiya, M., 1989. Circulation of the Antarctic Intermediate

Water in the North Atlantic Ocean. J. Mar. Res. 47,

747e755.

Tsuchiya, M., Talley, L.D., McCartney, M.S., 1992. An eastern

Atlantic section from Iceland southward across the

equator. Deep-Sea Res. 39, 1885e1917.

Tsuchiya, M., Talley, L.D., McCartney, M.S., 1994. Water

mass distributions in the western Atlantic: A section

from South Georgia Island (54S) northward across the

equator. J. Mar. Res. 52, 55e81.

Tully, J.P., 1949. Oceanography and prediction of pulp-mill

pollution in Alberni Inlet. Fish. Res. Bd. Can., Bulletin

83, 169 pp.

UCT Oceanography Department, 2009. Monthly sea surface

temperature (SST) composites. Marine remote sensing

unit at the Department of Oceanography. University of

Cape Town. <http://www.sea.uct.ac.za/projects/remsense/

index.php> (accessed 6.9.09).

UNESCO, 1981. The Practical Salinity Scale 1978 and the

International Equation of State of Seawater 1980. Tech.

Paper Mar., Sci. 36, 25 pp.

UNESCO, 1983. Algorithms for computation of fundamental

properties of seawater. Tech. Paper Mar., Sci. 44,

53 pp.

UNESCO, 1987. International oceanographic tables. Tech.

Paper Mar., Sci. 40, 196 pp.

Urick, R.J., 1983. Principles of Underwater Sound, 3rd ed.

McGraw-Hill, New York, 423 pp.

Vallis, G.K., 2006. Atmospheric and Oceanic Fluid

Dynamics: Fundamentals and Large-scale Circulation.

Cambridge University Press, Cambridge, UK, 745 pp.

van Aken, H.M., 2000. The hydrography of the mid-latitude

northeast Atlantic Ocean I: The deep water masses.


van Aken, H.M., van Veldhoven, A.K., Veth, C., de

Ruijter, W.P.M., van Leeuwen, P.J., Drijfhout, S.S., et al.,

2003. Observations of a young Agulhas ring, Astrid,

during MARE in March 2000. Deep-Sea Res. II 50, 167e195.

REFERENCES 541

Van der Vaart, P.C.F., Dijkstra, H.A., Jin, F.F., 2000. The Pacific

cold tongue and the ENSO mode: A unified theory within

the Zebiak-Cane model. J. Atmos. Sci. 57, 967e988.

Van Dorn, W.G., 1993. Oceanography and Seamanship,

second ed. Dodd, Mead Publishers, New York, 440 pp.

VanScoy, K.A., Druffel, E.R.M., 1993. Ventilation and transport

of thermocline and intermediate waters in the

northeast Pacific during recent El Ninos. J. Geophys.

Res. 98, 18083e18088.

Vellinga, M., Wood, R.A., 2002. Global climatic impacts of

a collapse of the Atlantic thermohaline circulation.

Climatic Change 43, 251e267.

Venegas, R.M., Strub, P.T., Beier, E., Letelier, R.,

Thomas, A.C., Cowles, T., et al., 2008. Satellite-derived

variability in chlorophyll, wind stress, sea surface

height, and temperature in the northern California

Current System. J. Geophys. Res. 113, C03015.

doi:10.1029/2007JC004481.

Veronis, G., 1966. Wind-driven ocean circulationepart II.

Numerical solution of the nonlinear problem. Deep-Sea

Res. 13, 30e55.

Vialard, J., Menkes, C., Anderson, D.L.T., Balmaseda, M.A.,

2003. Sensitivity of Pacific Ocean tropical instability waves

to initial conditions. J. Phys. Oceanogr. 33, 105e121.

Visbeck, M., 2002. The ocean’s role in climate variability.

Science 297, 2223e2224.

Visbeck, M., Chassignet, E.P., Curry, R.G., Delworth, T.L.,

Dickson, R.R., Krahmann, G., 2003. The ocean’s

response to North Atlantic Oscillation variability. In:

The North Atlantic Oscillation: Climate significance and

environmental impact. Geophys. Monogr. Ser. 134,

113e146.

Vivier, F., Provost, C., 1999. Direct velocity measurements in

the Malvinas Current. J. Geophys. Res. 104, 21083e21103.

Von Storch, H., Zwiers, F.W., 1999. Statistical Analysis in

Climate Research. Cambridge University Press, Cambridge,

UK, 496 pp.

Wacongne, S., 1990. On the difference in strength between

Atlantic and Pacific undercurrents. J. Phys. Oceanogr.

20, 792e800.

Wacogne, S., Piton, B., 1992. The near-surface circulation in

the northeastern corner of the South Atlantic Ocean.

Deep-Sea Res. A 39, 1273e1298.

Wadhams, P., Budéus, G., Wilkinson, J.P., Løyning, T.,

Pavlov, V., 2004. The multi-year development of longlived

convective chimneys in the Greenland Sea. Geophys.

Res. Lett. 31, L06306. doi:10.1029/2003GL019017.

Wadhams, P., Comiso, J., Prussen, E., Wells, S.T.,

Brandon, M., Aldworth, E., et al., 1996. The development

of the Odden ice tongue in the Greenland Sea during

winter 1993 from remote sensing and field observations.

J. Geophys. Res. 101, 18213e18235.

Wadhams, P., Holfort, J., Hansen, E., Wilkinson, J.P., 2002. A

deep convective chimney in the winter Greenland Sea.

Geophys. Res. Lett. 29, 10. doi:10.1029/2001GL014306.

Walin, G., 1982. On the relation between sea-surface heat flow

and thermal circulation in the ocean. Tellus 34, 187e195.

Wallace, J.M., 1992. Effect of deep convection on the regulation

of tropical sea surface temperature. Nature 357,

230e231.

Wallace, W.J., 1974. The development of the chlorinity/

salinity concept in oceanography. Elsevier Oceanography

Series 7, 227 pp.

Wang, C., 2002. Atlantic climate variability and its associated

atmospheric circulation cells. J. Clim. 15, 1516e1536.

Warren, B.A., 1981. Deep circulation of the world ocean. In:


Oceanography. MIT Press, Cambridge, MA, pp. 6e41.

Warren, B.A., 1990. Suppression of deep oxygen concentrations

by Drake Passage. Deep-Sea Res. 37, 1899e1907.

Warren, B.A., Johnson, G.C., 2002. The overflows across the

Ninetyeast Ridge. Deep-Sea Res. II, 1423e1439.

WCRP (World Climate Research Programme), 1998. CLI-

VAR Initial Implementation Plan. WCRP-103, WMO/TD

No. 869, ICPO No. 14, 367 pp.

Webb, D.J., 2000. Evidence for shallow zonal jets in the

South Equatorial Current region of the southwest

Pacific. J. Phys. Oceanogr. 30, 706e720.

Webster, P.J., Magana, V.O., Palmer, T.N., Shukla, J.,

Tomas, R.A., Yanai, M., et al., 1998. Monsoons: Processes,

predictability, and the prospects for prediction. J. Geophys.

Res. 103, 14451e14510.

Webster, P.J., Moore, A.M., Loschnigg, J.P., Leben, R.R., 1999.

Coupled ocean-atmosphere dynamics in the Indian

Ocean during 1997-98. Nature 401, 356e360.

Weiss, R.F., Bullister, J.L., Gammon, R.H., Warner, M.J., 1985.

Atmospheric chlorofluoromethanes in the deep equatorial

Atlantic. Nature 314, 608e610.

Weller, R., Dean, J.P., Marra, J., Price, J., Francis, E.A.,

Boardman, D.C., 1985. Three-dimensional flow in the

upper ocean. Science 118, 1e22.

Whitworth, T., Orsi, A.H., Kim, S.-J., Nowlin, W.D., 1998.

Water masses and mixing near the Antarctic slope front.

In: Jacobs, S.S., Weiss, R.F. (Eds.), Ocean, Ice, and

Atmosphere: Interactions at the Antarctic Continental

Margins. Antarctic Research Series 75, American

Geophysical Union, Washington, pp. 1e27.

Whitworth, T., Peterson, R., 1985. The volume transport of

the Antarctic Circumpolar Current from bottom pressure

measurements. J. Phys. Oceanogr. 15, 810e816.

Whitworth, T., Warren, B.A., Nowlin Jr., W.D., Rutz, S.B.,

Pillsbury, R.D., Moore, M.I., 1999. On the deep westernboundary

current in the Southwest Pacific Basin. Progr.

Oceanogr. 43, 1e54.

542

REFERENCES

Wick, G.A., 1995. Evaluation of the variability and predictability

of the bulk-skin sea surface temperature difference

with application to satellite-measured sea surface

temperature. Ph.D. Thesis, University of Colorado,

Boulder, CO, 146 pp.

Wick, G.A., Emery, W.J., Kantha, L., Schluessel, P., 1996. The

behavior of the bulk d skin temperature difference at

varying wind speeds. J. Phys. Oceanogr. 26, 1969e1988.

Wijffels,S.,Bray,N.,Hautala,S.,Meyers,G.,Morawitz,W.M.L.,

1996. The WOCE Indonesian Throughflow repeat hydrography

sections: I10 and IR6. International WOCE Newsletter

24, 25e28.

Wijffels, S., Firing, E., Toole, J., 1995. The mean structure and

variability of the Mindanao Current at 8 N. J. Geophys.

Res. 100, 18421e18436.

Wijffels, S.E., 2001. Ocean transport of fresh water. In:

Siedler, G., Church, J. (Eds.), Ocean Circulation and

Climate. International Geophysics Series. Academic

Press, pp. 475e488.

Wijffels, S.E., Schmitt, R.W., Bryden, H.L., Stigebrandt, A.,

1992. Transport of fresh water by the oceans. J. Phys.


Wijffels, S.E., Toole, J.M., Davis, R., 2001. Revisiting the

South Pacific subtropical circulation: A synthesis of

World Ocean Circulation Experiment observations along

32 S. J. Geophys. Res. 106, 19481e19513.

Wilks, D.S., 2005. Statistical Methods in the Atmospheric

Sciences. In: International Geophysics Series, second ed.,

vol. 91. Academic Press, 648 pp.

Williams, G.D., Aoki, S., Jacobs, S.S., Rintoul, S.R.,

Tamura, T., Bindoff, N.L., 2010. Antarctic Bottom Water

from the Adelie and George V Land coast, East Antarctica

(140e149 E). J. Geophys. Res. 115, C04027.

doi:10.1029/2009JC005812.

Williams, R.G., 1991. The role of the mixed layer in setting

the potential vorticity of the main thermocline. J. Phys.

Oceanogr. 21, 1803e1814.

Williams, W.J., Carmack, E.C., Ingram, R.G., 2007. Chapter 2

Physical oceanography of polynyas, in Polynyas:

Windows to the World. Elsevier Oceanogr. Ser. 74, 55e85.

Willis, J., Roemmich, D.L., Cornuelle, B., 2004. Interannual

variability in upper-ocean heat content, temperature and

thermosteric expansion on global scales. J. Geophys. Res.

109, C12036. doi:10.1029/2003JC002260.

Wilson, C., 2002. Newton and celestial mechanics. In:

Cohen, I.B., Smith, G.E. (Eds.), The Cambridge

Companion to Newton. Cambridge University Press,

Cambridge, UK.

Wilson, T.R.S., 1975. Salinity and the major elements in

seawater. Ch. 6. In: Riley, J.P., Skirrow, G. (Eds.),

Chemical Oceanography, Vol. 1 (second ed.). Academic


Winsor, P., Rodhe, J., Omstedt, A., 2001. Baltic Sea

ocean climate: An analysis of 100 years of hydrographic

data with focus on freshwater budget. Clim. Res. 18,

5e15.

Winther, N.G., Johannessen, J.A., 2006. North Sea circulation:

Atlantic inflow and its destination. J. Geophys. Res.

111, C12018. doi:10.1029/2005JC003310.

Witte, E., 1902. Zur Theorie der Stromkabbelungen. Gaea,

38, 484e487(in German).

Wolanski, E. (Ed.), 2001. Oceanographic Processes of Coral

Reefs: Physical and Biological Links in the Great Barrier

Reef. CRC Press, Boca Raton, Florida, 356 pp.

Wolfram, 2009. Wolfram Demonstrations Project and

Wolfram MathWorld. Wolfram Research Inc. <http://

demonstrations.wolfram.com/> and <http://mathworld.

wolfram.com> (accessed 4.3.09).

Wolter, K., 2009. Multivariate ENSO Index (MEI).

NOAA Earth System Research Laboratory. http://www.

cdc.noaa.gov/people/klaus.wolter/MEI/ (accessed 3.26.09).

Wolter, K., Timlin, M.S., 1993. Monitoring ENSO in COADS

with a seasonally adjusted principal component index.

Proc. of the 17th Climate Diagnostics Workshop,

Norman, OK, NOAA/NMC/CAC, NSSL, Oklahoma

Climate Survey, CIMMS and the School of Meteor.,

University of Oklahoma, pp. 52e57.

Wong, A.P.S., Bindoff, N.L., Church, J.A., 2001. Freshwater

and heat changes in the North and South Pacific

Oceans between the 1960s and 1985e94. J. Clim. 14,

1613e1633.

Woodgate, R.A., Aagaard, K., 2005. Revising the Bering

Strait freshwater flux into the Arctic Ocean. Geophys.

Res. Lett. 32, L02602. doi:10.1029/2004GL021747.

Woods, J.D., 1985. The World Ocean Circulation Experiment.

Nature 314, 501e511.

Wooster, W.S., Reid, J.L., 1963. Eastern boundary currents.

In: Hill, M.N. (Ed.), The Sea, Vol. 2: Ideas and Observations.

Wiley-Interscience, New York, pp. 253e280.

World Meteorological Organization, 2005a. Natural hazards.

WMO. <http://www.wmo.int/pages/themes/hazards/

index_en.html> (accessed 9.22.05).

World Meteorological Organization, 2005b. Our World: International

Weather. WMO. <http://www.wmo.int/pages/

about/wmo50/e/world/weather_pages/chronicle_e.html>

(accessed 3.28.09).

Worthington, L.V., 1953. Oceanographic results of project

Skijump 1 and Skijump 2 in the Polar Sea 1951e1952.

Eos T. Am. Geophys. Union 34, 543.

Worthington, L.V., 1959. The 18 Water in the Sargasso Sea.


Worthington, L.V., 1976. On the North Atlantic circulation.

Oceanographic Studies. The Johns Hopkins University,

Baltimore, Maryland, 110 pp.

REFERENCES 543

Worthington, L.V., 1981. The water masses of the world ocean:

some results of a fine-scale census. In: Warren, B.A.,

Wunsch, C. (Eds.), Evolution of Physical Oceanography.

MIT Press, Cambridge, MA.

Worthington, L.V., Wright, W.R., 1970. North Atlantic Ocean

Atlas of potential temperature and salinity in the deep

water. Woods Hole Oceanographic Institution Atlas

Series, 2, 24 pp and 58 plates.

Wu, Y., Tang, C.L., Sathyendranath, S., Platt, T., 2007. The

impact of bio-optical heating on the properties of the

upper ocean: A sensitivity study using a 3-D circulation

model for the Labrador Sea. Deep-Sea Res. II 54,

2630e2642.

Wunsch, C., 1996. The Ocean Circulation Inverse Problem.

Cambridge University Press, New York, 458 pp.

Wunsch, C., 2009. The oceanic variability spectrum and

transport trends. Atmosphere-Ocean 47, 281e291.

Wunsch, C., Ferrari, R., 2004. Vertical mixing, energy, and

the general circulation of the oceans. Annu. Rev. Fluid

Mech. 36, 281e314.

Wüst, G., 1935. Schichtung und Zirkulation des Atlantischen

Ozeans. Die Stratosphäre. In Wissenschaftliche

Ergebnisse der Deutschen Atlantischen Expedition auf

dem Forschungs- und Vermessungsschiff “Meteor”

1925e1927,6 1st Part, 2, 109e288 (in German).

Wüst, G., 1957. Wissenschaftliche Ergebnisse der deutschen

atlantischen Expedition “Meteor”, vol. 6. Walter de

Gruyter, Berlin, part 2, pp. 1e208 (in German).

Wüst, G., 1961. On the vertical circulation of the Mediterranean

Sea. J. Geophys. Res. 66, 3261e3271.

Wyrtki, K., 1971. Oceanographic Atlas of the International

Indian Ocean Expedition. National Science Foundation

Publication. OCE/NSF 86-00-001, Washington, D.C,

531 pp.

Wyrtki, K., 1973. An equatorial jet in the Indian Ocean.

Science 181, 262e264.

Wyrtki, K., 1975. Fluctuations of the dynamic topography in

the Pacific Ocean. J. Phys. Oceanogr. 5, 450e459.

Wyrtki, K., Kilonsky, B., 1984. Mean water and current

structure during the Hawaii-to-Tahiti shuttle experiment.


Xue, H., Chai, F., Pettigrew, N., Xu, D., Shi, M., Xu, J., 2004.

Kuroshio intrusion and the circulation in the South

China Sea. J. Geophys. Res. 109, C02017. doi:10.1029/

2002JC001724.

Yanigomoto, D., Kawabe, M., Fujio, S., 2010. Direct velocity

measurements of deep circulation southwest of the

Shatsky Rise in the western North Pacific. Deep-Sea Res.

I 57, 328e337.

Yashayaev, I., 2007. Hydrographic changes in the Labrador

Sea, 1960-2005. Progr. Oceanogr. 73, 242e276.

Yasuda, I., Hiroe, Y., Komatsu, K., Kawasaki, K., Joyce, T.M.,

Bahr, F., et al., 2001. Hydrographic structure and transport

of the Oyashio south of Hokkaido and the formation

of North Pacific Intermediate Water. J. Geophys.

Res. 106, 6931e6942.

Yates, M.L., Guza, R.T., O’Reilly, W.C., Seymour, R.J., 2009.

Overview of seasonal sand level changes on southern

California beaches. Shore Beach 77 (1), 39e46.

Yoshikawa, Y., Church, J.A., Uchida, H., White, N.J., 2004.

Near bottom currents and their relation to the transport

in the Kuroshio Extension. Geophys. Res. Lett. 31,

L16309. doi:10.1029/2004GL020068.

Yu, Z., McCreary, J.P., Kessler, W.S., Kelly, K.A., 2000.

Influence of equatorial dynamics on the Pacific North

Equatorial Countercurrent. J. Phys. Oceanogr. 30,

3179e3190.

Yuan, X., 2004. ENSO-related impacts on Antarctic sea ice:

a synthesis of phenomenon and mechanisms. Antarct.

Sci. 16, 415e425. doi:10/1017/S0954102004002238.

Yun, J.-Y., Talley, L.D., 2003. Cabbeling and the density of

the North Pacific Intermediate Water quantified by an

inverse method. J. Geophys. Res. 108, 3118. doi:10.1029/

2002JC001482.

Zamudio, L., Hurlburt, H.E., Metzger, E.J., Morey, S.L.,

O’Brien, J.J., Tilburg, C.E., et al., 2006. Interannual variability

of Tehuantepec eddies. J. Geophys. Res. 111,

C05001. doi:10.1029/2005JC003182.

Zaucker, F., Broecker, W.S., 1992. The influence of atmospheric

moisture transport on the fresh water balance of

the Atlantic drainage basin: General circulation model

simulations and observations. J. Geophys. Res. 97,

2765e2773.

Zemba, J.C., 1991. The structure and transport of the Brazil

Current between 27 and 36 South. Ph.D. Thesis,

Massachusetts Institute of Technology and Woods Hole

Oceanographic Institution, 160 pp.

Zenk, W., 1975. On the Mediterranean outflow west of

Gibraltar. "Meteor" Forschungsergebnisse A16, 23e34.

Zhang, R., Vallis, G.K., 2006. Impact of Great Salinity

Anomalies on the low-frequency variability of the North

Atlantic Climate. J. Clim. 19, 470e482.

Zhang, Y., Hunke, E., 2001. Recent Arctic change simulated

with a coupled ice-ocean model. J. Geophys. Res. 106,

4369e4390.

Zhong, A., Hendon, H.H., Alves, O., 2005. Indian Ocean

variability and its association with ENSO in a global

coupled model. J. Clim. 18, 3634e3649.

Zhurbas, V., Oh, I.S., 2004. Drifter-derived maps of lateral

diffusivity in the Pacific and Atlantic Oceans in relation

to surface circulation patterns. J. Geophys. Res. 109,

C05015. doi:10.1029/2003JC002241.

Index

AABW, see Antarctic Bottom Water

AAIW, see Antarctic Intermediate

Water

Absolute salinity, 35, 37

Abyssal circulation, 218, 219e220

Abyssal hill, 11

Abyssal plain, 14

ACC, see Antarctic Circumpolar

Current

Acceleration, 188

Accuracy, 150, 186

ACoC, see Antarctic Coastal Current

Acoustic Doppler Current Profiler

(ADCP), 148, 165, 451

Acoustic tomography, 50

Acoustics, see Sound

ACW, see Alaskan Coastal Water

ADCP, see Acoustic Doppler Current

Profiler

Adiabatic compression, 33, 50

Adiabatic lapse rate, 33

Adiabatic temperature gradient, 33

Adjacent seas, 244

Advection, 115e116, 192

Age, ocean water, 102e103

Age, Seafloor, 10

Agulhas Current, 373, 381, 386e389,

474

Agulhas Retroflection, 375

Agulhas rings, 272e273, 375, 502, 507

AIW, see Arctic Intermediate Water

Alaska Current, 322

Alaskan Coastal Water (ACW),

420e421

Alaskan Stream, 322

Albedo, 124e125, 130e131

Aliasing, 149, 151, 168, 171, 185

AMM, see Atlantic Meridional Mode

AMO, see Atlantic Multidecadal

Oscillation

Amphidrome, 241

Amplitude, wave, 224

Angola Dome, 265

Angola-Benguela Front, 272

Anomaly, 185

Antarctic Bottom Water (AABW), 90,

91, 185, 218, 268, 297, 361,

397, 445, 449, 456, 461e465,

498

Antarctic Circumpolar Current

(ACC), 73, 142, 163, 165, 207,

216, 245, 247, 268, 304, 363,

364, 437, 438, 444, 445, 449,

450, 474, 475

Antarctic Coastal Current (ACoC),

441, 444


(AAIW), 90, 282, 295, 334,

351, 355, 358, 379, 444, 445,

448, 458e460, 490, 494

Antarctic Slope Front (ASF), 441, 444,

445, 446

Antarctic Surface Water (ASW), 445,

448, 456, 457

Antarctic Zone (AZ), 438, 441, 444,

447

Anticyclonic flow, 206

Antipodal point, 238

Apogee, 240

Apparent optical properties,

seawater, 55

Arabian Sea surface water, 388

Arctic Intermediate Water (AIW),

358, 379, 444, 458

Arctic Ocean Deep Water, 407

Arctic Ocean

circulation

intermediate and deep circulation,

412, 414

upper layer, 414, 429

climate variation, 435

ice

build-up and break-up, 432e434

distribution, 430e432

drift, 412e414

drift and wind forcing, 412e414

icebergs, 434e435

overview, 23, 26, 417e420

transports and budgets, 427e430

water masses

Atlantic Water, 421

deep and bottom water, 421e427

overview, 417e419

surface and near-surface waters,

419e421

Arctic Oscillation, 175, 435

ARM program, see Atmospheric

Radiation Monitoring

program

ASF, see Antarctic Slope Front

Aspect ratio, 6

ASW, see Antarctic Surface Water

Atlantic Meridional Mode (AMM),

301

Atlantic Multidecadal Oscillation

(AMO), 249, 301, 412, 435

Atlantic Niño, 301

Atlantic Ocean

buoyancy forcing, 250

climate variability, 301

depth-dependence of circulation

Deep Western Boundary Currents,

277e279

wind-driven circulation,

273e275

meridional overturning circulation,

245, 248, 280e281

North Atlantic circulation

Canary Current System,

257e259

eddy variability and Gulf Stream

rings, 262e263

Gulf Stream, see Gulf Stream

North Atlantic Current, 259e260

overview, 245e249

Portugal Current System, 257

545

546

Atlantic Ocean (Continued)

subpolar circulation, 260e262

subtropical circulation, 252

overview, 21, 25e26, 245

South Atlantic circulation

Benguela Current System,

271e272

Brazil Current, 270

eddy variability and Agulhas

rings, 272e273

Malvinas Current, 271

overview, 268e269

Subantarctic Front, 271

subtropical gyre, 269e270

tropical circulation, 263e268

water masses

Antarctic Intermediate Water, 295

Central Water and Subtropical

Underwater, 288

deep and bottom waters, 295e301

Labrador Sea Water, 290e292

Mediterranean Water, 292e295

Mode Water, 288e289

potential temperature versus

salinity and oxygen, 283e286

surface water and mixed layer,

286e288

wind forcing, 249e250

Atlantic Water (AW), 421, 429, 430

Atlas, generation, 177e179

Atmospheric Radiation Monitoring

(ARM) program, 124

Autocorrelation, 155

Autocovariance, 155

Available potential energy, 211

AW, see Atlantic Water

AZ, see Antarctic Zone

Azores Current, 257, 496, 503, 507

Baffin Bay, 410e412

Baltic Sea, circulation properties, 248

Baroclinic deformation radius, 210

Baroclinic instability, 211

Barotropic instability, 211

Basins, see specific oceans

Bay of Bengal surface water, 388

BCS, see Benguela Current System

Beach, 15, 225e229

Beaufort Gyre, 430, 431

Benguela Current System (BCS), 245,

271, 272, 277

Bering Sea, circulation properties,

244

INDEX

Bering Strait, 306, 358, 475

Bias error, 150e154

Bjerknes feedback, 218, 349

Black body, 127

Black Sea

circulation properties, 245

positive water balance, 119e120

Brazil Current, 136, 280, 286, 507

Brazil-Malvinas confluence, 277, 280

Breakers, 228e232

Break-up, ice, 63

Brine rejection, 61e62

BrunteVäisälä frequency,

see Buoyancy frequency

Buoyancy flux, 140e142

Buoyancy forcing

abyssal circulation and Deep

Western Boundary Currents,

220

Atlantic Ocean, 250

buoyancy gain, 220

buoyancy loss processes, 220e221

Indian Ocean, 381

Pacific Ocean, 316, 348e349

Southern Ocean, 457e459

thermohaline oscillators, 224

Buoyancy frequency, 44, 196

Cabbeling, 41e43, 186

Calcium ion, seawater composition,

34

California Current, 164, 203, 312

California Current System (CCS),

316e318, 323, 343, 351

California Undercurrent (CUC), 326

Canada Basin, 417

Canadian Basin, 417, 443e445

Canadian Basin Deep Water

(CBDW), 436e440

Canary Current System, 252,

259e262

Capillary wave, 30

Carbon-14, 46e47, 101

Cariaco Trench, 97

Caribbean Current, 254

Cayman Current, 254

CBDW, see Canadian Basin Deep

Water

CCS, see California Current System

CDW, see Circumpolar Deep Water

Celsius scale, 30

Central Indian Ridge, 26, 249

Central Water, 76, 294, 363, 402

Centrifugal force, 190e192

Centripetal force, 190

CFCs, see Chlorofluorocarbons

Channel, 19

Charlie Gibbs Fracture Zone, 261

Chi-squared distribution, 156

Chlorine ion, seawater composition,

32

Chlorinity, 33e34

Chlorofluorocarbons (CFCs), 47,

101e103, 297, 308, 366, 436

Chlorophyll, 56e57, 107, 109

Circumpolar Deep Water (CDW), 99,

220, 472, 474, 476e482,

486e487

Climate change, versus variability,

534

Climatology, generation, 178e180,

188

Co-oscillation tide, 242

COADS, see Comprehensive Ocean

Atmosphere Data Set

Coast, 15e16

Coastal-trapped wave, 236

Cold tongue, 220, 352

Color, ocean, 55e58, 108e109

Comprehensive Ocean Atmosphere

Data Set (COADS), 120, 134

Conductivity, seawater, 34e36

Conductivity-temperature-depth

profiler (CTD), 150, 152,

160, 162

Confidence interval, 154, 157e158

Confused sea, 225

Conservation of heat, see Heat budget

Conservation of salt, 115e117

Conservation of volume

closed box, 112

open ocean continuity, 112e114

overview, 111e112

Conservative tracer, 44

Continental rise, 17

Continental shelf, 17

Continental shelf wave, 237

Continental slope, 17

Convection, 220e221

Coral reef, 244

Core sampling, bottom material,

22e23

Coriolis acceleration, 188, 201

Coriolis force, 75, 194, 200, 202e207

Coriolis parameter, 190, 207e209,

217, 233

INDEX 547

Correlation, 185

Costa Rica Dome, 318

Cotidal line, 241

Coupled Model Intercomparison

Project, 485

Covariance, 155, 185

CTD, see Conductivity-temperaturedepth

profiler

CUC, see California Undercurrent

Cyclone, 202

Cyclonic flow, 206

Data assimilation, 159

Davidson Current, 316

Deacon cell, 486

Deep-sea bottom, 17e19

Deep-water wave, 224

Deep Western Boundary Current

(DWBC), 220, 245, 252, 256,

260, 273, 278, 290, 297, 305,

328, 361, 384, 397, 407, 455,

478, 501

Degrees of freedom, 155e157

Denmark Strait Overflow Water

(DSOW), 277, 285, 286, 295

Density, water

distribution

overview, 96

potential density depth

distribution, 82e94

pycnocline, 96e97

sea surface and upper layer,

94e96

evolution equations, 192

ice, 60

neutral density, 41e43, 94

potential density, 40e41

pressure effects, 39e40

temperature and salinity effects, 39

Density anomaly, 204

Depositional coast, 16

Descriptive physical oceanography,

definition, 1

Determination, 147

Diapycnal downwelling, 218e219

Diapycnal mixing, 191

Diapycnal upwelling, 219

Dichothermal layer, 71, 353

Diffusion, 113

Diffusivity, 193

Dimensions, ocean, 7e9

Dispersion relation, wave, 210

Dissipation, see Viscous force

Dissipative beach, 227

Dissolved oxygen, 98e99

Doppler shift, 53

Double diffusion, 44, 77, 195, 196

Downwelling, global circulation, 480

Drake Passage, 271, 449e451, 455,

457, 469, 475, 497

DSOW, see Denmark Strait Overflow

Water

Duration, wind, 225

DWBC, see Deep Western Boundary

Current

Dynamic height, 208e210

Dynamic physical oceanography,

definition, 1

Dynamic viscosity, 191

EAC, see East Australian Current

EACC, see East African Coastal

Current

EAP, see East Atlantic Pattern

Earth Radiation Budget Experiment

(ERBE), 124, 129, 138

East African Coastal Current

(EACC), 364, 366, 373, 387,

395

East Atlantic Pattern (EAP), 301

East Auckland Current, 325

East Australian Current (EAC), 136,

304, 323e324, 327, 353, 366,

474

East China Sea, circulation

properties, 245

Eastern boundary current, see Winddriven

circulation

Eastern Gyral Current, 377

East Greenland Current (EGC), 245,

262, 404, 407, 421, 429, 432,

474

East Indian Coastal Current, 368

East Kamchatka Current,

see Oyashio/East

Kamchatka Current

East Madagascar Current (EMC),

368, 372

East Pacific Rise (EPR), 11, 24

EBDW, see Eurasian Basin Deep Water

Echo sounder, 19, 52

Ecliptic orbit, 240

ECMWF, see European Center for

Medium-range Weather

Forecasts

Eddy diffusivity, 115, 193

Eddy kinetic energy (EKE), 262e263,

272, 327, 375, 377, 467,

502e503, 507

Eddy viscosity, 191, 198

Edge wave, 229

EDW, see Eighteen Degree Water

EGC, see East Greenland Current

EIC, see Equatorial Intermediate

Current

Eighteen Degree Water (EDW), 286,

288

EKC, see Oyashio/East Kamchatka

Current

EKE, see Eddy kinetic energy

Ekman layer, 197e199, 212, 217

Ekman number, 191

Ekman pumping, 199, 212

Ekman response, wind forcing, 75

Ekman transport, 199, 212, 218, 314,

316, 317, 320, 341, 439, 465

El NiñoeSouthern Oscillation

(ENSO), 2, 152, 249, 301,

303, 346, 379, 469, 471

description, 347e349

mechanisms, 349e350

EMC, see East Madagascar Current

Empirical orthogonal function (EOF),

174e177, 185

ENSO, see El NiñoeSouthern

Oscillation

Entrainment, 194

EOF, see Empirical orthogonal

function

EOS 80, 38

EPR, see East Pacific Rise

Equation of state (EOS), 38

Equatorial Intermediate Current

(EIC), 266, 337, 339

Equatorial region, 68

Equatorial stacked jets, 339

Equatorial Undercurrent (EUC), 218,

265, 304, 325, 337, 368

Equatorial Water, 347

Equilibrium tide, 237e240

ERBE, see Earth Radiation Budget

Experiment

Erosional coast, 16

Estimation, 185

Estuarine circulation, 111

Estuary, classification, 243e244

EUC, see Equatorial Undercurrent

Eulerian framework, 190

Euphotic zone depth, 108

548

Eurasian Basin, 415, 442

Eurasin Basin Deep Water (EBDW),

437e440, 444

European Center for Medium-range

Weather Forecasts

(ECMWF), 134

Evaporation, 118e120, 129e132

Expendable bathythermograph

(XBT), 146, 154, 161, 466

Feedback

Bjerknes, 217e218

Ice-albedo, 130

Positive, 218

Fetch, wind, 225

Fick’s law of diffusion, 114

Filtering, data, 172e174, 188

Fine structure, 197

Flinders Current, 384

Florida Current, 253, 254, 257

Florida Strait, 252

Flushing time, see Residence time

Flux, 114

Flux convergence, 114

Flux divergence, 114

Fofonoff model, inertial current, 217

Folger, Timothy, 3

Fourier analysis, see Spectral analysis

Fracture zone, 10e11, 19

Franklin, Benjamin, 3

Freezing point, seawater, 43e44

Freezing, see Ice

Freshwater transport, 116, 118, 119e121

Fully developed sea, 225

Garrett-Munk spectrum, 234

Gaussian distribution, see Normal

distribution

GBRUC, see Great Barrier Reef

Undercurrent

Geoid surface, 192

Geopotential anomaly, 205

Geostrophic balance

dynamic topography and sea

surface height maps, 209e210

geopotential and dynamic height

anomalies and reference level

velocities, 203e205

pressure gradient force and Coriolis

force balance, 198e202

two-layer ocean model, 201e206

Geostrophic velocity shear, 202

Global circulation

INDEX


eddy variability and diffusivity

diapycnal diffusion and nearinertial

motion, 505e507

energy and lateral diffusivity


scales, speeds, and coherence of

variability, 505e507

heat and freshwater transports, 113,

115, 136e139, 494e496


509e512

mass transports in layers into closed

regions, 494e502

overturning circulation schematics,

509e512

overturning transport

streamfunction, 504e509

overview, 490e495

sea level, 507e510

upper ocean systems, 495

upwelling and downwelling,

501e508

water mass distribution, 498e501

GRACE, see Gravity and Earth

Climate Experiment

Gravitational force, 189e192, 236

Gravity and Earth Climate

Experiment (GRACE), 205

Great Barrier Reef, 244

Great Barrier Reef Undercurrent

(GBRUC), 354

Great Whirl, 379

Greenland Sea Deep Water, 420, 438

Group velocity, 224e225

Guinea Dome, 265e270

Gulf of Alaska, 322

Gulf of Mexico, 230, 248e254

Gulf of Tehuantepec, 145

Gulf Stream, 3e5, 12, 135, 158, 214,

251e259, 278, 505, 507

Gulf Stream Extension, 252

Gulf Stream rings, 264e266, 507

Guyot, 12

Haline contraction coefficient, 41

Halmahera Eddy (HE), 352, 507

Halocline, 69, 74e76

HE, see Halmahera Eddy

Heat

global circulation, 501e508

ice melting, 60

water content, 31

Heat budget

components

annual mean values, 135e136

seasonal variations, 135e137

terminology, 119e121

shortwave radiation

absorbance in sea, 125e126

definition, 121e123

factors affecting penetrance,

122e123

input to sea, 124

longwave radiation

definition, 124

factors affecting, 128

net rate of heat loss, 126

outgoing longwave radiation,

129e130

sea surface temperature and

penetration depth, 127e129

ice and snow cover effects, 129e131

latent heat flux, 129e131

sensible heat flux, 132e135

meridional heat transport, 136e138

Heat flux, 31

Heat of vaporization, water, 26

Heat transport, 114

Helium-3, 45e47, 101

Horizontal variation, 162e164

Hotspot, 12

Hovmöller diagram, 165

Hudson Bay, 424e426

Hydrostatic balance, 28, 202

Hysteresis, 222

Ice

Arctic Ocean

build-up and break-up, 449e451


drift, 425e427

icebergs, 449e450

break-up, 62

brine rejection, 58e59

density and thermodynamics, 60

freezing point of seawater, 43e44

freezing process, 58e60

heat budget effects, 120e121

mechanical properties, 62e63

motion, 63

polynya, 64e65

Southern Ocean

cover, 468e469

motion, 469e470

types, 63

INDEX 549

Ice-albedo feedback, 130

Iceland-Faroe Front, 262

Iceland-Scotland Overflow Water

(ISOW), 286

IDW, see Indian Deep Water

IIW, see Indonesian Intermediate

Water

Indian Deep Water (IDW), 100, 396,

460, 483, 484

Indian Ocean




384e387

monsoonal and tropical ocean

circulation, 367e370

overview, 2, 363e365

Persian Gulf outflow, 381, 384

Red Sea outflow, 381e384

subtropical circulation

Agulhas Current, 373e375

Indonesian Throughflow,

379e381

Leeuwin Current, 377e379

subtropical gyre, 370e373

water masses

deep and bottom waters,

396e399

intermediate waters, 394e396

upper ocean, 387e394

wind forcing, 365e367

Indian Ocean Dipole, 399

Indonesian Intermediate Water

(IIW), 381, 390

Indonesian Throughflow (ITF), 342,

363, 365, 379, 458, 475,

483, 494

Indonesian Throughflow Water, 390

Inertial current, 196, 214

Integral timescale, 155e158

Interleaving, 196

Internal gravity wave

generation and observation, 235

interfacial internal gravity wave,

232e233

overview, 223

stratification, 233e235

International Arctic Buoy Program,

412

International Practical Temperature

Scale of 1968 (IPTS-68), 32

International Practical Temperature

Scale of 1990 (IPTS-90), 32

International Research Institute for

Climate and Society (IRI),

350

International Satellite Cloud

Climatology Project

(ISCCP), 124

Intertropical Convergence Zone

(ITCZ), 39, 88, 118, 128, 250,

264e265, 286, 307, 335, 343,

351

IPTS-68, see International Practical

Temperature Scale of 1968

IPTS-90, see International Practical

Temperature Scale of

1990

IRI, see International Research

Institute for Climate and

Society

Irminger Current, 262

Irradiance, 55

ISCCP, see International Satellite

Cloud Climatology Project

Isopleth, 15

Isopycnal surfaces, 38

Isopycnic potential vorticity, 45

Isotherm, 15

Isotope tracers, 48e49

ISOW, see Iceland-Scotland Overflow

Water

ITCZ, see Intertropical Convergence

Zone

ITF, see Indonesian Throughflow

Izu Ridge, 311

Jan Mayen Current, 408

Japan Sea, circulation properties, 219

Kelvin scale, 32

Kelvin wave, 209, 211, 343, 367

Kinetic energy, 211

Kuroshio, 304, 311, 321, 353, 474, 477

Kuroshio Countercurrent, 311

Kuroshio Extension, 308, 310, 311

Labrador Current, 252, 257, 260

Labrador Current, 403, 475

Labrador Sea Water (LSW), 90, 92,

262, 295e296, 298, 496, 497

LADCP, see Lowered acoustic

Doppler current profiler

Lagrandian framework, 188

Langmuir circulation (LC), 197

La Niña, 339, 342, 346

Latent heat flux, 131e132

Latent heat polynya, 64

LC, see Langmuir circulation

LCDW, see Lower Circumpolar Deep

Water

Least squares analysis, 158e160, 186

Leeuwin Current, 136, 363, 367, 372,

377e379, 474

Level of no motion, current, 203

Light

penetration and attenuation, 54,

56e57

seawater

ocean color, 57e60, 109e110

optical properties, 54e57,

106e110

LNADW, see Lower North Atlantic

Deep Water

Lomonosov Ridge, 417

Longwave radiation

definition, 121

factors affecting, 127

net rate of heat loss, 121

outgoing longwave radiation, 129

sea surface temperature and

penetration depth, 129

Loop Current, 254

Lower Circumpolar Deep Water

(LCDW), 331, 332, 361, 384,

397, 448, 464e466, 483, 498

Lowered acoustic Doppler current

profiler (LADCP), 161

Lower North Atlantic Deep Water

(LNADW), 292, 297, 299

LSW, see Labrador Sea Water

Madden-Julian oscillation, 349

Madeira Mode Water, 289

Magnesium ion, seawater

composition, 34

Makarov Basin, 417

Malvinas-Brazil Current, 401

Malvinas Current, 269, 271, 395, 441,

450

MAR, see Mid-Atlantic Ridge

Marginal seas, 7, 241

Mariana Trench, 9, 81

Mass transport, 113

Matlab, 170

Maud Rise polynya, 469

Maury, Matthew Fontaine, 2

ME, see Mindanao Eddy

Mean, 152, 186

550

Mediterranean Overflow Water, 458

Mediterranean salt tongue, 299

Mediterranean Sea

circulation properties, 246

negative water balance, 117

Mediterranean seas, 7e8

Mediterranean Water (MW), 68, 163,

292e295, 458, 497

Meridional direction, 68

Meridional overturning circulation

(MOC), 245, 248, 480

Meridional overturning transport

streamfunction, 485

Meriodonal heat transport,

140e142

Mesoscale, 3, 5, 502

Mesothermal layer, 71, 368

Microstructure, 196

Mid-Atlantic Ridge (MAR), 7, 10e11,

14, 24, 245, 293, 299, 479

Middle North Atlantic Deep Water

(MNADW), 299

Mindanao Current, 305, 307, 318, 333,

342

Mindanao Eddy (ME), 342

Mixed layer, 71, 74e76

bottom mixed layers, 194

internal mixing layers, 194e196

surface mixed layer, 194

MNADW, see Middle North Atlantic

Deep Water

MOC, see Meridional overturning

circulation

Mode Water, 75, 79, 97, 288e289

Modified Circumpolar Deep Water,

476

MODIS, 128

Molecular diffusivity, 193

Momentum balance

acceleration, 188

advection, 188

centrifugal force, 189e190

Coriolis force, 190

gravitational force, 188e189

mathematical expression, 192

pressure gradient force, 188

viscous force

eddy viscosity, 191

molecular viscosity, 190e191

Monsoon, 366e370

Multidimensional sampling

climatology and atlas generation,

177e179

INDEX

empirical orthogonal function,

174e177

overview, 173

time series data, 165

Multiple equilibria, 220

Munk model, western boundary

currents, 214

MW, see Mediterranean Water

NAC, see North Atlantic Current

NADW, see North Atlantic Deep

Water

NAO, see North Atlantic Oscillation

National Centers for Environmental

Prediction (NCEP),

134, 249

National Geophysical Data Center

(NGDC), 7

National Oceanic and Atmospheric

Administration (NOAA), 7

National Oceanographic Data Center

(NODC), 178

National Oceanography Centre,

Southampton (NOCS), 120,

135

NBC, see North Brazil Current

NCEP, see National Centers for

Environmental Prediction

NEADW, see Northeast Atlantic

Deep Water

NEC, see North Equatorial Current

NECC, see North Equatorial

Countercurrent

NEMC, see Northeast Madagascar

Current

Neutral density, 41e43, 93

Neutrally stable flow, 210

New Guinea Coastal Undercurrent

(NGCUC), 315, 340, 348,

354, 356

Newtonian fluid, 191

NGCUC, see New Guinea Coastal

Undercurrent

NGDC, see National Geophysical

Data Center

NICC, see North Intermediate

Countercurrent

Nitrate, 100e102, 181

NOAA, see National Oceanic and

Atmospheric

Administration

NOCS, see National Oceanography

Centre, Southampton

NODC, see National Oceanographic

Data Center

Non-conservative tracer, 47

Non-dispersive wave, 224

Non-dimensional parameter, 7

Nordic Seas

circulation, 405

overview, 402e405

vertical convection, 407e410

water masses, 405e407

Nordic Seas Overflow Waters

(NSOW), 295e297

Normal distribution, 154

North Atlantic Current (NAC), 245,

259e260, 261, 323, 405, 474

North Atlantic Deep Water (NADW),

90e91, 140, 164, 185, 248, 282,

297, 298e301, 306, 358, 359,

361, 373, 401, 402, 460, 473

North Atlantic Oscillation (NAO),

249, 301

North Brazil Current (NBC), 265, 268,

273

Northeast Atlantic Deep Water

(NEADW), 286

Northeast Madagascar Current

(NEMC), 364, 368


(NECC), 245e247, 265, 307,

326, 337

North Equatorial Current (NEC),

246, 252, 253, 257, 265, 304,

308, 318, 364, 475

Northern Annular Mode, see Arctic

Oscillation

North Intermediate Countercurrent

(NICC), 266

North Pacific Central Water (NPCW),

352

North Pacific Current, 308, 311e313

North Pacific Gyre Oscillation

(NPGO), 175, 362

North Pacific Index (NPI), 362

North Pacific Intermediate Water

(NPIW), 90, 358, 458, 473,

483, 486, 491, 497

North Pacific Subtropical Mode

Water (NPSTMW), 358

North Queensland Current (NQC),

342

North Sea, circulation properties, 244

North Subsurface Countercurrent

(NSCC), 334, 339, 342

INDEX 551

North Subtropical Front, 274

North Water Polynya, 434

Northwest Corner, 260

Northwest Monsoon Current, 370

Norwegian Atlantic Current, 417,

422, 429

Norwegian Sea Deep Water, 422e423

NPCW, see North Pacific Central Water

NPGO, see North Pacific Gyre

Oscillation

NPI, see North Pacific Index

NPIW, see North Pacific Intermediate

Water

NPSTMW, see North Pacific

Subtropical Mode Water

NQC, see North Queensland Current

NSCC, see North Subsurface

Countercurrent

NSOW, see Nordic Seas Overflow

Waters

Nutrients

distribution, 98e99

tracers, 45e46

Nyquist frequency, 151, 171, 188

Obduction, 216

Objective mapping, 162, 180

Observation, 147

Observational error, 147e148

Odden-Nordbukta, 408e409

Okhotsk Sea, circulation properties,

244

OLR, see Outgoing longwave

radiation

OMP, see Optimum multiparameter

analysis

Optical properties, seawater, 52e53

Optimum multiparameter analysis

(OMP), 148, 185e188, 501

Outgoing longwave radiation (OLR),

129e130

Overturning circulation, 189, 219

Overturning transport

streamfunction, 502e507

Oxygen

dissolved oxygen, 95e97

utilization rate, 45

Oxygen-18, 46

Oyashio/East Kamchatka Current

(EKC), 304, 321e322, 475

Pacific Decadal Oscillation (PDO),

175, 322, 362

Pacific Deep Water (PDW), 306, 325,

332, 334, 359e361, 407,

460e461, 483e486,

491e493, 498

Pacific Ocean



depth dependence of circulation,

338e340

El Niño, see El NiñoeSouthern

Oscillation

mesoscale eddy variability, 338

North Pacific circulation

subpolar circulation

Gulf of Alaska, 332

overview, 330e331

western boundary currents,

330e332

subtropical circulation

California Current System,

318e323

depth dependence, 323e324

Kuroshio, 318e322

North Equatorial Current, 298

North Pacific Current, 320

overview, 315e316

overview, 20, 25, 311e314

South Pacific subtropical circulation

East Australian Current, 333e335

overview, 332e333

Peru-Chile Current System,

336e338

South Equatorial Current, 338

South Pacific Current, 335e336

tropical circulation

equatorial current structure,

346e348

equatorial property distributions,

354e356

low latitude western boundary

currents, 352e354

overview, 346

seasonal variability, 355e356

wind and buoyancy forcing,

346e348

water masses

bottom water, 373

deep waters, 373e375



salinity, 360e362

upper waters, 363e365

wind forcing, 314

PAR, see Photosynthetically available

radiation

Particulate organic carbon (POC),

106, 107e109

Passage, 19

Passive margin, 16

PCCS, see Peru-Chile Current System

PDF, see Probability density function

PDO, see Pacific Decadal Oscillation

PDW, see Pacific Deep Water

Perigee, 240

Period, wave, 226

Persian Gulf, 244, 395, 396

Peru Current, 312

Peru-Chile Current System (PCCS),

307, 323, 325e326, 350e352

Peru-Chile Undercurrent, 336

PF, see Polar Front

PFZ, see Polar Frontal Zone

Phase velocity, 224

Phosphate, 100, 185

Photosynthetically available

radiation (PAR), 55, 109

Physical oceanography

overview, 1e3

space and time scales, 3e6,

14e15

Planetary vorticity, 209

Plate techtonics, 9e13

Plume, 194

Plunging breaker, 228

PML, see Polar Mixed Later

POC, see Particulate organic carbon

Polar Front (PF), 439, 444, 446,

467

Polar Frontal Zone (PFZ), 431e438,

444

Polar Mixed Later (PML), 419

Polar region, 68

Polar Surface Water, 421

Poleward undercurrent, 217

Polynya, 64e65, 432e434, 490

Portugal Current System, 252, 257

Potassium ion, seawater

composition, 34

Potential density, 39e40, 94

Potential energy, 211


Arctic Ocean, 427

Atlantic Ocean, 273e280

deep-water temperature, 81e83

overview, 33e34, 71

Pacific Ocean, 350e355

552

Potential temperature (Continued)

vertical sections, 83

volumetric distribution, 92

volumetric potential temperaturesalinity,

184e185, 187

Potential vorticity, 45, 184, 208e209

Practical salinity, 36e37

Practical salinity unit (psu), 36, 62, 69

Precipitation, 120e121

Precision, 150, 186

Pressure gradient force, 187,

188e189, 200e203

Pressure, water, 30e31

Probability density function (PDF),

154, 186

Production rate, see Ventilation rate

Progressive vector diagram, 167

PSS 78, 36e37, 38

psu, see Practical salinity unit

Pycnocline, 45, 72, 76e79, 96e97

Pycnostad, 97

Queen Charlotte Eddy, 507

QuikSCAT satellite, 142, 143

Radiance, 54e56

Radiation, 113

Random error, 151

Recirculation

Gulf Stream, 252

Kuroshio, 308

Recirculation gyre, 257, 311

Red Sea, circulation properties, 244

Red Sea Overflow Water (RSOW),

381, 382, 389

Red Sea Water (RSW), 486, 497

Redfield ratio, 48

Reference level, current, 204

Reference salinity, 37

Reflectance, 55, 125

Reflected radiance, 59

Reflective beach, 228

Refraction, 228

Remineralization, 25

Residence time, 106, 117

Reverberation, 53

Reynolds number, 191

Richardson number, 196

Rim Current, 262, 275, 421

Rip current, 16, 229e230

Romanche Fracture Zone, 11, 297

Root mean square error, see Standard

error

INDEX

Rossby deformation radius, 210

Rossby number, 6, 190

Rossby wave, 209e210, 505, 507

Ross Sea gyre, 452e454, 475

RSOW, see Red Sea Overflow Water

RSW, see Red Sea Water

Runoff, 118e120

SAC, see South Atlantic Current

SACCF, see Southern Antarctic

Circumpolar Current Front

SAF, see Subantarctic Front

SAFZ, see Subarctic Frontal Zone

Salinity

Arctic Ocean, 419, 420


compressibility effects, 41

conservation of salt, 115e117

deep-water salinity, 90e91

definitions, 35, 37

density effects, 39

determination, 35e36

evolution equations, 192e193

freezing point depression, 45e46

Intermediate depth salinity, 89e90

Pacific Ocean, 363e365

range in world ocean, 69, 83, 87

surface salinity, 87e88

temperature relationship, 85

temporal variation, 91e92

upper layer salinity, 88e89


184e185, 187

Salt finger, 196

Salt transport, 113e114, 116

SAM, see Southern Annular Mode

Sampling, 149e150, 161

SAMW, see Subantarctic Mode Water

SAZ, see Subantarctic Zone

SB, see Southern Boundary

sBSW, see Summer Bering Strait

Water

Seafloor

age, 10

bottom material, 20e25

deep-sea bottom, 17e19

features, 14

mapping, 19e20

roughness, 11

spreading, 10e11

Sea level, 16e17, 494e495

Sea level pressure (SLP), 176,

412e413

Seamount, 11e12

Seas, definition, 8

Sea surface height (SSH), 168,

206e207, 502, 505

Sea surface temperature (SST),

32e33, 69e70, 80, 129, 132,

149, 176, 321

Seawater, see Water

SEC, see South Equatorial Current

SECC, see South Equatorial

Countercurrent

Secchi disk, 58, 106e107

SEISAMW, see Southeast Indian

Subantarctic Mode Water

Sensible heat flux, 132e136

Sensible heat polynya, 64

Separation point, 256

Set, swell, 227

Set-up, wave, 231

Shadow zone, 216

Shallow salinity minimum, 89

Shelf Water, 458

Shore, 15

Shoreline, see Coast

Shortwave radiation

absorbance in sea, 126e127

definition, 124e125

factors affecting penetrance,

125e126

input to sea, 124

SICC, see South Intermediate

Countercurrent

Significant wave height, 227

Sill, 19

Sill depth, 19

Slope Water Current, 257

SLP, see Sea level pressure

Snell’s law of refraction, 228

Snow, heat budget effects,

128e129

Sodium ion, seawater composition,

34

SOFAR channel, see Sound fixing and

ranging channel

Solar constant, 125

Solitary wave, 235

Somali Current, 372, 395

Somali Jet, 369

SONAR, see Sound navigation and

ranging

Sound

Doppler shift, 53

reverberation, 53

INDEX 553

sources, 49

speed, 49e50

wave properties, 49

Sound fixing and ranging (SOFAR)

channel, 50

Sound navigation and ranging

(SONAR), 52

Source water type, 69

South Atlantic Current (SAC),

273e275

South China Sea, circulation

properties, 245

Southeast Indian Subantarctic Mode

Water (SEISAMW), 395e411

Southern Annular Mode (SAM), 174,

249, 374, 494

Southern Antarctic Circumpolar

Current Front (SACCF),

458, 460, 465, 467, 474

Southern Boundary (SB), 456, 460,

464e465, 474

Southern Gyre, 379

Southern Ocean

buoyancy forcing, 457e459

circulation

Antarctic Circumpolar Current,

464e467

mid-depth to bottom circulation,

455e456

Ross Sea gyre, 469e470

Weddell gyre, 468e469


eddies, 468e469

fronts, 455e456

ice

cover, 480e481

motion, 492e493

overview, 23, 26e27, 449e450

water masses

Antarctic Bottom Water, 482e485

Antarctic Intermediate Water,

473e474

Circumpolar Deep Water, 475e483

overturning budgets, 488

surface waters, 473e474

wind forcing, 455

zones, 457, 464e465


(SECC), 268, 332, 366e368

South Equatorial Current (SEC), 217,

245e247, 265, 269, 323,

326e327, 332, 337, 363e366,

368, 377

South Indian Current, 384, 386

South Intermediate Countercurrent

(SICC), 274

South Pacific Central Water (SPCW),

363

South Pacific Current (SPC), 332,

334e335

South Pacific Subtropical Mode

Water (SPSTMW), 362e364

South Subsurface Countercurrent

(SSCC), 346, 349, 352

South Subtropical Front, 274

Southwest Indian Ridge, 26

Southwest Monsoon Current, 380

SPC, see South Pacific Current

SPCW, see South Pacific Central Water

Specific volume, 40e41

Specific volume anomaly, 41

Spectral analysis, 167e168, 170e172,

176

Spilling breaker, 228

SPMW, see Subpolar Mode Water

Spreading center, 10

Spring tide, 240

SPSTMW, see South Pacific


SPZ, see Subpolar Zone

SSCC, see South Subsurface

Countercurrent

SSH, see Sea surface height

SST, see Sea surface temperature

Stable flow, 212

Standard deviation, 152, 188

Standard error, 152, 154

Static stability, water column, 44e45

Statistics

autocovariance, 155

confidence interval, 155, 157e158

covariance, 155

degrees of freedom, 157e158

integral timescale, 156e157

least squares analysis, 159e160

mean, 152

probability density function,

153e154

standard deviation, 154, 156

standard error, 154

variance, 154

variation, see Horizontal variation;

Temporal variation; Vertical

variation

STCC, see Subtropical

Countercurrent

StefaneBoltzmann constant, 127

Stefan’s Law, 123, 127

Steric height, 206

Steric height anomaly, 206

Stick plot, 166

STMW, see Subtropical Mode

Water

Stommel model

thermohaline oscillators, 220

western boundary currents, 215

Stommel-Arons model, 219

Storm surge, 230

Strait, 19

Strait of Gibralter, 117, 163, 248, 260,

299

Streamfunction, 204

STUW, see Subtropical Underwater

Subantarctic Front (SAF), 245, 269,

271, 286, 323e324, 350, 352,

358, 379, 391, 393, 438e441,

456, 486

Subantarctic Mode Water (SAMW),

269, 287, 379, 445, 486,

496e497

Subantarctic Surface Water, 468

Subantarctic Zone (SAZ), 456, 466,

486, 496

Subarctic Current, 318, 332

Subarctic Front, 264, 274e275

Subarctic Frontal Zone (SAFZ), 318,

322

Subarctic intermediate water, 86

Subduction, 11, 220, 278

Sublunar point, 237

Submesoscale, 318

Subpolar circulation, 260e262

Subpolar gyre, 206, 245, 275

Subpolar Mode Water (SPMW), 248,

275, 284, 287

Subpolar Region, 68, 441

Subpolar Zone (SPZ), 441, 444, 449

Substantial derivative, 188

Subtropical circulation, 252

Subtropical Countercurrent (STCC),

257, 308, 327

Subtropical Front, 375

Subtropical Frontal Zone, 308

Subtropical gyre, 206, 245, 269e270,

274e275, 325

Subtropical Mode Water (STMW), 69,

248, 284, 286, 351, 353, 392,

496

Subtropical region, 68

554

Subtropical Underwater (STUW),

248, 286, 288, 352, 391

Sulfate ion, seawater composition, 34

Summer Bering Strait Water (sBSW),

420e421

Sun glint, 125e126

Sunda Trench, 26

Surf zone, 227

Surface gravity wave

definition, 224

dispersion relation, 224e225

shore effects, 227e230

storm surge, 230

tsunami, 230e232

wind-forced surface gravity waves,

225e227

Surface tension, water, 30

Surging breaker, 228

Sverdrup balance, 211e213

Sverdrup transport, 143, 213, 249,

307, 373

Swash, 229

Swash zone, 227

Systematic error, 186

Tasman Front, 324

Temperature

compressibility effects, 41e43

density effects, 39

evolution equations, 192e193

mean value for world ocean, 69

potential temperature

deep-water temperature, 81e82

overview, 33e34, 71

vertical sections, 82e83

volumetric distribution, 92

sea surface temperature, 32e33,

71e74

temporal variation in upper layer

and thermocline, 79e81

upper layer temperature and mixed

layer, 74e76

water, 32e33

Temperature-salinity-time (T-S-t)

diagram, 183

Temporal variation

filtering data, 172e173

spectral analysis, 166, 168e172

time series data display, 166

vector data time-series analysis,

166

TEOS-10, 38

Thermal expansion coefficient, 43

INDEX

Thermal wind relation, 204

Thermobaricity, 44

Thermocline, 71e72, 76e82

Thermohaline circulation, 189, 220,

245

Thermohaline forcing, see Buoyancy

forcing

Thermostad, 71, 79

Tide

dynamic tides, 240e243

equilibrium tide, 237e240

Time variation, see Temporal

variation

TOGA, see Tropical Ocean Global

Atmosphere

Topography

deep sea, 9e13

mapping, 19e20

TPD, see Transpolar Drift

Tracers, seawater, 46e49, 101

Transient tracer, 48e49

Transmittance, 58, 125

Transpolar Drift (TPD), 402, 412,

414e417, 420

Trench, 11

Tritium, 49, 101

Tropical dipole mode, 399

Tropical Instability Wave (TIW),

266, 343

Tropical Ocean Global Atmosphere

(TOGA), 346

Tropical region, 68

T-S-t diagram, see Temperaturesalinity-time

diagram

Tsuchiya jets, 339

Tsunami, 230e232

Turbidity current, 17e18

Turnover time, 100e101, 103e105,

117

UCDW, see Upper Circumpolar Deep

Water

UNADW, see Upper North Atlantic

Deep Water

Undertow, 229

Unstable flow, 212

uPDW, see Upper Polar Deep

Water

Upper Circumpolar Deep Water

(UCDW), 281, 286, 334, 361,

441, 448, 460, 465

Upper North Atlantic Deep Water

(UNADW), 299

Upper Polar Deep Water (uPDW),

407, 422

Upwelling

global circulation, 474e478

irradiance, 55

coastal, 217

U.S. Geological Survey (USGS), 7

USGS, see U.S. Geological Survey

Variance, 152, 186

Ventilated region, 216

Ventilated thermocline, 216

Ventilation, 215

Ventilation rate, 100e101, 103

Vertical exaggeration, 15

Vertical variation

profiles, 161e162

sampling, 161

sections, 162

Viscous force

eddy viscosity, 191

molecular viscosity, 191

Volume transport, 112e113

Volumetric potential temperaturesalinity,

181e182

Vorticity, 207e211

Walker circulation, 218

Warm pool, 218

Water, see also Ice

color of ocean, 57e60, 108e109

compressibility, temperature and

salinity effects, 41e43

density

pressure effects, 39e40

temperature and salinity effects,

39

equation of state linearity

and nonlinearity, 38,

43e44

freezing point of seawater, 45e46

heat, 33

molecular properties, 29e30

optical properties of seawater,

54e57, 106e110

potential temperature, 33e34

pressure, 30e31

salinity and conductivity, 34e37

specific volume, 40e41

static stability and BrunteVäisälä

frequency, 44e45

temperature, 32e33

tracers, 46e49

INDEX 555

Water mass

Arctic Ocean

Atlantic Water, 421

deep and bottom water, 421e427

overview, 418e420

surface and near-surface waters,

419e421

Atlantic Ocean

Antarctic Intermediate Water, 295

Central Water and Subtropical

Underwater, 288

deep and bottom waters, 295

Labrador Sea Water, 290e292

Mediterranean Water, 292e295

Mode Water, 288e289


salinity and oxygen, 283e286

surface water and mixed layer,

286e288


global distribution, 494e501

Indian Ocean

deep and bottom waters, 396e399


upper ocean, 387e394

Nordic Seas, 402e405

optimum multiparameter analysis,

148, 183e185

overview, 67e68

Pacific Ocean

bottom water, 361

deep waters, 359e361



salinity, 350

upper waters, 350e355

Southern Ocean

Antarctic Bottom Water, 361e365

Antarctic Intermediate Water,

458e460

Circumpolar Deep Water,

460e461

overturning budgets, 465

surface waters, 456e458


184e185

Water-leaving radiance, 58e59

Water type, 69

Wave

coastal-trapped wave, 237

continental shelf wave, 237

general properties, 223e224

internal gravity wave

generation and observation, 235

interfacial internal gravity wave,

232e233

overview, 232

stratification, 233e235

Kelvin wave, 207, 209, 211, 349

Rossy wave, 209, 347, 349, 505

surface gravity wave

definition, 224

dispersion relation, 224e225

shore effects, 227e230

storm surge, 230

tsunami, 230e232

wind-forced surface gravity

waves, 225e227

Wavelength, 223

Wavelet analysis, 172

Weddell gyre, 452e454, 472

Weddell Sea Bottom Water,

465e466

Weddell Sea Deep Water, 465

West Australian Current, 352

West Greenland Current, 252

West Indian Coastal Current, 350

West Spitsbergen Current, 412

Western Subarctic Gyre, 328

Westward intensification, 213e215,

217e219

Whitecap, 225

Wien’s Law, 124

Wind-driven circulation

definition, 190

eastern boundary currents

coastal upwelling, 217e218

near-surface equatorial currents,

217e218

large scale inertial current, 219

Sverdrup balance, 215e217

ventilation, 216, 219

western boundary currents,

217e219

Wind forcing

Arctic Ocean, 427e429


Indian Ocean, 377e379

overview, 142, 147

Pacific Ocean, 314, 346e349

responses

Ekman layer, 201e202

Ekman transport convergence

and wind stress curl,

202e203

inertial current, 197

Langmuir circulation, 197e198

Southern Ocean, 455

surface gravity waves, 223e227

Wind-sea, 226

Wind stress, 200

Wind stress curl, 202e203

WOCE, see World Ocean Circulation

Experiment

World Ocean Atlas, 178e181

World Ocean Circulation Experiment

(WOCE), 363, 370, 372

World Ocean Data, 181

Worthington Gyre, 257

Wyrtki Jets, 368

XBT, see Expendable

bathythermograph

Yellow Sea, circulation properties,

244

Yucatan Current, 253

Zonal direction, 68

Added Refs from Supplemental Material

Chapter S1

Bleck, R., Boudra, D.B., 1981. Initial testing of a numerical

ocean circulation model using a hybrid quasi-isopycnal

vertical coordinate. J. Phys. Oceanogr 11, 755e770.

Bryan, K., 1969. A numerical method for the study of the

circulation of the world ocean. J. Comp. Phys. 4,

347e376.

Bryan, K., Cox, M.D., 1968. A nonlinear model of an ocean

driven by wind and differential heating. Part 1. J. Atmos.

Sci. 25, 945e978.

Defant, A., 1936. Die Troposphäre des Atlantischen Ozeans.

In Wissenschaftliche Ergebnisse der Deutschen Atlantischen


"Meteor" 1925e1927, 6 (1), 289e411 (in

German).

Inman, D.L., 2003. Scripps in the 1940s: the Sverdrup era.


Munk, W.H., 2000. Achievements in physical oceanography.

In 50 Years of Ocean Discovery: National Science

Foundation 1950e2000. National Academy Press,

Washington, D.C. 44e50.

Nierenberg, W.A., 1996. Harald Ulrik Sverdrup 1888e1957.

Biographical Memoirs 69. National Academies Press,


Shor, E., Day, D., Hardy, K., Dalton, D., 2003. Scripps time

line. Oceanography 16, 109e119.

Sverdrup, H.U., Munk, W.H., 1947. Wind, Sea, and Swell:

Theory of Relations for Forecasting. U.S. Navy Dept.,

Hydrographic Office, H.O. Pub. No, 601, 44 pp.

Chapter S5


available online. NASA Goddard Institute for Space



Chapter S9

Schott, F., Zantopp, R., Stramma, L., Dengler, M., Fischer, J.,

Wibaux, M., 2004. Circulation and deep water export at

the western exit of the subpolar North Atlantic. J. Phys.


Chapter S10

Australian Government Bureau of Meteorology, 2009. S.O.I

(Southern Oscillation Index) Archives d 1876 to present.

http://reg.bom.gov.au/climate/current/soihtm1.shtml


1

C H A P T E R

S1

Brief History of Physical Oceanography

Supplementary Web Site Materials for Chapter 1

This supplementary chapter contains an

eclectic and necessarily truncated treatment of

the history of physical oceanography. Numerous

books, journal issues, and memoirs provide

diverse resources. Among these, the Scripps

Institution of Oceanography’s library archive

provides a webpage that is an excellent place to

begin searching for original materials, biographies,

and institutional histories (SIO, 2011).

While the ocean has been the object of many

ancient science applications, the science of

oceanography is fairly young. Its origins are in

a great variety of earlier studies including

some of the earliest applications of physics

and mathematics to Earth processes. Archimedes,

the Greek physicist and mathematician,

can also be considered one of the earliest physical

oceanographers. The familiar Archimedes

principle describes the displacement of water

by a body placed in the water. Archimedes

also made extensive studies of harbors to

fortify them against enemy attack. Pytheas

was another early physical oceanographer; he

correctly hypothesized that the moon causes

the tides.

Many early mathematicians used their skills

to study the ocean. Sir Isaac Newton did not

directly work on problems of the ocean, but

his principle of universal gravitation was an

essential building block in understanding the

tides. Both Laplace and Legendre, who were

mathematicians, advanced the formal theory of

the tides (Laplace, 1790); Laplace’s equation is

a fundamental element in a description of the

tides. English mathematicians worked on

a mathematical description of the ocean waves

that surrounded their homeland. All of these

studies are clearly part of what we now know

as physical oceanography.

Early charting of the ocean’s surface currents

came hand in hand with exploration of coastlines

and ocean basins and was performed by the

earliest seafaring nations. Peterson, Stramma,

and Kortum (1996) provided an excellent review

of the history of ocean circulation mapping, from

the earliest Greek times, through the middle ages

and rise of the Arabian empire, through the

Renaissance and into the eighteenth and nineteenth

centuries. In the late eighteenth century,

John Harrison’s development of the chronometer

to measure longitude was a watershed, making

more accurate mapping possible. By the nineteenth

century, descriptions of subsurface and

even deep circulation were becoming possible.

S1.1. SCIENTISTS ON SHIPS

Early charts of the ocean circulation were

produced by mariners. Benjamin Franklin,

1

2

S1. BRIEF HISTORY OF PHYSICAL OCEANOGRAPHY

among his many different accomplishments,

was also a scientist, was one of the first to

make measurements at sea specifically to chart

its features (Figure 1.1b in the textbook). His

goal was to decrease the time required for mail

packets to cross the Atlantic from Europe to

the United States. Another source of sea-going

physical studies of the ocean came from studies

made by “naturalists” who went along on

British exploring expeditions. One example

was Charles Darwin, who went along as

the ship’s naturalist of the HMS Beagle on

a voyage to chart the southeast shore of South

America. This journey included many long

visits to the South American continent where

Darwin formulated many of his ideas about

the origin of species. During the cruise he

took measurements of physical ocean parameters

such as surface temperature and surface

salinity.

There were so many naturalists traveling on

British vessels in the early 1800s that the Royal

Society in London decided to design a set of

uniform measurements. Then Royal Society

secretary, Robert Hooke, was commissioned to

develop the suite of instruments that would be

carried by all British government ships. One

noteworthy device was a system to measure

the bottom depth of the deep ocean. It consisted

of a wooden ball float attached to an iron

weight. The pair was to be dropped from the

ship to descend to the ocean floor where the

weight would be dropped; the wooden ball

would then ascend to the surface where it

would be spotted and collected by the ship.

S1.2. ORGANIZED EXPEDITIONS

PRIOR TO THE TWENTIETH

CENTURY

In the eighteenth century, organized ocean

expeditions contributed valuable knowledge of

the oceans. One of the most successful ocean

explorers was Captain James Cook who made

three major exploring voyages between 1768

and 1780. On these cruises, British naturalists

observed winds, currents, and subsurface

temperatures; among other discoveries they

found the temperature inversion in the Antarctic,

with cold surface water lying over a warmer

subsurface layer.

In 1838 the U.S. Congress had the Navy organize

and execute the United States Exploring

Expedition to collect oceanographic information

from all over the world (see Chapman, 2004).

Many of the backers of this expedition saw it

as a potential economic boon, but others were

more concerned with the scientific promise of

the expedition. In 1836, $150,000 had been

appropriated for this expedition. As originally

conceived, the expedition was to benefit natural

history, including geology, mineralogy, botany,

vegetable chemistry, zoology, ichthyology, ornithology,

and ethnology. Some practical studies

such as meteorology and astronomy were also

included in the program. Most of the science

was to be done by a civilian science complement;

the Navy was to provide the transportation

and some help with the sampling. The

Navy did not like this arrangement and insisted

that a naval officer lead the entire expedition.

This responsibility was given to Lieutenant

Charles Wilkes who had earned the reputation

of being interested in and able to work on scientific

problems. At the same time it was widely

known that Wilkes was proud and overbearing,

with his own ideas on how this expedition

should be executed. Most of the scientific

positions were filled with naval personnel.

Only nine positions were offered to civilians

who were subject to all the rules and conditions

of behavior applying to the naval staff.

Unlike other later and more significant

single-ship expeditions, five naval vessels

carried out the United States Exploring Expedition.

Starting in Norfolk, Virginia, the expedition

sailed across the Atlantic to Madeira,

re-crossed to Rio de Janeiro, then south around

Cape Horn and into the Pacific Ocean. By the

ORGANIZED EXPEDITIONS PRIOR TO THE TWENTIETH CENTURY 3

time the ships had sailed up the west coast of

South America to Callao, Peru, storms had put

three ships out of commission. What remained

of the expedition crossed the Pacific and while

the “scientific gentlemen” were busy making

collections in New Holland and New Zealand,

two ships, the Vincennes and the Porpoise, sailed

south into the Antarctic region where Wilkes

believed that there was a large land mass behind

a barrier of ice. In the austral summer of

1839e1840, Wilkes sailed his ships south until

blocked by the northern edge of the pack ice.

He then sailed west along the ice barrier and

was able to get close enough to see the land.

At one point he came within a nautical mile of

the coast of “Termination Land” as Wilkes

named it. This was the most interesting part of

the expedition as far as Wilkes was concerned.

His alleged discovery of Antarctica was strongly

contested by the British explorer Sir James Clark

Ross, but it remains as the only well-known

benefit of this mission. Other possible claimants

to having discovered Antarctica were Captain

Nathaniel Palmer, an American sealing captain

who claimed to have sighted it in 1820, and

the Russian Fabian von Bellingshausen who circumnavigated

the Antarctic continent from 1819

to 1821 as part of a Russian Navy expedition.

During this same period there was an important

development in the United States. A Navy

lieutenant, Matthew Fontaine Maury, was seriously

injured in a carriage accident and was not

able to go to sea for many years. Instead he was

put in charge of a fairly obscure Navy office

called the Depot of Charts and Instruments

(1842e1861). This later became the U.S. Naval

Observatory. This depot was responsible for the

care of the navigation equipment in use at that

time. In addition it received and sent out logs to

be filled out by the bridge crew ships. Maury

soon realized that the growing number of ship

logs in his keeping was an important resource

that could be used to benefit many. His first

idea was to make use of the estimates of winds

and currents from the ships to develop

a climatology of the currents and winds along

major shipping routes. At first most people

were skeptical about the utility of such maps.

Luckily one of the clipper ship captains plying

the route between the east and west coasts of

the United States decided to see if he could use

these charts to select the best course of travel

for his next voyage. He found that this new information

made it possible to cut many days off of

his regular travel. As word got around, other

clipper ship captains wanted the same information

to help to improve their travel times. Soon

other route captains were doing the same and

Maury’s information became a publication

known as “sailing directions.” Even today the

U.S. Coast Guard continues to publish “Sailing

Directions,” although the publication has little

to do with sailing and more to do with harbor

approaches and changes in coastal conditions.

This publication was so successful that many

European nations decided to adopt similar practices.

Maury was invited to advise the European

nations on how to develop and implement similar

systems. In the United States he expanded his use

of these archived data and also expanded his

“depot” to include other oceanographic measurements.

Itwas underhis guidance thataLieutenant

Baker developed one of the first deep-sea

sounding devices. Baker stuck with the age-old

concept of measuring the ocean depth by dropping

a line from the surface. The problem had

been that in 4000 m of water the line became too

heavy to retrieve from the surface, so he designed

a new metal line whose cross section varied from

a very narrow gauge wire at the bottom to a much

thicker wire nearer the surface. In addition, Baker

followed one aspect of Hooke’s design and dropped

the weight at the bottom, again making the

system much lighter for retrieval. A later addition

was a small corer added to the end of the line to

collect a short (a few centimeters) core of the top

layer of sediment. This device led to the first

comprehensive map of bottom topography of

the North Atlantic. Unfortunately for Maury,

when the civil war broke out he returned to his

4


FIGURE S1.1 Track of the HMS Challenger Expedition 1872e1876.

native south and spent most of the war developing

explosive devices to destroy enemy ships

and to barricade harbors. An important part of

Maury’s legacy is a book, the Physical Geography

of the Sea, which remarkably is still in print

(Maury, 1855).

The first global oceanographic cruise was

made on the British ship the HMS Challenger.

This three-year (1872e1876) expedition (Figure

S1.1) was driven primarily by the interest of

a pair of biologists (William B. Carpenter and

Charles Wyville Thomson) in determining

whether or not there is marine life in the great

depths of the open ocean. Thomson was a Scot

educated as a botanist at the University of

Edinburgh, and in the late 1860s he was

a professor of natural history at Belfast, Ireland.

He had been working with his friend Carpenter,

a medical doctor, to discover if the contention by

another British naturalist (Edward Forbes) that

there was no life below 600 m (called the azoic

zone) was true. Even in the early phase of the

Challenger expedition dredges of bottom material

from as much as 2000 m had demonstrated

the great variety of life that exists at the ocean

bottom. In addition to biological samples, this

expedition collected a great number of physical

measurements of the sea such as sea-surface

SCANDINAVIAN CONTRIBUTIONS AND THE DYNAMIC METHOD 5

temperature and samples of the min-max

temperatures at various depths.

Along with Thomson and Carpenter, the

Challenger scientific staff consisted of a naturalist,

John Murray, and a young chemist, John

Young Buchanan, both from the University of

Edinburgh. The youngest scientist on the staff

was 25-year-old German naturalist Rudolf von

Willemoës-Suhm who gave up a position at

the University of Munich to join the expedition.

Henry Nottidge Moseley, another British naturalist

who had also studied both medicine and

science, joined the expedition after returning

from a Government Expedition to Ceylon.

Completing the staff was the expedition’s artist

and secretary, James John Wild. Much of the

visual documentation that we have from the

Challenger expedition came from the able pen

of James Wild. The addition of John Murray

was fortuitous in that he later saw to the publication

of the scientific results of the expedition.

Upon return, it was soon found that the

Challenger expedition had exhausted the funds

available for the publication of the results.

Fortunately Murray, who was really a student

from the University of Edinburgh, recognized

the value of the phosphate formations that

dominated Christmas Island. Claiming the

island for England, Murray later set up mining

operations on the island. The income from this

operation was later used to publish the

Challenger reports.

S1.3. SCANDINAVIAN

CONTRIBUTIONS AND THE

DYNAMIC METHOD

In the last quarter of the nineteenth century

a group of Scandinavian scientists began to

investigate the theoretical complexities of the

sea in motion. In the late 1870s, a Swedish

chemist, Gustav Ekman, began studying the

physical conditions of the Skagerrak, part of

the waterway connecting the Baltic and the

North Sea. Motivated by fisheries problems,

Ekman wanted to explain shoals of herring that

had suddenly reappeared in the Skagerrak after

an absence of 70 years. He discovered that in the

Skagerrak there are layers of less-saline water

from the Baltic “floating” over the deeper, more

saline North Sea water. At the same time he

found that herring preferred a particular water

layer of intermediate salinity. This shelf, or

bank water, as it was called, moved in and out

of the Inland Sea and with it went the fish.

Ekman knew that his results would not be of

any use to the fishermen unless the shelf water

and the other layers could be mapped. He joined

forces with another Swedish chemist, Otto

Pettersson, and together they organized a very

thorough series of hydrographic investigations.

Pettersson was to emerge from this experience

as one of the first physical oceanographers. It

should be noted that in Swedish “hydrography”

translates as “physical oceanography.

Pettersson and Ekman both understood that

to obtain a useful picture of the circulation

a series of expeditions involving several vessels

that could work together at many times

throughout each year would have to be organized.

This was a new approach to the study

of the sea. In the name of fisheries research

such a series of research cruises was begun in

the early 1890s. These were some of the first

cruises that emphasized the physical parameters

of the ocean. For the vertical profiling of

the ocean temperature a new device was available.

Since 1874, the English firm Negretti and

Zambra had manufactured a reversing thermometer

that recorded accurate temperatures

at depth.

During this time, another Scandinavian broke

new ground in the rush to reach the North Pole.

As a young man of 16, Norwegian Fridtjof

Nansen was the first person to walk across

Greenland. This exploring spirit led Nansen to

propose a Norwegian effort to reach the North

Pole. After studying evidence, Nansen decided

that there was a northwestward circulation of

6


ice in the Arctic. Instead of mounting a large

attack on the Arctic, Nansen wanted to build

a special ship that could withstand the pressures

of the sea ice when the ship was frozen into the

Arctic pack ice (Figure 12.7 in the textbook). He

believed that if he could sail as far east as

possible in summer he could then freeze his

ship into the pack ice and be carried to the

northwest. His plan was to get as close as

possible to the North Pole at which time he

and a companion would use dog sleds to reach

the pole and then return to the ship. Named the

Fram (“forward” in Norwegian), this unique

ship was too small to carry a large crew. Instead

Nansen gathered a group of nine men who

would be able to adapt to this unique experience.

Always a scientist, Nansen planned a large

number of measurements to be made during the

Fram’s time in the ice pack.

On March 1895 the Fram reached 84 N, about

360 miles from the pole (Figure 12.7). Nansen

believed that this was about as far north as the

Fram was likely to get. In the company of

Frederik Hjalmar Johansen and a large number

of dogs, Nansen left the relative comfort of the

Fram and set off to drive the dog sleds to the

North Pole. They drove slowly north over drifting

ice until they were within 225 miles of their

goal, farther north than any person had been

before. For three months they had traveled

over extremely rough ice, crossing what Nansen

referred to as “congealed breakers” and they

had lost their way. From their farthest north

point they turned south eventually reaching

Franz Josef Land where they hoped to encounter

a fishing boat in the short summer season.

Surviving by eating their dogs, Nansen and

Johansen were very fortunate to meet a British

expedition led by Frederick Jackson. In the

summer of 1896 they sailed home to Oslo aboard

the Windward. Meanwhile the Fram drifted

further west and south and emerged from the

ice pack just north of Spitsbergen. She sailed

back to Oslo and arrived just a week after Nansen

and Johansen.

One of Nansen’s primary objectives in the Fram

expedition was to form a more complete idea of

the circulations of the northern seas. This was

achieved by taking systematic measurements of

the temperatures and salinities of the Arctic

water. Using one of Pettersson’s insulated water

bottles, Nansen had attached a reversing thermometer

to sample the temperature and salinity

profiles. This arrangement, known as a “Nansen

bottle,” is still in use. Working in the Geophysical

Institute of the University of Bergen, Norway,

Nansen tried to explain the measurements made

by the Fram. The hydrographic measurements

suggested a very complex connection between

the Norwegian and Arctic Seas. The daily position

information from the Fram was also of great

interest for this study. As a young student, Ekman

worked on this problem with Nansen. Both were

interested to note that the Fram did not drift in the

same direction as the prevailing wind, instead it

differed from the wind by about 20 to 40 degrees

to the right.

Using the measurements made by the Fram

along with simple tank models of the Fram,

Ekman developed his theory of the wind-driven

circulation of the ocean. Published as part of

the Fram report, Ekman (1905) postulated the

response of the ocean to a steady wind in

a uniform direction. Making some simple

assumptions about the turbulent viscosity of

the ocean, Ekman could show how the ocean

current response to a steady wind must have

a surface current 45 degrees to the right of the

wind in the Northern Hemisphere. Below that

there is a clockwise (Northern Hemisphere)

spiral of currents (called the Ekman spiral)

down to a depth where the current vanishes.

In spite of these successes with the Fram data,

Nansen realized that he could have done much

more. This was motivated by the development

of the “dynamic method” for estimating geostrophic

ocean currents (see Chapter 7 in the textbook).

Developed also in Bergen, this method

made it possible to map currents at every level

from a detailed knowledge of the vertical density

THE METEOR EXPEDITION 7

structure. The Fram’s measurements were not

detailed enough to take advantage of this technique.

This theory was furthered developed by

Wilhelm Bjerknes, a professor of meteorology

at the University of Oslo, who coined the term

“geostrophy” from the Greek geo for earth and

strophe meaning turning.

Two other Scandinavian physical oceanographers

of this period were Johan Sandström and

Bjorn Helland-Hansen, both of whom were

interested in the ocean circulation and its

measurement. The Norwegian Board of Sea

Fisheries had invited Helland-Hansen, Nansen,

and Johan Hjort to participate in the first cruise

of their new research vessel. They were responsible

for the collection of hydrographic measurements.

A new problem surfaced while they were

collecting their measurements. In their process

of measuring salinity it was necessary to have

a “reference sea water” to make the measurement

precise, since slightly different methods

and procedures were being used. At this time

a Danish physicist, Martin Knudsen, was

working on a set of hydrographical tables that

would clearly define the relationship between

temperature, salinity, and density. At the 1899

meeting of the International Council for the

Exploration of the Sea (ICES), Knudsen had

proposed that such tables be published to facilitate

the standardization of hydrographic work

(Knudsen, 1901). For this same reason Knudsen

suggested that a standard or normal water

be created and distributed to oceanographic

laboratories throughout the world as a standard

against which all salinity measurements

could be compared. Knudsen then proceeded

to set up the Hydrographical Laboratory for

ICES in Copenhagen and the standard seawater

later became known as “Copenhagen Water.”

He also published standard tables called

“Knudsen Tables,” which displayed the relationships

between chlorinity, salinity, densities,

and temperature.

Nansen and Helland-Hansen’s careful study

of the Norwegian Sea made it the most

thoroughly studied and best-known body of

water in the world. The new method of

computing geostrophic currents had played

a large role in defining the circulation of the

Norwegian Sea. This “dynamic method,” as it

was called, was slow to spread to other regions.

Then, around 1924, a German oceanographer

named Georg Wüst applied the dynamic

method to the flows at different levels through

the Straits of Florida. He compared the results

to the current profiles collected in the 1880s by

a Lieutenant Pillsbury in the same area with

a current meter. The patterns of the currents

were essentially the same and confidence in

the dynamic method increased. Another test of

the dynamic method arose when the International

Ice Patrol (IIP) began to compute the

circulation of the northwest Atlantic to track

the drift of icebergs. Created after the tragic

sinking of the Titanic, the IIP was charged

with mapping the positions and drifts of

icebergs released into Baffin Bay from the

glaciers on Ellesmere Island.

S1.4. THE METEOR EXPEDITION

German scientists performed the real test of

the dynamic method on the Meteor expedition

in the Atlantic from 1925 to 1927 (Spiess, 1928).

This expedition was conceived by a German

naval officer, Captain Fritz Spiess, to create an

opportunity for a German navy vessel to visit

foreign ports (prohibited by the treaty at the

end of World War I) in the capacity of an ocean

research vessel. Captain Spiess had served both

prior to and during the war as a hydrographer

in the German navy. He realized that to be

successful he must find a recognized German

scientist to be the “father” of the expedition.

Spiess presented his idea to Professor Alfred

Merz, then the head of the Oceanographic Institute

in Berlin. Merz had been educated as a physical

geographer, but had always worked on the

physics of the ocean. He was happy to accept

8


FIGURE S1.2 Overturning circulation of the Atlantic Ocean according to Merz and Wüst (1923).

the role of scientific leader of the future ocean

expedition. This interest included the participation

of his son-in-law and former student Georg

Wüst, who was previously mentioned with

respect to his use of the dynamic method.

Prior to the Meteor expedition, Merz and

Wüst collected all of the German and British

hydrographic observations and presented

a new vision of the horizontal and vertical circulation

in the Atlantic with different water

masses in thick layers (Figure S1.2). Our present

view of the Atlantic’s “overturning circulation”

is not very different from their concept. Richardson

(2008) provided an excellent overview of the

history of charting the overturning circulation

from these early attempts to the present.

The verification and improved resolution of

this proposed circulation became the focus for

the expedition. Because the Meteor was not

a very large ship, it was decided that the crew

would have to help out in many measurement

programs. Consequently, many crewmembers

were sent to school at the Oceanography Institute

in Berlin. In addition it was decided to execute

a “test or shakedown cruise” to determine if all

the equipment was working properly. This cruise

went from Wilhelmshaven on the North Sea to

theAzoresandback.Thispre-cruiseturnedout

to be a very wise move, resulting in a number

of very basic changes. The smokestack was

lengthened in an effort to get the heat of the

engines higher off the deck. In the tropics the

lack of good ventilation on the ship became

a serious problem and a lot of work had to be

done on the deck. The unique system developed

for the Meteor to anchor in the deep ocean had to

be corrected. In addition, the forward mast was

set up to carry more sail to save coal on some

of the longer sections (Figure S1.3).

There were also some interesting personnel

changes that were arranged after the preexpedition.

Most important was the fact that

a chemist who was to be in charge of the salinity

titrations was found to be colorblind. (The titration

has a color change at the end point.) It was

then necessary to find someone who could do

the salinity titrations. The solution was that

Wüst, although not originally slated to participate

in the expedition, was taken along to

titrate the salinity samples. This later became

very important since the expedition leader,

Dr. Merz, passed away in Montevideo after the

THE METEOR EXPEDITION 9

FIGURE S1.3 Meteor after refit. Source: From Spiess (1928).

first of the Meteor’s east-west sections had been

completed. This left the ship without a science

leader. Although Wüst was the most knowledgeable,

he was considered too junior to take

over as expedition leader. Instead Captain

Spiess officially took over both as scientific

leader and naval captain. In practice, however,

it was Wüst who guided the execution of the

many measurements in physical oceanography.

He was committed to testing the scheme that he

and Merz had developed for the circulation of

the Atlantic. He was also a careful and painstaking

collector of new measurements, making

sure that no “shortcuts” were taken in collecting

or processing the measurements.

On April 16, 1925, the Meteor left Wilhelmshaven

on her way to Buenos Aires, Argentina,

which was to be the starting point of the expedition.

Outfitted with every new instrument

possible, the Meteor was the first ocean research

cruise to concentrate primarily on the physical

aspects of the ocean. She carried not one but

two new echo-sounding systems, which were

to accurately measure the depth of the ocean

beneath the ship. With no computer or even

analog storage machines it was necessary for

someone to “listen” continually to the “pings”

of the unit. Crewmen were enlisted in this operation

and two sailors had to be in the room 24

hours a day listening to pings and writing

down the travel times.

In addition the Meteor had a new system that

enabled it to anchor in the deep ocean. Because

the Meteor was able to moor itself in the deep

ocean, Ekman developed a current meter that

could be used multiple times when suspended

from the main hydrographic wire (Figure

S1.4). Ekman had gone on the pre-expedition

trip to the Azores, but did not go along on the

main cruise. His current meter was used repeatedly

during the deep-sea anchor stations.

Before returning to Germany in the spring of

1927, the Meteor made 14 sections across the

Atlantic, traveled 67,000 miles, made 9 deep-sea

anchor stations, and occupied a total of 310 hydrographic

stations. In addition over 33,000 depth

soundings had been made in an area where only

about 3000 depth soundings already existed.

During this voyage she encountered more than

one hurricane that greatly challenged her seaworthiness.

She had also suffered due to the problem

of storing sufficient coal for the crossings.

10


FIGURE S1.4 Ekman repeating current meter. Source:

From Spiess (1928).

It was indeed fortunate that Wüst was

present on the cruise to take over the scientific

leadership. He worked on later analyses of the

Meteor results with Albert Defant of the Oceanographic

Institute in Berlin (Wüst, 1935; Defant,

1936). Defant joined the Meteor for the last

section across the Atlantic.

S1.5. WORLD WAR II AND MID-

TWENTIETH CENTURY PHYSICAL

OCEANOGRAPHY

Before World War II a number of oceanographic

institutions were founded in various

parts of the world. In the United States two

very notable institutions were created. In

California, the San Diego Marine Biological

Association was founded in 1903, becoming the

Scripps Institution for Biological Research in

1912 and renamed Scripps Institution of Oceanography

(SIO) in 1925 (Shor, Day, Hardy, &

Dalton, 2003), while in Massachusetts the

Marine Biological Laboratory (MBL) located in

Woods Hole spun off the Woods Hole Oceanographic

Institution (WHOI) in January of 1930.

Both organizations became and continue to be

leading American institutions for the study of

the ocean. At WHOI Henry Bigelow was made

the first director in spite of his genuine distaste

for administrative duties. Originally WHOI

was only to be operated in the summer leaving

Bigelow the rest of the year for his scientific

research and hobbies (fishing). Bigelow was so

convinced of the importance of having a fine,

seaworthy vessel capable of making long

voyages in the stormy North Atlantic that he

dodged the efforts of many to donate old pleasure

yachts or tired fishing vessels. Instead he

agreed to spend $175,000 on the largest steelhulled

ketch in the world. A sailing ship with

a powerful auxiliary engine was chosen over

a steamship because of the inability to carry

sufficient coal for long distance cruising. The

contract was awarded to a Danish shipbuilding

company and included two laboratories, two

winches, and quarters for 6 scientists and 17

crewmembers. After delivery in the summer

of 1931 Bigelow hired his former student,

Columbus O’Donnel Iselin, as master of the

research vessel named Atlantis. Iselin later

became the director of WHOI and left a legacy

of important developments in the study of the

water masses of the ocean.

At SIO, Harald Sverdrup was hired as the

new director in 1936, bringing from the Bergen

school an emphasis on physical oceanography.

Within a year of his arrival, SIO purchased

a movie star’s pleasure yacht and converted

her into the research vessel E.W. Scripps.

Sverdrup had earlier been involved with an

international effort to sail a submarine under

the North Polar ice cap. During a test it was

discovered that the submarine, named the

WORLD WAR II AND MID-TWENTIETH CENTURY PHYSICAL OCEANOGRAPHY 11

Nautilus, had lost a diving rudder and would

not be able to cruise beneath the ice. (It was

not until 1957 that another submarine named

Nautilus cruised beneath the North polar ice

cap and surfaced in one of the larger leads in

the ice pack.)

As is usually the case, war prompted some

new developments in physical oceanography.

At WHOI, a naval Lieutenant William Pryor

came looking for an explanation as to why the

destroyer he was working on as a soundman

could not find the “target” submarine in the

afternoon after being able to do it well in the

morning. At WHOI, Bigelow and Iselin were

happy to collaborate with the navy and an

experiment was set up in the Atlantic and in

Guantanamo Bay where for two weeks two

ships “pinged” on each other. From the Atlantis,

closely spaced water bottles and thermometers

were let down into the water. As Iselin expected,

the results showed that Pryor’s assumption that

bubbles created by plankton were not the cause

of the acoustic problems; instead the vertical

temperature profile was found to alter dramatically

during the day. The change of the vertical

temperature distribution caused the sound

pulses to be refracted away from the target

location making it impossible to detect the

submarine. What was needed was a detailed

knowledge of the vertical temperature profile

in the shallow upper layers of the ocean.

Detailed studies of the generation and propagation

of ocean waves led by Harald Sverdrup

and his student Walter Munk at SIO began during

World War II, driven by the importance of forecasting

wave conditions for military operations,

including beachhead assaults (Sverdrup &

Munk, 1947; Nierenberg, 1996; Inman, 2003).

In the 1940s and 1950s, Sverdrup and Munk

at SIO were also studying the dynamics of

wind-driven currents. At WHOI, Henry Stommel

was also involved in these studies. Basic models

of the wind-driven circulation emerged from

these studies starting with Sverdrup’s model,

which explained the basic balance between the

major currents and the pressure gradients,

followed by Stommel’s model and its explanation

of the westward intensification that closed

the major ocean gyres at the western end

(Section 7.8 in the textbook). Munk’s model,

with a slightly different explanation for the

westward intensification, put it all together,

presenting a realistic circulation in response to

a simplification of the meridional wind profile.

These models were the basis for future more

complex and eventually numerical models of

the ocean circulation.

Continuations of basin-scale measurements of

temperature, salinity, and other properties from

research ships continued in the 1950s with the

International Geophysical Year (IGY). In the

1960s, the international Indian Ocean Experiment

completed the global scale observations

begun in the IGY. In the 1970s, the International

Southern Ocean Study (ISOS) concentrated on

more restricted regions and involved many

different countries.

Meanwhile, understanding of the shorter

time and space scales in the ocean began to

emerge thanks to development of reliable

moored current meters, with studies of eddies

in the 1970s beginning with a Russian experiment,

Polygon 70, which established the

importance of large-scale “synoptic” eddies in

the ocean. Considered the “weather” of the

ocean, these mesoscale features carry heat,

momentum, and other properties as they move

about the ocean. The work was definitively

expanded by the U.S. Mid Ocean Dynamics

Experiment of the early 1970s and the subsequent

joint U.S.-Russian Polymode Experiment,

which began to reveal the rich variability that

occupies much of the ocean (Munk, 2000). In

the 1970s in the North Pacific, an ambitious

program of temperature profiling from merchant

ships began to define the time and space variability

of a large swath of ocean.

There has been a dramatic shift in emphasis of

research in physical oceanography near the end of

the twentieth century. A global survey of ocean

12


circulation (WOCE), whose main purpose was to

assist through careful observations; the development

of numerical ocean circulation models

used for climate modeling; and an intensive

ocean-atmosphere study of processes governing

El Niño in the tropical Pacific (Tropical Ocean

Global Atmosphere; TOGA) were completed.

Many of the programs that have continued

beyond these studies focus on the relationship

between ocean physics and the climate. At the

same time the practical importance of ocean

physics in the coastal ocean is emerging. The

need for military operations in the ocean has

shifted to the coasts largely in support of other

land operations. Oil operations are primarily

restricted to the shallow water of the coastal

regions where tension with the local environment

requires even greater study of the coastal ocean.

The most dramatic shifts in physical oceanographic

methods at the turn of the twenty-first

century are to extensive remote sensing, in the

form of both satellite and more automated

in situ observations, and to ever-growing reliance

on complex computer models. Satellites

measuring sea-surface height, surface temperature,

and most of the components of forcing for

the oceans are now in place. Broad observational

networks measuring tides and sea level and

upper ocean temperatures in the mid-to-late

twentieth century have been greatly expanded.

These networks now include continuous current

and temperature monitoring in regions where

the ocean’s conditions strongly affect climate,

such as the tropical Pacific and Atlantic, and

growing monitoring of coastal regions. Global

arrays of drifters measuring surface currents

and temperature, and subsurface floats

measuring deeper currents and ocean properties

between the surface and about 2000 m depth are

now expanding. Meanwhile the enormous

growth in available computational power and

numbers of scientists engaged in ocean modeling

is expanding our modeling capability and ability

to simulate ocean conditions and study particular

ocean processes. With increasing amounts of

globally distributed data available in near real

time, numerical ocean modelers are now beginning

to combine data and models to improve

ocean analysis and possibly prediction of ocean

circulation changes in a development similar to

that for numerical weather prediction in the

twentieth century. Full climate modeling includes

ocean modeling, and many oceanographers are

beginning to focus on the ocean component of

climate modeling. These trends are likely to

continue for some time.

S1.6. A BRIEF HISTORY OF

NUMERICAL MODELING IN

PHYSICAL OCEANOGRAPHY

Numerical modeling is a major component

of contemporary ocean science, along with theory

and observation. Models are quantitative expressions

of our understanding of the ocean and its

interactions with the atmosphere, solid earth,

and biosphere. They provide a virtual laboratory

that allows us to test hypotheses about particular

processes, predict future changes in the ocean,

and to estimate the response of the ocean to perturbations

in external conditions. The complexity and

nonlinearity of the physical laws governing the

system preclude solution by analytical methods

in all but the most idealized models. The most

comprehensive models, known as ocean general

circulation models, are solved by numerical

methods, often on the most powerful computers

available. Blending of models and observations

to provide comprehensive descriptions of the

actual state of the ocean, through a process of

data assimilation similar to that used in numerical

weather forecasting, has become a reality in the

past decade, due to advances in observing

systems, increases in computer power, and dedication

of scientific effort.

The growth and evolution of ocean modeling

is paced, to a certain degree, by the growth in

computing power over time. The computational

cost of a model is determined by its resolution,

A BRIEF HISTORY OF NUMERICAL MODELING IN PHYSICAL OCEANOGRAPHY 13

that is, the range of scales represented; the size of

the domain (basin or global, upper ocean or full

depth); and the comprehensiveness and

complexity of the processes, both resolved and

parameterized, that are to be represented. An

ocean model is typically first formulated in terms

of the differential equations of fluid mechanics,

often applying approximations that eliminate

processes that are of no interest to the study at

hand. For example, in the study of large-scale

ocean dynamics, sound wave propagation

through the ocean is not of great importance, so

seawater is approximated as an incompressible

fluid filtering sound waves out of the equations.

The continuous differential equations must

then be discretized, that is, approximated by

a finite set of algebraic equations that can be

solved on a computer. In ocean models this step

is most often done with finite-difference or

finite-volume methods, although finite-element

methods have also been employed. In addition

to the choice of numerical method, a major point

of diversity among ocean general circulation

models is the choice of vertical coordinate. In

the upper ocean, where vertical mixing is strong,

a discretization based on surfaces of constant geopotential

or depth is the most natural. In the ocean

interior, where transport and mixing occur

primarilyalongneutraldensitysurfaces,a vertical

discretization based on layers of constant density,

or isopycnal coordinates, is the most natural.

Near the ocean bottom, a terrain-following coordinate

provides a natural and accurate framework

for representing topography and applying

the boundary conditions for the flow.

The earliest three-dimensional ocean general

circulation models, originally developed in the

1960s by Kirk Bryan and colleagues at the

NOAA Geophysical Fluid Dynamics Laboratory,

were based on finite-difference methods using

depth as the vertical coordinate (Bryan & Cox,

1968; Bryan, 1969). Models descended from this

formulation still comprise the most widely used

class of ocean general circulation models, particularly

in the climate system modeling

community. The first global ocean simulations

carried out with this type of model were limited

by the then available computational resources to

resolutions of several hundred kilometers, insufficient

to represent the hydrodynamic instability

processes responsible for generating mesoscale

eddies.

In the 1970s observational technology

emerged that showed the predominance of

mesoscale eddies in the ocean. A new class of

numerical models with simplifications to the

physics, such as using the quasi-geostrophic

rather than the primitive equations and limited

domain sizes with resolutions of a few tens of

kilometers, was developed by Bill Holland, Jim

McWilliams, and colleagues at the National

Center for Atmospheric Research (NCAR).

Models of this class have contributed greatly to

the development of our understanding of the

interaction of mesoscale eddies and the largescale

ocean circulation, and to the development

of parameterizations of eddy-mixing processes

for use in coarser resolution models, such as

those used in climate simulations. Initially

developed as a generalization to the quasigeostrophic

eddy-resolving models, isopycnal

coordinate models such as that developed by

Bleck and co-workers at the University of Miami

(Bleck & Boudra, 1981) became increasingly

popular for ocean simulation through the 1980s

and 1990s. Today global eddy-resolving models

have spatial resolution of less than 10 km, with

regional models achieving much higher spatial

resolution. A recent overview of progress was

published in Hecht and Hasumi (2008) by

many of the principal groups.

Terrain-following coordinate models, also

known as “sigma coordinate” models initially

developed primarily in the coastal ocean

modeling community by Mellor and co-workers

at Princeton University, were used in basin- to

global-scale ocean studies throughout the

1980s and 1990s. A model of this type widely

used at present in regional studies is the

Regional Ocean Modeling System (ROMS).

14


Ocean general circulation models are important

in coupled climate modeling, although

they must be run in much coarser spatial configurations

than the eddy-resolving versions to

attain the many decades of integration required.

Many of the major international modeling

groups have participated in the Intergovernmental

Panel on Climate Change assessments,

which included more than 20 coupled models

in its summaries (Meehl et al., 2007).

In the twenty-first century we are witnessing

both a tighter integration of modeling with

observational oceanography, for example,

through the use of data assimilation techniques,

and significant merging and cross-fertilization

of the various approaches to ocean modeling

described earlier. Computer power has reached

a level where the ocean components of fully

coupled climate system models have sufficient

resolution to permit mesoscale eddies, blurring

the distinction between ocean models used for

climate applications and those used to study

mesoscale processes. Several new models have

emerged with hybrid vertical coordinates,

bringing the best features of depth, isopycnal,

and terrain-following coordinates into a single

model framework.

References

Bleck, R., Boudra, D.B., 1981. Initial testing of a numerical

ocean circulation model using a hybrid quasi-isopycnal

vertical coordinate. J. Phys. Oceanogr 11, 755e770.

Bryan, K., 1969. A numerical method for the study of the

circulation of the world ocean. J. Comp. Phys. 4, 347e376.

Bryan, K., Cox, M.D., 1968. A nonlinear model of an ocean

driven by wind and differential heating. Part 1. J. Atmos.

Sci. 25, 945e978.

Chapman, B., 2004. Initial visions of paradise: Antebellum

U.S. government documents on the South Pacific. J. Gov.

Inform. 30, 727e750.

Defant, A., 1936. Die Troposphäre des Atlantischen Ozeans. In

Wissenschaftliche Ergebnisse der Deutschen Atlantischen


"Meteor" 1925e1927 6 (1), 289e411 (in German).

Ekman, V.W., 1905. On the influence of the Earth’s rotation on

ocean currents. Arch. Math. Astron. Phys. 2 (11), 1e53.

Hecht, M.W., Hasumi, H. (Eds.), 2008. Ocean Modeling in

an Eddying Regime. AGU Geophysical Monograph

Series, 177, 350 pp.

Inman, D.L., 2003. Scripps in the 1940s: the Sverdrup era.


Knudsen, M. (Ed.), 1901. Hydrographical Tables. G.E.C.

Goad, Copenhagen, 63 pp.

Laplace, P.S., 1790. Mémoire sur le flux et reflux de la mer.

Mém. Acad. Sci Paris, 45e181 (in French).

Maury, M.F., 1855. The Physical Geography of the Sea.

Harper and Brothers, New York, 304 pp.

Meehl, G.A., Stocker, T.F., Collins, W.D., Friedlingstein, P.,

Gaye, A.T., Gregory, J.M., et al., 2007. Global climate

projections. In: Solomon, S., Qin, D., Manning, M.,

Chen, Z., Marquis, M., Averyt, K.B., Tignor, M.,

Miller, H.L. (Eds.), Climate Change 2007: The Physical

Science Basis. Contribution of Working Group I to the

Fourth Assessment Report of the Intergovernmental

Panel on Climate Change. Cambridge University Press,

Cambridge, UK.


3 Beitrag. Zeitschr. D.G.F.E, Berlin (in German).

Munk, W.H., 2000. Achievements in physical oceanography.

In 50 Years of Ocean Discovery: National Science

Foundation 1950e2000. National Academy Press,


Nierenberg, W.A., 1996. Harald Ulrik Sverdrup 1888e1957.

Biographical Memoirs. National Academies Press,

Washington, D.C. 69, 339e375.

Peterson, R.G., Stramma, L., Kortum, G., 1996. Early

concepts and charts of ocean circulation. Progr. Oceanogr

37, 1e115.

Richardson, P.L., 2008. On the history of meridional overturning

circulation schematic diagrams. Progr. Oceanogr

76, 466e486.

Shor, E., Day, D., Hardy, K., Dalton, D., 2003. Scripps time

line. Oceanography 16, 109e119.

SIO, 2011. Scripps Institution of Oceanography Archives.

UC San Diego. http://libraries.ucsd.edu/locations/sio/

scripps-archives/index.html (accessed 3.25.11).

Spiess, F., 1928. Die Meteor Fahrt: Forschungen und

Erlebnisse der Deutschen Atlantischen Expedition,

1925e1927. Verlag von Dietrich Reimer, Berlin, 376 pp.

(in German d English translation Emery, W.J., Amerind

Publishing Co. Pvt. Ltd., New Delhi, 1985).

Sverdrup, H.U., Munk, W.H., 1947. Wind, Sea, and Swell:

Theory of Relations for Forecasting. U.S. Navy Dept.,

Hydrographic Office, H.O. Pub. No. 601, 44 pp.





1925e1927, 6 1st Part, 2, 109e288 (in German).

C H A P T E R

S4

Typical Distributions of Water

Characteristics: Supplementary Materials

FIGURE S4.1 Satellite infrared sea-surface temperature ( C; nighttime only), averaged to 50 km and 1 week, for (a) July 3,

2008 (austral winter) and (b) January 3, 2008 (also Figure 4.1b in the textbook, where it appears in gray scale only). White is

sea ice. Source: From NOAA NESDIS, (2009b).

1

2

S4. TYPICAL DISTRIBUTIONS OF WATER CHARACTERISTICS: SUPPLEMENTARY MATERIALS

FIGURE S4.1

(Continued).

TYPICAL DISTRIBUTIONS OF WATER CHARACTERISTICS: SUPPLEMENTARY MATERIALS 3

FIGURE S4.2 (a) Chlorophyll (mg m 3 ), (b) particulate organic carbon (POC; mg m 3 ), derived from SeaWiFS ocean

color data, averaged May-August, 1997e2002, and (c) chlorophyll as % of POC. Source: From Gardner, Mishov, and Richardson,

(2006).

4

S4. TYPICAL DISTRIBUTIONS OF WATER CHARACTERISTICS: SUPPLEMENTARY MATERIALS

FIGURE S4.3

NASA (2009b).

Photosynthetically available radiation (PAR; Einsteins m 2 day 1 ) from the SeaWiFS satellite. Source: From

References

Gardner, W. D., Mishonov, A. V., & Richardson, M. J. 2006.

Global POC concentrations from in-situ and satellite

data. Deep-Sea Res, II, 53, 718e740.

NASA, 2009b. Ocean Color Web. NASA Goddard Space

Flight Center. http://oceancolor.gsfc.nasa.gov/(accessed

2.18.09).

NOAA NESDIS, 2009b. Ocean Products Page, NOAA/

NESDIS/OSDPD. http://www.osdpd.noaa.gov/PSB/

EPS/SST/SST.html (accessed 2.18.09).

C H A P T E R

S5

Mass, Salt, and Heat Budgets and Wind

Forcing: Supplementary Materials

FIGURE S5.1 Mean (1983e2004) shortwave radiation (W/m 2 ) from the International Satellite Cloud Climatology Project

(ISCCP). (a) annual, (b) January, (c) July, and (d) July 1992 monthly mean. Source: From ISCCP (2007).

1

2

S5. MASS, SALT, AND HEAT BUDGETS AND WIND FORCING: SUPPLEMENTARY MATERIALS

FIGURE S5.2 Monthly mean shortwave radiation (W/m 2 ) for (a) January, (b) April, (c) July, and (d) October. Data are

from the NOCS product of Grist and Josey (2003).

S5. MASS, SALT, AND HEAT BUDGETS AND WIND FORCING: SUPPLEMENTARY MATERIALS 3

FIGURE S5.3 Cloud cover (%) for (a) January, (b) April, (c) July and (d) October. Data are from the climatology of

da Silva, Young, and Levitus (1994), based on surface observations.

4


FIGURE S5.4 Monthly mean longwave heat flux (W/m 2 ) for (a) January, (b) April, (c) July, and (d) October. Data are from

the NOCS product of Grist and Josey (2003).


FIGURE S5.5 Monthly mean latent heat flux (W/m 2 ) for (a) January, (b) April, (c) July, and (d) October. Data are from the

NOCS product of Grist and Josey (2003).

6


FIGURE S5.6 Monthly mean sensible heat flux (W/m 2 ) for (a) January, (b) April, (c) July, and (d) October. Data are from

the NOCS product of Grist and Josey (2003).


FIGURE S5.7 Monthly mean net heat flux (W/m 2 ) for (a) January, (b) April, (c) July, and (d) October. Data are from the

NOCS product of Grist and Josey (2003).

8


FIGURE S5.8 Annual mean airesea (a) buoyancy flux, (b) heat flux, and (c) freshwater flux (precipitation, evaporation,

and runoff) with the buoyancy and freshwater fluxes converted to equivalent heat fluxes (W/m 2 ), based on Large and

Yeager (2009) airesea fluxes. Positive values (yellows-reds) indicate that the ocean is becoming less dense, warmer, or

fresher in the respective maps. Contour interval is 25 W/m 2 ; in (c) dotted contours are 10 and 20 W/m 2 .


FIGURE S5.9 Annual mean meridional transports of (a) heat (PW) and (b) freshwater (Sv). Uncertainties in the global

estimates are given in green. The symbols with error bars show direct transport estimates, from Bryden and Imawaki (2001).

Source: From Large and Yeager (2009).

10


FIGURE S5.10 Annual mean wind stress (N/m 2 ) (vectors) and wind-stress curl ( 10 7 N/m 3 ; color shading), multiplied

by -1 in the Southern Hemisphere. (a) Pacific Ocean, (b) Atlantic Ocean, and (c) Indian Ocean. Data are from the NCEP

reanalysis 1968e1996 (Kalnay et al., 1996).


FIGURE S5.10

(Continued).

References

Bryden, H.L., Imawaki, S., 2001. Ocean heat transport. In

G. Siedler, & J. Church (Eds.), Ocean Circulation and

Climate, International Geophysics Series (pp. 455e474).

Academic Press.

da Silva, A.M., Young, A.C., Levitus, S., 1994. Atlas of

surface marine data, Vol. 1. NOAA Atlas of surface

marine data.

Grist, J.P., Josey, S.A., 2003. Inverse analysis adjustment of

the SOC air-sea flux climatology using ocean heat

transport constraints. J. Clim., 20, 3274e3295.


available online. NASA Goddard Institute for Space



Kalnay, E., Kanamitsu, M., Kistler, R., Collins, W.,

Deaven, D., Gandin, L., et al., 1996. The NCEP-NCAR

40-year reanalysis project. Bull. Am. Meteorol. Soc. 77,

437e471.


interannually varying air-sea flux data set. Clim. Dyn.,

33, 341e364.

C H A P T E R

S6

Data Analysis Concepts and

Observational Methods: Supplementary

Materials

FIGURE S6.1 Objective mapping of density and acoustic Doppler current profiler velocity data. Azores Front: potential

density and geostrophic velocity at 68 m in February, 1992. Source: From Rudnick (1996).

1

2

S6. DATA ANALYSIS CONCEPTS AND OBSERVATIONAL METHODS: SUPPLEMENTARY MATERIALS

References



J. Phys. Oceanogr, 6, 249e266.

Rudnick, D.L., 1996. Intensive surveys of the Azores Front 2.

Inferring the geostrophic and vertical velocity fields.

J. Geophys. Res. 101, 16291e16303.

Stammer, D., Wunsch, C., 1999. Temporal changes in eddy

energy of the oceans. Deep-Sea Res. 46, 77e108.

FIGURE S6.2 Spectral aliasing. Frequency spectral

density of satellite altimetric sea-surface height. The small

sharp peak at 60 days is an alias of the semi-diurnal tide.

Source: From Stammer and Wunsch (1999).

FIGURE S6.3 The six principal empirical orthogonal functions describing the sea level pressure anomalies in the North

Pacific. ÓAmerican Meteorological Society. Reprinted with permission. Source: From Davis (1976).

C H A P T E R

S7

Dynamical Processes for Descriptive

Ocean Circulation

This is the complete chapter concerning

dynamical processes; a truncated version

appears in the print text. Many additional figures

are included here, along with expanded descriptions

and derivations. Tables for Chapter 7

appear only on this Web site.

Water in the ocean at all time and space scales

is subject to the same small set of forces and

accelerations. What distinguishes one type of

motion from another, for instance a surface

wave from the Gulf Stream, is the relative

importance of the different accelerations and

forces within this small set. In this chapter we

introduce a basic dynamical framework for the

major circulation and water mass structures

described in ensuing chapters. We use as little

mathematics as possible, relying principally on

word descriptions of the physical processes.

Students are directed to dynamical oceanography

textbooks for complete coverage of these

topics, including scale analysis and derivations,

such as Gill (1982), Pedlosky (1987), Cushman-

Roisin (1994), Knauss (1997), Salmon (1998),

Vallis (2006), and Huang (2010).

We proceed from the basic equations of

motion (Sections 7.1 and 7.2) and density

evolution (Section 7.3) to mixing layers (Section

7.4); direct wind response including Ekman

layers (Section 7.5); geostrophic flow (Section

7.6); vorticity, potential vorticity, and Rossby

waves (Section 7.7); wind-driven circulation

models of the gyre circulations (Section 7.8);

equatorial and eastern boundary circulations

(Section 7.9); and finally thermohaline forcing,

abyssal circulation, and overturning circulation

(Section 7.10).

7.1. INTRODUCTION:

MECHANISMS

Ultimately, motion of water in the ocean is

driven by the sun, the moon, or tectonic

processes. The sun’s energy is transferred to

the ocean through buoyancy fluxes (heat fluxes

and water vapor fluxes) and through the winds.

Tides create internal waves that break, creating

turbulence and mixing. Earthquakes and

turbidity currents create random, irregular

waves including tsunamis. Geothermal

processes heat the water very gradually with

little effect on circulation.

Earth’s rotation profoundly affects almost all

phenomena described in this text. Rotating

fluids behave differently from non-rotating

fluids in ways that might be counterintuitive.

In a non-rotating fluid, a pressure difference

between two points in the fluid drives the fluid

1

2

S7. DYNAMICAL PROCESSES FOR DESCRIPTIVE OCEAN CIRCULATION

toward the low pressure. In a fluid dominated

by rotation, the flow can be geostrophic, perpendicular

to the pressure gradient force, circling

around centers of high or low pressure due to

the Coriolis effect.

Ocean circulation is often divided conceptually

into two parts, the wind-driven and the thermohaline

(or buoyancy-dominated) components.

Wind blowing on the ocean initially causes

small capillary waves and then a spectrum of

waves and swell in the ocean (Chapter 8).

Impulsive changes in wind lead to short timescale

inertial currents and Langmuir cells.

Steady or much more slowly changing wind

(in speed and direction) creates the ocean’s

near-surface frictional Ekman layer, which

involves the Coriolis effect. As the wind

momentum transfer persists, the geostrophic,

wind-driven circulation results.

Thermohaline circulation is associated with

heating and cooling (“thermo”), and evaporation,

precipitation, runoff, and sea ice formation

d all of which change salinity (“haline”).

Thermohaline-dominated circulation is mostly

weak and slow compared with wind-driven

circulation. Thermohaline forcing ranges from

very local to very broad scale. An example of

local forcing is the deep overturn driven by cooling

and/or evaporation in which the horizontal

scale of convection is at most a few kilometers.

Broad-scale buoyancy forcing is associated

with vertical diffusion that acts on the large-scale

temperature and salinity structure. Vertical

diffusion is very weak in the interior of the

ocean, but is essential for maintaining the

ocean’s vertical stratification. In discussing thermohaline

effects, it is common to refer to the

meridional overturning circulation (MOC)

(Section 14.2.3). Overturning does not have to

be meridional to be of interest, and it is generally

useful to simply refer to overturning circulation.

The energy source for thermohaline circulation

importantly includes the wind and tides that

produce the turbulence essential for the diffusive

upwelling across isopycnals that closes the

thermohaline overturning. Both the winddriven

and thermohaline circulations are almost

completely in geostrophic balance, with the

forcing that drives them occurring at higher

order.

7.2. MOMENTUM BALANCE

Fluid flow in three dimensions is governed by

three equations expressing how velocity (or

momentum) changes, one for each of the three

physical dimensions. Each of the three

momentum equations includes an acceleration

term (how velocity changes with time), an

advection term (see Section 5.1.3), and forcing

terms. These are the same Newton’s Laws taught

in physics. Since a fluid is continuous, the mass

of a single object is replaced by the mass per

unit volume (density); forces are also expressed

per unit volume. In “word” equations:

Density ðAcceleration þ AdvectionÞ

¼ Forces per unit volume (7.1)

Forces per unit volume

¼ Pressure gradient force þ Gravity

þ Friction (7.2)

Expressions (7.1) and (7.2) are each three equations,

one for each of the three directions (e.g.,

east, north, and up). The terms in Eqs. (7.1)

and (7.2) are illustrated in Figure S7.1. For

ocean dynamics, these equations are usually

written in Cartesian coordinates (x, y, z), where

x and y are westeeast and southenorth, and z

is upward. Atmospheric dynamicists and some

ocean modelers use spherical coordinates

instead (longitude, latitude, and the local

vertical).

The inclusion of advection means that

Eq. (7.1) is the expression of momentum change

in a Eulerian framework, where the observer

sits at a fixed location relative to Earth.

MOMENTUM BALANCE 3

(a)

Acceleration

(b)

Advection V T

x

x 2 x 3 x 4

t 1 t 2 t 3

x 1 x 2 x 3

v 1

a

v 2

Time

Position

Velocity

Acceleration

x 1

Time t 1

T = 2° 3° 4° 5°

Time t 2

2° 3° 4° 5°

(c)

Pressure gradient force

(d)

Gravitational force – g

High

pressure

pressure gradient

Low

pressure

– g

x A

P A

dp P B – P

=~ A

dx x B – x A

x B

P B

pressure gradient

force

(e)

Acceleration associated with friction and viscosity

z




fluid

velocity u(z)

x-momentum flux

= u/ z


time: just after top plate starts

High flux divergence

High acceleration

x


time: later

Lower flux divergence

Lower acceleration


time: -->

No flux divergence

No acceleration

FIGURE S7.1 Forces and accelerations in a fluid: (a) acceleration, (b) advection, (c) pressure gradient force, (d) gravity,

and (e) acceleration associated with viscosity y.

Equation (7.1) can be written without the

advection term, in a Lagrangian framework,

where the observer drifts along with the fluid

flow. (See also Section S16.5 in the online

supplement.)

For a rotating geophysical flow, we, as

observers, sit within a rotating “frame of

reference” attached to the rotating Earth.

For this reference frame, the acceleration

term on the left-hand side of Eq. (7.1) is

4


rewritten to separate local acceleration due

to an actual local force from the effects of

rotation. The effects that are separated out

are the centrifugal and Coriolis accelerations

(Section 7.2.3).

The “pressure gradient” force in Eq. (7.2)

arises from external forcing. The frictional force

in Eq. (7.1) leads to dissipation of energy due to

the fluid’s viscosity.

7.2.1. Acceleration and Advection

Acceleration is the change in velocity with

time. If the vector velocity is expressed in Cartesian

coordinates as u ¼ (u, v, w) where the bold

u indicates a vector quantity, and u, v, and w are

the positive eastward (x-direction), northward

(y-direction), and positive upward (z-direction)

velocities, then

x-direction acceleration ¼ vu=vt

(7.3a)

with similar expressions for the y- and z-directions.

In a rotating frame of reference, such as

on the surface of Earth, the acceleration term

includes two additional terms: centrifugal and

Coriolis acceleration (Section 7.2.3).

Advection is defined in Section 5.1.3. Advection

is how the flow moves properties

(including scalars such as temperature or

salinity) and vectors (such as the velocity).

Advection can change the flow property if there

is a gradient in the property through which the

fluid moves. The advection term thus is

a product of the velocity and the difference in

the property from one location to another. There

are three advection terms in each momentum

equation, since the flow bringing in a different

property can come from any of the three directions.

(The vertical advection term is sometimes

called convection.) In the x-momentum equation,

the advection term is

x-direction advection

¼ uvu=vx þ vvu=vy þ wvu=vz

(7.3b)

The substantial derivative is the sum of the

acceleration and advection terms:

Du=Dt ¼ vu=vt þ uvu=vx

þ vvu=vy þ wvu=vz (7.4)

Eq. (7.4) represents the change in u at a fixed

point (Eulerian framework). In a Lagrangian

framework, following the particle of water,

only the time derivative appears; the three

advection terms do not appear since they are

contained in the movement of the particle.


Gravitational Force

Pressure is defined in Section 3.2. The flow of

fluid due to spatial variations in pressure (the

pressure gradient force) is also described. In mathematical

form, the pressure gradient force is

x-direction pressure gradient force

¼

vp=vx

(7.5)

The pressure gradient force has a negative sign

because the force goes from high pressure to

low pressure.

The gravitational force between Earth and

the object or fluid parcel is directed toward

the center of mass of Earth. Gravitational

force is mass of the object gravitational

acceleration g, equal to 9.780318 m 2 /sec (at

theequator).Thegravitationalforceperunit

volume is

z-direction gravitational force per unit volume

¼ rg

(7.6)

7.2.3. Rotation: Centrifugal and

Coriolis Forces

Earth rotates at a rate of U ¼ 2 p/T where T is

the length of the (sidereal) day, which has 86,164

MOMENTUM BALANCE 5

seconds; hence U ¼ 0.729 10 4 sec 1 . 1 We look

at motions and write our theories sitting in

a “rotating reference frame,” that is, attached to

the rotating Earth. However, the reference frame

that is correct for Newton’s Laws (Eq. 7.1) isan

“inertial reference frame,” which is not rotating.

To look at motions from within our rotating

reference frame, we must add two terms due to

the Earth’s rotation. The first is the “Coriolis

force” and the second is the “centrifugal force”

(Figure S7.2). A derivation of these two

pseudo-forces is given at the end of this section.

7.2.3.1. Centrifugal and Centripetal Force

Centrifugal force is the apparent outward force

on a mass when it is rotated. Think of a ball on the

end of a string that is being twirled around, or

the outward motion you feel when turning

a curve in a car. In an inertial frame, there is no

outward acceleration since the system is not

rotating. The ball or your body just moves in

the straight line that they were following originally.

But in the rotating reference frame of the

string or the car, they appear to be accelerated

away. Since Earth rotates around a fixed axis,

the direction of centrifugal force is always

outward away from the axis. Thus it is opposite

to the direction of gravity at the equator; at

Earth’s poles it is zero. (Centripetal force is the

necessary inward force that keeps the mass

from moving in a straight line; it is the same

size as centrifugal force, with the opposite sign.

Centripetal force is real; centrifugal force is just

an apparent force. For the rotating Earth, centripetal

force is supplied by the gravitational force

towards Earth’s center.)

If Earth was a perfect, rigid sphere, the ocean

would be 20 km deeper at the equator than at

the poles. But this is not observed, because the

(a)

(b)

150°

Coriolis

"deflection"

180°

centripetal

force

actual path

actual final

location

actual path

150°

string

intended

path

intended

target

intended

path

solid Earth is deformed by centrifugal force.

That is, Earth is a spheroid rather than a sphere,

with the radius at the equator approximately

20 km greater than at the poles. Therefore the

60°

120°

60°

centrifugal

force

Earth

rotation

90°

ball

FIGURE S7.2 (a) Centrifugal and centripetal forces and

(b) Coriolis force.

30°

30°

V

0°

1 The solar day, which is the time between consecutive highest points of the sun in the sky, is 24 hours, or 86,400 seconds. The

sidereal day is the rotation period relative to the fixed stars, which is the inertial reference frame. The sidereal day is slightly

shorter than the solar day, with 23 hours, 56 minutes, and 4.1 seconds. One pendulum day is one sidereal day/sin4, where

a sidereal day is the time it takes for Earth to rotate 360 degrees and where 4 ¼ latitude. For 4 ¼ 10 ,45 ,60 , 1 pendulum

day ¼ 5.7, 1.4, 1.2 sidereal days.

6


centrifugal force at the equator is balanced

(canceled) by the extra gravitational force there

(this is referred to as “effective gravity”).

The mathematical expression for centrifugal

acceleration (force divided by density) is

centrifugal acceleration ¼ U 2 r (7.7)

where U is the rotation rate of Earth, equal to

2p/T where T is the length of day, and r is

Earth’s radius. Because the centrifugal acceleration

is nearly constant in time and points

outward, away from Earth’s axis of rotation,

we usually combine it formally with the gravitational

force, which points toward Earth’s center.

We replace g in Eq. (7.6) with the effective

gravity g, which has a weak dependence on latitude.

Hereafter, we do not refer separately to the

centrifugal force. The surface perpendicular to

this combined force is called the geoid. If the

ocean were not moving relative to Earth, its

surface would align with the geoid.

7.2.3.2. Coriolis Force

The second term in a rotating frame of reference

included in the acceleration equation (7.1)

is the Coriolis force. When a water parcel, air

parcel, bullet, hockey puck, or any other body

that has little friction moves, Earth spins out

from under it. By Newton’s Law, the body

moves in a straight line if there is no other force

acting on it. As observers attached to Earth, we

see the body appear to move relative to our location.

In the Northern Hemisphere, the Coriolis

force causes a moving body to appear to move

to the right of its direction of motion (Figure

S7.2b). In the Southern Hemisphere, it moves

to the left.

The Coriolis force is non-zero only if the body

is in motion, and is important only if the body

travels for a significant period of time. Coriolis

force is larger for larger velocities as well. For

the flight of a bullet there is no need to consider

the Coriolis force because the travel time is

extremely short. For missiles that fly long paths

at high speeds, Coriolis force causes significant

deflections. For winds in the atmosphere’s Jet

Stream, the timescale of motion is several days

to several weeks, so Earth’s rotation is very

important and the winds do not blow from

high to low pressure. The same holds true in

the ocean, where currents last for weeks or years

and are strongly influenced by the Coriolis

force.

For large-scale ocean currents, and to some

extent winds, the vertical velocity is much

weaker than the horizontal velocity. Certainly

the distance that a water parcel can move in

the vertical is much more limited than in the

horizontal, because of both the difference in

depth and width of the ocean, and because of

the ocean’s stratification. Therefore, Coriolis

effects act mostly on the horizontal velocities

and not on the vertical ones. As noted previously,

the Coriolis force apparently sends

objects to the right in the Northern Hemisphere

and to the left in the Southern Hemisphere. At

the equator, the Coriolis effect acting on horizontal

velocities is zero. Its magnitude is largest

at the poles.

Mathematically, the Coriolis force is

x-momentum equation:

2U sin 4 vh fv (7.8a)

y-momentum equation:

2U sin 4 uh fu

Coriolis parameter:

f ¼ 2U sin 4

(7.8b)

(7.8c)

where “h” denotes a definition, U is the rotation

rate, 4 is latitude, u is velocity in the x-direction,

v is velocity in the y-direction, and where the

signs are appropriate for including these terms

on the left-hand side of Eq. (7.1). The Coriolis

parameter, f, is a function of latitude and changes

sign at the equator, and it has units of sec 1 . (The

non-dimensional parameter called the Rossby

number introduced in Section 1.2 is Ro ¼ 1/fT

MOMENTUM BALANCE 7

or Ro ¼ U/fL, where U, L, and T are characteristic

velocity, length, and timescales for the flow.)

7.2.3.3. Derivation of Centrifugal and

Coriolis Terms

The Coriolis and centrifugal terms are

derived by transforming Newton’s law of

motion (Eq. 7.1) from its true inertial system,

relative to the fixed stars, to the rotating Earthcentric

system. This derivation is available in

advanced textbooks on classical mechanics,

and is included here for completeness. Equations

are numbered separately to maintain

consistent numbering because they are not

included in the print text. We write the threedimensional

vector version of Eq. (7.1) as

v ! v s

¼ ! F =r

(S7.1)

vt

where the subscript “s” means that the velocity

of the particle is measured in the inertial frame

of reference relative to the stars. Rewrite this

velocity as the sum of the particle’s velocity relative

to Earth’s surface and the velocity of Earth’s

surface due to rotation:

!

v s ¼ ! v e þ / U ! r (S7.2)

where ! v e is the particle velocity relative to local

coordinates on Earth’s surface, /

U is Earth’s

rotation vector, pointing northward along the

axis of rotation with magnitude equal to the

rotation rate, and ! r is the vector position of the

particle. Substituting this back into Eq. (S7.1)

yields

v

! v s v

! v e

¼ þ v vt s vt s vt ðU/ ! r Þ s

v

! v e

¼

vt sþ vU/ !

r þ / v

! r

U

vt

vt s

(S7.3)

Since Earth’s rotation is essentially constant

compared with the timescales of atmospheric

and oceanic circulation, and using Eq. (S7.2),

we find that

v

! v s

vt s

v

! v e

¼

vt

v

! v e

¼

vt

þ U /

þ / U

s

v

! r

vt

þ / U ! v e þ / U

e

vt e

/

U

! r

(S7.4)

s

v

! r

Since the derivative of any vector in the fixed

frame is related to the derivative in the rotating

frame as

v

! q v

! q

vt

¼

s

we find finally that

v

! v s

vt

¼

s

vt

v v

! e

vt

þ / U ! q

e

þ 2U / ! v e

e

(S7.5)

þ / U / U ! r (S7.6)

The first term on the right-hand side is the acceleration

relative to the rotating (Earth) frame

of reference, the second term is the Coriolis

term, and the third term is the centrifugal

acceleration.

7.2.4. Viscous Force or Dissipation

Fluids have viscous molecular processes that

smooth out variations in velocity and slow

down the overall flow. These molecular

processes are very weak, so fluids can often be

treated, theoretically, as “inviscid” rather than

viscous. However, it is observed that turbulent

fluids like the ocean and atmosphere actually

act as if the effective viscosity were much larger

than the molecular viscosity. Eddy viscosity is

introduced to account for this more efficient

mixing (Section 7.2.4.2).

7.2.4.1. Molecular Viscosity

We can think of molecular viscosity by considering

two very different types of coexisting

8


FLOW U

L‘

FIGURE S7.3 Illustration of molecular processes that

create viscosity. The mean flow velocity is indicated in gray

(U). L 0 is the distance between molecules. U 0 is the speed of

the molecules. Random molecule motions carry information

about large-scale flow to other regions, thus creating

(viscous) stresses. Viscous stress depends on the mean

molecular speed jU 0 j and mean molecular free path jL 0 j.

motion: the flow field of the fluid, and, due to

their thermal energy, the random motion of molecules

within the flow field. The random molecular

motion carries (or advects) the larger scale

velocity from one location to another, and then

collisions with other molecules transfer their

momentum to each other; this smoothes out the

larger scale velocity structure (Figure S7.3).

The viscous stress within a Newtonian fluid

is proportional to the velocity shear. The proportionality

constant is the dynamic viscosity, which

has meter-kilogram-second (mks) units of

kg/m-sec. The dynamic viscosity is the product

of fluid density times a quantity called the kinematic

viscosity, which has mks units of m 2 /sec.

For water, the kinematic viscosity is

1.8 10 6 m 2 /sec at 0 C and 1.0 10 6 m 2 /sec

at 20 C(Table S7.1).

Flow is accelerated or decelerated if there is

a variation in viscous stress from one location

U‘

to another. This is illustrated in Figure S7.1e,

where a viscous stress is produced by the

motion of a plate at the top of the fluid, with

a stationary plate at the bottom. The fluid

must stay with each plate, so the fluid velocity

at each boundary equals that plate velocity.

1. At very small times (leftmost panel), just after

the top plate starts to move, there is a large

variation in velocity in the fluid close to the

top plate, which means there is a large stress

there. The stress is associated with flux of

x-momentum down into the fluid from the

plate. Since there is much smaller stress

farther down in the fluid, there is a net

deposit of x-momentum in the fluid, which

accelerates it to the right.

2. At a later time (center panel), this acceleration

has produced velocity throughout the fluid

and the change in viscous stress from top to

bottom is reduced.

3. At a very large time (rightmost panel), the

viscous stress is the same at all locations

and there is no longer any acceleration; at

this time the velocity varies linearly from

top to bottom. (This is known as “Couette

flow.”) There is a stress on the fluid as

a whole, which is balanced by the frictional

stress of the fluid back on the plates; there

is dissipation of energy throughout the

fluid even though there is no local

acceleration.

Formally, for a Newtonian fluid, which is

defined to be a fluid in which stress is proportional

to strain (velocity shear), and if viscosity

TABLE S7.1

Molecular and Eddy Viscosities and Diffusivities (m 2 /sec)

Molecular, at salinity [ 35

Eddy: horizontal

(along-isopycnal)

Eddy: vertical

(diapycnal)

Viscosity 1.83 10 6 m 2 /sec at 0 C 1.05 10 6 m 2 /sec at 20 C 10 2 to 10 4 m 2 /sec 10 4 m 2 /sec

Thermal diffusivity 1.37 10 7 m 2 /sec at 0 C 1.46 10 7 m 2 /sec at 20 C 10 2 to 10 4 m 2 /sec 10 5 m 2 /sec

Haline diffusivity 1.3 10 9 m 2 /sec 10 2 to 10 4 m 2 /sec 10 5 m 2 /sec

MOMENTUM BALANCE 9

has no spatial dependence, viscous stress enters

the momentum equations as


¼ y v 2 u=vx 2 þ v 2 u=vy 2 þ v 2 u=vz 2 (7.9)

where y is the molecular (kinematic) viscosity.

(The dynamic viscosity is ry.) This expression

comes from the divergence of the viscous stress

in the x-direction. For the example shown in

Figure S7.1.e, this stress is yvu/vz, and there is

an acceleration of the fluid only if this stress

varies with z.

Molecular viscosity changes flow very

slowly. Its effectiveness can be gauged by

a non-dimensional parameter, the Reynolds

number, which is the ratio of the dissipation

timescale to the advective timescale: Re ¼ UL/y.

When the Reynolds number is large, the flow

is nearly inviscid and most likely very turbulent;

this is the case for flows governed by molecular

viscosity. However, from matching observations

and theory we know that the ocean currents

dissipate energy much more quickly than we

can predict using molecular viscosity. How

this happens is described next.

7.2.4.2. Eddy Viscosity

Mixing at spatial scales larger than those

quickly affected by molecular viscosity is generally

a result of turbulence in the fluid. Turbulent

motions stir the fluid, deforming and pulling it

into elongated, narrow filaments. A stirred fluid

mixes much faster than one that is calm and

subjected only to molecular motion. While stirring

is technically reversible, mixing is not. It

is easier to think about this for a property,

such as milk in a coffee cup, or salinity in the

ocean, than for velocity, but the same principles

apply to both. The filaments are deformed by

turbulence on a smaller spatial scale. Eventually

molecular viscosity takes over, when the spatial

scales become very small. We refer to the effect

of this turbulent stirring/mixing on the fluid as

eddy viscosity.

There is no obvious way to derive the size

of eddy viscosity from molecular properties.

Instead, it is determined empirically, either

directly from observations, or indirectly from

models that work relatively well and include

eddy viscosity. For large-scale ocean circulation,

the “turbulent” motions are mesoscale

eddies, vertical fine structure, and so on,

with spatial scales smaller than the larger

scales of interest. Like molecular viscosity,

eddy viscosity should be proportional to the

product of turbulent speed and path length.

Therefore, horizontal eddy viscosity is generally

much larger than vertical eddy viscosity

(Table S7.1). More specifically, although we

often refer to “horizontal” and “vertical”

eddy viscosity, the relevant directions are

along isopycnals (adiabatic surfaces) and

across isopycnals (diapycnal mixing), since these

are the natural coordinates for uninhibited

quasi-lateral motion and the most inhibited

quasi-vertical motion (Redi, 1982; Gent &

McWilliams, 1990).

To mathematically include eddy viscosity, the

viscous terms in Eqs. (7.1) and (7.9) are replaced

by the eddy viscosity terms:


¼ A H v 2 u=vx 2 þ v 2 u=vy 2 þ A V v 2 u=vz 2

(7.10a)

where A H is the horizontal eddy viscosity and A V

is the vertical eddy viscosity. (The use of the

symbol A is from the early German definition

of an “Austausch” or exchange coefficient to

represent eddy viscosity.) A H and A V have units

of kinematic viscosity, m 2 /sec in mks units.

(Although we often use these Cartesian coordinates,

the most relevant stirring/mixing directions

are along isopycnals (adiabatic surfaces)

and across isopycnals (diapycnal mixing), so the

coordinate system used in Eq. 7.10a is better

modeled by rotating it to have the “vertical”

direction perpendicular to isopycnal surfaces,

and replace A H and A V with eddy viscosities

10


that are along and perpendicular to those

surfaces.)

In many applications and observations, it is

useful to include spatial dependence in the

eddy viscosity coefficients because turbulence

is unevenly distributed. Equation (7.10a) is

then written in its original form, which includes

spatial variation of stress:


¼ v=vxðA H vu=vxÞþv=vyðA H vu=vyÞ

þ v=vzðA V vu=vzÞ

(7.10b)

Eddy viscosity coefficients (Table S7.1 and

Section 7.3), also called eddy momentum

diffusion coefficients, are inferred from observations

of microstructure (very small scale

variations in velocity) and from eddy diffusivities

acting on temperature and salinity that are

also derived from observations, given that both

are due to similar structures that mix the

ocean. (Formally, in fluid mechanics, the nondimensional

ratio of viscous diffusivity to

thermal diffusivity is called the Prandtl number;

if we assume that eddy viscosity and eddy

diffusivity were equal, we are assuming

a turbulent Prandtl number of 1.) Numerical

models typically use higher eddy viscosities

than eddy diffusivities (e.g., Smith, Maltrud,

Bryan, & Hecht, 2000; Treguier, 2006).

7.2.5. Mathematical Expression of

Momentum Balance

The full momentum balance with spatially

varying eddy viscosity and rotation is

Dv=Dt þ fu ¼ vv=vt þ u vv=vx þ v vv=vz

¼

þ w vv=vz þ fu

ð1=rÞvp=vy þ v=vxðA H vv=vxÞ

þ v=vyðA H vv=vyÞ

þ v=vzðA V vv=vzÞ

Dw=Dt ¼ vw=vt þ u vw=vx þ v vw=vy

þ w vw=vz

(7.11b)

¼ ð1=rÞvp=vz g þ v=vxðA H vw=vxÞ

þ v=vyðA H vw=vyÞ

þ v=vzðA V vw=vzÞ

(7.11c)

Here the standard notation “D/Dt” is the

substantial derivative defined in Eq. (7.4).

The full set of equations describing the physical

state of the ocean must also include the mass

conservation equation (Section 5.1):

Dr=Dt þ rðvu=vx þ vv=vy þ vw=vzÞ ¼0

(7.11d)

If density changes are small, Eq. 7.11d is approximated

as

vu=vx þ vv=vy þ vw=vz ¼ 0

(7.11e)

which is known as the continuity equation.

The set is completed by the equations governing

changes in temperature, salinity, and

density, which are presented in the following

section.

Du=Dt

fv ¼ vu=vt þ u vu=vx þ v vu=vy

þ w vu=vz fv

¼ ð1=rÞvp=vx þ v=vxðA H vu=vxÞ

þ v=vyðA H vu=vyÞ

þ v=vzðA V vu=vzÞ

(7.11a)

7.3. TEMPERATURE, SALINITY,

AND DENSITY EVOLUTION

Evolution equations for temperature and

salinity d the equation of state that relates

density to salinity, temperature, and pressure,

and thus an evolution equation for density d

complete the set of equations (7.11aed) that

TEMPERATURE, SALINITY, AND DENSITY EVOLUTION 11

describe fluid flow in the ocean. The boundary

and initial conditions required for solving the

systems of equations are beyond our scope.

7.3.1. Temperature, Salinity, and

Density Equations

Temperature is changed by heating, cooling,

and diffusion. Therefore the most basic equation

would be that for heat (or enthalpy), but most

dynamical treatments and models use an

explicit temperature equation. Salinity is

changed by addition or removal of freshwater,

which alters the dilution of the salts. Most

modeling uses an explicit salinity equation

rather than a freshwater equation. Density is

then computed from temperature and salinity

using the equation of state of seawater. The

“word” equations for temperature, salinity,

and density forcing include:

temperature change

þ temperature advection=convection

¼ heating=cooling term þ diffusion

(7.12a)

salinity change

þ salinity advection=convection

¼ evaporation=precipitation=runoff

=brine rejection þ diffusion

(7.12b)

equation of state ðdependence of density

on salinity; temperature; and pressureÞ

(7.12c)

density change þ density advection=convection

¼ density sources þ diffusion

(7.12d)

Written in full, these are

DT=Dt ¼ vT=vt þ u vT=vx þ v vT=vy

þ w vT=vz

¼ Q H =rc p þ v=vx k H vT=vx

þ v=vy k H vT=vy þ v=vz k V vT=vz

(7.13a)

DS=Dt ¼ vS=vt þ u vS=vx þ v vS=vy þ w vS=vz

¼ Q s þ v=vxðk H vS=vxÞ

þ v=vyðk H vS=vyÞþv=vzðk V vS=vzÞ

(7.13b)

r ¼ rðS; T; pÞ

(7.13c)

Dr=Dt ¼ vr=vt þ u vr=vx þ v vr=vy þ w vr=vz

¼ðvr=vSÞDS=Dt þðvr=vTÞDT=Dt

þðvr=vpÞDp=Dt

(7.13d)

where Q H is the heat source (positive for heating,

negative for cooling, applied mainly near

the sea surface), c p is the specific heat of

seawater, and Q S is the salinity “source” (positive

for evaporation and brine rejection, negative

for precipitation and runoff, applied at

or near the sea surface). (See also Chapter 5 for

discussion of heat and salinity.) k H and k V are

the horizontal and vertical eddy diffusivities,

analogous to the horizontal and vertical eddy

viscosities in the momentum equations

(7.11aed; Table S7.1. The full equation of state

appears in Eq. (7.13c), from which the evolution

of density in terms of temperature and salinity

change can be computed (Eq. 7.13d). The coefficients

for the three terms in Eq. (7.13d) are the

haline contraction coefficient, the thermal

expansion coefficient, and the adiabatic

compressibility, which is proportional to the

inverse of sound speed (Chapter 3).

12


7.3.2. Molecular and Eddy Diffusivity

The molecular diffusivity k for each

substance depends on the substance and the

fluid. The molecular diffusivity of salt in

seawater is much smaller than that for heat

(Table S7.1). This difference results in a process

called “double diffusion” (Section 7.4.3).

Eddy diffusivity is the equivalent of eddy

viscosity for properties like heat and salt. Eddy

diffusivity and eddy viscosity typically have

similar orders of magnitude (Table S7.1) since

the same turbulent processes create both. For

lack of observations and for simplicity, diapycnal,

quasi-vertical eddy diffusivity was once

considered to be globally uniform (e.g.,

Stommel & Arons, 1960a, b; Section 7.10.3). A

globally averaged vertical eddy diffusivity of

k v ¼ 1 10 4 m 2 /sec accounts for the observed

average vertical density structure (Section

7.10.2; Munk, 1966). However, the directly

observed vertical (or diapycnal) eddy diffusivity

in most of the ocean is a factor of 10 lower: k v ~

1 10 5 m 2 /sec, based on direct measurements

of mixing rates using dye release and spread

(Ledwell, Watson, & Law, 1993, 1998), measurements

of very small scale vertical structure

(Osborn & Cox, 1972; Gregg, 1987), and largescale

property distributions within the pycnocline

(e.g., Olbers, Wenzel, & Willebrand, 1985).

This implies regions of much higher diffusivity

to reach the global average.

Measurements show huge enhancements of

diapycnal eddy diffusivity in bottom boundary

regions, especially where topography is rough

(Figure S7.4; Polzin, Toole, Ledwell, & Schmitt,

1997; Kunze et al., 2006), and on continental

shelves where tidal energy is focused (Lien &

Gregg, 2001). In these regions, the tides move

water back and forth over hundreds of meters

horizontally (Egbert & Ray, 2001). If the bottom

is rough, as it is over most mid-ocean ridge

systems (Figures 2.5 and 2.6), the internal tide

can break, causing enhanced turbulence and

diffusivity. Internal tides have been directly

observed and related to turbulence along the

Hawaiian Ridge (Rudnick et al., 2003). If the

interaction is strong, then the enhanced diffusivity

can reach high into the water column,

even reaching the pycnocline, as is seen over

the topographic ridges in Figure S7.4.

Diapycnal eddy diffusivity also depends on

latitude (Figure S7.4b). It is small at low latitudes

(order of 10 6 m 2 /sec), increasing to

a peak at 20e30 latitude, and then declining

somewhat toward higher latitudes (0.4 to

0.5 10 4 m 2 /sec). The relation of this diffusivity

distribution to the actual efficiency of mixing,

which also depends on the currents, has not

yet been mapped.

Within the water column, away from the top

and bottom boundaries, internal waves are

generally relatively quiescent, without much

breaking, but nonlinear interactions between

internal waves and encounters with mesoscale

eddies could also produce higher velocity

shears that result in a low level of breaking

and turbulence.

In the surface layer, eddy diffusivities and

eddy viscosities are also much greater than the

Munk value (e.g., Large, McWilliams, & Doney,

1994). In Section 7.5.3 on Ekman layers, we

describe eddy viscosities in the surface layer

on the order of 100 to 1000 10 4 m 2 /sec. Large

lateral variations in diapycnal diffusivity result

from the processes that create the turbulence,

such as strongly sheared currents (such as the

Gulf Stream) and wind-forced near-inertial

motions near the base of the mixed layer.

Horizontal eddy diffusivities k H are estimated

to be between 10 3 and 10 4 m 2 /sec, with large

spatial variability (e.g., Figure 14.17). k H is

much larger than k V . The larger size is related

to the larger horizontal length and velocity scales

than in the vertical; turbulent motions and mixing

are enhanced in the horizontal. Observational

estimates of horizontal diffusivity have

been based on dye release (Ledwell et al.,

1998) and on dispersion of floats and surface

drifters (Section 14.5.1). Estimates of horizontal

MIXING LAYERS 13

FIGURE S7.4 (a) Observed diapycnal diffusivity (m 2 /s 2 ) along 32 S in the Indian Ocean, which is representative of other

ocean transects of diffusivity. (b) Average diapycnal diffusivity as a function of latitude range (color codes). Source: From

Kunze et al. (2006).

diffusivity are also made from choices required

to match observed and theoretical phenomena

such as boundary current widths. It is emphasized

that, unlike molecular diffusivities, eddy

diffusivities are not a property of the flow in

general, but depend on which space and timescales

are “resolved” and “unresolved.”

7.4. MIXING LAYERS

Mixing occurs throughout the ocean. Mixing

of momentum is the frictional process while

mixing of properties is the diffusion process.

While it is weak, it is essential for maintaining

the observed stratification and can regulate the

14


strength of some parts of the circulation. In this

section we look at mixing in the surface layer

where there is direct atmospheric forcing; in

bottom layers where mixing can be caused by

interaction between ocean currents, tides, and

waves and the bottom topography; and within

the water column, away from top and bottom

boundaries.

7.4.1. Surface Mixed Layer

The surface layer (Section 4.2.2) is forced

directly by the atmosphere through surface

wind stress and buoyancy (heat and freshwater)

exchange. The surface “mixed layer” is seldom

completely mixed, so it is sometimes difficult

to define. We consider it to be the upper

boundary layer of the ocean, forced directly by

the atmosphere through: (1) surface stress of

the wind and (2) buoyancy (heat and freshwater)

exchange through the surface. Wind

stress generates motion, which is strongest at

the surface and decreases with depth, that is,

with vertical shear in the velocity. These

motions include waves that add turbulent

energy to increase mixing, particularly if they

break. Wind-driven Langmuir circulations

(Section 7.5.2) can promote mixing, possibly

through the full depth of the mixed layer.

For a surface layer that is initially stably

stratified (Figure S7.5a), sufficiently large

wind stress will create turbulence that mixes

and creates a substantially uniform density or

mixed layer (Figure S7.5b). This typically

results in a discontinuity in properties at the

mixed layer base.

The upper layer can also be mixed by buoyancy

loss through the sea surface, increasing

the density of the top of the surface layer and

causing it to overturn (convect) to a greater depth

(Figure S7.5cee). This type of mixed layer typically

has no discontinuity in density at its base.

Heat or freshwater gain decreases the density of

the top of the surface layer, resulting in a more

stably stratified profile. If the wind then mixes

it, the final mixed layer is shallower than the

initial mixed layer (Figure S7.5feh).

Mixed layer observations typically show

much more vertical structure than might be

expected from these simple ideas. This is

because the mixed layer is subject to greatly

varying forcing, including diurnal heating that

restratifies the mixed layer, cooling that convectively

mixes the layer, wind-generated turbulence

that mechanically stirs the layer, and

small-scale instabilities of the many localized

fronts within the mixed layer that can change

its stratification (e.g., Boccaletti, Ferrari, & Fox-

Kemper, 2007).

The thickest mixed layers occur at the end of

winter (Figure 4.5), after an accumulation of

months of cooling that deepens the mixed layer

and increases its density. For large-scale oceanographic

studies, these end-of-winter mixed

layers set the properties that are subducted

into the ocean interior (Section 7.8.5). Maps of

late winter mixed layer depth and also

maximum mixed layer depth are shown in

Figure 4.5.

Several different parameterizations of surface

layer mixing due to winds and buoyancy fluxes

have been used. The first parameterization used

(“K-T”) was developed by Kraus and Turner

(1967). The Price, Weller, Pinkel (1986; PWP)

model largely replaced the K-T model and is

still used widely. Large et al. (1994) proposed

the most commonly used modern approach,

called “K-Profile Parameterization” (KPP),

where “K” is shorthand for diffusivity k. The

KPP model extends the response to surface

forcing to below the completely mixed layer,

since turbulence set up at the base of the wellmixed

layer penetrates downward; for instance,

through near-inertial motions (Sections 7.5.1

and 14.5.3).

7.4.2. Bottom Mixed Layers

Near the ocean bottom, turbulence, and hence

mixing, can be generated by currents or current

MIXING LAYERS 15

shear caused by the interaction with the bottom.

In shallow (e.g., coastal) waters, complete mixing

of the water column occurs if the depth (H) is

shallow enough and the tidal currents (U) are

fast enough (see reviews in Simpson, 1998 &

Brink, 2005). Complete mixing of the water

column occurs if the depth (H) is shallow enough

and the tidal currents (U) are fast enough. From

energy dissipation arguments, a useful critical

parameter based on depth and velocity is

H/U 3 .WhenH/U 3 < a, where a is proportional

to the empirically determined mixing efficiency

and the buoyancy flux, there can be complete

mixing that destroys the stratification. Considerable

observational efforts have been made and

are ongoing to quantify and understand the

turbulence that creates the mixing (Doron,

Bertuccioli, Katz & Osborn, 2001; Polzin, Toole,

Ledwell, & Schmitt, 1997; Kunze et al., 2006;

Lien & Gregg, 2001 and many others).

At longer timescales on the shelf, a bottom

Ekman layer can develop in which frictional

and Coriolis forces balance (Ekman, 1905 and

Section 7.5.3), with the bottom slope also

affecting the layer. The bottom slope on the

shelf, and its intersection with the water column’s

density structure, is important for

bottom Ekman layers, which can have both

upslope and downslope flow. Eddy viscosity

also has important variations in space and

time, which affects the Ekman layer structure

(Lentz, 1995).

Enhanced turbulence in a bottom boundary

layer can be created by movement of water

across rough topography and by breaking of

internal waves that reflect off the topography

and result in higher eddy diffusivity values

(Figure S7.5). The higher turbulence creates

locally mixed bottom boundary layers that can

then be advected away from the bottom topography,

creating “steppy” vertical profiles near

the bottom some distance from the mixing site

(Figure S7.6a).

Bottom currents due to density differences

can also cause mixing. One example is

a turbidity current down an underlying bottom

slope. (See Section 2.6. Another example is the

overflow of dense water across a sill, as seen at

the Strait of Gibraltar (Chapter 9.) The dense

water flows down the continental slope as

a plume, mixing vigorously with the lighter

water around it (Figure S7.6b). This turbulent

process is called entrainment.

As it entrains, the outflow reaches an equilibrium

density with the ambient water and

spreads thereafter along that isopycnal surface.

The entrainment rate and the final density of

the plume depend on the density of the strait

outflow and the density profile in the ambient

waters outside the strait.

Density differences due to the injection of

lighter water into the ocean also cause mixing

and entrainment. An example is hot hydrothermal

water injected at mid-ocean ridges

and hotspots that entrain ambient waters as

the plumes rise. A man-made example is water

from a sewer outfall where the discharged fluid

is less dense than the seawater. In both cases,

mixing (entrainment) takes place as the plumes

rise due to their buoyancy.

7.4.3. Internal Mixing Layers

In the interior of the ocean (i.e., away from

boundaries), continuous profiling instruments

have shown that vertical profiles of water properties

d temperature and salinity, and hence

density d are often not smooth (Figure S7.5i)

but “stepped” (Figure S7.5j). The vertical scale

of the steps can be decimeters to many meters.

Turbulence (Section 7.4.3.1) and/or double

diffusion (Section 7.4.3.2) mix the water column

internally and can create such steps.

7.4.3.1. Turbulent Mixing

Breaking internal waves can create internal

mixing (Section 8.4; Rudnick et al., 2003).

Vertical shear from other sources can also result

in turbulence. On the other hand, vertical stratification

stabilizes the mixing. One way to

16


FIGURE S7.5 Mixed layer development. (a, b) An initially stratified layer mixed by turbulence created by wind stress;

(c, d, e) an initial mixed layer subjected to heat loss at the surface, which deepens the mixed layer; (f, g, h) an initial mixed

layer subjected to heat gain and then to turbulent mixing presumably by the wind, resulting in a thinner mixed layer; (i, j) an

initially stratified profile subjected to internal mixing, which creates a stepped profile. Notation: s is wind stress and Q is

heat (buoyancy).

MIXING LAYERS 17

(a)

FIGURE S7.6 (a) Bottom

boundary layers and their advection

away from the bottom Source:

From Armi (1978). (b) Mixing of

a plume of dense water as it flows

out over a strait into less dense

ambient waters. After Price and

Baringer (1994).

(b)

Air-Sea

Exchange

(buoyancy loss)

ρ = constant

Descent and Entrainment

Equilibration

Ocean-Sea

Exchange

Marginal sea

dense water production

Ocean

express this trade-off is through a non-dimensional

quantity called the Richardson number:

where

R i ¼ N 2 =ðvu=vzÞ 2 (7.14)

N 2 ¼ g vr=vz r 0 (7.15)

N is the Brunt-Väisälä frequency (Section

3.5.6) and the vertical shear of the horizontal

speed is (vu/vz). If the Richardson number

is small, the stratification is weak and the

shear is large, so we expect mixing to be

vigorous. From theory and observations,

vigorous mixing starts when the Richardson

number falls below 1/4.

The initial steps of mixing between two horizontally

adjacent waters with strong temperature/salinity

differences are visible at the

front between the waters. Stirring at the front

draws layers of the adjacent waters into each

other along isopycnals, resulting in interleaving

or fine structure, with layering of one to tens

of meters on both sides of the front. The interleaving

facilitates local vertical (diapycnal)

mixing between the two water masses, which

is the next step to actual mixing between

them. The actual mixing can take place through

turbulent processes or the double diffusive

processes described in the next subsection. In

both cases, much smaller scale vertical structure,

on the order of centimeters (microstructure),

is an indication of the actual mixing at

18


the interfaces between the interleaving layers.

Such interleaving has been observed in the

western equatorial Pacific, in the Antarctic

Circumpolar Current (ACC), in the Kuroshio

and Gulf Stream, and so forth; that is, in every

region where there are strong water mass

fronts.

7.4.3.2. Double Diffusion

Heat diffuses about 100 times faster than salt

(Table S7.1). Double diffusion is due to these

differing molecular diffusivities, acting at scales

of centimeters to meters, and can also create

well-mixed internal layers. When warm, salty

water lies above cold, fresh water, and the interface

between the two is disturbed so that small

columns of warm, salty water are next to cold,

fresh ones, the fast heat exchange between

them will cool the saltier water and warm the

fresher water while the salinity will mix much

less. The saltier water becomes denser and tends

to sink into the lower layer and vice versa

(Figure S7.7a). The alternating columns are

called salt fingers. In the laboratory, salt fingers

can be produced easily and can grow to a few

millimeters across and up to 25 cm long. Lateral

diffusion occurs between the “fingers” and

produces a uniform layer. Then the process

FIGURE S7.7 Double diffusion:

(a) salt fingering interface (cold,

fresh water warms and rises;

warm, salty water cools and sinks).

(b) Diffusive interface. (c) North

Atlantic Mediterranean eddy

salinity profile with steps due to

salt fingering (25 23’N, 26 W). (d)

Arctic temperature profile with

diffusive layering. Source: From

Kelley et al. (2003).

(a)

Warm

Salty

Lower density

Salt fingering

layer(s)

Cold

Fresh

Higher density

(c)

Salt

fingering

layers

Salty

Fresh

500

Pressure (dbar)

1000

Warm

Cold

(b)

Salt fingering

Cold

Fresh

Lower density

(d)

-2

150

35 36

Salinity (psu)

5 10 15


N. Atlantic “Meddy” salt fingering

Temperature (°C)

-1 0 1

Diffusive layer(s)

Warm

Salty

Higher density

Diffusive layering

Depth (m)

200

250

Arctic diffusive layering (Kelley et al., 2003)


may start again at the two interfaces that are

now present, and eventually a number of individually

homogeneous layers develop with

sharply defined temperature and salinity interfaces

(as in Figure S7.5j). In the ocean the layers

may be meters to tens of meters thick, separated

by thinner interface zones of sharp gradients of

temperature and salinity. External horizontal

velocities can disturb the growing fingers, so

prediction of salt finger growth under all

oceanic conditions is complex.

When cold, fresh water lies above warm, salty

water (Figure S7.7b), the lower layer, losing heat

through the interface but not much salt, will

become denser and water will tend to sink, again

within its own layer. This is called the diffusive

form of double diffusion. The original stratification

is strengthened by this double diffusive

process. An important difference from salt

fingering is that fluid does not cross the interface.

Salt fingering effects are observed in the ocean

where there are strong contrasts in salinity, for

instance, where salty Mediterranean Water

enters the Atlantic through the Strait of Gibraltar

(Figure S7.7c). The saline water intrudes at

mid-depth (about 1000e2000 m) into the cooler,

less saline Atlantic water (Section 4.3 and

Figure 4.10). Step structures in temperature/

depth and salinity/depth traces due to double

diffusion are clear in CTD profiles below the

Mediterranean water (Figures S7.7c and S9.33)..

Diffusive interfaces are observed in high latitude

regions where there is a fresh, cold layer at the

surface with an underlying saltier temperature

maximum layer (a dichothermal layer; Sections

4.2 and 4.3.2 and Figure S7.7d).

7.5. RESPONSE TO WIND FORCING

The wind blows over the sea surface exerting

stress and causing the water to move within the

top 50 m. Initially the wind excites small capillary

waves that propagate in the direction of the wind.

Continued wind-driven momentum exchange

excites a range of surface waves (Chapter 8). The

net effect of this input of atmospheric momentum

is a stress on the ocean (wind stress) (Section 5.8).

For timescales of about a day and longer, Earth’s

rotation becomes important and the Coriolis

effect enters in, as described in the following

subsections.

7.5.1. Inertial Currents

The ocean responds initially to a wind stress

impulse with transient motions known as “inertial

currents.” These are a balance of the Coriolis

force and the time derivatives of the initial horizontal

velocities caused by the wind stress. In

the Northern Hemisphere, Coriolis force acts

to the right of the velocity. So if the current is

initially to the north, then Coriolis will move it

to the east, and then to the south, and so forth.

Thus, the water particles trace out clockwise

circles (Figure S7.8a). In the Southern Hemisphere,

Coriolis force acts to the left and inertial

currents are counterclockwise.

(Mathematically, inertial currents are the

solution of

vu=vt ¼ fv (7.16a)

vv=vt ¼ fu (7.16b)

which is taken from Eq. 7.11a and b assuming

that advection, pressure gradient forces, and

dissipation are very small and can be neglected.)

Since the Coriolis force is involved, inertial

currents vary with latitude. They have shorter

time and length scales for higher latitudes. The

frequency of an inertial current (time for a full

circle) is the Coriolis parameter f, so the time it

takes for the circle (the period) is 2p/f. Since the

rotation is to the right of the initial stress (wind

impulse), the average flow over a full circle of

the inertial current is perpendicular to the wind

stress and to the right in the Northern Hemisphere

and left in the Southern Hemisphere.

Inertial currents are often observed in surface

drifter trajectories and surface velocity

20


(a)

(b)

Northern hemisphere

Southern hemisphere

T = 2π/f

North

T = 2π/f

(c)

49.0

East

48.5

48.0

Latitude

47.5

47.0

46.5

46.0

-143 -142 -141 -140 -139 -138 -137

Longitude

FIGURE S7.8 (a) Schematics of inertial currents in the Northern and Southern Hemispheres. (b) Hodograph of inertial

currents at 45 N for a wind blowing in the y-direction; the numbers are in pendulum hours. Source: From Ekman (1905).

(c) Observations of near-inertial currents. Surface drifter tracks during and after a storm. Source: From d’Asaro et al. (1995).

moorings in the wake of a storm (Figure S7.8c).

Inertial periods are often very close to tidal

periods, so separating tidal and inertial effects

in time series is sometimes difficult.

After the wind starts to blow impulsively, the

current will initially oscillate around and then,

after several days, settle frictionally to a steady

flow at an angle to the wind (Figure S7.8b

from Ekman, 1905). This becomes the surface

Ekman velocity (Section 7.5.3).

7.5.2. Langmuir Circulation

“Langmuir circulation” is another transient

response to impulsive wind forcing, in which

helical vortices form near the sea surface.

Langmuir cells (LCs) were first discussed by

Langmuir (1938) who carried out a number of

experiments to identify their character. LCs

are visually evident as numerous long parallel

lines or streaks of flotsam (“windrows”) that

are mostly aligned with the wind, although

they can deviate by 20 degrees (Figure S7.9).

The streaks are formed by the convergence

caused by the vortices (Figure S7.10). Alternate

cells rotate in opposite directions so that convergence

and downwelling occurs at the surface (to

form streaks of flotsam) between pairs of adjacent

cells, while divergence and upwelling

occurs between alternate pairs. (LCs only

become apparent to the eye when there is flotsam

on the surface to be brought together by the


FIGURE S7.9 “Windrows” of foam, associated with the Langmuir circulation in Loch Ness. The surface wave field suggests

the wind direction, which is parallel to the narrow bands of foam. Source: From Thorpe (2004).

Waves

Convergence

Wind

5° to 15°

u' ~ 1% Wind

Convergence

FIGURE S7.10 Langmuir circulation,

first described by Langmuir

(1938). Source: From Smith (2001).

v' asymmetric?

v' ~ u'?

bubbles

bubbles

w' ~ u'

Ekman

Spiral

Plankton in "zones of retention"

Thermocline

(Form of bottom part not well known)

convergences.) The water in the cells progresses

downwind as well, so that its motion is helical.

LCs have typical depth and horizontal

spacing of 4e6 m and 10e50 m, but they can

range up to several hundred meters horizontal

separation and up to two to three times the

mixed layer depth. The cells can be many

kilometers long. Multiple scales have been

observed simultaneously in strong wind conditions

(Assaf, Gerard, and Gordon, 1971). The

22


downwelling zones are concentrated in jets

occupying one-third or less of the cell width

under the streaks while upwelling is more

widely distributed at smaller speeds. Velocities

within an LC are only a fraction of the wind

velocities that create them. Thus, the horizontal

flow speed at the surface in the streaks can add

10 cm/sec to the non-Langmuir currents elsewhere

in the surface layer. The vertical downwelling

at the convergences is about one-third of

the surface water speed as driven by the wind.

Downwelling velocities are several centimeters

per second and up to 20 cm/sec.

Langmuir circulations only appear when

there are wind waves on the water surface, as

in Figure S7.9. Surface films that dampen small

waves tend to inhibit the formation of the cells.

LCs generally occur only for wind speeds

greater than 3 m/sec and appear within a few

tens of minutes of wind onset. The mechanism

for producing Langmuir circulation is beyond

the scope of this text. See Smith (2001) and

Thorpe (2004) for further discussions.

Langmuir circulations provide a mechanism

for converting wave energy to turbulent energy

and mixing and causing the upper layer to

deepen. Mixed layer observations suggest that

Langmuir downwelling can penetrate to at least

the middle of the mixed layer, therefore, it is

expected that the downwelling plumes can penetrate

to the bottom of the actively mixing layer

(Weller et al., 1985; Smith, 2001). Langmuir cells

can generate internal waves in the stratified layer

below the mixed layer that contribute to moving

momentum from the mixed layer into the interior

(Polton, Smith, Mackinnon, & Tejada-Martinez,

2008). Thus LCs are one of several processes

that may contribute to surface mixing.

Note that Ekman’s theory of the wind drift

(Section 7.5.3) yields an upper layer motion

that is about 45 degrees to the right of the

wind, whereas LCs are more closely aligned

to the wind. This is because the timescales of

the two mechanisms are quite different. LCs

are generated within minutes of the wind

onset and die out soon after the strong wind

pulse, whereas the Ekman circulation takes

many hours to develop.

7.5.3. Ekman Layers

Wind stress is communicated to the ocean

surface layer through viscous (frictional)

processes that extend several tens of meters

into the ocean. For timescales longer than

a day, the response is strongly affected by Coriolis

acceleration. This wind-driven frictional

layer is called the Ekman layer after Walfrid

Ekman (1905), who based his theory on ship

drift observations of the Fram in the Arctic. 2

The classical surface Ekman layer is the

steady frictional response to a steady wind

stress on the ocean surface (Figure S7.11). The

physical processes in an Ekman layer include

only friction (eddy viscosity) and Coriolis acceleration.

Velocity in the Ekman layer is strongest

at the sea surface and decays exponentially

downward, disappearing at a depth of about

50 m. It coexists with, but is not the same as,

the mixed layer depth or euphotic zone depth.

The two most unusual characteristics of an

Ekman layer (compared with a frictional flow

that is not rotating) are (1) the horizontal

velocity vector spirals with increasing depth

(Figure S7.11) and (2) the net transport integrated

through the Ekman layer is exactly to

the right of the wind in the Northern Hemisphere

(left in the Southern Hemisphere).

The surface water in an Ekman layer moves

at an angle to the wind because of Coriolis acceleration.

If eddy viscosity is independent of

depth, the angle is 45 degrees to the right of

2 Collected as part of Fridtjof Nansen’s Fram expedition, the ship drift and wind measurements were given to Ekman to

explain as his Ph.D. thesis, which focused on the response of water movement in the upper ocean to the wind stress. Later

analysis of these data showed that the sea ice drifted 20 to 40 degrees to the right of the wind (Nansen, 1922).


D E

V 7

V 6

V 9

45°

V 9

V 7

V 6

“Ekman flow” - Water velocities decreasing

and rotating with increasing depth:

V 5

Wind

V 5

W

V 4

Ekman spiral

V 4

Surface

V 3

V 3

Horizontal

Plane

Resultant volume transport

at right angles to wind

FIGURE S7.11 Ekman layer velocities (Northern Hemisphere).

Water velocity as a function of depth (upper

projection) and Ekman spiral (lower projection). The large

open arrow shows the direction of the total Ekman transport,

which is perpendicular to the wind.

the wind in the Northern Hemisphere (and to

the left of the wind in the Southern Hemisphere).

If viscosity is not constant with depth,

for instance, if the turbulence that creates

the eddy viscosity changes with depth, then

the angle between the surface velocity and the

wind will differ from 45 degrees.

As the surface parcel moves, a frictional stress

develops between it and the next layer below.

This accelerates the layer below, which moves

off to the right (Northern Hemisphere) of the

surface parcel. This second layer applies stress

to the third layer, and so on. The total stress

decays with depth, at a rate that depends on

the eddy viscosity coefficient A V . Since each

successively deeper layer is accelerated to the

V 2

V 2

V 1

V 1

V o

V o

right of the layer above it (Northern Hemisphere)

and has a weaker velocity than the layer above it,

the complete structure is a decaying “spiral.” If

the velocity arrows are projected onto a horizontal

plane, their tips form the Ekman spiral

(Figure S7.11). The whole spiral is referred to as

the “Ekman layer.”

The Ekman layer depth is the e-folding depth

of the decaying velocity:

D E ¼ð2A v =fÞ 1=2 (7.17)

Using a constant eddy viscosity of 0.05 m 2 /sec

from within the observed range (Section 7.5.5),

the Ekman layer depths at latitudes 10, 45, and

80 degrees are 63, 31, and 26 m, respectively.

The vertically integrated horizontal velocity in

the Ekman layer is called the Ekman transport:

Z

U E ¼ u E ðzÞ dz (7.18a)

Z

V E ¼ v E ðzÞ dz

(7.18b)

where u E and v E are the eastward and northward

velocities in the Ekman layer, and U E

and V E are the associated Ekman transports.

(Ekman “transport” has units of depth times

velocity, hence m 2 /sec, rather than area times

velocity.) Ekman transport in terms of the

wind stress is derived from Eq. (7.11):

U E ¼ s ðyÞ =ðrfÞ

(7.19a)

V E ¼ s ðxÞ =ðrfÞ (7.19b)

where s (x) and s (y) are the wind stresses positive

in the east and north directions, assuming no

time acceleration, advection, or pressure

gradient force, and setting the eddy friction

stress at the sea surface equal to the wind stress.

The Ekman transport is exactly perpendicular

and to the right (left) of the wind in the

Northern (Southern) Hemisphere (large arrow

in Figure 7.5).

24


For applications of Ekman layers to general

circulation (Sections 7.8 and 7.9), only the

Ekman transport matters. Thus, the actual

eddy viscosity and Ekman layer thickness are

unimportant.

Ekman layers also form in the ocean’s surface

layer below sea ice. When the ice is blown by the

wind, friction between the sea ice and water

drives the water. If the timescale is longer than

a day, the Coriolis effect is important, and an

Ekman layer develops.

Ekman layers also occur at the ocean bottom

(Section 7.4.2). Because of friction, the flow at the

bottom must be zero. When the timescale of the

deep flow is longer than a day, Coriolis acceleration

is important, and an Ekman layer

develops, also 50 to 100 m thick above the

bottom like the surface Ekman layer. If there is

a current (e.g., a geostrophic current), flowing

in the lower part of the water column over the

sea bottom, which we will assume for simplicity

to be flat, then there is a bottom frictional stress

on the water. The frictional stress at the bottom

acts in the opposite direction to the current.

The result of the stress is a frictional transport

to the right of the stress (Northern Hemisphere).

Therefore the frictional transport (bottom

Ekman layer transport) is to the left of the

current. The total current (interior plus Ekman)

must be zero at the bottom. The net result is an

Ekman current spiral in the bottom layer with

the total current rotating to the left as the bottom

is approached.

In shallow water, the top and bottom Ekman

layers can overlap, so that the right-turning

tendency in the top layer (Northern Hemisphere)

will overlap the left-turning tendency

in the bottom layer. The opposing right- and

left-turning effects will tend to cancel more

and more as the water depth decreases. If there

is a wind stress at the top surface that would

produce an Ekman layer of depth D E in deep

water, then in water of depth h, the approximate

angle a between the wind and the surface

flow is as listed in Table S7.2.Thatis,aswater

TABLE S7.2

Angle of Surface Flow, a to the Right of

the Wind Direction (Northern Hemisphere),

with Overlapping Surface and

Bottom Ekman Layers

h/D E a Net flow direction in the water column

>1 45 At 90 to right of wind

0.5 45 About 60 to right of wind



depth decreases, the net flow is more in the

direction of the wind.

Tides or internal waves (Chapter 8) rubbing

against the bottom can also generate bottom

Ekman-like layers, but with time-dependent

spiraling currents in the frictional layer.

7.5.4. Ekman Transport Convergence

and Wind Stress Curl

When the wind stress varies with position so

that Ekman transport varies with position, there

can be a convergence or divergence of water

within the Ekman layer. Convergence results

in downwelling of water out of the Ekman layer.

Divergence results in upwelling into the Ekman

layer. This is the mechanism that connects the

frictional forcing by wind of the surface layer

to the interior, geostrophic ocean circulation

(Section 7.8).

Divergence and convergence occur if the

transport varies in the same direction as the

transport. In Figure S7.12, withvaryingzonal

(west to east) wind, the Ekman transport is to

the right of the wind, and is convergent

because the zonal wind varies with latitude.

Note that it is not necessary for the Ekman

transports to be in opposite directions to have

divergence or convergence, just that the transports

change.

The vertical velocity w E at the base of the

Ekman layer is obtained from the divergence

of the Ekman transport, by vertically integrating


Up

North

East

Ekman

transport

the continuity equation Eq. (7.11e) over the

depth of the Ekman layer:

ðvU E =vx þ vV E =vyÞ ¼V,U E

¼ ðw surface w E Þ¼w E

(7.20)

where U E is the horizontal vector Ekman transport

and it is assumed that the vertical velocity

at the sea surface, w surface , is 0. When

Eq. (7.20) is negative, the transport is convergent

and there must be downwelling below the sea

surface (increasingly negative w E ). The relation

of Ekman transport divergence to the wind

stress from Eq. (7.19a, b) is

V,U E ¼ v=vxðs ðyÞ =ðrfÞÞ

Wind

Ekman

convergence/

pumping

(downwelling)

Ekman

divergence/

suction

(upwelling)

FIGURE S7.12 Ekman transport convergence and

divergence in the Northern Hemisphere due to variations in

a zonal (eastward) wind. Ekman transport is southward, to

the right of the wind. Divergent transport causes downwelling,

denoted by circles with a cross. Convergent transport

causes upwelling, denoted by circles with a dot.

v=vyðs ðxÞ =ðrfÞÞ

¼ k,V ðs=rfÞ

(7.21)

where s is the vector wind stress and k is the

unit vector in the vertical direction. Therefore,

in the Northern Hemisphere (f > 0), upwelling

into the Ekman layer results from positive

wind stress curl, and downwelling results

from negative wind stress curl. Downwelling

is referred to as Ekman pumping. Upwelling is

sometimes referred to as Ekman suction.

A global map of wind stress curl was shown

in Figure 5.16d, and is referred to frequently in

subsequent chapters because of its importance

for Ekman pumping/suction, although the

mapped quantity should include the Coriolis

parameter, f, to be related directly to upwelling

and downwelling.

Equatorial upwelling due to Ekman transport

results from the westward wind stress

(trade winds). These cause northward Ekman

transport north of the equator and southward

Ekman transport south of the equator. This

results in upwelling along the equator, even

though the wind stress curl is small because

of the Coriolis parameter dependence in

Eq. (7.21).

At the equator, where the Coriolis parameter

changes sign, zonal (east-west) winds can cause

Ekman convergence or divergence even without

any variation in the wind (Figure S7.13a,b).

Right on the equator, there is no Ekman layer

since the Coriolis force that would create it is

zero (f ¼ 0). However, it has been shown from

observations (Eriksen, 1982) that the Coriolis

force is important quite close to the equator in

the ocean, starting at about 1/4 latitude. If the

equatorial wind is westward (a trade wind),

then the Ekman transport just north of the

equator is northward, and the Ekman transport

just south of the equator is southward, and there

must be upwelling into the surface layer on the

equator. This is roughly included in Eq. (7.21)

because of the variation in f, although the equation

is not accurate right on the equator where f

vanishes.

The coastline is the other place where Ekman

transport divergence or convergence can occur,

and it is not included in Eq. (7.21), becausethis

divergence is due to the boundary condition at

the coast and not wind stress curl. If the wind

blows along the coast, then Ekman transport

is perpendicular to the coast, so there must be

26


(a)

Ekman transport (northward)

Northern

Hemisphere

Trade Winds

Equator

Ekman transport (southward)

Southern

Hemisphere

(b)

Sea surface

warm

Ekman

transport

Thermocline

Southern

Hemisphere

(c)

cold

Upwelling

Equator

North

Trade

Winds

warm

Ekman

transport

Northern

Hemisphere

Alongshore wind

Northern

Hemisphere

Upwelling region

Eastern boundary curr.

Isopycnal

Ekman transport

East

Isopycnal

Onshore

transport

Poleward undercurrent

FIGURE S7.13 Ekman transport divergence near the equator driven by easterly trade winds. (a) Ekman transports. (b)

Meridional cross-section showing effect on the thermocline and surface temperature. (c) Coastal upwelling system due to an

alongshore wind with offshore Ekman transport (Northern Hemisphere). The accompanying isopycnal deformations and

equatorward eastern boundary current and poleward undercurrent are also shown (see Section 7.9).

either downwelling or upwelling at the coast to

feed the Ekman layer (Figure S7.13c). This is

one mechanism for creation of coastal

upwelling and subtropical eastern boundary

current systems. The other mechanism is

wind stress curl in the near-coastal region that

drives upwelling (Section 7.9). One such

example is the California-Oregon coast, where

the mean wind during most of the year

includes a component that blows southward

along the coast. This causes westward

(offshore) Ekman transport to the right of the

wind. This means there must be upwelling at

the coast.


7.5.5. Observations of Ekman Response

and Wind Forcing

The Ekman theory has major consequences

for wind-driven ocean circulation. Thus it has

been important to confirm and refine Ekman’s

theory with ocean observations, beyond the

original ice, wind, and ship drift observations

used by Ekman (1905) and Nansen (1922).

For instance, one assumption, that the eddy

viscosity in the water column is constant with

depth, is not accurate. (Recall that the Ekman

transport is independent of viscosity, so the

variability of eddy viscosity does not matter

for large-scale circulation.) Eddy viscosity is

highest near the sea surface because of turbulence

resulting from wind waves and inertial

currents generated at the surface. Also, Ekman

assumed a steady wind. The speed with which

the Ekman circulation develops depends on

latitude, because the Coriolis force depends

on latitude. Observations of Ekman spirals

and Ekman response are very difficult because

of the time dependence of the wind. It takes

about one pendulum day for inertial and Langmuir

responses (Sections 7.5.1 and 7.5.2) to die

out and an essentially Ekman circulation to

develop.

Ekman layer observations are also difficult

because the spiral is thin compared with the

usual vertical resolution of current measurements.

Davis, deSzoeke, and Niiler (1981)

measured currents in the mixed layer in the

northeast Pacific. By filtering the data and lookingatresponsesatshortandlongtimescales,

they found that the currents at timescales of

longer than about one day looked like Ekman’s

theory. Chereskin (1995) measured currents in

the mixed layer in the California Current using

an Acoustic Doppler Current Profiler (Section

S16.5.5.1 of Chapter S16 in the online supplement).

Because the wind direction there was

relatively steady, the Ekman-like response

was clear (Figure S7.14) even without filtering

the data. The high eddy viscosity values that

(a)

(b)

North Velocity (cm/s)

8

6

4

2

0

−2

−4

−6

Average currents and wind (6 Jun − 4 Oct 1993)

OBSERVATIONS

(slab extrapolation)

8

12 16

Mean wind

(m/s)

EKMAN THEORY

De = 25 m

2

A = 274.2 cm /s 16

12

8

4

−8

−8 −6 −4 −2 0 2 4 6 8

East Velocity (cm/s)

0

De = 48 m

2

A = 1011 cm /s

FIGURE S7.14 Observations of an Ekman-like response

in the California Current region. (a) Progressive vector

diagrams (Section 6.5.2) at 8, 16, 24, 32, and 40 m depth.

Because of the way the ADCP measures, the currents are

shown relative to a deeper depth, rather than as absolute

currents. The wind direction and speed for each day is

shown by the small arrows on the 8 m progressive vector

curve. (b) Observed mean velocities (left) and two theoretical

Ekman spirals (offset) using different eddy diffusivities

(274 and 1011 cm 2 /S). The numbers on the arrows are

depths. The large arrow is the mean wind. Source: From

Chereskin (1995).

28


Chereskin reported (Section 7.5.3) were obtained

by fitting the observed spiral to an Ekman layer

with depth-dependent viscosity.

An Ekman response to the wind for a large

part of the Pacific Ocean is apparent in the

average 15 m velocity from surface drifters

deployed in the 1980s and 1990s. The surface

drifters were drogued at 15 m depth, within the

Ekman layer. The drifter velocities from many

years of observations were averaged and the

average geostrophic velocity was subtracted.

The resulting “ageostrophic” velocities, which

are likely the Ekman response, are to the right

of the wind stress in the Northern Hemisphere

and to the left in the Southern Hemisphere

(Figure S7.15).

The Ekman volume transports (horizontal

and vertical) for each ocean and for the World

Ocean are shown in Figure S7.16. The easterly

trade winds (blowing westward) cause poleward

horizontal Ekman flows in the tropical

Atlantic and Pacific. The westerlies (blowing

eastward) cause equatorward flows at higher

latitudes. The Pacific Ekman transports are

larger than the Atlantic transports mainly

because the Pacific is so much wider, not

because the wind stress differs. The near-equatorial

Indian transports are of the opposite sign

compared with the Pacific, Atlantic, and total

transports because of the large annual monsoon

cycle; the westerly winds dominate the annual

mean in the equatorial Indian Ocean.

Associated with the convergences and divergences

of the horizontal Ekman flows are

vertical flows due to Ekman pumping (Figure

S7.16b). Between approximately 40 S and

40 N, downwelling prevails and the winds

cause convergent Ekman transport. Poleward

of about 40 degrees, there is upwelling caused

by divergent Ekman transport. The narrow

region of Ekman upwelling at about 5 to 10 N

is associated with the Intertropical Convergence

Zone in the winds. Not shown is the major

upwelling along the equator that must result

from the divergent Ekman transports there

due to the change of sign in the Coriolis parameter.

Again the Pacific and Atlantic have similar

distributions and the Indian Ocean differs

because of its strong annual (monsoonal) variation

north of the equator.

FIGURE S7.15 Ekman response. Average wind vectors (red) and average ageostrophic current at 15 m depth (blue).

The current is calculated from 7 years of surface drifters drogued at 15 m, with the geostrophic current based on

average density data from Levitus, Boyer, and Antonov (1994a) removed. (No arrows were plotted within 5 degrees of the

equator because the Coriolis force is small there.) This figure can also be found in the color insert. Source: From Ralph and

Niiler (1999).


FIGURE S7.16 (a) Zonally integrated meridional Ekman fluxes (Sv) for the three oceans by latitude and month. (Positive

is northward, negative is southward.) (b) Zonally integrated vertical Ekman volume flux (Sv) at the base of the Ekman layer

per 10 degrees latitude belt by latitude and month. (Positive is up, negative is down.) Source: From Levitus (1988).

30


7.6. GEOSTROPHIC BALANCE


Coriolis Force Balance

Throughout most of the ocean at timescales

longer than several days and at spatial scales

longer than several kilometers, the balance of

forces in the horizontal is between the pressure

gradient and the Coriolis force. This is called

“geostrophic balance” or geostrophy. 3

In a “word” equation, geostrophic balance is

horizontal Coriolis acceleration

¼ horizontal pressure gradient force (7.22)

This is illustrated in Figure S7.17. The pressure

gradient force vector points from high pressure

to low pressure. In a non-rotating flow, the

water would then move from high to low pressure.

However, with rotation, the Coriolis force

exactly opposes the pressure gradient force, so

that the net force is zero. Thus, the water parcel

does not accelerate (relative to Earth). The parcel

moves exactly perpendicular to both the pressure

gradient force and the Coriolis force.

A heuristic way to remember the direction of

geostrophic flow is to think of the pressure

gradient force pushing the water parcel from

high to low pressure, but Coriolis force moves

the parcel off to the right (Northern Hemisphere)

or the left (Southern Hemisphere). In

the resulting steady geostrophic state, the water

parcel moves exactly perpendicular to the pressure

gradient force.

The vertical force balance that goes with

geostrophy is hydrostatic balance (Section 3.2).

The vertical pressure gradient force, which

points upward from high pressure to low pressure,

is balanced by gravity, which points

downward. Thus vertical acceleration, advection,

and diffusion are assumed to be very

(a)

(b)

PGF

Low

pressure

v (in)

CF

PGF

Low

pressure

FIGURE S7.17

v (velocity)

z

x

High

pressure

CF

High

pressure

small, just as in the horizontal momentum

equations. (We note that in full treatments of

rotating fluid dynamics, the student will learn

that hydrostatic balance holds for a very large

range of fluid flows, not just those that are

geostrophic.)

The mathematical expression of geostrophy

and hydrostatic balance, from Eq. (7.11a, b, c), is

y

CF

x

PGF

. v (out)

Low

pressure

Geostrophic balance: horizontal forces

and velocity. (a) Horizontal forces and velocity in

geostrophic balance. PGF ¼ pressure gradient force.

CF ¼ Coriolis force. (b) Side view showing elevated pressure

(sea surface) in center, low pressure on sides, balance of

PGF and CF, and direction of velocity v (into and out of

page).

3 The other terms in the force balance d the actual acceleration, the advection, and diffusion d never completely vanish, so

no flow is exactly geostrophic.


fv ¼ ð1=rÞvp=vx (7.23a)

fu ¼ ð1=rÞvp=vy (7.23b)

0 ¼ vp=vz rg (7.23c)

An alternate form for Eq. (7.23c), used for

dynamic height calculations (Section 7.6.3), is

0 ¼ a vp=vz g (7.23d)

where a is specific volume. Note how many of the

terms in Eq. (7.11) have been assumed to be very

small and therefore are left out in Eq. 7.23a,b). 4

From Eq. (7.23a,b), if the Coriolis parameter

is approximately constant (f ¼ f o ) and if density

in Eq. (7.23a,b) is also very nearly constant

(r ¼ r o ; the “Boussinesq approximation”), the

geostrophic velocities are approximately nondivergent:

vu=vx þ vv=vy ¼ 0

(7.23e)

Formally in fluid dynamics, such a non-divergent

velocity field can be written in terms of

a streamfunction j:

u ¼ vj=vy and v ¼ vj=vx (7.23f)

From Eqs. (7.23a, b) the streamfunction for

geostrophic flow is j ¼ p/(f o r o ). Therefore,

maps of pressure distribution (or its proxies

like dynamic height, steric height, or geopotential

anomaly; Section 7.6.2) are maps of the

geostrophic streamfunction, and flow approximately

follows the mapped contours.

Geostrophic balance is intuitively familiar to

those with a general interest in weather reports.

Weather maps show high and low pressure

regions around which the winds blow (Figure

S7.18). Low pressure regions in the atmosphere

are called cyclones. Hurricanes, dramatic winter

storms, and tornados are all cyclones. Flow

around low-pressure regions is thus called

cyclonic (counterclockwise in the Northern

Hemisphere and clockwise in the Southern

Hemisphere). Flow around high-pressure

regions is called anticyclonic.

In the ocean, higher pressure can be caused

by a higher mass of water lying above the observation

depth. At the “sea surface,” pressure

differences are due to an actual mounding of

water relative to Earth’s geoid. Over the

complete width of the Atlantic or Pacific Ocean

anticyclonic gyres, the total contrast in seasurface

height is about 1 m.

The geostrophic velocities at the sea surface

could be calculated if the appropriately timeaveraged

sea-surface height were known (as

yet not possible for the time mean, but definitely

possible from satellite altimetry for variations

from the mean). The geostrophic velocity at

the sea surface in terms of sea-surface height h

above a level surface is derived from Eqs.

(7.23a,b):

fv ¼ gvh=vx (7.24a)

fu ¼ gvh=vy (7.24b)

4 Rigorous justification of geostrophic balance is based on small Rossby and Ekman numbers, where the Rossby number is

defined in Section 7.2.3, and the Ekman number is the non-dimensional parameter that is the ratio of the size of the viscous

term to the size of the Coriolis term. For the vertical direction, the Ekman number is E V ¼ 2A V /fH 2 , where f is the Coriolis

parameter and H is a characteristic vertical length scale; note the resemblance of this parameter to the Ekman layer depth

in Eq. (7.17). Hydrostatic balance (Eq. 7.23c,d) is valid when the non-dimensional aspect ratio, which is the ratio of the

vertical scale of motion (H) to the horizontal scale of motion (L); that is, d ¼ H/L, is small. Hydrostatic balance is even

more strongly justified when the Rossby number is small; that is, the substantial derivative in the z-momentum equation

scales as the square of the aspect ratio times the Rossby number. Performing a complete “scale analysis” in which these

assumptions are rigorously applied to the full set of momentum equations, thus deriving the balances in Eq. (7.23), is far

beyond the scope of this text.

32


FIGURE S7.18 Example of a daily weather map for North America, showing high- and low-pressure regions. Winds are

generally not from high to low, but rather clockwise around the highs and counterclockwise around the lows. Source: From

NOAA National Weather Service (2005).

To calculate the horizontal pressure difference

below the sea surface, we have to consider

both the total height of the pile of water above

our observation depth and also its density,

since the total mass determines the actual pressure

at our observation depth (Figure S7.19).

(This is where the vertical hydrostatic balance

in Eq. 7.23c enters.) Therefore, if a mound of

less dense water lies above us in one location

and a shorter column of denser water in

another location, the total mass in the two places

could be the same. Close to the sea surface,

there would be a pressure difference between

the two places since the sea surface is higher

in one location than in the other, but at depth

the pressure difference would vanish because

the difference in densities cancels the difference

in heights. Therefore there would be

a geostrophic flow at the sea surface, which

would decrease with depth until it vanishes at

our observation depth (h 3 in Figure S7.19a),

where the total mass of the two columns of

water is the same.

The variation in geostrophic flow with depth

(the geostrophic velocity shear) is therefore proportional

to the difference in density of the two

water columns on either side of our observation

location. The relation between the geostrophic

velocity shear and the horizontal change

(gradient) in density is called the thermal wind

relation, since it was originally developed by

meteorologists measuring temperature and

wind, rather than by oceanographers measuring

density and currents. 5

5 The thermal wind balance should not be confused with the thermohaline circulation, which refers to ocean overturning

directly involving buoyancy fluxes (Section 7.10).


(a)

Η Β

Η Α PGF

h 1

ρ Α

v

ρ Β

The thermal wind relation is illustrated in

Figure S7.19b. The sea surface is sloped, with

surface pressure higher to the right. This creates

a pressure gradient force to the left, which

drives a surface geostrophic current into the

page (Northern Hemisphere). The density r

increases with depth, and the isopycnals are

tilted. Therefore the geostrophic velocity

changes with depth because the pressure

gradient force changes with depth due to the

tilted isopycnals. Because the isopycnals are

sloped in the opposite direction to the seasurface

height, the into-the-page geostrophic

velocity is reduced with depth. That is, when

there is light water under a high sea surface

and dense water under a low sea surface, the

horizontal pressure gradients become smaller

with depth, since the mass of the two columns

becomes more equalized with depth.

A useful rule of thumb for geostrophic flows

that are surface-intensified is that, when facing

downstream in the Northern Hemisphere, the

h 2

h 3

(b)

p 1

PGF

ρ 1

p 2

ρ 2

p 3

ρ 3 B

FIGURE S7.19 Geostrophic flow and thermal wind

balance. (a) Schematic of change in pressure gradient

force (PGF) with depth, assuming that the left column (A)

is shorter and denser than the right column (B), that is,

r A > r B and H A < H B . The horizontal geostrophic velocity

V is into the page for this direction of PGF and is strongest

at the top, weakening with depth, as indicated by the

circle sizes. (If the densities of the two columns were the

same, then the PGF and velocity V are the same at all

depths.) (b) Same, but for density (red) increasing with

depth, and isopycnals tilted, and assuming that the sea

surface at B is higher than at A so that the PGF at the sea

surface (h 1 ) is to the left. The PGF decreases with

increasing depth, as indicated by the flattening of the

isobars p 2 and p 3 .

A

z

x

v

v

v

“light/warm” water is to your right. (In the

Southern Hemisphere, the light water is to

the left when facing downstream.) This can be

safely recalled by memorizing the example

for the Gulf Stream recalling that the current

flows eastward with warm water to the south.

Geostrophic flow with vertical shear, which

requires sloping isopycnals, is often called baroclinic.

Geostrophic flow without any vertical

shear is often called barotropic. Barotropic

flow is driven only by horizontal variations

in sea-surface height. Most oceanic geostrophic

flows have both barotropic and baroclinic

components.

Mathematically, the thermal wind relations

are derived from the geostrophic and hydrostatic

balance Eq. (7.23):

fvv=vz ¼ðg=r 0 Þvr=vx

fvu=vz ¼ðg=r o Þvr=vy

(7.25a)

(7.25b)

(Here we have again used the Boussinesq

approximation, where r is replaced by the

constant r o in the x- and y-momentum equations,

whereas the fully variable density r must

be used in the hydrostatic balance equation.)

To calculate geostrophic velocity, we must

know the absolute horizontal pressure difference

between two locations. If we have only

the density distribution, we can calculate only

the current at one level relative to that at

another, that is, the geostrophic vertical shear.

To convert these relative currents into absolute

currents, we must determine or estimate the

absolute current or pressure gradient at some

level (reference level).

The selection of a reference level velocity is

one of the key problems in using the geostrophic

method to compute currents. A common, but

usually inaccurate, referencing approach has

been to assume (without measuring) that the

absolute current is zero at some depth (level of

no motion). In the case of western boundary

currents or in the ACC where the currents

34


extend to great depth, this is not a good assumption.

Nevertheless the relative geostrophic

surface current calculation can be revealing

since the surface currents are usually much

stronger than the deep ones, and small relative

error in the surface currents might be tolerated

while the same amount of error in the deep

currents is untenable.

In the next subsection, we introduce the

“dynamic” method widely used to calculate

geostrophic velocities (shear), and continue the

discussion of reference velocity choices.

7.6.2. Geopotential and Dynamic

Height Anomalies and Reference Level

Velocities

Historically and continuing to the present, it

has been too difficult and too expensive to

instrument the ocean to directly observe

velocity everywhere. Density profiles, which

are much more widely and cheaply collected,

are an excellent data set for estimating

geostrophic velocities using the thermal wind

relations and estimates of the reference level

velocity. This approach to mapping ocean

currents is called the “dynamic method”; it

was developed in the School of Geophysics in

Bergen, Norway, more than a century ago.

This is the same school where both Nansen

and Ekman worked, contributing some very

significant ideas to physical oceanography. The

dynamic method has origins in both oceanography

and meteorology. In the dynamic method,

the distribution of mass in the ocean is used to

compute an important component of the current

field. In this text the emphasis will be on how

this method is commonly used in descriptive

studies of ocean circulation rather than derivation

of the method.

In the ocean the distribution of mass is represented

by the distribution of density over both

the horizontal and vertical dimensions. Once

the density profiles at two locations (“stations”)

are calculated from observed temperature and

salinity, the distribution of mass at the two

stations can be used to calculate the vertical

shear of geostrophic velocity at the midpoint

between the two stations at all depths that are

common to the two stations (Section 7.6.1).

Then, if the velocity is known at one depth (or

is assumed to have a certain value, e.g., zero),

the vertical shear can be used to give the velocity

at all other depths. The assumed or measured

velocity at one depth is called the reference

velocity, and its depth is called the “reference

depth” or reference level. Thus in this method,

the horizontal change in the distribution of

mass creates the horizontal pressure gradient,

which drives the geostrophic flow.

Oceanographers have created two closely

related functions, geopotential anomaly and

dynamic height, whose horizontal gradients

represent the horizontal pressure gradient force.

Another closely related concept, steric height, is

used to study variations in sea level. All are

calculated from the density profiles computed

from the measured temperature and salinity

profiles. Sverdrup, Johnson, and Fleming

(1942), Gill and Niiler (1973), Gill (1982), Pond

and Pickard (1983), and Stewart (2008) are

a few of the many useful references for these

practical quantities.

The gradient of the geopotential, F, is in the

direction of the local force due to gravity (modified

to include centrifugal force). The geopotential

gradient is defined from hydrostatic balance

(Eq. 7.23c) as

dF ¼ gdz ¼ a dp (7.26a)

where a is specific volume. The units of geopotential

are m 2 /sec 2 or J/kg. For two isobaric

surfaces p 2 (upper) and p 1 (lower), the geopotential

is

Z

Z

F ¼ g dz ¼ gðz 2 z 1 Þ¼ adp (7.26b)


Geopotential height is defined as

Z ¼

9:8 ms 2 1 Z gdz

¼ 9:8 ms 2 1 Z a dp (7.26c)

and is nearly equal to geometric height. This

equation is in mks units; if centimeter-gramsecond

(cgs) units are used instead, the multiplicative

constant would change from 9.8 m s 2 to

980 cm s 2 . Most practical calculations, including

common seawater computer subroutines, use the

specific volume anomaly

d ¼ aðS; T; pÞ að35; 0; pÞ (7.26d)

to compute the geopotential anomaly

Z

DF ¼ d dp:

(7.26e)

The geopotential height anomaly is then defined

as

Z

Z 0 ¼ ð9:8 ms 2 Þ 1 d dp: (7.26f)

Geopotential height anomaly is effectively

identical to steric height anomaly, which is

defined by Gill and Niiler (1973) as

Z

h 0 ¼ ð1=r o Þ r 0 dz (7.27a)

in which the density anomaly r’ ¼ r r o . Using

hydrostatic balance and defining r o as

r(35,0,p), Eq. (7.27a) is equivalent to Tomczak

and Godfrey’s (1994) steric height (anomaly)

Z

h 0 ¼ dr o dz (7.27b)

which can be further manipulated to yield

Z

h 0 ¼ð1=gÞ d dp: (7.27c)

This is nearly identical to the geopotential

height anomaly in Eq. (7.26f), differing only in

the appearance of a standard quantity for g. In

SI units, steric height is in meters.

Dynamic height, D, is closely related to geopotential,

F, differing only in sign and units of

reporting. Many modern publications and

common computer subroutines do not distinguish

between dynamic height and geopotential

anomaly. The unit traditionally used for

dynamic height is the dynamic meter:

1 dyn m ¼ 10 m 2 =sec 2 : (7.28a)

Therefore dynamic height reported in dynamic

meters is related to geopotential anomaly as

Z

DD ¼ DF=10 ¼ d dp=10: (7.28b)

Its relation to the geopotential height and steric

height anomalies is

10 DD ¼ 9:8 Z 0 ¼ gh 0 : (7.28c)

The quantities DD and Z 0 are often used interchangeably,

differing only by 2%. With use of

the dynamic meter, maps of dynamic topography

are close to the actual geometric height

of an isobaric surface relative to a level surface;

for example, a horizontal variation of 1 dyn m

means that the isobaric surface has a horizontal

depth variation of about 1 m. Note that the geopotential

height anomaly more closely reflects

the actual height variation, so a variation of

1 dyn m would be an actual height variation

closer to 1.02 m.

Geostrophic velocities at one depth relative to

those at another depth are calculated using Eq.

(7.25) with geopotential anomalies, steric height

anomalies, or dynamic heights. In SI units, and

using dynamic meters for dynamic height, the

difference between the northward velocity v

and eastward velocity u at the pressure surface

p 2 relative to the pressure surface p 1 is

36


f v 2 v 1

¼ 10 vDD=vx ¼ vDF=vx

¼ gvh 0 =vx

f u 2 u 1

¼ 10 vDD=vy ¼ vDF=vy

(7.29a)

¼ gvh 0 =vy (7.29b)

where the dynamic height or geopotential

anomalies are integrated vertically from p 1 to

p 2 . The surface p 1 is the reference level. (Comparison

of Eq. 7.29 with Eq. 7.23 shows that the

dynamic height and geopotential anomalies

are streamfunctions for the difference between

geostrophic flows from one depth to another.)

How is the velocity at the reference level

chosen? Since the strength of ocean currents

decreases from the surface downward in many

(but not all) regions, for practical reasons,

a deep level of no motion has often been

presumed. A much better alternative is to use

a “level of known motion”. For example, current

meter measurements, or the tracks of subsurface

floats, may be used to define the current

at some level, and then dynamic or steric

heights can be used to compute currents at all

other levels relative to the known reference

level. Another modern practice is to require

that the entire flow field, which is defined by

many density profiles, satisfy some overall

constraints. An obvious one is that there can

be no net transport into a region enclosed by

a set of stations (otherwise there would be an

increasing mound or hole in that region).

Another one is that the chemistry must make

sense d there can no net production of oxygen

within the ocean outside the surface layer for

instance. Another type of constraint is that the

flows match measured velocities from current

meters or floats, but allowing for some error in

the match. The constraints then help narrow

the choices of reference level velocities. Formal

versions of these methods, first applied to the

reference level problem by Carl Wunsch in

the 1970s, are called inverse methods because of

the mathematics used to connect the constraints

to the choices of reference velocities (see

Wunsch, 1996).

Another apparently attractive option is to use

satellite altimeters to measure the sea-surface

height, which would give the pressure distribution

and hence geostrophic currents at the sea

surface. These can be used to reference the

geostrophic velocities calculated at all depths

below the surface using dynamic height

profiles. However, while the sea surface elevation

is measured very precisely by satellite

altimeters, the height includes Earth’s geoid,

which has large spatial variations that are not

yet well measured; this leads to spurious surface

currents if one simply calculates the gradient in

measured surface height. The geoid does not

vary in time, so satellite altimetry does provide

excellent information on time changes of the

surface geostrophic currents. The GRavity and

Earth Climate Experiment (GRACE) satellite,

launched in 2002 to measure the shorter spatial

scales of Earth’s gravity field, is helping to

resolve this geoid problem. Satellite altimeters

and GRACE are described in the online supplementary

Chapter S16.

As an example of the geostrophic method,

we calculate dynamic height and a geostrophic

velocity profile from two density profiles that

straddle the Gulf Stream (Figure S7.20 and

Table S7.3). The isopycnals sloping upward

toward the north between 38 and 39 Nmark

the horizontal pressure gradient associated

with the Gulf Stream (Figure S7.20a). The

geostrophic velocity profile is calculated

between stations “A” and “B” relative to an

arbitrary level of no motion at 3000 m. (If it

were known, the velocity at 3000m can be

added later to the full velocity profile.) Station

A has lower specific volume (higher potential

density) than station B (Figure S7.20b). The

surface dynamic height at A is therefore lower

than at B (Figure S7.20c) and the surface pressure

gradient force is toward the north, from

B to A. Therefore, the geostrophic velocity at

the midpoint between the stations (Figure


(a)

0

500

Stations

26.5

26.0

B

A

(b)

A

B

(c)

A

B

(d)

27.0

1000

27.5

27.7

1500

Depth (meters)

2000

2500

3000

27.8

3500

27.88

4000

4500

Potential

density

(kg/m 3 )

5000

38°N 39°N 40°N

100 200 300

Distance (km)

0 200 400 0 2 0 50 100

Specific volume anomaly

(x 10 –8 m 3 /kg)

Dynamic height

(dyn m)

Geostrophic velocity

(cm/sec)

FIGURE S7.20 (a) Potential density section across the Gulf Stream (66 W in 1997). (b) Specific volume anomaly

d ( 10 8 m 3 /kg) at stations A and B. (c) Dynamic height (dyn m) profiles at stations A and B, assuming reference level at

3000 m. (d) Eastward geostrophic velocity (cm/sec), assuming zero velocity at 3000 m.

S7.20d) is eastward and is largest at the sea

surface. This means that the sea surface must

tilt downward from B to A. The vertical shear

is largest in the upper 800 m where the difference

in dynamic heights is largest.

For practical applications in which the

maximum depths of density profiles vary, it is

often most convenient to first calculate dynamic

height by integrating from the surface downward

for every profile (Table S7.3), and then

calculate the associated geostrophic velocity

relative to 0 cm/sec at the sea surface (column

4inTable S7.3). This geostrophic velocity profile

is likely not close to the actual velocity profile,

since velocities are usually small at depth and

not at the sea surface. Then the assumed or independently

measured velocity at the chosen deep

reference level is compared with the velocity at

the reference level calculated relative to 0 cm/

sec at the sea surface, and the entire geostrophic

velocity profile is offset by the difference. For

instance, if our reference velocity choice is

0 cm/sec at 3000 m, then we look for the calculated

geostrophic velocity at 3000 m relative to

0 at the sea surface and subtract this from the

velocities at all depths (column 5 in Table

38

TABLE S7.3


Computation of Dynamic Height and Geostrophic Current Between Stations A and B Relative to an

Assumed Zero Velocity at 3000 m and at the Deepest Common Level (DCL)

Eastward speed (cm/sec)

Depth (m) D A (dyn m) D B (dyn m) Relative to sea surface Ref. 0 cm/sec at 3000 m Ref. 0 cm/sec at DCL*

0 0 0 0 94.83 95.13

50 0.230 0.211 2.78 92.04 92.32

100 0.416 0.340 10.70 84.12 84.40

150 0.552 0.425 17.89 76.94 77.22

200 0.657 0.495 22.85 71.98 72.26

300 0.835 0.611 31.52 63.31 63.59

400 1.005 0.708 41.80 53.03 53.31

500 1.164 0.785 53.27 41.56 41.84

600 1.300 0.846 63.79 31.04 31.32

800 1.511 0.947 79.37 15.45 15.73

1000 1.652 1.039 86.29 8.53 8.81

1500 1.904 1.267 90.19 4.64 4.92

2000 2.142 1.489 91.87 2.96 3.24

2500 2.377 1.712 93.45 1.37 1.65

3000 2.602 1.928 94.85 0.0 0.28

3500 2.814 2.136 95.34 0.52 0.24

4000 3.024 2.347 95.26 0.43 0.15

4500 3.243 2.566 95.13 0.31 0.03

4710 (DCL*) 3.343 2.667 95.10 0.28 0.0

Note: Although D is called “height” and is quoted in units of “dynamic meters,” it has physical dimensions of energy per unit mass as it

represents work done against gravity.

Distance between the two stations ¼ 78.0 km; latitude ¼ 38.65 N.

The SI units for D are J/kg ¼ m 2 /s 2 .

S7.3). If our best estimate of a reference velocity

is 0 cm/sec at the bottom (deepest common

level; DCL), then we offset the velocities by the

value at the bottom (column 6 in Table S7.3). If

we have measured the bottom current to be

5 cm/sec, then we add an offset to the complete

velocity profile so as to yield 5 cm/sec at the

bottom.

The DCL is the maximum depth at which

geostrophic velocity can be calculated for this

particular station pair, since the shallower of

the two stations extends to 4710 dbar (the

deeper of the pair extends to 4810 dbar). Especially

for transport calculations in which the

bottom current is of interest, the geostrophic

velocity below the DCL is needed, but there is


only one density profile available. There are

a variety of ways to assign velocity to this

“bottom triangle,” including (1) no assignment,

(2) assignment of velocity at the DCL, (3) extrapolation

of velocity profile from above the DCL,

(4) extrapolation of the velocity horizontally

from the next station pair if there is one, or (5)

objective mapping of the velocity field into the

triangle. The last is the best way, but an objective

mapping scheme might not be readily available.

7.6.3. Dynamic Topography and

Sea-Surface Height Maps

Dynamic height at one surface relative to

another is the streamfunction for the

geostrophic flow at that surface relative to the

other, as an extension of Eq. (7.23f). Flows are

along the contours with the high “hills” to the

right of the flow in the Northern Hemisphere

(to the left in the Southern Hemisphere). The

speed at any point is proportional to the steepness

of the slope at that point; in other words,

it is inversely proportional to the separation of

the contours.

Dynamic topography maps (equivalently, steric

height or sea-surface height) are shown in

Chapter 14 and throughout the ocean basin

chapters (9e13) to depict the geostrophic flow

field. As an illustration of the common features

for all basins, we show here dynamic topography

maps for the Pacific and Atlantic Oceans

(Figures S7.21 and S7.22). These were the first

modern basin-wide maps in common use and

thus have some historical interest; both show

dynamic height relative to a deep level of no

motion. For comparison, Figures 9.2a and 10.2a

are the surface steric height maps from Reid

(1994, 1997) that we use to illustrate circulation

in the Pacific and Atlantic Ocean chapters. The

steric height in these maps has been adjusted

to represent the full flow, hence incorporating

estimates of deep geostrophic velocities at all

station pairs.

At the sea surface, all five ocean basins have

highest dynamic topography in the west in the

subtropics. The anticyclonic flows around these

highs are called the subtropical gyres. The

Northern Hemisphere oceans have low

dynamic topography around 50e60 N; the

cyclonic flows around these lows are the

subpolar gyres. Tightly spaced contours along

the western boundaries indicate the swift

western boundary currents for each of the gyres.

Low values are found all the way around Antarctica;

the band of tightly spaced contours to

its north marks the eastward ACC. The contrast

in dynamic height and sea-surface height from

high to low in a given gyre is about 0.5 to 1

dynamic meters.

In the subtropical gyres in Figures S7.21 and

S7.22, close contour spacings, hence large

geostrophic velocities, are found at the western

boundaries. These include the energetic

subtropical western boundary currents just

east of Japan (Kuroshio), east of North America

(Gulf Stream), east of Australia (East Australian

Current), east of Brazil (Brazil Current), and east

of southern Africa (Agulhas Current).

The similarity between the two Pacific

surface maps (and the two Atlantic surface

maps) indicates that indeed the flow at 1000

dbar (700 dbar) is relatively weak. The additional

advantage of the Reid (1994, 1997) analyses

is that he also produced maps of absolute

dynamic topography at 1000 dbar, and at 500

dbar intervals to the ocean bottom, whereas

the simple dynamic topography method

assuming a level of no motion clearly does not

yield a reasonable flow field at these depths.

7.6.4. A Two-Layer Ocean

It is frequently convenient to think of the

ocean as composed of two layers in the vertical,

with upper layer of density r 1 and lower layer of

density r 2 (Figure S7.23). The lower layer is

assumed to be infinitely deep. The upper layer

40


FIGURE S7.21 Mean annual dynamic topography of the Pacific Ocean sea surface relative to 1000 dbar in dyn cm

(DD ¼ 0/1000 dbar). Source: From Wyrtki (1975).

thickness is h + H, where h is the varying height

of the layer above the sea level surface and H is

the varying depth of the bottom of the layer. We

sample the layers with stations at “A” and “B.”

Using the hydrostatic equation (7.23c), we

compute the pressure at a depth Z at the

stations:

p A

¼ r 1 g h A þ H A

þ r2 g Z

H A

(7.30a)

p B

¼ r 1 g h B þ H B

þ r2 g Z H B

: (7.30b)

Here Z represents a common depth for both

stations, taken well below the interface. If we


FIGURE S7.22 Dynamic topography of 100 dbar surface relative to 700 dbar surface (DD ¼ 100/700 dbar) in dyn cm in the

Atlantic Ocean. Source: From Stommel, Niiler, and Anati (1978).

(a)

z

ρ 2

ρ

ρ 1

Pycnocline

(b)

h A

H A

h B

H B

Sea surface

Ideal level surface

Pycnocline

FIGURE S7.23 The two-layer ocean. (a) Vertical density profile with upper and lower layers of density r 1 and r 2 . (b) Sea

surface and pycnocline for two stations, A and B, where the thickness of the layer above the “ideal level surface” is h A and

h B and the thickness of the layer below the level surface is H A and H B , respectively. Both h and H are part of the “upper”

layer shown in (a).

42


FIGURE S7.24 Two-layer ocean

depiction of a (a) “cold,” cyclonic

ocean circulation showing the

“ideal” sea surface and the subsurface

thermocline structure and a (b)

“warm” anticyclonic circulation.

(a)

Sea surface

Level surface

Pycnocline

Cyclonic eddy

(b)

Sea surface

Level surface

Pycnocline

Anticyclonic eddy

assume that p A ¼ p B , which amounts to assuming

a “level of no motion” at Z, we can compute

a surface slope, which we cannot measure in

terms of the observed density interface slope:

h A

Dx

h B

¼ r 2 r 1 H A H B

r 1 Dx

(7.31a)

We then use Eq. (7.30a) to estimate the surface

velocity v:

h B

fv ¼ g h A

¼ g r 2 r 1 H A H B

(7.31b)

Dx r 1 Dx

This says that we can estimate the slope of the sea

surface (h A h B ) from knowledge of the subsurface

slope of the density interface (H A H B ),

which then allows us to estimate the surface

flow velocity from the shape of the pycnocline.

This simple construct is also useful in depicting

various forms of geostrophic circulation

features. For example, a cyclonic feature in either

hemisphere is drawn in Figure S7.24a where the

subsurface pycnocline slope is much greater than

the surface topographic change. These cyclonic

features are also known as cold features due to

the upwelling of the central isopycnals in the

center of the feature. This is true even if the

feature is not a closed circulation. Likewise

a warm feature looks like Figure S7.24b regardless

of hemisphere. What will change with the

hemisphere is the direction of the flow where

a warm feature rotates anticyclonically (clockwise

in the Northern Hemisphere) and a cold

feature is cyclonic (counterclockwise). The twolayer

depiction of the ocean is convenient for

quickly evaluating new measurements in terms

of the corresponding geostrophic currents. Note

that the two-layer assumption results in

a mapping of only geostrophic currents and not

the entire current field.


7.7. VORTICITY, POTENTIAL

VORTICITY, ROSSBY AND KELVIN

WAVES, AND INSTABILITIES

Ocean currents are mostly geostrophic. This

means that the equation for velocity includes

only the pressure gradient force and Coriolis

force. This poses an apparent problem: How

do we insert external forces such as the wind?

In formal geophysical fluid dynamics, we

would show that these forces are in the

momentum equations, but are so weak that we

safely consider the flows to be geostrophic (to

lowest order). To reinsert the external forces,

we have to consider the “vorticity” equation,

which is formally derived from the momentum

equations by combining the equations in a way

that eliminates the pressure gradient force

terms. (It is straightforward to do.) The resulting

equation gives the time change of the vorticity,

rather than the velocities. It also includes dissipation,

variation in Coriolis parameter with latitude,

and vertical velocities, which can be set

externally by Ekman pumping.

7.7.1. Vorticity

Vorticity in fluids is similar to angular

momentum in solids, and many of the intuitions

developed about angular momentum from

a standard physics course can be applied to

understanding vorticity.

Vorticity is twice the angular velocity at a point

in a fluid. It is easiest to visualize by thinking of

a small paddle wheel immersed in the fluid

(Figure S7.25). If the fluid flow turns the paddle

wheel, then it has vorticity. Vorticity is a vector,

and points out of the plane in which the fluid

turns. The sign of the vorticity is given by the

“right-hand” rule. If you curl the fingers on

your right hand in the direction of the turning

paddle wheel and your thumb points upward,

then the vorticity is positive. If your thumb

points downward, the vorticity is negative.

Vorticity is exactly related to the concept of

curl in vector calculus. The vorticity vector u

is the curl of the velocity vector v, expressed

here d in Cartesian coordinates:

u ¼ V v

¼ iðvv=vz vw=vyÞþjðvw=vx vu=vzÞ

þ kðvv=vx

vu=vyÞ

(7.32)

where (i, j, k) is the unit vector in Cartesian coordinates

(x, y, z) with corresponding velocity

components (u, v, w). Vorticity, therefore, has

units of inverse time, for instance, (sec) 1 .

(a)

Right-hand rule, thumb up:

North positive vorticity

(b)

North



Up

Current

Up

Current

East

East

Right-hand rule, thumb down:

negative vorticity

FIGURE S7.25 Vorticity. (a) Positive and (b) negative vorticity. The right-hand rule shows the direction of the vorticity by

the direction of the thumb (upward for positive, downward for negative).

44


Fluids (and all objects) have vorticity simply

because of Earth’s rotation. This is called planetary

vorticity. We do not normally appreciate

this component of vorticity since it is only important

if a motion lasts for a significant portion of

a day, and most important if it lasts for many

days, months, or years. Since geostrophic motion

is essentially steady compared with the rotation

time of Earth, planetary vorticity is very important

for nearly geostrophic flows. The vector

planetary vorticity points upward, parallel to

the rotation axis of Earth. Its size is twice the

angular rotation rate U of Earth:

u planetary ¼ 2U (7.33)

where U ¼ 2p/day ¼ 2p/86160 sec ¼ 7.293

10 5 sec 1 ,sou planetary ¼ 1.4586 10 4 sec 1 .

The vorticity of the fluid motion relative to

Earth’s surface (Eq. 7.32) is called the relative

vorticity. It is calculated from the water velocities

relative to Earth’s surface (which is rotating).

The total vorticity of a piece of fluid is the sum

of the relative vorticity and planetary vorticity.

The total vorticity is sometimes called absolute

vorticity, because it is the vorticity the fluid has

in the non-rotating reference frame of the stars.

For large-scale oceanography, only the local

vertical component of the total vorticity is

used because the fluid layers are thin compared

with Earth’s radius, so flows are nearly horizontal.

The local vertical component of the planetary

vorticity is exactly equal to the Coriolis

parameter f (Eq. 7.8c) and is therefore maximum

and positive at the North Pole (4 ¼ 90 N),

maximum and negative at the South Pole

(4 ¼ 90 S), and 0 at the equator.

The local vertical component of the relative

vorticity from Eq. (7.32) is

z ¼

vv

vx

vu

¼ curl z v (7.34)

vy

The local vertical component of the absolute

vorticity is therefore (z + f). The geostrophic

velocities calculated from Eq. (7.23) (Section

7.6) are often used to calculate relative vorticity.

7.7.2. Potential Vorticity

Potential vorticity is a dynamically important

quantity related to relative and planetary

vorticity. Conservation of potential vorticity is

one of the most important concepts in fluid

dynamics, just as conservation of angular

momentum is a central concept in solid body

mechanics. Potential vorticity takes into account

the height H of a water column as well as its

local spin (vorticity). If a column is shortened

and flattened (preserving mass), then it must

spin more slowly. On the other hand, if a column

is stretched and thinned (preserving mass), it

should spin more quickly similar to a spinning

ice skater or diver who spreads his or her arms

out and spins more slowly (due to conservation

of angular momentum). Potential vorticity,

when considering only the local vertical components,

is

Q ¼ðz þ fÞ=H (7.35)

where H is the thickness, if the fluid is unstratified.

When the fluid is stratified, the equivalent

version of potential vorticity is

Q ¼ ðz þ fÞð1=rÞðvr=vzÞ: (7.36)

When there are no forces (other than gravity) on

the fluid and no buoyancy sources that can

change density, potential vorticity Q is

conserved:

DQ=Dt ¼ 0 (7.37)

where “D/Dt” is the substantial derivative (Eq.

7.4). This means that a water parcel keeps the

value of Q that it obtains wherever a force acts

on it. For instance, parcels of water leaving the

ocean surface layer, where they are subject to

wind forcing, which changes their potential

vorticity, keep the same value of potential


vorticity after they enter the ocean interior where

forces (primarily friction) are much weaker.

Considering the potential vorticity (Eq. 7.35),

there are three quantities that can change: relative

vorticity z, the Coriolis parameter f, and

the thickness H (or equivalent thickness

(1/r)(vr/vz) in Eq. 7.36). The variation in f

with latitude has huge consequences for ocean

currents and stratification. Therefore, a special

symbol b is introduced to denote the change in

f with northward distance y, or in terms of latitude

f and Earth’s radius R e :

b ¼ df=dy ¼ 2U cos F=R e (7.38)

We often refer to the “b-effect” when talking

about how changes in latitude affect currents,

or the very large-scale, mainly horizontal

Rossby waves for which the b-effect is the

restoring force, described in Section 7.7.3.

All three components of potential vorticity

can change together, but we learn more about

what happens if we consider just two at a time.

First we consider changes in relative vorticity

z and Coriolis parameter f, holding thickness H

constant (Figure S7.26). When a water parcel is

moved northward, it experiences an increase in

f. Its relative vorticity z must then decrease to

keep the numerator of Eq. (7.35) constant. If z is

zero to start with, z will become negative and

the water parcel will rotate clockwise. If the

parcel is moved southward, f decreases and its

relative vorticity will have to become more positive;

the parcel will rotate counterclockwise.

Secondly, we consider changes in relative

vorticity z and thickness H, holding the latitude

and hence f constant (Figure S7.27). (This would

be appropriate for mesoscale eddies with high

relative vorticity that do not move far from their

initial latitude, Arctic dynamics, or for flows in

rotating laboratory tanks.) An increase in thickness

H (“stretching”) must result then in an

increase in relative vorticity, and the water

parcel will rotate more in the counterclockwise

direction. A decrease in thickness (“squashing”)

results in a decrease in relative vorticity, and the

water parcel will rotate more in the clockwise

direction.

Thirdly, if the thickness H is allowed to vary,

and if the relative vorticity is very small (such as

in the very weak currents in the mid-ocean),

then a northward move that increases f must

result in column stretching (Figure S7.28). Similarly,

a southward move would cause H to

decrease or squash. (Since neither thickness

H

Relative

vorticity

ζ = 0

Relative

vorticity

ζ < 0

FIGURE S7.26 Conservation of

potential vorticity: changes in relative

vorticity and Coriolis parameter

f, if thickness is constant.

move northwards

Latitude θ 1

Latitude θ 2

Q = f(θ 1

)/H

Q = (f(θ 2

) + ζ)/H = f(θ 1

)/H

Conservation of potential vorticity Q in the absence of

stretching (northern hemisphere):

balance of planetary vorticity and relative vorticity

46


Relative

vorticity

ζ = 0

Relative

vorticity

ζ > 0

H 2

H 1

Relative

vorticity

ζ = 0

Relative

vorticity

ζ = 0

H 2

Latitude θ 1

(same latitude) Latitude θ 1

Q = f(θ 1

)/H 1

Q = (f(θ 1

) + ζ)/H 2

= f(θ 1

)/H 1


planetary vorticity change (northern hemisphere):

balance of relative vorticity and stretching

FIGURE S7.27 Conservation of potential vorticity:

changes in thickness and relative vorticity, assuming

constant latitude (constant f).

nor relative vorticity can change without limit,

there is an inherent restoring force to northward

and southward movements in the ocean and

atmosphere. This restoring force creates Rossby

waves.)

In the Southern Hemisphere, f is negative. A

southward move of a water column makes f

even more negative, and requires stretching

(increase in H). A northward move makes f

less negative, and requires squashing (decrease

in H). Therefore, looking at both hemispheres,

we can say that poleward motion, toward larger

magnitude f, requires stretching. Equatorward

motion requires squashing.

The equator is a special place in terms of

potential vorticity, since f changes from negative

to positive crossing the equator and is zero on

the equator. Any water parcels moving into

the equatorial region must become more dominated

by relative vorticity, as in Figure S7.28.

We see this in the much stronger horizontal

current shears near the equator than at higher

latitudes. (Geostrophy also breaks down right

on the equator; slightly off the equator, small

pressure gradients result in large geostrophic

currents, so we also see high velocities in the

equatorial region compared with other

latitudes.)

Latitude θ move northwards

1 Latitude θ 2

Q = f(θ 1

)/H1

Q = f(θ 2

)/H 2

= f(θ 1

)/H 1


relative vorticity (northern hemisphere):

balance of planetary vorticity and stretching

FIGURE S7.28 Conservation of potential vorticity:

changes in thickness and latitude (Coriolis parameter f),

assuming negligible relative vorticity (Northern

Hemisphere).

7.7.3. Rossby Waves

The adjustment of any fluid to a change in

forcing takes the form of waves that move out

and leave behind a steady flow associated

with the new forcing. We describe some general

properties of waves in Chapter 8. The largescale,

almost geostrophic circulation adjusts to

changing winds and buoyancy forcing mainly

through “planetary” or Rossby waves and Kelvin

waves (Section 7.7.6). Pure Rossby and Kelvin

waves are never found except in simplified

models and lab experiments. However, much

of the ocean’s variability can be understood in

terms of Rossby wave properties, particularly

the tendency for westward propagation relative

to the mean flow. We describe these waves

without derivations, which can be found in the

many geophysical fluid dynamics textbooks

referenced at the start of this chapter.

A first important fact is that Rossby waves

have wavelengths of tens to thousands of kilometers.

Since the ocean is only 5 to 10 km deep

and is stratified, particle motions in Rossby

waves are almost completely transverse (horizontal,

parallel to the surface of Earth), which


differs from intuition that we build from watching

surface gravity waves.

Second, the restoring force for Rossby waves

is the variation in Coriolis parameter f with latitude,

so all dispersion information includes

b (Eq. 7.38). As a water column is shoved off

to a new latitude, its potential vorticity must

be conserved (Eq. 7.35). As with all waves, the

column overshoots, and then has to be restored

again, creating the wave. Therefore the water

column height or relative vorticity begin to

change. These cannot change indefinitely

without external forcing, so the water column

is restored back toward its original latitude. As

with all waves, the column overshoots, and

then has to be restored back again, creating the

wave. For a short wavelength Rossby wave,

the relative vorticity changes in response to the

change in Coriolis parameter f as in Figure

S7.26 d for a parcel moving northward to

higher f, the relative vorticity becomes negative.

This pushes columns to the east of the parcel

toward the south and pulls columns to the

west of parcel toward the north. The net effect

is a westward propagation of the wave. For

a long wavelength Rossby wave (Figure S7.29)

the column height changes in response to the

change in f, as in Figure S7.28. For northward

motion of the parcel, height increases; the downward

slope in height to the east causes southward

geostrophic flow on that side while the

downward slope in height to the west of the

perturbation causes northward geostrophic

flow on the west. The net effect again is westward

propagation of the disturbance.

Third, Rossby wave crests and troughs move

only westward (relative to any mean flow,

which could advect them to the east) in both

the Northern and Southern Hemispheres;

that is, the phase velocity is westward (plus

a northward or southward component). On

the other hand, the group velocity of Rossby

waves can be either westward or eastward.

The group velocity of Rossby waves is westward

for long wavelengths (more than about

50 km) and eastward for short wavelengths

(even though the zonal phase velocity is

westward).

westward phase propagation

H 2

Northward

geostrophic flow

Southward

geostrophic flow

North

Column moved north

Stretched

Latitude θ 2

H 1

Latitude θ 1

East

Long Rossby wave: f/H conserved (f 1 /H 1 = .f 2 /H 2 )

Geostrophic flow due to pressure ridges, moves columns northward or southward,

producing westward propagation of the wave

FIGURE S7.29 Schematic of a long wavelength Rossby wave.

48


Fourth, velocities in Rossby waves are almost

geostrophic. Therefore, they can be calculated

from variations in pressure; for instance, as

measured by a satellite altimeter, which

observes the sea-surface height. Behavior

similar to a Rossby wave (westward phase

propagation) can be seen at almost all latitudes

in each of the subtropical oceans in the satellite

altimetry images in Figures 14.18 and 14.19.

Although pure Rossby waves do not occur,

many variable flows such as eddies or meanders

of currents like the Gulf Stream Extension or

AAC can be interpreted in terms of Rossby

wave properties, in the sense that Rossby waves

are the basic set of linear wave solutions for

flows with a small Rossby number and small

aspect ratio. If the mean flows are removed

from observed variability, the variability often

appears to move westward. The atmosphere

has the same Rossby-wave-like phenomena,

such as those seen in daily weather maps

showing large loops or meanders in the Jet

Stream (Figure S7.18).

7.7.4. Rossby Deformation Radius and

Rossby Wave Dispersion Relation

Turning slightly more analytical, we introduce,

again without derivation, the Rossby

deformation radius and the dispersion relation

for Rossby waves with simple stratification.

The length scale that separates long from

short wavelength Rossby waves is called the

Rossby deformation radius. It is the intrinsic horizontal

length scale for geostrophic or nearly

geostrophic flows, relative to which all length

scales are compared. The Rossby radius characterizes

the observed mesoscale (eddy) length

scales and also the spatial decay scale of

boundary-trapped waves such as Kelvin waves

(Section 7.7.6) and the latitudinal width of equatorially

trapped waves.

The Rossby deformation radius in an unstratified

ocean is

R E ¼ðgHÞ 1=2 =f

(7.39a)

where H is the ocean depth scale. R E is called the

barotropic Rossby deformation radius or “external”

deformation radius. Barotropic deformation

radii are on the order of thousands of kilometers.

In an unstratified ocean, the horizontal

velocities for geostrophic flows are the same

(in magnitude and direction) from the top of

the ocean to the bottom. In the more realistic

stratified ocean, there is a similar “barotropic

mode,” with velocities in the same direction at

all depths, and with a barotropic Rossby deformation

radius also given by Eq. (7.39a).

The Rossby deformation radius associated

with the ocean’s stratification is

R I ¼ NH S =f

(7.39b)

where N is the Brunt-Väisälä frequency (Eq.

7.14), and H s is an intrinsic scale height for the

flow. R I is called the baroclinic deformation radius

(or “internal” deformation radius). “Baroclinic”

means that the velocity structure changes within

the water column, associated with isopycnal

slopes. The first baroclinic mode has a single

velocity reversal within the water column. The

vertical length scale H s associated with the first

baroclinic mode is about 1000 m, which is the

typical pycnocline depth. (The second baroclinic

mode has two velocity reversals and hence

a shorter vertical length scale, and so on for

the higher modes.) The vertical length scale H s

associated with the first baroclinic mode is

about 1000 m, which is the typical pycnocline

depth. R I for the first baroclinic mode varies

from more than 200 km in the tropics to around

10 km at high latitudes (Figure S7.30a; Chelton

et al., 1998).

The dispersion relation (Section 8.2) for first

mode baroclinic Rossby waves is

u ¼

bk

k 2 þ l 2 þð1=R I Þ 2 (7.40)


(b)

+120 ° +150 ° –180 ° –150 ° –120 ° – 90 ° – 60 ° –30 ° 0 ° +30 ° +60 ° +90 °

500

1000

+45 ° +60 ° +75 ° 50

0 °

100

200

200

50

50

100

0 °

–45 °

–60 °

–75 °

200

1000

500

FIGURE S7.30 (a) Rossby deformation radius (km) for the first baroclinic mode. Source: From Chelton et al. (1998).

(b) Shortest period (in days) for the first baroclinic mode, based on the deformation radius in (a). Note that the annual cycle,

at 365 days, occurs around latitudes 40 to 45 degrees; poleward of this, all such waves are slower. Source: From Wunsch (2009).

50


where u is the wave frequency, k and l are the

wavenumbers in the east-west (x) and northsouth

(y) directions, b is as in Eq. (7.38), and R I

is as given in Eq. (7.39b). Highest frequency

(shortest period) occurs at the wavelength associated

with the Rossby deformation radius

(Figure S7.31). The shortest periods vary from

less than 50 days in the tropics to more than 2

to 3 years at high latitudes (Figure S7.30b from

Wunsch, 2009). Poleward of about 40 to 45

degrees latitude there is no first baroclinic

mode at the annual cycle, so seasonal atmospheric

forcing cannot force the first baroclinic

mode at these higher latitudes. This results in

a fundamentally different response to atmospheric

variability at higher latitudes than in

the tropics and at mid-latitudes.

In much of the ocean away from the equator,

the barotropic and first baroclinic modes dominate

the variability, and hence the space and

timescales of the eddy field. At the equator,

a much larger set of baroclinic modes is typically

observed, resulting in much more complex

vertical velocity structure than at higher latitudes.

Equatorial Rossby waves are slightly

different from non-equatorial Rossby waves

since geostrophy does not hold at the equator,

but the vertical structures and behavior are

similar, with the restoring force for the equatorial

Rossby waves the same as at mid-latitude d the

change in Coriolis parameter with latitude.

7.7.5. Instability of Geostrophic Ocean

Currents

Almost all water flows are unsteady. When

gyre-scale flows break up, they do so into large

eddies, on the order of tens to hundreds of kilometers

in diameter or larger (see Section 14.5).

The size of the eddies is often approximately

the Rossby deformation radius. The eddies

usually move westward, like Rossby waves.

Instabilities of flows are often studied by considering

a mean flow and then finding the small

perturbations that can grow exponentially. This

approach is called “linear stability theory”; it is

linear because the perturbation is always

assumed to be small relative to the mean flow,

which hardly changes at all. When perturbations

are allowed to grow to maturity, when they

might be interacting with each other and

affecting the mean flow, the study has become

nonlinear.

We define three states: stable, neutrally stable,

and unstable. A stable flow returns to its original

state after it is perturbed. A neutrally stable flow

remains as is. In an unstable flow, the perturbation

grows.

The two sources of energy for instabilities are

the kinetic energy and the potential energy of the

mean flow. Recall from basic physics that kinetic

energy is ½ mv 2 where m is mass and v is speed;

for a fluid we replace the mass with density r,or

just look at the quantity ½ v 2 . Also recall from

basic physics that potential energy comes from

raising an object to a height; the work done in

raising the object gives it its potential energy.

In a stratified fluid like the ocean, there is no

available potential energy if isopycnals are flat,

which means that nothing can be released. For

there to be usable or available potential energy,

isopycnals must be tilted.

Barotropic instabilities feed on the kinetic

energy in the horizontal shear of the flow. For

instance, the Gulf Stream and similar strong

currents are jet-like, with large horizontal shear.

Their speeds exceed 100 cm/sec in the center of

the jet and decay to 0 cm/sec over about 50 km

on either side of the jet. Such currents also have

large kinetic energy because of their high

speeds. The kinetic energy can be released if

special conditions on the potential vorticity

structure of the current are met. These conditions

are that the horizontal shear be “large

enough” compared with a restoring b-effect

(Eq. 7.38), which creates Rossby waves in the

absence of sheared flow (previous subsection).

Barotropic instabilities can be thought of as the

(unstable) waves that occur in the presence of

a horizontally sheared current and possibly


0.1

(a) Period =

Rossby wave dispersion relation

2π/frequency

0.09

70 days

Latitude 20°N

Deformation radius R

0.08

I = 50 km

Frequency (day -1 )

0.07

0.06

0.05

0.04

0.03

100 days

200 days

(b)

0.02

0.01

20 km

100 km

200 km

0

-0.1 −0.09 −0.08 −0.07 −0.06 −0.05 −0.04 −0.03 −0.02 −0.01 0

Wavenumber (km -1 )

R I

500

450

400

350

Rossby wave dispersion relation

Latitude 20°N

Deformation radius R I = 50 km

300 days

400 days

500 days

Period (day)

300

250

200

150

100

50

length scales =

wavelength/2π

0

3000 2500 2000 1500 1000 500 0

R I

Wavelength (km)

FIGURE S7.31 Dispersion relation for first mode baroclinic Rossby waves (Eq. 7.40), assuming a deformation radius R I of

50 km, latitude 20 degrees (north or south) and y-wavenumber l ¼ 0. (a) Frequency u versus x-wavenumber k and (b) period

versus wavelength. The Rossby radius is shown with the dashed line. The highest frequency and shortest period are at the

Rossby radius length scale.

200 km

100 km

20 km

10 km

52


also the b-effect; see Pedlosky (1987). The net

effect of the barotropic instability is to reduce

the size of the horizontal shear. For the Gulf

Stream, for instance, this results in decreasing

the maximum speed at the core of the jet, and

inducing flows in the opposite direction on the

outskirts of the jet. These flows look like “recirculations”

(Section 9.3.2).

Baroclinic instabilities draw on the available

potential energy of the flow. The relatively

recent study of sub-mesoscale eddies and instabilities

generated in the ocean’s mixed layer is

essentially that of baroclinic instability operating

on density fronts within the mixed layer

(Boccaletti et al., 2007). The fronts are strongly

tilted isopycnals, which are then subject to this

kind of potential energy release.

Baroclinic instability is peculiar to geostrophic

flows, because Earth’s rotation makes it possible

to have a mean geostrophic flow with mean tilted

isopycnals. On the other hand, barotropic instability

is similar to instabilities of all sheared flows

including those without Earth’s rotation.

7.7.6. Kelvin Waves

Coastlines and the equator can support

a special type of hybrid wave called a “Kelvin

wave,” which includes both gravity wave and

Coriolis effects. Kelvin waves are “trapped” to

the coastlines and trapped at the equator, which

means that their amplitude is highest at the

coast (or equator) and decays exponentially

with offshore (or poleward) distance. Kelvin

waves are of particular importance on eastern

boundaries since they transfer information poleward

from the equator. They are also central to

how the equatorial ocean adjusts to changes in

wind forcing, such as during an El Niño

(Chapter 10).

Kelvin waves propagate with the coast to the

right in the Northern Hemisphere and to the left

in the Southern Hemisphere. At the equator,

which acts like a boundary, Kelvin waves propagate

only eastward. In their alongshore

direction of propagation, Kelvin waves behave

just like surface gravity waves and obey the

gravity wave dispersion relation (Section 8.3).

However, unlike surface gravity waves, Kelvin

waves can propagate in only one direction.

Kelvin wave wavelengths are also very long,

on the order of tens to thousands of kilometers,

compared with the usual surface gravity waves

at the beach. Although the wave propagation

speed is high, it can take days to weeks to see

the transition from a Kelvin wave crest to

a Kelvin wave trough at a given observation

point.

In the across-shore direction, Kelvin waves

differ entirely from surface gravity waves. Their

amplitude is largest at the coast. The offshore

decay scale is the Rossby deformation radius

(Section 7.7.4).

Lastly, Kelvin wave water velocities in the

direction perpendicular to the coast are exactly

zero. The water velocities are therefore exactly

parallel to the coast. Moreover, the alongshore

velocities are geostrophic, so they are associated

with pressure differences (pressure gradient

force) in the across-shore direction.

7.8. WIND-DRIVEN

CIRCULATION: SVERDRUP

BALANCE AND WESTERN

BOUNDARY CURRENTS

The large-scale circulation in the ocean basins

is asymmetric, with swift, narrow currents along

the western boundaries, and much gentler flow

within the vast interior, away from the side

boundaries. This asymmetry is known as westward

intensification of the circulation; it occurs in

both the Northern and Southern Hemispheres

and in the subtropical and subpolar gyres.

The Gulf Stream is the prototype of these

western boundary currents, as the first that

was extensively studied, and as the example

for which theories of westward intensification

were developed. In a book that summarizes


these theories, Stommel (1965) reviewed early

knowledge of the Gulf Stream, dating back to

the first explorations of the North Atlantic, and

summarized theoretical attempts to understand

it, dating back to the nineteenth century. When

the subtropical gyre and Gulf Stream were

finally modeled theoretically in the mid-twentieth

century, the resulting theory was breathtakingly

simple. The long delay in arriving at

this theory was due to the similarity between

the wind patterns above the subtropical North

Atlantic and the circulation d both are high

pressure systems d with anticyclonic flow

(clockwise in the Northern Hemisphere). But

the winds are clearly not westward intensified

relative to the ocean boundaries.

The primary originators of the theories that

provide our present understanding were

Harald Sverdrup, Henry Stommel, Walter

Munk, and Nicholas Fofonoff. Sverdrup (1947)

first explained the mid-ocean vorticity balance,

created by variations in Ekman transport, that

creates what we now call the “Sverdrup interior”

solution (Section 7.8.1). Just a few years

earlier, Sverdrup et al. (1942) were still suggesting

that the Ekman transport variations would

simply pile water up in the central gyre, with

a resulting geostrophic flow around the pile.

Stommel (1948) and Munk (1950) provided the

first (frictional) explanations for the western

boundary currents (Section 7.8.2), and Fofonoff

(1954) showed how very different the circulation

would be without friction.

Most of the physical effects described in this

section occur because the Coriolis parameter

varies with latitude, that is, because of the

b-effect (Eq. 7.38).

7.8.1. Sverdrup Balance

The gentle interior flow of the (non-equatorial)

oceans can be described in terms of their

meridional (north-south) direction. In the

subtropical gyres, the interior flow is toward

the equator in both the Northern and Southern

Hemispheres. In the subpolar gyres, the interior

flow is poleward in both hemispheres. These

interior flow directions can be understood

through a potential vorticity argument introduced

by Sverdrup (1947), so we call the applicable

physics the “Sverdrup balance.”

Consider a schematic of the subtropical

North Pacific (Figure S7.32). The winds at the

sea surface are not spatially uniform (Figure

5.16 and Figure S10.2 in the online supplement).

South of about 30 N, the Pacific is dominated by

easterly trade winds. North of this, it is dominated

by the westerlies. This causes northward

Ekman transport under the trade winds, and

southward Ekman transport under the westerlies.

As a result, there is Ekman convergence

throughout the subtropical North Pacific

(Figures 5.16d and S10.2).

The convergent surface layer water in the

subtropics must go somewhere so there is

downward vertical velocity at the base of the

(50 m thick) Ekman layer. At some level

between the surface and ocean bottom, there

is likely no vertical velocity. Therefore there

is net “squashing” of the water columns in

the subtropical region (also called Ekman

pumping; Section7.5.4).

This squashing requires a decrease in either

planetary or relative vorticity (Eq. 7.35). In the

ocean interior, relative vorticity is small, so planetary

vorticity must decrease, which results in

the equatorward flow that characterizes the

subtropical gyre (Figure S7.28).

The subpolar North Pacific lies north of the

westerly wind maximum at about 40 N. Ekman

transport is therefore southward, with

a maximum at about 40 N and weaker at higher

latitudes. Therefore there must be upwelling

(Ekman suction) throughout the wide latitude

band of the subpolar gyre. This upwelling

stretches the water columns (Eq. 7.35), which

then move poleward, creating the poleward

flow of the subpolar gyre.

The Sverdrup transport is the net meridional

transport diagnosed in both the subtropical

54


FIGURE S7.32 Sverdrup

balance circulation (Northern

Hemisphere). Westerly and trade

winds force Ekman transport

creating Ekman pumping and

suction and hence Sverdrup

transport.

North

Subpolar gyre

Westerlies

Subtropical gyre

Trades

Ekman transport

East

Ekman

upwelling

Tropical gyre

Ekman transport

Thermocline

Ekman

downwelling

Ekman

upwelling

Northern Hemisphere

Sverdrup transport

and subpolar gyres, resulting from planetary

vorticity changes that balance Ekman pumping

or Ekman suction.

All of the meridional flow is returned in

western boundary currents, for reasons described

in the following sections. Therefore, subtropical

gyres must be anticyclonic and subpolar gyres

must be cyclonic.

Mathematically, the Sverdrup balance is

derived from the geostrophic equations of

motion with variable Coriolis parameter f

(Eq. 7.23a,b). The x- and y-momentum equations

are combined to form the vorticity equation,

recalling that b = df/dy:

fðvu=vx þ vv=vyÞþbv ¼ 0 (7.41)

Using the continuity equation

vu=vx þ vv=vy þ vw=vz ¼ 0 (7.42)

Eq. (7.41) becomes the potential vorticity

balance

bv ¼ f vw=vz: (7.43)

This important equation states that water

column stretching in the presence of rotation is

balanced by a change in latitude (Figure S7.28).

In Eq. (7.43), the vertical velocity w is due to

Ekman pumping. From Eqs. (7.20) and (7.21):

w ¼ v=vx s ðyÞ =rf

v=vy s ðxÞ =rf

¼ }curl s} (7.44)

where s is the vector wind stress, s (x) is the zonal

wind stress, and s (y) is the meridional wind

stress. Assuming that the vertical velocity w is

zero at great depth, Eq. (7.43) can be vertically

integrated to obtain the Sverdrup balance:

b

M ðyÞ s ðxÞ =f


¼ v=vx s ðyÞ

s ðxÞ

¼ }curl s}

v=vy

(7.45)

where the meridional (south-north) mass transport

M (y) is the vertical integral of the meridional

velocity v times density r. The second

term on the left side is the meridional Ekman

transport. Thus, the meridional transport in the

Sverdrup interior is proportional to the wind

stress curl corrected for the Ekman transport.

The meridional transport M (y) is the Sverdrup

transport. A global map of the Sverdrup transport

integrated from the eastern to the western

boundary is shown in Figure 5.17. The size of

the integral at the western boundary gives the

western boundary current transport since

Sverdrup’s model must be closed with a narrow

boundary current that has at least one additional

physical mechanism beyond those in the

Sverdrup balance (a shift in latitude because of

water column stretching driven by Ekman transport

convergence). Physics of the boundary

currents are discussed in the following sections.

7.8.2. Stommel’s Solution: Westward

Intensification and Western Boundary

Currents

In the late 1940s, Henry Stommel (1948)

added simple linear friction to Sverdrup’s

model of the gentle interior flow in a basin

with eastern and western boundaries (Section

7.8.1). Mathematically this is an addition of

dissipation of potential vorticity Q on the

right-hand side of Eq. (7.37). The remarkable

result was that the returning flow can only be

in a narrow jet along the western boundary

(Figure S7.33). The potential vorticity balance

in this jet is change in planetary vorticity

balanced by bottom friction.

Figure S7.33a shows the ocean circulation if

there were no latitudinal variation in Coriolis

parameter (no b-effect; Stommel, 1965). This

is the solution if Earth were a rotating, flat

disk with westerlies in the north and trades

in the south. In this solution, the potential

vorticity input from the wind cannot be

balanced by a change in latitude, so the flow

builds up relative vorticity (negative sign)

that is balanced throughout the basin by

bottom friction; the Sverdrup balance (Eq.

7.40) cannot apply. In Figure S7.33b, for the

realistic spherical Earth with a b-effect, the

flow is southward throughout the interior

(Sverdrup balance), and returns northward in

a swift jet on the western boundary. This

idealized circulation resembles the Gulf

Stream and Kuroshio subtropical gyres in

which the Gulf Stream and Kuroshio are the

narrow western boundary currents returning

all southward Sverdrup interior flow back to

the north.

-10

0

+50

+100

+150

+10

40

30

20

10

1000 km

0

-10

1000 km

FIGURE S7.33 Stommel’s wind-driven circulation solution for a subtropical gyre with trades and westerlies like the

central latitudes of Figure S7.32: (a) surface height on a uniformly rotating Earth and (b) westward intensification with the

b-effect. After Stommel (1965).

56


Stommel’s frictional solution is somewhat

unrealistic since the friction is between the

boundary current and the ocean bottom. This

means that the wind-driven flow must reach

to the ocean bottom. However, with stratification,

it is not at all obvious that the circulation

reaches so deep (although in fact one characteristic

of strong western boundary currents such

as the Gulf Stream system is that the narrow

current does reach to the bottom even if the

Sverdrup interior flow does not). A subsequent

study by Walter Munk avoids this restriction

and still yields westward intensification, as

seen next.

7.8.3. Munk’s Solution: Western

Boundary Currents

A few years after Stommel’s work, Walter

Munk considered the effect of more realistic

friction on the ocean gyre circulations between

the currents and the side walls rather than

between the currents and the ocean bottom.

Munk’s (1950) result was very similar to

Stommel’s result, predicting westward intensification

of the circulation. A narrow, swift jet

along the western boundary returns the

Sverdrup interior flow to its original latitude

(Figure S7.34).

WEAK EASTERLIES

POLAR CURRENT

CYCLONES

SUBPOLAR GYRE

WESTERLY WINDS

(ROARING FORTIES)

WEST WIND DRIFT

SUBTROPICAL

ANTICYCLONES

Western current

Western boundary

vortices

SUBTROPICAL GYRE

Wind-spun vortex

Eastern current

TRADE WINDS

EQUATORIAL CURRENT

DOLDRUMS

EQUATORIAL COUNTER CURRENT

TRADE WINDS

EQUATORIAL CURRENT

EAST WEST

ZONAL WINDS

N

S

MERIDIONAL WINDS

FIGURE S7.34 Munk’s wind-driven circulation solution: zonal wind profiles on left and circulation streamlines in the

center. After Munk (1950).


How does the potential vorticity balance

work in Munk’s model (which is combined

with Sverdrup’s model)? Why do we find the

boundary current on the western side rather

than the eastern side, or even within the middle

of the basin (if considering Stommel’s bottom

friction)? In the Sverdrup interior of a subtropical

gyre, when the wind causes Ekman pumping,

the water columns are squashed, they

move equatorward to lower planetary vorticity.

To return to a higher latitude, there must be

forcing that puts the higher vorticity back into

the fluid. This cannot be in the form of planetary

vorticity or very, very narrow wind forcing,

since the first is already contained in the

Sverdrup balance, and the second is unphysical

except in one or two extremely special locations

(e.g., Arabian coast, Chapter 11). Therefore, the

input of vorticity must affect the relative

vorticity.

(a)

North

Western boundary (coastline)

Input

positive

relative

vorticity

Frictional

boundary

layer

Western boundary current

Frictional western boundary

layer (Munk, 1950): input of

positive relative vorticity allows

northward boundary current

(increasing planetary vorticity)

East

Southward interior

(Sverdrup) flow

FIGURE S7.35 (a) Vorticity

balance at a western boundary,

with side wall friction (Munk’s

model). (b) Hypothetical eastern

boundary vorticity balance,

showing that only western boundaries

can input the positive relative

vorticity required for the flow to

move northward.

(b)

What happens if the boundary current is on

the eastern boundary? Input of negative

relative vorticity cannot allow northward boundary

current. This solution is not permissible as a

balance for southward Sverdrup interior flow.

West

Southward interior

(Sverdrup) flow

Input

negative

relative

vorticity

Frictional

boundary

layer

Impermissible eastern boundary current

North

Eastern boundary (coastline)

58


Consider a western boundary current for

a Northern Hemisphere subtropical gyre

(Figure S7.35), with friction between the current

and the side wall (Munk’s model). The effect of

the side wall is to reduce the boundary current

velocity to zero at the wall. Therefore, the

boundary current has positive relative vorticity.

This vorticity is injected into the fluid by the friction

at the wall, and allows the current to move

northward to higher Coriolis parameter f. (Note

that there is negative relative vorticity in the

boundary current offshore of its maximum

speed, but the current changes much more

slowly and the negative relative vorticity there

is much lower than the positive relative vorticity

at the boundary.) On the other hand, if the

narrow jet returning flow to the north were on

the eastern boundary, the side wall friction

would inject negative relative vorticity, which

would make it even more difficult for the

boundary current fluid to join the interior flow

smoothly. Therefore, vorticity arguments

require that frictional boundary currents be on

the western boundary. The reader can go

through this exercise for subpolar gyres as

well as for both types of gyres in the Southern

Hemisphere and will find that a western

boundary current is required in all cases.

7.8.4. Fofonoff’s Solution: Large-Scale

Inertial Currents

In one further important simplified approach

to large-scale ocean circulation, Nicholas Fofonoff,

in 1954, showed that circulation can arise as

a free, unforced mode. The idea is that a very

small amount of wind, with very little friction

anywhere in the system, could set up such

a circulation. Indeed, aspects of the Fofonoff

solution are found in highly energetic regions,

such as in the neighborhood of the Gulf Stream

(which in actuality is not highly frictional, and

which is stronger than predicted from the

Sverdrup interior balance). This type of circulation

is called an “inertial circulation.” It is

easiest to describe using Fofonoff’s own figure

(Figure S7.36).

In the Fofonoff circulation, there is no

Sverdrup interior with flow moving northward

or southward. The interior flow is exactly zonal

(east-west). This is because there is no wind input

of vorticity, so flow cannot change latitude since

it would then have to change its planetary

vorticity. This exact zonality therefore results

from the b-effect. However, there are strong

boundary currents on both the western and

eastern boundaries, and there can be strong,

exactly zonal jets crossing the ocean in its interior.

How do these strong currents with so much

relative vorticity connect to each other?

Consider westward flow across the middle of

the ocean, as illustrated in Figure S7.36. This reaches

the western boundary and must somehow

get back to the eastern boundary to feed back

into the westward flow. It can do this by moving

along the western boundary in a very narrow

current that has a large amount of relative

vorticity. This current can be to either the north

or the south. Suppose it is to the north. Then the

relative vorticity of this frictionless current is

positive, allowing it to move to higher latitude.

It then jets straight across the middle of the

ocean, reaches the eastern boundary, and moves

southward, feeding into the westward flow in

the interior. There is no net input of vorticity

anywhere in this model (no wind, no friction).

Following the Sverdrup, Stommel, Munk,

and Fofonoff models, a number of theoretical

papers explored various combinations of the

different types of friction, inertia, and boundary

geometries on the mean ocean flow, but their

results can all be understood in terms of these

basic models. Some of the earliest ocean circulation

models (Veronis, 1966; Bryan, 1963) illustrated

the dynamical processes for various

strengths of friction and inertia. Further real

breakthroughs in theoretical understanding of

wind-driven ocean circulation occurred thirty

to forty years later, with treatment of the effect

of stratification, as discussed next.


N

W

E

1000 km

S

1000 km

FIGURE S7.36

Fofonoff (1954).

Inertial circulation, in the absence of friction and wind, but in the presence of the b-effect. Source: From

7.8.5. Wind-Driven Circulation in

a Stratified Ocean

What happens to the wind-driven circulation

theories in a stratified ocean? Water moves

down into the ocean, mostly along very gradually

sloping isopycnals. Where streamlines of flow are

connected to the sea surface, we say the ocean is

directly ventilated (Figure S7.37). Where there is

Ekman pumping (negative wind stress curl), the

60


Sverdrup interior flow is equatorward (Section

7.8.1). Water columns at the local mixed layer

density move equatorward and encounter less

dense water at the surface. They slide down into

the subsurface along isopycnals, still moving

equatorward. This process is called subduction

(Luyten, Pedlosky, & Stommel, 1983), using

a term borrowed from plate tectonics. The subducted

waters then flow around the gyre and

enter the western boundary current if they do

not first enter the tropical circulation. The

details of this process are beyond the scope of

this text.

In each subducted layer, there can be three

regions (Figure S7.37): (1) a ventilated region connected

from the sea surface as just described, (2)

a western unventilated pool with streamlines

that enter and exit from the western boundary

current without entering the surface layer, and

(3) an eastern quiet (shadow) zone between the

easternmost subducting streamline and the

eastern boundary. A continuous range of

surface densities is found in the subtropical

gyre; the water column is directly ventilated

over this full range, with waters at each density

coming from a different sea-surface location

FIGURE S7.37 (a) Subduction

schematic (Northern Hemisphere).

(b) Streamlines for idealized

subduction on an isopycnal

surface. The light gray regions are

the western pool and eastern

shadow zone, where streamlines

do not connect to the sea surface.

The heavy dashed contour is

where the isopycnal meets the sea

surface (surface outcrop); in the

dark gray area there is no water of

this density. After Williams (1991).

N

W

(a)

WESTERN

UNVENTILATED

POOL

WIND

SURFACE

OUTCROP

SUBDUCTED

REGION

EASTERN BOUNDARY

WIND

EASTERN

SHADOW

ZONE

WESTERN BOUNDARY

ABYSSAL

OCEAN

(b)

40

Surface outcrop

Latitude (degree)

Western

Pool

30

20

Ventilated

region

–60 –50 –40 –30 –20

Longitude (degree)

Eastern

shadow

zone

WIND-DRIVEN CIRCULATION: EASTERN BOUNDARY CURRENTS AND EQUATORIAL CIRCULATION 61

depending on the configuration of streamlines

on that isopycnal. This is called the “ventilated

thermocline”; in water mass terms, this process

creates the Central Water. The maximum

density of the ventilated thermocline is set by

the maximum winter surface density in the

subtropical gyre (Stommel, 1979). This usually

occurs at the most poleward edge of the gyre,

around 40 to 50 degrees latitude. The

maximum depth of the ventilated thermocline

is the depth of this densest isopycnal, and is

between 500 and 1000 m depending on the

ocean (see Chapters 9e11).

Subducting waters can leave the surface layer

in two distinct ways: they can be pushed downward

along isopycnals by Ekman pumping, and

they can also be included in the subsurface layer

through seasonal warming and cooling of the

surface layer while they flow southward. In

winter the surface layer is of uniform density.

Entering spring and summer, this is glazed

over by a surface layer of much lower density.

All the while the geostrophic flow is southward.

When the next winter arrives, the water column

is farther south and winter cooling does not

penetrate down to it. Therefore it has effectively

entered the subsurface flow and does not

re-enter the surface layer until it emerges from

the western boundary, possibly many years

later. Therefore the properties of the subsurface

flows are set by the late winter conditions. The

other seasons have no impact other than to

provide seasonal isolation of the winter layer

until it has subducted. Stommel (1979) called

this phenomenon the “Ekman demon,” analogous

to Maxwell’s demon of thermodynamics,

which is a thought experiment about separating

higher and lower energy molecules.

The opposite of subduction is obduction, borrowed

again from plate tectonics by Qiu and

Huang (1995). In obducting regions, waters

from subsurface isopycnals come up and into

the surface layer. These are generally upwelling

regions such as the cyclonic subpolar gyres and

the region south of the ACC.

Wind-driven circulation occurs in unventilated

stratified regions as well. It is most

vigorous in regions connected to the western

boundary currents where water can enter and

exit the western boundary. In these regions,

the western boundary currents and their separated

extensions usually reach to the ocean

bottom. In a region that is closer and closer to

the western boundary with increasing depth,

there can be a closed circulation region that

connects in and out of the western boundary

without connection to the sea surface; such

regions are characterized by constant potential

vorticity (stretching and planetary portions

only, or f/H). These dynamics are beyond the

scope of this text.

7.9. WIND-DRIVEN

CIRCULATION: EASTERN

BOUNDARY CURRENTS AND

EQUATORIAL CIRCULATION

7.9.1. Coastal Upwelling and Eastern

Boundary Currents

The eastern boundary regions of the subtropical

gyres have strong but shallow flow that is

dynamically independent of the open ocean

gyre regimes. Upper ocean eastern boundary

circulation is driven by alongshore wind stress

that creates onshore (or offshore) Ekman transport

that creates upwelling (or downwelling;

Section 7.5.4). Beneath or inshore of the equatorward

eastern boundary currents there is a poleward

undercurrent or countercurrent. Coastal

upwelling systems are not restricted to eastern

boundaries; the southern coast of the Arabian

peninsula has the same kind of system. These

circulations are fundamentally different from

western boundary currents, which are tied to

potential vorticity dynamics (Section 7.8).

The classical explanation of eastern boundary

currents is that equatorward winds force Ekman

flow offshore, which drives a shallow upwelling

62


(on the order of 200 m deep) in a very narrow

region adjacent to the coast (on the order of

10 km; Figure S7.13c). The upwelling speed is

about 5e10 m/day. Because of stratification,

the source of upwelled water is restricted to

layers close to the sea surface, usually between

50 and 300 m.

The zone of coastal upwelling can be

extended to more than 100 km offshore by an

increase in longshore wind strength with

distance offshore; this is observed in each

eastern boundary upwelling system due to

topographic steering of the winds by the

oceaneland boundary. The offshore Ekman

transport therefore increases with distance

offshore, which requires upwelling through

the whole band (Bakun & Nelson, 1991). The

zone is identified by positive wind stress curl,

notably in the California Current and Peru-

Chile Current regions and the Arabian

upwelling zone (Figure 5.16d).

Upwelled water is cooler than the original

surface water. It originates from just below the

euphotic zone and therefore is also rich in nutrients,

which results in enhanced biological

productivity characterized by high chlorophyll

content (Section 4.6). Cool surface temperatures

and enhanced biological productivity are clear

in satellite images that record sea-surface

temperature and ocean color (Figure 4.28).

Upwelling is strongly seasonal, due to seasonality

in the winds. Onset of upwelling can

be within days of arrival of upwelling-favorable

winds. In one example, off the coast of Oregon,

the surface temperature dropped by 6 C in two

days after a longshore wind started.

Coastal upwelling is accompanied by a rise in

upper ocean isopycnals toward the coast (Figure

7.6). This creates an equatorward geostrophic

surface flow, the eastern boundary current. These

currents are narrow (<100 km width and near

the coast), shallow (upper 100 m), strong

(40e80 cm/sec), and strongly seasonal. The

actual flow in an eastern boundary current

system includes strong, meandering eddies

and offshore jets/filaments of surface water,

often associated with coastline features such as

capes (Figure 10.6). Actual eastern boundary

currents are some distance offshore at the axis

of the upwelling front created by the offshore

Ekman transport.

Poleward undercurrents are observed at about

200 m depth beneath the equatorward surface

currents in each eastern boundary upwelling

system. When upwelling-favorable winds

weaken or disappear, the equatorward flow

also disappears and the poleward undercurrent

extends up to the surface (there is no longer an

undercurrent). Poleward undercurrents are

created mainly by the alongshore pressure

gradient that drives the onshore subsurface

geostrophic flow that feeds the upwelling. There

may also be a contribution from positive wind

stress curl throughout the eastern boundary

region that leads to poleward Sverdrup transport

(Section 7.8.1; Hurlburt & Thompson,

1973).

The only ocean without an equatorward

eastern boundary current is the Indian Ocean.

The Leeuwin Current along the west coast of

Australia flows poleward, even though the

winds are upwelling favorable and would drive

a normal eastern boundary current there in the

absence of other forces. However, there is

a much larger poleward pressure gradient force

along this boundary than along the others, due

to the flow of water westward through the Indonesian

archipelago from the Pacific to the Indian

Ocean.

7.9.2. Near-Surface Equatorial

Currents and Bjerknes Feedback

Circulation within about 2 degrees latitude of

the equator is very different from non-equatorial

circulation because the Coriolis parameter

f vanishes at the equator. The narrowness of

this equatorial influence, that is, the equatorial

baroclinic deformation radius, is set by the variation

in Coriolis parameter with latitude and the


stratification of the ocean. Equatorial circulation

is driven by easterly trade winds in the Pacific

and Atlantic and by the seasonally reversing

monsoonal winds in the Indian Ocean. We

describe here only the equatorial circulation

that results from trade winds.

Since the Coriolis parameter vanishes and

there is no frictional Ekman layer, the easterly

trade winds drive equatorial surface flow due

westward in a frictional surface layer (Figure

S7.38a). The westward surface current is

shallow (50 to 100 m) and of medium strength

(10 to 20 cm/sec). In each of the three oceans,

this westward surface flow is a part of the

South Equatorial Current. The water piles up

gently in the west (to about 0.5 m height) and

leaves a depression in the east. This creates

an eastward pressure gradient force (from

high pressure in the west to low pressure in

the east). The pressure gradient force drives

an eastward flow called the Equatorial Undercurrent

(EUC). The EUC is centered at 100 to

200 m depth, just below the frictional surface

layer. The EUC is only about 150 m thick. It is

among the strongest ocean currents

(>100 cm/sec). (See illustrations of the Pacific

EUC in Section 10.7.3 and of the Atlantic EUC

in Section 9.4.)

The pileup of waters in the western equatorial

region results in a deepened pycnocline

called the warm pool and a shallow pycnocline

in the eastern equatorial region. Coriolis effects

become important a small distance from the

equator; the resulting off-equatorial Ekman

transport enhances upwelling in the equatorial

band. This creates upwelling along the equator

and shoaling of the pycnocline toward the

equator that drives a westward, nearly

geostrophic flow at the sea surface. This

broadens the frictional westward flow found

right on the equator.

Upwelling in the eastern equatorial region

draws cool water to the surface because of the

shallow thermocline there. This creates a cold

surface feature along the equator called the cold

tongue (see the sea-surface temperature map in

Figure 4.1). Because of the thickness of the

warm pool in the west, even intense upwelling

cannot cause cold surface temperatures. The

warm pool’s high surface temperature, in excess

of 28 C, is maintained through radiative equilibrium

with the atmosphere (Jin, 1996).

The east-west contrast in temperature along

the equator maintains the atmosphere’s Walker

circulation, which has ascending air over the

warm pool and descending over the eastern

colder area. The Walker circulation is an important

part of the trade winds that creates the

warm pool and cold tongue, so there can be

a feedback between the ocean and atmosphere;

this is called the Bjerknes feedback (Bjerknes,

1969; Figure S7.38b). If something weakens the

trade winds, as at the beginning of an El Niño

event (Chapter 10), the westward flow at the

equator weakens and upwelling weakens or

stops. Surface waters in the eastern regions

therefore warm. Water in the deep warm pool

in the west sloshes eastward along the equator,

thinning the pool. The change in sea-surface

temperature weakens the Walker circulation/

trade winds even more, which further exacerbates

the ocean changes. This is an example of

a positive feedback.

In the Indian Ocean, the prevailing equatorial

winds are monsoonal, meaning that trade winds

are only present for part of the year. This creates

seasonally reversing equatorial currents and

inhibits the formation of the warm pool/cold

tongue structure. The Indian Ocean sea-surface

temperature is high at all longitudes.

7.10. BUOYANCY

(THERMOHALINE) FORCING AND

ABYSSAL CIRCULATION

Heating and cooling change the ocean’s

temperature distribution, while evaporation,

precipitation, runoff, and ice formation change

the ocean’s salinity distribution (Chapters 4

64


(a)

Normal Conditions

Convective

Circulation

Equator

Thermocline

(b)

120ϒE

80ϒW

Bjerknes tropical feedback

Trade wind strength

Equatorial upwelling.

High sea level and deep

thermocline in west.

Low sea level and

shallow thermocline in east.

+ +

Zonal tropical SST difference

(+ positive feedback)

FIGURE S7.38 (a) Schematic of upper ocean equatorial circulation (large white arrows), surface temperature (red is

warm, blue is cold), and thermocline depth and upwelling, driven by the Walker circulation (“convective loop”). Source:

From NOAA PMEL (2009b). (b) “Bjerknes feedback” between the trade wind strength and zonal (east-west) difference in

tropical surface temperature. (Arrows mean that increase in one parameter results in an increase in the second parameter.) In

this positive feedback loop, increased trade winds cause a larger sea-surface temperature difference, which in turn increases

the trade wind strength.

and 5). Collectively, these are referred to as

buoyancy, or thermohaline, forcing. Buoyancy

processes are responsible for developing the

ocean’s stratification, including its abyssal

properties, pycnocline, thermocline, halocline,

and upper layer structure (other than in windstirred

mixed layers). Advection by currents

also changes temperature and salinity locally,

but it cannot change the overall inventory of

either.

Abyssal circulation refers to the general category

of currents in the deep ocean. The overturning

circulation, also called the thermohaline

circulation, is the part of the circulation associated

with buoyancy changes, and overlaps

spatially with the wind-driven upper ocean

circulation; it also includes shallow elements

that are independent of the abyssal circulation.

In the overturning circulation, cooling and/or

salinification at the sea surface causes water to

sink. This water must rise back to the warm

surface, which requires diffusion of heat (buoyancy)

downward from the sea surface. The

source of eddy diffusion is primarily wind and

tidal energy. Thus aspects of the thermohaline

circulation depend on the magnitude of nonbuoyancy

processes through the eddy diffusivity

(Wunsch & Ferrari, 2004).

Studies of the overturning circulation

originated in the 1800s and early 1900s with

German, British, and Norwegian oceanographers.

J. Sandström (1908) presented experiments

and ideas about the simplest overturning

cells driven by high-latitude cooling and a deep


tropical warm source that we now identify with

downward heat diffusion (see Figure S7.40).

H. Stommel, in the 1960s, produced a series of

elegant papers on abyssal flow driven by isolated

sources of deep water and broad scale upwelling

that returns the water back to the upper ocean

(Section 7.10.2). At the same time, Stommel presented

simple theories of the complementary

idea of ocean flows driven by very large-scale

density contrasts (warm, saline tropics and

cold, fresh poles: Section 7.10.3).

7.10.1. Buoyancy Loss Processes

(Diapycnal Downwelling)

Water becomes denser through net cooling,

net evaporation, and brine rejection during

sea ice formation. We have already described

brine rejection (Section 3.9.2); it is responsible

for creating the densest bottom waters in the

global ocean (Antarctic Bottom Water and parts

of the Circumpolar Deep Water) and also in the

regional basins where it is operative (Arctic

Ocean, Japan Sea, etc.). Here we focus on

convection created by net buoyancy loss in the

open ocean, when surface water becomes

denser than water below, and advects and

mixes downward. Convection creates a mixed

layer, just like wind stirring (Section 7.3).

However, a convective mixed layer can be

hundreds of meters thick by the end of winter,

whereas a wind-stirred mixed layer is limited

to about 150 m by the depth of wind-driven

turbulence.

Convection happens on different timescales.

Diurnal (daily) convection occurs at night in

areas where the surface layer restratifies

strongly during the day. During the annual

cycle, cooling usually starts around the

autumnal equinox and continues almost until

the spring equinox. The resulting convection

eats down into the surface layer, reaching

maximum depth and density at the end of

winter when the cumulative cooling reaches its

maximum (FebruaryeMarch in the Northern

Hemisphere and AugusteSeptember in the

Southern Hemisphere).

Ocean convection is usually driven by

surface cooling. Excess evaporation can also

create convection, but the latent heat loss associated

with evaporation is usually stronger.

“Deep” convection is a loose term that usually

refers to creation of a surface mixed layer that

is thicker than about 1000 m. Deep convection

has three phases: (1) preconditioning (reduction

in stratification), (2) convection (violent

mixing), and (3) sinking and spreading. Preconditioning

for deep convection includes

reduced stratification through the water

column and some sort of dynamical feature

that allows stratification to become even

more. The convection phase occurs when there

is large heat loss, usually due to high wind

speeds along with very cold, dry air usually

blowing from the land. Adjustment or restratification,

followed by spreading, occurs as the

convective features collapse (Killworth, 1983;

Marshall & Schott, 1999.)

Convective regions have a typical structure

(Figure S7.39). These include: (1) a chimney,

which is a patch of tens to more than hundreds

of kilometers across within which preconditioning

can allow convection and (2) convective

plumes that are the actual sites of

convection and are about 1 km or less across

(Killworth, 1979; Marshall & Schott, 1999).

Chimney (50-100 km)

Eddies (~10 km)

Mixed water

Stratified water

Plumes (< 1 km)

FIGURE S7.39 Processes in a deep convection region.

After Marshall and Schott (1999).

66


The plumes are about the same size across as

they are deep. It is not yet clearly known

what the vertical velocity structure is within

the convective plumes. Observations in the

Labrador Sea have suggested that there is

more downward motion than upward motion,

and that the required upward motion might

occur more slowly over a broader area within

the chimney.

Deep convection occurs only in a few special

locations around the world: Greenland Sea, Labrador

Sea, Mediterranean Sea, Weddell Sea,

Ross Sea, and Japan (or East) Sea. These sites,

with the exception of the isolated Japan Sea,

ventilate most of the deep waters of the global

ocean. (The denser bottom waters, particularly

in the Southern Hemisphere, result from the

brine rejection process around Antarctica.)

7.10.2. Diapycnal Upwelling

(Buoyancy Gain)

The structure of the basin and global scale

overturning circulations depends on both the

amount of density increase in the convective

source regions and the existence of a buoyancy

(heat) source at lower latitudes that is at least

as deep as the extent of the cooling (Sandström,

1908; Figure S7.40). Since there are no significant

local deep heat sources in the world ocean,

waters that fill the deep ocean can only return

to the sea surface as a result of diapycnal eddy

diffusion of buoyancy (heat and freshwater)

Up

Heating

Warming through

diffusion

Cooling

Equatorward

Poleward

FIGURE S7.40 The role of vertical (diapycnal) diffusion

in the MOC, replacing Sandström’s (1908) deep tropical

warm source with diapycnal diffusion that reaches below

the effect of high latitude cooling.

downward from the sea surface (Sections 5.1.3

and 7.3.2).

Munk’s (1966) diapycnal eddy diffusivity

estimate of k v ¼ 1 10 4 m 2 /sec (Section 7.3.2)

was based on the idea of isolated sources of

deep water and widespread diffusive upwelling

of this deep water back to the surface. From all

of the terms in the temperature and salt equations

(7.12 7.13), Munk assumed that most of

the ocean is dominated by the balance

vertical advection ¼ vertical diffusion (7.46a)

w vT=vz ¼ v=vzðk V vT=vzÞ

(7.46b)

Munk obtained his diffusivity estimate from an

average temperature profile and an estimate of

about 1 cm/day for the upwelling velocity w,

which can be based on deep-water formation

rates and an assumption of upwelling over the

whole ocean. The observed diapycnal eddy

diffusivity in the open ocean away from boundaries

is an order of magnitude smaller than

Munk’s estimate, which must be valid for the

globally averaged ocean structure. This means

that there must be much larger diffusivity in

some regions of the ocean d now thought to be

at the boundaries d at large seamount and island

chains, and possibly the equator (Section 7.3).

7.10.3. Stommel and Arons’ Solution:

Abyssal Circulation and Deep Western

Boundary Currents

Deep ocean circulation has been explained

using potential vorticity concepts that are very

familiar from Sverdrup balance (Section 7.8.1).

Stommel (1958), Stommel, Arons, and Faller

(1958), and Stommel & Arons (1960a,b) considered

an ocean with just two layers, and solved

only for the circulation in the bottom layer.

They assumed a source of deep water at the

northernmost latitude, and then assumed that

this water upwells uniformly (at the same rate)

everywhere (Figure S7.41). This upwelling

stretches the deep ocean water columns.


(a)

S 0

φ 1 φ 2

Equator

(b)

FIGURE S7.41 (a) Abyssal circulation model. After Stommel and Arons (1960a). (b) Laboratory experiment results looking

down from the top on a tank rotating counterclockwise around the apex (So) with a bottom that slopes towards the apex.

There is a point source of water at So. The dye release in subsequent photos shows the Deep Western Boundary Current, and

flow in the interior Si beginning to fill in and move towards So. Source: From Stommel, Arons, & Faller (1958).

Stretching requires a poleward shift of the water

columns to conserve potential vorticity (Eq.

7.35). The predicted interior flow is therefore

counterintuitive d it runs toward the deepwater

source. (Actual abyssal flow is strongly

modified from this by the major topography

that modifies the b-effect by allowing stretched

columns to move toward shallower bottoms

rather than toward higher latitude.)


connect the isolated deep-water sources and

the interior poleward flows. Whereas unambiguous

poleward flow is not observed in the deep

ocean interior (possibly mostly because of

topography), DWBCs are found where they

are predicted to occur by the Stommel and

Arons abyssal circulation theory (Warren,

1981). One such DWBC runs southward beneath

the Gulf Stream, carrying dense waters from the

Nordic Seas and Labrador Sea. Swallow and

Worthington (1961) found this current after

being convinced by Stommel to go search for

68


it. Maps from the 1920s Meteor expedition

(Wüst, 1935) show the large-scale consequences

of this particular DWBC for deep salinity and

oxygen distributions over the whole length of

the Atlantic (see Chapter 9). Stommel’s (1958)

map (Figure S7.42), revisited later by Kuo and

Veronis (1973) using a numerical ocean model,

shows the conceptual global pattern of DWBCs

and abyssal circulation, including the deepwater

source in the northern North Atlantic

(the source of North Atlantic Deep Water,

Chapter 9) and in the Antarctic (the source of

Antarctic Bottom Water, Chapter 13).

7.10.4. Thermohaline Oscillators:

Stommel’s solution

An entirely different approach to the MOC

from the Stommel-Arons abyssal circulation

model considers changes in overturn associated

with changing rates of dense water production.

The prototype of these models is a very simple

reduction of the ocean to just a few boxes, and

was also developed by Stommel (1961), who

can be appreciated at this point as a giant of

ocean general circulation theory. Such box

models show how even the simplest model of

climate change, for example, can lead to

complex results. In this case, multiple equilibria

result, that is, the system can jump suddenly

between quite different equilibrium states.

Stommel (1961) reduced the ocean to two

connected boxes representing dense, cold,

fresh high latitudes and light, warm, saltier

low latitudes (Figure S7.43). The boxes are connected,

with the amount of flow between them

dependent on the density difference between

the boxes. (This is a simplification of sinking

of dense water to the bottom and flowing

toward a region of lower bottom density, to

be fed in turn by upwelling in the lower

density box, and return flow at the sea surface.)

In each box, the temperature and salinity are

set by (1) flux of water between the boxes (thermohaline

circulation) that depends on the

density difference between the boxes and (2)

restoring temperature and salinity to a basic

state over some set time period. Then the

effects on the flow between the boxes of slow

heating and cooling, or of freshwater fluxes

(evaporation and precipitation for instance),

are studied.

Stommel (1961) found that several different

thermohaline circulation strengths exist for

a given set of choices of model parameters

(externally imposed temperature and salinity,

FIGURE S7.42 Global abyssal

circulation model, assuming two

deep water sources (filled circles

near Greenland and Antarctica).

Source: From Stommel (1958).


(a)

Warming

Evaporation

Cooling

Freshening

surface flow

T L

, S L

T H

, S H

bottom flow

Low latitudes

High latitudes

(b)

North Atlantic SST

1

2

North Atlantic SST

1

3

2

High latitude Freshwater input

High latitude Freshwater input

FIGURE S7.43 (a) Schematic of the Stommel (1961) two-box model of the meridional overturning circulation. The

direction of the arrows assumes that the higher latitude box (blue) has higher density water. Each box is well mixed. (b)

Schematic of the hysteresis in North Atlantic sea-surface temperature resulting from hysteresis in MOC strength. The

starting point in freshwater is denoted by 1; starting at lower freshwater, hence higher salinity, in the left panel. Freshening is

denoted by the blue arrow, with the same total amount in both panels. Salinification is denoted by red arrow, and should be

exactly opposite to the freshwater arrow. In the left panel, starting at higher salinity, the freshening allows the system to

remain on the top branch, and so subsequent evaporation returns the system to original state. In the right panel, with

a fresher starting point, the same freshening causes transition to lower curve (2), and subsequent evaporation returns system

to a different state, denoted by 3. After Stocker and Marchal (2000).

restoration timescales for temperature and

salinity, and factor relating the flow rate to the

density difference between the boxes). As the

basic state was slowly changed, perhaps by

reduction of the basic high-latitude salinity

(which reduces its density), the flow rate slowly

changed and then suddenly jumped to

a different equilibrium rate. When the basic

state salinity was then slowly increased, the

system jumped back to a higher flow rate but

70


at a very different basic salinity than during its

decreasing phase. Thus this system exhibits

hysteresis: it has different equilibrium states

depending on whether the state is approached

from a much higher salinity or a much lower

salinity.

The coupled atmosphere-sea-ice-landphysics-biology-chemistry

climate system is

far more complex than the two simple boxes in

this very simple Stommel oscillator model. Yet

its multiple equilibria and hysteresis behavior

have been useful in demonstrating the potential

for abrupt and relatively large changes in

climate and, more specifically, for interpretation

of numerical models of the changes in overturning

circulation that could result from changes in

external forcing.

References

Armi, L., 1978. Some evidence for boundary mixing in the

deep ocean. J. Geophys. Res. 83, 1971e1979.

Assaf, G., Gerard, R., Gordon, A., 1971. Some mechanisms

of oceanic mixing revealed in aerial photographs.

J. Geophys. Res. 76, 6550e6572.

Bakun, A., Nelson, C.S., 1991. The seasonal cycle of windstress

curl in subtropical eastern boundary current

regions. J. Phys. Oceanogr. 21, 1815e1834.

Bjerknes, J., 1969. Atmospheric teleconnections from the

equatorial Pacific. Mon. Weather Rev. 97, 163e172.

Boccaletti, G., Ferrari, R., Fox-Kemper, B., 2007. Mixed layer

instabilities and restratification. J. Phys. Oceanogr. 37,

2228e2250.

Brink, K.H., 2005. Coastal physical processes overview. In:

Robinson, A.F., Brink, K.H. (Eds.), The Sea, Vol. 13: The

Global Coastal Ocean: Multiscale interdisciplinary

Processes. Harvard University Press, Cambridge, MA,

pp. 37e60.

Bryan, K., 1963. A numerical investigation of a nonlinear

model of a wind-driven ocean. J. Atm. Sci. 20, 594e606.

Chelton, D.B., deSzoeke, R.A., Schlax, M.G., El Naggar, K.,

Siwertz, N., 1998. Geographical variability of the first

baroclinic Rossby radius of deformation. J. Phys.


Chereskin, T.K., 1995. Direct evidence for an Ekman balance

in the California Current. J. Geophys. Res. 100,

18261e18269.

Cushman-Roisin, B., 1994. Introduction to Geophysical Fluid

Dynamics. Prentice Hall, Englewood Cliffs, N.J., 320 p.

d’Asaro, E.A., Eriksen, C.C., Levine, M.D., Paulson, C.A.,

Niiler, P., Van Meurs, P., 1995. Upper-ocean inertial currents

forced by a strong storm. Part 1: Data and comparisons

with linear theory. J. Phys. Oceanogr. 25, 2909e2936.

Davis, R.E., deSzoeke, R., Niiler, P., 1981. Variability in the

upper ocean during MILE. Part II: Modeling the mixed

layer response. Deep-Sea Res. 28A, 1453e1475.

Doron, P., Bertuccioli, L., Katz, J., Osborn, T.R., 2001.

Turbulence characteristics and dissipation estimates in

the coastal ocean bottom boundary layer from PIV data.


Egbert, G.D., Ray, R., 2001. Estimates of M2 tidal energy

dissipation from TOPEX/Poseidon altimeter data. J.

Geophys. Res. 106, 22475e22502.

Ekman, V.W., 1905. On the influence of the Earth’s rotation

on ocean currents. Arch. Math. Astron. Phys. 2 (11),

1e53.

Eriksen, C.C., 1982. Geostrophic equatorial deep jets. J. Mar.

Res. 40 (Suppl), 143e157.

Fofonoff, N.P., 1954. Steady flow in a frictionless homogeneous

ocean. J. Mar. Res. 13, 254e262.

Gent, P.R., McWilliams, J.C., 1990. Isopycnal mixing in ocean

circulation models. J. Phys. Oceanogr. 20, 150e155.

Gill, A.E., 1982. Atmospheric-Ocean Dynamics. Academic

Press, New York, 662 p.

Gill, A.E., Niiler, P., 1973. The theory of seasonal variability

in the ocean. Deep-Sea Res. 20, 141e177.

Gregg, M.C., 1987. Diapycnal mixing in the thermocline:

a review. J. Geophys. Res. 94, 5249e5286.

Huang, R.-X., 2010. Ocean Circulation: Wind-driven and

Thermohaline Processes. Cambridge University Press,

Cambridge, UK, 806 p.

Hurlburt, H.E., Thompson, J.D., 1973. Coastal upwelling on

a b-plane. J. Phys. Oceanogr. 19, 16e32.



Science 274, 76e78.




Killworth, P.D., 1979. On chimney formation in the ocean.


Killworth, P.D., 1983. Deep convection in the world ocean.

Rev. Geophys. 21, 1e26.

Knauss, J.A., 1997. Introduction to Physical Oceanography,

second ed., Waveland Press, Long Grove, IL, 309 pp.

Kraus, E.B., Turner, J.S., 1967. A one-dimensional model of

the seasonal thermocline, II. The general theory and its

consequences. Tellus 19, 98e105.

Kunze, E., Firing, E., Hummon, J.M., Chereskin, T.K.,

Thurnherr, A.M., 2006. Global abyssal mixing Inferred

from Lowered ADCP shear and CTD strain profiles.


REFERENCES 71

Kuo, H.-H., Veronis, G., 1973. The use of oxygen as a test

for an abyssal circulation model. Deep-Sea Res. 20,

871e888.

Langmuir, I., 1938. Surface motion of water induced by

wind. Science 87, 119e123.

Large, W.G., McWilliams, J.C., Doney, S.C., 1994. Oceanic

vertical mixing: A review and a model with a non-local

K-profile boundary layer parameterization. Rev.

Geophys. 32, 363e403.


slow mixing across the pycnocline from an openeocean

tracererelease experiment. Nature 364, 701e703.

Ledwell, J.R., Watson, A.J., Law, C.S., 1998. Mixing of

a tracer in the pycnocline. J. Geophys. Res. 103,

21499e21529.

Lentz, S.J., 1995. Sensitivity of the inner-shelf circulation to

the form of the eddy viscosity profile. J. Phys. Oceanogr.

25, 19e28.

Levitus, S., 1988. Ekman volume fluxes for the world ocean

and individual ocean basins. J. Phys. Oceanogr. 18,

271e279.

Levitus, S., Boyer, T.P., Antonov, J., 1994a. World ocean

atlas: Volume 5: Interannual variability of upper ocean

thermal structures. NOAA/NESDIS, Tech. Rpt., OSTI

ID: 137204.

Lien, R.-C., Gregg, M.C., 2001. Observations of turbulence in

a tidal beam and across a coastal ridge. J. Geophys. Res.

106, 4575e4591.

Luyten, J.R., Pedlosky, J., Stommel, H., 1983. The ventilated

thermocline. J. Phys. Oceanogr. 13 292e309.

Marshall, J., Schott, F., 1999. Open-ocean convection:

observations, theory, and models. Rev. Geophys. 37,

1e64.

Munk, W., 1966. Abyssal recipes. Deep-Sea Res. 13,

707e730.

Munk, W.H., 1950. On the wind-driven ocean circulation.

J. Atm. Sci. 7, 80e93.

Nansen, F., 1922. In: Brodhaus, F.U. (Ed.), Nacht und Eis.

Leipzig, Germany, 355 pp. (in German).

NOAA National Weather Service, 2005. Hydrometeorological

Prediction Center (HPC) Home Page. National

Weather Service. http://www.hpc.ncep.noaa.gov/

(accessed 1.3.05).

NOAA PMEL TAL Project Office, 2009b. El Niño theme

page: access to distributed information on El Niño.

NOAA Pacific Marine Environmental Laboratory.

http://www.pmel.noaa.gov/tao/elnin/nino-home.html

(accessed 3.26.09).

Olbers, D.J.M., Wenzel, Willebrand, J., 1985. The inference of

North Atlantic circulation patterns from climatological

hydrographic data. Rev. Geophys. 23, 313e356.

Osborn, T.R., Cox, C.S., 1972. Oceanic fine structure. Geophys.

Astrophys. Fluid Dyn. 3, 321e345.

Pedlosky, J., 1987. Geophysical Fluid Dynamics, second ed.),

Springer-Verlag, New York, 732 p.

Polton, J.A., Smith, J.A., MacKinnon, J.A., Tejada-

Martínez, A.E., 2008. Rapid generation of highfrequency

internal waves beneath a wind and wave

forced oceanic surface mixed layer. Geophys. Res. Lett.

35, L13602. doi:10.1029/2008GL033856.

Polzin, K.L., Toole, J.M., Ledwell, J.R., Schmitt, R.W., 1997.

Spatial variability of turbulent mixing in the abyssal

ocean. Science 276, 93e96.

Pond, S., Pickard, G.L., 1983. Introductory Dynamical

Oceanography, second ed., Pergamon Press, Oxford.

329 p.


production by marginal seas. Progr. Oceanogr. 33,

161e200.

Price, J.F., Weller, R.A., Pinkel, R., 1986. Diurnal cycling:

Observations and models of the upper ocean response to

diurnal heating, cooling and wind mixing. J. Geophys.

Res. 91, 8411e8427.

Qiu, B., Huang, R.X., 1995. Ventilation of the North Atlantic

and North Pacific: Subduction versus obduction. J. Phys.

Oceanogr. 25, 2374e2390.

Ralph, E.A., Niiler, P.P., 1999. Wind-driven currents in the

tropical Pacific. J. Phys. Oceanogr. 29, 2121e2129.

Redi, M.H., 1982. Oceanic isopycnal mixing by coordinate

rotation. J. Phys. Oceanogr. 12, 1154e1158.





Pacific Ocean: Flow patterns, tracers and transports.


Rudnick, D.L., Boyd, T.J., Brainard, R.E., Carter, G.S.,

Egbert, G.D., Gregg, M.C., Holloway, P.E., Klymak, J.M.,

Kunze, E., Lee, C.M., Levine, M.D., Luther, D.S.,

Martin, J.P., Merrifield, M.A., Moum, J.N., Nash, J.D.,

Pinkel, R., Rainville, L., Sanford, T.B., 2003. From

tides to mixing along the Hawaiian Ridge. Science 301,

355e357.

Salmon, R., 1998. Lectures on Geophysical Fluid Dynamics.

Oxford University Press, New York, 378 p.

Sandström, J., 1908. Dynamische Versuche mit Meerwasser.

Annalen der Hydrographie und Maritimen Meteorologie,

pp. 6e23 (in German).

Simpson, J.H., 1998. Tidal processes in shelf seas. In:

Brink, K.H., Robinson, A.R. (Eds.), The Sea, Vol. 10: The

Global Coastal Ocean: Processes and Methods. Harvard


Smith, J.A., 2001. Observations and theories of Langmuir

circulation: a story of mixing. In: Lumley, J.L. (Ed.), Fluid

Mechanics and the Environment: Dynamical

Approaches. Springer, New York, pp. 295e314.

72


Smith, R.D., Maltrud, M.E., Bryan, F.O., Hecht, M.W., 2000.

Numerical simulation of the North Atlantic Ocean at

1/10 . J. Phys. Oceanogr. 30, 1532e1561.

Stewart, R.H., 2008. Introduction to Physical Oceanography.

Open-source textbook. http://oceanworld.tamu.edu/

ocean410/ocng410_text_book.html (accessed 3.28.09).

Stocker, T.F., Marchal, O., 2000. Abrupt climate change in

the computer: Is it real? Proc. Natl. Acad. Sci., USA 97l,

1362e1365.

Stommel, H., 1948. The westward intensification of winddriven

currents. Trans. Am. Geophys. Union 29, 202e206.

Stommel, H.M., 1958. The abyssal circulation. Deep-Sea Res.

5, 80e82.

Stommel, H.M., 1961. Thermohaline convection with two

stable regimes of flow. Tellus 13, 224e230.

Stommel, H.M., 1965. The Gulf Stream: A Physical and

Dynamical Description, second ed., University of

California Press, Berkeley, and Cambridge University

Press, London, 248 p.

Stommel, H.M., 1979. Determination of water mass properties

of water pumped down from the Ekman layer to

the geostrophic flow below. Proc. Nat. Acad. Sci., USA

76, 3051e3055.

Stommel, H.M., Arons, A., 1960a. On the abyssal circulation

of the World Ocean d I. Stationary planetary flow

patterns on a sphere. Deep-Sea Res. 6, 140e154.

Stommel, H.M., Arons, A., 1960b. On the abyssal circulation

of the World Ocean d II. An idealized model of the

circulation pattern and amplitude in oceanic basins.


Stommel, H.M., Arons, A., Faller, A., 1958. Some examples

of stationary planetary flow patterns in bounded basins.

Tellus 10, 179e187.

Stommel, H.M., Niiler, P.P., Anati, D., 1978. Dynamic

topography and recirculation of the North Atlantic.

J. Mar. Res. 36, 449e468.

Sverdrup, H.U., 1947. Wind-driven currents in a baroclinic

ocean. Proc. Nat. Acad. Sci., USA 33, 318e326.

Sverdrup, H.U., Johnson, M.W., Fleming, R.H., 1942. The

Oceans: Their Physics, Chemistry and General Biology.

Prentice-Hall Inc., Englewood Cliffs, NJ, 1057 pp.

Swallow, J.C., Worthington, L.V., 1961. An observation of

a deep countercurrent in the western North Atlantic.


Thorpe, S.A., 2004. Langmuir circulation. Annu. Rev. Fluid

Mech. 36, 55e79. doi:10.1146/annurev.fluid.36.052203.

071431.

Tomczak, M., Godfrey, J.S., 1994. Regional Oceanography, An

Introduction. Pergamon Press, Oxford, England. 422 p.

Treguier, A.M., 2006. Ocean models. In: Chassignet, E.P.,

Verron, J. (Eds.), Ocean Weather Forecasting: An Integrated

View of Oceanography. Springer, The

Netherlands.

Vallis, G.K., 2006. Atmospheric and Oceanic Fluid

Dynamics: Fundamentals and Large-scale Circulation.

Cambridge University Press, Cambridge, UK, 745 p.

Veronis, G., 1966. Wind-driven ocean circulation d part II.

Numerical solution of the nonlinear problem. Deep-Sea

Res. 13, 30e55.

Warren, B.A., 1981. Deep circulation of the world ocean. In:


Oceanography. MIT Press, Cambridge MA, pp. 6e41.

Weller, R., Dean, J.P., Marra, J., Price, J., Francis, E.A.,

Boardman, D.C., 1985. Three-dimensional flow in the

upper ocean. Science 118, 1e22.

Williams, R.G., 1991. The role of the mixed layer in setting

the potential vorticity of the main thermocline. J. Phys.

Oceanogr. 21, 1803e1814.

Wunsch, C., 1996. The Ocean Circulation Inverse Problem.

Cambridge University Press, New York, 458 pp.

Wunsch, C., 2009. The oceanic variability spectrum and

transport trends. Atmosphere-Ocean 47, 281e291.

Wunsch, C., Ferrari, R., 2004. Vertical mixing, energy, and

the general circulation of the oceans. Annu. Rev. Fluid

Mech. 36, 281e314.





1925-1927,6 1st Part 2, 109e288 (in German).

Wyrtki, K., 1975. Fluctuations of the dynamic topography in

the Pacific Ocean. J. Phys. Oceanogr. 5, 450e459.

C H A P T E R

S8

Gravity Waves, Tides, and Coastal

Oceanography: Supplementary Materials

This web-based content continues the topics

of Chapter 8, covering several different aspects

of coastal oceanography (river runoff, estuaries,

and coral reefs in Sections S8.7, S8.8, and S8.9),

followed by an extended discussion of adjacent

seas (Section S8.10), including the Mediterranean,

Black, Baltic, and North Seas from the

Atlantic; the Bering, Okhotsk, Japan (East),

Yellow, East China, South China Seas, and Gulf

of California from the Pacific Ocean; and the

Red Sea and Persian Gulf from the Indian Ocean.

Figure numbering in this portion of Chapter 8

continues from the last figure of the print text

but with “S” denoting online material, thus

starting with Figure S8.16.

S8.7. WATER PROPERTIES IN

COASTAL REGIONS: RIVER

RUNOFF

River runoff affects coastal regions. It reduces

the salinity of the surface layer and even of the

deeper water if there is sufficient vertical mixing.

It often carries a large amount of suspended sediment,

as seen for the Mississippi River outflow

and the outflows from the Himalayas into the

Bay of Bengal, including the Ganges River (Figure

S8.16). Generally, river runoff has a pronounced

seasonal variation, resulting in much larger

seasonal fluctuations of salinity in coastal waters

than in the open ocean. In a coastal region where

precipitation occurs chiefly as rain, the seasonal

salinity variation will closely follow the local

precipitation pattern. In regions where rivers are

fed by meltwater from snowfields or glaciers,

the river runoff increases in the summer to

many times the winter rate and causes a corresponding

decrease of salinity that lags the snowfall

by several months.

Fresh river water flowing out over saltier

open ocean water creates a strong halocline,

with high vertical stability. This can inhibit mixing

with water below the halocline. In the warm

seasons, this can result in higher temperatures

in the surface layer. In winter in high northern

latitudes, the halocline permits the surface layer

to cool to below the temperature of the water

beneath the halocline, producing a temperature

inversion. Ice thus tends to form first in coastal

waters, as “fast ice” (Section 3.9.1; the shallowness

of the coastal region also contributes). In

regions of multiyear ice such as the Arctic, the

new coastal ice spreads seaward until it contacts

the first-year ice spreading shoreward from the

multiyear pack ice.

Since river water frequently carries suspended

sediment (Figure S8.16), coastal waters

1

2

S8. GRAVITY WAVES, TIDES, AND COASTAL OCEANOGRAPHY: SUPPLEMENTARY MATERIALS

often have low optical transparency (Section

3.8.1). Sometimes this sediment is carried in

the surface low-salinity layer for some distance

while the deeper, more saline water remains

clear. The deposition of this sediment causes

shoaling and consequent hazards to navigation.

Frequently the location of the deposition is

influenced by the salinity distribution because

increases in salinity can cause flocculation of

the sediment and rapid settling.

The effect of runoff, especially from large

rivers, can often be traced a long way from the

coast, both by reduced salinity and by the sediment

in the water. Some examples of major influences

on the open ocean include the Amazon

and Congo Rivers in the tropical Atlantic, in

the northeast Pacific from the many rivers flowing

off the North American continent, and in the

Bay of Bengal from the large rivers whose runoff

is strongly influenced by the monsoon. Low

salinity from these river sources can be seen

even in the global surface salinity map (Figure

4.16). The net freshwater input from these sources

is an important part of the ocean’s freshwater

budget as apparent in the list of outflows of the

major rivers of the world in order of volume

transport in Dai and Trenberth (2002). Runoff is

comparable to open ocean precipitation and

evaporation because it deposits net precipitation

over land into the ocean.

The result of the sediment deposition from

runoff into the Bay of Bengal is apparent even

in open ocean bottom topography, as a smooth

sediment fan spreading down to 5000 m depth,

across the equator in the Indian Ocean, evident

in the bathymetry in Figure 4.13 (Curray,

Emmel, & Moore, 2003).

FIGURE S8.16 (a) Sediments in the Ganges River plume

in the northern Bay of Bengal. The Himalayas are the line of

snow-covered mountains across the top of the image; the

whites in the center right are clouds. (b) Mississippi River

estuary. Images from the Moderate Resolution Imaging

Spectroradiometer (MODIS). Source: From NASA Goddard

Earth Sciences, (2007c, 2008).

S8.8. ESTUARIES

An estuary, in the strictest definition, is

formed at the mouth of a river, where the river

meets the sea (Dyer, 1997). Cameron and Pritchard

(1963) defined an estuary as “a semi-enclosed

ESTUARIES 3

coastal body of water having a free connection to

the open sea and within which the sea-water is

measurably diluted with fresh water deriving

from land drainage.” They restrict the definition

to coastal features and exclude large bodies of

water such as the Baltic Sea. The river water,

which enters the estuary, mixes to some extent

with the salt water therein and eventually flows

out to the open sea in the upper layer. The mixing

processes are mainly due to tides and the

wind. A corresponding inflow of seawater takes

place below the upper layer. The inflow and

outflow are dynamically associated so that while

an increase in river flow tends to reduce the

salinity of the estuary water, it also causes an

increased inflow of seawater, which tends to

increase the salinity. Thus an approximate

steady state prevails.

The defining characteristic of estuarine circulation

is that inflow is denser than outflow,

which is diluted relative to the inflow. Sometimes

this concept is applied heuristically to

much larger bodies of water, such as the Black

Sea, or even the Indian and Pacific Oceans, but

the study of estuarine circulation is defined to

be within the confined coastal regions.

Extended descriptions of estuaries and estuarine

circulation can be found in the texts by Dyer

(1997), Officer (1976), and Hardisty (2007).

Compilations of papers on specific estuaries or

types of estuaries have appeared over the years.

Beardsley and Boicourt (1981) summarized the

Middle Atlantic Bight and Gulf of Maine;

Farmer and Freeland (1983) reviewed the physical

oceanography of fjord estuaries; papers in

Neilson, Kuo, & Brubaker (1989) summarized

then-contemporary ideas about estuaries; and

the Estuarine and Coastal Science Association

periodically produces compilations of papers.

S8.8.1. Types of Estuaries

There are many types of estuaries and many

types of flow in estuaries. A classification system

is useful as an introduction, but inevitably results

in oversimplification (Pritchard, 1989). Estuaries

are classified in terms of both their shape and their

stratification. They can also be classified in terms

of tidal and wind forcing. The inland end of an

estuary is called the head and the seaward end

the mouth. “Positive” estuaries have a river or

rivers emptying into them, usually at the head.

In terms of geology, three specific types of

estuary are recognized: the coastal plain

(drowned river valleys), the deep basin (e.g.,

fjords), and the bar-built estuary; there are also

types that do not fit in these categories (Dyer,

1997; Pritchard, 1989). The first is the result of

land subsidence or a rise of sea level that floods

a river valley; North American examples are the

St. Lawrence River valley and Chesapeake Bay.

Typical examples of the deep basin are the fjords

of Norway, Greenland, Canada, South America,

and New Zealand. Most of these have a sill or

region toward the seaward end, which is shallower

than both the main basin of the fjord

and the sea outside, so it restricts the exchange

of deep water. The third type is the narrow

channel between the shore and a bar, which

has built up close to shore through sedimentation

or wave action.

In terms of stratification and salinity structure,

estuaries have been classified based on

the distribution of water properties as (a) vertically

mixed, (b) slightly stratified, (c) highly

stratified, and (d) salt wedge estuaries (Figure

S8.17; Dyer, 1997; Pritchard, 1989). The stratification

is due to salinity, because density in estuaries

is determined mainly by salinity rather

than by temperature. The classification system

is not rigid. In the left-hand column of Figure

S8.17, the salinity distributions are shown as

vertical profiles at each of four stations between

the head and the mouth of the estuary (see schematic

plan view at the top). The right-hand

column shows simplified longitudinal sections

of salinity from head to mouth for the full depth

of the estuary. In most estuaries, unlike the schematics

in Figure S8.17, the bottom depth is shallowest

at the head.

4


FIGURE S8.17 Typical salinity/depth profiles (left) and longitudinal salinity sections (right) in different types of estuaries:

(a) vertically mixed, (b) slightly stratified, (c) highly stratified, and (d) salt wedge.

The vertically mixed estuary (Figure S8.17a) is

generally shallow and the water is mixed vertically

making it homogeneous from the surface

to the bottom at any particular place along the

estuary. The salinity increases with distance

along the estuary from head to mouth. The river

water in this type of estuary flows toward the

mouth while the salt may be considered to progress

from the sea toward the head by eddy diffusion

at all depths. In the right-hand figure, the

vertical isohalines indicate the homogeneity of

the water at each location while the straight

ESTUARIES 5

arrows indicate that the direction of net flow of

the water is seaward at all depths. (The circular

arrows symbolize the mixing taking place at all

depths.) The Severn River in England is an

example of a vertically mixed estuary.

In the slightly stratified estuary (Figure S8.17b),

which is usually also shallow, the salinity

increases from head to mouth at all depths.

The water is essentially in two layers with the

upper layer a little less saline than the deeper

one at each position along the estuary, with

a mixing layer between them (symbolized by

the circular arrows in Figure S8.17b). In this

type of estuary there is a net seaward (outward)

flow in the upper layer and a net inward flow in

the deeper layer as shown by the straight arrows

in the vertical salinity section. In addition to

these flows at both levels there is the vertical

mixing of both fresh and salt water giving rise

to the longitudinal variation of salinity in both

layers. The James River in Chesapeake Bay is

an example of this type of estuary.

In the highly stratified estuary (Figure S8.17c),

of which fjords are typical, the upper layer

increases in salinity from near zero in the river

at the head to a value close to that of the outside

sea at the mouth. The deep water, however, is of

almost uniform salinity from head to mouth.

Again there is a net outflow in the upper layer

and inflow in the deeper water as shown by

the straight arrows in the salinity section. In

these estuaries there is a very strong halocline

between the upper water and the deep water,

particularly at the head where vertical salinity

gradients of 10 to 20 psu per meter may occur

in summer during the period of greatest river

runoff. There is vertical mixing, but this results

predominantly in an upward movement of salt

water from below into the upper layer, with

little downward movement of fresh water. One

explanation for this almost unidirectional mixing

is that internal waves are generated by the

velocity shear between the upper low salinity

layer and the deeper more saline water, and

that the tops of these waves break and throw

off a “spray” of saline water into the upper layer

into which it mixes. There is much less breaking

at the bottom of the internal waves and therefore

no spray of fresh water downward into

the saline water.

For the salt wedge estuary (Fig S8.17d), the

longitudinal section indicates the reason for its

name. The saline water intrudes from the sea

as a wedge below the river water. This situation

is typical of rivers of large volume transport

such as the Fraser or Mississippi Rivers (Figure

S8.16b). It should be noted that the section in

Figure S8.17 is exaggerated in the vertical direction;

the salt wedge really has a much smaller

angle than shown, with almost horizontal

isohalines.

The salt wedge estuary has features in

common with the stratified estuaries. There is

a horizontal gradient of salinity at the bottom as

in a slightly stratified estuary and a pronounced

vertical salinity gradient as in a highly stratified

estuary. The distinction is in the lack of saline

water at the surface until it reaches the sea at the

mouth of the estuary, because of the large river

flow. In this type of estuary the salt wedge

migrates up and down the estuary as the tide

floods and ebbs, sometimes by several kilometers.

In terms of mixing, Stommel (reported by

Pritchard, 1989) suggested that estuaries be classified

in terms of tidal and wind forcing, which

are the main modes of mixing in estuaries. The

tides that are important in estuaries are almost

always co-oscillation tides (Section 8.6.2). This

classification has evolved to consider the tidal

range and effect of friction on tides in the estuary

(Dyer, 1997). Both the winds and tides have

temporal and spatial variations. The stratification

in an estuary can therefore vary significantly

with time as well as location in the estuary.

S8.8.2. Estuarine Circulation

In an estuary, the flow is out to the ocean in

the upper layer and into the estuary in the

bottom layer. In stratified estuaries, the depth

6


of the halocline (thickness of the upper, low

salinity layer) remains substantially constant

from head to mouth of an estuary for a given

river runoff. If the estuary width does not

change much, then the depth remains constant,

which means that the cross-sectional area of the

upper layer outflow remains the same while its

volume transport increases because of the

entrainment of salt water from below. Consequently,

the speed of the outflowing surface

layer markedly increases along the estuary

from head to mouth. The increase in volume

and speed can be considerable, with the outflow

at the mouth as much as 10 to 30 times the

volume flow of the river. In his classical study

of Alberni Inlet d a typical, highly stratified,

fjord-type estuary in British Columbia d Tully

(1949) demonstrated the above features. He

also showed that the depth of the upper layer

decreased as the river runoff increased up to

a critical value and thereafter increased as

runoff increased.

Estuarine circulation depends on several

factors: the sill depth, river runoff rate, and the

character of the outside water density distribution.

Tides and mixing also impact the circulation.

If the sill is so shallow that it penetrates

into the low-salinity, out-flowing upper layer,

the full estuarine circulation cannot develop

and the subsurface inflow of saline water does

not occur regularly. As a result, the deep water

is not exchanged regularly and tends to become

stagnant. This situation occurs in some of the

smaller Norwegian fjords, but is by no means

typical of deep basin estuaries. Most of the

fjords in Norway, as well as on the west coasts

of North and South America and New Zealand,

have sills that are deeper than the upper layer.

Therefore the estuarine circulation is developed

sufficiently to affect continual renewal of the

deep water and stagnation does not occur (Pickard,

1961; Pickard & Stanton, 1980). The rate of

renewal is proportional to the circulation, which

is proportional to the river runoff. Fjord estuaries

with small river runoff show more

evidence of limited circulation in the form of

low oxygen values than those with large runoff.

The depth of the sill has little effect as long as it

is greater than the depth of the low-salinity, outflowing

upper layer.

The other major factor influencing the

exchange of the deep basin water is seasonal

variation in the density structure of the outside

seawater. Although the downward mixing of

fresh water in an estuary is small, it does occur

to some extent. Therefore the salinity, and hence

the density of the basin water, tends to decrease

slowly. If a change then occurs in the outside

water such that the density outside becomes

greater than that inside at similar levels above

the sill depth, then there will be an inflow of

water from the sea. The inflowing water is likely

to sink, although not necessarily to the bottom,

in the estuary basin and displace upward and

outward some of the previously resident water.

In this way the basin water becomes refreshed.

In deep-sill estuaries this refreshment may

occur annually, but in shallow-sill estuaries it

may occur only at intervals of many years; the

disturbance to the biological regime may be

cataclysmic on these occasions (by displacing

upward into the biotic zone the low-oxygen

water from the bottom). This type of basinwater

replacement has been well documented

for some Norwegian fjords (with very shallow

sills), but it should not be considered characteristic

of all fjord estuaries.

The previous remarks only briefly describe

some of the salient characteristics of stratified

estuaries; the property distributions in Figure

S8.17 are smoothed and schematic. Real distributions

show fine and mesoscale structure and

detailed features, some general and some local.

In particular, because the density structure is

determined largely by the salinity distribution,

temperature maxima and minima are quite

common in the water column. Mixing between

fresh and salt water is largely governed by tidal

movements and the effects of internal waves.

The circulation that was just reviewed for

CORAL REEFS 7

stratified estuaries is greatly modulated by the

strong tidal currents in the estuaries. This brief

description also neglects the horizontal variability

and horizontal circulation in estuaries.

Estuarine characteristics and processes are

observed in ocean areas as well as near the coast.

In the northeast Pacific and in the Bay of Bengal,

where there is considerable river runoff, the

density of the upper layer is controlled by the

salinity rather than by temperature as is usually

the case in the open ocean. The upper, lowsalinity

layer of perhaps 100 m depth in the

northeast Pacific is much less dense than the

deeper, more saline water and the stability in

the halocline between them inhibits mixing.

Consequently, the summer input of heat is trapped

in the surface layer and a marked seasonal

thermocline develops as shown in Figure 4.8.

S8.8.3. Flushing Time of Estuaries

The time that it takes to replace the freshwater

within an estuary through river discharge

is called the flushing time. This is important for

water quality within estuaries. The flushing

time has significant temporal variation, especially

since river flows have strong variability.

Following Dyer (1997), the flushing time is the

freshwater volume (V F in units of m 3 ) divided

by the river discharge (R, in units of m 3 /sec).

Both the freshwater volume and the river

discharge can be time dependent. Using observations

of the average salinity <S> within the

estuary compared with the seawater salinity S o

outside the estuary, the freshwater fraction can

be estimated as F ¼ (S o <S>)/S o . The freshwater

volume is the total volume, V, multiplied

by the freshwater fraction. The flushing time is

then

t F ¼ V F =R ¼ FV=R:

(S8.10)

Flushing times range from several days to

a year. Dyer lists flushing times for several estuaries:

Narragansett Bay, Massachusetts e 12 to

40 days depending on river flow; Mersey e 5.3

days; Bay of Fundy e 76 days; and the Severn

Estuary e 100 to 300 days depending on river

flow.

Observing the salinity at all locations in the

estuary at all times is unrealistic, so various

approximate methods are used to determine

the flushing time. Dyer (1997) is a good source

for these different methods.

S8.9. CORAL REEFS

The physical oceanography of coral reefs was

of particular interest to George Pickard, the original

author of this text. He published several

papers and a book on the Great Barrier Reef in

1977 (Pickard, Donguy, Henin, & Rougerie,

1977). Therefore we retain this section of

Chapter 8, but it has not been updated. Many

papers and some books have been published

on the physical oceanography of coral reefs

since the previous edition, and Wolanski (2001)

and Monismith (2007) are suggested as starting

points.

Prior to about 1970, most physical oceanography

in coral reef areas had been carried out

as ancillary to biological or chemical studies.

Studies of the dynamics started in the late

1970s with most studies carried out in the Great

Barrier Reef of Australia. Through the 1980s, of

the 200 publications on the physical oceanography

of coral reef regions, 70% refer to the Great

Barrier Reef and only about 30% to other reef

areas. Because of observed degradation of coral

reefs worldwide, coral reef research expanded

greatly in the 1990s, with numerous reefs now

the focus of careful, ongoing studies d many

with physical oceanographic components.

S8.9.1. Topography of Coral Reefs

Coral reefs are features of many coastal

regions between the tropics, along the continental

shelves, around islands, and also on the

tops of shoals and seamounts in the open ocean

8


with little or no emergent land in their vicinity,

such as atolls. Reefs act as complex barriers to

flow in their neighborhood. Living coral cannot

withstand exposure above water so it only

occurs below low-tide level; it also requires light

for growth so it generally cannot survive below

about 50 m depth.

Along a land boundary, two types of reef are

recognized: (1) the fringing reef, which extends

out from the shore and (2) the barrier reef, which

is located away from the shore with a relatively

reef-free region (lagoon) between it and the

shore. Despite its name, a barrier reef is rarely

continuous for very long distances; instead it

consists of a series of reefs with gaps between

them. Water exchange with the ocean can take

place through the gaps as well as over the reefs.

In some cases the outer reefs may be long

(parallel to the coast) and narrow; in others,

the “barrier” may consist of a number of individual

reefs dotted over a relatively wide band

(50 km or more) parallel to the shore, with irregular

passages between them connecting the

lagoon with the open ocean outside. Open ocean

reefs, based on shoals or seamounts, usually

extend around the shoals; the “lagoon” is the

body of shallow water within the reef perimeter,

which usually has gaps that permit some direct

exchange with the ocean. Atolls are reefs where

sufficient material has collected on parts of the

reef to raise the level a few meters above sea

level and on which shrubs and trees may grow.

S8.9.2. Water Properties in Coral Reefs

Extensive and long-term measurements of

water properties have been made in the Great

Barrier Reef; in the southwest lagoon of New

Caledonia; and in the Hawaiian, Floridian, and

Caribbean reefs. The annual variation of water

temperature is generally approximately sinusoidal

with the maximum in the local summer

and closely correlated with the air temperature.

The annual range of temperature variation

decreases toward the equator.

Salinity variations are less regular than those

of temperature. Near land, decreases occur due

to local precipitation and river runoff associated

with monsoons, whereas for atolls only precipitation

is effective. Increases of salinity are due to

evaporation. An increase from an oceanic value

of 35.7 psu near the pass into Canton Island

lagoon to 39.5 psu at the back of the lagoon

approximately 15 km away was recorded, but

this is probably an extreme example.

Water depths in lagoons and around reefs are

generally small, less than 50e100 m, and the

water is usually unstratified, for example,

well-mixed vertically due to turbulence from

wind-wave effects and the rough character of

the bottom over the reefs. Some stratification

in the upper 10e20 m occurs near river mouths

during periods of heavy runoff, but even then

the variations in the vertical are generally less

than 1 C in temperature and 1psu in salinity.

Along the Great Barrier Reef, intrusions of

cooler, more saline water can be evident near

the bottom of passes through the outer reefs

with Dt ¼ -5 C and DS ¼ +1 psu relative to the

upper layer reef-area waters. In the very shallow

water over fringing reefs, diel (day-night) variations

of as much as 10 to 12 C have been

observed and attributed to solar heating during

the day and radiant cooling at night. It is probable

that the relatively reef-free lagoon between

the shore and off-lying barrier reefs occurs

because coral is intolerant both of the fresh

water and silt carried in by rivers.

S8.9.3. Currents in Coral Reefs

To describe the currents we divide them into

three classes: drift or long-period (periods of

weeks or more), weather band (periods of

days), and tidal (periods of hours). Drift currents

are generated by steady wind stress (e.g., the

tradewinds)orlong-shorepressuregradients.

The oceanic equatorial currents generated by

the trade winds cause flow over the mid-ocean

reefs, such as at Bikini. In the central Great

CORAL REEFS 9

Barrier Reef, a 25-year time series of current

measurements showed equatorward currents of

20 cm/sec during the south-east Trade Wind

season, while at other times there was a poleward

current of about 30 cm/sec attributable to the

downward slope to the south associated with

the southward flow of the East Australian

Current outside the Great Barrier Reef.

Weather band currents associated with continental

shelf waves have also been documented

for the central and southern Great Barrier Reef.

They are a consequence of fluctuations of wind

stress as weather systems move eastward with

their centers over the southern part of Australia.

The weather systems have periods of 10 to 20

days and speeds of some 500 km/day (equatorward).

As the resulting ocean currents have

a vertical range of only 10 to 30 cm (near the

shore and diminishing to zero outside the

reef), they are not evident to the eye and can

only be identified by analysis of tide or current

records. Water particle speeds are approximately

20e40 cm/sec or 17e35 km/day. This

can result in long-shore displacements of water

of 100 to 200 km that can be very significant in

transporting pollutants or plankton over such

distances in the reef area.

Tidal-flows through reef passes of approximately

200 cm/sec are common; speeds as high

as 370 cm/sec (13 km/h) at Aldabra Atoll have

been recorded. It should be noted that although

such tidal speeds through passes can be large,

the inflowing water then spreads out in the

lagoon and the distance of penetration of ocean

water during the flood (which only lasts about

6 hours) may be only a few kilometers. This is

small compared to the diameter of many atoll

lagoons and therefore tidal flows may only

have a limited effect on water exchange and

flushing. Also, it has been observed that there

is often little mixing between the intruding ocean

water and the resident lagoon water.

The term “reef flat” refers to extensive areas of

coral of relatively uniform height; the water

depths over them are generally only a few

meters. Flow over them may be due to drift

currents, shelf waves, and tidal currents as well

as to the local wind stress. Note that as the water

in these areas is very shallow, bottom friction is

more important than Coriolis force so the

wind-driven flow is downwind, not to the left

or right of the wind direction as in deeper water

(see Section 7.5.3). Tidal currents over the reef

flats may have speeds of 100 cm/sec or more.

These flows (over the surrounding reef) will

not necessarily go into a lagoon during the flood

or out during the ebb. For instance, in the Great

Barrier Reef, the tide wave approaches from the

northeast so that the tidal flow in the area during

the flood is to the southwest and can be over

a surrounding reef into its lagoon on the northeast

side but out of the lagoon on the southwest

side at the same time. Wave-overtopping also

contributes to the water in a reef lagoon. This

occurs when ocean waves or swells break on

the outside of a reef generating a slope across

the reef, which causes flow across it. This component

can contribute as much transport across

a reef as the other mechanisms combined.

S8.9.4. Circulation in Lagoons

The circulation in individual reef lagoons may

be forced by some or all of the current mechanisms

described previously. In the extensive

open lagoon between the shore and the barrier

reef in the Great Barrier Reef, steady drift currents

due to wind stress and pressure gradients, and

periodic currents due to tides and continental

shelf waves, all contribute to the water circulation.

In the New Caledonia lagoon, about 20 80 km

long inside a narrow barrier reef, the tide appears

to be the main contributor to water circulation

during light wind conditions. During strong

southeast trade winds the wind stress superimposes

a general northwest motion over the barrier

reef and across the lagoon.

Within the Great Barrier Reef, individual reefs

form partial obstacles to the general flow d

“partial” because, except at very low water,

10


some flow occurs continually over the coral reefs.

Eddies often form downstream of reefs, particularly

during flows associated with tidal currents.

Such eddies can increase mixing on a small scale

(tens to hundreds of meters) as well as form

closed volumes in which plankton can be held

for hours or longer. Shedding of eddies behind

reefs probably does not occur very often, because

flow speeds are not great enough.

For the roughly circular lagoons within individual

reefs in the Great Barrier Reef, again the

drift and periodic currents contribute to the

circulation together with inflow due to waveovertopping.

Studies within atoll lagoons have

demonstrated that in addition to inflow due to

ocean currents, tides, and wave-overtopping,

wind stress causes downwind flow in the upper

layer at speeds of about 3% of the wind speed

while a compensating upwind flow develops

in the deeper water.

Residence times for water within lagoons

cover a wide range. For lagoons of 2e10 km

diameter in the Great Barrier Reef, times of 0.5

to 4 days have been estimated, for Bikini Atoll

40 to 80 days, and for very shallow lagoons

such as at Fanning Atoll (18 km long but only

a few meters deep) periods of up to 11 months

were estimated. Rougerie (1986) estimated residence

times in the New Caledonia lagoon as 2 to

28 days, depending on the particular area and

the runoff, wind, and tide characteristics.

S8.10. ADJACENT SEAS

S8.10.1. General Inflow and Outflow

Characteristics

Adjacent seas affect the open ocean’s stratification

and circulation through water mass

transformation. Transformation can be from

dense to light water (as in the Black Sea and

Baltic, where fresh water is added within the

sea, greatly reducing the salinity, Chapter 9),

from light to dense (as in the Nordic Seas and

the Mediterranean Sea, where there is large

cooling and evaporation, Chapter 9), or due to

vigorous mixing (as in the Indonesian passages,

Chapter 11).

Exchanges between basins (inflow and

outflow) can be principally separated in the

vertical or in the horizontal directions (Figure

S8.18). Adjacent seas that are separated from

a larger basin by a narrow strait usually have

vertically stratified exchange, with inflow in

one layer and outflow in a layer of a different

density. The inflow and outflow “layers” are

separated by an interfacial layer within which

vigorous vertical mixing modifies both the

inflow and outflow. The Mediterranean, Black,

and Baltic Seas, discussed in the following

sections, have this type of exchange. Other

examples include the Red Sea and Persian

Gulf in the Indian Ocean (Section 11.6 and

Section S8.10.7 below). For the Mediterranean,

Red Sea and Persian Gulf, inflow is in the upper

layer and outflow in the lower layer, with

a density increase within the sea. In the Black

and Baltic Seas, inflow is in the lower layer

and outflow is in the upper layer, with a density

decrease within the sea.

When the exchanges between basins can

occur over a much broader region than just

a narrow strait, the inflow/outflow geometry

can be more horizontal, with water of one

density entering in one region and transformed

water exiting in another (Figure S8.18b). The

inflow and outflow might adjoin each other, or

occur through separate passages. The Caribbean

Sea and Gulf of Mexico (Section 9.3.1) are of this

type as are Fram Strait between Greenland and

Spitsbergen (Chapter 12); the Indonesian

passages (Section 11.5); and the Bering,

Okhotsk, and Japan Seas (Sections S8.10.5 and

S8.10.6).

Many exchanges are a mixture of horizontal

and vertical. The exchange between the North

Atlantic and North Sea (Section S8.10.4) is mostly

horizontal, with inflow through Dover Strait and

near the Shetland Islands and outflow along the

ADJACENT SEAS 11

FIGURE S8.18 Schematic diagram of various types of circulation in seas adjacent to the oceans. (a) Vertically separated

inflow and outflow, typified by the Mediterranean and Red Seas (surface inflow and subsurface outflow with transformation

within the sea), Baltic Sea and Hudson Bay (subsurface inflow and surface outflow, with transformation within the sea), and

Black Sea (subsurface inflow and surface ouflow, with no deep ventilation within the sea). (b) Horizontally separated inflow

and outflow, typified by the Arctic Ocean and Caribbean Sea and also by the North Pacific marginal seas (surface inflow and

surface outflow, usually through a different strait from the inflow), and by the Nordic Seas, Labrador Sea/Baffin Bay, and

also the Persian Gulf (surface inflow and both surface and subsurface outflow).

Norwegian coast. But there is also inflow of saline

water within the Norwegian Trench, beneath the

fresher outflow along the Norwegian coast. For

the Nordic Seas exchange with the North Atlantic,

the exchange is mostly horizontal, with generally

less dense Atlantic Water (AW) entering the

Nordic Seas along the eastern boundary in the

Norwegian Atlantic Current, and denser outflow

occurring across each of the three main deep sills.

However, the easternmost of these outflows, over

12


the Faroe-Shetland Ridge, is beneath the northward

inflow of AW.

S8.10.2. Mediterranean Sea

The Mediterranean Sea (Figure S8.19) is

a nearly enclosed marginal sea in the eastern

North Atlantic, connected to the Atlantic

through the Strait of Gibraltar, which has a sill

depth of 284 m. The maximum depths within

the sea are about 3400 m in the western basin

and 4200 m in the eastern, separated by the

Strait of Sicily, which is 430 m deep. The Mediterranean

is connected to the Black Sea in the

(a)

10˚W 5˚ 0˚ 5˚ 10˚ 15˚ 20˚ 25˚ 30˚ 35˚E

45˚N

40˚

35˚

Gulf of

Cadiz

Ebro R.

Alboran Sea

Strait of

Gibraltar

Gulf of

Lions

Northern Current

Rhone R.

Balearic Sea

Ligurian

Sea

Algerian Basin

Algerian Current

Str. of

Sicily

Po R.

Adriatic Sea

Tyrrhenian

Sea

Ionian Sea

Black Sea

Bosphorus

Dardanelles

Aegean

Sea

Rhodes

Crete

Cyprus

Levantine Sea

45˚N

40˚

35˚

30˚

Nile R.

30˚

10˚W 5˚ 0˚

5˚

10˚

15˚

20˚

25˚

30˚

35˚E

Bottom depth (m)

1000 2000 3000 4000 5000 6000 7000

FIGURE S8.19 Mediterranean Sea. (a) Surface circulation schematic. Principal features are in red. After Millot & Taupier-

Letage (2005) and Robinson et al. (1991). Blue shows dense flow direction through the straits and away from formation areas,

which are roughly indicated with purple. Etopo2 topography (m). Source: From NOAA NGDC (2008). (b) Schematic of

overturning circulation (surface, intermediate, and deep layers); curved arrows indicate that only the lighter part of the layer

can flow over the sill.

ADJACENT SEAS 13

northeast through the Dardanelles and

Bosporus.

The tidal range in the Mediterranean is

small, decreasing from 0.8 m at Gibraltar in

the west to 0.4 m at Port Said in the east (to as

low as 0.2 m along the French coast in the

north). Sea level decreases in a northeasterly

direction by about 0.7 m from the African coast

to the Aegean Sea.

The Mediterranean Sea is the prototype of

a “negative” basin in terms of water balance

(Section 5.3.1), with evaporation exceeding

precipitation and runoff. There is also net cooling

within the sea. Therefore, the outflow from

the Mediterranean is denser (saltier and cooler)

than the inflow. The sea is well ventilated to its

bottom as a result. The Mediterranean Sea has

a profound impact on North Atlantic water

properties because of the high salinity and

density of its outflow. While the Mediterranean

contributes only about one-third of the net evaporation

of the Atlantic Ocean, its cooling of the

saline surface waters allows them to sink to

depth in the Atlantic, which is important for

the entire North Atlantic Deep Water formation

process.

S8.10.2.1 Exchange at the Strait of Gibraltar

The net exchange through the Strait of

Gibraltar is small, on the order of 0.7 Sv (Bryden,

Candela, & Kinder, 1994). The salinity difference

between the Atlantic inflow in the surface layer

and Mediterranean outflow below is large: 2.3

psu (from 36.1 to 38.4 psu). The inflow temperature

is strongly seasonal, but averages around

15e16 C, hence a potential density of s q ¼ 26.6

to 26.8 kg/m 3 . The temperature of the outflow

is around 13.3 C, so the potential density of

the outflow is about 28.95 kg/m 3 (Figure 9.23b).

The flow through the Strait of Gibraltar is

hydraulically controlled and modulated by

tides (Armi & Farmer, 1988). Minimum sill

depth is at the Camarinal Sill, while the

minimum horizontal constriction is farther to

the east, at Tarifa Narrows; both affect the

outflow. The interface depth between the AW

and outflowing Mediterranean Water is about

100 m at the sill, and slopes downward to almost

250 m depth on the Atlantic side (Bray, Ochoa, &

Kinder, 1995). This downward slope is typical of

hydraulically supercritical flow. The interface

also slopes upward toward the northern side

of the strait because of the Coriolis force, so

the saltiest, densest water is banked to the north

(Figure 9.23a).

The saline Mediterranean Water at the Strait

of Gibraltar is one of the densest water masses

in the world ocean; it is denser than the various

Nordic Seas Overflow Waters (NSOW) at their

overflow sills. However, instead of sinking to

the bottom of the North Atlantic like the

NSOW, the Mediterranean Water equilibrates

at about 1200 m depth due to the difference in

the stratification of the entrained waters for

these two overflows (Price & Baringer, 1994;

Figure S7.6b). The warmth of the Mediterranean

outflow also means that it compresses less than

NSOW as both descend to high pressure; in

terms of potential density referenced to 4000

dbar, the NSOW is actually denser than the

Mediterranean outflow.

S8.10.2.2 Circulation of the Mediterranean

Sea

The horizontal and vertical circulations in the

Mediterranean Sea are strongly affected by the

basin geometry, which is separated into western

and eastern basins by the Strait of Sicily and

which has a saddle depth of about 430 m (Figure

S8.19). The general sense of mean circulation in

the Mediterranean Sea is cyclonic. Surface water

enters from the North Atlantic through the Strait

of Gibraltar. Within the Alboran Sea close to the

strait, the circulation is anticyclonic (Alboran

gyre), but then becomes cyclonic as the AW

flows into the Algerian Basin, following the

North African coastline. This eastward coastal

flow is called the Algerian Current. The flow

splits at the Strait of Sicily into a branch that

continues eastward through the strait and

14


a northward branch feeding the cyclonic

western Mediterranean circulation. The mean

westward flow along the northern side can be

called the Northern Current (Millot, 1991).

In the eastern Mediterranean, the Adriatic

and Aegean circulations are each cyclonic.

Circulation in the Levantine Basin is more

complex, with a quasi-permanent, meandering

eastward jet in mid-basin (the mid-Mediterranean

or mid-Levantine jet; e.g., Robinson et al.,

1991 and Özsoy et al., 1991). South of the jet,

there are quasi-permanent, anticyclonic, subbasin-scale

gyres. North of the jet, there are

cyclonic, sub-basin-scale gyres. Of the several

commonly occurring gyres, the cyclonic Rhodes

gyre between Crete and Cyprus is singled out

here, as it is the site of Levantine Intermediate

Water (LIW) formation (see later in this section).

The Mediterranean circulation is markedly

time dependent. In the western Mediterranean,

the Algerian Current regularly spawns large

(100e200 km) anticyclonic mesoscale features,

while the Northern Current region is also

eddy-rich, although the features are not as

coherent (Millot, 1991). In the eastern Mediterranean,

the sub-basin-scale anticyclonic and

cyclonic gyres are strongly time dependent

(Özsoy et al., 1991; Robinson et al., 1991).

The mean horizontal circulation in the Mediterranean

at both intermediate and deep levels

is cyclonic, similar to the surface circulation

(Millot & Taupier-Letage, 2005).

Transports of the main currents in the Mediterranean

are on the order of 1 to 3 Sv. That is,

the mean circulation is on the order of the

exchange through the Strait of Gibraltar, and

does not reach the strength of the open North

Atlantic currents.

The vertical circulation of the Mediterranean

is best described in terms of its water masses

(Figure S8.19b and Section S8.10.2.3). AW flows

in through the Strait of Gibraltar and circulates

at the surface through the Mediterranean. In

the eastern Mediterranean, LIW is formed in

the vicinity of the Rhodes Gyre and then

spreads cyclonically at mid-depth (200e600 m).

LIW is the source of the deep waters in both

the eastern and western Mediterranean. LIW

and the lighter part of the Eastern Mediterranean

Deep Water flow back to the west

through the Strait of Sicily beneath the eastward

flow of AW. This joins the Western Mediterranean

Deep Water (WMDW). Outflow

through the Strait of Gibraltar originates in

the Northern Current along the coast of Spain

and is composed of LIW and the upper part of

the WMDW.

S8.10.2.3 Properties and Water Masses

Within the Mediterranean Sea

The Mediterranean is a saline, warm, wellventilated

basin. Due to prevailing dry northwest

winds and frequent sunny days, there is

a large excess (about 100 cm/year) of evaporation

over precipitation in the eastern part of

the Mediterranean Sea. The high temperatures

and salinities are surpassed only in the Red

Sea. Salinity ranges from 36.1 psu, in the

entering AW, to 39.1 at the surface in the eastern

Mediterranean. Bottom temperatures are above

12.5 C even at 4000 m (Wüst, 1961) and have

become warmer (Klein et al., 1999). Bottom

water densities exceed s q ¼ 29.25 kg/m 3 due

to the high salinity and deep oxygen exceeds

200 mmol/kg. These properties differ greatly

from those at the same depth in the adjacent

North Atlantic, which are 2.4 C, 34.9 psu, and

s q ¼ 27.8 kg/m 3 .

The nomenclature for Mediterranean water

masses has varied. An official list of names

and acronyms is maintained by CIESM (2001).

We limit our discussion to four primary water

masses: AW in the surface layer; LIW in the

intermediate layer; and two dense waters,

WMDW and the denser Eastern Mediterranean

Deep Water (EMDW). LIW is formed in the

northern Levantine Basin, off the south coast

of Turkey near the island of Rhodes. The Deep

Waters are formed at the northern edges of the

basins, chiefly in the Gulf of Lions in the western

ADJACENT SEAS 15

basin (WMDW) and in the southern Adriatic

and in the Aegean (EMDW).

LIW is recognized throughout the Mediterranean

by a subsurface vertical maximum of

salinity between 200 and 600 m (Figure S8.20;

Wüst, 1961). At formation, LIW salinity is

greater than 39.1 psu and its temperature is

around 15 C. After passing westward through

the Strait of Sicily its core becomes colder (13.5

C), fresher (38.5), slightly less dense, and somewhat

deeper than in the eastern Mediterranean.

LIW formation was observed in the cyclonic

Rhodes Gyre in early 1995 (Malanotte-Rizzoli

et al., 2003). Deep convection to 900 m occurred

in this gyre (Figure S8.21), acting as a classic

convective chimney (Section 7.10). The dense

FIGURE S8.20 (a) Salinity at the vertical salinity maximum characterizing LIW. Source: From Wüst (1961). (b) Longitudinal

salinity section in winter to show the LIW.

16


(a)

(b)

(c)

FIGURE S8.21 Eastern Mediterranean convection. Formation of dense water in the Levantine Basin in January 1995.

(a) Surface dynamic height relative to 800 dbar. The cyclonic low centered at 28 30’E, 35 N is the Rhodes gyre. (b) Salinity

and (c) potential density s q along section A. LIW is the salinity maximum at 50e100 m. The convective chimney in the center

is forming a deep water. Source: From Malanotte-Rizzoli et al. (2003).

ADJACENT SEAS 17

water in the chimney in that year was a newly

identified deep water, rather than LIW. New

LIW was the shallow salinity maximum on the

outside of the gyre (right and left at 50e100 m

depth in Figure S8.21b,c). LIW spread much

farther into the eastern Mediterranean at the

end of that winter than did the new deep water,

which was trapped in the Rhodes Gyre.

The Deep Waters originate at several locations

along the northern coast through winter

cooling of the widespread high salinity LIW. In

the Adriatic and Aegean, cold winter outbreaks

with intense winds (the “Bora”) create EMDW

from LIW. The Adriatic has historically been

the site of the densest formation (Schlitzer et

al., 1991). However, changes in winter conditions

shifted the densest water production to

the Aegean in the 1980s, and then back to the

Adriatic in the 1990s (Klein et al., 2000). The

presence of EMDW affects the properties of

new LIW in the eastern Mediterranean.

In the western Mediterranean, dense water

formation contributing to WMDW occurs

primarily in the Gulf of Lions in a sub-basinscale

cyclonic gyre (Figure S8.22), in response

to cold, dry winter winds (the “Mistral”). The

first observations of the classic stages of deep

convection (preconditioning, convective mixing

and spreading; Section 7.10.1) were made here

in 1969 (MEDOC Group, 1970; Sankey, 1973).

Deep convection has occurred reliably in this

region in many other years within a cyclonic

dome of about 100 km scale. Dense water properties

here are around 12.8 C, 38.45 psu, and

s q ¼ 29.1 kg/m 3 (Marshall & Schott, 1999).

Thus WMDW is not as dense as EMDW, but

the Strait of Sicily blocks the densest EMDW

from flowing into the western Mediterranean.

Thus the deep water of the western Mediterranean

is a mixture of local Gulf of Lions dense

waters and the shallower part of EMDW.

Water mass properties and formation rates in

the Mediterranean are demonstrably affected by

the North Atlantic Oscillation and by the warming

and drying trends of global climate change

(Section S15.6). Because the sea is relatively

small, these changes affect the relative balance

and properties of the different deep and intermediate

waters. The net effect is observed in

changes of salinity and temperature of the

outflow at the Strait of Gibraltar. This has

affected Mediterranean Water properties within

the North Atlantic (Potter & Lozier, 2004).

S8.10.3. Black Sea

The Black Sea is an almost completely isolated

marginal sea with a maximum depth

of over 2200 m. It is connected to the northeastern

Mediterranean Sea through the narrow

Bosphorus and Dardanelles, which have depths

of only 33 and 70 m, respectively (Figure S8.23).

The small Sea of Marmara lies between the two

straits. Inflow from the Mediterranean is more

saline and denser than the fresh outflow from

the Black Sea. The Black Sea thus represents

a classic estuarine circulation with inflow at

the bottom and outflow at the surface in the

straits; it is a “positive sea” because it has

a net input of freshwater (Section 5.3.2). The

Black Sea is also a prototypical anoxic sea with

no dissolved oxygen below the pycnocline

because of the very long residence time of its

deep water. Black Sea physical oceanography

was reviewed by Özsoy and Ünlüata (1998)

and Oguz et al. (2006).

The surface circulation of the Black Sea is

cyclonic overall, with a cyclonic gyre in each

of the west and east basins, including cyclonic

eddies (Figure S8.23a andOguz et al., 2006). A

Rim Current circulates around the exterior,

roughly following the continental shelf break.

Its maximum velocity is 50e100 cm/sec at

the surface. Inshore of the Rim Current is

a series of anticyclonic eddies or small gyres

that connect the coastal regions with the

cyclonic circulation (light contours in Figure

S8.23a). The whole is dominated by timedependent

eddies and seasonal changes in

circulation.

18


FIGURE S8.22 Western Mediterranean convection. (a) Deep convection region in the Gulf of Lions for three winters.

Circulation (arrows) and isopycnal depth (m) showing cyclonic doming. (b) Potential density through the convection region

in February 1992. Source: From Marshall and Schott (1999).

ADJACENT SEAS 19

The narrowness and shallowness of the

passages between the Black Sea and Mediterranean

result in high current speeds and vertical

shear. The consequent turbulence causes

vertical mixing between the inflow and outflow

layers. Therefore the surface water that leaves

the Black Sea with a salinity of about 17 psu reaches

the Mediterranean with its salinity

increased to about 30 psu, while the salinity of

38.5 psu of incoming subsurface Mediterranean

water is reduced to about 34 psu by the time that

it reaches the Black Sea.

Exchange with the Mediterranean includes

inflow on the order of 300 km 3 yr 1 (300 km 3

yr 1 is equal to 9.5 10 3 m 3 sec 1 , hence 0.0095

Sv) and outflow of the order of 600 km 3 yr 1

(Oguz et al., 2006). The higher outflow is due

to net freshwater input. Evaporation and precipitation

within the Black Sea are each on the order

of 300 km 3 yr 1 each, so they are nearly

(a)

25˚

30˚

35˚

40˚

Dniester R.

Dnieper R.

Azov Sea

45˚

Danube R.

45˚

Western Gyre

Eastern Gyre

Bosphorus

Dardanelles Sea of Marmara

40˚

Aegean

Sea

40˚

25˚

30˚

35˚

40˚

1000 2000 3000 4000 5000 6000 7000

FIGURE S8.23 Black Sea. (a) Surface circulation schematic. Heavy contours: principal circulation. Light contours: shelf

circulation. Dashed contours: eddy-like circulation in interior. Blue: subsurface inflow through the Dardanelles and

Bosphorus. After Oguz et al. (2006), with Etopo2 topography from NOAA NGDC (2008). (b) Water properties in the upper 200 m

in the Black Sea, 1988. Adapted from Murray et al. (1989). (c) Overturn and transport balances (km 3 yr 1 ). Source: From Oguz

et al. (2006).

20


FIGURE S8.23

(Continued).

balanced. Inflow from the large rivers, including

the Danube, Dniester, and Dnieper in the northwest,

is also on the order of 300 km 3 yr 1 . Thus

without the river inflow, salinity in the Black Sea

would be nearly neutrally balanced. In Section

5.3.2, we calculated a residence time for the

Black Sea of 1000 to 2000 years.

As a result of the net input of freshwater and

the long residence time, the Black Sea is one of

the world’s major brackish (low salinity) seas.

A halocline between 50 and 100 m separates

the fresher surface layer from the subsurface

water column (Figure S8.23b). The sharp halocline/pycnocline

separates the upper lowsalinity,

oxygenated (oxic) water from the

deeper oxygen-free (anoxic) water.

A subsurface temperature minimum of <8 C

near 100 m is called the Cold Intermediate Layer.It

is most likely a remnant of the winter surface

mixed layer. Because the surface layer is so

ADJACENT SEAS 21

fresh, this density structure is stable; it is similar

to the subpolar structures found in the northern

North Pacific and parts of the Southern Ocean.

The Deep Water in the Black Sea has a salinity

of 22.3 psu and potential temperature of 8.9 C.

The salinity structure and isolation of the

deep Black Sea produce interesting double

diffusive and deep geothermal convective

features. The overall vertical structure is diffusive,withcolder,freshwaterintheColdIntermediate

Layer overlying warmer, saltier water

below the pycnocline (Özsoy & Ünlüata, 1998;

Kelley et al., 2003). Inflowing salty Mediterranean

water enters the pycnocline and below in

intrusions that are double diffusive. Weak

geothermal heating at the bottom of the Black

Sea, along with the long residence time,

creates a very thick (>450 m) convective

bottom layer, rivaled only in the deep Arctic

(Section 12.5.3; Timmermans, Garrett, &

Carmack, 2003).

Over geological time, the Black Sea has

varied from being fresh (as recently as 7000

years ago), to moderately saline. As global sea

levels rose and fell during glacials and interglacials

and river outlets changed location, the

exchange between the Black Sea and Aegean

may have reversed direction and even ceased

because the Bosphorus and Dardanelles are so

shallow. There is an ongoing paleoclimate

debate about whether overflow in through the

Bosphorus 5600 years ago created a sea level

rise of tens of meters and massive flooding

around the Black Sea or milder changes (Giosan,

Filip, & Constatinescu, 2009).

The considerable river runoff into the Black

Sea has decreased by 15% in the last several

decades due to the diversion of the river water

for agricultural purposes. Observations in 1988

compared with 1969 (Figure S8.23b) showed

higher salinity by about 0.1 psu, ascribed to

the change in runoff (Murray et al., 1989). The

lower temperature in the surface layer in Figure

S8.23b is due to a month of observation in the

two years.

S8.10.4. Baltic and North Seas

The North Sea is the semi-enclosed, shallow,

continental shelf sea of about 100 m depth

between the British Isles, Norway, and Europe;

it is connected to the open North Atlantic

through a broad region between Scotland and

Norway at 61 e62 N and through Dover Strait

(Figure S8.24). The Baltic Sea is the nearly

enclosed sea east of Denmark. The Baltic is

connected to the North Sea at its southwest

end through a complex of passages with a sill

depth of 18 m, leading to the Kattegat and

the North Sea. The Kattegat is the small sea

between Denmark and Sweden. The Baltic,

which includes the Gulf of Bothnia to the north

and the Gulf of Finland to the east, is the

largest area of brackish (nearly fresh) water in

the ocean system. It has irregular bottom

topography, with a mean depth of 57 m, and

a number of basins of which the deepest is

459 m deep.

The physical oceanography of both seas was

reviewed in Rodhe (1998) and Rodhe, Tett, and

Wulff, (2006). Since 1992, the Baltic Sea has

been the focus of an intensive hydrological cycle

study called “Baltex” (the Baltic Sea Experiment);

as an outcome of the study, Leppäranta

and Myrberg (2009) provided a thorough overview

of Baltic Sea physical oceanography.

Surface salinity clearly illustrates the connection

of the North Sea to the open ocean and the

much greater isolation of the Baltic (Figure

S8.24b). Surface salinity in the North Sea is close

to oceanic values, with a tongue of high salinity

(>35) entering from the north. Through the Kattegat

and into the Baltic, there is an enormous

decrease, with salinity in the southern Baltic

between 7 and 8 psu, dropping to less than 2

psu in the northernmost Gulf of Bothnia and

easternmost Gulf of Finland.

The North Sea circulation is cyclonic, with

most water entering and leaving across the

continental shelf break in the north. The

exchange transport is about 2 Sv. Properties

22


and circulation along the western side are

strongly modulated by tides, which vertically

mix the inflowing waters, with some dilution

of salinity due to river inflow. Inflow from the

Baltic through the Kattegat introduces much

lower salinity waters into the North Sea; the

northward flow along the coast of Norway

that feeds the outflow is strongly stratified in

salinity as a result. Its circulation is estuarine

because of this net freshwater input.

The Baltic Sea, like the North Sea, has an estuarine

circulation (Figure S8.24c) with the upper

layer outflow in the Kattegat having a salinity

of 20 psu and the bottom layer inflow having

a salinity of 30e34 psu. Large-scale meteorological

conditions can override the estuarine circulation

and result in full depth inflow or outflow

at times.

The mean circulation in the Baltic Sea is weak

and cyclonic with mean surface currents of

about 5 cm/sec and no major stable features.

The complicated geography of this sea creates

a complex deeper circulation. However, the

circulation is extremely time dependent and

(a)

10˚W 5˚

0˚

5˚

10˚

15˚

20˚

25˚

30˚E

65˚

Gulf of Bothnia

65˚N

Faroes

60˚

Shetlands

Neva

River

Gulf of Finland

60˚

Norwegian Trench

North Sea

Skagerrak

Kattegat

Baltic Sea

55˚

55˚

Elbe R.

Rhine R.

50˚

Dover Strait

50˚

10˚W 5˚

0˚

5˚

10˚

15˚

20˚

25˚

30˚E

1000 2000 3000 4000 5000

FIGURE S8.24 North Sea and Baltic Sea. (a) Surface circulation schematic. After Winther and Johannessen (2006) and

Leppäranta and Myrberg (2009) with Etopo2 topography from NOAA NGDC (2008). Blue indicates subsurface flow through the

Kattegat. (b) Surface salinity in August in the Baltic and North Seas. Source: From Rodhe (1998). (c) Physical processes in the

Baltic. Source: From Winsor, Rodhe, and Omstedt. (2001).

ADJACENT SEAS 23

FIGURE S8.24

(Continued).

strongly coupled to the wind (through the

Ekman layer) because of the shallowness of

the sea; currents can reach 50 cm/sec in the

open sea and up to 100 cm/sec in straits during

storms (Leppäranta & Myrberg, 2009).

The very low salinity of the Baltic Sea results

primarily from river runoff and a long residence

time of waters within the Baltic of more

than 30 years (see Section 4.7). Evaporation

and precipitation are estimated to be nearly

24


equal at about 47 cm/year. The freshwater

budget is always positive, even on a monthly

level, because of the river inflows, which are

equivalent to 123 cm of water over the whole

area of the Baltic Sea. The bulk is due to the

Neva River into the Gulf of Finland, equivalent

to 400 cm/year over that gulf, and 170 cm/year

to the Gulf of Bothnia, with significant year-toyear

variations.

The Baltic is basically a two-layer system,

with a well-mixed upper layer (in terms of

salinity) that is 30e 50 m deep in the south,

increasing to 60e70 m in the central Baltic. The

upper layer temperature in summer is over 10

C with a thermocline at 15e20 m, which is

significantly shallower than the halocline.

Surface layer salinities are 6e 8 psu while the

deeper waters are usually 10e 13 psu, sometimes

exceeding 17 psu in the south when large

inflows occur. These bursts of inflow typically

occur no more than once a year, with the strongest

events sometimes separated by a decade or

more (Jakobsen, 1995). Surface salinity in the

gulfs is even lower at 2e7 psu. Dissolved

oxygen reaches 100% saturation in the surface

layers, but is relatively low in the deep water,

with variations on a decadal timescale related

to variations of inflows from the Kattegat.

Anoxic conditions occur in many of the deeper

Baltic basins where the residence times are

several years.

Starting in January, sea ice forms in the north

and east gulfs (Gulfs of Bothnia and Finland)

along the coast. The ice often extends to midgulf

but is less extensive in the central Baltic.

S8.10.5. Subtropical North Pacific

Marginal Seas

Along the western boundary of the tropical

and subtropical Pacific lies a set of marginal

seas that interact differently with the open

North Pacific circulation (Figure 10.1). From

south to north, these are the South China Sea,

the East China Sea, the Yellow Sea, and the

Japan (or East) Sea. Farther to the north lie the

subpolar Okhotsk and Bering Seas (Section

S8.10.6). Wind forcing for the southern part of

this region is strongly monsoonal, with the seasonality

weakening to the north. These seas are

connected through shallow straits or are located

on the continental shelf; the net transport

exchanges between them and with the Kuroshio,

which lies in deep water to the east, therefore

are limited and of the order of 2 Sv.

The South China Sea circulation is highly

seasonal, driven by the Asian monsoon. Inflow

from the Pacific occurs throughout the year

through Luzon Strait between Luzon and Taiwan;

these are Kuroshio waters. Exit is northward

through Taiwan Strait, between Taiwan

and the continent, also throughout the year.

The exchange rate is approximately 2 Sv (Xue

et al., 2004). Within the South China Sea, the

summer circulation is mostly anticyclonic,

driven by southwesterly winds, while in winter

it is mostly cyclonic, driven by northeasterly

winds (Hu, Kawamura, Hong, & Qi, 2000). In

Figure 10.1, only the cyclonic winter circulation

is depicted. The western boundary current

along Malaysia, Vietnam, Hainan, and southern

China reverses from southward in winter to

northward in summer. However, in the north,

the South China Sea Warm Current flows northward

through Taiwan Strait throughout the

year, with maximum transport in summer

when the full western boundary current

complex is northward.

The East China Sea and Yellow Sea constitute

the broad continental shelf region that lies east

of China, north of Taiwan, and west and south

of Korea. The eastern edge of the continental

shelf is the effective western boundary for the

North Pacific’s circulation, hence for the Kuroshio,

which flows northward along the continental

slope. Flow enters the East China Sea

from the South China Sea through Taiwan Strait

in the Taiwan Warm Current. Exit is to the north

into the Japan Sea through Tsushima Strait

(Korea Strait) in the Tsushima Warm Current.

ADJACENT SEAS 25

Within the Yellow Sea, the circulation is cyclonic

and much stronger in winter than in summer

due to the strong northerly monsoonal wind

forcing (Naimie, Blain, & Lynch, 2001). This

overall region also absorbs the major freshwater

output from the Changjiang River (Yangtze

River). Cross-shelf exchange with the Kuroshio

modifies the properties of the East China Sea

waters. The Bohai Sea, which is the gulf north

of the Yellow Sea, forms sea ice in winter; this

is the southernmost ice-covered region in the

Northern Hemisphere.

The Japan Sea (East Sea) lies between Asia and

Japan. It is deep with bottom depths exceeding

3000 m, but it is connected to the North Pacific

and Okhotsk Sea only through shallow straits.

Water enters the Japan Sea from the south

through Tsushima Strait (140 m deep). The

source of this warm, saline subtropical water is

the East China Sea with some possible input

from an onshore branch of the Kuroshio. The

net transport into the Japan Sea is estimated at

a little less than 2 Sv (Teague et al., 2006). Water

exits from the Japan Sea mainly through Tsugaru

Strait, between Honshu and Hokkaido

(130 m deep). There is also small but important

transport into the Okhotsk Sea through Soya

Strait between Hokkaido and Sakhalin, and

through the very shallow Tatar Strait far to the

north, between Siberia and Sakhalin.

Within the Japan Sea, there are typical

subtropical and subpolar circulations driven

by Ekman downwelling in the south and

upwelling in the north, and separated by a zonal

subarctic front that is similar to the North Pacific’s

subarctic front. The northward subtropical

western boundary current is the East Korean

Warm Current. The subpolar western boundary

current is the Primorye (or Liman) Current

where it flows along the coast of Russia and

the North Korean Cold Current where the flow

intrudes southward along the Korean coast.

The Japan Sea circulation deviates from

a typical open ocean gyre system because of

its vigorous eastern boundary current, the

Tsushima Warm Current, which flows northward

along the coast of Honshu. This results

from the “island effect,” which is related to

the wind forcing of the entire North Pacific

circulation east of Japan with open straits on

the southern and northern sides of the island

(this is outside the scope of this text).

A principal role of the Japan Sea in the North

Pacific circulation is to carry warm, saline

subtropical water northward west of Japan,

(cool and freshen it), and then expel the stillsaline

water north of the Kuroshio’s separation

point. This impacts details of formation of the

salinity minimum of North Pacific Intermediate

Water east of Japan (Section 10.9.2; review in

Talley et al., 2006).

S8.10.6. Bering and Okhotsk Seas

The Bering and Okhotsk Seas are separated

from the North Pacific by the long Aleutian

and Kuril Island chains. The North Pacific’s

cyclonic subpolar circulation partially loops

through these adjacent seas. The two seas are

intrinsically part of the North Pacific’s circulation,

but the island chains create leaky barriers

that partially support boundary currents and

a large amount of mixing in the island passages

due to tides. Both seas have sea ice formation

and brine rejection processes in the winter that

create denser shelf waters. Because the Okhotsk

Sea has a salty external source of water from the

Japan (East) Sea, through Soya Strait, its brine

rejection process produces denser water than

in the Bering Sea. The Bering Sea’s special role

is as a small conduit of Pacific waters to the

Atlantic Ocean.

In the Bering Sea, cyclonic circulation enters

from the Alaskan Stream beginning with the

easternmost passages through the Aleutians;

the principal deep inlet straits are Amchitka

Pass at about the date line (1155 m), and Near

Strait at 170 E (2000 m), just west of Attu Island

(Stabeno & Reed, 1995). Most of the exit flow

is through Kamchatka Strait (4420 m depth)

26


between Kamchatka and the Komadorskiy

Islands with a smaller amount exiting to the

Arctic through the shallow Bering Strait. The

Bering Sea is well known for its vigorous eddy

field that obscures much of the mean circulation

in synoptic data (Reed, 1995). Flow around

groups of islands in the Aleutian chain is anticyclonic

and tidally driven.

Within the Bering Sea, the inflow from the

central and eastern straits proceeds cyclonically,

with the principal northwestward flow

following the continental shelf topography as

the Bering Slope Current (Kinder, Coachman, &

Galt, 1975). When it encounters the Kamchatka

boundary, the Bering Slope Current splits. The

southward flow along the coast of Kamchatka

is the East Kamchatka Current (EKC), which

becomes the principal western boundary

current for the North Pacific’s subpolar gyre.

This is joined by water circulating in the deeper

basin of the Bering Sea, and exits to the North

Pacific following the Kamchatka coast. The

EKC outflow transport through Kamchatka

Strait is 6e12 Sv (Stabeno & Reed, 1995).

The flow of water northward through Bering

Strait to the Arctic Ocean is one of the principal

pathways in the global overturning circulation,

despite the shallowness of the strait (~50 m)

and its small transport (~0.8 Sv; Roach et al.,

1995). This is the only northern connection

between the Pacific and Atlantic. The flow is

relatively fresh (~32.5 psu) relative to global

mean salinities and is thus one of the freshwater

exits from the Pacific (Wijffels, Schmitt, Bryden,

& Stigebrandt, 1992; Talley, 2008). The water

flowing into Bering Strait comes from a western

boundary flow, the Anadyr Current, which is

the northward branch of the Bering Slope

Current, and a warmer eastern boundary flow,

the Alaskan Coastal Current, which is fed by

cyclonic circulation around the broad Bering

Sea shelf (Woodgate & Aagaard, 2005).

The Okhotsk Sea is connected to the North

Pacific through the Kuril Island chain. It is the

source of the densest water in the North Pacific,

which contributes to the North Pacific Intermediate

Water. The predominantly cyclonic circulation

of the Okhotsk Sea enters through the

northernmost strait close to the southern end

of the Kamchatka peninsula (Kruzenshtern

Strait, ~1400 m depth), and through Bussol’

Strait, which lies in the center of the Kurils

and is the deepest passage (~2300 m depth).

Net outflow is mainly through Bussol’ Strait,

which, like the other straits, has bidirectional

flow associated with anticyclonic flow around

each island. The net transport in and out of the

Okhotsk Sea is approximately 3e4 Sv(Gladyshev

et al., 2003. This water is greatly modified

within the Okhotsk Sea.

Within the Okhotsk Sea, the cyclonic circulation

flows northward along the western side of

Kamchatka as the West Kamchatka Current, and

westward along the broad continental shelves

on the northern Siberian boundary, where sea

ice formation produces especially dense shelf

waters (Section 10.9.2). The Amur River injects

fresh water in the northwest. The complex

then moves southward along the east coast of

Sakhalin, as the East Sakhalin Current, which is

a typical narrow western boundary current.

Waters from the East Sakhalin Current enter

the region south of Sakhalin, join an anticyclonic

circulation there, and then head for Bussol’

Strait. They are joined by eastward flow in the

Soya Current along the northern coast of Hokkaido

that enters the Okhotsk Sea from the

Japan Sea through Soya Strait.

S8.10.7. Red Sea and Persian Gulf

Geographically, the Red Sea, west of the

Arabian Peninsula, is a rift valley, resulting

from the separation of Africa and the Arabian

Peninsula, which is closed at the north and

opens to the Gulf of Aden, Arabian Sea, and

the Indian Ocean at the south through the

narrow strait of the Bab el Mandeb (or Bab al

Mandab; see Ross, 1983 in Ketchum). The

depth averages 560 m, with maximum values

ADJACENT SEAS 27

of2900mandasillofabout110mdepthatthe

BabelMandebinthesouth.

In contrast, the Persian or Arabian Gulf, east of

the Arabian Peninsula, is shallow with a

maximum depth of 105 m and average depth of

35 m (Swift & Bower, 2002). The Persian Gulf is

connected to the Arabian Sea through the Strait

of Hormuz (86 m deep) and the Gulf of Oman.

This brief introduction to these marginal seas

is well complemented by the greater detail in

Tomczak and Godfrey (1994).

A major aspect of the northwestern Arabian

Sea is the high evaporation rate of approximately

100e 200 cm/year, while precipitation

averages about 7e10 cm/year. There are no

major rivers flowing into the Red Sea. The Tigris

and Euphrates Rivers drain into the Persian

Gulf, but their freshwater contribution is far

smaller than the net evaporation.

The water structure in the Red Sea consists of

a shallow upper layer and a thick deep layer

separated by a thermocline/halocline at about

200 m depth. At the surface, the temperature

in summer (June eSeptember) is 26 e30 C,

and in winter (October eMay) it is 24e28 C.

Below the thermocline the deep layer is nearly

isothermal, at 21.6e21.8 C. The Red Sea is the

most saline large body of ocean water, with

surface layer values of 38e 40 psu (with higher

values to 42.5 psu in the north) and deep water

values of 40.5e40.6 psu. The deep water is

formed by winter cooling in the north. The

surface layer is saturated with dissolved

oxygen, but the absolute values are low because

of the high temperature (less than 175 mmol/kg).

There is an oxygen minimum of 20e60 mmol/kg

at 400 m below the thermocline/halocline,

whereas the deep water below this has a content

of 80e90 mmol/kg.

A schematic Red Sea mean circulation is presented

in Figure S8.25a. The central Red Sea

ocean surface pressure (sea level) is dominated

by two highs with flanking lows in the north

and south. Intermediate water forms in the

northern low. All of the mean boundary

FIGURE S8.25 Schematic circulations: (a) Red Sea and

(b) Persian Gulf. Source: From Johns et al. (1999).

28


currents flow to the north, reflecting the net

thermohaline overturn in the Red Sea.

The Red Sea circulation’s seasonal variation is

related to the winds (Johns et al., 1999; Sofianos &

Johns, 2003). In summer (Southwest Monsoon)

the winds are to the south over the whole Sea

(Figure S8.26); the surface flow is southward

with outflow through the Bab el Mandeb, while

there is a subsurface inflow to the north and

weak outflow at the bottom through that strait.

In the winter (Northeast Monsoon) the winds

over the southern half of the Red Sea change to

northward; there is a northward surface flow

over the whole of the Red Sea and a subsurface

southward flow with outflow through the Bab

el Mandeb. The residence time for the upper

layer has been estimated at 6 years and for the

deep water at 200 years.

Hot brine pools are found in some of the deepest

parts of the Red Sea (Karbe, 1987). Very high

temperatures of 58 C and salinities of 320 psu are

due to hydrothermal activity. The heat flow up

through the bottom is much greater than the

world average of 4 10 2 W/m 2 . The high

FIGURE S8.26 Wind stress (dyn/cm 2 ) for the Red Sea and Persian Gulf: (a) January (Northeast Monsoon) and (b) July

(Southwest Monsoon). Sources: From Johns et al. (1999); see also Sofianos and Johns (2003).

REFERENCES 29

salinity value is not directly comparable to ocean

water salinities because the chemical constitution

of these brines is quite different. They have

a much higher content of metal ions. (For

comparison, a saturated solution of sodium chloride

in water has a salinity value in the oceanographic

sense of about 270.) The favored

explanation for the origin of the chemical constituents

is that this is interstitial water from sediments

or solutions in water of crystallization

from solid materials in the sea bottom, which

are released by heating from below and forced

out through cracks into the deep basins of the

Red Sea.

Circulation and formation of hypersaline

water in winter in the Persian Gulf was briefly

summarized in Section 11.6, based on Johns et

al. (2003) and Swift and Bower (2003). The

cyclonic circulation, exchange through the

Straits of Hormuz, and formation region in the

south of the high salinity water mass are illustrated

in Figure S8.25b. Winter surface salinity

in the southern region exceeds 42 psu. Because

the Persian Gulf is so shallow, bottom salinity

largely mirrors surface salinity. Inflows of

fresher waters from the Gulf of Oman in the

southeast and from the Tigris and Euphrates

rivers in the northwest bracket the high salinity

region along the coast of the Arabian Peninsula.

Outflow of the dense water occurs throughout

the year, with only a weak seasonal signal, at

a mean salinity of 39.5 psu (Johns et al., 2003).

Part of the outflow occurs in the surface layer

in the southern part of the Straits of Hormuz.

The surface layer transport has a strong seasonal

cycle.

References

Armi, L., Farmer, D.M., 1988. The flow of Mediterranean

Water through the Strait of Gibraltar. Progr. Oceanogr.

21, 1e105 (Also Farmer and Armi, 1988.).

Beardsley, R.C., Boicourt, W.C., 1981. On estuarine and

continental-shelf circulation in the Middle Atlantic

Bight. In: Warren, B.A., Wunsch, C. (Eds.), Evolution of

Physical Oceanography. MIT Press, Cambridge, MA,

pp. 198e223.

Bray, N., Ochoa, J., Kinder, T., 1995. The role of the interface

in exchange through the Strait of Gibraltar. J. Geophys.

Res. 100, 10755e10776.

Bryden, H.L., Candela, J., Kinder, T.H., 1994. Exchange

through the Strait of Gibraltar. Progr. Oceanogr. 33,

201e248.

Cameron, W.M., Pritchard, D.W., 1963. Estuaries. In:

Hill, M.N. (Ed.), Ideas and Observations. The Sea, Vol. 2.

Wiley-Interscience, pp. 306e324.

CIESM, 2001. CIESM Round table session on Mediterranean

water mass acronyms. 36th CIESM Congress, Monte

Carlo, 26 September 2001. https://www.ciesm.org/

catalog/WaterMassAcronyms.pdf (accessed 6.5.09).

Curray, J.R., Emmel, F.J., Moore, D.G., 2003. The Bengal Fan:

Morphology, geometry, stratigraphy, history and

processes. Mar. Petrol. Geol. 19, 1191e1223.

Dai, A., Trenberth, K.E., 2002. Estimates of freshwater

discharge from continents: Latitudinal and seasonal

variations. J. Hydromet. 3, 660e687.

Dyer, K.R., 1997. Estuaries: A Physical Introduction, second

ed. Wiley, New York, p. 195.

Farmer, D.M., Freeland, H.J., 1983. The physical oceanography

of fjords. Progr. Oceanogr. 12, 147e219.

Giosan, L., Filip, F., Constatinescu, S., 2009. Was the Black

Sea catastrophically flooded in the early Holocene?

Quaternary Sci. Rev. 28, 1e6.

Gladyshev, S., Talley, L., Kantakov, G., Khen, G.,

Wakatsuchi, M., 2003. Distribution, formation and

seasonal variability of Okhotsk Sea Intermediate Water.

J. Geophys. Res. 108 (C6), 3186. doi:10.1029/2001JC

000877.

Hardisty, J., 2007. Estuaries: Monitoring and Modeling the

Physical System. Blackwell Publishing, Maiden, MA.

p. 157.

Hu, J., Kawamura, H., Hong, H., Qi, Y., 2000. A review on

the currents in the South China Sea: Seasonal circulation,

South China Sea Warm Current and Kuroshio intrusion.

J. Oceanogr. 56, 607e624.

Jakobsen, F., 1995. The major inflow to the Baltic Sea during

January 1993. J. Marine Syst. 6, 227e240.

Johns, W.E., Jacobs, G.A., Kindle, J.C., Murray, S.P.,

Carron, M., 1999. Arabian Marginal Seas and Gulfs:

Report of a Workshop held at Stennis Space Center,

Miss. 11e13 May, 1999. University of Miami RSMAS.

Technical Report 2000e01.

Karbe, L., 1987. Hot brines and the deep sea environment.

In: Edwards, A.J., Head, S.M. (Eds.), Red Sea. Pergamon

Press, Oxford, p. 441.




30


Kinder, T.H., Coachman, L.K., Galt, J.A., 1975. The Bering

Slope Current System. J. Phys. Oceanogr. 5, 231e244.

Klein, B., Roether, W., Civitarese, G., Gacic, M., Manca, B.B.,

d’Alcalá, M.R., 2000. Is the Adriatic returning to dominate

the production of Eastern Mediterranean Deep

Water? Geophys. Res. Lett. 27, 3377e3380.


Kovacevic, V., Luchetta, A., 1999. The large deep water

transient in the Eastern Mediterranean. Deep-Sea Res. I

46, 371e414.

Leppäranta, M., Myrberg, K., 2009. Physical Oceanography

of the Baltic Sea. Springer, Berlin, p. 378 with online

version.

Malanotte-Rizzoli, P., Manca, B.B., Salvatore Marullo, Ribera

d’ Alcalá, M., Roether, W., Theocharis, A., Bergamasco, A.,

Budillon, G., Sansone, E., Civitarese, F. Conversano, F.,

Gertman, I., Hernt, B., Kress, N., Kioroglou, S., Kontoyannis,

H., Nittis, K., Klein, B., Lascaratos, A., Latif,

M.A., Özsoy, E., Robinson, A.R., Santoleri, R., Viezzoli, D.,

Kovacevic, V., 2003. The Levantine Intermediate Water

Experiment (LIWEX) Group: Levantine basin d Alaboratory

for multiple water mass formation processes. J.

Geophys. Res. 108(C9), 8101. doi:10.1029/2002JC001643.

Marshall, J., Schott, F., 1999. Open-ocean convection: observations,

theory, and models. Rev. Geophys. 37, 1e64.

MEDOC Group, 1970. Observations of formation of deepwater

in the Mediterranean Sea, 1969. Nature 227,

1037e1040.

Millot, C., 1991. Mesoscale and seasonal variabilities of the

circulation in the western Mediterranean. Dynam.


Millot, C., Taupier-Letage, I., 2005. Circulation in the

Mediterranean Sea. In: Saliot, E.A. (Ed.), The Handbook

of Environmental Chemistry, Vol. 5. Part K. Springer-

Verlag, Berlin Heidelberg, pp. 29e66.

Monismith, S.G., 2007. Hydrodynamics of coral reefs. Annu.

Rev. Fluid Mech. 39, 37e55.

Murray, J.W., Jannasch, H.W., Honjo, S., Anderson, R.F.,

Reeburgh, W.S., Top, Z., Friedrich, G.E., Codispoti, L.A.,

Izdar, E., 1989. Unexpected changes in the oxic/anoxic

interface in the Black Sea. Nature 338, 411e413.

Naimie, C.E., Blain, C.A., Lynch, D.R., 2001. Seasonal mean

circulation in the Yellow Sea d A model-generated

climatology. Continental Shelf Res. 21, 667e695.

NASA Goddard Earth Sciences, 2007c. Sedimentia. NASA

Goddard Earth Sciences Ocean Color. http://disc.gsfc.

nasa.gov/oceancolor/scifocus/oceanColor/sedimentia.







Neilson, B.J., Kuo, A., Brubaker, J., 1989. Estuarine Circulation.

Humana Press, Clifton, N.J., p. 377.

Noaa Ngdc, 2008. Global Relief Data d ETOPO. NOAA

National Geophysical Data Center. http://www.ngdc.

noaa.gov/mgg/global/global.html (accessed 9.24.08).

Officer, C.B., 1976. Physical Oceanography of Estuaries (and

Associated Coastal Waters). Wiley, New York, p. 465.

Oguz, T., Tugrul, S., Kideys, A.E., Ediger, V., Kubilay, N.,

2006. Physical and biogeochemical characteristics of the

Black Sea. In: Robinson, A.R., Brink, K.H. (Eds.), The Sea.


Studies and Syntheses, Vol., 14A. Harvard University

Press, pp. 1333e1372.

Özsoy, E., Hecht, A., Ünlüata, Ü., Brenner, S., Oguz, T.,

Bishop, J., Latif, M.A., Rozentraub, Z., 1991. A review of

the Levantine Basin circulation and its variability during

1985e1988. Dynam. Atmos. Oceans 15, 421e456.

Özsoy, E., Ünlüata, U., 1998. The Black Sea. In:

Robinson, A.R., Brink, K.H. (Eds.), The Sea. The Global

Coastal Ocean: Regional Studies and Syntheses, Vol. 11.

Harvard University Press, pp. 889e914.

Pickard, G.L., 1961. Oceanographic features of inlets in the

British Columbia mainland coast. J. Fish. Res. Bd. Can.

18, 907e999.

Pickard, G.L., Donguy, J.R., Hénin, C., Rougerie, F., 1977. A

review of the physical oceanography of the Great Barrier

Reef and western Coral Sea. Australian Institute of

Marine Science, 2. Australian Government Publishing

Service, p. 134.

Pickard, G.L., Stanton, B.R., 1980. Pacific fjords d A review

of their water characteristics, pp. 1e51. In: Freeland, H.J.,

Farmer, D.M., Levings, C.D. (Eds.), Fjord Oceanography.

Plenum Press.

Potter, R.A., Lozier, M.S., 2004. On the warming and salinification

of the Mediterranean outflow waters in the

North Atlantic. Geophys. Res. Lett. 31, L01202.

doi:10.1029/2003GL018161.


production by marginal seas. Progr. Oceanogr. 33,

161e200.

Pritchard, D.W., 1989. Estuarine classification d A help or

a hindrance. In: Neilson, B.J., Kuo, A., Brubaker, J. (Eds.),

Estuarine Circulation. Humana Press, Clifton, N.J,

pp. 1e38.

Reed, R.K., 1995. On geostrophic reference levels in the

Bering Sea basin. J. Oceanogr. 51, 489e498.

Roach, A.T., Aagaard, K., Pease, C.H., Salo, S.A.,

Weingartner, T., Pavlov, V., Kulakov, M., 1995. Direct

measurements of transport and water properties through

the Bering Strait. J. Geophys. Res. 100, 18443e18458.

Robinson, A.R., Golnaraghi, M., Leslie, W.G., Artegiani, A.,

Hecht, A., Lazzoni, E., Michelato, A., Sansone, E.,

Theocharis, A., Ünlüata, Ü, 1991. The eastern

REFERENCES 31

Mediterranean general circulation: features, structure

and variability. Dynam. Atmos. Oceans 15, 215e240.

Rodhe, J., 1998. The Baltic and North Seas: A processoriented

review of the physical oceanography. In:

Robinson, A.R., Brink, K.H. (Eds.), The Sea. The Global

Coastal Ocean: Regional Studies and Syntheses, Vol. 11.

Harvard University Press, pp. 699e732.

Rodhe, J., Tett, P., Wulff, F., 2006. The Baltic and North Seas:

A regional review of some important physical-chemicalbiological

interaction processes. In: Robinson, A.R.,

Brink, K.H. (Eds.), The Sea. The Global Coastal Ocean:

Interdisciplinary Regional Studies and Syntheses, Vol.,

14A. Harvard University Press, pp. 1033e1076.

Ross, D.A., 1983. The Red Sea. In: Ketchum, B.H. (Ed.),

Estuaries and Enclosed Seas. Ecosystems of the World,

26. Elsevier, pp. 293e307.

Rougerie, F., 1986. Le lagon sud-ouest de Nouvelle-

Calédonie: spécificité hydrologique, dynamique et productivité.

Etudes et Thèses. ORSTOM, Paris, p. 234.

Sankey, T., 1973. The formation of deep water in the Northwestern

Mediterranean. Progr. Oceanogr. 6, 159e179.

Schlitzer, R., Roether, W., Oster, H., Junghans, H.-G.,

Hausmann, M., Johannsen, H., Michelato, A., 1991.

Chlorofluoromethane and oxygen in the Eastern Mediterranean.

Deep-Sea Res. 38, 1531e1551.

Sofianos, S.S., Johns, W.E., 2003. An Oceanic General

Circulation Model (OGCM) investigation of the Red Sea

circulation: 2. Three-dimensional circulation in the Red

Sea. J. Geoph. Res. 108, 3066. doi: 10, 1029/200IJC001185.

Stabeno, P.J., Reed, R.K., 1995. Circulation in the Bering Sea

basin observed by satellite-tracked drifters: 1986e1993. J.

Phys. Oceanogr. 24, 848e854.

Swift, S.A., Bower, A.S., 2003. Formation and circulation of

dense water in the Persian/Arabian Gulf. J. Geophys.

Res. 108 (C10). doi:10.1029/2002JC001360.




doi:10.1016/j.pocean.2008.05.001.

Talley, L.D., Min, D.-H., Lobanov, V.B., Luchin, V.A.,

Ponomarev, V.I., Salyuk, A.N., Shcherbina, A.Y.,

Tishchenko, P.Y., Zhabin, I., 2006. Japan/East Sea water

masses and their relation to the sea’s circulation.


Teague, W.J., Ko, D.S., Jacobs, G.A., Perkins, H.T.,

Book, J.W., Smith, S.R., Chang, K.-I., Suk, M.-S., Kim, K.,

Lyu, S.J., Tang, T.Y., 2006. Currents through the Korea/

Tsushima Strait. Oceanography 19, 50e63.

Timmermans, M.L., Garrett, C., Carmack, E., 2003. The

thermohaline structure and evolution of the deep waters

in the Canada Basin, Arctic Ocean. Deep-Sea Res. I 50,

1305e1321.

Tomczak, M., Godfrey, J.S., 1994. Regional Oceanography,

an Introduction. Pergamon Press, Oxford, UK, p. 422.

Tully, J.P., 1949. Oceanography and prediction of pulp-mill

pollution in Alberni Inlet. Fish. Res. Bd. Can. Bull. 83,

169.

Wijffels, S.E., Schmitt, R.W., Bryden, H.L., Stigebrandt, A.,

1992. Transport of fresh water by the oceans. J. Phys.


Winsor, P., Rodhe, J., Omstedt, A., 2001. Baltic Sea ocean

climate: An analysis of 100 yr of hydrographic data with

focus on freshwater budget. Climate Res. 18, 5e15.

Winther, N.G., Johannessen, J.A., 2006. North Sea circulation:

Atlantic inflow and its destination. J. Geophys. Res.

111 C12018. doi:10.1029/2005JC003310.

Wolanski, E. (Ed.), 2001. Oceanographic Processes of Coral

Reefs: Physical and Biological Links in the Great Barrier

Reef. CRC Press, Boca Raton, FL, p. 356.

Woodgate, R.A., Aagaard, K., 2005. Revising the Bering

Strait freshwater flux into the Arctic Ocean. Geophys.

Res. Lett. 32 L02602. doi:10.1029/2004GL021747.

Wüst, G., 1961. On the vertical circulation of the Mediterranean

Sea. J. Geophys. Res. 66, 3261e3271.

Xue, H., Chai, F., Pettigrew, N., Xu, D., Shi, M., Xu, J., 2004.

Kuroshio intrusion and the circulation in the South

China Sea. J. Geophys. Res. 109 C02017. doi:10.1029/

2002JC001724.

C H A P T E R

S9

Atlantic Ocean: Supplementary

Materials

FIGURE S9.1 Atlantic Ocean surface height (cm) and surface current names (Table S9.1). Data from Niiler, Maximenko,

and McWilliams (2003).

1

2

S9. ATLANTIC OCEAN: SUPPLEMENTARY MATERIALS

FIGURE S9.2 Geostrophic circulation at (a) 250 dbar, (b) 1000 dbar, and (c) 1500 dbar. The contours are steric height (10 m 2 s 2 ), adjusted to represent

the absolute circulation. Source: From Reid (1994).

S9. ATLANTIC OCEAN: SUPPLEMENTARY MATERIALS 3

FIGURE S9.2 (Continued).

4


FIGURE S9.3 Annual mean winds. (a) Wind stress (N/m 2 ) (vectors) and wind-stress curl ( 10 7 N/m 3 ) (color),

multiplied by 1 in the Southern Hemisphere. (b) Sverdrup transport (Sv), where blue is clockwise and yellow-red is

counterclockwise circulation. Data are from the NCEP reanalysis 1968e1996 (Kalnay et al., 1996).


FIGURE S9.4 Annual mean buoyancy forcing, using fluxes for 1997e2006. Data are from Large and Yeager (2009). (a) Net

airesea heat flux (W/m 2 ). (b) Buoyancy forcing (equivalent W/m 2 ). (c,d) Net evaporation minus (precipitation + runoff)

(cm/yr and equivalent W/m 2 ). Values less than 10 W/m 2 are white.

6


FIGURE S9.5 (a) Modeled transport streamfunction. Source: From Johns, Townsend, Fratantoni, and Wilson (2002). (b) Mean

velocity from surface drifters (1968e2003); velocities greater than 25 cm/sec are in red. Source: From Richardson (2005).


FIGURE S9.6 Flow through Yucatan Channel. (a) Mooring locations for August 1999 to June 2001. (b) Mean northward

velocity (cm/sec). Gray is northward flow, white is southward flow. Source: From Candela et al. (2003).

8


FIGURE S9.7 (a) Florida Current, Gulf Stream, Deep Western Boundary Current, and common eddy features. Source:

From Schmitz (1996), after Bane (1994). (b) Average 25e75 m velocity from one crossing. Source: From Beal et al. (2008).

(c) Annual cycle and (d) long-term record of Florida Current transports (solid) with the monthly mean North Atlantic

Oscillation index (dashed). Source: From Baringer and Larsen (2001).


FIGURE S9.8 Florida Strait in September, 1981. (a) Geostrophic velocity (cm/sec). ÓAmerican Meteorological Society.

Reprinted with permission. Source: From Roemmich (1983). (b) Potential density s q (kg/m 3 ), (c) potential temperature ( C),

and (d) salinity.

10


(a)

0

200

Sargasso

Sea

24

25

25.5

26

26.1

26.2

26.3

Gulf Stream

Slope Water

(b)

0

200

Sargasso

Sea

21

20

19

Gulf Stream

28

25

22

Slope Water

23

26.4

400

400

18

17

26.5

16

26.6

26.7

26.8

15

14

600

26.9

27

600

13

12

800

27.1

27.2

27.3

27.4

800

11

10

9

8

5

4.8

4.6

27.5

7

4.4

27.6 27.7

1000

0 50 100 150 200 250 300 350 400 km

6

4.2

1000

0 50 100 150 200 250 300 350 400 km

(c)

0

200

37°N 38 39 40°N

Potential density σθ at 66°W

36.1

34.1

36.5

35

36.6 36.7

36.8

36.4

36.2

(d)

σθ

23.0

23.5

24.0

37°N 38 39 40°N

Potential temperature (°C) 66°W

28

27

26

24

22

23

24.5

25

400

25.0

25.5

22

21

20

18

17

600

800

36.1

36

35.8

35.6

35.5

35.4

35.3

35.2

35.1

35

26.0

26.5

27.0

14

12

10

15

16

34.98

27.5

8

6

5

1000

0 50 100 150 200 250 300 350 400 km

28.0

0 50 100 150 200 250 300 350 400 km

37°N 38 39 40°N

Salinity 66°W

37°N 38 39 40°N

Potential temperature (°C) 66°W

FIGURE S9.9 Gulf Stream properties at 66 W (World Ocean Circulation Experiment section A22 in August, 1997).

(a) Potential density s q , (b) potential temperature ( C), (c) salinity, (d) potential temperature with potential density s q as

the vertical coordinate.


FIGURE S9.10 North Atlantic Current formation region: (a) including the Deep Western Boundary Current (DWBC)

ÓAmerican Meteorological Society. Reprinted with permission. Source: From Pickart, McKee, Torres, and Harrington (1999), and

(b) including eastward detrainments. NAC (North Atlantic Current), GS (Gulf Stream), LC (Labrador Current), SPF

(Subpolar Front), FC (Flemish Cap), TGB (Tail of the Grand Banks). Source: From Rossby (1999).

12


FIGURE S9.11 Eddy kinetic energy of the North Atlantic Ocean (cm 2 /s 2 ) from surface drifter observations from 1990 to

1999. Source: From Fratantoni (2001).


FIGURE S9.12 Gulf Stream rings. (a) Schematic of cold core ring formation from multi-ship observations in Operation

Cabot. After Parker (1971). (b,c) Warm core ring observations in April 1982: velocity at 28 m depth and depth of the 10 C

isotherm, and azimuthal velocity near the sea surface (bars) with modeled velocity (solid curve). Source: From Olson, Schmitt,

Kennelly, and Joyce (1985).

14


FIGURE S9.13 (a) Surface circulation of the South Atlantic using a temperature/salinity climatology and a b-spiral

inverse. ÓAmerican Meteorological Society. Reprinted with permission. Source: From Juliano and Alvés (2007) with labels

added. (b) Streamfunction for the barotropic flow (contours), with pathways of eddies and Rossby waves (yellow) and

Kelvin waves (red). Shading is the eddy kinetic energy where it is especially high. Source: From Biastoch, Böning, and

Lutjeharms (2008).


FIGURE S9.14 (a) Schematic of currents and fronts in the confluence region of the Malvinas (Falkland) and Brazil

Currents. Source: From Peterson (1992). (b) Brazil Current transports, in the thermocline layer (Central Water), based on

hydrographic sections (straight lines). Each solid contour is 5 Sv. After Zemba (1991). (c) Location and trajectories of warm

core Brazil Current rings from 1993 to 1998, superimposed on the rms sea-surface height variability from altimetry data.

Source: From Lentini, Goni, and Olson (2006).

16


FIGURE S9.15 Agulhas ring “Astrid” in March 2000. (a) SST satellite image. Source: From Peeters et al. (2004). (b) Velocity

(m/s) from an LADCP, (c) potential temperature, and (d) salinity. Source: From van Aken et al. (2003).


FIGURE S9.15

(Continued).

18


FIGURE S9.16 Deep Gulf Stream structure. (a,b) Mean velocity at 700 and 2000 m, averaged in 1 bins, from acoustically

tracked floats. Vectors show mean flow direction and speed; ellipses are the variance. Source: From Owens (1991). (c) Mean

velocities at 4000 m from current meter observations in the 1970s and suggested streamlines. Source: From Hogg (1983).


FIGURE S9.17 Deep Brazil and Malvinas Current structure. (a) Mean velocity and (b) circulation schematic at intermediate

depth (650e1050 m) based on subsurface floats from different experiments during 1989e1996. Source: From Boebel

et al. (1999).

20


FIGURE S9.18 Mid-depth Labrador Sea circulation. (a) Velocity (cm/sec) and (b) geostrophic pressure (cm) at 700 m

from profiling float observations. Source: From Lavender, Davis, and Owens. (2000).


FIGURE S9.19 Mid-depth subpolar circulation. Mean transport streamfunctions on isopycnal s q ¼ 27.5 kg/m 3 , based on

acoustically tracked isopycnal floats. Source: From Bower et al.(2002).

22


FIGURE S9.20 Mean velocities from current meter deployments in Denmark Strait and along the coast of Greenland

from 1986 to 1991. Left to right: Cape Farewell (southern tip of Greenland), 63, 64, and 65 S. Small inset at bottom: just south

of the strait. The map shows the location of each line of moorings. Source: From Dickson and Brown (1994).


FIGURE S9.21 Deep Western Boundary Current (DWBC) east of the Grand Banks. (a) Mean velocity (color) and

transports (numbers in Sv) and (b) transport time series for the DWBC, all deep water and the North Atlantic Current, from

current meters at 42 N, 45 W east of the Grand Banks (location in Figure 9.44). Acronyms: LSW, Labrador Sea Water; uLSW,

upper LSW; DSOW, Denmark Strait Overflow Water; and GFZW, Gibbs Fracture Zone Water, which is called Northeast

Atlantic Deep Water or Iceland Scotland Overflow Water by others. ÓAmerican Meteorological Society. Reprinted with

permission. Source: From Schott et al. (2004).

24


FIGURE S9.22 DWBC east of Florida. (a) DWBC and Antilles Current mean velocity section (Lowered ADCP observations)

and (b) location of Abaco moorings (26.5 N). ÓAmerican Meteorological Society. Reprinted with permission.

Source: From Johns et al. (2008).


FIGURE S9.23 DWBC east of Brazil. (a) Schematic of the DWBC (blue) and its breakup into eddies south of 8 S.

Acronyms for currents are as in Table S9.3. Source: From Dengler et al. (2004). (b) NADW (2500 m) flows in the Brazil Basin

based on acoustically tracked floats: float displacements over 800 days. Source: From Hogg and Owens (1999). (c) Mean

velocities at current meters at the 30 S moored array across Vema and Hunter Channels. Shaded regions are northward flow.

ÓAmerican Meteorological Society. Reprinted with permission. Source: From Hogg, Siedler, & Zenk (1999).

26


FIGURE S9.23

(Continued).

FIGURE S9.24 Meridional transport (geostrophic velocity) at 24e26.5 N in the North Atlantic for five different years.

(a) Full depth, (b) 0 to 1000 m, and (c) 1000 m to bottom. Source: From Bryden, Longworth, and Cunningham (2005b).


FIGURE S9.25 Northward transports (Sv) across zonal sections in isopycnal layers, integrated upward from zero at the

bottom. Section latitudes are indicated in parentheses. Ekman transport is not included. Gray indicates the uncertainty.

Source: From Ganachaud (2003).

28


FIGURE S9.26

Schematic temperature d salinity diagram for the main water masses of the Atlantic Ocean.


FIGURE S9.27 Atlantic Ocean: mean T-S curves and one standard deviation curves by 5 squares. Extended caption: The

largest variability (large standard deviation) is at 40e50 N off the east coast of North America, due to the Gulf Stream and

North Atlantic Current, which each separate strongly contrasting water masses (fresh, cold on the north/west side and

saltier, warmer on the south/east side). At 4e6 C the salinity minimum of the AAIW is well marked in the tropics. It erodes

to the north, losing its character by about 20 N. The North Atlantic Central Water connects the AAIW to the high salinity

near-surface or surface waters. The STUW (near-surface salinity maximum) is present throughout the tropics south of 20 N.

Salinity is highest at the sea surface in the central subtropical gyre; this is the surface source region of the subducted STUW

to the south. The Gulf Stream system also has a near-surface salinity maximum, due to northward advection of STUW and

saline Central Water, which is overrun by fresher slope water. At mid-latitudes, off the Strait of Gibraltar, the saline outflow

from the Mediterranean leads to the salinity maximum of the Mediterranean Water at mid-depth at about 10 C. From the

sharp bend in the T-S curves this maximum can be traced as it spreads north, west, and south. In northern latitudes, all

temperatures are below 15 C. In the Labrador Sea, the upper ocean waters are colder (and fresher) than the underlying

salinity maximum of the NEADW and NSOW. In contrast, the waters in the central far North Atlantic appear almost isohaline

over the entire temperature range.

30


FIGURE S9.28 Subtropical Underwater. Vertical sections of (a) salinity and (b) oxygen (mmol/kg) with selected potential

density contours, along approximately 25 W in the Atlantic Ocean. (c) Salinity at s q ¼ 25.0 kg/m 3 .


FIGURE S9.29 South Atlantic Subtropical Mode Water. (a) Thickness of the 14e16 C layer. (b) Vertical temperature

derivative along the east-west section in red on the map; 12, 14, 16, and 18 C isotherms are indicated as black contours.

Source: From Provost et al. (1999).

32


FIGURE S9.30 Subpolar Mode Water (SPMW). (a) Potential density (s q ) and (b) thickness (m) of the late winter mixed

layer, shown only where the mixed layer is more than 200 m thick. This thick mixed layer is the SPMW. ÓAmerican

Meteorological Society. Reprinted with permission. Source: From McCartney and Talley (1982).


FIGURE S9.31 Labrador Sea Water. (a) Mixed layer depths in winter (1996e1998) from profiling floats. Red: >800 m.

Blue: 400 to 800 m. Green: <400 m. Source: From Lavender et al. (2000). (b) Potential density profiles from a deep convection

region in the Labrador Sea in late winter of 1997 (“119” in the deep convection patch, “62” in a western boundary convection

regime, “59” typical of stratified water). (c) Vertical section through the Labrador Sea in late winter 1997 that includes deep

convection stations. ÓAmerican Meteorological Society. Reprinted with permission. Source: From Pickart, Torres, and Clarke

(2002).

34


FIGURE S9.32 Meddies. (a) Occurrences in historical hydrographic data, superimposed on a map of salinity anomaly of

Mediterranean Water near 1100 m depth. (b) Float trajectories in Meddies. Source: From Richardson, Bower, and Zenk (2000).


(a)

0

(b)

0

500

500

Pressure

1000

Meddy profiles

1000

50˚N

35.4

35.3

35.7

35.8

1500

40˚

35.1

35.2

35.5

36.3

1500

35.6

35.4

(26°N, 26°W)

30˚

35.3

35.1

35.2

2000

35.0 35.5 36.0 36.5 37.0

Salinity

20˚

50˚W 40˚ 30˚ 20˚ 10˚W

2000

26.5 27.0 27.5 28.0

Potential density

FIGURE S9.33 Meddy structure. (a) Salinity and (b) potential density through the only two Meddies found on a synoptic

section in 1988, with adjacent profiles, at 26 N, 26 W (large dots on inset map). Inset map: salinity contoured at potential

density referenced to 1000 dbar ¼ 32.2 kg/m 3 (around s q ¼ 27.65 kg/m 3 ), with all stations from the 1988 section. After

Tsuchiya, Talley, and McCartney (1992).

36


FIGURE S9.34 Antarctic Intermediate Water. (a) Salinity at the AAIW salinity minimum. (b) Oxygen (ml/L) at the AAIW

oxygen maximum. Source: From Talley (1996b).


FIGURE S9.35 North Atlantic Deep Water and Circumpolar Deep Water. Oxygen (ml/L) on the isopycnal s 3 ¼ 41.44 kg/m 3

(referenced to 3000 dbar), which lies at approximately 2500 m depth. Source: From Reid (1994).

0

0

35.7

220

100

35.1

36

35.2

34.7

100

80

500 34.6 34.65

500

34.5

120

34.8

34.6

35.5

140

160

100

34.65

140

1000

34.7

1000

180

160

34.8

190

34.85

180

34.9

200

1500

1500

34.97 35

240

230

2000

34.96

2000

255

34.94

250

34.95

240

250

250

2500

2500

34.92

245

34.93

250

34.91

250

2

250

3000

3000

245

34.9

34.91

255 255

34.9

9

3500

3500

260

245

34.85

245

240

4000

34.8

4000

240

235

245

4500

34.75

4500

230

255

250

5000

5000

225

34.7

5500

5500

225

245

Salinity

Oxygen

6000

6000

30°S 20° 10° 0° 10° 20° 30°N 30°S 20° 10° 0° 10° 20° 30°N

Mid-Atlantic Ridge

FIGURE S9.36 Upper, middle, and lower NADW in the tropical Atlantic. (a) Salinity and (b) oxygen (mmol/kg) along

25 W. Data were collected in 1988e1989. (World Ocean Circulation Experiment section A16).

38


TABLE S9.1

Major Upper Ocean Circulation Systems, Currents and Fronts of the Subtropical North Atlantic

(Figures 9.1 and S9.1 ) *

Name Description Approximate Latitudes

Subtropical gyre Anticyclonic gyre at mid-latitudes 10 e40 N

North Equatorial

Current (NEC)

Westward flow of the subtropical gyre and northern

tropical gyre

10e20 N

Gulf Stream System Subtropical western boundary current complex 10e40 N

Caribbean Current

Yucatan Current

Loop Current (LC)

Antilles Current

Florida Current

Gulf Stream

Subtropical western boundary current portion

within the Caribbean Sea


passing through Yucatan Channel


looping through the Gulf of Mexico

Subtropical western boundary current portion east

of the Antilles and Bahamas


through Florida Strait

Subtropical western boundary current north of

Florida Strait and the separated extension of the

subtropical western boundary current

10e22 N

20e23 N

23e27 N

18e28 N

12e28 N

28e35 N

Gulf Stream Extension Eastern part of the separated Gulf Stream 35 N

Subtropical

Countercurrent (STCC)

Azores Current

Canary and Portugal

Current Systems

Subtropical Frontal

Zone

* Shading indicates the basic set.

Eastward flow of the western subtropical gyre, south

of the recirculation; continues into the Subtropical

Front

Zonal eastward flow in the central and eastern

subtropical gyre

Subtropical eastern boundary current system

Zonal frontal band in the subtropical gyre; close to

the maximum Ekman transport convergence

22e25 N

33e36 N

23e46 N

30e35 N


TABLE S9.2

Major Upper Ocean Circulation Systems, Currents and Fronts of the Subpolar North Atlantic

(Figures 9.1 and S9.1)*

Name Description Approximate Location

Subpolar gyre Cyclonic circulation at mid to high latitudes 45e65 N

North Atlantic Current (NAC)

Subtropical western boundary current and

eastward flow of the subtropical and subpolar

gyres; northeastward flow of subpolar gyre

with three distinct branches

40e65 N

Labrador Current Subpolar western boundary current 40e65 N

East Greenland Current (EGC)

West Greenland Current (WGC)

Irminger Current

Iceland Basin branch of the NAC

Subpolar western boundary current on east

coast of Greenland

Subpolar eastern boundary current on west

coast of Greenland

Northward flow along the western flank of the

Reykjanes Ridge

Northward flow of the NAC in the Iceland

Basin

North of 62 N

60e65 N

55e64 N

55e62 N

Rockall Trough branch of the NAC North flow of the NAC in Rockall Trough 54e64 N

Iceland-Faroe Front (IFF) Eastward flow along the Iceland-Faroe Ridge 62e66 N

North Iceland Current (NIC) Eastward flow along north side of Iceland 66 N

Subarctic Frontal Zone


Frontal band separating subpolar and

subtropical waters, within the North Atlantic

Current

55e65 N

40


TABLE S9.3

Name

Tropical and South Atlantic Circulation Systems and Currents*

Description


(NECC)

North Brazil Current (NBC)

South Equatorial Current (SEC)

Equatorial Undercurrent (EUC)

Equatorial Intermediate Current (EIC)

South Equatorial Countercurrent (SECC)

Guinea Current

Guinea Dome

Angola Current

Angola Dome

Subtropical gyre

Brazil Current (BC)

South Atlantic Current (or Westwind

Drift; SAC)

Benguela Current System (BCS)

Agulhas Retroflection

Subantarctic Front (SAF)

Malvinas (Falkland) Current

Deep Western Boundary Current

(DWBC)

Eastward flow at 5e10 N

Northward-flowing low latitude western boundary current

Westward flow in equatorial region (“North” and “Central” SEC) and on

the equatorial side of the South Atlantic’s subtropical gyre (“South” SEC)

Eastward subsurface flow along equator

Westward flow beneath the Equatorial Undercurrent along the equator

Eastward flow at 7e9 S

Eastward flow along the coast of Africa north of the equator

Upwelling region with cyclonic circulation in the eastern tropical North

Atlantic

Southward tropical eastern boundary current between the equator and 16 S

Upwelling region with cyclonic circulation in the eastern tropical South

Atlantic

Anticyclonic gyre at mid-latitudes

Western boundary current of the subtropical gyre along the coast of Australia

Eastward flow of the subtropical gyre

Eastern boundary current system for the subtropical gyre; 34e14 S

Retroflection of the Indian Ocean’s Agulhas Current

Eastward flow in the northernmost front of the Antarctic Circumpolar

Current

Western boundary current that is a northward loop of the Subantarctic Front

Deep boundary currents in the NADW and AABW layers



TABLE S9.4

Principal Atlantic Ocean Water Masses*

Water Mass Characteristic in the Vertical Layer Formation Process

North Atlantic Central

Water (NACW)

Subtropical thermocline waters

Upper

0e1000 m

Subduction

South Atlantic Central

Water (SACW)


Upper

0e1000 m

Subduction

North Atlantic Subtropical

Underwater (NASTUW)

Subtropical/tropical salinity

maximum

Upper

50e100 m

Subduction of high salinity

subtropical surface waters

South Atlantic Subtropical

Underwater (SASTUW)


maximum

Upper

50e100 m



North Atlantic Subtropical

Mode Water or Eighteen

Degree Water (EDW)

Subtropical stability (potential

vorticity) minimum

Upper

0e400 m

Subduction of thick,

convective winter mixed

layer

Subpolar Mode Water

(SPMW)

North Atlantic stability (potential

vorticity) minimum

Upper

0e700 m

Thick, convective winter

mixed layer

South Atlantic Subtropical

Mode Water (SASTMW)


vorticity) minimum

Upper

0e300 m

Subduction of thick,

convective winter mixed

layer


(SAMW)

Labrador Sea Water (LSW)

Potential vorticity minimum and

oxygen maximum in subtropical

South Atlantic

Salinity and potential vorticity

minimum in subpolar and western

North Atlantic

Upper

0e600 m

Intermediate

200e2000 m

Subducted thick winter

mixed layers north of

Subantarctic Front

Deep convection in the

Labrador Sea

Mediterranean Water (MW;

or Mediterranean Overflow

Water or Mediterranean

Outflow Water, MOW)

Salinity maximum in North Atlantic

subtropical gyre and tropics

Intermediate

700e1700 m


Mediterranean Sea,

overflow through Strait of

Gibraltar


Water (AAIW)

Salinity minimum in subtropical

S. Atlantic and tropical Atlantic

Intermediate

500e1200 m

Advection of fresh

subantarctic surface water

Nordic Seas Overflow Water

(NSOW)

Oxygen maximum in the northern

North Atlantic

Deep

600e4500 m


Greenland Sea, overflow

into the North Atlantic

Denmark Strait Overflow

Water (DSOW)

Oxygen maximum in the deep

northern North Atlantic

Deep

600e4500 m

Nordic Seas overflow

through Denmark Strait

Iceland Scotland Overflow

Water (ISOW)

Salinity maximum in the deep

northern North Atlantic

Deep

2500e3500 m

Nordic Seas overflow across

the Iceland-Scotland ridge

(Continued)

42

TABLE S9.4


Principal Atlantic Ocean Water Masses* dCont’d

Water Mass Characteristic in the Vertical Layer Formation Process

Northeast Atlantic Deep

Water (NEADW)


(NADW)

Oxygen minimum, salinity

maximum in deep eastern North

Atlantic

Oxygen minimum, nutrient

maximum, salinity maximum

Deep

2500e4500 m

Deep

1000e4500 m

Mixture of all NADW

sources and AABW

Mixing and aging of deep

waters

Upper Circumpolar Deep

Water (UCDW)

Low oxygen

Deep

~1000e 3000 m

Mixture of IDW, PDW, and

deep waters in the Southern

Ocean

Weddell Sea Deep Water

(WSDW)


(AABW)

Cold, dense

Deep salinity and oxygen maxima,

nutrient minima

Near surface to

bottom

Bottom

3000 m to bottom

Brine rejection and

convection in the Southern

Ocean

Brine rejection in the

Southern Ocean, mixed

with NADW, PDWand IDW


References

Bane, J.M., 1994. The Gulf Stream System: An observational

perspective. Chapter 6. In: Majumdar, S.K., Miller, E.W.,

Forbes, G.S., Schmalz, R.F., Panah, A.A. (Eds.), The

Oceans: Physical-Chemical Dynamics and Human

Impact. The Pennsylvania Academy of Science,

pp. 99e107.

Baringer, M., Larsen, J., 2001. Sixteen years of Florida

Current transport at 27 N. Geophys. Res. Lett. 28,

3179e3182.

Beal, L.M., Hummon, J.M., Williams, E., Brown, O.B.,

Baringer, W., Kearns, E.J., 2008. Five years of Florida

Current structure and transport from the Royal Caribbean

Cruise Ship Explorer of the Seas. J. Geophys. Res.

113 C06001. doi:10.1029/2007JC004154.

Biastoch, A., Böning, C.W., Lutjeharms, J.R.E., 2008. Agulhas

leakage dynamics affects decadal variability in the

Atlantic overturning circulation. Nature 456, 489e492.

Boebel, O., Davis, R.E., Ollitrault, M., Peterson, R.G.,

Richardson, P.L., Schmid, C., Zenk, W., 1999. The

intermediate depth circulation of the western. South

Atlantic. Geophys. Res. Lett. 26, 21. doi:10.1029/

1999GL002355.

Bower, A.S., LeCann, B., Rossby, T., Zenk, W., Gould, J.,

Speer, K., Richardson, P.L., Prater, M.D., Zhang, H.-M.,

2002. Directly measured mid-depth circulation in the

northeastern North Atlantic Ocean. Nature 410,

603e607.



at 26.5 N. Nature 438, 655e657.

Candela, J., Tanahara, S., Crepon, M., Barnier, B.,

Sheinbaum, J., 2003. Yucatan Channel flow: Observations

versus CLIPPER ATL6 and MERCATOR PAM

models. J. Geophys. Res. 108, 3385. doi:10.1029/

2003JC00196.

Dengler, M., Schott, F.A., Eden, C., Brandt, P., Fischer, J.,

Zantopp, R.J., 2004. Break-up of the Atlantic deep

western boundary current into eddies at 8 S. Nature 432,

1018e1020.

Dickson, R., Brown, J., 1994. The production of North

Atlantic Deep Water: Sources, rates, and pathways.

J. Geophys. Res. 99, 12319e12341.

Fratantoni, D.M., 2001. North Atlantic surface circulation

during the 1990s observed with satellite-tracked drifters.

J. Geophys. Res. 106, 22067e22093.




data. J. Geophys. Res. 108 (C7), 3213. doi: 10.1029/

2002JC002565.

Hogg, N.G., 1983. A note on the deep circulation of the

western North Atlantic: Its nature and causes. Deep-Sea

Res. 30, 945e961.

Hogg, N.G., Owens, W.B., 1999. Direct measurement of the

deep circulation within the Brazil Basin. Deep-Sea Res. II

46, 335e353.


Hogg, N.G., Siedler, G., Zenk, W., 1999. Circulation and

variability at the southern boundary of the Brazil Basin.

J. Phys. Oceanogr 29, 145e157.

Johns, W.E., Townsend, T.L., Fratantoni, D.M., Wilson, W.D.,

2002. On the Atlantic inflow to the Caribbean Sea. Deep-

Sea Res. I 49, 211e243.

Juliano, M.F., Alvés, M.L.G.R., 2007. The Subtropical Front/

Current systems of Azores and St. Helena. J. Phys.

Oceanogr 37, 2573e2598.


Deaven, D., Gandin, L., et al., 1996. The NCEP-NCAR

40-year reanalysis project. B. Am. Meteorol. Soc. 77,

437e471.


interannually varying air-sea flux data set. Clim. Dynam

33, 341e364.

Lavender, K.L., Davis, R.E., Owens, W.B., 2000. Mid-depth

recirculation observed in the interior Labrador and

Irminger seas by direct velocity measurements. Nature

407, 66e69.

Lentini, C.A.D., Goni, G.J., Olson, D.B., 2006. Investigation

of Brazil Current rings in the confluence region. J. Geophys.

Res, 111. C06013. doi:10.1029/2005JC002988.

McCartney, M.S., Talley, L.D., 1982. The Subpolar Mode

Water of the North Atlantic Ocean. J. Phys. Oceanogr 12,

1169e1188.


Dynamically balanced absolute sea level of the global

ocean derived from near-surface velocity observations.


Olson, D., Schmitt, R., Kennelly, M., Joyce, T., 1985. A twolayer

diagnostic model of the long-term physical

evolution of warm-core ring 82B. J. Geophys. Res. 90,

8813e8822.

Owens, W.B., 1991. A statistical description of the mean

circulation and eddy variability in the northwestern

Atlantic using SOFAR floats. Progr. Oceanogr 28, 257e303.

Parker, C.E., 1971. Gulf Stream rings in the Sargasso Sea.


Peeters, F.J.C., Acheson, R., Brummer, G.-J.A., de

Ruijter, W.P.M., Schneider, R.R., Ganssen, G.M.,

Ufkes, E., Kroon, D., 2004. Vigorous exchange between

the Indian and Atlantic oceans at the end of the past five

glacial periods. Nature 430, 661e665.

Peterson, R.G., 1992. The boundary currents in the western

Argentine Basin. Deep-Sea Res. 39, 623e644.

Pickart, R.S., McKee, T.K., Torres, D.J., Harrington, S.A.,

1999. Mean structure and interannual variability of the

slopewater system south of Newfoundland. J. Phys.

Oceanogr 29, 2541e2558.

Pickart, R.S., Torres, D.J., Clarke, R.A., 2002. Hydrography

of the Labrador Sea during active convection. J. Phys.

Oceanogr 32, 428e457.

Provost, C., Escoffier, C., Maamaatuaiahutapu, K.,

Kartavtseff, A., Garçon, V., 1999. Subtropical mode

waters in the South Atlantic Ocean. J. Geophys. Res. 104,

21033e21049.



Progr. Oceanogr 33, 1e92.

Richardson, P.L., 2005. Caribbean Current and eddies as

observed by surface drifters. Deep-Sea Res. II 52, 429e463.

Richardson, P.L., Bower, A.S., Zenk, W., 2000. A census of

Meddies tracked by floats. Progr. Oceanogr 45, 209e250.

Roemmich, D.L., 1983. Optimal estimation of hydrographic

station data and derived fields. J. Phys. Oceanogr 13,

1544e1549.

Rossby, T., 1999. On gyre interactions. Deep-Sea Res. II 46,

139e164.

Schmitz, W.J., 1996b. On the World Ocean Circulation:

Volume I: Some global features/North Atlantic circulation.

Woods Hole Oceanographic Institution Technical

Report, WHOI-96-03. Woods Hole, MA, 141 pp.

Schott, F., Zantopp, R., Stramma, L., Dengler, M., Fischer, J.,

Wibaux, M., 2004. Circulation and deep water export at

the western exit of the subpolar North Atlantic. J. Phys.

Oceanogr 34, 817e843.


reviewed for the CNLS conference. Physica D 98,

625e646.

Tsuchiya, M., Talley, L.D., McCartney, M.S., 1992. An eastern

Atlantic section from Iceland southward across the

equator. Deep-Sea Res. 39, 1885e1917.

van Aken, H.M., van Veldhoven, A.K., Veth, C., de

Ruijter, W.P.M., van Leeuwen, P.J., Drijfhout, S.S.,

Whittle, C.P., Rouault, M., 2003. Observations of a young

Agulhas ring, Astrid, during MARE in March 2000.

Deep-Sea Res. II 50, 167e195.

Zemba, J.C., 1991. The structure and transport of the Brazil

Current between 27 and 36 South. Ph.D. Thesis.

Massachusetts Institute of Technology and Woods Hole

Oceanographic Institution, 160 pp.

C H A P T E R

S10

Pacific Ocean: Supplementary Materials

FIGURE S10.1 Pacific Ocean: mean surface geostrophic circulation with the current systems described in this text. Mean

surface height (cm) relative to a zero global mean height, based on surface drifters, satellite altimetry, and hydrographic

data. (NGCUC ¼ New Guinea Coastal Undercurrent and SECC ¼ South Equatorial Countercurrent). Data from Niiler,

Maximenko, and McWilliams (2003).

1

2

S10. PACIFIC OCEAN: SUPPLEMENTARY MATERIALS

FIGURE S10.2 Annual mean winds. (a) Wind stress (N/m 2 ) (vectors) and wind-stress curl (10 7 N/m 3 ) (color),

multiplied by 1 in the Southern Hemisphere. (b) Sverdrup transport (Sv), where blue is clockwise and yellow-red is

counterclockwise circulation. Data from NCEP reanalysis (Kalnay et al.,1996).

S10. PACIFIC OCEAN: SUPPLEMENTARY MATERIALS 3

(a)

0

STFZ

SAFZ

PF

Depth (m)

(b)

Depth (m)

(c)

100

200

300

400

500

0

100

200

300

400

500

0

18

17

16

9

15

14

8

13

Subtropical Domain

34.5

34.3

12

11

10

35.2 34.6

34.2

9

8

7

6.5

Transition Zone

STFZ

SAFZ

34.1

34

34

33

6

5.5

5

4.5

4

Subarctic Domain

33.9

34

33.7

33.8

34.1

30°N 40°N 50°N

Potential

temperature

(°C)

3.5

32.7

32.8

Salinity

Alaskan

Stream

100

Depth (m)

(d)

200

300

400

500

24.0

24.5

12

1 2 4

14

16

8

6

10

20

12 14 16

25

44

44

30 35

20

40

Nitrate (μmol/kg)

30°N Latitude 40°N 50°N

Sea surface density

Nitrate (μmol/kg)

Potential density σ θ

25.0

25.5

26.0

26.5

27.0

2

1

4

8

10

12

16

1

2

12

14

16

20

25

30

35

40

30°N 40°N 50°N

FIGURE S10.3 The subtropical-subarctic transition along 150 W in the central North Pacific (MayeJune, 1984). SAFZ

and STFZ: subarctic and subtropical frontal zones. (a) Potential temperature ( C), (b) salinity, (c) nitrate (mmol/kg), and

(d) nitrate versus potential density. After Roden (1991). Data from WOCE Pacific Ocean Atlas; Talley (2007).

10

4


FIGURE S10.4

and Nasu (1976).

Subpolar gyre regimes. Only the major features are described in the text. Source: From Favorite, Dodimead,

FIGURE S10.5 Oyashio. Acceleration potential anomaly (similar to geopotential anomaly) on the isopycnal s q ¼ 26.52

kg/m 3 (150 cl/ton) referenced to 1500 dbar in September 1990. Source: From Kono and Kawasaki (1997).


FIGURE S10.6 (a) The Oyashio, Kuroshio, and Mixed Water Region east of Japan. (b) The southernmost latitude of the

first Oyashio intrusion east of Honshu. Source: From Sekine (1999).

6


FIGURE S10.7 (a) Ocean color from the SeaWIFS satellite, showing an anticyclonic Haida Eddy in the Alaska Current on

June 13, 2002. Source: From NASA Visible Earth (2008). (b) Tracks of Sitka and Haida Eddies in 1995 and 1998 (top right) and in

remaining years between 1993 and 2001 (bottom right). Source: From Crawford (2002).


FIGURE S10.8 Mean steric height at (a) 150 m and (b) 500 m relative to 2000 m; contour intervals are 0.04 and 0.02 m,

respectively. Source: From Ridgway and Dunn (2003).

8


FIGURE S10.10 Surface chlorophyll concentration in

austral winter (June, July, August) and summer (December,

January, February), derived from SeaWiFS satellite observations.

Source: From Mackas, Strub, Thomas, and Montecino

(2006).

FIGURE S10.9 Sea level (m). (a) Total sea level, and

(b) RMS sea level anomalies, from satellite altimetry. The

3000 m isobath is shown (purple). Source: From Mata,

Wijffels, Church, and Tomczak (2006).


FIGURE S10.11 Eastern South Pacific zonal vertical sections at 33 S: (a) temperature with meridional current directions,

(b) salinity, (c) dissolved oxygen, and (d) phosphate.

10


FIGURE S10.12 RMS variability in surface height (cm) from satellite altimetry, high-passed with half power at 180 days

to depict the mesoscale eddy energy. ÓAmerican Meteorological Society. Reprinted with permission. Source: From Qiu, Scott,

and Chen (2008).


FIGURE S10.13 Mean flow at 900 m depth in the tropical and South Pacific based on subsurface float observations.

(a) Velocity (cm/sec). (b) Geostrophic streamfunction (1000 m 2 /s). ÓAmerican Meteorological Society. Reprinted with

permission. Source: From Davis (2005).

12


FIGURE S10.14

Kitagawa (2006).

Pacific abyssal circulation schematics. Low latitude North Pacific. Source: From Kawabe, Yanagimoto, and


FIGURE S10.15 Northward transports (Sv) across zonal sections in isopycnal layers, integrated upwards from zero at

the bottom. Section latitudes are indicated in parentheses. Ekman transport is not included. Gray indicates the uncertainty.

Source: From Ganachaud (2003).

14


(a)

20˚

100˚ 120˚

February

0

140˚

160˚

180˚

200˚

220˚

240˚

260˚

280˚

0

300˚

20˚

0

0˚

0

0˚

0

-20˚ -20˚

0

(b)

20˚

100˚

August

120˚

140˚

160˚

0

180˚

200˚

0

220˚

240˚

260˚

280˚

0

300˚

20˚

0

0

0

0˚

0˚

0

0

0

0

-20˚ -20˚

0.1 N/m 2

100˚ 120˚ 140˚ 160˚ 180˚ 200˚ 220˚ 240˚ 260˚ 280˚ 300˚

Wind stress curl (N/m 3 )

-0.20 -0.10 0

0.10

(x 1) (NH) or (x -1) (SH)

0.20

FIGURE S10.16 Climatological wind stress (N/m 2 ) (vectors) and wind stress curl (N/m 3 , multiplied by 1 in Southern

Hemisphere; contours, shading is negative): (a) February and (b) August. Monthly mean wind data are from the NCEP

reanalysis (Kalnay et al., 1996). (c) Sverdrup transport (Sv) in the tropical Pacific Ocean, calculated from Hellerman-

Rosenstein (1983) wind stress. Positive (negative) values yield clockwise (counterclockwise) circulation. ÓAmerican

Meteorological Society. Reprinted with permission. Source: From Qu and Lindstrom (2002).


FIGURE S10.17 Dynamic height (dyn cm) along the equator; transport (Sv) of the Equatorial Undercurrent is shown in

the inset. ÓAmerican Meteorological Society. Reprinted with permission. Source: From Leetmaa and Spain (1981).

FIGURE S10.18 Chlorophyll composite images from SeaWiFS (January 1998 during El Niño and July 1998 during

transition to La Niña). Red ¼ highest chlorophyll contents, dark purple ¼ lowest chlorophyll. Source: From SeaWiFS Project

(2009).

16


FIGURE S10.19 Currents in the western tropical Pacific. NEC ¼ North Equatorial Current; NECC ¼ North Equatorial

Countercurrent; SEC ¼ South Equatorial Current; EUC ¼ Equatorial Undercurrent; NSCC and SSCC ¼ North and South

Subsurface Countercurrent; MC ¼ Mindanao Current; MUC ¼ Mindanao Undercurrent; ME ¼ Mindanao Eddy; HE ¼

Halmahera Eddy; NGCC ¼ New Guinea Coastal Current; NGCUC ¼ New Guinea Coastal Undercurrent; GBRUC ¼ Great

Barrier Reef Undercurrent; EAC ¼ East Australian Current; LC ¼ Leeuwin Current; AAIW ¼ Antarctic Intermediate Water.

Source: From Lukas, Yamagata, and McCreary (1996); after Fine et al. (1994).


FIGURE S10.20 Tropical sea surface temperature from the Tropical Rainfall Mapping Mission (TRMM) Microwave

Imager (TMI) for ten-day intervals from June 1 to August 30, 1998. Source: From Remote Sensing Systems (2004).

18


FIGURE S10.21 Winds and SST along the equator in the Pacific. Climatological zonal wind speed in (a) February and

(b) August. Source: From TAO Project Office (2009b). (c) Monthly zonal wind speed (m/s) and SST ( C). Positive wind is

towards the east. Source: From TAO Project Office (2009a).


(a)

FIGURE S10.22 (a) February mean winds (vectors) from COADS and February mean SST. The large arrows emphasize

the gaps through the American cordillera. From north to south: Tehuantepec, Papagayo, and Panama. Source: From Kessler

(2009). (b) SST in the Gulf of Tehuantepec, January 22, 1996. Source: From Zamudio et al. (2006).

20


(a)

SOI (Australia BOM)

25

20

15

10

5

0

-5

-10

-15

-20

-25

(b)

2

Southern Oscillation Index (Australia BOM)

La Nina

El Nino

1880 1890 1900 1910 1920 1930 1940 1950 1960 1970 1980 1990 2000

Oceanic Nino Index (NOAA CPC)

El Nino

Index

0

-2

La Nina

1950 1955 1960 1965 1970 1975 1980 1985 1990 1995 2000 2005

FIGURE S10.23 (a) Southern Oscillation Index (SOI) time series from 1876 to 2008 (annual average). Data from

Australian Government Bureau of Meteorology (2009). (b) “Oceanic Nino Index” based on SST in the region 5 Ne5 S,

170 We120 W, as in Figure 10.28b. Red and blue in both panels correspond to El Niño and La Niña, respectively. (c) SST

reconstructions from the region 5 Ne5 S, 150 We 90 W. Source: From IPCC (2001). (d) Correlation of monthly sea level

pressure anomalies with the ENSO Nino3.4 index, averaged from 1948 to 2007. The Nino3.4 index is positive during the

El Niño phase, so the signs shown are representative of this phase. Data and graphical interface from NOAA ESRL (2009b).


FIGURE S10.23

(Continued).

FIGURE S10.24 Global precipitation anomalies for Northern Hemisphere summer (left) and winter (right) during

El Niño. Source: From NOAA PMEL (2009d).

22


FIGURE S10.25 Anomalies of United States winter (JFM) (a) temperature ( C) and (b) precipitation (mm) during

composite El Niño events from 1950 to 2008. Source: From NWS Internet Services Team (2008).

FIGURE S10.26 Subtropical Mode Water. (a) Vertical profile through North Pacific Subtropical Mode Water, at 29 5’N,

158 33’E. Source: From Hanawa and Talley (2001). (b) North Pacific: thickness of the 17e18 C layer. Source: From Masuzawa

(1969). (c) South Pacific: thickness of the 15e17 C layer. ÓAmerican Meteorological Society. Reprinted with permission.

Source: From Roemmich and Cornuelle (1992).


FIGURE S10.26

(Continued).

FIGURE S10.27 Salinity at the NPIW salinity minimum. Outer dark contour is the edge of the salinity minimum.

ÓAmerican Meteorological Society. Reprinted with permission. Source: From Talley (1993).

24


FIGURE S10.28 (a) Salinity, (b) oxygen (mmol/kg), and (c) silicate (mmol/kg) along 165 W. Neutral densities 28.00 and

28.10 kg/m 3 are superimposed. Station locations are shown in inset in (c). Source: From WOCE Pacific Ocean Atlas, Talley

(2007).


FIGURE S10.29 (a) Salinity, (b) silicate (mmol/kg), (c) D 14 C (/mille), and (d) d 3 He (%) at neutral density 28.01 kg/m 3

(s 2 ~ 36.96 kg/m 3 ), characterizing PDW/UCDW at mid-depth. The depth of the surface is approximately 2600e2800 m

north of the Antarctic Circumpolar Current. Source: From WOCE Pacific Ocean Atlas, Talley (2007).

FIGURE S10.30 (a) Salinity, (b) silicate (mmol/kg), and (c) depth (m) at neutral density 28.10 kg/m 3 (s 4 ~ 45.88 kg/m 3 ),

characteristic of LCDW. (d) Potential temperature at 4000 m. Source: From WOCE Pacific Ocean Atlas, Talley (2007).

26

TABLE S10.1


Major Upper Ocean Circulation Systems, Currents and Fronts of the Mid and High Latitude North

Pacific (Figure S10.1)*

Name Description Approximate Location

Subtropical gyre Anticyclonic gyre at mid-latitudes 10e40 N

Subpolar gyre Cyclonic gyre at mid to high latitudes 40e65 N

Western Subarctic Gyre Intense cyclonic sub-gyre in the western subpolar gyre 40e55 N, Kuril Islands to

180

Alaska Gyre Intense cyclonic sub-gyre in the eastern subpolar gyre 40 N to Alaskan coast, 180

to eastern boundary

Bering Sea gyre Cyclonic circulation in the Bering Sea Bering Sea

Okhotsk Sea gyre Cyclonic circulation in the Okhotsk Sea Okhotsk Sea

Kuroshio Subtropical western boundary current 12e35 N

Kuroshio Extension Subtropical western boundary current extension 35 N

Kuroshio recirculation or

Kuroshio Countercurrent

Westward flow just south of the Kuroshio Extension

30 N

Subtropical Countercurrent Eastward flow of the western subtropical gyre, south of 25 N

the recirculation; continues into the Subtropical Front

California Current System Subtropical eastern boundary current system 23e52 N

Oyashio

East Kamchatka Current

Subpolar western boundary current south of central

Kuril Islands

Subpolar western boundary current north of central

Kuril Islands

40eN

45e65 N

Alaska Current Subpolar eastern boundary current North of 52 N

Alaskan Stream


Southwestward flow of the subpolar gyre along the

northern boundary

Westward flow of the subtropical gyre and northern

tropical gyre

180e145 W

10e20 N

North Pacific Current Eastward flow of the subtropical and subpolar gyres 20e50 N

Subtropical Frontal Zone

Subarctic Frontal Zone


Zonal frontal band in the subtropical gyre; close to the

maximum Ekman transport convergence

Zonal frontal band separating the subpolar and subtropical

gyre regimes; close to maximum westerly wind stress

30e35 N

40e N


TABLE S10.2

Name

Subtropical gyre

South Pacific Circulation Systems and Currents*

Description


East Australian Current (EAC)

Tasman Front

East Auckland Current

South Pacific Current (or Westwind Drift)


Peru-Chile Current System (PCCS)

Peru-Chile Current (PCC)

Poleward Undercurrent (PUC)

Peru-Chile Countercurrent (PCCC)

Cape Horn Current


Western boundary current of the subtropical gyre along the coast

of Australia

Eastward current connecting the East Australian Current and the

East Auckland Current

Western boundary current of the subtropical gyre along the coast of

New Zealand


Eastward flow in the northernmost front of the Antarctic

Circumpolar Current

Eastern boundary current system for the subtropical gyre

Northward flow in the PCCS

Southward undercurrent in the PCCS

Southward surface flow within the PCCS

Southward eastern boundary current

Westward flow of the subtropical gyre


28


TABLE S10.3

Name

Tropical Pacific Currents*

Description

Location Upper Ocean

Unless Otherwise Noted

North Equatorial Current (NEC)


Westward flow of the North Pacific

subtropical gyre

Westward flow of the South Pacific

subtropical gyre and westward flow in the

equatorial region

8e30 N

30 Sto3 N

North Equatorial Countercurrent (NECC) Eastward flow between the NEC and SEC 3e8 N

South Equatorial Countercurrent (SECC) Eastward flow embedded in the SEC 8e11 S, western and

central Pacific only

Equatorial Undercurrent (EUC)

Eastward subsurface flow, just below the

surface layer

1 Sto1 N

50e250 m

Equatorial Intermediate Current (EIC) Westward subsurface flow, below the EUC 1 Sto1 N

250e1000 m

Equatorial stacked jets

North and South Subsurface Countercurrents

(NSCC, SSCC; “Tsuchiya jets”)

Mindanao Current

New Guinea Coastal Undercurrent (NGCUC)

North Queensland Current (NQC) and Great

Barrier Reef Undercurrent (GBRUC)


Reversing subsurface eastward-westward

jets, beneath the EIC

Eastward subsurface flows, off the equator

Southward western boundary current

connecting the NEC and NECC

Northward tropical western boundary

current connecting the SEC and NQC to

the EUC, NSCC and SSCC

Northward western boundary current for

the SEC

1 Sto1 N

1000 m to bottom

6e2 S; 2e6 N

150e500 m

6e14 N

12 Sto6 N

15e12 S (NQC)

23e15 S (GBRUC)


TABLE S10.4

Principal Pacific Ocean Water Masses*

Water Mass Characteristic in the Vertical Layer Process

North Pacific Central Water

(NPCW)


Upper

0e1000 m

Subduction

South Pacific Central Water

(SPCW)


Upper

0e1000 m

Subduction

North Pacific Subtropical

Underwater (NPSTUW)


maximum

Upper

100e200 m



South Pacific Subtropical

Underwater (SPSTUW)


maximum

Upper

100e200 m



North Pacific Subtropical

Mode Water (NPSTMW)


vorticity) minimum

Upper

0e400 m

Subduction of thick winter mixed

layer

South Pacific Subtropical

Mode Water (SPSTMW)


vorticity) minimum

Upper

0e300 m


layer


(SAMW)

Southern subtropical stability

(potential vorticity) minimum

Upper

0e600 m


layer from Subantarctic Front

Dichothermal Water (DtW)

North Pacific subpolar

temperature minimum

Upper

0e150 m

Advection of cold subpolar surface

waters

Mesothermal Water (MtW)

North Pacific Intermediate

Water (NPIW)

North Pacific subpolar

temperature maximum


North Pacific

Upper

200e500 m

Intermediate

200e800 m

Advection of warmer near-surface

subpolar waters

Advection of fresh subpolar surface

water


(AAIW)


North Pacific and tropical Pacific

Intermediate

500e1200 m

Advection of fresh subantarctic

surface water

Pacific Deep Water (PDW)

Oxygen minimum, nutrient

maximum

Deep

1000e4000 m

Mixing and aging of deep waters


Water (UCDW)

High oxygen, low nutrients, high

salinity on isopycnal surfaces

Deep

~ 1000e3000 m

Mixture of deep waters in the

Southern Ocean

Lower Circumpolar Deep

Water (LCDW)

Deep salinity and oxygen

maxima, nutrient minima

Bottom

3000 m to bottom

Brine rejection in the Southern Ocean

mixed with NADW, PDW, and IDW


30


References

Australian Government Bureau of Meteorology, 2009. S.O.I

(Southern Oscillation Index) Archives d 1876 to present.

http://reg.bom.gov.au/climate/current/soihtm1.shtml


Crawford, W., 2002. Physical characteristics of Haida

Eddies. J. Oceanogr 58, 703e713.



floats. J. Phys. Oceanogr 35, 683e707.

Favorite, F., Dodimead, A.J., Nasu, K., 1976. Oceanography

of the subarctic Pacific region, 1960e71. International

North Pacific Fisheries Commission, ew, 33,

187 pp.

Fine, R.A., Lukas, R., Bingham, F.M., Warner, M.J.,

Gammon, R.H., 1994. The western equatorial Pacific:

A water mass crossroads. J. Geophys. Res. 99,

25063e25080.




data. J. Geophys. Res. 108 (C7), 3213. doi: 10.1029/

2002JC002565.

Hanawa, K., Talley, L.D., 2001. Mode Waters. In: Siedler, G.,

Church, J. (Eds.), Ocean Circulation and Climate. International

Geophysics Series. Academic Press,

pp. 373e386.

IPCC, 2001. Climate Change 2001: The Scientific Basis. In:

Houghton, J.T., Ding, Y., Griggs, D.J., Noguer, M., van

der Linden, P.J., Dai, X., Maskell, K., Johnson, C.A.

(Eds.), Contribution of Working Group I to the Third

Assessment Report of the Intergovernmental Panel on

Climate change. Cambridge University Press, Cambridge,

UK and New York, 881 pp.


Deaven, D., Gandin, L., 1996. The NCEP-NCAR 40-year

reanalysis project. B. Am. Meteorol. Soc. 77, 437e471.

Kawabe, M., Yanagimoto, D., Kitagawa, S., 2006. Variations of

deepwesternboundarycurrentsintheMelanesianBasin

in the western North Pacific. Deep-Sea Res. I 53, 942e959.

Kessler, W.S., 2009. The Central American mountaingap

winds and their effects on the ocean. http://

faculty.washington.edu/kessler/t-peckers/t-peckers.html

(accessed 3.27.09).

Kono, T., Kawasaki, Y., 1997. Modification of the western

subarctic water by exchange with the Okhotsk Sea.


Leetmaa, A., Spain, P.F., 1981. Results from a velocity

transect along the equator from 125 to 159 W. J. Phys.

Oceanogr. 11, 1030e1033.

Lukas, R., Yamagata, T., McCreary, J.P., 1996. Pacific lowlatitude

western boundary currents and the Indonesian

throughflow. J. Geophys. Res. 101, 12209e12216.

Mackas, D.L., Strub, P.T., Thomas, A., Montecino, V., 2006.

Eastern ocean boundaries pan-regional overview. In:

Robinson, A.R., Brink, K.H. (Eds.), The Sea, Vol. 14A:



pp. 21e60.

Masuzawa, J., 1969. Subtropical Mode Water. Deep-Sea Res.

16, 453e472.

Mata, M.M., Wijffels, S.E., Church, J.A., Tomczak, M., 2006.

Eddy shedding and energy conversions in the East

Australian Current. J. Geophys. Res. 111, C09034.

doi:10.1029/2006JC003592.

McClain, C., Christian, J.R., Signorini, S.R., Lewis, M.R.,

Asanuma, I., Turk, D., Dupouy-Douchement, C., 2002.

Satellite ocean-color observations of the tropical Pacific

Ocean. Deep-Sea Res. II 49, 2533e2560.

NASA Visible Earth, 2008. Eddies off the Queen Charlotte

Islands. NASA Goddard Space Flight Center. http://

visibleearth.nasa.gov/view_rec.php?id=2886 (accessed

3.26.09).


Dynamically balanced absolute sea level of the global

ocean derived from near-surface velocity observations.


NOAA ESRL, 2009b. Linear correlations in atmospheric

seasonal/monthly averages. NOAA Earth System

Research Laboratory Physical Sciences Division. http://

www.cdc.noaa.gov/data/correlation/ (accessed 10.30.09).

NOAA PMEL, 2009d. Impacts of El Niño and benefits of

El Niño prediction. NOAA Pacific Marine Environmental

Laboratory. http://www.pmel.noaa.gov/tao/

elnino/impacts.html (accessed 3.26.09).

NWS Internet Services Team, 2008. ENSO temperature and

precipitation composites. http://www.cpc.noaa.gov/

products/precip/CWlink/ENSO/composites/EC_LNP_


Qiu, B., Scott, R.B., Chen, S., 2008. Length scales of eddy

generation and nonlinear evolution of the seasonally

modulated South Pacific Subtropical Countercurrent.


Qu, T., Lindstrom, E., 2002. A climatological interpretation

of the circulation in the western South Pacific. J. Phys.

Oceanogr. 32, 2492e2508.

Remote Sensing Systems, 2004. TMI sea surface temperatures

(SST). http://www.ssmi.com/rss_research/tmi_

sst_pacific_equatorial_current.html (accessed 3.27.09).

Ridgway, K.R., Dunn, J.R., 2003. Mesoscale structure of the

mean East Australian Current System and its relationship

with topography. Progr. Oceanogr. 56, 189e222.

Roden, G.I., 1991. Subarctic-subtropical transition zone of

the North Pacific: Large-scale aspects and mesoscale

structure. In: Wetherall, J.A. (Ed.), Biology, Oceanography

and Fisheries of the North Pacific Transition Zone


and the Subarctic Frontal Zone. NOAA Technical Report

105, 1e38.

Roemmich, D., Cornuelle, B., 1992. The Subtropical Mode

Waters of the South Pacific Ocean. J. Phys. Oceanogr. 22,

1178e1187.

SeaWiFS Project, 2009. SeaWiFS captures El Nino-La Nina

transitions in the equatorial Pacific. NASA Goddard

Space Flight Center. http://oceancolor.gsfc.nasa.

gov/SeaWiFS/BACKGROUND/Gallery/pac_elnino.jpg

(accessed 3.26.09).

Sekine, Y., 1999. Anomalous southward intrusions of the

Oyashio east of Japan 2. Two-layer numerical model.

J. Geophys. Res. 104, 3049e3058.

Talley, L.D., 1993. Distribution and formation of North

Pacific Intermediate Water. J. Phys. Oceanogr 23,

517e537.

Talley, L.D., 2007. Hydrographic Atlas of the World Ocean

Circulation Experiment (WOCE). In: Sparrow, M.,

Chapman, P., Gould, J. (Eds.), Pacific Ocean, Volume 2.

International WOCE Project Office, Southampton, U.K.

ISBN 0-904175-54-5.

TAO Project Office, 2009a. TAO/TRITON data display and

delivery. NOAA Pacific Marine Environmental Laboratory.

http://www.pmel.noaa.gov/tao/disdel/disdel.


TAO Project Office, 2009b. TAO Climatologies. NOAA

Pacific Marine Environmental Laboratory. http://www.

pmel.noaa.gov/tao/clim/clim.html (accessed 7.5.09).

Zamudio, L., Hurlburt, H.E., Metzger, E.J., Morey, S.L.,

O’Brien, J.J., Tilburg, C.E., Zavala-Hidalgo, J., 2006.

Interannual variability of Tehuantepec eddies. J. Geophys.

Res. 111, C05001. doi:10.1029/2005JC003182.

C H A P T E R

S11

Indian Ocean: Supplementary Materials

FIGURE S11.1 Indian Ocean surface circulation (Tables S11.1, S11.2 and Figure 11.1). Surface height (cm). Data from

Niiler, Maximenko, and McWilliams (2003).

1

2

S11. INDIAN OCEAN: SUPPLEMENTARY MATERIALS

FIGURE S11.2 Surface circulation: (a) Southwest Monsoon (JulyeAugust) and (b) Northeast Monsoon (Januarye

February). Most current names are in Table S11.2. Red numbers are transports (Sv) taken from Schott and McCreary (2001).

Source: Modified from Schott, Dengler, and Schoenefeldt (2002).

S11. INDIAN OCEAN: SUPPLEMENTARY MATERIALS 3

FIGURE S11.3 Annual mean winds. Data from the NCEP reanalysis (Kalnay et al.,1996). (a) Wind stress (N/m 2 ) (vectors)

and wind-stress curl (10 7 N/m 3 ) (color), multiplied by 1 in the Southern Hemisphere. (b) Sverdrup transport (Sv),

where blue is clockwise and yellow-red is counterclockwise circulation. See also Figures 5.16 and 5.17.

4


FIGURE S11.3

(Continued).


FIGURE S11.4 Monthly mean wind stress (N/m 2 ) from NCEP climatology. (a) January: Southwest Monsoon. (b) April:

transitions. (c) July: Northeast Monsoon. (d) October: transitions. Monthly mean surface temperature ( C) from Levitus and

Boyer (1994) is shown in color. Source: From Schott, Dengler, and Schoenefeldt (2002). See also Figure 5.16.

6


FIGURE S11.5 Ocean surface color for (a) JulyeSept., 1979 (Southwest Monsoon) and (b) AprileJune, 1979 (Northeast

Monsoon), indicating the presence of high productivity (high chlorophyll) by red, orange, and yellow colors. Source: From

NASA Goddard Earth Sciences (2008).

FIGURE S11.6 Geostrophic stream function (1000 m 2 /s) at 900 m, from mean velocities measured by profiling floats.

ÓAmerican Meteorological Society. Reprinted with permission. Source: From Davis (2005).


FIGURE S11.7 Infrared boundary of the Agulhas Current from 1985 to 1988: (a) DecembereFebruary (summer) and

(b) JuneeAugust (winter). ÓAmerican Meteorological Society. Reprinted with permission. Source: From Quartly and

Srokosz (1993).

8


FIGURE S11.8 Changes in (a) surface temperature ( C), (b) surface salinity, (c) precipitation, and (d) surface winds due

to the closure of the Indonesian throughflow in a coupled ocean-atmosphere model. ÓAmerican Meteorological Society.

Reprinted with permission. Source: From Song, Vecchi, & Rosati (2007).


FIGURE S11.9 (a) Salinity at 32 S with isopycnal layers (black contours) and water mass labels (Table S11.3). (b)

Meridional transport in the isopycnal layers. Source: From Talley (2008). See also Figure 11.15.

10


FIGURE S11.10 Oxygen (mmol/kg) at 33 S in 1987. Station locations are on the inset maps. Source: From WOCE Indian

Ocean Atlas, Talley (2011a); see also Toole and Warren (1993). See also Figures 11.15, 11.16, and 11.21.


TABLE S11.1

Name

Subtropical gyre

Agulhas Current

Indian Ocean Mid-Latitude Circulation Elements (Southern Hemisphere)*

Description


Western boundary current of the subtropical gyre along the coast of

Africa

Agulhas Retroflection and Agulhas Return Current

Southeast Madagascar Current (SEMC)

South Indian Current (SIC)


West Australia Current

Leeuwin Current

Leeuwin Undercurrent


Deep Western Boundary Currents

Agulhas loop and return eastward flow off coast of southern Africa

Western boundary current of the subtropical gyre, flowing

southward along the coast of Madagascar


Eastward flow in the northernmost front of the Antarctic

Circumpolar Current

Broad northward flow in the eastern subtropical gyre

Poleward eastern boundary current along Australia

Equatorward undercurrent beneath the Leeuwin Current

Westward flow of the subtropical gyre

Northward flow on the western side of the deep basins


12


TABLE S11.2

Name

Indian Ocean Tropical and Monsoonal Circulation Systems (Southwest Monsoon: JulyeAugust;

Northeast Monsoon: JanuaryeFebruary)

Description

Arabian Sea circulation

Somali Current (SC)

Southern Gyre (SG)

Great Whirl (GW)

West Indian Coastal Current (WICC)

Bay of Bengal circulation

East Indian Coastal Current (EICC)

Northeast Monsoon Current (NMC), also

called North Equatorial Current (NEC)

Southwest Monsoon Current (SMC)

South Equatorial Countercurrent (SECC)

South Java Current (SJC)


Northeast Madagascar Current (NEMC)

and East African Coastal Current (EACC)

Reversing with monsoon

Low latitude, monsoonally reversing western boundary current for the

Arabian Sea circulation

Large eddy at western boundary on the equator during the Southwest

Monsoon

Large eddy at western boundary at 10 N during the Southwest Monsoon

Eastern boundary current of Arabian Sea gyre, along west coast of India,

reverses with monsoon

Reversing with monsoon

Western boundary current in the Bay of Bengal, along east coast of India,

reverses with monsoon

Westward flow north of and at the equator during the northeast monsoon

Eastward flow north of and at the equator during the southwest monsoon

Eastward open ocean zonal flow south of equator

Southward eastern boundary current connecting the SECC and SEC during

the Northeast Monsoon

Westward open ocean zonal flow

Low-latitude western boundary currents along coast of Madagascar and

coast of Africa, flowing northward connecting the SECC and SEC


TABLE S11.3

Water Mass

Principal Indian Ocean Water Masses*

Characteristic in the

Vertical Layer Process

South Indian Central Water

(CW)

Subtropical thermocline

waters

Upper

0e1000 m

Subduction

South Indian Subtropical

Underwater (STUW)

Subtropical/tropical

salinity maximum

Upper

100e200 m




(STMW)

Subtropical stability

(potential vorticity)

minimum

Upper

0e300 m


layer from Agulhas Current


(SAMW) and Southeast

Indian SAMW (SEISAMW)

Subantarctic stability

(potential vorticity)

minimum

Upper

0e700 m


layer from Subantarctic Front

Arabian Sea surface water

Warm, high salinity surface

water

Upper

0e200 m

Net evaporation in Arabian Sea,

Persian Gulf, and Red Sea

Bay of Bengal surface

water

Warm, low salinity surface

water

Upper

0e100 m

Net runoff and precipitation in

Bay of Bengal

Gulf Overflow Water

(GOW)

Indonesian Throughflow

Water (ITFW)

High salinity subsurface

water

Low salinity in South

Equatorial Current

Upper

200e350 m

Upper

0e500 m

Evaporation and cooling in

Persian Gulf and overflow into

Arabian Sea

Throughflow from Pacific Ocean

Indonesian Intermediate

Water (IIW)

Red Sea Overflow Water

(RSOW)

Low salinity in South

Equatorial Current

Salinity maximum

Intermediate

800e1200 m

Intermediate

400e1200 m

Throughflow from Pacific Ocean

Evaporation and cooling in Red

Sea and overflow into Arabian Sea


Water (AAIW)

Salinity minimum

Intermediate

500e1200 m

Advection of fresh subantarctic

surface water

Indian Ocean Deep Water

(IDW)

Oxygen minimum,

nutrient maximum

Deep

2000e3500 m

Mixing and aging of deep waters

including RSOW


(NADW)

Salinity maximum

Deep

2200e3500 m

Atlantic Ocean


Water (UCDW)

High oxygen, low

nutrients, high salinity on

isopycnal surfaces

Deep

~1000e3000 m

Mixture of deep waters in the

Southern Ocean

Lower Circumpolar Deep

Water (LCDW)

Deep salinity and oxygen

maxima, nutrient minima

Bottom

3000 m to bottom

Brine rejection in the Southern

Ocean mixed with NADW, PDW

and IDW


14


Reference




Kalnay, E., Kanamitsu, M., Kistler, R., Collins, W., Deaven, D.,

Gandin, L., et al., 1996. The NCEP-NCAR 40-year reanalysis

project. Bull. Am. Meteorol. Soc. 77, 437e471.

Levitus, S., Boyer, T.P., 1994. World Ocean Atlas 1994

Volume 4: Temperature. NOAA Atlas NESDIS 4. U.S.








Dynamically balanced absolute sea level of the

global ocean derived from near-surface velocity observations.

Geophys. Res. Lett. 30, 22. doi:10.1029/

2003GL018628.

Quartly, G.D., Srokosz, M.A., 1993. Seasonal variations in

the region of the Agulhas retroflection: studies with

Geosat and FRAM. J. Phys. Oceanogr. 23, 2107e2124.

Schott, F.A., Dengler, M., Schoenefeldt, R., 2002. The shallow

overturning circulation of the Indian Ocean. Progr.




Song, Q., G.A., Vecchi, G.A., Rosati, A.J., 2007. The role of

the Indonesian Throughflow in the Indo-Pacific climate

variability in the GFDL coupled climate model. J. Clim.

20, 2434e2451.


global overturning circulation: shallow, deep and


doi:10.1016/j.pocean.2008.05.001.

Talley, L.D., 2011a. Hydrographic Atlas of the World Ocean

Circulation Experiment (WOCE). Volume 3: Indian

Ocean. Sparrow, M., Chapman, P., Gould, J. (Eds.),

International WOCE Project Office, Southampton, U.K.

Online version: http://www-pord.ucsd.edu/whp_

atlas/indian_index.htm (accessed 4.20.09).

Toole, J.M., Warren, B.A., 1993. A hydrographic section

across the subtropical South Indian Ocean. Deep-Sea

Res. I 40, 1973e2019.

C H A P T E R

S12

The Arctic Ocean and Nordic Seas:

Supplementary Materials

FIGURE S12.1 Principal currents of the Nordic Seas. Shaded currents show upper ocean circulation; thin black

arrows show deep circulation. ÓAmerican Meteorological Society. Reprinted with permission. Source: From Østerhus and

Gammelsrød (1999).

1

2

S12. THE ARCTIC OCEAN AND NORDIC SEAS: SUPPLEMENTARY MATERIALS

(a)

(b)

FIGURE S12.2 Classical and recent structure of the Nordic Seas water column: (a) with deep convection and (b) with

intermediate depth convection. PW, Polar Water; RAW, Return Atlantic Water; AODW, Arctic Ocean Deep Water; and

NSDW, Norwegian Sea Deep Water. Source: From Ronski and Budéus (2005b).

S12. THE ARCTIC OCEAN AND NORDIC SEAS: SUPPLEMENTARY MATERIALS 3

FIGURE S12.3 (a) Salinity, (b) potential density (kg/m 3 ), and (c) potential temperature ( C) sections at 75 N across the

Greenland Sea “chimney.” The last shows the whole section, while the first two are expanded views of the chimney itself.

Source: From Wadhams, Holfort, Hansen, and Wilkinson (2002).

4


FIGURE S12.4 Monthly mean Arctic sea ice motion from 1979e2003 from Special Sensor Microwave Imager (SSM/I)

passive microwave satellite data. Extended from Emery, Fowler, and Maslanik (1997); data from NSIDC (2008a).


FIGURE S12.5 Mean sea level pressure from ERA-15 and NCEP-NCAR reanalyses for (a) winter (DecembereFebruary)

and (b) summer (JulyeSeptember) ÓAmerican Meteorological Society. Reprinted with permission. Source: From Bitz, Fyfe,

and Flato (2002).

-

-

-

-

6


(a)

Nansen B Amundsen B Lomo R

Pressure (dbar)

0

250

500

1000

2000

3000

4000

-29

-30

-33 - -35

-37

-38

-39

-40

-41

-42

-43

-44

-45

-

-47

-48

0.50

2.50

1.00

2.00

0.50

1.50

-0.50

-0.80

-0.95

-49

-50

-51

-52

-53

-54

1.00

1.00

0.00

-0.90

0.00

-55

-56

-1.70

0.50

-0.50

-57

-58

-59

-1.50

-60

-62

- -64

- -66

-

-68

-69

-1.00

1.50

-1.50

-1.70

1.00

-70 -

-72

(b)

0

0 100 200 300 400 500 600 700 800 900

-29

-30

-33 - -35

-37

-38

-39

-40

-41

-42

-43

-44

-45

-

-47

-48

34.00

Distance (km)

-49

-50

-51

-52

-53

-54

-55

-56

-57

-58

-59

-60

-62

- -64

- -66

-

-68

-69

34.00

33.50

-70 -

-72

Pressure (dbar)

250

500

1000

2000

3000

34.95

34.92

34.90

34.93

34.85

34.92

34.93

34.80

34.50

34.93

34.90 34.90

34.95

4000

34.94

S

(c)

Pressure (dbar)

0

250

500

1000

2000

3000

4000

0 100 200 300 400 500 600 700 800 900

-29

-30

-33 - -35

-37

-38

-39

-40

-41

-42

-43

-44

-45

-

-47

-48

27.95

27.50

Distance (km)

28.05

28.08

28.09

28.10

-49

-50

-51

-52

-53

-54

27.90

27.95

28.00

28.05

0 100 200 300 400 500 600 700 800 900

Distance (km)

-55

-56

27.80

-57

-58

-59

27.50

-60

-62

- -64

- -66

-

-68

-69

FIGURE S12.6 (a) Potential temperature ( C), (b) salinity, and (c) potential density s q in the Eurasian Basin, with the

Russian coast on the left and Lomonosov Ridge at the right. Source: From Schauer et al. (2002).

27.20

-70 -

-72

o


FIGURE S12.7 Volume transport budget for the Nordic Seas. Source: From Hansen et al. (2008).

8


(a)

(b)

FIGURE S12.8 (a) Winter (spring 1949) and (b) melting (spring 1950) sea ice in the Beaufort Sea. http://www.photolib.

noaa.gov/htmls/corp1014.htm and http://www.photolib.noaa.gov/htmls/corp1104.htm (NOAA Photo Library, accessed

2009.) (Photographer: Harley D. Nygren.)


FIGURE S12.9 Ice cover in the Barents Sea in early June 1994, using NASA AVHRR near-infrared imaging. Black

indicates lack of ice (open water and polynyas). Source: From Anselme (1998).

FIGURE S12.10 Kara Sea polynya distribution for JanuaryeApril 2001. Light gray indicates land: Novaya Zemlya is at

the bottom of the image. Dark gray is masked regions. Source: From Kern (2008).

10


FIGURE S12.11 The Laptev Sea flaw polynya, imaged using Envisat advanced synthetic aperture radar, 1 May 2008. The

polynya region is labeled as “new ice.” Source: From Dmitrenko et al. (2010).


TABLE S12.1

Major Nordic Seas Water Masses

Water Mass Acronym Depth Characteristic Properties Source

Polar (Arctic) Surface

Water

PSW (ASW) Surface to 25e50 m 1.5 to l.9 C

28 to 33.5 psu (polar mixed

layer and halocline)

S: 28 to 33.5

Local (associated with

sea ice) and inflow from

Arctic

Atlantic Water AW 200e900 m >3 C

>34.9 psu

(q and salinity maximum)

Norwegian Atlantic

Current flow

Arctic Intermediate

Water

AIW

Upper ocean to

1200 m

1.2 C, 34.88 psu (salinity

minimum at ~800 m)

Intermediate depth

convection in the

Greenland and Iceland

Seas

Upper Polar Deep

Water

uPDW

800e1500 m

(Nordic Seas)

0.5e0 C

34.85 to 34.9 psu (salinity

minimum)

Upper Polar Deep

Water from the Arctic

Ocean

Arctic Ocean Deep

Water

Greenland Sea Deep

Water

AODW 2000 m to bottom 0.53 C, >34.95 psu

-0.4 to -0 2 C

GSDW 2000 m to bottom < 1.2 C

34.88e34.90 psu

Canadian and Eurasian

Basin Deep Waters

from the Arctic Ocean

Greenland Sea deep

convection

Norwegian Sea Deep

Water

NSDW 2000 m to bottom 1.055 C, 34.91 psu GSDW and mixing

Source: After Aagaard, Swift, and Carmack, 1985 and Rudels et al., 2005.

12


TABLE S12.2

Name

Arctic Ocean and Nordic Seas Surface Circulation Elements (Partial List)

Description

Transpolar Drift (TPD)

Beaufort Gyre

“Rim” current

Siberian Coastal Current

Alaskan Coastal Current

Bering Strait inflow

West Spitsbergen Current (WSC)

Norwegian Atlantic Current (NAC)

Norwegian Coastal Current (NCC)

East Greenland Current (EGC)

Jan Mayen Current (JMC)

East Iceland Current (EIC)

Iceland-Faroe Front (IFF)

Irminger Current (IC) and North Irminger

Current (NIC)

West Greenland Current

Baffin Current

Labrador Current

Canadian Archipelago flows

Broad drift across Arctic from Siberian region to Greenland

Anticyclonic gyre in the Canadian Basin

Cyclonic coastal flow around the Arctic

Portion of rim current along the Siberian coast

Portion of rim current along the Alaskan coast

Inflow to the Arctic from the Bering Sea

Northward flow through Fram Strait

Northward eastern boundary flow in the Nordic Seas

Northward coastal current along Norway in the Nordic Seas

(rim current along Norwegian coast)

Southward western boundary current along Greenland coast

Eastward flow branching from the EGC into the Greenland

Sea toward Jan Mayen

Southeastward flow branching from the EGC in the Iceland

Basin

Eastward flow along the Iceland-Faroe Ridge

Northward flow in the North Atlantic along the western flank

of the Reykjanes Ridge, and its northeastward branch around

Iceland

Northward eastern boundary flow along Greenland in the

Labrador Sea and Baffin Bay

Southward western boundary flow in Baffin Bay

Southward western boundary flow in the Labrador Sea

Outflow from Arctic to Baffin and Hudson Bays through the

many island passages, including Lancaster Sound, Jones

Sound, and Nares Strait

TABLE S12.3

Major Arctic Ocean Water Masses a

Water Mass Acronym Depth Characteristic Properties Source

Polar Surface Water PSW Surface to 25e50 m 1.5 to l.9 C

31e34 psu

S: 28 to 33.5

Alaskan Coastal Water ACW 1.1 to 1.2 C

31e32 psu (temperature

maximum)

summer Bering Strait

Water (Pacific Summer

Water)

winter Bering Strait Water

(Pacific Winter Water)

Atlantic Water (Atlantic

Layer)

sBSW 70e130 m 1.3 C

32e33 psu (temperature

maximum)

Local, associated with sea

ice, river runoff; includes

Polar Mixed Layer and

halocline

Surface water with river

runoff

Bering Strait summer flow

wBSW 33.1 psu Bering Strait winter flow

AW 200e1000 m 0e 3 C

>34.9 psu (q maximum)

upper Polar Deep Water uPDW 1000e1700 m 0.5e 0 C 34.85e34.9 psu

Canadian Basin Deep

Water; bottom water

Eurasian Basin Deep

Water; bottom water

Source: From Jones, 2001 and Steele et al., 2004.

CBDW 1700 m to bottom 0.53 C

~34.95 psu

0.4 to 0.2 C

EBDW 1700 m to bottom 0.95 C

~34.94 psu

Norwegian Atlantic

Current inflow, modified

within the Arctic

Eurasian Basin; bottom

water isolated

Brine-rejected shelf waters

and Greenland Sea Deep

Water


14


References

Aagaard, K., Swift, J.H., Carmack, E.C., 1985. Thermohaline

circulation in the arctic mediterranean seas. J. Geophys.

Res. 90, 4833e4846.

Anselme, B., 1998. Sea ice fields and atmospheric

phenomena in Eurasiatic arctic seas as seen from the

NOAA-12 satellite. Int. J. Remote Sens. 19, 307e316.

Bitz, C.M., Fyfe, J.C., Flato, G.M., 2002. Sea ice response

to wind forcing from AMIP models. J. Clim. 15,

522e536.

Dmitrenko, I.A., Wegner, C., Kassens, H., Kirillov, S.A.,

Krumpen, T., Heinemann, G., et al., 2010. Observations

of supercooling and frazil ice formation in the Laptev

Sea coastal polynya. J. Geophys. Res. 115 C05015.

doi:10.1029/2009JC005798.

Emery, W.J., Fowler, C.W., Maslanik, J.A., 1997. Satellite

derived Arctic and Antarctic sea ice motions: 1988e1994.

Geophys. Res. Lett. 24, 897e900.

Hansen, B., Østerhus, S., Turrell, W.R., Jónsson, S.,

Valdimarsson, H., Hátún, H., et al., 2008. The inflow of

Atlantic water, heat, and salt to the Nordic Seas across

the Greenland-Scotland Ridge. In: Dickson, R.R.,

Meincke, J., Rhines, P. (Eds.), Arctic-Subarctic Ocean

Fluxes: Defining the Role of the Northern Seas in

Climate. Springer, The Netherlands, pp. 15e44.

Jones, E.P., 2001. Circulation in the Arctic Ocean. Polar Res.

20, 139e146.

Kern, S., 2008. Polynya area in the Kara Sea, Arctic, obtained

with microwave radiometry for 1979e2003. IEEE Geosc.

Remote Sens. Lett. 5, 171e175.

NSIDC, 2008a. Polar Pathfinder Daily 25 km EASE-Grid

Sea Ice Motion Vectors. National Snow and Ice Data

Center. http://nsidc.org/data/docs/daac/nsidc0116_

icemotion.gd.html (accessed 02.01.09).

Østerhus, S., Gammelsrød, T., 1999. The abyss of the Nordic

Seas is warming. J. Clim. 12, 3297e3304.


convection in the Greenland Sea. J. Geophys. Res. 110

C04015. doi:10.1029/2004JC002318.

Rudels, B., Bjork, G., Nilsson, J., Winsor, P., Lake, I., Nohr, C.,

2005. Interaction between waters from the Arctic Ocean

the Nordic Seas north of Fram Strait and along the East

Greenland Current: results from the Arctic Ocean-20

Oden expedition. J. Marine. Syst. 55, 1e30.

Schauer, U., Rudels, B., Jones, E.P., Anderson, L.G.,

Muench, R.D., Björk, G., et al., 2002. Confluence and

redistribution of Atlantic water in the Nansen, Amundsen

and Makarov basins. Ann. Geophys. 20, 257e273.



water in the Arctic Ocean. J. Geophys. Res. 109,

C02027. doi:10.1029/2003JC002009.

Wadhams, P., Holfort, J., Hansen, E., Wilkinson, J.P., 2002.

A deep convective chimney in the winter Greenland Sea.


C H A P T E R

S13

Southern Ocean: Supplementary

Materials

FIGURE S13.1 Mixed layer thickness in the southern hemisphere in late winter (September), based on Argo float profiles

(depth that is 0.03 kg m 3 denser than the surface value). Source: From Dong, Gille, Sprintall, and Talley (2008).

1

2

S13. SOUTHERN OCEAN: SUPPLEMENTARY MATERIALS

FIGURE S13.2 Cyclonic eddy just south of the SAF at 132 E. (a) Temperature section along 132 E in 1977. (b) Depth of

the 3.5 C isotherm; arrows are the ship drift. Source: From Savchenko, Emery, and Vladimirov (1978).

S13. SOUTHERN OCEAN: SUPPLEMENTARY MATERIALS 3

FIGURE S13.3

Antarctic ice shelves that are monitored using satellite imagery. Source: From NSIDC (2009c).

4

TABLE S13.2

Southern Ocean Water Masses

S13. SOUTHERN OCEAN: SUPPLEMENTARY MATERIALS

Water Mass and

Acronym Location Characteristic Properties Source

Subantarctic Surface

Water (SASW)

Surface north of the SAF Warm, salty Local

Subantarctic Mode

Water (SAMW)

Upper ocean north of

the SAF

Vertical thickness

maximum

Thick mixed layers just

north of the SAF

Antarctic Surface Water

(ASW)

Surface layer south of

the PF

Cold, fresh, extends down

to temperature minimum

(“Winter Water”) at the top

of the CDW

Local

Continental Shelf Water Surface to the shelf bottom Freezing temperature,

density of AABW

Properties set by sea ice

formation.


Water (AAIW)

North of the SAF at

500e1500 m depth

Vertical salinity minimum

Fresh surface water north

of the SAF around the

Drake Passage

Upper Circumpolar

Deep Water (UCDW)

Throughout the Southern

Ocean, at 200 to 1700 m

depth

Vertical oxygen minimum

Pacific and Indian Deep

Waters provide the oxygen

minimum

Lower Circumpolar

Deep Water (LCDW)


Ocean, at 200 to 4000 m

depth

Vertical salinity maximum


provides the salinity

maximum


(AABW)


Ocean, near-bottom layer

that spreads north from

the ACC

Cold, dense, relatively

fresh bottom layer

Dense shelf waters formed

by brine rejection

Weddell Sea Deep

Water (WSDW)

Weddell Sea, most of the

water column

Cold, dense, thick layer

between the CDW above

and the bottom water

below

Mixture of CDW and dense

shelf waters formed by

brine rejection

Weddell Sea, Ross Sea

and Adélie Land

Bottom Water (WSBW,

RSBW, ALBW)

Weddell Sea, Ross Sea,

coast of Adélie Land,

bottom

Cold, densest bottom

layers

Densest shelf waters

formed in these seas

References

Dong, S., Gille, S., Sprintall, J., Talley, L., 2008. Southern

Ocean mixed-layer depth from Argo float profiles.

J. Geophys. Res. 113, C06013. doi:10.1029/2006JC004051.

NSIDC, 2009c. Images of Antarctic Ice Shelves. National


iceshelves_images/index.html (accessed 3.5.09).

Savchenko, V.G., Emery, W.J., Vladimirov, O.A., 1978.

A cyclonic eddy in the Antarctic Circumpolar Current

south of Australia: results of Soviet-American observations

aboard the R/V Professor Zubov. J. Phys. Oceanogr.

8, 825e837.

C H A P T E R

S14

Global Circulation and Water Properties:

Supplementary Materials

(a)

Southern Ocean



SAMW, AAIW

Low, mid- latitude upper ocean waters

LCDW

UCDW

Pacific-Indian

upwelling &

diffusion

PDW/IDW

Antarctica

AABW

formation

(brine

rejection)

NADW

PDW/IDW

formation

(diffusion)

NADW

formation

(convection)

AABW

FIGURE S14.1 (a) Two-dimensional schematic of the interconnected NADW, IDW, PDW, and AABW cells of Figure

14.13. (b). Global overturning schematic that mirrors the globally-averaged overturning streamfunction, hence concealing

deep upwelling in the Indian and Pacific Oceans. (c) Implied global overturning in the Broecker schematic of Figure 14.12,

which ignores the Southern Ocean upwelling and AABW formation. ÓAmerican Meteorological Society. Reprinted with

permission. Source: From Talley (submitted, 2011b).

1

2

S14. GLOBAL CIRCULATION AND WATER PROPERTIES: SUPPLEMENTARY MATERIALS

(b)

Southern Ocean



SAMW, AAIW

LCDW

Antarctica

AABW

formation

(brine

rejection)

NADW

NADW

formation

(convection)

AABW

(c)

Low, mid- latitude upper ocean waters

ITF

Pacific-Indian

upwelling &

diffusion

PDW/IDW

NADW

formation

(convection)

FIGURE S14.1

(Continued).

S14. GLOBAL CIRCULATION AND WATER PROPERTIES: SUPPLEMENTARY MATERIALS 3

40˚N

20˚N

0˚

20˚S

40˚S

60˚N

(BS)

80˚N

0.38 PW

0.17 Sv

80˚W

40˚W

-0.01 PW

0.06 Sv (BS)

-0.10 PW

0.20 Sv

0˚

40˚E

80˚E

120˚E

160˚E

160˚W

Upper ocean

Heat transport divergences (total) (PW) (BS)

Freshwater transport divergences (Sv)

-0.10 PW

0.23 (ITF)

-0.33 PW

0.18 Sv

0.55 PW

-0.08 Sv

120˚W

0.08 PW

-0.16 Sv (ITF)

-0.35 PW

0.23 Sv

80˚N

60˚N

40˚N

0.06 PW

-0.07 Sv (BS)

40˚S

20˚N

0˚

20˚S

60˚S

60˚S

80˚S

80˚W

40˚W

0˚

40˚E

80˚E

FIGURE S14.2 Estimated mean annual meridional (a) heat transport (1 PW ¼ 10 15 W) and (b) freshwater transport (Sv)

by the subtropical gyres (black contours), resulting from poleward mass transport in the western boundary currents and

subducted equatorward return flow, all within and above the main pycnocline. The contributions of the Indonesian

Throughflow (magenta contours) and Bering Strait (magenta) are also shown. After Talley (2003, 2008).

120˚E

160˚E

160˚W

120˚W

80˚S

4


(a)

20˚N

40˚N

60˚N

80˚N

80˚W 40˚W 0˚ 40˚E 80˚E 120˚E 160˚E 160˚W 120˚W

0.91 PW (S/AAIW to LSW/NADW)

-0.02 PW (AABW to NADW)

Heat transports

Surface

Intermediate

Deep

Bottom

80˚N

60˚N

40˚N

0.13 PW (S to NPIW)

-0.003 PW (AABW to PDW)

20˚N

0˚

0˚

20˚S

40˚S

0.38 PW (S/AAIW to NADW)

-0.03 PW (AABW to NADW)

-0.16 PW (AABW to S)

-0.08 PW (AABW to IDW/AAIW)

40˚S

20˚S

-0.13 PW (AABW to PDW/AAIW)

60˚S

60˚S

80˚S

80˚W 40˚W 0˚ 40˚E 80˚E 120˚E 160˚E 160˚W 120˚W

80˚S

(b)

20˚N

40˚N

60˚N

80˚N

Freshwater transports

Surface

Intermediate

Deep

-0.54 Sv (S/AAIW to LSW/NADW) Bottom

0.01 Sv (AABW to NADW)

80˚N

60˚N

40˚N

-0.07 Sv (S to NPIW)

0.01 Sv (AABW to PDW)

20˚N

0˚

0˚

20˚S

40˚S

0.01 Sv (S/AAIW to NADW)

0.01 Sv (AABW to NADW)

0.01 Sv (AABW to S)

-0.05 Sv (AABW to IDW/AAIW)

40˚S

20˚S

-0.03 Sv (AABW to PDW/AAIW)

60˚S

60˚S

S

AAIW

LSW

80˚S

80˚W 40˚W 0˚ 40˚E 80˚E 120˚E 160˚E 160˚W 120˚W

Surface


Labrador Sea Water

NADW

PDW

IDW

AABW

80˚S


Pacific Deep Water

Indian Deep Water


FIGURE S14.3 Estimates of mean annual meridional (a) heat (PW) and (b) freshwater (Sv) transport by elements of the

overturning circulation. Acronyms indicate the type of overturn; for instance, AABW to NADW means that the listed

transport is associated with overturn of AABW to NADW (upwelling in this instance). After Talley (2003, 2008).


FIGURE S14.4 Fraction of waters on the isoneutral surface g N ¼ 28.06 kg/m 3 (s 4 ~ 45.84 kg/m 3 , at a depth of 2500e3000 m

north of the ACC) that are (top) North Atlantic Deep Water and (bottom) Antarctic Bottom Water, calculated as in Figure S14.5.

Personal communication, Gregory C. Johnson (2009).

6


FIGURE S14.5 Fraction of bottom waters that are (top) North Atlantic Deep Water and (bottom) Antarctic Bottom Water,

from an optimum multiparameter analysis using as inputs the properties of NADW at a location just south of Greenland,

downstream from the Nordic Seas Overflows, and of AABW in the Weddell Sea. Source: From Johnson (2008).


FIGURE S14.6 Eddy kinetic energy from geostrophic velocities calculated from satellite altimetry from 1992 to 1998. The

equatorial band is blank because geostrophic velocities cannot be calculated there. This is a companion to Figure 14.16 in the

text. Source: From Ducet, Le Traon, and Reverdin, (2000).

8


FIGURE S14.7 Horizontal eddy diffusivity (m 2 /sec) at the sea surface (color) in the Atlantic Ocean, with mean velocity

vectors, based on surface drifter observations. This is a companion to Figure 14.17a. Source: From Zhurbas and Oh (2004).


FIGURE S14.8 Eddy diffusivity ellipses at 900 m in the Indian Ocean based on subsurface float velocities. Colors indicate

different scales (see figure header). This is a companion to Figure 14.17b. ÓAmerican Meteorological Society. Reprinted with

permission. Source: From Davis (2005).

References




Ducet, N., Le Traon, P.Y., Reverdin, G., 2000. Global highresolution

mapping of ocean circulation from TOPEX/

Poseidon and ERS-1 and -2. J. Geophys. Res. 105,

19477e19498.

Johnson, G.C., 2008. Quantifying Antarctic Bottom Water

and North Atlantic Deep Water volumes. J. Geophys.

Res. 113, C05027. doi:10.1029/2007JC004477.

Talley, L.D., 2003. Shallow, intermediate, and deep overturning

components of the global heat budget. J. Phys.





doi:10.1016/j.pocean.2008.05.001.

Talley, L.D., 2011b. Schematics of the global overturning

circulation. J. Phys. Oceanogr., submitted.

Zhurbas, V., Oh, I.S., 2004. Drifter-derived maps of lateral

diffusivity in the Pacific and Atlantic Oceans in relation

to surface circulation patterns. J. Geophys. Res. 109,

C05015. doi:10.1029/2003JC002241.

C H A P T E R

S15

Climate and the Oceans

This chapter on climate variability and

climate change appears only on the textbook

Web site http://booksite.academicpress.com/

DPO/. The first section introduces climate variability

and climate change. This is followed by

a section on climate modes and variability for

each of the ocean basins in the order of the basin

chapters 9 through 13 (Atlantic, Pacific, Indian,

Arctic, and Southern Ocean). The final section

summarizes global ocean observations indicative

of climate change. Each of the basinoriented

subsections can be read as an

addendum to the print chapter for that basin.

S15.1. INTRODUCTION

S15.1.1. Definitions

Climate is variability in any part of the oceanatmosphere-land-ecology

system on timescales

that are longer than seasonal. Climate variations

can be due to natural, internal interactions

between components of the system; to natural,

external forcing; or to anthropogenic external

forcing. In present-day usage, climate variability

usually refers to natural climate variability and

climate change refers to anthropogenically forced

variations in climate. Examples of internal

interactions include some of the feedbacks

described in this text (i.e., the Bjerknes tropical

ocean-atmosphere feedback or the ice-albedo

feedback), and a plethora of others that are

described in books and journals devoted to the

topic of climate variations. Examples of natural

external forcing include variations in the Earthmoon-Sun

orbits, and random but ongoing

volcanism. Examples of anthropogenic forcing

include burning of fossil fuels, and changes in

land surfaces due to patterns of land use;

changes in these forcings can then set off feedbacks

that could move the climate system to

a different state. We categorize climate timescales

as interannual (roughly longer than

a year and shorter than about eight years),

decadal (roughly one to several decades), centennial,

millennial, and longer. For the shorter timescales,

time series observations of the property

of interest are used. For the ocean these include

temperature, salinity, oxygen, nutrients, carbon

parameters, current velocities, surface height,

and so forth. For the longer timescales, which

are the realm of paleoclimate studies, “proxy”

records of something that depends on the

property of interest are used such as the prevalence

of different types of benthic foraminifera

or the size of a tree ring, which can be related

to changes in temperature, precipitation, and

so on.

Most of this text is concerned with the mean

structure of the ocean circulation and properties,

and for some regions, its seasonal variability.

Climate is included in this descriptive

oceanography text because it affects ocean

1

2

S15. CLIMATE AND THE OCEANS

variability in properties and circulation on timescales

of years to millennia. Climate variability

and climate change usually result in small

changes, of the order of 10%, in the mean structures.

The most energetic modes of variability,

which are the tropical modes such as El Niño-

Southern Oscillation, result in much larger

changes in the structure, but even these do not

eliminate the mean pycnocline or vertical

temperature structure.

As another example, no climate variability or

change would ever remove the importance of

western boundary currents, because their existence

is due to Earth’s rotation and the presence

of ocean boundaries, neither of which will

vanish although the boundaries do change on

geological timescales. Moreover, Earth will

continue to be heated in the tropics and cooled

at high latitudes, and the organization of the

major wind systems is unlikely to change.

(These include the easterly and westerly winds

associated with the Hadley, Ferrel, and Polar

cells, and the tropical circulations such as the

Walker circulation and monsoons.) Climate

variations are therefore unlikely to cause the

demise or complete reorganization of the Gulf

Stream or any of the other western boundary

currents. The strength might change, the position

of the separated currents might shift somewhat,

and the advected properties might be

somewhat altered, but the basic structure would

remain as long as the general surface wind

forcing patterns exist.

On the other hand, major changes in ocean

stratification due to heating, precipitation and

evaporation, and sea ice could change the

strength and structure of the overturning circulation.

Such changes are apparent in paleoclimate

records such as the production and

properties of North Atlantic Deep Water were

vastly altered during the last glaciation.

Changes in stratification in the tropics could

alter the El Niño-Southern Oscillation (ENSO)

feedbacks, with a warmer, more stratified ocean

much less capable of producing the cold tongue

of the eastern tropical Pacific. Since ENSO

affects temperature and precipitation over

a large part of the globe, such changes could

have widespread consequences.

S15.1.2. Natural Modes of Climate

Variability

In each of the subsequent sections focused on

each ocean basin, some of the most energetic

(natural) modes of interannual and decadal

climate variability are described. These are

summarized in Table S15.1. Each mode has

been described in terms of an index that can be

calculated and plotted over many decades.

Correlations of surface temperature and sea level

pressure with many of the indices are shown in

later sections and are listed in the table.

Climate modes with dominantly interannual

variability are (a) the El Niño-Southern Oscillation

(ENSO), (b) the Atlantic Meridional Mode

(AMM), Atlantic Niño, and (c) the Indian Ocean

Dipole (IOD) mode. These are all tropical modes

of variability with their relatively high

frequency set by tropical dynamics; all of these

modes include strong feedback between the

atmosphere and ocean. ENSO is the most energetic

of these globally by far, with a sea level

pressure pattern that includes the centraleastern

tropical Pacific (one sign), western tropical

Pacific and eastern Indian Ocean (opposite

sign), and the tropical Atlantic (opposite sign),

with a signature over North America and in

the western tropical Atlantic (Figure 10.28 and

Section S15.2.1). The intrinsic Atlantic and

Indian interannual modes are mostly confined

within their own ocean basins.

Climate modes with dominantly decadal

variability are (a) the North Atlantic Oscillation

(NAO) and closely related Arctic Oscillation

(AO or Northern Annular Mode, NAM), (b)

the Antarctic Oscillation (AAO or Southern

Annular Mode, SAM), and (c) the Pacific

Decadal Oscillation (PDO) with its closely

related modes that are defined within the North

INTRODUCTION 3

TABLE S15.1

Some of the Principal Modes of Natural Climate Variability

Climate Mode

Acronym

Approximate

Timescale

Section and Map

Atlantic Meridional Mode AMM Interannual Sections S15.2.1 and S15.5,

Figure S15.1

Atlantic Niño d Interannual Section S15.2.1, Figure S15.1

Arctic Oscillation (also called Northern

Annular Mode) and the closely related

North Atlantic Oscillation

AO (NAM)

NAO

Decadal

Sections S15.2.2 and S15.5,

Figure S15.2

East Atlantic Pattern EAP Decadal Section S15.2.2, Figure S15.2

Atlantic Multidecadal Oscillation AMO Multidecadal Section S15.2.2, Figure S15.2

Section S15.5

El Niño-Southern Oscillation (Southern

Oscillation Index)

ENSO (SOI) Interannual Section 10.8, Figure 10.28

Pacific Decadal Oscillation (closely related:

North Pacific Index and Pacific North

American teleconnection)

PDO (NPI)

(PNA)

Decadal Section S15.3, Figure S15.5

North Pacific Gyre Oscillation NPGO Decadal Section S15.3, Figure S15.6

Indian Ocean Dipole mode IOD Interannual Section S15.4

Antarctic Oscillation (also called the

Southern Annular Mode)

AAO (SAM) Decadal Section S15.6, Figure S15.15

Pacific. These modes do not tend to have strong,

obvious feedbacks between the ocean and atmosphere,

although much work has been done and

continues to be done on such mechanisms.

The only centennial mode described herein

and listed in Table S15.1 is the AMO, which is

defined in terms of basin-wide sea-surface

temperature (SST) averages and is presumed

to be linked to variations in the meridional overturning

circulation (MOC). The AMO also

affects the Arctic Ocean.

The AAO (SAM) and the PDO have similar

spatial patterns, and both have similarities

with the interannual ENSO pattern. That is, all

three have strongest signatures in the Pacific,

Indian, and Antarctic, as if the Pacific region is

connected primarily zonally to the Indian and

meridionally to the Antarctic. There is little

correlation with the Arctic. The NAO (AO)

pattern, on the other hand, connects the Atlantic

Ocean meridionally with the Arctic with little

signature in the Southern Hemisphere, even in

the tropical Atlantic.

These modes of climate variability are

described in subsequent sections as simply as

possible, as if they were standing patterns,

with the oceans and land determining to some

extent the location of the nodes. Many of the

modes are also analyzed in terms of lagged

correlations and large-scale wavelike propagation,

but this is beyond the scope of this text.

These natural modes of climate variability

not only have importance for regional climate

variation in the ocean, but are also the natural

modes of the entire system that could be forced

anthropogenically. Shifts into a particular phase

of modes such as ENSO, the PDO, SAM, and

NAO/AO are sometimes suggested by climate

4


prediction models. As we begin to consider

consequences of continuing changes in greenhouse

gases, particulates in the atmosphere,

and land use, some of the hypotheses naturally

involve projection of climate change forcing on

these climate modes.

S15.2. CLIMATE AND THE

ATLANTIC OCEAN

Atlantic climate research tends to be focused

on decadal and longer term variability centered

on the northern North Atlantic’s deep-water

formation processes and on sea ice processes

in the Nordic Seas and Arctic (Section S15.5).

This is because the mean ventilation age of

northern North Atlantic deep waters is on the

order of decades or less with associated measurable

variability. However, climate variability at

all timescales from interannual to decadal,

centennial, and millennial has been documented

and affects all regions of the Atlantic

(Table S15.1). Trends that have been related to

climate change (anthropogenic forcing) have

also been documented.

S15.2.1. Tropical Atlantic Variability

Interannual variability studies are focused on

the tropical Atlantic, where there are several

modes, including two intrinsic to the Atlantic.

These are (1) the Atlantic Meridional Mode

(AMM), which is a cross-equatorial mode; (2)

the Atlantic Niño, which is a zonal equatorial

mode that is dynamically similar to ENSO

with a tropical Bjerknes feedback (Section

7.9.2); and (3) remote forcing from the Pacific

ENSO. None of these modes is overwhelmingly

dominant in the sense of the Pacific’s ENSO.

Variability in the upper ocean is linked to these

modes. Variability at intermediate and abyssal

depths may have other sources and timescales.

Tropical Atlantic variability is regularly monitored

with the PIRATA array (Section S16.5.6.2

and Figure S16.38), which was designed to

sample both the meridional and zonal modes

(Bourlès et al., 2008).

AMM has SST anomalies of opposite sign on

either side of the equator: warm SST to the north

and cold SST to the south and vice versa (Figure

S15.1a). Because of these opposing anomalies,

the AMM is also called the “tropical dipole

mode.” Surface wind anomalies blow toward

the warm SST. During positive AMM, the Intertropical

Convergence Zone (ITCZ), which lies in

the Northern Hemisphere, is displaced northward.

The AMM’s full Atlantic hemispheric

pattern includes alternating highs and lows

from the Nordic Seas to the Southern Ocean,

but its amplitude is largest in the tropics, while

the North Atlantic Oscillation, whose spatial

pattern it resembles, has highest amplitude in

the north. The AMM has a seasonal cycle, peaking

in boreal spring, and interannual to decadal

variability. Decadal variation in the AMM has

been described in terms of a wind-evaporation-SST

feedback 1 (Chang, Ji, & Li, 1997; Kushnir,

Seager, Miller, & Chiang, 2002; Figure

S15.1e), but the feedback is weak (Sutton, Jewson,

& Rowell, 2000; Chiang &Vimont, 2004).

External forcing, for instance from the NAO or

Pacific’s ENSO, appears to be necessary to maintain

the decadal energy.

The Atlantic Niño, also known as the

“Atlantic zonal equatorial mode” (Figure

1 A feedback diagram is shown in Figure S15.1e. Starting with a positive SST dipole (warm north of the equator), the surface

winds blow northwestward south of the equator and northeastward north of the equator. This decreases the easterly trade

winds in the Northern Hemisphere, which reduces the evaporative heat flux in the Northern Hemisphere, since evaporative

heat flux is proportional to wind speed. This enhances the SST anomaly there, hence is a positive feedback. The

system is restored by a slower negative feedback involving advective heat flux in the ocean with the cooler southern waters

advected northward by the North Brazil Current that is strengthened by the winds.


(e)

High N. hem. SST

Low S. hem. SST

Atlantic Meridional Mode feedback (Chang et al., 1997)

(+ positive feedback) (fast)

Weaker N. Hem. trades

Stronger S. Hem. trades

Reduced N. Hem. evap. heat flux

Stronger S. Hem. evap. heat flux

Stronger northward cross-equatorial ocean flow N. hem. advective cooling

(- negative feedback) (slow)

FIGURE S15.1 Tropical Atlantic interannual climate modes.(a, b) Atlantic Meridional Mode: SST correlation with the

AMM index for 1948e2007, all months and monthly time series (light) of the AMM index, with a one-year running mean

(heavy). (Data and graphical interface from NOAA ESRL, 2009b). (c, d) Atlantic “Niño” (zonal equatorial mode): SSTanomalies and

time series of temperature averaged in the cold-tongue region 3 Se3 N, 20 We0 (“ATL3 index”). High values correspond to the

Niño state (weak or absent cold tongue). ÓAmerican Meteorological Society. Reprinted with permission. Source: From Wang

(2002). (e) Feedbacks for AMM decadal variability. Arrowheads mean an upward trend in the cause results in an upward trend

in the result, circles indicate upward trend resulting in negative trend. (Based on Chang et al., 1997; Kushnir et al., 2002).

S15.1c,d) has the typical Bjerknes tropical feedback

between the ocean’s SST and atmosphere’s

winds (Section 7.9.2). The timescale of the

Atlantic Niño is interannual, on the order of 30

months, but with considerable randomness. In

the normal seasonal cycle, a cold tongue appears

in the central and eastern Atlantic every boreal

summer (Figure S15.1c). The seasonal cold

tongue occupies a large fraction of the equatorial

Atlantic, with coldest temperatures less

6


than 24 C, comparable to the Pacific’s seasonal

cold tongue temperatures (Section 10.7.3). The

western warm pool in the Atlantic, at about

28 C, is cooler and more spatially limited than

the Pacific’s warm pool (>30 C; Figure 10.25).

During an Atlantic Niño, warm SST anomalies

almost obliterate the cold tongue (e.g., 1998 in

Figure S15.1d). This is accompanied by an eastward

shift and weakening of the Atlantic’s

Walker circulation, with rising air over the

maximum anomaly in the central Atlantic, and

a strengthening of the Hadley circulation

(Wang, 2002).

The Atlantic Niño has lower amplitude and

a smaller geographical impact than the Pacific’s

ENSO. The simplest explanation is that the

Atlantic is much narrower than the Pacific, so

the thermocline depth variation in the east and

the associated SST anomalies are weaker in the

Atlantic (Jin, 1996). Since the mean western

warm pool is much narrower and cooler than

in the Pacific, Atlantic anomalies there are also

weaker than in the Pacific.

The Pacific’s ENSO reaches eastward into the

tropical Atlantic (Wang, 2002). During an El

Niño warm event, the Pacific’s Walker circulation

shifts eastward with ascending air moving

to the central and eastern equatorial Pacific.

The descending branch of this anomalous

Walker circulation is in the central Atlantic

with strongest effects on SST in the tropical

North Atlantic. Tropical Atlantic SST anomalies

lag a Pacific El Niño warm event by five to six

months.

S15.2.2. Decadal and Multidecadal

Variability

North Atlantic decadal variability is often

interpreted in terms of the North Atlantic Oscillation

(NAO) and East Atlantic Pattern (EAP),

which are internal modes of the atmosphere at

short timescales that have important decadal

and longer term variability that might involve

feedbacks with the ocean. The NAO is closely

related to the AO (NAM) (Section S15.5). In

the South Atlantic, decadal climate variability

is associated with the SAM (Section S15.6). The

Atlantic Multidecadal Oscillation (AMO) represents

a longer timescale natural mode of the

Atlantic overturning circulation associated

with surface temperatures throughout the

North Atlantic.

The NAO is one of the most vigorous and

best described of Earth’s natural climate modes

(Hurrell, Kushnir, Ottersen, & Visbeck, 2003;

Visbeck et al., 2003). In the mean, the North

Atlantic’s westerly winds are forced by the

lower atmosphere’s pressure difference

between the subtropical (Bermuda) high and

subpolar (Iceland) low. When the pressure

systems shift or change in strength, the westerly

wind location and strength also change. The

traditional NAO index is the difference in pressure

between Portugal and Iceland, although

other indices are also used. When the NAO is

positive, the pressure difference is large and

the westerlies are shifted northward relative to

their mean position; that is, with maximum

strength between Portugal and Iceland, and

vice versa. NAO variability is only roughly

decadal and includes seasonal to multidecadal

timescales (Figure S15.2d). A high NAO with

strong westerlies, a cold subpolar gyre, and

warm Nordic Seas and Gulf Stream region

dominated from the 1970s to 1990s. A low

NAO dominated from the 1950s to 1960s.

Shifts in the NAO affect North Atlantic circulation

and the production and properties of its

water masses. Associated with high NAO, the

Gulf Stream and its separation point move

slightly but measurably northward and transport

increases, lagging the NAO by several

years (Curry & McCartney, 2001; Visbeck et al.,

2003). Also during high NAO, the subpolar

gyre circulation north of about 50 N shifts westward

and intensifies (Flatau, Talley, & Niiler,

2003; Häkkinen & Rhines, 2004). Because the

North Atlantic forms intermediate and deep

water, its properties are highly variable from


FIGURE S15.2 Atlantic decadal to multidecadal climate modes. (a) North Atlantic Oscillation (NAO), (b) East Atlantic

Pattern (EAP), and (c) Atlantic Multidecadal Oscillation (AMO). Maps of SST correlation with each index: positive is warm

and negative is cold. (Data and graphical interface from NOAA ESRL, 2009b.) (d) NAO index (Hurrell, 1995, 2009):

difference of sea level pressure between Lisbon, Portugal and Stykkisholmur, Iceland. Source: Updated by Hurrell (personal

communication, 2011). (e) EAP index: amplitude of second EOF. Source: From NOAA ESRL (2009b). (f) AMO: amplitude of the

principal component of proxy temperature records. Source: From Delworth and Mann (2000). (g) Time series, each with a 10-

year running mean and “normalized” by its maximum amplitude. NAO and EAP as above. The AMO is the Enfield et al.

(2001) SST-based index.

8


top to bottom. Labrador Sea Water (LSW),

Greenland Sea Deep Water, and Eighteen

Degree Water (EDW) all vary with the NAO

(Dickson et al., 1996). During positive NAO,

when the subpolar gyre and Labrador Sea are

cold, LSW production is strong and anomalously

cool. The Greenland Sea, on the other

hand, is warmer during high NAO, and Greenland

Sea Deep Water production is weakened

and warmer (Section S15.5.3). EDW production

is also weaker during periods of high NAO,

nearly ceasing in the mid-1970s and shifting to

lower densities in the 1990s (Dickson et al.,

1996; Talley, 1996b).

Decadal variability in the northern North

Atlantic is also associated with the East Atlantic

Pattern (EAP) (Barnston & Livezey, 1986; Josey

& Marsh, 2005; Figure S15.2b and e). Decades

long freshening of the subpolar gyre (described

in the following section) appears to be related to

increased precipitation associated with

increasing EAP. The EAP and NAO are independent.

The EAP is the second empirical

orthogonal function (EOF) of climate variability

for the Atlantic, while the NAO can be defined

as the first EOF. The EAP has a zero crossing

around 35 N that is farther south than that of

the NAO, and a symmetric shape about the

equator. It appears to be the lowest order

symmetric (sinelike) meridional mode for the

Atlantic.

The Atlantic’s longer term variability is of

interest because of its potential relationship to

variability in the MOC. The Atlantic Multidecadal

Oscillation (AMO) or “Atlantic Multidecadal

Variability” is an index of Atlantic SST used to

quantify variability at timescales longer than

decadal. The AMO index is the average SST

anomaly for the entire North Atlantic

(0e70 N), detrended, and with a 10-year

running mean applied (Enfield, Mestas-Nuñez,

& Trimble, 2001). When the index is positive,

the North Atlantic as a whole is warm and the

South Atlantic is cool; this is thus an “interhemispheric

mode” (Figure S15.2c). Monthly values

of the index since 1856 are available through

the NOAA ESRL (2009b) Web site, listed as an

updated “Kaplan SST” product (Kaplan et al.,

1998). The AMO timescale is 65e80 years with

a range of several 0.1 C. There are only two

“cycles” in the SST record. However, longer

paleoclimate proxy records also show an AMO

(Delworth & Mann, 2000; Figure S15.2f). The

AMO can also be reproduced with coupled

ocean-atmosphere models that include meridional

overturning circulation changes.

An interhemispheric mode described as the

“bipolar seesaw” has been introduced to

explain much longer timescale (millennial)

variability in paleoclimate records at the end

of the last glaciation during the Younger Dryas

interval (Broecker, 1998). These records include

signals in the far Northern and far Southern

Hemispheres that are out of phase with each

other. These could be explained by a climate

mode with north-south structure like that of

the AMO. With a strong MOC, there would

be enhanced northward transport of heat into

the subpolar gyre and Nordic Seas, and SST

would then be higher there. There would

also be enhanced northward cross-equatorial

flow of warm water removing heat from the

South Atlantic and moving it to the North

Atlantic.

The NAO also has a multidecadal timescale

(Delworth & Mann, 2000; Visbeck, 2002), as

shown using a 10-year running mean of its

index (Figure S15.2g). The EAP also has decadal

variability (also seen in Figure S15.2g), which

has some resemblance to the AMO index after

about 1970. The EAP has been associated with

the decadal Great Salinity Anomalies described

in the next section.

S15.2.3. Atlantic Ocean Property

Variability

Changes in the Atlantic’s MOC, whether

natural or anthropogenic, both reflect and have

the potential for affecting Earth’s climate


(Vellinga & Wood, 2002). Since the late 1990s,

there have been coordinated programs to

monitor the Nordic Seas overflows, the Labrador

Sea, the Strait of Gibraltar, and meridional

overturn at several latitudes in the North

Atlantic with most resources across 24 N.

Broad-scale observations are providing the

larger context for the changes. SST is monitored

widely; its variations relative to the various

Atlantic climate modes were shown in Figures

S15.1 and S15.2.

We focus here on variability in surface salinity

as it provides a control on mixed layer depth and

density. Over the past century, subpolar North

Atlantic SSS variations have been significant

(Figure S15.3a). After a fresh period centered

around 1910, salinity was relatively high at

60 N until the 1970s. Salinity then declined and

remained low until about 2000, with the

continuing freshening a subject of great interest

because of its potential for slowing the North

Atlantic MOC (Curry, Dickson, & Yashayaev,

2003; Dickson, Curry, & Yashayaev, 2003). After

2000, the salinity trend reversed, with salinity

now increasing throughout the subpolar gyre.

This has joined the upward salinity trend over

the past 50 years in the remainder of the Atlantic

(Figure S15.20 in Section S15.7).

Looking at decadal timescales within the

longer term salinity record, there are clear

pulses of fresher water in the mid-1970s, 1980s,

and 1990s. Along with the 1910 event, these

have been called Great Salinity Anomalies

(GSAs; Dickson, Meincke, Malmberg, & Lee,

1988; Belkin, 2004). The low salinity GSAs

form coherent, time-lagged patterns around

the northern North Atlantic and Nordic Seas.

The 1970s GSA emerged from Fram Strait into

the East Greenland Current in 1968 (Dickson et

al., 1988; Figure S15.3b). The pulse of freshwater

appeared to move down around Greenland,

into the Labrador Sea, out into the North

Atlantic Current, and into the eastern subpolar

gyre, returning to the Nordic Seas about 10

years later. Low salinity anomalies with similar

propagation patterns occurred in the 1980s and

the 1990s, with both events originating from the

Canadian archipelago into the Labrador Sea.

The GSAs in the northern Labrador Sea are

closely related to sea ice extent in Davis Strait,

which is the northern entrance to the Labrador

Sea from the Arctic (Deser, Holland, Reverdin,

& Timlin, 2002).

Because it is unlikely that anomalies of the

magnitude of GSAs could advect all the way

around the cyclonic circulation for up to 10

years with little change in amplitude, it has

been hypothesized that GSAs arise at least

partially in response to adjustment of the circulation

and its fronts (Sundby & Drinkwater,

2007). Whatever the mechanism, when upper

ocean low salinity anomalies arrive, SST

patterns are altered, convection is inhibited,

and there may be feedbacks into the climate

modes (Zhang & Vallis, 2006).

Large-scale salinity changes have been

observed at depth in the northern North

Atlantic (e.g., in the Labrador Sea Water in

Figure S15.4). In the 1960s, during a period of

low NAO, the Labrador Sea was warm and

saline at all depths; LSW formation was weak

and relatively warm and saline. By the 1990s,

LSW properties had shifted to fresher, colder,

and denser (high NAO). Freshening occurred

throughout the northern North Atlantic at intermediate

depths (Dickson et al., 2002). In the later

1990s, LSW production again was interrupted,

and its temperature and salinity began to

increase (declining NAO; Yashayaev, 2007;

Schott, Stramma, Giese, Zantopp, 2009).

Although salinity had shifted to a fresher range

in the1990s, the spatial structure of salinity did

not change: lowest salinity in the Labrador Sea

with tongues of low salinity extending to the

Irminger Sea, Iceland Basin, and Rockall Trough

and a weak tongue extending southward

around Newfoundland. Thus the overall circulation

pattern was mostly preserved.

The freshening of deep waters throughout

the northern North Atlantic and the southern

10


FIGURE S15.3 North Atlantic

surface salinity variability. (a)

Salinity anomalies relative to longterm

mean along 60 N. Source:

From Reverdin et al. (2002). (b) Great

Salinity Anomaly: timing in years

for the 1990s GSA. Source: From

Belkin (2004).

Nordic Seas through the 1990s was accompanied

by freshening of the Nordic Seas overflows

as well (Dickson et al., 2002, 2003). These overflow

property variations result from (1) changes

upstream in the Nordic Seas and (2) the previously

mentioned changes in the entrained

upper ocean and intermediate waters as the

overflows plunge toward the ocean bottom.

On the other hand, the overflows have been

remarkably steady in terms of velocities, transport,

and temperature, mainly because of the

importance of hydraulic control at the straits,

governed by the large upstream reservoir in

the Nordic Seas (Girton, Pratt, Sutherland, &

Price, 2006). In Faroe Bank Channel, transports,

bottom velocities, and bottom temperature held

steady at 2.1 Sv, >100 cm/sec, and -0.4 C for

the directly observed period from 1995 to 2005

with indirect evidence for similar stability based

on observations starting in 1948 (Olsen, Hansen,

Quadfasel, & Østerhus, 2008). In Denmark

Strait, four years of monitoring showed more


move southward into the subtropics and

tropics, mainly through the Deep Western

Boundary Current. Changes in properties

including oxygen have been documented at

the Grand Banks (~43 N), at Abaco (26.5 N),

and all the way to the equator with appropriate

time lags of 2 to 10 years from changes at the

subpolar sources (Molinari et al., 1998; Stramma

et al., 2004; Bryden, Longworth, & Cunningham,

2005b). On the other hand, DWBC transport

variations that can be associated with changes

in the MOC have been difficult to document

from sparse decadal hydrographic sampling

because large seasonal variability is aliased to

longer timescales. However, with the now

continuous monitoring at ~25 N, including the

Florida Current and basin-wide Ekman transport

as well as the DWBC, the prognosis for

monitoring interannual variations in the total

overturn is good to within about 10% of the total

overturn (Cunningham et al., 2007).

FIGURE S15.4 Salinity at the density of LSW in two

different decades: (a) 1960s and (b) 1990s. Source: From

Yashayaev (2007).

variability in overflow velocities, transports,

and temperatures than at Faroe Bank

(Macrander et al., 2005). Transports ranged

from 3.1 to 3.7 Sv and temperatures varied by

0.5 C. Higher transports corresponded roughly

with colder water. Hydraulic control is important,

as in Faroe Bank Channel, but there are

other dynamical processes, such as wind-driven

northward flow in the eastern Denmark Strait,

that can modulate the overflow transport.

The variable properties in the intermediate

and deep waters of the northern North Atlantic

S15.2.4. Climate Change and the

Atlantic Ocean

As introduced in Section S15.1, climate change

is the response of the climate system to anthropogenic

forcing, as distinguished from natural

climate variability. There is great interest in determining

if the Atlantic’s MOC is changing in

response to anthropogenic forcing, because this

would presumably change the amount of ocean

heat transport to high northern latitudes.

However, because of the large amplitude of

natural variability in the northern North Atlantic,

mostly associated with the NAO and possibly

also the EAP, and because of the short length of

observational records, attribution of circulation

and local water mass variations to anthropogenic

forcing has not yet been possible (Bindoff et al.,

2007). Detection of long-term trends indicative

of climate change has only been possible for

properties averaged over very large areas.

Heat content in the Atlantic Ocean has

increased overall during the last several decades

12


(Figures S15.18 and S15.19 in Section S15.7). The

upper ocean warmed except between 50 and

60 N where cooling was due to a positive trend

in the NAO index, which peaked in the early

1990s; positive NAO is associated with cooling

in the Labrador and Irminger Seas (Figure

S15.2a). In the North Atlantic, the deep penetration

of warming in the subtropics was due to

warming Mediterranean Water and reduced

production of LSW. The world ocean as a whole

warmed during those five decades; the Atlantic

contributed the most to the overall trend (Levitus,

Antonov, & Boyer, 2005). Attribution of the

warming in the North and South Atlantic to

anthropogenic change has been made by use

of coupled climate model simulations run with

and without anthropogenic forcing (Barnett

et al., 2005; see Chapter 14).

Salinity trends in the Atlantic during the

same five decades included regions that

increased and decreased in salinity (Figure

S15.20 in Section S15.7). Freshening in the

northern North Atlantic between 45 and 75 N

(Section S15.2.3) began in the mid-1970s. This

has reversed to increasing salinity since the

year 2000. Overall, the Atlantic salinity has

increased (while the Indian has increased and

the Pacific salinity has decreased for a global

balance of no net change). Attribution of the

observed salinity changes to anthropogenic

forcing is more indirect than for temperature

change. However, the changes in salinity are

consistent with anthropogenic change since

a warmer atmosphere can hold more moisture

(Section S15.7).


PACIFIC OCEAN

The Pacific Ocean represents a large fraction

of the global ocean’s surface and therefore

a large potential for coupled atmosphere-ocean

feedbacks. The interannual ENSO (Section

10.8), which has maximum amplitude in the

tropics, is an excellent example of efficient

coupling. The decadal and longer timescale

climate modes are characterized by much larger

north-south spatial patterns with extratropical

amplitudes that are similar to tropical amplitudes.

Outside the tropics, coupling of the ocean

and atmosphere is much weaker, so feedbacks

are much weaker and harder to discern.

Good resources for the many climate modes

are found on the National Oceanic and Atmospheric

Administration’s (NOAA) various

climate Web sites, including the Climate Diagnostics

Center (Climate Analysis Branch;

http://www.cdc.noaa.gov/) and the National

Weather Service’s Climate Prediction Center

(http://www.cpc.ncep.noaa.gov/).

Pacific “decadal” climate variability has

a timescale of 15 to 20 years, which is longer

than the dominant Atlantic decadal timescales.

The difference in timescale may reflect the size

of the ocean basins and hence timescale for

planetary wave propagation. Decadal Pacific

modes include the Pacific Decadal Oscillation

(PDO) and the North Pacific Gyre Oscillation

(NPGO; Mantua et al., 1997; Di Lorenzo et al.,

2008). These are the first and second EOFs of

SST in the North Pacific (Figures 6.11 and 6.12

from Davis, 1976; Cayan, 1992). Two related

North Pacific indices based on atmospheric

pressure are the Pacific North American teleconnection

pattern (PNA) and the North Pacific

Index (NPI; Trenberth & Hurrell, 1994). The

Southern Annular Mode (SAM) is a circumpolar

mode with major impacts on the South Pacific

(Thompson & Wallace, 2000).

The PDO spans the whole Pacific, although it

is most robust in the North Pacific (Figure

S15.5a). Its pattern is nearly symmetric about

the equator, with high amplitude centered

broadly on the equator and out-of-phase amplitude

centered in the subtropical/subpolar

North and South Pacific. The strength of the

Aleutian Low is associated with the PDO.

When the Aleutian Low is strong (high PDO),

the westerly winds are strong and displaced

CLIMATE AND THE PACIFIC OCEAN 13

FIGURE S15.5 Pacific Decadal Oscillation (PDO) and North Pacific Index (NPI). (a) SST correlation with the PDO index.

(b) Annual mean PDO index (red/blue) and with a 10-year running mean (black). (Updated from Mantua et al., 1997 and

Trenberth et al., 2007). (c) NPI SST pattern. (Data and graphical interface for a, b, and c from NOAA ESRL, 2009b). (d) NPI

index. ÓAmerican Meteorological Society. Reprinted with permission. Source: From Deser, Phillips, & Hurrell (2004).

somewhat to the south; the ocean’s subpolar

gyre is strong and less subpolar water enters

the California Current system. This means that

the entire eastern boundary region, for both

the subpolar and subtropical gyres, is warmer

than normal, while the central Pacific, beneath

the strengthened westerlies, is abnormally

cold. The Oyashio is strong and penetrates

farther southward along the coast of Japan.

A well-documented shift from low to high

PDO occurred around 1976. At this point,

a lengthy period of a particularly strong Aleutian

Low began. The changes in ocean temperatures

and circulation resulted in marked shifts in

almost every environmental variable measured

in the North Pacific d fish, birds, crabs, salinity,

nutrients, and so forth (Mantua et al., 1997).

The NPGO has a tighter spatial pattern than

the PDO since it is a higher mode EOF (Figure

S15.6). The NPGO is much better correlated

than the PDO with environmental variables

such as upwelling and ecosystem production

along some large portions of the North Pacific

coastline.

The NPI is the mean sea level pressure over

the region 30e65 N, 160 We140 W and as

14


FIGURE S15.6 (a) Pacific Decadal Oscillation (PDO)

and (b) North Pacific Gyre Oscillation (NPGO) patterns of

sea level pressure (color) and surface wind stress (vectors).

The PDO/NPGO are correlated well with upwelling in the

red-circled/blue-circled region off Oregon/California.

Source: From Di Lorenzo et al. (2008).

such is a direct measure of the strength of the

Aleutian Low (Trenberth & Hurrell, 1994). The

PNA is an older index of atmospheric geopotential

height, summed from four locations,

including two over North America. Their SST

patterns are virtually identical. The NPI and

PDO patterns and time series are very similar

(Figure S15.5). The PDO could be considered

a combination of ENSO and the NPI, that is,

a combination of tropical forcing and Aleutian

Low forcing (Schneider & Cornuelle, 2005).

The Southern Annular Mode (SAM), also

known as the Antarctic Oscillation (AAO),

dominates decadal variability at high southern

latitudes (Section S15.6). One of the centers of

maximum amplitude of the SAM pattern is in

the western South Pacific, centered at New Zealand

(Figure S15.15 in Section S15.6). Variability

in circulation in the South Pacific subtropical

gyre has been linked to the SAM (Roemmich

et al., 2007).

Climate change in response to anthropogenic

forcing has been documented in the Pacific as

well as in the other main oceans (Section

S15.7). The upper 500 m of the Pacific has

warmed over the past 50 years (Figure S15.19)

as part of the general warming of the global

ocean (Levitus et al., 2005). Basin-averaged

salinity has decreased slightly but measurably

(Boyer, Antonov, Levitus, & Locarnini, 2005).

Fresh intermediate water masses such as North

Pacific Intermediate Water (NPIW) and Antarctic

Intermediate Water (AAIW) (Section 10.9.2) have

freshened (Wong, Bindoff, & Church, 2001).

Oxygen content has been decreasing in most

parts of the upper Pacific Ocean over the past 50

years. The tropical oxygen minimum zones

have expanded (Stramma, Johnson, Sprintall,

& Mohrholz, 2008) and oxygen has declined

throughout the upper ocean in the North Pacific

and in the Antarctic Circumpolar Current

(ACC) region (Deutsch, Emerson, & Thompson,

2005; Aoki, Bindoff, & Church, 2005). The Pacific

Ocean has become more acidic; it appears that

there is no possibility of reversing the trend

given the relentless increase in atmospheric

CO 2 content. Stresses on ecosystems such as

coral reefs and continental shelves resulting

from increased temperatures and acidity are

beginning to be observed.


INDIAN OCEAN

Climate variability at interannual to decadal

timescales has been documented in the Indian

Ocean. Because of its importance to agriculture,

interannual and longer term variability in the

monsoon has been of special interest. In fact,

CLIMATE AND THE INDIAN OCEAN 15

FIGURE S15.7 Correlation of

SST anomalies with the ENSO

index for 1982e1992, at (a) 0 month

lag and (b) 4 month lag.


Reprinted with permission. Source:

From Klein, Soden, and Lau (1999).

while working in the early twentieth century on

understanding the sources of monsoon variability

(including an especially devastating

monsoon failure in 1899), Sir Gilbert Walker

detected and documented the Southern Oscillation,

which is the interannual variability in the

zonal atmospheric pressure gradient between

the central tropical Pacific and the western

Pacific/eastern Indian Ocean (Section 10.8).

His was the first major step toward documenting

and understanding the interannual ENSO

of the Pacific Ocean.

Although the airesea coupling process that

creates ENSO is centered in the tropical Pacific,

ENSO dominates interannual climate variability

in the Indian Ocean (Tourre & White, 1995,

1997). During an El Niño event in the Pacific,

SSTs in the tropical Indian Ocean rise 3e6

months later (Figure S15.7). El Niño also affects

precipitation in the Indian Ocean region,

including dry conditions in India in the

Northern Hemisphere summer and in eastern

Africa in austral summer (see global ENSO

precipitation anomaly maps in online supplementary

Figure S10.43).

At the onset of an El Niño event in the Pacific,

anomalously easterly winds in the Indian Ocean

cause upwelling in its eastern region and

depress the thermocline in the western region,

initially resulting in cooling in the east and

warming in the west as observed. As El Niño

progresses, changes in surface heat flux cause

the entire tropical Indian Ocean to warm (Klein,

Soden, & Lau, 1999). Feedbacks between the

ocean and atmosphere within the Indian Ocean

then affect the local response to El Niño(Zhong,

Hendon, & Alves, 2005). The development of

the Indian Ocean response to El Niño depends

on the phase of a given El Niño event relative

to the monsoon, because the seasonal monsoon

affects SST in the western tropical Indian Ocean,

which then affects the local airesea feedbacks

(Krishnamurthy & Kirtman, 2003). The Indian

Ocean is “upstream” of the Pacific Ocean in

terms of an atmospheric signal called the

Madden-Julian Oscillation (MJO). The MJO

has a period of 30 to 60 days and affects all

atmospheric variables including winds, clouds,

rainfall, and airesea fluxes (Madden & Julian,

1994; NOAA CPC, 2005). MJO events begin in

the Indian Ocean and propagate eastward.

They often provide the westerly wind bursts

that affect the onset of El Niño in the western

Pacific (Section 10.8).

16


RELATIVE PRODUCTION

RAINFALL

160

140

120

100

80

60

40

1400

1200

1000

800

Relationship of Indian Rice Production

and Indian Rainfall

(a) Indian Rice Production

(% of 1978)

El Niño

La Niña

(b) All-India Rainfall (mm)

1960 1970 1980 1990 2000

Year

FIGURE S15.8 Indian Ocean. Changes in (a) rice

production and (b) rainfall in India with El Niño and La

Niña events indicated. The long-term trend in production is

due to improved agricultural practices. (Adapted by WCRP,

1998 from Webster et al., 1998.)

The Indian monsoon is affected by ENSO. The

Southwest Monsoon is weak during El Niño

events, leading to the “dry” years with reduced

agricultural production in India (Figure S15.8;

Webster et al., 1998).

Beyond its response to ENSO, the Indian

Ocean has internal interannual variability. A

tropical IOD mode has been described whose

positive phase is characterized by warm SST

anomalies in the western tropical Indian Ocean

and cool anomalies in the eastern tropical region

(Figure S15.9a). These SST anomalies are accompanied

by zonal wind anomalies that blow from

the cool region to the warm and higher amounts

of rainfall over the warm region. The simplest

index is the east-west difference in tropical SST

(Figure S15.9b; Saji, Goswami, Vinayachandran,

& Yamagata, 1999; Webster, Moore, Loschnigg,

& Leben, 1999). Studies of the complete

Indian-Pacific region suggest that the mode

might not be entirely independent of ENSO,

and that this internal Indian Ocean mode can

be excited by ENSO (Krishnamurthy & Kirtman,

2003; Zhong et al., 2005). High correlation

between ENSO and the dipole mode occurred

from 1960 to 1983 and after 1993, but not in

the intervening period (Clark, Webster, & Cole,

2003). It is also possible that the relationship

between ENSO and the dipole mode has

changed over time and that the dipole mode

could be excited by other climate variability

(Annamalai, Xie, McCreary, & Murtugudde,

2005; Ihara, Kushnir, & Cane, 2008).

Ocean circulation in the northern Indian

Ocean, including the Arabian Sea and Bay of

Bengal, is dominated by the seasonal monsoon

wind forcing and so it should respond to interannual

variability in the monsoon strength.

The monsoon response is especially strong in

the upper 200 m, with almost no effect in the

abyss, so that strong response might be expected

in the upper ocean (Dengler, Quadfasel, Schott,

& Fischer, 2002). However, interannual variability

observed in Arabian Sea circulation is

more complicated than a simple direct response

to changing monsoons. Planetary (Rossby)

wave propagation across the Arabian Sea is

important for adjustment of the circulation to

changes in winds and affects the time phasing

of the circulation response (Schott & McCreary,

2001).

Comprehensive, long-term in situ observations

that could be used to describe decadal

and longer term climate variability within the

Indian Ocean’s water column are sparse and

descriptions are therefore lacking. Relative to

climate change, trends and changes observed

in Indian Ocean water properties and circulation

were summarized in the IPCC Fourth

Assessment Report (Bindoff et al., 2007). The

Indian Ocean’s upper layer warmed over the

past 50 years except at the base of the mixed

layer at the equator and southward through

CLIMATE AND THE ARCTIC OCEAN 17

FIGURE S15.9 Indian Ocean Dipole mode. (a) Anomalies of SST (shading) and wind velocity (arrows) during

a composite positive IOD event. These are accompanied by higher precipitation in the warm SST region and lower

precipitation in the cool SST region. (b) IOD index (blue: difference in SST anomaly between the western and eastern tropical

Indian Ocean), plotted with the anomaly of zonal equatorial wind (red) and the Nino3 index from the Pacific (black line).

Source: From Saji et al., (1999).

the SEC, similar to the warming pattern in the

Pacific (Figures S15.18 and S15.19 in Section

S15.7; Levitus et al., 2005). Salinity in the upper

ocean also increased overall, similar to salinity

increase in the Atlantic but opposite the slight

freshening of the Pacific (Boyer et al., 2005).

Circulation in the south Indian Ocean’s

subtropical gyre likely increased by about 20%

between the late 1980s and early 2000s based

on tracers indicating ventilation (McDonagh et

al., 2005). There was a slowdown of similar

strength between the 1960s and 1980s (Bindoff

& McDougall, 2000). The increased circulation

through the 1990s was associated with strengthening

westerlies, as measured by the SAM index

(Section S15.6). It is not yet clear whether these

shifts in circulation and forcing are due to

climate change or are a natural climate

fluctuation.


ARCTIC OCEAN

S15.5.1. Arctic Oscillation, Atlantic

Multidecadal Oscillation, and Global

Change

Three modes of climate variability/change

are frequently used for describing Arctic

18


variability: the Arctic Oscillation (AO; also called

the Northern Annular Mode), the Atlantic Multidecadal

Oscillation (AMO), and global change

driven by anthropogenic forcing. The AO and

AMO are natural modes. The AO has decadal

to centennial variability, and is closely related

to the North Atlantic Oscillation (Section

S15.2). AO variability affects the winds, so variations

in the ocean circulation and ice drift in the

Arctic and northern North Atlantic are closely

tied to the AO. The AO does not appear to be

a coupled ocean-ice-atmosphere climate mode,

but rather a mode of the atmosphere. The

AMO might be a natural mode of multidecadal

variability of the Atlantic MOC (Section S15.2).

Global change effects are detected as long-term

trends with attribution based on the distinctive

signature of changes in winds, atmospheric

pressure, temperature, and so forth.

The AO is a variation in the atmospheric

pressure and wind pattern in middle and high

northern latitudes (Thompson & Wallace,

1998). The prevailing wind pattern over the

Arctic is the westerly “polar vortex” with low

atmospheric pressure at the pole and higher

pressure at mid-latitudes (see schematic in

NSIDC, 2009b). This can be described in terms

of the dominant EOF of the sea level pressure

pattern north of 20 N, using only winter months

(JFMA) for the EOF (Thompson & Wallace,

1998). The AO index (Figure S15.10c) is the

amplitude of this dominant EOF.

When the AO is in its positive phase (illustrated

by the correlation pattern in Figure

S15.10b), polar pressure is lower than usual,

the difference in pressure between the highs at

mid-latitudes and the polar low is larger, and

the polar vortex is stronger and is shifted

toward the north. There is wetter, warmer

weather in the subpolar regions (and drier

conditions in mid-latitudes) since the mid-latitude

high pressure extends farther north, and

the storm track is shifted farther north. Temperatures

over Labrador, Greenland, and the

western subpolar North Atlantic drop. When

the AO is in its negative phase, the difference

in pressure is reduced, the polar low-pressure

region is larger, the polar vortex is weaker, and

the storm tracks shift to the south; this results

in colder, drier weather at the higher latitudes.

During positive AO, the Transpolar Drift

(TPD) flows nearly directly across the Arctic

from Bering Strait to the northern side of Greenland

and the Beaufort Gyre is restricted to the

side of the Canadian Basin. During negative

AO, the Beaufort Gyre expands and strengthens

and the TPD shifts toward the Lomonosov

Ridge (Rigor, Wallace, & Colony, 2002; Figure

12.14a from Steele et al., 2004).

Over the past century, the AO was relatively

high into the 1930s, then alternated or shifted

to low in the 1960s and 1970s. Since a sudden

peak in 1989, it has been generally high, with

much interannual and decadal variability about

these longer term signals (Figure S15.10c).

The AMO is a natural climate mode of the

Atlantic Ocean (Section S15.2.2) and also affects

the Nordic Seas and Arctic. It is entirely independent

of the AO and NAO. During periods of high

AMO, northern North Atlantic SSTs are high and

the MOC is strong, advecting warm waters

farther northward. During low AMO, the overturning

circulation is weak and the northern

North Atlantic and Nordic Seas cool off. The

AMO index time series (Figure S15.2) is remarkably

similar to multidecadal fluctuations in the

Atlantic Water temperature within the Arctic

Ocean (Figure S15.13), suggesting a link between

the meridional overturning strength and upper

ocean temperatures far into the Nordic Seas/

Arctic (Polyakov et al., 2005).

Long-term variability in the Arctic and

Nordic Seas can involve climate system feedbacks.

The simplest and possibly most important

is the “ice-albedo feedback” (Figure 5.10).

Changes in sea ice also affect atmospheric

winds, which sense the large temperature difference

between the ice and open water. A feedback

involving sea ice and wind in the

Beaufort Gyre is centered on the difference in


FIGURE S15.10 Arctic Oscillation

(AO). (a) Schematics of (left)

the positive phase and (right) the

negative phase (NSIDC, 2009b). (b)

Correlation of surface pressure (20

e90 N) with the AO index for 1958

to 2007. (Data and graphical interface

from NOAA ESRL, 2009b.) (c)

Arctic Oscillation index 1899e2002.

Source: From JISAO (2004).

surface currents when driven directly by wind

or by wind acting on the ice (Shimada et al.,

2006).

Global change resulting from anthropogenic

forcing, mainly greenhouse warming, has

a strong signature in the Arctic, which has

warmed nearly twice as much as the global

average over the past 100 years (IPCC, 2007).

Greatest surface warming as a result of global

change is predicted for the highest northern

20


latitudes; this is referred to as “polar amplification,”

although it is a feature of just the Arctic

and not the Antarctic. Greater warming occurs

at the Arctic’s surface than at lower latitudes

for a number of reasons: the ice-albedo feedback,

the increased amount of water vapor in

the atmosphere that changes the polar radiation

balance in the atmosphere, and the thinness of

the polar troposphere over which the heat is

distributed.

S15.5.2. Variations in Arctic Sea Ice

Arctic sea ice extent and volume have been

decreasing and the ice has become younger

and thinner since the late 1970s (Rothrock, Yu,

& Maykut, 1999; Fowler, Emery, & Maslanik,

2004). Multi-year ice has been declining in areal

extent (Figures 12.22 and S15.11), even taking

into account considerable interannual variation.

Each year has brought continued decrease in ice

cover and additional diagnoses of the causes

(Serreze, Holland, & Stroeve, 2007). Causes of

the sea ice reduction have been linked to both

the positive phase of the naturally occurring

AMO and to global warming. Indeed, global

warming might contribute to a positive AO,

and could likely impact the AMO (IPCC, 2007).

From the beginning of the satellite observations

of ice cover in 1978 to the present, Arctic

ice cover has decreased relentlessly, although

the record includes the usual short-term fluctuations/noise

of climate records (Figure S15.11). 2

(This highlights the importance of observing

continuously for many years before it is possible

to discern a trend.) Ice cover in 2007 relative to

the mean extent for previous decades illustrates

the dramatic trend in ice cover (Figure S15.11).

Once the ice decline had progressed long

enough to be noticed even in the noisy climate

record, it became apparent that much of the

ice loss is in the perennial ice pack (multi-year

ice), which means that thinner, seasonal ice has

increased in relative importance (Johannessen,

Shalina, & Miles, 1999). Largest changes have

taken place in the Beaufort and Chukchi Seas,

with more than 25% reduction (Maslanik, Serreze,

& Agnew, 1999; Shimada et al., 2006). Since

first-year, thin ice melts away more readily in

summer, reducing the overall albedo of the

Arctic, the stage continues to be set for further

decline of ice cover through ice-albedo feedback

(Figure 5.10). (Note though that there are

multiple reasons for the decline in addition to

this feedback.)

Large interannual variations in ice cover that

persist for 5e7 years are associated with

changes in the wind-forced circulation (Proshutinsky

&Johnson, 1997), possibly associated

with variations in the AO. In the mid-1990s,

during positive AO (Figure S15.10c), the cold

halocline weakened, which increased ocean

heat flux and a decrease in sea ice (Martinson

& Steele, 2001). The AO returned closer to

neutral in the mid-1990s, but sea ice continued

to decrease, with an even greater loss in 2007,

repeated in 2008, than expected from the trend

(Figure S15.11c). The Pacific sector north of

Bering Strait has shown the greatest decline

(Figure S15.11a,b and Shimada et al., 2006).

The AMO shifted to a high phase in the late

1990s and the incursion of warm Atlantic Water

into the Arctic continued, separate from the AO

variability; this likely has contributed to the

decreasing sea ice.

There are many ways to reduce sea ice during

the positive phase of the AO. Changes in Arctic

winds (direction) can result in warmer air that

melts ice. Changes in circulation during a positive

AO include greater penetration of warm

Atlantic Water into the Arctic, which also melts

ice. Changing winds in positive AO increase

movement of sea ice away from the coasts of

the Eurasian Basin, resulting in younger, thinner

2 The maps and time series in Figure S15.11 show September ice cover changes because September is at the end of summer.

The September sea ice record represents the amount of ice that can remain as multi-year ice during the succeeding year.


FIGURE S15.11 Arctic Ocean. (a) Sea ice extent 9/25/2007. Pink indicates the average extent for years 1979e2000. Source:

From NSIDC (2007). (b) Sea ice concentration anomaly (%) for September 1998e2003 minus 1979e1997. Source: From Shimada

et al. (2006). (c) Arctic sea ice extent in September (1978e2008), based on satellite microwave data. Source: From NSIDC

(2008b); after Serreze et al. (2007).

22


ice there that is more likely to melt in summer.

Sea ice transport in the Transpolar Drift

increases with greater export through Fram

Strait, even as ice moves more slowly from the

Beaufort Sea across the North Pole; the Beaufort

Gyre contracts and weakens (Rigor et al., 2002).

Other effects that have reduced sea ice

include greater incursion of Pacific Summer

Water through Bering Strait into the Beaufort

Gyre from 1998 to 2003; a positive feedback in

the ocean-ice-winds system has been proposed

(Shimada et al., 2006). Sea ice reduction leads

to a positive feedback, as the greater area of

open water absorbs more heat from the sun

(ice-albedo feedback).

The loss of Arctic sea ice is now faster than

predicted by the ensemble of climate models

in the fourth assessment report of the IPCC

(2007), based on greenhouse gas forcing. It is

now expected that the Arctic will be ice-free in

summer within the next several decades, which

is much earlier than the prediction of the end of

the twenty-first century given in the IPCC

report. The conclusion is that the polar-amplification

of global warming is indeed operable and

that the trend of sea ice loss is due to global

change (Serreze et al., 2007).

S15.5.3. Variations in Nordic Seas and

Arctic Water Properties

The Nordic Seas have been warming at all

depths since the 1980s. This has affected and

also resulted from changes in convection. Deep

convection occurs in the Greenland Sea, renewing

intermediate through bottom waters

throughout the Nordic Seas. However, the

depth and properties of convection have varied

greatly. Convection reaching to the bottom is

likely to have occurred in the early 1980s, but

was rare enough that Carmack and Aagaard

(1973) and Clarke, Swift, Reid, & Koltermann

(1990) hypothesized other mechanisms for

bottom water renewal (Section 12.2). Since the

early 1980s, top-to-bottom convection has been

replaced by annual intermediate depth convection

(down to 1000e2000 m; Ronski & Budéus,

2005b; Hughes, Holliday, & Beszczynska-Möller,

2008). Because of the cessation of very

deep convection, the vertical structure in the

Greenland Sea has been replaced by a two-layer

structure with active convection in the upper

layer and older water beneath. Reduced convection

in the Greenland Sea reduces the amount of

carbon dioxide that can be pulled down into

deep water and could boost global warming.

Regardless of whether it reaches to intermediate

depths or the bottom, renewal of Greenland Sea

Deep Water is important for North Atlantic

Deep Water formation, since this convection is

still deeper than the Greenland-Scotland ridge

sill depths and the renewed water can spill

southward into the deep North Atlantic.

Decades long freshening throughout the

Nordic Seas until the late 1990s mirrored freshening

of the subpolar North Atlantic at the

same time (Chapter 9). A long time series just

west of the Norwegian Atlantic Current, at

Ocean Weather Station Mike (66 N, 2 E), shows

the freshening trend in the upper 1500 m from

the mid-1970s until 1998 (Figure S15.12).

However, the whole region, including the

Nordic Seas and subpolar North Atlantic,

started to become saltier in the late 1990s after

the end of the record shown in Figure S15.12

(Hughes et al., 2008).

Within the Arctic, north of Svalbard, there

has been remarkable warming in the 1990s up

to the present (Figures S15.13 and S15.14). The

strongest signal is warming of the Atlantic

Water temperature maximum. A new pulse of

warm Atlantic Water entered the Arctic in

2004, captured by the annually repeated hydrographic

section in Figure S15.14 (Polyakov,

personal communication, 2009), but past the

end of the time series depicted in Figure

S15.13. After 2004, the core temperature of the

Atlantic Water layer increased by almost 2 C

in the Laptev Sea. Although this warm Atlantic

layer is capped by a much colder surface layer,

CLIMATE AND THE SOUTHERN OCEAN 23

FIGURE S15.12 Salinity in the Norwegian Sea, at Ocean

Weather Station Mike offshore of the Norwegian Atlantic

Current (66 N, 2 E). Source: From Dickson, Curry, and

Yashayaev (2003, Recent changes in the North Atlantic, Phil.

Trans. Roy. Soc. A, 361, p. 1922, Fig. 2).

supported by the Arctic halocline, the warm

layer is expanding upward and can represent

an additional source of heat that reduces sea

ice cover above it.

There are few time series that can be used to

study changes over decades in the Arctic. The

data sets, which are primarily Russian, are

geographically sparse during some decades,

and unevenly distributed in time (Swift,

Aagaard, Timokhov, & Nikiforov, 2005). Likely

because of the sparse data, there are two

competing views of the evolution of temperature

within the Atlantic Water layer, whose

temperature has become something of an index

of Arctic Ocean climate variability. The first

view (Polyakov et al., 2005), represented in

Figure S15.13a, which shows the average of

temperature anomalies in all regions, suggests

a multidecadal timescale, which matches the

timescale of the AMO index although they are

not in phase with each other. The current warming

of the layer is comparable to the warming in

the 1920se1940s, with cool periods before 1920

and in the 1960s and 1970s.

The second view (Figure S15.13b from Swift

et al., 2005), in which the geographic distribution

of the temperature changes is retained,

does not find a cool period throughout the

1960s and 1970s. Rather there was a warming

in the mid to late 1960s in all regions except

the Eurasian continental shelf. There was

a cyclonically propagating Atlantic layer warm

event in the second half of the 1950s similar to

that of the 1990s. Rapid onset of Arctic-wide

warmth during the 1960s was apparently followed

by an equally widespread, but even

longer cold period that lasted until the late

1980s. The 1950s event is obscured in Figure

S15.13 because of basin-wide averaging; the

rapidity of the onset of the 1960s warming is

also not as apparent in Figure S15.13 as in the

more comprehensive data set.

At this point it is unclear whether the present

warming of the Atlantic Water layer is natural or

anthropogenic, but this warming can only reinforce

the present increased loss of Arctic sea ice,

which does have its roots in global change.


SOUTHERN OCEAN

Climate variability in the Southern Ocean is

still being characterized because of the shortness

of good time series. It appears to be dominated

by (1) a circumpolar pattern (SAM; (Figure

S15.15) and (2) higher mode patterns with large

amplitude in the central Pacific sector of the

Antarctic. Both have interannual and longer

timescales, and are manifested in SST, circulation,

and sea ice extent. ENSO (Chapter 10) has

an impact on these modes, especially at interannual

timescales. Longer timescales appear to be

tied in part to anthropogenic change.

ENSO variability in the tropical Pacific is connected

to the Southern Ocean through the atmosphere.

The resulting Antarctic response to

ENSO has a dipole character with the two

centers in the central Pacific (Ross Sea) and the

24


(b)

(c)

FIGURE S15.13 Atlantic Water core temperature in the Arctic Ocean: (a) Mean anomaly ( C), averaged from anomalies

in ten regions relative to the mean over the record. Source: From Polyakov et al. (2005). (b) Variability for all geographic boxes

shown in (c), leaving blank those with too few observations. Source: From Swift et al. (2005).

CLIMATE AND THE SOUTHERN OCEAN 25

FIGURE S15.14 Arctic Ocean.

Sections of potential temperature

( C) for 2003e2008, and a time

series of temperature northeast of

Svalbard. (I. Polyakov, personal

communication, 2009.)

western Atlantic (Weddell Sea) out of phase

with each other. ENSO warm events are associated

with warm SST/reduced sea ice in the

Pacific center and the opposite in the Atlantic

center. La Niña cold events have the opposite

response. This Antarctic Dipole has its own

internal air-sea-ice feedbacks and persists for

several years after being triggered by ENSO

(Yuan, 2004).

The Southern Annular Mode, also known as

the Antarctic Oscillation , is the dominant decadal

climate mode of the Southern Hemisphere

(Thompson & Wallace, 2000). It subsumes various

Southern Hemisphere climate modes that

were described in the 1990s. The SAM index is

the amplitude of an EOF (Section 6.6.1) of the

atmosphere’s geopotential anomaly (similar to

dynamic height; (Figure S15.15). The SAM

pattern has a center of pressure of one sign

over Antarctica and the opposite sign in a ring

(annulus) at 40 to 50 S. Regions of greatest

amplitude occur in the Ross Sea region and are

of opposite sign in the southwest Pacific and

central Indian Ocean. When the SAM index is

positive and high, the south-north pressure

difference is large (higher pressure in the red

areas and lower pressure in the blue area in the

figure). This means that the westerly winds are

more intense. There is also a southward shift in

the maximum westerlies. Its SST pattern is

more complex and is related to the changes in

both zonal and meridional winds. Stronger westerlies

cause stronger northward Ekman transport

in the (ACC), and changes in patterns of

upwelling around Antarctica (Hall & Visbeck,

2002).

The SAM index had been rising noisily for

over 40 years, reaching a maximum in the early

2000s, after which it began to decline (Figure

S15.15). The positive SAM trend has been

26


FIGURE S15.15 Southern Ocean. Correlation of the Southern Annular Mode index (from Thompson and Wallace, 2000), for

all months from 1979 to 2005, with (a, b) sea level pressure and (c) SST. (Data and graphical interface from NOAA ESRL,

2009b.) (d) Time series of SAM index. Source: From the IPCC AR4, Trenberth et al., 2007; Climate Change 2007: The Physical

Science Basis. Working Group I Contribution to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change,

Figure 3.32. Cambridge University Press.)

GLOBAL OCEAN CLIMATE CHANGE 27

which could increase the poleward heat transport

(Meredith & Hogg, 2006).

S15.7. GLOBAL OCEAN

CLIMATE CHANGE

FIGURE S15.15

related to anthropogenic change (Thompson &

Solomon, 2002; Marshall, 2003). The rise in

SAM has been related to a southward shift and

strengthening of the ACC and to an increase in

subtropical circulation in the western South

Pacific (Roemmich et al., 2007). The southward

shift resulted in incursion of warmer waters at

depth on the north side of the ACC. Such warming

has been observed over the past 70 years at

900 m depth (Figure S15.16 from Gille, 2002;

Fyfe, 2006). The change in the ACC might also

have resulted in an increase in eddy activity,

60 W

90 W

120 W

30 W

150 W

(Continued).

0

180

30 E

150 E

60 E

120 E

90 E

o C/yr

0.40

0.03

0.02

0.01

0.00

-0.01

-0.02

-0.03

-0.40

FIGURE S15.16 Temperature change at 900 m in the

Southern Ocean from the 1930s to 2000, including shipboard

profile and ALACE profiling float data. The largest warming

occurs in the Subantarctic Zone, and a slight cooling to

the north. Source: From Gille (2002).

Observing anthropogenic climate change

using in situ observations is difficult, as its

imprint is mostly sought in terms of long-term

trends in a particular observed field. For the

global ocean, long time series are not available

in many regions, especially away from the

coasts, islands, and heavily trafficked shipping

lanes. Our best long time series are from tide

stations at scattered sites around the globe.

Our best time series with good spatial coverage

are from satellites, but this coverage only began

in earnest in the 1980s. Nevertheless, reconstructions

of global ocean heat content, surface

temperature, sea level, and near-surface salinity

are of high enough quality, with quantifiable

error estimates, to begin to discern unambiguous

trends over many decades. When interpreted

along with similar time series of

atmospheric observations, and in the context

of simple to complex climate models, the ocean

observations provide support for concluding

that there is discernible evidence for global

change in response mainly to greenhouse gas

forcing (IPCC, 2007).

Heat content of the upper ocean (0e700 m)

has been increasing since the 1950s, with

a possible decline through the 1960s (Figure

S15.17). The heat content increase from 1961 to

2003 was about 16 10 22 J. This was associated

with an average temperature increase of about

0.1 C in the upper 700 m (Levitus et al., 2005).

The SST changed about 0.4 C over that time.

The heat content change is equivalent to

a change in surface heat flux of about 0.4 W/

m 2 (Domingues et al., 2008), which is far smaller

than the error in airesea heat flux observations,

of order 10 W/m 2 . Even though the task of

mapping and analyzing ocean temperature is

28


FIGURE S15.17 Global ocean

heat content change (10 22 J) for

the upper 0e700 m (black), 0e100

m (red), and SST change (blue).

One standard deviation of error is

indicated in gray (for 0e700 m)

and thin red lines (for 0e100 m).

The optical thickness of the

stratosphere is indicated at the

bottom, with three major volcanoes

labeled. Source: From Domingues

et al. (2008).

formidable, it is a far more robust indicator of

climate change than could ever be derived in

the foreseeable future from direct observations

of airesea heat flux.

The heat content of the entire global system

has increased since the 1950s. The oceans have

absorbed 90% of this heat increase because there

is much more heat storage capacity in water than

in the atmosphere, sea and land ice, or the continents.

The much greater specific heat of water

compared with gas means that the 0.1 C change

in upper ocean temperature would be equivalent

to an almost 100 C change in atmospheric

temperature (Levitus et al., 2005).

The spatial distribution of the ocean’s heat

content and SST trend over the past 50 years is

not uniform (Figures S15.18 and S14.12).

Climate change models predict non-uniform

changes when projected over the next century,

with greatest warming in the Arctic and little

change in the subpolar North Atlantic and

ACC (IPCC, 2007). Observed warming is widespread

and is indeed more exaggerated in the

Arctic where sea ice cover has been retreating

significantly (Chapter 12). Heat content appears

to have decreased in the subpolar North Pacific

and North Atlantic and in the Pacific’s tropical

warm pool. SST trends (Figure S15.18) differ

somewhat from water column heat content,

showing warming in the warm pool, and also

a band of cooling along the ACC in the region

where airesea fluxes warm the ocean in the

annual mean (Figure 5.15 and S5.9). This band

is associated with wind-driven upwelling,

which can decrease SST. An increased AAO

(SAM) index for the 1950s through 2000

(Marshall, 2003) lowered sea level pressure

and strengthened westerly winds and

upwelling, which could have decreased the

surface temperature.

Warming has been mostly confined to the

upper ocean (Figure S15.19; Levitus et al.,

2005). The tropical cooling in the world average

derives from the tropical Pacific and is due to an

ENSO signal (Section 10.8). The subpolar

Northern Hemisphere cooling is from the

Atlantic (Section S15.2.4). Details for each of

the ocean basins were described in the basin

chapters (9e13) and are not repeated here.

Abyssal and bottom temperatures have also

been increasing worldwide based on highly

accurate observations from research ships

(Kawano et al., 2006, 2010; Purkey & Johnson,

2010). The changes are small, but within the

uncertainty of the observations. The largest

changes are found near the obvious sources of

deep water. Changes that are far downstream

from these sources, such as in the deep North

Pacific where bottom water is hundreds of years

old, can result from adjustment of the deep

circulation such that the whole complex of

waters shifts northward without having to

advect warmer water all the way from the

distant source (Nakano & Suginohara, 2002).


FIGURE S15.18 (a) Correlation

of SST for 1970e2007 with a linear

trend, based on the NCEP/NCAR

reanalysis. Positive correlation

means warming and negative

correlation means cooling. (Data

and graphical interface from

NOAA ESRL, 2009b.) (b) Linear

trend of change in ocean heat

content per unit surface area (W

m e2 ) for the 0 to 700 m layer from

1955 to 2003, based on Levitus et al.

(2005). Red shading is values above

0.25 W m e2 and blue shading is

below e0.25 W m e2 . Source: From

the IPCC AR4, Bindoff et al., 2007;

Climate Change 2007: The Physical

Science Basis. Working Group I

Contribution to the Fourth Assessment

Report of the Intergovernmental Panel

on Climate Change, Figure 5.2. Cambridge

University Press.

Large-scale but weak trends in upper ocean

salinity from 1955 to 1998 have been demonstrated

(Figure S15.20 from Boyer et al., 2005,

and more recent results from Durack & Wijffels,

2010). These represent a redistribution of freshwater,

rather than a net change in freshwater

content. The average ocean salinity should

change (decrease) in response to net melting of

land ice, which is an expected result of global

warming. A quick calculation of the impact of

such melt shows that detection is feasible if the

meltoff is large enough. However, at this time,

such a change in average ocean salinity has

not been observed. There has been a weak

increase in Atlantic and Indian Ocean salinity

and a decrease in Pacific Ocean salinity. This is

suggestive of an increase in the cycle of precipitation

and evaporation. This could result from

warming of the atmosphere, which increases

its capacity to hold water and hence cycle it in

greater amounts from evaporation regions to

precipitation regions (Bindoff et al., 2007; Talley,

2008).

Global mean sea level has increased over the

130 years of reconstructed records (Figure S15.

21). Uncertainties in this reconstruction are

30


FIGURE S15.19 Zonally averaged linear temperature trend for 1955 to 2003 (contour interval of 0.05 C per decade) for

the world ocean. Pink: increasing trend. Blue: decreasing trend. Source: From the IPCC AR4, Bindoff et al., 2007; Climate Change

2007: The Physical Science Basis. Working Group I Contribution to the Fourth Assessment Report of the Intergovernmental Panel on

Climate Change, Figure 5.3. Cambridge University Press.

large, but the increasingly accurate data sets of

recent decades show the same increasing trend.

Since 1961, which begins a period of improved

data coverage, the average rate has been about

2 mm/year. Since 1993, with even better observations,

it has been about 3 mm/year (Bindoff

et al., 2007). About half of the sea level increase

since 1993 can be attributed to changes in

thermal expansion due to the warming ocean

and half to glacier and ice cap/ice sheet melting.

Similar to changes in ocean heat content, there is

large-scale spatial variation in the sea level

change, as can be expected from the large contribution

of thermal expansion to sea level change.

The ocean’s chemical constituents have also

been changing. Much attention is focused on

changes in carbon parameters, since the ocean

is a sink for excess anthropogenic carbon

dioxide (Sabine et al., 2004), such that the ability

to quantify the ocean’s uptake of excess CO 2 is

important for future projections of climate

change. Increasing the amount of CO 2 dissolved

in the ocean also increases the ocean’s acidity,

which is also receiving wide attention (Feely et

al., 2004; National Research Council, 2010).

Because this text provides no background on

the complexities of the ocean’s carbon budget,

these changes are beyond the scope of this book.

The ocean’s oxygen distribution is also

changing. Oxygen and nutrient changes are

mostly measured from infrequent research

ship reoccupations of long sections. From such


FIGURE S15.20 Zonally averaged linear salinity trend for 1955 to 2003 (contour interval of 0.01 psu per decade) for the

world ocean. Pink: increasing trend. Blue: decreasing trend. Source: From the IPCC AR4, Bindoff et al., 2007; Climate Change

2007: The Physical Science Basis. Working Group I Contribution to the Fourth Assessment Report of the Intergovernmental Panel on

Climate Change, Figure 5.5. Cambridge University Press.

observations, oxygen at the base of the pycnocline

has declined at mid to high latitudes over

the past several decades. In the northern and

subtropical North Pacific, the changes are widespread

and have been attributed to changes in

ocean circulation and in the winter surface

outcrop densities (Deutsch et al., 2005). In the

northeastern North Atlantic, similar oxygen

declines have been similarly attributed (Johnson

& Gruber, 2007). As the ocean surface warms,

the outcropping isopycnals in the circulation

also shift. In subtropical gyres, the ventilated

isopycnals become less dense. This leads to

reduced oxygen on the underlying isopycnals

that would have been vigorously ventilated in

previous decades. In the Southern Ocean

oxygen declines in the pycnocline in the ACC

have been observed (Aoki, Bindoff, & Church,

2005). In the tropics the great oxygen minima

of the denitrification regions of the upper ocean

have been expanding (Stramma et al., 2008).

Thus we see that over the past several

decades the ocean has been warming; its salinity

has been redistributed in a manner consistent

with a warmer, more humid atmosphere; sea

level has been rising in response to ocean warming

and land ice melt; and oxygen in the upper

ocean may be declining.

Are these signals indicative of climate

change? Two major issues are that the data

sets are not optimized for global spatial

coverage or continuous temporal coverage,

32


that the observed pattern of temperature change

is inconsistent with variability arising from

natural sources alone, therefore the ocean shows

an imprint of global change.

FIGURE S15.21 Global mean sea level (mm) relative to

the 1961e1990 average, with 90% confidence intervals,

based on sparse tide gauges (red), coastal tide gauges (blue),

and satellite altimetry (black solid). Source: From the IPCC

AR4, Bindoff et al., 2007; Climate Change 2007: The Physical

Science Basis. Working Group I Contribution to the Fourth

Assessment Report of the Intergovernmental Panel on Climate

Change, Figure 5.13. Cambridge University Press.

and interpretation in terms of climate change is

usually based on fitting linear trends to time

series. The data sets have become much better

in recent years, confirming the long-term trends

or providing much more information for understanding

the observed trends.

But even if the data coverage was perfect,

trend fitting is subject to two problems: the

natural climate state at the end points of the

time series, and interpretation of the signal as

a trend rather than part of a longer term climate

variation. Thus to distinguish between natural

and anthropogenic climate variability, “attribution”

studies are useful. Barnett et al. (2005)

approached this by examining how natural

climate variability in terms of modes (Table

S15.1), volcanic activity, and solar activity affect

regional changes in upper ocean temperature in

an ocean model. Locally, anthropogenic climate

change is not necessarily distinguishable from

natural variability. But when averaged over

very large regions, essentially entire ocean

basins, in order to obtain robust results (90%

confidence level), Barnett et al. (2005) showed

References

Annamalai, H., Xie, S.P., McCreary, J.P., Murtugudde, R.,

2005. Impact of Indian Ocean sea surface temperature on

developing El Niño. J. Clim. 18, 302e319.

Aoki, S., Bindoff, N.L., Church, J.A., 2005. Interdecadal

water mass changes in the Southern Ocean between 30E

and 160E. Geophys. Res. Lett. 32 doi:10.1029/

2004GL022220.

Barnett, T.P., Pierce, D.W., AchutaRao, K., Gleckler, P.,

Santer, B., Gregory, J., Washington, W., 2005. Penetration

of human-induced warming into the World’s Oceans.

Science 309, 284e287.

Barnston, A.G., Livezey, R.E., 1987. Classification, seasonality

and persistence of low-frequency atmospheric

circulation patterns. Mon. Weather Rev. 115, 1083e1126.

Belkin, I.M., 2004. Propagation of the "Great Salinity

Anomaly" of the 1990s around the northern North

Atlantic. Geophys. Res. Lett. 31 L08306. doi:10.1029/

2003GL019334.

Bindoff, N.L., McDougall, T.J., 2000. Decadal changes along

an Indian Ocean section at 32 degrees S and their

interpretation. J. Phys. Oceanogr. 30, 1207e1222.

Bindoff, N.L., Willebrand, J., Artale, V., Cazenave, A.,

Gregory, J., Gulev, S., Hanawa, K., Le Quere, C.,

Levitus, S., Nojiri, Y., Shum, C.K., Talley, L.D.,

Unnikrishnan, A., 2007. Observations: Oceanic climate

change and sea level. In: Solomon, S., Qin, D.,

Manning, M., Chen, Z., Marquis, M., Averyt, K.B.,

Tignor, M., Miller, H.L. (Eds.), Climate Change 2007: The

Physical Science Basis. Contribution of Working Group I

to the Fourth Assessment Report of the Intergovernmental

Panel on Climate Change. Cambridge University

Press, Cambridge, UK and New York.


Nobre, P., Campos, E., Yu, L., Planton, S., Busalacchi, A.,

Moura, A.D., Servain, J., Trotte, J., 2008. The Pirata

program: History, accomplishments, and future directions.

B. Am. Meteorol. Soc. 89, 1111e1125.

Boyer, T.P., Antonov, J.I., Levitus, S., Locarnini, R., 2005.

Linear trends of salinity for the world ocean: 1955e1998.

Geophys. Res. Lett. 32 L01604. doi:1029/2004GL021791.

Broecker, W.S., 1998. Paleocean circulation during the last

deglaciation: A bipolar seesaw? Paleoceanography 13,

119e121.



at 26.5 N. Nature 438, 655e657.

REFERENCES 33

Carmack, E., Aagaard, K., 1973. On the deep water of the

Greenland Sea. Deep-Sea Res. 20, 687e715.

Cayan, D.R., 1992. Latent and sensible heat flux anomalies

over the northern oceans: Driving the sea surface


Chang, P., Ji, L., Li, H., 1997. A decadal climate variation in

the tropical Atlantic Ocean from thermodynamic air-sea

interactions. Nature 385, 516e518.

Chiang, J.C.H., Vimont, D.J., 2004. Analogous Pacific and

Atlantic meridional modes of tropical atmosphereeocean


Clark, C.O., Webster, P.J., Cole, J.E., 2003. Interdecadal

variability of the relationship between the Indian Ocean

zonal mode and East African coastal rainfall anomalies.

J. Clim. 16, 548e554.

Clarke, R.A., Swift, J.H., Reid, J.L., Koltermann, K.P., 1990.

The formation of Greenland Sea Deep Water: Double

diffusion or deep convection? Deep-Sea Res. Part A 37,

1385e1424.

Cunningham, S.A., Kanzow, T., Rayner, D., Baringer, M.O.,

Johns, W.E., Marotzke, J., Longworth, H.R., Grant, E.M.,

Hirschi, J.J.-M., Beal, L.M., Meinen, C.S., Bryden, H.L.,

2007. Temporal variability of the Atlantic meridional

overturning circulation at 26.5 N. Science 317, 935e938.

Curry, R., Dickson, B., Yashayaev, I., 2003. A change in the

freshwater balance of the Atlantic Ocean over the past

four decades. Nature 426, 826e829.

Curry, R.G., McCartney, M.S., 2001. Ocean gyre circulation

changes associated with the North Atlantic Oscillation. J.





Delworth, T.L., Mann, M.E., 2000. Observed and simulated

multidecadal variability in the northern hemisphere.

Clim. Dynam. 16, 661e676.

Dengler, M., Quadfasel, D., Schott, F., Fischer, J., 2002.

Abyssal circulation in the Somali Basin. Deep-Sea Res. II

49, 1297e1322.

Deser, C., Holland, M., Reverdin, G., Timlin, M., 2002.

Decadal variations in Labrador Sea ice cover and North

Atlantic sea surface temperatures. J. Geophys. Res. 107,

3035. doi:10.1029/2000JC000683.

Deser, C., Phillips, A.S., Hurrell, J.W., 2004. Pacific interdecadal

climate variability: Linkages between the tropics

and the North Pacific during boreal winter since 1900.

J. Clim. 17, 3109e3124.

Deutsch, C., Emerson, S., Thompson, L., 2005. Fingerprints

of climate change in North Pacific oxygen. Geophys. Res.

Lett. 32 L16604. doi:10.1029/2005GL023190.

Di Lorenzo, E., Schneider, N., Cobb, K.M., Franks, P.J.S.,

Chhak, K., Miller, A.J., McWilliams, J.C., Bograd, S.J.,

Arango, H., Curchitser, E., Powell, T.M., Rivière, P., 2008.

North Pacific Gyre Oscillation links ocean climate and

ecosystem change. Geophys. Res. Lett. 35 L08607.

doi:10.1029/2007GL032838.

Dickson, R., Lazier, J., Meincke, J., Rhines, P., Swift, J., 1996.

Long-term coordinated changes in the convective

activity of the North Atlantic. Progr. Oceanogr. 38,

241e295.

Dickson, R., Yashayaev, I., Meincke, J., Turrell, B., Dye, S.,

Holfort, J., 2002. Rapid freshening of the deep North

Atlantic Ocean over the past four decades. Nature 416,

832e837.

Dickson, R.R., Curry, R., Yashayaev, I., 2003. Recent changes

in the North Atlantic. Phil. Trans. Roy. Soc. A 361,

1917e1933.

Dickson, R.R., Meincke, J., Malmberg, S.-A., Lee, A.J., 1988.

The "Great Salinity Anomaly" in the northern North

Atlantic 1968e1982. Progr. Oceanogr. 20, 103e151.

Domingues, C.M., Church, J.A., White, N.J., Gleckler, P.J.,

Wijffels, S.E., Barker, P.M., Dunn, J.R., 2008. Improved

estimates of upper-ocean warming and multidecadal

sea-level rise. Nature 453, 1090e1093.

Durack, P.J., Wijffels, S.E., 2010. Fifty-year trends in global

ocean salinities and their relationship to broad-scale

warming. J. Clim. 23, 4342e4362.

Enfield, D., Mestas-Nuñez, A., Trimble, P., 2001. The

Atlantic Multidecadal Oscillation and its relation to

rainfall and river flows in the continental U.S. Geophys.

Res. Lett. 28, 2077e2080.

Feely, R.A., Sabine, C.L., Lee, K., Berelson, W., Kleypas, J.,

Fabry, V.J., Millero, F.J., 2004. Impact of anthropogenic

CO 2 on the CaCO 3 system in the oceans. Science 305,

362e366.

Flatau, M.K., Talley, L.D., Niiler, P.P., 2003. The North

Atlantic Oscillation, surface current velocities, and SST

changes in the subpolar North Atlantic. J. Clim. 16,

2355e2369.

Fowler, C., Emery, W.J., Maslanik, J., 2004. Satellite-derived

evolution of Arctic sea ice age: October 1978 to March

2003. IEEE Remote Sensing Lett. 1, 71e74.

Fyfe, J.C., 2006. Southern Ocean warming due to human

influence. Geophys. Res. Lett. 33 L19701. doi:10.1029/

2006GL027247.

Gille, S.T., 2002. Warming of the Southern Ocean since the

1950s. Science 295, 1275e1277.

Girton, J.B., Pratt, L.J., Sutherland, D.A., Price, J.F., 2006. Is

the Faroe Bank Channel overflow hydraulically

controlled? J. Phys. Oceanogr. 36, 2340e2349.

Häkkinen, S., Rhines, P.B., 2004. Decline of subpolar North

Atlantic circulation during the 1990s. Science 304,

555e559.

Hall, A., Visbeck, M., 2002. Synchronous variability in the

southern hemisphere atmosphere, sea ice and ocean

resulting from the annular mode. J. Clim. 15, 3043e3057.

34


Hughes, S.L., Holliday, N.P., Beszczynska-Möller, A. (Eds.),

2008. ICES Report on Ocean Climate 2007. ICES Cooperative

Research Report No. 291, p. 64. http://www.

noc.soton.ac.uk/ooc/ICES_WGOH/iroc.php (accessed

7.1.09).

Hurrell, J., 1995. Decadal trends in the North Atlantic

Oscillation: Regional temperatures and precipitation.

Science 269, 676e679.

Hurrell, J., 2009. Climate Indices. NAO Index Data provided

by the Climate Analysis Section, NCAR, Boulder,

Colorado (Hurrell, 1995). http://www.cgd.ucar.edu/

cas/jhurrell/nao.stat.winter.html (accessed 6.23.09).

Hurrell, J.W., Kushnir, Y., Ottersen, G., Visbeck, M., 2003.

An overview of the North Atlantic Oscillation. In: The

North Atlantic Oscillation: Climate Significance and

Environmental Impact. Geophys. Monogr. Ser. 134,

1e35.

Ihara, C., Kushnir, Y., Cane, M.A., 2008. Warming trend of

the Indian Ocean SST and Indian Ocean dipole from

1880 to 2004. J. Clim. 21, 2035e2046.

IPCC, 2007. Summary for Policymakers. In: Solomon, S.,

Qin, D., Manning, M., Chen, Z., Marquis, M.,

Averyt, K.B., Tignor, M., Miller, H.L. (Eds.), Climate

Change 2007: The Physical Science Basis. Contribution of


Intergovernmental Panel on Climate Change. Cambridge

University Press, Cambridge, UK, New York.



Science 274, 76e78.

JISAO, 2004. Arctic Oscillation (AO) time series, 1899 d

June 2002. JISAO. http://www.jisao.washington.edu/

ao/ (accessed 3.18.10).

Johannessen, O.M., Shalina, E.V., Miles, M.W., 1999. Satellite

evidence for an arctic sea ice cover in transformation.

Science 286, 1937e1939.

Johnson, G.C., Gruber, N., 2007. Decadal water mass variations

along 20 W in the northeastern Atlantic Ocean.


Josey, S.A., Marsh, R., 2005. Surface freshwater flux variability

and recent freshening of the North Atlantic in the

eastern subpolar gyre. J. Geophys. Res. 110 C05008.

doi:10.1029/2004JC002521.

Kaplan, A., Cane, M., Kushnir, Y., Clement, A., Blumenthal, M.,

Rajagopalan, B., 1998. Analyses of global sea surface

temperature 1856e1991. J. Geophys. Res. 103,

18567e18589.

Kawano, T., Doi, T., Uchida, H., Kouketsu, S., Fukasawa, M.,

Kawai, Y., Katsumata, K., 2010. Heat content change in

the Pacific Ocean between 1990s and 2000s. Deep-Sea

Res. II 57, 1141e1151.

Kawano, T., Fukasawa, M., Kouketsu, S., Uchida, H., Doi, T.,

Kaneko, I., Aoyama, M., Schneider, W., 2006. Bottom

water warming along the pathway of Lower Circumpolar

Deep Water in the Pacific Ocean. Geophys. Res.

Lett. 33 L23613. doi:10.1029/2006GL027933.


Kovacevic, V., Luchetta, A., 1999. The large deep water

transient in the Eastern Mediterranean. Deep-Sea Res. I

46, 371e414.

Klein, S.A., Soden, B.J., Lau, N.C., 1999. Remote sea surface

temperature variations during ENSO: Evidence for

a tropical atmospheric bridge. J. Clim. 12, 917e932.

Krishnamurthy, V., Kirtman, B.P., 2003. Variability of the

Indian Ocean: relation to monsoon and ENSO.Q.J. Roy.


Kushnir, Y., Seager, R., Miller, J., Chiang, J.C.H., 2002. A

simple coupled model of tropical Atlantic decadal

climate variability. Geophys. Res. Lett. 29, 23.

doi:10.1029/2002GL015874.

Levitus, S., Antonov, J.I., Boyer, T.P., 2005. Warming of the

world Ocean, 1955e2003. Geophys. Res. Lett. 32 L02604.

doi:10.1029/2004GL021592.

Macrander, A., Send, U., Valdimarsson, H., Jónsson, S.,

Käse, R.H., 2005. Interannual changes in the overflow

from the Nordic Seas into the Atlantic Ocean through

Denmark Strait. Geophys. Res. Lett. 32 L06606.

doi:10.1029/2004GL021463.

Madden, R., Julian, P., 1994. Observations of the 40-50 day

tropical oscillation: A review. Mon. Weather Rev. 112,

814e837.

Mantua, N.J., Hare, S.R., Zhang, Y., Wallace, J.M.,

Francis, R.C., 1997. A Pacific interdecadal climate oscillation

with impacts on salmon production. B. Am.

Meteor. Soc. 78, 1069e1079.

Marshall, G., 2003. Trends in the Southern Annular Mode

from observations and reanalyses. J. Clim. 16,

4134e4143.

Martinson, D.G., Steele, M., 2001. Future of the Arctic sea ice

cover: Implications of an Antarctic analog. Geophys.

Res. Lett. 28, 307e310.

Maslanik, J., Serreze, M., Agnew, T., 1999. On the record

reduction in 1998 western Arctic sea-ice cover. Geophys.

Res. Lett. 26, 1905e1908.

McDonagh, E.L., Bryden, H.L., King, B.A., Sanders, R.J.,

Cunningham, S.A., Marsh, R., 2005. Decadal changes in

the south Indian Ocean thermocline. J. Clim. 18,

1575e1590.

Meredith, M.P., Hogg, A.M., 2006. Circumpolar response of

Southern Ocean eddy activity to a change in the

Southern Annular Mode. Geophys. Res. Lett. 33 L16608.

doi:10.1029/2006GL026499.

Molinari, R.L., Fine, R.A., Wilson, W.D., Curry, R.G., Abell, J.,

McCartney, M.S., 1998. The arrival of recently formed

Labrador Sea Water in the Deep Western Boundary Current

at 26.5 N. Geophys. Res. Lett. 25, 2249e2252.

REFERENCES 35

Nakano, H., Suginohara, N., 2002. Importance of the eastern

Indian Ocean for the abyssal Pacific. J. Geophys. Res. 107

(C12) doi:10.1029/2001JC001065.

National Research Council, 2010. Ocean acidification: A

national strategy to meet the challenges of a changing

ocean. National Academies Press, Washington D.C. pp 152.

NOAA CPC, 2005. Madden/Julian Oscillation (MJO).

NOAA/National Weather Service. http://www.cpc.

ncep.noaa.gov/products/precip/CWlink/MJO/mjo.


NOAA ESRL, 2009b. Linear correlations in atmospheric

seasonal/monthly averages. NOAA Earth System

Research Laboratory Physical Sciences Division. http://

www.cdc.noaa.gov/data/correlation/ (accessed 10.30.09).

NSIDC, 2007. Arctic sea ice news fall 2007. National Snow

and Ice Data Center. http://nsidc.org/arcticseaicenews/


NSIDC, 2008b. Arctic sea ice down to second-lowest extent;

likely record-low volume. National Snow and Ice Data

Center. http://nsidc.org/news/press/20081002_seaice_

pressrelease.html (accessed 3.17.09).

NSIDC, 2009b. Arctic climatology and meteorology primer.

National Snow and Ice Data Center. http://nsidc.org/

arcticmet/ (accessed 3.1.09).

Olsen, S.M., Hansen, B., Quadfasel, D., Østerhus, S., 2008.

Observed and modeled stability of overflow across the

Greenland-Scotland ridge. Nature 455, 519e523.

Polyakov, I.V., 22 co-authors, 2005. One more step toward

a warmer Arctic. Geophys. Res. Lett. 32 L17605.

doi:10.1029/2005GL023740.

Proshutinsky, A.Y., Johnson, M.A., 1997. Two circulation

regimes of the wind-driven Arctic Ocean. J. Geophys.

Res. 102, 12493e12514.

Purkey, S.G., Johnson, G.C., 2010. Antarctic bottom water

warming between the 1990s and 2000s: Contributions to

global heat and sea level rise budgets. J. Clim. 23, 6336e6351.

Reverdin, G., Durand, F., Mortensen, J., Schott, F.,

Valdimarsson, H., 2002. Recent changes in the surface

salinity of the North Atlantic subpolar gyre. J. Geophys.

Res. 107 (C12) doi:10.1029/2001JC001010.

Rigor, I.G., Wallace, J.M., Colony, R.L., 2002. Response of sea

ice to the Arctic Oscillation. J. Clim. 15, 2648e2663.

Roemmich, D., Gilson, J., Davis, R., Sutton, P., Wijffels, S.,

Riser, S., 2007. Decadal spin-up of the South Pacific

subtropical gyre. J. Phys. Oceanogr. 37, 162e173.


convection in the Greenland Sea. J. Geophys. Res. 110

C04015. doi:10.1029/2004JC002318.

Rothrock, D., Yu, Y., Maykut, G., 1999. Thinning of the

Arctic sea-ice cover. Geophys. Res. Lett. 26, 3469e3472.

Sabine, C.L., Feely, R.A., Gruber, N., Key, R.M., Lee, K.,

Bullister, J.L., Wanninkhof, R., Wong, C.S.,

Wallace, D.W.R., Tilbrook, B., Millero, F.J., Peng, T.-H.,

Kozyr, A., Ono, T., Rios, A.F., 2004. The oceanic sink for

anthropogenic CO 2 . Science 305, 367e371.

Saji, N.H., Goswami, B.N., Vinayachandran, P.N.,

Yamagata, T., 1999. A dipole mode in the tropical Indian

Ocean. Nature 401, 360e363.

Schneider, N., Cornuelle, B.D., 2005. The forcing of the

Pacific Decadal Oscillation. J. Clim. 18, 4355e4373.



Schott, F.A., Stramma, L., Giese, B.S., Zantopp, R., 2009.

Labrador Sea convection and subpolar North Atlantic

Deep Water export in the SODA assimilation model.


Serreze, M.C., Holland, M.M., Stroeve, J., 2007. Perspectives

on the Arctic’s shrinking sea-ice cover. Science 16,

1533e1536.

Shimada, K., Kamoshida, T., Itoh, M., Nishino, S.,

Carmack, E.C., McLaughlin, F., Zimmerman, S.,

Proshutinsky, A., 2006. Pacific Ocean Inflow: Influence

on catastrophic reduction of sea ice cover in the Arctic

Ocean. Geophys. Res. Lett. 33 L08605. doi; 10.1029/

2005GL025624.



water in the Arctic Ocean. J. Geophys. Res. 109

C02027. doi:10.1029/2003JC002009.

Stramma, L., Johnson, G.C., Sprintall, J., Mohrholz, V., 2008.

Expanding oxygen minimum zones in the tropical

oceans. Science 320, 655e658.

Stramma, L., Kieke, D., Rhein, M., Schott, F., Yashayaev, I.,

Koltermann, K.P., 2004. Deep water changes at the

western boundary of the subpolar North Atlantic during

1996 to 2001. Deep-Sea Res. I 51, 1033e1056.

Sundby, S., Drinkwater, K., 2007. On the mechanisms

behind salinity anomaly signals of the northern North

Atlantic. Progr. Oceanogr. 73, 190e202.

Sutton, R.T., Jewson, S.P., Rowell, D.P., 2000. The elements of

climate variability in the tropical Atlantic region. J. Clim.

13, 3261e3284.

Swift, J.H., Aagaard, K., Timokhov, L., Nikiforov, E.G., 2005.

Long-term variability of Arctic Ocean Waters: Evidence

from a reanalysis of the EWG data set. J. Geophys. Res.

110 C03012. doi:10.1029/2004JC002312.


reviewed for the CNLS conference. Physica. D 98,

625e646.



throughflow components. Progr. Oceanog. 78, 257e303.

doi:10.1016/j.pocean.2008.05.001.

Thompson, D.W.J., Solomon, S., 2002. Interpretation of

recent southern hemisphere climate change. Science 296,

895e899.

36


Thompson, D.W.J., Wallace, J.M., 1998. The Arctic- Oscillation

signature in the wintertime geopotential height and

temperature fields. Geophys. Res. Lett. 25, 1297e1300.

Thompson, D.W.J., Wallace, J.M., 2000. Annular modes in

the extratropical circulation. Part I: Month-to-month


Tourre, Y.M., White, W.B., 1995. ENSO Signals in global

upper-ocean temperature. J. Phys. Oceanogr. 25,

1317e1332.

Tourre, Y.M., White, W.B., 1997. Evolution of the ENSO

signal over the Indo-Pacific domain. J. Phys. Oceanogr.

27, 683e696.

Trenberth, K.E., Hurrell, J.W., 1994. Decadal atmosphereocean

variations in the Pacific. Clim. Dyn. 9, 303e319.

Trenberth, K.E., Jones, P.D., Ambenje, P., Bojariu, R.,

Easterling, D., Klein Tank, A., Parker, D.,

Rahimzadeh, F., Renwick, J.A., Rusticucci, M., Soden, B.,

Zhai, P., 2007. Observations: Surface and Atmospheric

Climate Change. In: Solomon, S., Qin, D., Manning, M.,

Chen, Z., Marquis, M., Averyt, K.B., Tignor, M.,

Miller, H.L. (Eds.), Climate Change 2007: The Physical

Science Basis. Contribution of Working Group I to the

Fourth Assessment Report of the Intergovernmental

Panel Climate Change. Cambridge University Press,

Cambridge, UK and New York.

Vellinga, M., Wood, R.A., 2002. Global climatic impacts of

a collapse of the Atlantic thermohaline circulation.

Climatic Change 43, 251e267.

Visbeck, M., 2002. The ocean’s role in climate variability.

Science 297, 2223e2224.

Visbeck, M., Chassignet, E.P., Curry, R.G., Delworth, T.L.,

Dickson, R.R., Krahmann, G., 2003. The ocean’s response

to North Atlantic Oscillation variability. In: The North

Atlantic Oscillation: Climate significance and environmental

impact. Geophys. Monogr. Ser. 134, 113e146.

Wang, C., 2002. Atlantic climate variability and its associated

atmospheric circulation cells. J. Clim. 15, 1516e1536.

WCRP, 1998. CLIVAR Initial Implementation Plan. WCRP-

103, WMO/TD No. 869, ICPO No. 14, 367 pp.

Webster, P.J., Magana, V.O., Palmer, T.N., Shukla, J.,

Tomas, R.A., Yanai, M., Yasunari, T., 1998. Monsoons:

Processes, predictability, and the prospects for prediction.

J. Geophys. Res. 103, 14451e14510.

Webster, P.J., Moore, A.M., Loschnigg, J.P., Leben, R.R., 1999.

Coupled ocean-atmosphere dynamics in the Indian

Ocean during 1997e98. Nature 401, 356e360.

Wong, A.P.S., Bindoff, N.L., Church, J.A., 2001. Freshwater

and heat changes in the North and South Pacific Oceans

between the 1960s and 1985e94. J. Clim. 14, 1613e1633.

Yashayaev, I., 2007. Hydrographic changes in the Labrador

Sea, 1960e2005. Progr. Oceanogr. 73, 242e276.

Yuan, X., 2004. ENSO-related impacts on Antarctic sea ice:

a synthesis of phenomenon and mechanisms. Antarct.

Sci. 16, 415e425. doi:10/1017/S0954102004002238.

Zhang, R., Vallis, G.K., 2006. Impact of Great Salinity

Anomalies on the low-frequency variability of the North

Atlantic Climate. J. Clim. 19, 470e482.

Zhong, A., Hendon, H.H., Alves, O., 2005. Indian Ocean

variability and its association with ENSO in a global

coupled model. J. Clim. 18, 3634e3649.

C H A P T E R

S16

Instruments and Methods

This chapter on methods for measuring the

large-scale circulation and water properties

of the ocean, emphasizing instrumentation, is

published solely online at http://booksite.

academicpress.com/DPO/; “S” denotes online

supplemental material. Many of the methods for

measuring basic properties such as temperature,

salinity, and pressure were described briefly in

Chapter 3. Some of the satellite observations

were described in Chapters 3e5. Many of these

techniques are also used for smaller scale

phenomena such as waves. Every decade brings

new advances and thus the descriptions presented

in succeeding editions of this text have been

quickly outdated. Nevertheless, it is useful to

understand what types of instruments have been

available at different points in oceanographic

development and their resolution, precision, and

accuracy. The information here primarily supports

Chapter6,DataAnalysisConceptsandObservational

Methods, in the printed textbook.

In Section S16.1 some of the sampling issues for

physical oceanography are discussed, augmenting

the discussion in Chapter 1. In Section S16.2

platforms for observations are described. In

Sections S16.3 through S16.8 instruments for in

situ observations (within the water column) are

reviewed. Section S16.9 is an overview of satellite

remote sensing, and Section S16.10 briefly

describes oceanographic archives. A recent review

of oceanographic instrumentation by Howe and

Chereskin (2007) is also recommended.

S16.1. THE IMPACT OF SPACE AND

TIMESCALES ON SAMPLING AND

INSTRUMENTATION

The time and space scales of physical oceanographic

phenomena were summarized in

Chapter 1 (Figure 1.2). Data collection requirements

to study motions with so many time

and space variations are demanding, calling

for a wide variety of sampling methods. As

described in Chapter 6, studies at almost every

scale require averaging or filtering to remove

space and timescales that are not of interest. It

is not possible to measure every space and timescale,

however, to form perfect averages and

statistics. Therefore observational oceanographers

must understand the sources of error

and uncertainty, which can be due to instrumental

or sampling limitations, or to signals at

different frequencies and wavelengths.

For example, traditional deep oceanographic

profiles (Section S16.4) were and continue to be

made from research ships to study the very

largest spatial and temporal scales of the ocean

circulation and property distributions. These

remain the only way to measure the deep

ocean with high accuracy, and the only way

to make most chemical measurements. A deep

oceanographic station can take up to three hours

and a cross-section across an ocean can take up

to two months, posing limitations to interpretation.

The individual, widely separated profiles

1

2

S16. INSTRUMENTS AND METHODS

cannot be used to study tides, internal waves, or

eddies, for instance, but these and other smaller

scale motions affect the individual station

measurements. There are, however, useful

ways to process and analyze the data so that

they can be used to study the large space and

timescales of interest.

As a second example, satellite altimeters

(Section S16.9.9) measure the ocean’s surface

height, passing over each point on the ocean’s

surface every week or two. Surface height

depends on several things: the ocean circulation,

surface waves and tides, expansion and contraction

due to more or less heat or salt in the water,

and the uneven distribution of mass in the solid

earth (variations in the geoid). The geoid, which

does not vary in time, dominates the altimetric

signal. Therefore the time-dependent altimetry

measurements have been most useful, providing

significant information about the time-dependent

“mesoscale” (tens to hundreds of kilometers)

and large-scale time dependence in seasurface

height, which is associated with changes

in large scale circulation, climate variability such

as El Niño, and global sea level rise.

Interpretation of the altimetry measurements

in the presence of thermal expansion requires

information on the temperature and salinity

structure beneath the surface, which a satellite

cannot see. Therefore in situ measurements are

combined with altimetry. Since the different

data sets are mismatched in sampling frequency

and location, the combination poses significant

data analysis challenges, dealt with most

recently through use of data assimilation

(Section 6.3.4). And as a third example drawn

from altimetry, the many days between satellite

passes over a given location means that shorter

timescales, due for instance to tides, are

measured at different times in their cycles on

each satellite pass. This “aliasing” produces

a false long timescale (Section 6.5.3). Great care

is taken in choosing satellite orbital frequency

and in interpretation of the data to properly

deal with these shorter timescales, to remove

them as much as possible from the longer

timescales.

Returning to observing the largest scale circulation

from the top to the bottom of the ocean,

which is the primary focus of this text, it might

appear that employing numerous instruments

that measure the currents directly would be

the best approach. Indeed, at the onset of the

twenty-first century a global program (Argo,

described in Section S16.5.2) to continuously

monitor velocity within the water column was

initiated using relatively inexpensive subsurface

floats that follow the subsurface currents (mostly

at a single depth) and report back to satellites at

regular intervals. This program has already

revolutionized observing of the ocean interior,

primarily because of the temperature and

salinity profiles collected on every trip to the

surface, which has been standardized at tenday

intervals; the velocity data have been less

utilized. A global deployment of surface drifters

accomplishes the same objective at the sea

surface (Section S16.5.1). These ocean-wide

Lagrangian sampling methods were not

possible prior to the beginning of global satellite

communications, and it is still prohibitively

expensive to instrument the ocean at all depths.

Current meters, both mechanical and acoustic,

directly measure flow at a given point for

several years; they were developed and

deployed widely after the 1950s. Current meters

give information on the velocity (speed and

direction) of the water only close to the location

(in time and space) of the instrument itself;

experience indicates that large variations in

velocity can occur over small distances as well

as over small time intervals. Because of these

spatial scales and because of the high expense

of current meter deployments, it has not proven

possible to widely instrument the ocean. Current

meters are now used primarily in well-defined

currents of no more than several hundred kilometers

width, or in specific target areas to sample

all of the temporal scales (the full time spectrum)

in that area, sometimes for many years. All of the

PLATFORMS 3

direct current measurements of subsurface

currents have provided just a small proportion

of our observed knowledge of the ocean circulation.

On the other hand, where they have been

used they provide invaluable information; for

instance, quantifying the total transport and variations

of strong, relatively narrow currents like

the Gulf Stream or Kuroshio.

In the absence of sufficient direct measurements

of ocean currents, oceanographers

studying the circulation use indirect methods.

One of the oldest, remaining in very common

use, is the geostrophic or dynamic method, which

relates the horizontal pressure distribution to

horizontal currents (Section 7.6). Most currents

with timescales greater than a few days (except

at the equator) are in geostrophic balance, which

is a balance between the horizontal change

(gradient) in pressure and the Coriolis force.

The geostrophic velocity is perpendicular to the

pressure gradient direction due to Earth’s rotation.

The pressure distribution depends on seasurface

height and also on the vertical profile of

seawater density at a given latitude and longitude.

Thus the chief method for mapping ocean

circulation has been to measure the temperature

and salinity distribution of the ocean. The density

distribution is then calculated, from which the

horizontal pressure gradient is calculated at

every depth, given an assumption of the pressure

gradient at one depth (which could be at the

surface, due to surface height). The geostrophic

currents are then calculated.

The step of estimating the pressure gradient

at one depth is nontrivial, given the general

lack of distributed velocity observations. (The

subsurface float deployments starting in the

1990s were first motivated by providing such

a velocity field at one depth.) The traditional

approach has been to require mass conservation

within ocean regions and then to make educated

guesses about the velocity distribution at a given

depth, based on mapping property distributions

within the ocean. “Inverse methods” (introduced

but not developed in Section 6.3.4)

formalize the use of constraints based on mass

conservation and on property distributions,

which are affected by mixing.

Some water properties also are inherent

tracers of time (Sections 3.6 and 4.7). These

include tracers that are biologically active and

are reset at specific locations. For example,

oxygen content is saturated through contact

with the atmosphere in the surface layer, and is

then consumed by bacteria within the water

column, yielding a rough age for a given water

parcel. The built-in clock of radioactive decay

in transient tracers offers more promise, as it is

independent of the physical and biological character

of the environment. Anthropogenic tracers

such as chlorofluorocarbons (CFCs) have been

injected into the earth system by mankind. If

the history of their release into the environment

is known, as is the case for CFCs, then they are

useful tracers of the paths taken by surface ocean

waters as they move into the interior ocean.

S16.2. PLATFORMS

Manned measurement platforms are described

here. Autonomous (unmanned) platforms such as

floating or moored instruments, or satellites, are

described in later sections.

S16.2.1. Ocean Research Vessels

The majority of oceanographic measurements

have been made from research ships with auxiliary

measurements from merchant ships (ocean

temperature and salinity, weather) and from

coastal stations (tide gauges, wave staffs, lighthouse

temperature and salinity observations,

etc.). Today the research vessel continues to be

essential for oceanographic research, but rapid

improvements in technology, including satellite

communications and long-lived mooring capabilities,

have introduced new options. These

include wider use of commercial vessels as platforms

for expendable devices and deployment

4


FIGURE S16.1 The R/V Roger

Revelle is a modern research vessel.

(Photo courtesy of Katy Hill.)

of autonomous instruments that can profile the

ocean while reporting their data via satellite.

New options also include drifting and moored

platforms as well as a new potential for interactive

devices, such as gliders. In addition, the

advantages of observing the earth from aircraft

and satellites have further motivated the

continued development of these measurement

technologies. The need to validate and verify

satellite surface measurements has given rise in

turn to new in situ sampling programs to

provide these calibration data.

Research vessels have evolved in the past few

decades from rather large all-purpose vessels to

smaller, more automated ships that can perform

the same large variety of tasks at a lower cost of

operation. The need to deploy deep-sea moorings,

launch open ocean sampling systems,

and make specific process measurements

ensures the continued need for ocean research

vessels. A good research vessel is reliable,

maneuverable, stable at sea, and has comfortable

living and working spaces. The R/V Revelle

(Figure S16.1) is a typical large research vessel

(Scripps Institution of Oceanography, 2009).

The Revelle was built in 1996. Its overall length

is 277 feet and its displacement is 3180 tons. It

carries a crew of 23 with room for 38 scientists.

For work in ice-covered regions, icebreakers

are required. Most of the icebreakers used for

research have dual purposes, including as

supply and rescue ships. The U.S. Coast Guard’s

icebreakers are primarily used for research in

the Arctic and Antarctic. The Alfred Wegener

Institute’s FS Polarstern, which began operation

in 1982, is a dedicated research ship (Figure

S16.2). Icebreakers have double hulls and

rounded bows, as seen in this figure. Ice is

broken by running the ship up onto the ice.

S16.2.2. Propulsion and

Maneuverability

Maneuverability is a critical factor in research

vessel operations, which primarily involve

working with instruments deployed over the

side of the ship into the ocean. Many research

operations are improved if the ship can remain at

PLATFORMS 5

FIGURE S16.2 The FS Polarstern

is a modern icebreaking

research ship. (Photo courtesy of

P. Lemke/Alfred Wegener Institute.)

a geographical location and if the angle between

the sea surface and cables deployed

over the side remains constant. This level of

control is usually achieved by a variety of

methods including twin propellers and

various types of thrusters.

S16.2.3. Winches, Wires, and Support

Systems

The hydrographic winch is an essential piece

of equipment on an oceanographic vessel. The

winch has a drum holding wire rope on

which instruments are lowered into the sea.

For water sampling without electronic instruments

(which is now rare), or sampling with

instruments with internal batteries and data

recording, a medium-duty winch with 2000 to

6000 m of 4 mm diameter wire rope and a

7-to 15-kW motor may be used. For heavier

work, such as dredging, coring, and so forth,

winches with up to 15,000 m of 10-to 20-mm

wire and 75 to 150 kW have been used. The

wire rope used is multi-strand for flexibility,

and made of galvanized or stainless steel

(more expensive) to resist corrosion. (Seawater

is one of the most corrosive substances known,

given time to act.) The winches must be capable

of reeling the wire in or out at speeds up to 100

m/min but must also be controllable in speed so

that an instrument can be brought accurately to

a position for operation or to where it can be

reached for recovery.

For instruments that telemeter their information

to the surface, a steel cable incorporating

one or more insulated electrical conductors is

used. The winch must have “slip rings” to

transmit the electrical signals from the wire to

the deck instruments while the winch drum is

turning. Early versions of these slip rings

were simple copper brushes slipping over

a rotating steel shaft. More recently the slip

rings are mercury baths in which the steelconducting

shaft rotates. Either way the

purpose is to transmit electrical signals for

a rotating system. Since most electronic

profiling instruments transmit multiple observables

to the sea surface, these signals are

6


frequency-multiplexed in the instrument and

then transmitted on the same single conductor

wire.

Most research ships are designed with open

deck workspaces to allow easy access to the

water, often in the form of a lowered deck at

the rear of the ship or a science deck on one

side of the ship. Multiple support systems

(cranes, A-frames, etc.) are available to load

equipment as well as to lower equipment over

the side into the water and back again on to

the deck. Winches and cranes are placed in

appropriate locations for handling samplers

and sensors that go over the side. As an

example, a schematic of a portion of the “science

deck” of another research ship of the Scripps

Institution of Oceanography, the R/V Melville,

(Figure S16.3) shows the winches, A-frames,

and cranes. In addition to these winches and

A-frames, this deck has cranes for manipulating

equipment and storage vans. Thus, supplies and

equipment for particular experiments can be

stored in a container van that can then be loaded

intact on the ship.

Winches and their support/deployment

systems are often part of the ship’s equipment,

although individual research groups often

provide specialized winches. Many research

ships may have a weather-protected winch

house, as shown in Figure S16.4, which also

shows the “joystick” type of controls (black

knobs) used to operate the winch and the associated

booms and cranes.

S16.2.4. Workspaces: Dry Labs

and Wet Labs

Research vessels have various laboratory

spaces. Open or partially enclosed spaces near

the winches are used for sampling or servicing

instruments. Interior laboratories are divided

into “wet labs” and “dry labs.” In the former, water

samples can be analyzed, fish samples examined,

net tows examined, and so forth. In the latter,

samples are examined under a microscope, data

are analyzed, and other computer and

instrument hardware might be used or serviced.

The distribution of the different types of labs

on the main deck of the R/V Melville is shown in

Figure S16.5. Here the dry lab is referred to as

the “analytical lab” and the main lab serves

as the wet lab. Note the double-door access

from the main lab to the deck needed to bring

large pieces of equipment and oceanographic

samples into the lab. This main lab also has

good access to the other labs on the ship. Similar

lab spaces are found on all research ships.

S16.2.5. Navigation (also Section

S16.9.13.2)

Research vessels require accurate and

precise navigation as all oceanographic sampling

is associated with a time and a location.

As discussed in Chapter 1, early oceanographic

expeditions had to rely on traditional methods

of navigation using sextant and chronometer.

While it is surprising how well the early ship

captains were able to specify their locations,

it is clear that these methods cannot compare

in accuracy with modern navigation methods,

which are now based mainly on GPS satellites

(Section S16.9.13.2). It is important to remember

this navigation limitation when analyzing

historical data, particularly when they are

mixed with more modern data.

S16.2.6. Alternative Sampling

Platforms

S16.2.6.1. Aircraft

A rapid sampling alternative to working

from a research vessel is to use sensors mounted

on aircraft. Many airborne systems sense

remotely using radiation either emitted or

reflected from the ocean’s surface. Infrared

sensors are used to map sea-surface temperature

(SST) patterns while visible light sensor

channels are used to measure patterns of ocean

color related to biological productivity and the

FIGURE S16.3 The science deck of the RV Melville. The CTD and hydrographic winches, A frames, and cranes used for maneuvering equipment

over the side are located on the open deck in the rear. Source: From Scripps Institution of Oceanography (2009).

PLATFORMS 7

8


FIGURE S16.4 FS Polarstern

winch station (foreground) and

deck operations. (Photo courtesy of

H. Grobe/Alfred Wegener Institute.)

amount of chlorophyll in the surface waters.

Multispectral scanners are capable of simultaneously

measuring radiation in both the visible

and thermal infrared channels. Recently, passive

and active microwave sensors have also been

flown on aircraft to sense ocean parameters.

One of the most useful of these is Synthetic

Aperture Radar (SAR), which uses the motion

of the aircraft to synthesize a larger antenna

than an aircraft could carry, making it possible

to greatly improve the ground resolution.

Aircraft SAR imagery has been particularly

useful in the detailed mapping of sea ice and

its motion.

Another important use of aircraft is in the

collection of upper layer temperature profile

data using expendable profilers, described

below in Section S16.4.2.5. Helicopters and

aircraft are often used in polar studies to ferry

instruments from the ship to and from the ice

and may also carry instrumentation over large

portions of the ice cover. The limited range of

helicopters limits their usefulness in normal

oceanographic sampling. When long transects

are required, such as in the Arctic, fuel must

be stored on the ice to refuel the helicopter

for continued operation. Routine sampling

throughout the Arctic is conducted in this manner

by Russia, using aircraft to reach sampling

sites on the ice.

S16.2.6.2. Ships of Opportunity

As early as the eighteenth century, Matthew

Fontaine Maury realized that routine observations

from ships operating at sea for other

than oceanographic measurement purposes

could be useful for a variety of applications.

Most of the maps of airesea heat fluxes in

Chapter 5 are based on the routine weather

observations made by various ships at sea

and not by dedicated research vessels. This

sampling concept was extended in the 1970s

to include routine deployment of expendable

temperature profilers (expendable bathythermograph

or XBTs, described in Section

S16.4.2.5) from merchant vessels to provide

repeat coverage of the upper layer (<700 m)

thermal structure. Some the programs also

FIGURE S16.5 Scientific laboratories on the main deck of the R/V Melville. Source: From Scripps Institution of Oceanography (2009).

PLATFORMS 9

10


include expendable conductivity, temperature

and depth profilers (XCTDs), which measure

both temperature and conductivity, so that

salinity profiles are also available. These Ship

of Opportunity (SOOP) or Volunteer Observing

Ship (VOS) programs started in the North

Pacific but quickly spread to other parts of the

Pacific and to the Atlantic and Indian Oceans.

Today many shipping lines collect XBT and

XCTD profiles on a routine basis.

Coastal vessels such as ferries are also used as

ships of opportunity. Ferries frequently travel

through interesting regions such as river

discharges or isolated seas between islands.

Instruments can be installed in ferries to continuously

monitor near surface temperature and

salinity to study the temporal changes associated

with the river outflow. For example,

a British Columbia ferry was instrumented to

study fluctuations of the Fraser River outflow

into Georgia Strait in western Canada and the

effect this outflow has on local pollution in the

Strait. Infrared radiometers can also be installed

on such ferries to continuously measure the skin

SST. Cruise ships in the Caribbean and Bahamas

are also collecting research data relevant to the

Gulf Stream system along their regular tracks.

Merchant vessels also collect continuous

records while underway (in addition to meteorology).

SSTobservations are relatively common.

Several ships are outfitted with research-specific

acoustic Doppler current profilers. Others collect

underway surface water carbon dioxide (pCO 2 )

or oxygen measurements.

S16.2.6.3. Special Sampling Platforms:

Floating Laboratory Instrument Platform

Some specialized sampling platforms have

been developed for oceanographic sampling.

The Floating Laboratory Instrument Platform

(FLIP) from Scripps Institution of Oceanography

is particularly unique. FLIP is not

a ship; it is a 355 foot spar buoy with research

and living quarters at one end. It is towed in

the horizontal position to its operating location.

(This limits its range in comparison with a standard

research vessel, which can work very far

from home port.) Once on site, part of FLIP is

flooded so that its lower half sinks. During

FLIP’s transition, everything must rotate

through 90 degrees. FLIP was developed by

Fred Spiess and Fred Fisher and built in 1962;

it had a major refit from 1994 to 1996, and

continues to operate on a full schedule.

FLIP provides a very stable platform for longterm

measurements at sea. Numerous booms

and radial supports allow various instruments

to be installed and suspended from the platform

(Figure S16.6). Instruments can also be mounted

on the submerged portion of the hull. Unlike

a research vessel, FLIP is designed to remain

relatively motionless. It does have a well-characterized

mode of vertical oscillation that must

be compensated for when analyzing time series

data it has collected.

FLIP provides an ideal platform for airesea

interaction studies. It has been equipped with

instruments to measure the airesea fluxes and

coincident upper ocean characteristics, and has

contributed immensely to knowledge of surface

fluxes and associated surface layer processes

(mixed layer development, internal wave generation,

Langmuir circulation, etc.). One limitation

of FLIP is that it is unsafe in high seas, so it is

difficult to measure airesea interaction at high

wind speeds and sea states.

S16.3. DEPTH AND PRESSURE

MEASUREMENTS

When instruments are lowered or dropped

into the ocean, it is necessary to measure their

depth. This has not always been easy. Depths

for instruments attached to cables were originally

estimated using the amount of cable

deployed. This is measured using a meter wheel,

which counts the number of turns of a pulley

over which the cable passes. In calm conditions

with negligible winds or currents, this is close

DEPTH AND PRESSURE MEASUREMENTS 11

FIGURE S16.6 FLIP on station.

Source: From Marine Physical

Laboratory, Scripps Institution of

Oceanography (2009).

to the actual depth. More often the ship drifts

with the wind or surface currents and the wire

is neither straight nor vertical, so the actual

depth is less than the length of wire paid out.

A much more accurate and modern method of

measuring the depth of an instrument is to

measure its pressure directly. The pressure is

related to depth through the hydrostatic relation

(Section 3.2 and Table 3.1). Pressure can be

precisely converted to depth using the local value

of gravity and the vertical density profile. Oceanographers

usually use a non-SI unit for pressure,

the decibar, where 1 dbar ¼ 10 4 Pa and the Pascal

(Pa) is the SI unit. Efforts on the part of major

publishing companies to change this practice

failed because the decibar is an intuitive unit: 1

dbar is nearly equal to 1 meter in depth.

Historically, the pressure difference recorded

in the mercury columns of paired protected and

unprotected reversing thermometers was used

to accurately measure the depth of the bottle

sample. (See Sections 3.3.1 and S16.4.2 on

temperature measurements.)

Pressure is now measured directly on most

instruments. Very accurate pressure measurements

can be made using a quartz crystal, whose

frequency of oscillation depends on pressure.

This technology is used in modern CTDs.

Temperature must be accurately measured for

the best pressure accuracy. In CTDs, a thermistor

is part of the quartz pressure transducer. The

accuracy is 0.01% and precision is 0.0001%

of full-scale values. A bourdon tube arrangement

is used to transfer pressure changes to the support

of the quartz sensor, and a “tuning fork” is used to

sense the change in oscillating frequency due to

the pressure changes (Figure S16.7).

Older devices that measured pressure

directly, but much less accurately, included the

“bourdon tube” with a sliding electrical potentiometer.

It may be accurate to 0.5e1.0%.

Another device is the electrical strain-gauge

pressure transducer, which uses the change of

electrical resistance of metals under mechanical

tension. Accuracies to 0.1% or better of fullscale

pressure range are claimed, with resolution

of 0.01% or better. Yet another device is

the “Vibratron” pressure gauge, in which the

water pressure varies the tension in a stretched

wire, which is caused to vibrate electromagnetically.

The frequency of vibration

depends on the wire tension and hence on the

12


FIGURE S16.7 Quartz pressure sensor, designed for

high pressures. The quartz crystal is in the “tuning fork.”

Source: From Paroscientific, Inc. (2009).

depth. The vibration frequency gives a measurement

of pressure to about 0.25% accuracy.

Expendable instruments such as the XBT

(Section S16.4.2.5), which measures temperature

using a thermistor, do not actually measure depth

directly but infer it from the time of a “freely

falling” body with an assumed “known” constant

fall rate. This is an error source for the XBT’s

temperature profile since there are many reasons

why an individual probe’s actual fall rate might

deviate from a known constant. First, since a spool

of copper wire pays out from the probe to achieve

“free fall,” the XBT is continually undergoing

a change of mass and hence fall rate. This change

is offset by the fact that the probe reaches its

maximum operating depth well before its wire

supply is depleted. At the same time the buoyancy

of an individual probe must be dictated by

the density structure at the location where the

probe is deployed. It is best assumed that fall

rate equations yield XBT “depths” that are not

accurate to more than a few meters, which is

consistent with the lack of individual XBT thermal

calibration, as discussed in Section S16.4.2.5.

Seafloor topography is mapped with acoustic

systems that use the round-trip travel time of an

acoustic pulse to determine the sea bottom depth

below the ship (Section 2.9). Often called “echo

sounders” or “depth sounders,” these systems

can take the form of less precise instruments often

called “fathometers” used for routine bottom

mapping from the ship’s bridge. More complex

systems are used to precisely map the seafloor.

The resolution of topographic features is a function

of the acoustic frequency used to sense the

bottom. Since acoustic transmission is inversely

proportional to sound speed, low frequency

sound penetrates deeper with a wider beam

and less spatial resolution, while higher frequencies

can resolve the detailed bottom structure

better but require a much greater input energy

to reach significant depths. Acoustic transponders,

called pingers, are often attached to lowered

instruments to ensure that an oceanographic

sensor is not accidentally run into the bottom.

S16.4. WATER PROPERTY

MEASUREMENTS (TEMPERATURE,

SALINITY, DENSITY, AND

TRACERS)

S16.4.1. Water-Sampling Bottles

To determine the properties of a seawater

sample, we must first obtain the sample. For

a “surface” sample, a bucket on a rope sufficed

in the past to obtain water for temperature and

salinity measurements, and is still sometimes

used (Figure S16.8). The older buckets were

wooden (Figure S16.8b), which worked well

with slowly moving sailing ships. These were

replaced with canvas buckets and then with

plastic. Since water temperature is often

measured from the buckets, the shift from

wood to canvas to plastic has had consequences

for constructing useful climate records (Folland

WATER PROPERTY MEASUREMENTS (TEMPERATURE, SALINITY, DENSITY, AND TRACERS) 13

FIGURE S16.8 (a) Special bucket samplers for SST measurements. The rightmost is a canvas bucket and the other two are

metal containers. (b) Traditional wooden bucket used to collect surface samples. Source: From Folland and Parker (1995).

& Parker, 1995). Standard buckets are small,

holding only about a liter.

For the past several decades, surface samples

have been routinely collected continuously

through clean water intake lines. A thermistor

in the intake line measures the water temperature.

These temperatures differ from a surface

bucket temperature or a satellite infrared

temperature, depending on the depth of the

intake. For research purposes, separate intake

lines can bring water directly and continuously

to the laboratory, where surface properties in

addition to temperature can also be measured.

Such properties include salinity and concentrations

of dissolved gases such as oxygen and CO 2 .

For subsurface samples, different types of

water-sampling “bottles” have been used. These

are generally metal or plastic tubes with either

plug valves at each end (“Nansen bottle,” Figure

S16.9) or spring-loaded end-caps with rubber

washers (“Niskin bottle,” Figure S16.10). Materials

for the bottles and parts are carefully

14


impractical. The bottles are closed by the tripping

action of a “messenger,” which is a small

metal weight that slides down the wire. Generally

a number of bottles (12 to 24) are attached

in series at predetermined intervals along the

wire (a “bottle cast”) and closed in succession.

Each in turn releases a messenger to close the

next bottle below it. When the bottles are

brought back on deck, the water samples are

drawn through a tap, following a routine

designed to obtain a pure sample. In some older

designs, the tripped bottle was released at its

upper end and rotated through 180 degrees

about a hinge at its lower end where it was

clamped to the wire. These “reversing water

FIGURE S16.9 Nansen bottle, circa 1960, for mounting

individually on a wire with reversing thermometer racks.

Source: From Dietrich, Kalle, Krauss, and Siedler (1980); Ocean

World (2009).

chosen to avoid contamination of the samples.

Prior to the 1980s, the sample bottle was

attached to the wire with the ends open and

lowered to the desired depth. This remains the

practice for analyses that require exceptionally

large water samples, for which a rosette sampler

such as that shown in Figure S16.10 is

FIGURE S16.10 Rosette sampler. Large sampler used in

the World Ocean Circulation Experiment, with 36 10-liter

Niskin bottles, an acoustic pinger (lower left), an LADCP

(center, yellow long), a CTD (bottom, horizontal), and

transmissometer (center, yellow short). (Photo courtesy of L.

Talley.)


bottles” permitted operation of the reversing

thermometers described in Section S16.4.2.2. In

other designs, the bottle remains stationary

while a frame carrying the reversing thermometers

rotates 180 degrees. A capacity of 1.25 L was

common for bottles prior to development of the

rosette sampler, which now typically collects up

to 10 L per bottle. For special purposes, such as

39 Ar analyses, or radiocarbon analyses prior to

the 1990s, much larger bottles are used with

up to several hundred liters capacity.

The most commonly used bottle is the Niskin

bottle. It is used with most rosette samplers

(Figure S16.10). These are plastic bottles with

stoppers at each end. The stoppers are held

together by a rubber cord or spring that pulls

them together from inside the bottle. To “cock”

these bottles, lanyards are used to pull the stoppers

away from the bottle, leaving the bottle

wide open for water to flow through. The bottle

is “tripped” by activating a firing mechanism

that releases the lanyard, allowing the stoppers

to close on the bottle trapping the seawater

sample. Niskin bottles can capture a much

larger volume of seawater than the older Nansen

bottles. Reversing thermometers on Niskin

bottles are mounted in a spring-loaded frame

that rotates the thermometers at the same time

that the Niskin bottle stoppers are closed.

The “rosette sampler” (Figure S16.10) is now

the most common arrangement for water bottles.

A single frame usually carries 24 sample bottles,

and might hold up to 36. The frame is attached to

the end of an oceanographic wire that contains

electrical conductors. The bottles can be closed

when desired by electrical command from on

deck. This rosette arrangement is generally

used with a continuously profiling CTD, measuring

pressure, temperature, and conductivity

(Section S16.4.2.3). CTD profiles can be plotted

while the sampler is in the water and can be

used to adjust rosette bottle sampling depths.

Rosette bottles are always open on the way

down and are closed on the way up because

the tremendous pressure at depth would cause

the bottles to leak if they were first closed and

then moved to a greater depth. The sampler is

usually stopped before each rosette bottle is

tripped so that the up and down movement of

the ship can help to flush the bottle out.

Reversing mercury thermometers are no longer

used on rosette samplers since CTD thermistor

accuracy is now higher than the accuracy of the

thermometers.

After sample bottles are brought back to the

ship’s deck, water samples are drawn for immediate

analysis or storage. At this stage, any problems

with sample collection should be noted for

future reference. Bottles suspected of leaking

can be checked to see if the measurements

(e.g., salinity) are consistent with the CTD.

Samples for dissolved oxygen and other dissolved

gases are collected as soon as possible

after the sample bottle is available to avoid

contamination from the air. Samples for other

properties are collected thereafter.

S16.4.2. Temperature Measurement

The concepts of temperature and heat were

discussed in Section 3.3, which also included

brief mention of measurement methods. Typical

distributions were shown in Section 4.2. Here

we describe thermometry methods in much

greater detail. In situ temperature is measured

using thermistors of various accuracies and

precisions. Historically, it was measured using

mercury thermometers. Satellite instruments

measure SST remotely, using radiometry

(Section S16.9).

S16.4.2.1. Sea-Surface Temperature

SST on research ships is measured either from

engine intake water, dedicated intake water

lines, surface seawater samples collected in

buckets or Niskin bottles, or thermistors

mounted in probes such as XBTs (S16.4.2.5) or

CTDs (S16.4.2.3). SST on buoys is measured

using thermistors. Satellites measure SST using

infrared or microwave radiometry (Section

16


S16.9.5). We describe these methods from the

oldest to the most recent.

The oldest method for measuring temperature

is from bucket samples (Section S16.4.1 and

Figure S16.8). Prior to the advent of digital

thermometers (thermistors), an ordinary

mercury-in-glass thermometer was used, taking

care not to expose the bucket to the sun (heating)

or to the evaporating influence of the wind (cooling).

For faster moving powered vessels, special

bucket samplers have smaller openings to reduce

the tension on the bucket support line when collecting

a sample. The thermometer is usually

installed and insulated as part of the bucket.

This type of SST measurement is limited by the

accuracy and readability of the thermometer

along with the ability of the sampling system to

meet the requirements for sample isolation

The change from wooden to canvas bucket

samplers around 1880 to 1890 resulted in an overall

cool bias (drop in the mean SST to a low just

after 1900 from the level between 1850 and 1880;

(Figure S16.11), due to the wind cooling of the

less well insulated canvas buckets on the ship’s

deck (Folland & Parker, 1995). This low bias

continued up through 1940 when the mean SST

again began to rise. Part of the work in computing

the most accurate heat budgets (Section 5.4) was

to correct for this bucket bias. The Hadley Centre

in England has modeled the effects of using

a wooden bucket for the SSTsample versus using

a canvas bucket and found a relationship

between the temperature anomaly and the wind

speed. This made it possible to adjust for this

bias to create temperature records useful for

studying climate variability.

Bucket samples for surface temperature have

been mainly replaced by “injection temperature”

measurements in the engine cooling intake

water. This shift in measurements began in the

1940s and continued through the 1950s. By the

early 1950s, almost all ship SST measurements

were made in this way. Because these measurements

are made in the warm engine room, they

tend to be biased high even though the engine

intake is usually 2 to 5 m below the sea surface.

An upward trend in global SST anomalies after

1940 was partially due to this change in method,

resulting in a need for correction for data sets

used to track climate trends (Folland & Parker,

1995). A separate bias results from the location

of the engine intake below the waterline. An

alternative is to measure the temperature of

the ship’s hull below the waterline (Emery,

Cherkauer, Shannon, & Reynolds, 1997). Since

the ship’s steel hull is a good thermal conductor,

it responds quickly to changes in the

surrounding SST. Due to changes in ship

loading, it is important to install a series of

thermistors in the vertical to ensure that a sensor

is below the ship’s waterline.

SST is also commonly measured on research

ships using a “thermosalinograph,” which

measures the properties of water collected

through a special inlet located on the ship’s

hull somewhere below the sea surface (Figure

S16.12). This intake is usually located as close

to the ship’s bow as possible to collect a sample

with little influence from the ship. The intake is

usually a few meters below the mean waterline,

so it is representative of the “bulk” SST. This

method avoids the engine room heating

problem that plagues the “ship injection”

temperatures. The research-oriented sensors

are also generally more accurate than those

used in engine intake lines and record internally

rather than having to be read by a ship’s officer

(a possible source of SST error). In addition,

thermosalinographs are often integrated with

other underway data collection systems.

SST measurement has been revolutionized

in terms of geographic and temporal coverage

with the advent of satellite remote sensing,

using thermal infrared (IR) sensors (1 km

resolution) and with passive microwave sensors

(25 km resolution, but can observe through

clouds). These instruments are described in

Section S16.9.5. Using satellite SSTs and all available

in situ measurements of bulk SST,

a “blended SST analysis” with 100 km resolution


FIGURE S16.11 Time series from 1856 to 1992 of Northern (a) and Southern (b) Hemisphere anomalies of SSTs from the

1951 to 1980 average. The dashed line is a correction to the canvas bucket SST measurements for the wind cooling,

developed by the British Hadley Centre. Source: From Folland and Parker (1995).

is distributed routinely (Reynolds, 1988;

Reynolds & Smith, 1994, 1995). This SST

product is used for many different applications,

including the initialization and boundary

conditions for ocean, weather forecasting, and

coupled climate numerical models.

Satellite systems measure the temperature of

the very thin (<1 mm) “skin” layer of the

surface of the ocean. This skin layer is the molecular

layer between a turbulent ocean and the

overlying turbulent atmosphere that affects

the heat and momentum exchange between

the two. Unfortunately it is not possible for

drifting buoys and ships to measure this skin

SST. As a result, the blended SST analysis is

a mix of skin and bulk SSTs. These generally

differ by more than 0.3 C, with the skin generally

cooler than the bulk SST. The effects of

diurnal heating strongly influence the relationship

between the skin and bulk SSTs, particularly

under low wind, high solar insolation

conditions. Higher wind speeds decrease the

mean difference between skin and bulk SSTs.

Shipboard and airborne radiometers are being

developed to routinely and accurately measure

the skin SST for satellite sensor validation.

18


FIGURE S16.12

(2009a).

Thermosalinograph: (a) Sea-Bird unit and (b) schematic of its operation. Source: Sea-Bird Electronics, Inc.


Most in situ comparisons with satellite SSTs

use temperatures measured at 0.5 to 1.5 m below

the surface by the buoys and down to 5 m by the

ships, since these are the most available, with

a long historical record and reasonably good

accuracy. Skin SST algorithms are being

improved and the connections between skin

and bulk temperature continue to be explored

with the goal of understanding the connections

between these temperature differences and

wind speed and airesea heat flux. Critical in

this understanding is the development of

methods to assimilate both skin and bulk SST

into ocean and coupled ocean-atmosphere

numerical models.

Aircraft also use radiation methods to

measure SST. In practice the temperature of

the sea is not measured absolutely but is

compared with that of two black bodies, one at

a constant temperature and one allowed to

“float” with the ambient temperature by a parabolic

mirror that rotates to view the sea and two

black bodies. The same principle is used on

shipboard radiometers.

S16.4.2.2. Mercury Reversing

Thermometers

Mercury thermometers were the traditional

method for measuring subsurface temperatures

prior to the 1980s. This method has been almost

completely replaced with digital thermometry

using thermistors. Most present oceanographic

instruments incorporate thermistors, including

vertically profiling instruments, single point

time series instruments, and floating instruments.

For sampling in the most traditional

mode with reversing thermometers, highly

accurate digital reversing thermometers have

been developed. For historical interest, the

description of mercury reversing thermometers

is retained here, since they were the basis of

oceanographic data sets prior to the advent of

CTDs with highly accurate thermistors (Section

S16.4.2.3). Reversing mercury thermometer

precision and accuracy was much lower (0.01

and 0.02 C) than is now possible with highquality

thermistors (0.001 and 0.005 C). However,

lower quality thermistors, such as are

used on many expendable instruments such as

XBTs (Section S16.4.2.5), may also have a low

precision of 0.01 C.

The protected reversing thermometer (Figure

S16.13) was developed especially for oceanographic

use to record temperature at depth

and then remain unchanged while the instrument

was brought back up through the water

column to the ship. It is a mercury-in-glass thermometer,

which is attached to a water-sampling

bottle. It is protected from temperature change

due to ambient water pressure by a surrounding

glass jacket with a vacuum. When the reversing

thermometer rack is flipped during collection of

a water sample, the mercury in the inverted

thermometer “breaks” and runs down to the

other end of the capillary, thus recording the

temperature at the depth of reversal. The break

occurs in the capillary stem above the bulb at

a point where a short side-arm (called the

“pigtail” appendix) is placed. It is really rather

surprising that the mercury should break

consistently d to better than 0.01 K d in

a good thermometer in laboratory tests.

The mercury thermometer is read when

brought back on deck. After corrections for scale

errors and for the small change in reading due

to any difference between the in situ temperature

and that on deck, the reversing thermometer

yields water temperature to an accuracy of

about 0.02 K in routine use. This final correction

is made possible by the presence of an

“auxiliary” thermometer parallel to the

reversing thermometer in the same enclosed

glass housing. The auxiliary thermometer

senses the ambient deck temperature, which is

then used to correct the reversing thermometer

for the temperature on the ship when it is

read. Normal practice is for each thermometer

(reversing and auxiliary) to be read twice by

two different persons using a microscope lens

20


deformations that might take place in the glass

thermometers. An older thermometer with an

accurate calibration history is more valuable

than a new thermometer with no calibration

history.

The most common way to accurately determine

the depth of a sampling bottle prior to

the use of continuously profiling devices with

pressure sensors such as CTDs (Section

S16.4.2.3) was to use an unprotected reversing

thermometer (Figure S16.13) together with the

protected one that recorded the temperature.

The unprotected thermometer differs from the

protected thermometer due to the absence of

a vacuum, thus allowing the ocean pressure to

alter the mercury column height and the

recorded temperature. The reading depends

on the thermometer’s compressibility and the

ambient water pressure when it was reversed.

Use of the protected and unprotected thermometer

measurements together yields the pressure

and hence depth, the latter to about 0.5% or

to 5 m, whichever is greater.

Mercury reversing thermometers have been

almost completely replaced by digital thermometers,

which use calibrated thermistors.

These are usually incorporated in a profiling

device such as a CTD (next section) or XBT

(S16.4.2.5). Reversing digital thermometers are

also available for use with Niskin bottles.

FIGURE S16.13 Protected and unprotected reversing

thermometers. Source: Emery and Thomson (2001).

(Figure S16.14). Reading these thermometers

takes some skill since it is necessary to interpolate

between gradations on the thermometers.

Good practice for reversing thermometers

includes regular calibration and maintenance

of a calibration history. With these calibration

values it is possible to correct for any “creep”

S16.4.2.3. Conductivity, Temperature, and

Depth Profiler

Continuous profiles of temperature (and

salinity) are more desirable than values at

discrete sample bottle depths. Sensor packages

known as STDs (Salinity-Temperature-Depth)

were developed in the 1950s, incorporating

newly developed seawater conductivity sensors

and thermistors. (Salinity calculation requires

concurrent temperature and conductivity

measurement, see Sections 3.4 and S16.4.3.) As

experience developed with processing STD

data, it became apparent that it would be best

to record the seawater conductivity directly


FIGURE S16.14 Reading reversing

thermometers. (Photo courtesy

of W. Emery.)

along with temperature and pressure to permit

post-cruise processing to improve the salinity

accuracy. The device that replaced the STD

by the mid-1970s was known as the CTD

(Conductivity-Temperature-Depth).

The standard CTD sensor includes high precision

thermistors (often two, either for redundancy

or to provide different response times

for processing the conductivity measurements),

a conductivity sensor and a pressure sensor

(Section S16.3), and often an oxygen sensor

(Figure S16.15). The unit is lowered through the

water on the end of an electrical conductor cable

that transmits the information to computers or

recorders onboard ship. The Guildline and Neil

Brown CTDs were the original instruments;

Neil Brown MK3 CTDs were used extensively

in the World Ocean Circulation Experiment

(WOCE) and are no longer produced. Sea-Bird

added a high-capacity pumping system to flush

the conductivity cell at a known rate, which

improves sensor response.

Internally recording CTDs eliminate the

complex infrastructure of having a conducting

wire to transfer the signal from the CTD to the

ship. This type of unit can be used with a simple

support cable or on a mooring. Upon return to

the surface, this CTD is plugged into a computer

and these data are downloaded for processing

and display. Internally recording CTDs are

used on all profiling floats, such as in the Argo

program (Section S16.5.2), and can be mounted

on moorings.

For highly accurate measurements, CTD

sensors, including thermistors and pressure

transducers, must be calibrated. Prior to the

1990s, thermistor calibration was accomplished

by adjusting laboratory calibrations of the

sensors with estimates of shifts and drifts monitored

at sea via reversing thermometers on the

rosette bottles that usually accompanied

a CTD profile. Since the early 1990s, improvements

in the stability of calibrated, precision

CTD thermistors has superseded the use of

reversing thermometers for calibration. This

has shifted best practice to using CTD thermistors

in redundant pairs, calibrated pre- and

post-cruise in the laboratory. This provides

both detection and correction for sensor drift

and now-rare sudden offsets. Pressure transducer

calibration is entirely accomplished in

specially equipped calibration facilities. Salinity

and oxygen calibration are more complicated

because these sensors are not stable; water

22


(b)

(a)

FIGURE S16.15 (a) Neil-Brown Mark III CTD. Source: From General Oceanics (2009). (b) Sea-Bird 911plus CTD. Source:

Sea-Bird Electronics, Inc. (2009b).

sample values are required for the highest

accuracy.

Modern CTD accuracy in ocean deployments

is approximately 0.001 C in temperature,

0.001 psu in salinity if routinely calibrated

with seawater sample salinity (in practice, on

every station), and 0.5 db in pressure.

S16.4.2.4. Mechanical Bathythermograph

Vertical profiles of temperature alone have

been collected from research and merchant

ships since the early 1950s. The first profiling

instrument in wide use from approximately

1951 to 1975, prior to the widespread use of

thermistors, was the mechanical bathythermograph

(MBT; Figure S16.16). Given the large

number of MBT temperature profiles in the

historical data records, it is important to

describe this instrument and its limitations. In

addition to providing upper layer temperature

profiles, the MBT could be operated while

a ship was underway. In the MBT, a liquid-inmetal

thermometer caused a metal point to

move in one direction over a smoked or goldplated

glass slide that was moved at right angles

to this direction by a pressure-sensitive bellows.

The instrument was lowered to its permitted

limit in the water (60, 140, or 270 m) and then


FIGURE S16.16 Mechanical bathythermograph (MBT), in use from 1951 to 1975. Source: Neumann and Pierson (1966).

brought back using a very fast electric winch.

Since pressure is directly related to depth, the

line scratched on the slide formed a graph of

temperature against depth. It was read against

a calibration grid to an accuracy of 0.2 K

(0.1 F) and 2 m if well calibrated. Since each

instrument had a nonlinear relationship

between the sensors and temperature, each

instrument was coupled with its own reader to

convert the scribed profiles to a temperature

profile.

The MBT had a “torpedo-like” profile

intended to provide minimal water resistance

as the probe was hauled in to the ship. Unfortunately

the MBT was hydrodynamically unstable

and would “swim” on its return, often

announcing its presence by a thud against the

hull of the ship. The high-speed electric winch

had a “neutral position” between up and brake

that would cause the probe to fall back into the

water. As a consequence most MBT probes were

hauled up by hand for the last few meters.

Most MBTs were in units of F rather than C

and in feet rather than meters, since the U.S.

Navy developed them. A sample temperature

profile is shown in Figure S16.17. Note the

nonlinear temperature and depth scales. Each

temperature profile had to be read by eye,

resulting in numerous data transcription errors.

S16.4.2.5. Expendable Bathythermograph

and Expendable CTD

The MBT was replaced by the expendable

bathythermograph (XBT; Figure S16.18), which

was introduced in 1966 and remains in wide

use, especially for profiling from volunteer

24


FIGURE S16.17 MBT temperature profile. Source: Neumann and Pierson (1966).

observing ships. The XBT has a thermistor and

electronic data acquisition system. A conductivity

sensor is included in the XCTD, which is

otherwise essentially the same as an XBT. XBT

profiles are deeper (400, 800, or 1500 m) than

MBT profiles. They can be launched from ships

moving at up to 30 knots. The XBT has a thin,

two-element wire insulated by clear resin. This

dual element wire pays out from a spool on

the probe as the probe falls, transferring the

temperature signal from a thermistor installed

on the head of the XBT probe back to the ship.

At the same time a spool of the same wire on

the ship pays out as the ship travels, thus

mechanically disconnecting the probe from the

ship while still retaining the electrical connection.

When all of the wire is out, the wire breaks

and the XBT is lost (hence “expendable”).

The probe is assumed to fall at a known and

constant rate thus making it possible to infer

the depth from the time the probe hit the sea

surface. Early versions of the XBT deck unit

recorded the temperature profiles on pressuresensitive

paper that rolled forward at the

assumed rate of fall for the XBT. More recent

systems are digital, but an assumed fall rate is

again used to estimate depth. The fall rate is

determined by a limited number of measurements

in a tower that is 250 feet high and then

extrapolated to the full XBT depth. This is known

to introduce some error in depth. In addition,

changes in the density of the probe as the wire

pays out alter the fall rate and introduce depth

errors. The present error in XBT depth, using

the most recent fall rates, is approximately 20%.

XBT probes are usually not individually calibrated.

Instead, a lot of about 2000 thermistors is

purchased by the manufacturer and about 250

of these are “calibrated” for accuracy. The accuracy

reported for the group of 250 thermistors is

then assigned to all of the 2000 thermistors.

Thermistors that do not meet the stated accuracy

of the XBT probes are discarded.

XBTs and XCTDs are launched from portable

or fixed launchers (Figure S16.19). For occasional

use, the portable unit is useful and


FIGURE S16.18 An expendable bathythermograph (XBT). Source: From NOAA UOTC (2009).

flexible. For volunteer observing ships operating

XBTs every hour, fixed launchers holding

multiple probes have been developed.

XBT’s have been developed for platforms

other than moving ships. One type can be

deployed from a submarine, with the buoy

floating up to the surface and then dropping the

probe. XBTs can also be dropped from aircraft

(AXBT). The AXBT deploys a small surface

buoy, which contains a radio transmitter to

send the temperature/depth information (from

300 to 800 m) to the aircraft, which continues its

flight. AXBT probes are usually dropped from

altitudes between 300 and 1000 m, but testing

has shown that they can survive the impact

of being dropped from 6000 m. AXBT designs

from different manufacturers vary, but all have

some type of atmospheric drag element (parachute,

wings, etc.) to slow their descent and

soften the impact with the sea. The combined

26


(a)

(b)

FIGURE S16.19 (a) Portable (hand-held) XBT launcher. (Photo courtesy of Valerie Cannon.) (b) XBT autolauncher developed

for multiple probes by Scripps Institution of Oceanography. (Photo courtesy of G. Pezzoli and D. Roemmich.)

XBT probe and radio transmitter makes the cost

per AXBT much higher than that of the normal

ship XBT, but the relatively lower cost per area

surveyed of using an aircraft often offsets the

higher cost per probe. Also, if near-simultaneity

is a requirement of the sampling program,

AXBT observations are one of only a very few

methods of carrying out this type of sampling.

Expendable instruments have provided the

oceanographer with simple tools for rapid

sampling. This has proved important for

synoptic sampling from multi-ship or aircraft

surveys and has led to wider use of ships of

opportunity. In an effort to extend such technology

to other important parameters, developments

in the 1980s produced an expendable

velocimeter (speed of sound) and an expendable

current profiler, using the electromagnetic

principle. These more exotic expendables are

considerably more expensive than the XBT and

are less widely used.

S16.4.2.6. Subsurface Temperature

Measurements from Floating and Moored

Instruments

Valuable temperature records with long

temporal coverage are collected on subsurface

instruments. Thermistors are robust, and can

maintain an acceptable calibration for years.

Most moored instruments (Sections S16.5.4

and S16.5.5) employ thermistors. Thermistors

spaced closely together in the vertical are sometimes

deployed as “thermistor chains” to obtain

dense vertical coverage, particularly useful for

studies of internal waves and smaller scale

phenomena. Most surface drifters and subsurface

floats also have thermistors, providing valuable

records that complement or sometimes


supersede the usefulness of the actual velocity

measurement.

Freely drifting subsurface floats are becoming

routine vehicles for temperature and salinity

profiling. Profiling floats are described in Section

S16.5.2. The “pop-up” type of float typically

includes a thermistor, and preferably also

a conductivity sensor. Every time the float

surfaces, it transmits a profile of temperature

(and conductivity if available) to a communications

satellite, which then relays the information

to a data acquisition center. A global deployment

of such profiling floats is now in progress (Argo).

Argo has reached a full global sampling capacity

of 3000 floats, providing profiles to 1800 m depth

every 10 days. This network is replacing much of

the current XBTship of opportunity sampling for

temperature, since volunteer ships do not cross

many areas of the ocean.

S16.4.3. Salinity Measurement

As already described in Section 3.4, salinity is

presently determined from conductivity, relative

to the conductivity of an international standard

seawater prepared in the UK. Prior to the widespread

use of conductivity methods beginning

in the 1960s, salinities were calculated by titration

with much lower precision and accuracy

than is achievable with conductivity methods.

Salinity is measured both on seawater samples

collected from bottles such as on a rosette

sampler, and through paired conductivity and

temperature sensors deployed in the water.

Conductivity sensors in CTDs are relatively

unstable and usually require frequent calibration.

Therefore highly accurate salinity observation

(especially in the open ocean) requires

calibration with measurements on seawater

samples in the laboratory. On research cruises,

salinity samples are analyzed onboard within

a day of sample collection to minimize evaporation

from the sample. However, the stability of

conductivity sensors is improving steadily to

the point where sensors can be moored and

measured for about a year, or can be deployed

on drifting floats, and produce reasonable

salinity values. All profiling floats now deployed

globally as part of the Argo array (Section

S16.5.2) include internally recording CTDs to

provide temperature and salinity profiles.

When measuring salinity from seawater

samples, water from sampling bottles is drawn

into 200 ml glass bottles (Figure S16.20) after

several “rinses” done with a minimal amount

of water from the sample bottle. These rinses

remove residue from earlier salinity samples.

The subsamples are then carefully capped

and left in the wet lab to reach the equilibrium

temperature of the lab, which can take about

12 hours, since conductivity is foremost a function

of temperature and secondarily of salinity.

The salinity samples are then processed with

a laboratory salinometer that measures the

conductivity of each sample in comparison

with a carefully prepared standard. The

conductivity and temperature of the lab

sample are then used to calculate the salinity

of the sample.

S16.4.3.1. Salinity Measurements Using

Titration

As described in Section 3.4, the classical

(Knudsen) method of salinity measurement, in

general use prior to about 1960, determined

the chlorinity by titration with standard silver

nitrate solution (Strickland & Parsons, 1972)

and calculated salinity from the formula (3.3).

In routine use, an accuracy of 0.02 is considered

reasonable, with rather better accuracy

with special care and replicate titrations. A careful

operator could titrate 50 samples per day.

This method was volumetric, whereas salinity

is defined gravimetrically (i.e., by mass). As

a consequence, it was necessary either to correct

for deviations of the temperature of the solutions

from the standard, or preferably to carry

out the titrations in a temperature-controlled

room. This titration method was not very convenient

to use onboard ship. It is also less precise

28


FIGURE S16.20

Drawing a salinity sample from a Nansen bottle. (Photo courtesy of W. Emery.)

than electronic methods, which are based on the

relationship between salinity and conductivity.


Conductivity

Salinity has been estimated through its relation

to electrical conductivity since about 1930,

when the U.S. Coast Guard introduced the

measurement for the International Ice Patrol in

the western North Atlantic. The method was

not widely used for many years because of the

bulk and expense of the equipment required.

This is because the conductivity is as much

a function of temperature as of salinity, which

necessitates stabilizing the temperature of the

samples to 0.001 C during measurement.

However, improvements in circuits and equipment

encouraged a number of laboratories to

bring this method into wider use from about

1956. An accuracy of 0.003 psu or better is

obtained in routine use. This is substantially

better than the titration method.

In 1957, Esterson (1957) of the Chesapeake

Bay Institute described an inductive (electrodeless)

salinometer, which was later developed by

Hamon (1955) and HamonandBrown(1958).

Hamon and Brown’s inductive salinometer

design was the basis for modern inductive salinometers.

In this instrument, the temperature

effect is taken care of by measuring the


temperature while the conductivity is

measured and correcting for its effect automatically

in the electrical circuit. The salinity may

be measured to a precision of 0.001 psu over

the range from 32 to 39 psu. With a little practice,

an operator can measure the salinity of

up to 45 samples per hour. The absolute accuracy

of the measurement depends on the accuracy

of the standard seawater (Section

S16.4.3.4). Since the 1990s, this accuracy has

been approximately 0.001 psu.

Conductive salinometers such as the

“Autosal” from Guildline (Figure S16.21) are

more accurate than inductive salinometers and

have virtually replaced them. This salinometer

uses a four-electrode conductance cell of small

dimensions in a thermostat bath (controlled to

0.001 K/day) with a precision of 0.001 K or

better. The seawater flows continuously from

the sample bottle through a heat exchanger in

the thermostat, to bring it to a specified temperature,

and then through the cell. The conductancebridge

is balanced semi-automatically and the

conductivity ratio of the sample relative to that

of Standard Seawater (see Section 16.4.3.4) is displayed

digitally. Salinity is then obtained from

the conductivity ratio and the temperature using

the UNESCO/N.I.O International Oceanographic

Tables or the Practical Salinity Scale 1978

Formula or Tables referred to in Section 3.4

(UNESCO, 1981). The circuits are such that

variations of electrode surface conditions do

not affect the measurement. The size of the

instrument is about 60 50 55 cm and it may

FIGURE S16.21

(2009).

Autosal inductive salinometer in common use in laboratories for salinity analyses. Source: From Guildline

30


be used on shipboard as well as in a shore

laboratory.

In situ conductivity measurements are made

by CTDs (and their predecessors the STDs)

and on other subsurface devices where salinity

observations are desired, such as moorings

and floats. Conductivity sensors are far less

stable than thermistors, primarily because they

are open cells, often with ceramic coatings,

and changes in the geometry of the cells affect

the calibration. Therefore, relatively frequent

calibration of these sensors using water samples

is required for high accuracy.

To obtain salinity measurements from CTDs

and similar instruments, temperature must be

measured simultaneously with conductivity,

since conductivity depends primarily on

temperature and only secondarily on salinity.

The conductivity and temperature are then

combined during processing that takes in

account the time lag in sensor response. Different

sensors usually have different response times to

changes in temperature; a conductivity cell

responds faster than the high precision thermistors.

Therefore, when processing CTD data, it is

important to account for the sensor response

time mismatch. Erroneous spiking in the derived

salinity is usually a result of this mismatch. Some

CTDs overcome the temperature response time

by combining a fast but less accurate thermistor

with a slower but more accurate precision resistance

thermometer (PRT) to yield a rapid and

accurate temperature estimate at each level in

the vertical.


Refractive Index

The refractive index of seawater is also related

to salinity (and to temperature). The interference

type of refractometer has been used in the past

for salinity measurements with a claimed accuracy

of 0.02 psu. A refractometer that can be

installed in a profiling instrument to measure

salinity in situ rather than in a laboratory setting

has been developed.

S16.4.3.4. Standard Seawater

All of the above conductivity methods are

comparative rather than absolute, and require

a chemical standard for calibration. The current

standard is set by the international body, IAPSO,

with documentation published by UNESCO

(1981). Practical Salinity is defined by the ratio

of the electrical conductivity of the seawater

sample to that of a standard potassium chloride

(KCl) solution, at 15 C and 1 atmosphere

(Section 3.4; Lewis, 1980). The standard solution

is known as Standard Seawater (SSW), since it

actually was seawater collected from a given

location near Copenhagen for many years.

Defined now as a KCl solution, salinity calibration

is much more robust and stable. Oceanographic

laboratories throughout the world use

samples of SSW, sealed in glass ampoules, to

standardize electrical conductivity salinometers.

The use of a common standard for salinity

reduces the possibility of systematic errors

making it possible to combine data from

different expeditions or surveys in the same

area or worldwide.

S16.4.4. Density Measurement

Standard laboratory methods to determine

density directly are not practical at sea because

of the motion of the ship, and are far too slow

for routine use on shore. Thus density is calculated

indirectly from salinity, temperature, and

pressure using the equation of state. The most

widely used modification to the internationally

recognized equation of state was made in 1980,

and is referred to as EOS 80. The method for

determining this equation of state was

described in Section 3.5, along with the newly

developed equation of state, TEOS-10, which is

replacing EOS 80.

S16.4.5. Other Water Properties

Many properties of seawater are measured

in addition to temperature and salinity.


Nutrient concentrations (nitrate, phosphate,

silicic acid, nitrite, and sometimes ammonium)

are routinely measured using relatively small

samples drawn from the rosette sampler with

a simple rinse to remove residue from previous

analyses. Samples are run using auto-analyzers

that mix the samples with chemical reagents.

This produces a colored compound whose

light absorbance is proportional to the nutrient

concentration being measured. The absorbance

is measured with a colorimeter. Separate analyses

are run for each nutrient.

Oxygen content is even more routinely

measured from water samples, and increasingly

from oxygen sensors incorporated in a CTD.

Sampling of dissolved gases such as oxygen

requires care to avoid contamination from the

atmosphere. The standard method for oxygen

is to rinse nominal 125 ml volume-calibrated

flasks with a cone-shaped neck once with

minimal agitation followed by a 10 second

inverted rinse (to minimize the amount of air/

oxygen getting in) with laminar flow from the

sample drawing tube (Figure S16.22), allowing

the sample to overflow for at least three flask

volumes. Reagents are then added to fix the

oxygen before stoppering the flask. Flasks are

shaken immediately after drawing, and then

again after 20 minutes to assure the dispersion

of the MnO(OH) 2 precipitate. These samples

are then analyzed for oxygen content within 4

to 36 hours of their collection. This processing

uses a Winkler titration with a colorimetric

end point to find the oxygen content of the

sample. Automated titration rigs are commonly

used to reduce the variability in end point detection

that can occur between human operators.

CFC analyses were introduced in the 1980s

and have become routine for large oceanographic

expeditions such as those in the

WOCE. Because extreme care must be taken to

avoid contamination, not just from the atmosphere

but also from shipboard sources, the

ships and laboratories must be scrupulously

free of refrigerants and propellants that contain

CFCs. Samples are drawn from the rosette

sample bottles into large syringes, which are

filled, ejected, and then refilled. Samples are

run in the shipboard laboratory using a gas

chromatograph.

Dissolved helium sampling has also become

widespread. These samples are drawn into

FIGURE S16.22

Drawing an O 2 sample from a Niskin bottle. (Photo courtesy of J. Swift.)

32


sample containers, which might be special,

nearly impermeable glass flasks or narrow

copper tubes, and overfilled to eject any air or

air bubbles. The samples are sealed tightly

and then taken to the laboratory for later

analysis.

In addition to the dissolved gases, nutrients,

and salinity, a number of other types of analyses

are regularly run. These include dissolved

inorganic carbon, pH, alkalinity, tritium, and

isotopes of carbon, nitrogen, and oxygen. These

require clean samples of seawater, but without

the extreme care to exclude air from the

samples.

In Section S16.5.3 as well as in Chapter 4 and

throughout the ocean basin chapters, the use of

some of these chemicals as tracers of circulation

and mixing is discussed.

S16.5. CURRENT MEASUREMENTS

There are two basic ways to describe fluid

flow: the Eulerian method in which the velocity

(i.e., speed and direction) is given or observed

at every point in the fluid, and the Lagrangian

method in which the path followed by each fluid

particle is given or observed as a function of time

(Section 7.2). Both approaches are used to map

ocean currents and it is possible to connect the

two methods using some approximations.

Typical horizontal current speeds in the ocean

range from about 200 cm/sec (about 200 km/day

or about 2 knots) in the swift western boundary

currents (Gulf Stream, Kuroshio), in the

Antarctic Circumpolar Current, and in the upper

ocean equatorial currents to a fraction of 1 cm/

sec in much of the surface layer and in the deep

waters. Vertical speeds associated with the

large-scale circulation are very much less, on

the order of 10 5 cm/sec or 1 cm/day; these are

essentially unmeasurable except with extremely

good instruments and data filtering. On the other

hand, vertical speeds associated with surface

and internal waves and tides are easily

measured as they are of the same order as

horizontal speeds for the same phenomena.

S16.5.1. Lagrangian Methods for

Surface Currents

The simplest Lagrangian current indicator is

an object floating in the water, carried by the

ocean current with a minimum of surface

exposed to the wind, or below the surface of

the water. The so-called drift pole, a wooden

pole a few meters long (also called a spar

buoy) and weighted to float with only 0.5 to

1 m emergent, was historically used to determine

surface currents close to landmarks

(where the movement of the pole can be

measured relative to the landmarks). Such

a pole was simply allowed to drift with the

water, its position determined at intervals either

from the shore or by approaching it in a boat

and fixing its position relative to the shore.

Sheets of paper or patches of dye, such as

sodium fluorescein, which can be photographed

at intervals from a high point of land or from an

aircraft, have also been used. Glass drift bottles

of about 10 cm length, with small cards inside to

be returned to the agency that deployed them,

were deployed in large numbers prior to the

1930s to map surface currents. Other near-shore

drifters have been built as wooden crosses or

even “hula hoops” covered with plastic garbage

bags. The latter have been found to closely

simulate oil being carried along by currents on

the surface of the ocean.

Historically, surface currents were mapped

using information about how much a ship

drifted from its intended course due to surface

currents. This information is a by-product of

sailing ship navigation. A comparison of the

actual track (checked by astronomical navigation,

landfall, etc.) with the intended track

gives a measure of the surface current every

time a positional fix is obtained from the

ship. Maury, in about 1853, first suggested

examining ships’ navigation logs to extract

CURRENT MEASUREMENTS 33

such information on currents. In his case he

first examined the logs for the Gulf Stream

region off the eastern United States. The

method was subsequently extended worldwide.

Most of the maps of surface currents

presented in marine atlases are based on the

accumulation of such ship-drift data. The

modern version of the Maury ship-drift

surface current compilation is Mariano, Ryan,

Perkins, and Smithers (1995), from which the

Gulf Stream map in Figure 1.1b was obtained.

The error is +10 cm/sec.

Modern measurements of surface currents

are made with freely drifting buoys (or surface

drifters) with a radio transmitter for satellite

tracking. Thousands of tracked surface drifters

have been used since the 1970s (Figure S16.23),

most with a lifetime of at least a year. There

are three items to consider for the instrument:

the surface buoy, the drogue, and the tracking

system.

Surface buoys were initially quite large and

made of such hardy materials as aluminum or

fiberglass. Smaller buoys were then developed

to reduce the “windage” or extent to which

the drifter was pushed by the wind versus

following the current. Thus, the floats became

small glass spheres or plastic platters (Figure

S16.24). Another popular alternative was

a very small spar buoy that was ballasted to

ride with most of the buoy below the surface.

As mentioned previously in Section S16.4.2 on

temperature, most surface drifters also include

a thermistor on the buoy to measure surface

temperature. Many, particularly in very remote

regions rarely visited by research ships, carry

other instruments used for meteorological

models and weather prediction, such as air pressure

sensors.

The surface drifter’s drogue is attached

beneath the buoy with a sturdy line. Usually

the connection at the bottom of the buoy float

is some form of hydraulic hose capable of flexing

a great number of times without breaking.

The drogue acts like a parachute, and the

buoy moves at the speed of the water at the

drogue depth. Most surface drifters deployed

in the 1990s and 2000s were drogued between

15 and 100 m depth and had lifetimes of 1 to 2

years. Measurements of buoy slip through the

water have shown that the “holey sock”

drogues follow the water better than any of

the other drogue configurations. These drogues

are wire cylinders covered with fabric with

several holes (Figure S16.24). Measurements

also showed that a larger diameter cylinder is

better than a longer cylinder of smaller

diameter.

Modern surface drifters report their positions

and data via satellite. Initially all

communication and location was accomplished

by the Argos system, carried on the

National Oceanic and Atmospheric Administration

(NOAA) polar orbiter satellite. This

communications systems predates the global

subsurface float program Argo (Section

S16.5.2), and should not be confused with it.

AFrenchsystemwithanAmericansubsidiary,

Argos is able to communicate with a large

number of transmitters at the same time.

From the Doppler shift of the radio signal,

the satellite is able to compute a fairly accurate

position (2 km) for the buoy’s location. In

addition, Argos is capable of transferring 250

data words, which is enough for the SST

samples and barometric pressure if available.

With the advent of the GPS satellite navigation

system, it is possible to have the buoy calculate

its position independently using GPS. Modern

communications satellites have much wider

bandwidths than available through Argos,

allowing transmission of much more detailed

data streams. However, many observational

programs such as the Global Drifter Program

continue to use Argos for continuity and

reliability.

In the United States, surface drifter data are

collected and processed by the Global Drifter

Data Center at NOAA Atlantic Oceanographic

and Meteorological Laboratory in Miami,

34


FIGURE S16.23 Starting points of surface drifter tracks from 1991 to 1997. Source: From WOCE (2009). (b) Number of

years of drifter data per year since 1988. Source: From NOAA Global Drifter Program (2009).

which currently serves as the international

data center for drifting buoy data. Another

global drifter data center is located in Ottawa,

Canada, at the Marine Environmental Data

Service.

S16.5.2. Lagrangian Methods for

Subsurface Currents

Subsurface currents are also observed using

Lagrangian instruments that follow the water,


(a)

(b)

FIGURE S16.24 Lagrangian free-drifting buoy elements: (a) drifter ready to deploy with buoy, coiled wire, and collapsed

holey sock drogue. Source: From NOAA Global Drifter Program (2009). (b) Holey sock drogue (blue cylinder) being recovered.

(Photo courtesy of K. Buesseler.) Source: From WHOI Image Galleries (2009).

at least approximately. The first category of

subsurface floats that we consider are acoustically

tracked. A great advantage of acoustically

tracked floats is that they can be followed continuously,

hence obtaining information at eddy

scales, which is not the case for the pop-up floats

described in the next section. John Swallow of the

National Institute of Oceanography in England

developed the first float in the 1950s (Swallow,

1955). The “Swallow float” (Figure S16.25a) and

its modern derivatives are neutrally buoyant,

which means that the float’s mass is adjusted

before launching so that it will sink to a selected

density. Actual seawater density is primarily

a function of pressure, hence depth, since the

compressibility of seawater causes a larger

density range than either temperature or salinity

(Section 3.5). The float then remains at this depth

and drifts with the water around it. The Swallow

float sent out sound pulses at intervals, which

was followed by listening to it through hydrophones

from the ship that chased the float and

simultaneously determined its own position. In

doing this, the direction and speed of drift of

the float was determined.

Subsequently, floats were developed that

could be tracked by moored instruments or by

satellite, removing the need to chase them, permitting

long deployments. A research ship is needed

to deploy and retrieve the moorings and to deploy

specialized floats. The first development, by

Rossby and Webb (1970), was the “SOFAR float”

(Figure S16.25b), which emitted a signal that was

picked up by at least three moored hydrophones,

so that the position of the float could be

continuously monitored through triangulation.

The word SOFAR refers to the main sound

channel in the ocean, which is at the depth of the

minimum in sound velocity (Section 3.7).

A reversed system called RAFOS (SOFAR

spelled backwards) was developed in the

1980s (Rossby, Dorson, & Fontaine, 1986), in

which the buoy is a simple listening device

and the moored stations are low frequency

acoustic sources (Figure S16.25c). This greatly

reduces the cost of the (much smaller)

36


expendable floats and puts the higher cost into

the moorings that are retrieved and reused.

The RAFOS float is much smaller than the

SOFAR float since the sound source is a very

long tube that resonates like an organ pipe

(Figure S16.25c). The RAFOS system is currently

the basis for all acoustically tracked subsurface

floats. At intervals, the buoy comes to the

surface, reports its data over the same satellite

system used to track surface drifters, and then

returns to its pre-selected depth to collect more

information. (Often this interval is just the

beginning and end of the experiment, so it

might take several years to obtain the data.)

RAFOS and SOFAR systems are restricted to

the region of the ocean that is insonified for

the experiments, which is typically no larger

than about 1000 km radius because of the range

of the tracking. RAFOS deployments in the

North Atlantic at various depths have been

used extensively to map the circulation and its

eddy statistics (Figure S16.26).

Other acoustically tracked floats that have

been developed and deployed in special experiments,

primarily in the North Atlantic, include

floats that continuously re-ballast themselves to

follow an isothermal or isopycnal surface (isopycnal

floats; Figure S16.26) and floats that

move up and down to sample a layer in the

ocean, usually to profile temperature and

salinity in that layer (“bobber floats”).

A lower cost alternative to acoustically

tracked floats, with the additional advantage

of permitting coverage everywhere in the (icefree)

ocean because they do not require moorings,

was developed by R. Davis and D. Webb

in the 1990s (Davis, Killworth, & Blundell,

1996) as part of the WOCE. These pop-up floats,

more commonly called “profiling floats” (Figure

S16.27), are neutrally ballasted for a preassigned

depth, and are tracked only by coming

up to the surface at regular, pre-assigned intervals

and transmitting to a satellite. The satellite

then records the float position and also any

temperature or salinity or other observations

FIGURE S16.25 Acoustically tracked floats: (a) Swallow

float, (b) SOFAR float, and (c) RAFOS float. Source: From

University of Rhode Island Graduate School of Oceanography

(2009). See also Rossby (2007).

that are profiled by the float (Section S16.4.2.6).

After reporting to the satellite and perhaps

remaining at the surface for half a day or more


FIGURE S16.25

(Continued).

to ensure the satellite transmission, the floats

return to their original depth. Because these

floats are only discretely tracked when they

pop to the surface, say every 10 days, the

velocity field is more coarsely resolved than

from acoustically tracked floats. The latter therefore

are preferable for experiments requiring

resolution of the ocean’s eddies, while the

pop-up floats are useful for studying larger scale

circulation with its longer timescales. They are

38


FIGURE S16.26 RAFOS floats on the isopycnal surface

27.2 s q in the North Atlantic for 1993 through 1995: (a)

release locations and (b) tracks. Source: From Zhang, Prater,

and Rossby. (2001).

also proving to be extremely useful for

providing global temperature/salinity profiling

when they pop to the surface.

The profiling floats ascend and descend by

using oil in a small cavity to control buoyancy.

When the float ascends, the oil is pumped into

an external bladder. When it descends, the oil

is pumped back inside the solid body of the

float. The floats are carefully ballasted for the

desired “parking depth”; for 20 m accuracy,

the ballast must be accurate to within a gram

compared with the much greater weight of the

float.

Profiling floats commonly carry a thermistor

and often a conductivity sensor so that temperature

and salinity can be vertically profiled and

measured during the full-submersed track if

desired. Additional sensors continue to be

desired; a popular add-on is an oxygen sensor.

The floats can be “parked” at one depth, say

1000 m, if this is the depth where the velocity

field is desired, and then profile down to say

2000 m just before popping to the surface to

provide a deeper profile. This vertical transit is

now used to measure oceanographic profiles,

which are then reported via satellite giving an

autonomous ocean profiling system.

Large-scale deployments of profiling floats

were made in the 1990s at 1000 m depth. A global

networkcalledArgo(nottobeconfusedwiththe

satellite communication service Argos) consisting

of the most recent designs of these floats is

now underway (Figure S16.27c) as part of the

long-term global ocean observing system. It is

intended that this kind of observational system

will remain in place for the foreseeable future

as a means of regularly mapping the temperature,

salinity, and velocity structure of the upper

half of the ocean. In the recent Argo program, the

floats are parked at 1000 m for 10 days, then sink

to 2000 m before they rise to the surface, collecting

temperature and salinity profiles along the

way. These floats carry GPS units for precise geolocation

while they are at the surface. The position

and profile data are then relayed via satellite.

Another advantage of the Argo floats is that

they can be deployed by a variety of methods.

Traditional deployments from research vessels


FIGURE S16.27 (a) Schematic of an Argo float. (b) The 3000th Argo float being deployed in July 2007. (Photo courtesy of

Kara Lavender.) Source: From Argo 3000 (2007). (c) Argo float profile locations in February, 2009. Source: From U.S. Argo

Program (2009) for a and c.

40


FIGURE S16.28 Operation of an Argo float in “park and profile” mode. Source: From U.S. Argo Program (2009).

can be augmented by deployments from

merchant vessels of opportunity and from

aircraft. This capability is particularly important

for regions such as the Southern Ocean where

visits by research vessels and merchant ships

are infrequent.

The operation of an Argo float is summarized

in Figure S16.28, which demonstrates the 10-day

repetition cycle, including its hours at the

surface, descent, 9 days at its “parking” depth,

and ascent. Not included in the diagram is the

possibility of parking at one depth for most of

the nine days and profiling over a larger range

of the water column.

S16.5.3. Lagrangian Methods

Employing Tracers

While they are not direct measurements of

current, we discuss flow tracers here, since

they provide direct evidence of some average

of the circulation and diffusion. Tracers are

useful if they have known regional or temporal

sources, such as intentional dyes or an actual

pollutant like sewage or industrial waste.

Samples of water are collected from a grid of

positions near the source and in likely directions

of flow. The tracer concentration is determined

by chemical analysis. In all tracer release studies

both advection and eddy diffusion are acting

three dimensionally to spread the tracer, thus

results cannot be interpreted solely in terms of

advection by currents.

Intentional tracers that have been used are

red dye rhodamine-B and sulfur hexafluoride

(SF 6 ). Rhodamine-B can be detected at

extremely small concentrations (less than 1

part in 10 10 of water) by its fluorescence, using

relatively simple instruments, and it is also

non-toxic at such dilutions. It is only of practical

use in coastal waters. SF 6 can be detected at

much lower levels and has been used in the

open ocean to study stirring and mixing of ocean

waters over periods of up to a year (Figure


S16.29). Eddy diffusivities in the horizontal and

vertical directions (Sections 5.1 and 7.3.2) have

been derived from this intentional tracer.

Materials released unintentionally in small

amounts for reasons other than oceanographic

research have been exploited as artificial tracers

of water movement. These materials are known

as “transient tracers” (Section 3.6). Primary

examples of radioactive or unstable materials

are tritium (decaying to 3 He) and D 14 C (radioactive)

released into the Northern Hemisphere

atmosphere in the early 1960s during atomic

bomb testing in the Pacific, and iodine ( 129 I)

(radioactive) released from a nuclear plant in

western England. D 14 C also occurs naturally

and is useful for relative dating of the ocean’s

deep waters. However, its anthropogenic

concentrations are much larger than its natural

concentrations, making D 14 C useful for

following surface waters into the subsurface

ocean. Primary examples of stable materials

created for industrial use are the CFCs, which

have been used as cleaning compounds, refrigerants,

and propellants. With known production

rates since their invention and introduction in

the twentieth century, and recent curtailment

because of their major impact on Earth’s ozone

layer, CFCs have been useful for determining

ventilation pathways. Because these tracers

have temporal source functions that are well

understood, they can also be used for bulk dating

of the tagged subsurface waters (Section 4.7).

Tritium, enhanced levels of D 14 C, and CFCs

have been traced through all the upper ocean

waters of the world. In the Pacific example of

Figure 4.25b, tritium is higher in the Northern

Hemisphere, demonstrating the predominantly

Northern Hemisphere atmospheric source

(primarily in the 1960s). Low values in the

Antarctic result from upwelling of deep waters.

Penetration of the purely anthropogenic tracers

to the ocean bottom in the Antarctic and North

Atlantic shows that these are regions of deep

water formation. 129 I has been traced through

much of the northern North Atlantic, following

FIGURE S16.29 Horizontal region covered by an

intentional release of sulfur hexafluoride after release at

a point in the subtropical North Atlantic. Source: From

Ledwell, Watson, and Law (1993).

the subpolar circulation. The concentration of

these released tracers is extremely low. They

are useful because there is either no natural

source or the natural source creates a much

lower concentration than the anthropogenic

source.

S16.5.4. Eulerian Current

Measurement: Mechanical Sensors

Current meters are deployed at a fixed location

and record the current speed and direction

over time. The instruments described in this

section have mechanical moving parts, and are

more subject to fouling than the acoustic and

electromagnetic current meters described in

Section S16.5.5.1. As a result, acoustic methods

have largely superseded mechanical sensors,

but we describe the most common mechanical

current meters because of the abundance of

historical data records.

All current meters have a sensor for speed,

a sensor for direction, and ideally a sensor for

42


pressure to detect mooring deflections in the

vertical. Ideal speed sensors have low inertia.

Compasses are used for direction, and must be

extremely well calibrated especially if used

where the horizontal component of Earth’s

magnetic field is small. The technology for

mooring and recovering the instrument is also

important. Most current meter systems record

internally and are retrieved by a research vessel

to recover the data, although some modern

moorings are equipped to transmit their data

via communications satellite.

Before 1960, the most widely used Eulerian

instrument was the Ekman current meter

(shown in a unique “repeating” form in Figure

S1.4). This consisted of a 10-cm-diameter ducted

propeller mounted in a frame with a large “tail”

to orient the meter with the current. The

assembly was attached to the end of a wire

and lowered to the desired depth. A metal

weight (messenger similar to that used for

a hydrographic “bottle” cast) was dropped

down the wire to free the propeller to rotate

and a second one was dropped after a measured

time to stop it. The number of revolutions was

recorded by a mechanical counter. The water

speed was then proportional to the number of

revolutions per minute. The current direction

on an Ekman current meter was recorded by

the rotating counting mechanism dropping

small metal balls at even intervals (controlled

by the rotation of the propeller) into a magnetic

compass tray with 100 sectors. Thus, the

number of balls in each tray gave a statistical

view of the directions. This instrument had to

be lowered and raised for each measurement d

a tedious business.

The Robert’s current meter was an improved

version of the Ekman current meter, and is the

forerunner of most current meters today. In the

Robert’s meter, speed (from a propeller) and

direction (from a compass) were transmitted

electrically to the surface and recorded shipboard

or transmitted by radio from the supporting

buoy to a mother ship. Since this required

considerable ship time to collect the measurements,

these current meters were eliminated in

favor of internally recording instruments that

can be moored for considerable periods of time.

One disadvantage of most propeller-type

current meters is that up-and-down motion

(when the ship rolls or the mooring moves),

which may cause the propeller to turn and cause

inaccuracies in the speed measurement. A

hollow cylinder (ducting) with its axis mounted

horizontally around the propeller minimizes

this effect. An alternative to the propeller is

the Savonius rotor (Figure S16.30), which is

less sensitive to vertical motion. It consists of

two half hollow cylinders mounted on a vertical

axis with flat end plates and produces a large

torque even in small horizontal currents. The

rotor is made of plastic to be neutrally buoyant

to reduce bearing friction so that it is sensitive

to currents of as little as 2 cm/sec. Even this

low threshold value can be a problem in parts

of the ocean where currents of this order prevail.

The rotor carries several small magnets. As each

magnet passes a coil on the frame it induces

a momentary electrical current pulse. The

number of pulses per second is proportional to

the current speed. The current direction is determined

electrically with reference to a magnetic

compass.

The Savonius rotor was used in Aanderaa

current meters (Figure S16.31, which were

widely used for several decades. (Aanderaa

now sells only acoustic sensors.) The Savonius

rotor on the Aanderaa current meter is affected

by vertical mooring motion, which causes an

alternating artificial current by what is called

“rotor pumping.” This effect can double the

recorded current speed and is most severe on

shallow meters and in coastal regions where

wave action is significant. To reduce this effect,

Aanderaa now mounts a semi-cylindrical

shroud around one-half of the rotor and uses

flat rotor blades rather than curved ones.

Note in Figure S16.31 that the current meter

is mounted on the mooring line with a hard bar


Aanderaa current meters, VACMs use a Savonius

rotor and vane. They were made by

EG&G Marine Instruments and remained in

wide use through the 1990s.

The Vector Measuring Current Meter (VMCM)

was developed in the 1970s (Weller & Davis, 1980)

to compensate for the rotor pumping problem

that affects Savonius rotors. The VMCM (Figure

S16.32) uses two orthogonal propellers. The

open fan-type rotors of the VMCM are susceptible

to fouling, and so, like Savonius rotor current

meters, VMCMs are gradually being replaced

with acoustic current meters.

S16.5.5. Eulerian Methods: Acoustic

and Electromagnetic Current Meters

FIGURE S16.30

Savonius rotor current meter.

while the current meter is supported by a gimble

that allows complete azimuth change with

limited vertical variation. Thus, the current

meter is expected to align itself with the current

that then measures the current direction with

an internal compass. All of the external sensors

areatthetopofthecurrentmeter.Thesecan

include temperature, pressure, and inductive

salinity. In the earlier Aanderaa current meters,

current speeds and directions were averaged

over a fixed period of time and stored

internally.

The Vector Averaging Current Meter (VACM)

(Figure S16.32a) was designed in the 1960s to

measure the velocity frequently, resolve it into

components, and record averages of these

components separately to give a more complete

record of the velocity. Presently available Aanderaa

meters follow this protocol as well. Like

Non-mechanical current-measuring devices

include acoustic and electrical-field and

magnetic-field sensing tools. These methods

are replacing mechanical current meters

because they are less subject to inaccuracies

resulting from fouling.

S16.5.5.1. Acoustic Current Measurements

The most widespread non-mechanical Eulerian

current measurement technology is acoustic,

which measures the travel time of pulses of

high-frequency sound reflecting off particles in

the water. The Doppler shift in frequency gives

a measure of fluid speed along the sound path.

These Doppler sonar profilers (Figure S16.33)

are equivalent to sonic anemometers used to

measure winds. Acoustic systems have no

moving parts that can foul or provide inertial

resistance to changes in ocean currents. Fouling

on transducer heads reduces instrument range

but not accuracy. Acoustic instruments can

also provide current measurements at

numerous depths within the range of the instrument,

which is usually several hundred meters.

Acoustic instruments in wide use include

acoustic Doppler current profilers (ADCPs),

acoustic Doppler velocimeters (ADVs), and acoustic

current meters (ACMs) (Figure 16.33).

44


FIGURE S16.31 Aanderaa RCM-7/8. Source: From Aanderaa Instruments (2000).

Acoustic current profilers are used both from

ship installations and as moored current

meters. For mooring use, both ADCPs and

ACMs are available. ADCPs are also sometimes

incorporated in a CTD/rosette package

and used to profile velocities in the water

column along with the temperature and

salinity profile from the CTD. In this configuration

they are known as Lowered ADCPs

(LADCPs).


FIGURE S16.32 (a) Vector Measuring Current Meter (VMCM) and Vector Averaging Current Meter (VACM). Source:

From Beardsley (1987). (b) Deploying a VMCM with some other instruments. Source: From USGS (2005).

The ADCP was developed in the 1970s from

the “Doppler log,” which measures currents

relative to a moving ship to yield ship speed.

(Originally, the speed of a sailing ship was

measured by measuring the travel time of a log

thrown into the water as it went from the ship’s

bow to its stern. From this practice, any speedmeasuring

device from a ship became known

as a log.) The Doppler log measures the speed

of the ship by sending out an acoustic pulse

that is reflected back to the ship by particles in

the water (such as plankton). The Doppler shift

of the returned signal’s frequency compared

with the original pulse makes it possible to

compute the ship’s speed relative to the water.

The use of ADCPs for oceanographic

research was pioneered by Rowe and Young

(1979) and Pinkel (1979). This same Doppler

technology allows the water motion relative to

the ship to be measured if the ship’s motion

can be accurately computed from an external

navigation system, such as the satellite Global

Positioning System (GPS). By controlling the

direction of the acoustic beam, the Doppler

system reflects the currents at different depths

below the ship. The principals of operation are

described in Howe and Chereskin (2007).

Using a three- or four-element sensor head,

an ADCP (Figure S16.33) is capable of resolving

both speed and direction of the water movements

relative to the sensor. Most oceanographic

research vessels carry an ADCP system on board

and may operate it continuously. The ADCP

profiling depth depends on the sound frequency

used in a given instrument. There is a trade-off

in depth coverage versus vertical resolution.

Greater depth coverage requires a lower

frequency, which results in lower vertical

46


(b)

SEACAT

Transmissometer

FIGURE S16.32

(Continued).

VMCM propellers

resolution. Commonly employed ADCPs profile

over 300 m. Lower frequency Doppler instruments

are becoming more common, profiling

to 800 or even 1500 m.

ADCP ACCURACY IS LIMITED BY

1. Theaccuracyofthefrequencyshift

measurement used to obtain the relative

velocity; this estimate is conducted by

software within the instrument and

strongly depends on the signal/noise ratio

and the velocity distribution among the

scatters.

2. The size of the footprint and the homogeneity

of the flow field; at a distance of 300 m from

the transducer, the spatial separation

between sampling volumes for opposite

beams is 300 m so that they are seeing

different parts of the water column, which

may have different velocities.

3. The actual passiveness of the drifters (i.e.,

how representative are they of the in situ

current?) and the concentration of the drifters

(limiting range in regions of exceptionally

clear water).

In the shipboard system, the ADCP can track

the bottom and obtain absolute velocity,

provided the acoustic beam ranges to the

bottom. Once out of range of the bottom, only

the velocity relative to the ship can be measured.

Erroneous velocity and backscatter data are

commonly obtained from shipboard ADCP

measurements due to vessel motions in

moderate to heavy seas; the transducer head

can be exposed and the acoustic signal attenuated

by air bubbles under the ship’s hull or

through the upper portion of the water column.

Much better data are collected from a ship

“running” with the seas than one lying in the

trough or hove-to in heavy seas. In deep water,

zooplankton aggregations can lead to the formation

of “false bottoms” in which the instrument

mistakes the high reflectivity from the scattering

layer as the seafloor.

ACMs and ADVs typically measure currents

at a point. Different instruments have been

developed for the wide range of fluid conditions

from rivers, lakes, and surf zones, to shallow

water and the deep ocean. Different instruments

have been developed for different conditions,

ranging from very shallow to deep water.

Commercial ADVs use three beams focused

a short distance from the instrument (tens of

centimeters); they measure all three components

of the velocity at one point with high spatial

resolution useful for studying turbulence and

waves.

ADCP ACCURACY IS LIMITED BY 47

lines of force of a magnetic field, an EMF is

generated:

E ¼ B$L$v

(S16.1)

FIGURE S16.33 Acoustic Doppler Current Profiler with

a 4-transducer head. Source: From Teledyne RD Instruments

(2011).

S16.5.5.2. Electrical and Magnetic Field

Current Measurements

A second technique for non-mechanical

current measurement is the hotwire anemometer,

commonly used to measure wind speed. In

this instrument, the rate of cooling of an electrically

heated wire is a measure of the fluid speed

past it. A thin wire or metal film about a millimeter

long is exposed to the flow and maintained

at a constant temperature by

automatically adjusting the electric current

through it so that the joule heating is exactly

equal to the rate of loss to the fluid. The magnitude

of the electric current is then a measure of

the fluid speed. This device is small and

responds rapidly to flow variations, which

makes it particularly suitable for the measurement

of turbulent fluctuations of flow speeds.

Problems with this system include its sensitivity

and tendency to foul. As a result, no reliable

moored version has been developed.

A third technique, the electromagnetic method,

uses a fundamentally different principle first

suggested by Faraday (1832). With this technique

an electromagnetic force (EMF) is induced

in a conductor when it moves through a magnetic

field. In oceanographic applications, seawater is

the conductor. When seawater flows across the

where v is the water speed, L is the width of the

current between the measurement points, and B

is the strength of the magnetic field component

in a direction mutually perpendicular to the

direction of both v and L. Depending on the

application and instrument, the magnetic field

can be that of Earth, or it can be generated internally

in the instrument. For a horizontal current

and a method that uses Earth’s magnetic field, B

is the local vertical component of the magnetic

field.

Faraday attempted to measure the flow of the

Thames using this method, but was unsuccessful

because of problems with copper electrodes.

Some of the earliest reported successful

measurements by this technique were of tidal

currents in the English Channel (Young, Gerard,

& Jevons, 1920); a long series of measurements

was made of the Florida Current between Key

West and Havana (Wertheim, 1954). The basic

equipment required is a recording milli-voltmeter

and two electrodes to dip in the sea. An

example of a modern electromagnetic current

meter, with two pairs of electrodes on opposing

sides of a small plastic sphere (25 cm or 10 inch

diameter), is pictured in Figure S16.34; this

particular instrument generates its own

magnetic field internally and the “current

width” is the distance between the opposing

electrodes. Because there are two pairs of

sensors mounted perpendicular to each other,

two orthogonal velocity components can be

measured.

Unused commercial undersea cables are often

used to make electromagnetic measurements

of currents. A very long time series of current

measurements through Florida Strait, between

Florida and the Bahamas, has been collected

using this method with abandoned telephone

cables that have easily accessible terminations

48


FIGURE S16.34 Electromagnetic current meter (S4

model). Source: From InterOcean Systems (2011).

on either end (Larsen & Sanford, 1985).

Undersea cables that have been used for such

transport measurements are shown in Figure

S16.35 and are monitored by ICPC (2007).

One source of error in the EMF method is the

finite, but usually unknown, electrical conductivity

of the sea bottom, allowing an electrical

current to flow due to the induced EMF and

thus reducing the observed EMF below the level

expected from the formula and speed of the

water. This introduces a constant scaling factor,

which must be determined by making some

water-current measurements with another type

of meter while the electromagnetic system is in

operation.

The Geomagnetic Electrokinetograph (GEK)

was an early adaptation of the electromagnetic

technique to permit underway shipboard

current measurement (Longuet-Higgins, Stern,

& Stommel, 1954; Von Arx, 1950). Two electrodes

were towed behind the ship with a cable

strung between the electrodes. The EMF

induced in the cable was recorded as a measure

of the component of the water velocity perpendicular

to the ship’s track. To obtain the total

water velocity, the ship was then turned at right

angles to the original track and a second

component measured. Combining the two

components gave the water velocity relative

to the solid earth. The difficulty of reducing

and interpreting GEK data led to a rapid

decline in its use. The small magnitude of

Earth’s magnetic field together with electrical

noise always present in nature makes the geoelectromagnetic

method practical only with

electrode separations of tens of meters or

more. Recently current meters employing this

principle with even smaller electrode spacing

have become available commercially. They

have no moving parts but do need a significant

electrical power supply.

Most electromagnetic current meters allow

for additional measurements of temperature,

conductivity, and pressure. Data can be averaged

over regular intervals of a few seconds to

tens of minutes, or set to burst sampling with

a specified number of samples per burst at

a given sampling interval. For example, one

can set the number of samples per burst (say

continuous sampling for two minutes every

hour) and set the number of times velocity is

sampled compared with conductivity and

temperature. The limitations are the storage

capacity of the instrument (thousands of kilobytes)

and the amount of power consumption.

For some electromagnetic current meters, the

surface of the housing is grooved to maintain

a turbulent boundary layer to prevent flow

separation at higher speeds.

S16.5.6. Mooring Technology

S16.5.6.1. Subsurface Current Meter

Moorings

Even with the development of improved

instruments, the measurement of currents from

a ship has several disadvantages. A major one is

that a ship cannot remain at sea for very long,

whereas it is very useful to obtain long records

of currents. A second, but minor problem is that

ship movement introduces spurious components

into the measured currents, which must be


FIGURE S16.35 (a) Map showing abandoned commercial undersea cables that have been exploited for observations of

transports. This map was prepared in 1998. (b) Close-up of Florida Current cable measurement locations. Source: From

Flosadottir (2004) and ICPC (2007).

filtered. (However, moored measurements can

have their own problems with motion.)

For these reasons, techniques for the successful

mooring and recovery of strings of current

meters in the deep ocean, supported on a cable

from a buoy beneath the surface to an anchor

weight on the bottom, have been developed

since the mid-1960s. There are still problems

associated with the movement of the mooring

in strong currents, but an autonomous, moored

string of current meters is nevertheless an efficient

way to resolve and monitor ocean current

behavior over a period of time. Instruments

can also be mounted on a frame that is fixed

on the bottom, which eliminates mooring

motion that is useful in shallow water or for

instrument types such as ADCPs or inverted

echo sounders that sample a large part of the

water column from a single instrument.

Moorings can be primarily surface or subsurface,

depending on location of the uppermost

instruments and mooring flotation (Figure

S16.36). Surface moorings have a surface float

with the instruments suspended below it on

a line. There is a loose tether to a bottom anchor.

Surface buoys are commonly used for surface

50


FIGURE S16.35

(Continued).

meteorological observations, such as wind,

pressure, and so forth; for surface layer observations

with a downward-looking ADCP; and

when regular data telemetry is desired. On

a subsurface mooring the uppermost flotation

for the line of instruments is below the sea

surface and includes clusters of intermediate

floats (Figure S16.37a) that provide buoyancy

to the mooring to keep it tight and provide the

buoyancy needed for recovery. The floats are

attached at intervals in the mooring to give

intermediate buoyancy making the mooring

tighter so that it resists being displaced by

currents. Mooring lines consist of steel cable,

nylon rope, or a synthetic called Kevlar. Steel

cable is very strong and inexpensive but is

subject to kinking while being deployed, which

can dramatically weaken the cable. Nylon rope

and Kevlar are most commonly used, because

both are pliable and thus unaffected by kinking

but are more expensive. Mooring anchors are

usually a set of used, hence low cost, railroad

wheels (Figure S16.37b,c).

To recover a mooring after it has been in the

water for the length of the experiment, it is

necessary to find it and release it. With GPS

navigation, it is now straightforward to return

to the deployment location for recovery. The

most common release method is an acoustic

release (Figure S16.37d) mounted between the

anchor and the end of the line; when it receives

an acoustic signal from the recovery ship, it

releases the line and the mooring surfaces.

Battery lifetime is the primary limiting factor

for acoustic releases. An early recovery technique

was a double anchor with a float at the

surface marking the second anchor. This second

anchor could be recovered and the system

brought in, ending with the current meters

and flotation. With this system, subsurface flotation

could be used, which greatly reduced

mooring motion from surface effects. A modification

of this system eliminated the surface

marker for the second anchor but required the

ship to grapple to find the cable between the

two anchors. The absence of a surface marker

is advantageous when there are ships or

icebergs in the area. These recovery methods

were time-consuming and risky compared

with the acoustic release method.

For deployment, a mooring is laid out behind

the ship, starting with the float that will be nearest

the surface. The top float usually has a light

and radio transmitter that activates when the

float reaches the surface during retrieval. The

top float is often equipped with a radar reflector

to make it easier to see on the ship’s radar. After

the near surface float is in the water the rest of the

mooring is played out one segment at a time with

current meters, thermistor chains, and intermediate

floats installed along the line as planned.

At the end, the acoustic release is mounted above

the anchor with the entire mooring floating out

away from the ship. The ship is then maneuvered

into position so that when the anchor is


FIGURE S16.36 Two current meter mooring configurations: (a) Surface mooring and (b) subsurface mooring. Internal

mooring flotation on the subsurface mooring is not shown. Source: From Emery and Thomson (2001).

dropped the mooring will fall to the position

desired. This is usually done by cutting a separate

line that has been used to support the anchor

when the ship is brought to the correct location.

When the mooring is recovered, it includes

a surface float with a radio transmitter and a light

to assist in locating it. Once spotted, the ship is

maneuvered so as to approach the buoy with

the working deck exposed to carefully get hold

of the line to pull the mooring in. The current

meters are strung out along the surface, supported

by the surface float and the various intermediate

floats. The mooring line is brought on

board, disconnecting the current meters from

the line.

S16.5.6.2. Deep-Sea Surface Moorings

Surface moorings have been used in coastal

oceans for many decades to measure

atmospheric conditions and sometimes the

currents and conditions in the water column.

In the past several decades, surface buoys

have also been moored in the deep ocean to

measure airesea interactions, particularly in

the tropical Pacific, which is the location of

the El NiñoeSouthern Oscillation (ENSO).

The tropical Pacific moorings were started as

part of the Tropical OceaneGlobal Atmosphere

(TOGA) program, and are called the

TOGA Atmosphere Ocean (TAO) buoys

(Section 10.8; Figure S16.38; TAO, 2009). The

TAO buoys relay their data in real time and

measure surface atmospheric conditions as

well as the water column. The tropical

measurements have proven so important for

interannual climate prediction that the arrays

have been extended into the tropical Atlantic

and Indian Oceans in arrays referred to as

52


FIGURE S16.37 (a) Intermediate floats being attached to a mooring line. (b) Railroad wheel anchors. (c) Cutting the

support line to drop anchor on a mooring. (d) Two different acoustic releases at the bottom of a mooring. (Photos courtesy of

W. Emery.)

PIRATA and RAMA respectively (Figure

S16.38).

A TAO surface float is shown in Figure S16.39

while it is being serviced. Because their data is

central to ENSO prediction, the floats carry

redundant sensors and are serviced on a regular

basis, at least once per year, requiring

a dedicated research vessel because of the large

number of moorings.

S16.5.6.3. Large Moored Buoy Programs

Prior to the TAO buoy program there were

a number of large buoys that were deployed to

measure primarily meteorological parameters

ACOUSTIC METHODS FOR OBSERVING CHANGES IN TEMPERATURE OR DENSITY 53

FIGURE S16.38 Tropical

moored buoy array. Source: From

NOAA PMEL (2009a); Bourlès et al.

(2008); McPhaden et al. (2009).

over the ocean, particularly in areas of critical

ocean operations such as the tanker route

between Alaska’s North Slope and Seattle.

Some of the earliest moored surface buoys

were the “bumble bee buoys” moored in the

North Pacific (Figure S16.40a) in the late 1960s

and early 1970s. These buoys were constructed

from old fuel tanks and fitted with meteorological

instrumentation. On the early buoys the

recording systems were optical with film

cameras recording the analog readings. This

meant that data processing was done by eye,

introducing the potential for errors.

Following the success of these programs in

collecting useful data from unvisited portions

of the North Pacific, there were plans to install

a large number of 12 m discus “monster buoys”

(Figure S16.40b) throughout the northern

oceans. The high cost of these buoys made this

expansive plan impractical. The National Data

Buoy Center (NDBC) was created within

NOAA; it operates monster buoys in the Bering

Sea, the Gulf of Mexico, and the western North

Atlantic. At this time there are seven such

buoys. There is also a set of five 10 m discus

buoys in the Caribbean and off the California

coast. A much larger number of 6- and 3-m

buoys have also been deployed. At present the

monster buoys report their data via satellite,

which makes them available both for operations

and for research applications. These are

primarily meteorological data that provide

needed forecast information for ship operations.

These data are available online from NDBC

(www.ndbc.noaa.gov/mooredbuoy.shtml).

S16.6. ACOUSTIC METHODS FOR

OBSERVING CHANGES IN

TEMPERATURE OR DENSITY

S16.6.1. Acoustic Tomography

As discussed in Section 3.8, electromagnetic

radiation penetrates only short distances in the

ocean, ~100e200 m for light. However, as discussed

in Section 3.7, the ocean is essentially

transparent to sound waves. The speed of sound

in seawater mainly depends on temperature

and pressure. Temperature changes are of great

interest, both at the short range and short timescale

of eddies and winter convection (tens to

hundreds of kilometers over several weeks or

months), and at the long range and long timescales

of basin-averaged temperature changes.

Thus the travel times of sound pulses between

a source and a receiver in a particular region

might be used to obtain information on the

changing temperature distribution. A technique

developed to take advantage of this is called

acoustic tomography (Munk & Wunsch, 1979;

Munk, Worcester, & Wunsch, 1995). It is analogous

with medical tomography in which brains

are mapped using radiation applied from

outside the head. Howe and Chereskin (2007)

provided a good review of this technique.

54


FIGURE 16.39 (a) TAO buoys being serviced. (b) Layout

of TAO ATLAS mooring. Source: From NOAA PMEL (2009b).

The first large-scale acoustic tomography

experiment used a moored array of sound sources

and receivers in a 300 km square (Figure

S16.41; Cornuelle et al., 1985). Each source and

receiver had accurate clocks so that travel times

for sound pulses could be determined along

each of the 20 possible paths joining the sources

and receivers, which were situated at about

2000 m depth in the sound channel. Along

each of the source/receiver directions, there

were 5 to 10 possible sound paths in the vertical

plane so that information about sound speeds

was available over a range of depths between

about 500 m and 3500 m. Analysis of traveltime

data by “inverse methods” (Chapter 6)

then yielded changes in sound speed. From

this the temperature structure in the volume

within the array is derived. Sound speed variations

of about 2 m/sec corresponded to temperature

variations of 0.4 C at 700 m depth. The

temperature structure derived from the tomography

corresponded well with CTD surveys,

which of course did not have the temporal

coverage of the tomographic array.

Ocean acoustic velocity tomography uses

measurements of the differences in travel time

between pairs of transceivers (co-located transmitter/receiver

pairs) to provide information

about the velocity of the water (Howe,

Worcester, & Spindel, 1987). For instance, if

the water between locations A and B is moving

from A toward B, then the travel time for

a sound pulse is less from A to B than from B

toA.Thedifferenceintraveltimeisproportional

to the water speed along the path AB

with an accuracy of a few centimeters per

second. With three moorings in a triangle, the

velocities measured along each leg of the

triangle provide the relative vorticity (rotation

about a vertical axis) of the water (Section

7.7). If extended to four, five, and more moorings

at the vertices of a square, pentagon, and

so forth, transmissions across the area can

also add information on the water motion

within the area.

ACOUSTIC METHODS FOR OBSERVING CHANGES IN TEMPERATURE OR DENSITY 55

FIGURE S16.40 (a) Bumble bee buoys used in the North Pacific. (Photo courtesy of W. Emery.) (b) 12 meter discus buoy.

Source: From NOAA NDBC (2008).

Acoustic tomography lends itself well to

intense regional studies for which repeated

cruises are impractical and for which traditional

moorings or floats would not provide the

required spatial and temporal coverage. Similar

regional tomographic arrays have been used

successfully to study winter convection in the

Greenland, Labrador, and Mediterranean Seas

(Morawitz et al., 1996; Avsic, Send, & Skarsoullis,

2005; Send, Schott, Gaillard, & Desaubies, 1995).

Acoustic tomography is proving very useful for

monitoring straits as well (Denmark Strait,

Fram Strait, Strait of Gibraltar).

Acoustic tomography has also been implemented

for basin scales in the ocean to detect

very large scale warming or cooling (Acoustic

Thermometry of Ocean Climate, ATOC;

Dushaw, 2002). Sound sources of sufficient

strength can be heard around the world since

acoustic waves propagate so easily in water.

Changes in travel time for acoustic waves along

the extremely long paths are related to changes

in the total temperature change (“integrated

temperature”) along the path. A test of the

concept was made with a sound source in the

southern Indian Ocean, which was readily

heard at receivers on both the east and west

coasts of the United States. An array of sources

and receivers around the North Pacific (Figure

S16.41b) is presently monitoring basin-scale

variations in temperature. A global-scale

deployment would be feasible (Figure S16.41c).

S16.6.2. Inverted Echo Sounder

Rossby (1969) suggested that variations in

travel times of acoustic pulses from the

seafloor to the sea surface could be related to

changes in the density structure and hence

depth of the thermocline. Moreover, since these

travel times are integrated measurements over

the water column, they effectively filter out all

but the fundamental mode of any vertical oscillations.

This led to the development of the

inverted echo sounder (IES; Watts & Rossby,

1977), in which the round-trip travel times of

regularly spaced acoustic pulses from the

seafloor are used to determine temporal variability

in the vertically integrated heat content.

This can be related to thermocline depth and

dynamic height if the temperatureesalinity

relationship is stable. With an array of IESs, it

is possible to map the changing dynamic

height field and hence changes in geostrophic

currents. This has good regional application,

especially for studying the evolving eddy field.

IES arrays have been used in studies of the

Gulf Stream, Antarctic Circumpolar Current,

Japan Sea, and Malvinas Current as well as

other regions.

56


FIGURE S16.41 (a) Moored array of sound sources (S), receivers (R), and “environmental moorings” (E) for the first

large-scale acoustic tomography trial in the western North Atlantic. Source: From Cornuelle et al. (1985). (b) Existing North

Pacific ATOC acoustic array. Source: From Dushaw (2003). (c) Prototypical ATOC long-distance acoustic array. Source: From

Dushaw (2002).

SEA-LEVEL MEASUREMENT 57

S16.7. SEA-LEVEL MEASUREMENT

Like SST, the measurement of sea level is one of

the oldest oceanographic observations. Observations

and increasing understanding of the tides

occupied ancient scientists from Greece to India,

Arabia, and China (Cartwright, 1999). Nineteenth

century sea level studies were related to

vertical movements of the coastal boundaries in

the belief that, averaged over time, the height of

the mean sea level was related to movements

of the land. Today sea level data are used to

resolve the tides, monitor wind-driven storm

surges and events such as El Niño in the western

tropical Pacific, monitor global sea level rise, and

calibrate/validate satellite altimeters (Section

S16.9.9). Tide gauges also form the backbone of

the tsunami warning system that alerts coastal

residents to seismically generated waves.

In addition to measuring the effects of coastal

erosion and global sea level rise, long-term sea

level changes are also related to changes in

global ocean currents. A map of sea-surface

topography is analogous to a meteorologist’s

map of surface pressure from which winds are

inferred. For this reason satellite-based radar

and laser altimeters can be related to the

geostrophic components of the ocean circulation

(Section S16.9.9 and Section 7.6).

Most operating tide gauges consist of a float

sitting in a “stilling well” (usually a pipe) with

a counterweight on the other side of a shaft

that rotates as the float (sea) level changes

(Figure S16.42). The system includes careful

benchmarking to measure changes in the land

level. Float tide gauges are the most standard.

In areas with large wave and wind action that

creates oscillations even within the stilling

well, a “bubbler” gauge can be used. With this

gauge a bubble is released at the bottom and

variations in pressure due to oscillations of the

sea level can be sensed by changes in the bubble.

Bubbler gauges actually measure the combined

effects of sea level and atmospheric pressure,

so it is necessary to correct for atmospheric pressure

when processing these data. Data within

the global network of sea level stations are digitally

encoded and transmitted by satellite to

data collection centers.

In the distant past, sea level was measured

with a staff installed so that the level of the

ocean’s surface moved up and down the graded

staff, and the level was read by eye. This was

then scribed on to a recording chart that moved

with time.

Satellite altimetry is an important new tool

for observing global sea level changes. (Altimetry

is not useful for observing tides, since it

does not have the correct temporal sampling.)

For the ocean interior far from the islands and

coastlines where tide gauges can be mounted,

altimetry is the only available tool for observing

sea level change. Its uncertainty is 3e4 mm,

with larger error near coastlines. Tide gauges

FIGURE S16.42 Tide gauge

measurement system. Source: From

Nerem (2009).

58


FIGURE S16.43 The global sea level observing system

(GLOSS) tide gauge network. Source: From WMO JCOMM

(2009).

serve as altimetric calibration points, so they

continue to be essential for observing global

sea level change. The Global Sea Level

Observing System (GLOSS) is the international

focal point for tide gauge data (WMO JCOMM,

2009). The tide gauge network currently

comprises 290 stations (Figure S16.43).

Observations of shortwave radiation are

made with a pyranometer. The sensing element

of the Eppley pyranometer consists of two flat

plates of copper, one painted with a flat black

paint and the other whitened with magnesium

oxide. The two plates are placed horizontally

with a clear view of the sun and sky and are

shielded from draughts by a clear hemispherical

cover. The black paint absorbs all shortwave

and longwave radiant energy falling

upon it and is heated above the surrounding

temperature. The white plate reflects practically

all of the energy between 0.3 and 5 mm

(shortwave radiation) but absorbs all longwave

energy. The white plate is consequently heated

lessthantheblackoneandthedifferencein

temperature between them is a measure of the

shortwave radiation (Q s ) falling on a horizontal

surface in the locality of the instrument. The

difference in temperature is measured by connecting

the “hot” junctions of a group of

S16.8. RADIATION AND OPTICAL

MEASUREMENTS

In Chapter 5 we discussed methods for

computing shortwave and longwave radiation

indirectly using bulk formulae and observations

of external quantities such as cloud cover and

albedo. These formulae are based on direct

measurements of shortwave and longwave radiation

(Q s and Q b ). Such direct measurements are

made at high-quality weather stations and

meteorological instrument packages (Figure

S16.44), which are often carried by research

ships. In Chapters 3 and 4 the inherent and

apparent optical properties of seawater, which

are observed in situ using various optical instruments,

were described. (Satellite observations

relevant to these quantities are described in

the next section.)

FIGURE S16.44 Meteorological sensor package: ASI-

MET system with dual sets of sensors for shortwave radiation

(pyranometer), longwave radiation (infrared

radiometer), barometric pressure, relative humidity and air

temperature, precipitation, and wind. Source: From WHOI

(2010).

RADIATION AND OPTICAL MEASUREMENTS 59

thermocouples to the black plate and the “cold”

junctions to the white plate. The difference in

temperature gives rise to a thermoelectric

EMF, which is measured by a recording galvanometer.

This instrument is calibrated by

exposing it to a standard source of energy,

such as a standard electric filament lamp.

The downward-directed component of the

longwave radiation Q b in the atmosphere is

measured with a radiometer. This Gier and Dunkle

instrument consists of two horizontal plates

of black material separated by a layer of material

of known heat conductivity. The upper sheet

of black material absorbs all the radiation falling

upon it from above and is heated above the

temperature of the lower sheet, which is

screened from radiation from below by a sheet

of polished metal. The difference in temperature

between the upper and lower sheets is

measured by thermocouples and is a measure

of the rate at which the sum total of longwave

and shortwave energy is coming down from

above. To determine the value of just the longwave

component, it is necessary to subtract the

shortwave radiation rate as measured with

a pyranometer. An alternative procedure is to

omit the polished metal screen from below the

black horizontal plate and arrange the instrument

so that the upper plate “looks at” the

atmosphere and the lower plate “looks at” the

sea below. In this “net radiometer” arrangement

the difference in temperature between the upper

and lower plates is a measure of the net amount

of radiant energy reaching a horizontal surface,

that is, it is a direct measure of (Q s Q b ).

The Secchi disk is the simplest device used to

determine the transmission of visible light

through the water, hence its clarity (Tyler,

1968; Preisendorfer, 1986). The Secchi disk

used for marine applications is a plate with

a 20 to 40 cm diameter that hangs horizontally

on the end of a rope marked in meters (Figure

S16.45); smaller disks are used for lakes

(Carlson, 2011). The upper surface is usually

painted white or with alternating white and

FIGURE S16.45 Secchi disk with alternating black and

white quadrants, which is typically used for lake applications.

Secchi disks for marine applications are larger and

usually all white. Source: From Carlson (2011).

black segments. The disk is lowered into the

sea and the depth at which it is lost to sight is

noted. This depth decreases as the vertical attenuation

coefficient of the seawater increases. In

very clear water the depth may be over 50 m,

in coastal waters 1 to 2 m, and in some river

estuaries less than 1 m. The Secchi disk

measurement is only semiquantitative, but has

often been used because it is so simple. After

just a little practice, it is possible to obtain

consistent readings to better than 10%, with

little variation from individual to individual.

Secchi disks are also used to estimate attenuation

coefficients resulting from dissolved and

particulate material. However, Preisendorfer

(1986) cautioned against using Secchi disks for

quantitative estimates. Secchi depths for the

Atlantic and Pacific range from less than 15 m

60


to more than 35 m (Figure 4.26), and are

inversely correlated with chlorophyll-a content.

Modern electronic instruments measure

optical properties directly and quantitatively.

There are many useful instruments, including

those measuring beam attenuation as a function

of depth and wavelength (transmissometer), fluorescence

(fluorometer), light scattering (optical

backscattering meter), and radiance and irradiance

sensors. The first three sensors are active instruments,

which emit their own light and measure

the response. The latter sensors are passive,

measuring the ambient light.

Transmissometers (Figure S16.46) measure

beam attenuation. The instrument emits light

at a specified wavelength, and then detects

how much light passed through the intervening

water to be intercepted at the other end of the

instrument. The wavelengths are chosen based

on what is being studied, such as chlorophyll,

dissolved organic matter, and other particles

(Richardson & Gardner, 1997). Transmissometers

compatible with full-depth CTD casts

were used throughout the WOCE to gather

full-depth profiles of attenuation. Typically

there is high attenuation in the surface layer as

FIGURE S16.46 (a) Transmissometer schematic. (b) Fluorometer schematic. Source: From Richardson and Gardner (1997).

(c) Irradiance radiometer. Source: From TriOS (2009).

SATELLITES 61

well as in a boundary layer of about 100 m thickness

at the ocean bottom where sediment is

stirred up.

Fluorometers measure fluorescence, which

indicates the presence of chlorophyll. They

emit flashes of light at specified wavelengths

to excite fluorescence and measure the emitted

light (which is at another wavelength). The light

and the receiver are located close together in

contrast to the setup on the transmissometer

(Figure S16.46).

Backscattering instruments emit a beam of

light and measure the backscattered light with

a sensor located next to the light.

Radiance sensors (radiometers) measure

visible light at a range of wavelengths within

a narrow field of view (e.g., 7 degrees), while

irradiance sensors measure the same within

a wide field of view (TriOS, 2009).

S16.9. SATELLITES

S16.9.1. Satellite Remote Sensing

One of the biggest problems in physical oceanography

is the mixture of time and space

sampling that results from the normal ship-based

measurement system. Because of ship speed limitations

(10e12 knots), it is impossible to observe

a large area in a “synoptic” (i.e., near simultaneous)

fashion. Thus oceanographers once had

to be satisfied with data collected over some

period of time. For many stationary and longer

period phenomena, such averages are adequate,

but as the study of time-variable oceanographic

processes has intensified, the need for more

nearly simultaneous sampling has increased.

One solution is the use of Earth-orbiting satellite

data to sense the ocean’s surface. While satellite

measurements are limited to near the ocean’s

surface, the large-scale, almost-synoptic sampling

(minutes for areas of thousands of square kilometers)

is an essential component of the global

observing system. This is complemented by the

increasing deployment of autonomous in situ

instruments as part of the Global Ocean

Observing System, including subsurface floats

and surface drifters (Section S16.5).

Oceanographic parameters that are routinely

measured by satellites include surface temperature,

sea ice distribution, wave height, surface

height, radar backscatter, and ocean color. A

sea-surface salinity sensor is set to be launched

in 2011 Meteorological parameters measured

by satellites that are important for ocean forcing

include wind speed, sea level pressure, cloudiness,

and water vapor content. The next subsections

describe some of the present (around 2004)

generation of primarily U.S. satellite sensors

used for oceanography.

There are many other satellite-based observational

systems not covered here. NASA

publishes an online Remote Sensing Tutorial

(http://rst.gsfc.nasa.gov/) that is an excellent

starting point. A good, older textbook is Methods

of Satellite Oceanography (Stewart, 1985). Many

texts address specific aspects of satellite remote

sensing. As any textbook presentation of satellite

methods is certain to be outdated within

just a few years, the student is encouraged to

seek the latest information from the space

agencies and data archive centers. Web sites d

particularly those of NASA, NOAA, the European

Space Agency (ESA), and the Japan Aerospace

Exploration Agency (JAXA) d provide

a large amount of information about these satellites,

and are also excellent starting points for

data sets. An especially useful gateway to

oceanographic data sets is the Web site for the

NASA Jet Propulsion Laboratory’s Physical

Oceanography DAAC (PO.DAAC), which has

been reliably available and updated for

a number of years.

S16.9.2. Satellite Orbits

At present much satellite-based information

comes from operational weather satellites in

both polar and geostationary orbits. A geostationary

satellite remains fixed over one location on

62


Earth’s surface orbiting Earth at the same rate as

Earth rotates. This specific orbit necessitates an

altitude of 35,800 km. The geostationary system

scans about one-third of Earth’s surface under

the satellite. Geostationary operational environmental

satellites (GOES) for weather analysis

and forecast purposes are operated by NOAA

(GOES East and GOES West), providing full

coverage of the American continents and most

of the Pacific and Atlantic Oceans. The principal

instruments on GOES are an imager (visible and

infrared radiation) and atmospheric sounder.

Among other products, these provide the

familiar satellite images of cloud cover used in

weather reporting (Figure S16.47).

ESA has operated geostationary satellites for

the same weather purposes since the 1970s. The

first ESA satellite was Meteosat (1977), stationed

over western Africa (intersection of the equator

and the Greenwich meridian). The series of

Meteosats was followed in 2000 by the Meteosat

Second Generation (MSG) satellites with 1 km

resolution in the visible light and 15 minute

reporting intervals as well as a number of new

sensors.

At much lower altitudes of about 800 km,

polar-orbiting weather satellites are generally

deployed in “sun-synchronous” orbits to maintain

the same local equatorial crossing time on

each orbit. Thus each polar orbiter passes over

a location at least twice per day (once at night

and once during the day). At higher latitudes

the approximately 2000 km scan width of these

instruments leads to overlap, providing many

more views of the surface. Polar-orbiting

weather satellites have progressed dramatically

since 1960, when the television camera on the

first Television Infrared Observing Satellite

(TIROS) could point only at North American

latitudes due to the spin stabilization of the

spacecraft. This first series of ten weather satellites

ended in 1966. The next TIROS spacecraft

series mounted a camera looking out along

the radius of the spinning satellite (called the

wheel satellite) and took a successive set of

pictures to provide the first global view of

Earth’ssurface.Today,three-axisstabilization

on the TIROS-N (N for new) satellites makes

it possible to keep instruments such as the

Advanced Very High Resolution Radiometer

(AVHRR), which senses surface temperature

as described in Section S16.9.5, pointed continuously

at Earth’s surface. As of 2011 there have

been nineteen TIROS-N satellites, known as

NOAA 1 through 9. The first was launched in

1970 and the most recent in 2009, with new

launches every two to four years. The AVHRR

is the principal instrument of oceanographic

interest on these satellites.

Another series of American polar-orbiting

satellites that collect data useful for oceanography

is the Defense Meteorological Satellite

Program (DMSP). They use essentially the

same spacecraft as the NOAA polar orbiters

but they carry different payloads. The principal

instruments useful for oceanographic studies

are visible and infrared sensors for SST and

a microwave sensor that has been especially

useful for sea ice. These data are collected by

an instrument called the Operational Line

Scanner (OLS). It has poorer spatial and radiometric

sensitivity than the AVHRR, but has

a unique characteristic The OLS has a “lowlight”

capability, which makes it possible to

view Earth lights from cities, polar Aurora,

and other low intensity light phenomena. This

capability was specified for the DMSP by the

U.S. military that wanted the ability to operate

even at night when the visible solar radiation

was not available. Currently there are four

DMSP satellites in operation.

ESA has launched numerous Earth observing

satellites. Polar-orbiting satellites include ERS-1

(1991e2000), ERS-2 (1995epresent), and Envisat

(2002epresent). Each of these satellites has carried

a wide range of sensors useful to oceanographic

applications such as imagers, radar altimeters,

scatterometers, and synthetic aperture radars.

The present approach to earth science from

satellites within NASA and ESA is to fly many

SATELLITES 63

FIGURE S16.47 Infrared image for June 2, 2009 from (a) NOAA GOES_East, (b) NOAA GOES_West, and (c) Meteosat.

Source: From NOAA NESDIS (2009).

small missions, each dedicated to one or just

a few parameters. Almost all of the many

NASA satellites in the Earth Science Mission

are polar orbiters in various orbits that depend

on the desired frequency, repetition, and spatial

resolution and range. A summary of these

missions can be found on NASA’s Earth

Observing System (EOS) Web site (http://

eospso.gsfc.nasa.gov/).

S16.9.3. Sensor Types

All satellite sensing is “remote,” most using

electromagnetic radiation at various wavelengths

64


and extensive signal processing to assemble

images and digital data sets of physical parameters.

(One notable exception to the dominance

of radiation sensors for oceanographic purposes

is the NASA GRACE satellite, which is used to

sense Earth’s gravitational field through

measuring its actual displacement as a function

of that field.) The radiation used in satellite

systems for oceanography ranges from microwaves

through infrared to visible (Figure

S16.48 a). Wavelengths for microwaves are

between 0.1 and 30 cm, for infrared radiation

from 0.7 to 1000 mm (0.1 cm), and for visible light

from 400 to 700 nm (0.7 mm). Specifications for

FIGURE S16.48 (a) The electromagnetic spectrum and (b) the atmospheric transmission % and solar spectra; the

emission spectrum for Earth is also shown (green). Source: After NASA GSFC (2004).

SATELLITES 65

satellite sensors are often listed in terms of

frequency of the radiation rather than wavelength.

The frequency n of electromagnetic radiation

is related to wavelength l through the

speed of light c:

ln ¼ c

(S16.2)

The speed of light is 3.00 10 8 m/sec. Most

satellite instruments measure several different

wavelengths. Scientists and engineers then use

the measurements at these various wavelengths

to construct the physical field.

Satellite radiation sensors are either “active”

or “passive.” In passive systems, the satellite

sensors detect radiation from Earth’s surface

or reflected/re-radiated solar energy. In active

systems, the satellite radiates energy downward

and then senses such energy reflected/re-radiated

from Earth. The latter is typical of radar

and lidar systems that emit radiation to sense

surface properties.

The curves in Figure S16.48b represent the

atmospheric transmission and the solar emissions

spectra. The wavelength axis is expanded

compared to Figure S16.48a, but it is possible to

compare the main components of the relevant

spectra. Transmission through the atmosphere

in the visible and near-infrared portions of the

EM spectrum is nearly complete; that is, the

atmosphere is nearly transparent to these wavelengths.

The solar emission maximum is also in

the visible range, so that most shortwave energy

reaching Earth from the sun is in this and

adjoining ranges, including near-infrared.

In the thermal infrared (10 e12 mm), transmission

is greater in well-defined atmospheric

windows. The best (most transparent) longwave

thermal channel is at 10 mm. There is

another atmospheric window at about 5 mm.

At these wavelengths the solar emission

decreases relative to its peak at the visible

wavelengths. At the shorter thermal wavelengths

Earth emission (green) is quite small.

There is another atmospheric window at about

60 to 80 mm where both the solar and Earth

emissions are quite low.

The atmosphere is almost completely transparent

for microwave wavelengths greater

than about 5 mm.

S16.9.4. SEASAT

The first satellite dedicated to ocean remote

sensing was the short-lived SEASAT satellite

(Figure S16.49). Because of its special place in

the history of satellite techniques for measuring

the ocean, we describe it and all of its sensors

together. Launched in mid-1978, SEASAT failed

after 90 days of operation. Even during its short

life, SEASAT proved the concept of a number of

unique passive and active microwave instruments

designed to measure various parameters

of the ocean. The instruments included an altimeter,

a visible-infrared radiometer similar to the

AVHHR, a Scanning Multi-channel Microwave

Radiometer (SMMR), a RADAR scatterometer,

and a Synthetic Aperture Radar (SAR). These

provided measurements of sea-surface height

(altimeter), SST (AVHRR and SMMR), ice distribution

(SMMR and SAR), and wind speed and

direction (RADAR scatterometer). These SEA-

SAT instruments are described in the following

paragraphs.

The passive SMMR on SEASAT provided

all-weather images of sea ice and SST. Since

microwaves are only slightly attenuated by

atmospheric moisture, they are excellent for

observing Earth’s surface, which is often

obscured by clouds. This is particularly important

for the ice-covered polar regions, which

are frequently cloud-covered. In addition, the

SMMR responds to changes in surface emissivity

related to ice age and thickness. A low

frequency (6.7 GHz) channel on SMMR was

intended to provide SST independent of cloud

cover. Unfortunately, calibration and noise

problems with the SMMR in this channel

resulted in inaccurate SST, a problem that has

been solved in more recent microwave sensors.

66


FIGURE S16.49 SEASAT satellite.

Source: From Fu and Holt (1982).

The RADAR Scatterometer on SEASAT was

an active microwave instrument, measuring

wind speed and direction from the RADAR

backscatter from the small wavelets that form

at the ocean’s surface when the wind blows.

This system accurately measures wind stress

over the ocean both in terms of magnitude and

direction. This is the best way to resolve the

wind patterns and their changes over the open

ocean. Oceanographers have continued to

pursue the launch and operation of these

systems; scatterometry is more completely

described in Section S16.9.10.

The (active) SAR on SEASAT was the first

SAR flown on a satellite. In its short period of

operation the SEASAT SAR produced some

very interesting images of the ocean’s surface

that are still being analyzed today, although it

is clear that we do not yet completely understand

the SAR signatures of ocean features.

SATELLITES 67

SAR has also proven very useful for the detailed

mapping of sea ice and its motion. Again the allweather

capability of an active microwave

sensor makes it possible to see through persistent

cloud cover. Also the antenna synthesis

available with SAR makes it possible to attain

high spatial resolution (25 m on SEASAT) with

a small satellite antenna. The biggest problem

with SAR is the large amount of data processing

necessary to retrieve the image of interest from

the large volume of data recorded. Originally

this was done using optical analog methods,

which were very fast but produced brightness

inconsistencies. Digital SAR processing was

shown to be much more consistent and now

all SAR processing is done digitally. In addition,

SAR systems have been developed that can

actually process the SAR data onboard the

spacecraft.

An early satellite altimeter was also flown on

SEASAT. Altimeters monitor the height of the

sea surface and its changes. Again a lot of experience

was gained in working with satellite

altimetry obtained from the short life of SEA-

SAT. As discussed in Section S16.9.9, because

Earth’s geoid (gravity field) has not yet been

mapped in enough detail to allow use of the

altimeter to map absolute sea-surface height,

altimeters have been used mainly to study variability

in sea-surface height. The SEASAT

altimeter also provided the first truly global

map (Cheney, Marsh, & Beckley, 1983) of

eddy energy from fluctuations of the ocean’s

surface height.

Finally, SEASAT carried a (passive) visibleinfrared

radiometer (similar to the AVHRR) to

provide single channel visible and thermal

infrared imagery simultaneously with the

microwave data. All of the SEASAT instruments

functioned well during the short lifetime

of the satellite. Only failure of the power

supply terminated the mission. The concepts

behind all of the microwave sensors described

below were established by the SEASAT

mission.

S16.9.5. Sea-Surface Temperature from

Satellite Remote Sensing

SST (Section S16.4.2.1) is measured from

satellites with two different methodologies:

visible-infrared radiometry and passive microwave

sensing. Starting in the early 1970s, procedures

were developed to routinely compute SST

from satellite infrared measurements. Clouds

completely block infrared radiation and normal

water vapor seriously attenuates it, resulting in

only partial SST coverage. Various methods

have been used to correct for clouds and water

vapor with most relying on statistics to correct

for these effects. A major development was

a shift from the 8 km spatial resolution of the

SAR to the 1 km resolution of the Very High

Resolution Radiometer (VHRR), flying on the

same spacecraft, allowing observation of almost

the smallest scale phenomena. Later an

improved version of this same instrument, the

Advanced VHRR (AVHRR), became the standard

instrument for satellite SST estimates.

Most archived data are from the AVHRR (see

Figure 4.1b). The Global Area Coverage archive,

at about 4 km resolution, goes back to late 1978.

There are also a number of archives of 1 km

AVHRR data from direct readout stations

around the globe. Microwave sensors have

poorer spatial resolution but provide images

even in cloudy conditions. Given that clouds

cover a large portion of the ocean at any given

time, and given that some regions in some

seasons are almost completely cloud covered,

microwave SST sensing is indispensable. Unfortunately

there have not been as many successful

passive microwave sensors with channels

appropriate for SST sensing.

The AVHRR is the primary SST sensor on the

TIROS-N (NOAA operational) satellites. The

AVHRR has five channels: one in the visible

(0.58e0.68 mm), one in the near-infrared

(0.725e1.10 mm), and three in the thermal

infrared (3.55e3.93 mm, 10.3e11.3 mm, and

11.5e12.5 mm) channel. This combination has

68


proven to be useful in studies of cloud patterns

and atmospheric temperatures, land-surface

vegetation, and SST. The thermal infrared channels

also provide meteorological images at night

when there is no visible radiation to reflect from

Earth.

The multiple AVHRR thermal infrared channels

make it possible to estimate the atmospheric

attenuation, by atmospheric water

vapor, of the infrared signal emitted from the

ocean’s surface. Using the difference between

the radiation received in channel 4 versus that

on channel 5, it is possible to estimate the

amount of infrared energy lost to atmospheric

moisture before reaching the AVHRR sensor

optics. The relatively weak surface temperature

gradients in the ocean make it necessary to carry

out the best atmospheric moisture correction

possible when working with SSTs computed

from AVHRR data in an effort to get the precision

of the satellite measurement to the 0.3 K

accuracy recommended for modern climate

studies.

One of the most important aspects of working

with the AVHRR data, and any other satellite

data, is the correction of imagery for Earth

distortion and the remapping of the corrected

image to a selected map projection for intercomparisons

between images and with other

images or in situ data. This step is generally

referred to as “image navigation” and is essential

for employing AVHRR imagery quantitatively.

Without this image navigation step,

satellite infrared imagery can only suggest the

complexity of scales on which the ocean varies.

The AVHRR provides three different types of

image data. The most commonly available form

is the direct readout mode, which is called High

Resolution Picture Transmission (HRPT) and is

directly read-out by a tracking antenna at

a ground station. These data have an approximately

1 km spatial resolution. Each station

receives between 4 and 6 passes per satellite

per day and, depending on the latitude of the

satellite pass relative to Earth station locations,

these passes include between 2000 and 6000

scan lines of 1 km AVHRR data. During each

orbit the satellite system records a lower spatial

resolution product (approximately 4 km square)

called the Global Average Coverage (GAC).

These data are only read out at receiving

stations at Wallops Island, Virginia, and Gilmore

Creek, Alaska. These stations are operated by

NOAA, the agency responsible for the U.S.

operational weather satellites. The GAC data is

valuable because each day a satellite provides

a day and night image of the entire globe.

Finally, each satellite has a number of tape

recorders that can record the full 1 km image

data during a short part of the orbit. These Local

Area Coverage (LAC) data are recorded at areas

selected and commanded by NOAA and are

then downloaded, or received at, one of the

two NOAA stations. In this way it is possible

to “see” parts of Earth that are out of range of

the NOAA operated receiving stations.

An example of the 1 km imagery available

from the AVHRR is shown in Figure S16.50

which is an infrared (channel 4) AVHRR image

of the North American west coast region

around San Francisco, California. The color

scale indicates that surface temperatures in

this image range from about 10 C near the coast

to almost 17 C farther offshore. The colder

water near the shore reflects the presence of

summer upwelling bringing colder water to

the surface. The rich patterns of meanders and

eddies reflect the well-known instabilities that

dominate this area, creating the mesoscale

features and what are sometimes known as

“jets and squirts” extending out from the coast.

Unlike the smooth temperature map of Figure

4.1a, based on averaged historical data, or

even the 50 km global SST satellite product in

Figure 4.1b, the SST gradients in this image

are quite complicated. Mesoscale features

populate the boundaries between warm and

cold water. This truly synoptic sample clearly

indicates the complex spatial variability of

many features.

SATELLITES 69

FIGURE S16.50 Thermal infrared image of the West

Coast of North America from the Advanced Very High

Resolution Radiometer (AVHRR) on June 12, 1993, at 23:52

Universal Time.

The primary SST imager on the DMSP satellites

is called the OLS. This is quite different

from the AVHRR OLS. In addition to visible

and thermal infrared channels, the DMSP satellite

OLS has a unique low-light imaging capability

designed to make it possible to sense

Earth surface conditions at nighttime. There

are two levels of spatial resolution: fine (~0.5

nautical mile) and smooth (~2 nm). Unfortunately

most of the data archived from the

DMSP satellites are stored only in the “smooth”

mode (~2 nm resolution) making them marginally

useful for oceanographic studies. DMSP

data are broadcast in an encrypted format

because these are military satellites. Over areas

such as the Antarctic the encryption is removed

so that the OLS data are available for scientific

use. Decrypted versions of the DMSP data are

available through NOAA’s National Geophysical

Data Center (NGDC) in Boulder, Colorado.

The MODerate resolution Imaging Spectrometer

(MODIS), on the Earth Observing TERRA

and AQUA satellites (launched in 1999 and

2002, respectively), has channels to compute

infrared SST and ocean color. There are additional

channels of optical data, which are now

being explored for additional ocean applications

(examples in Section 4.8).

A major limitation for SST imaging from

AVHRR and other visible-infrared sensors is

the presence of cloud cover. Passive microwave

sensors can observe through clouds because

they use longer wavelengths (6e12 GHz). Early

observations were made from 1978 to 1986, using

the SMMR, which had calibration and noise

problems for SST applications. Later success

has come from the use of the Tropical Rainfall

Mapping Mission (TRMM) Microwave Imager

(TMI). While not intended for measuring SST,

TMI has proven very useful for this application

(see example in Chapter 4), and has been

enhanced by availability of the Advanced Microwave

Spectral Radiometer (AMSR).

Unfortunately, spatial resolution for passive

microwave SST is 25 to 50 km instead of the 1

km AVHRR resolution. This is too large to

“see” SST gradients in detail. However the

microwave sensor provides information that

would otherwise be impossible to collect because

of clouds. A challenge now is to develop techniques

for merging passive microwave and

infrared satellite measurements of SST.

Many SST products based on satellite data

are now available. Some of these are based on

individual sensors and others are a blend of

different types of data, including AVHRR and

microwave sensing. Some include in situ observations

as well. The products differ in spatial

and temporal resolution and averaging.

S16.9.6. Sea Surface Salinity

Satellite sensing for sea surface salinity (SSS)

is on the horizon. A major effort has begun

through the NASA EOS to develop a passive

microwave radiometer that will provide information

on salinity. This mission is called

70


Aquarius and is scheduled for launch in 2011

The spacecraft will be provided by Argentina.

This instrument takes advantage of the dependence

of conductivity of seawater on salinity.

Variations in conductivity can be sensed by

microwave radiometers. Aquarius will carry

three radiometers sensitive to salinity and

a scatterometer to correct for surface roughness.

S16.9.7. Sea Ice

Sea ice observations have been revolutionized

by satellite measurements. Satellite imagery is

used to monitor the presence of sea ice cover

and to estimate its concentration. Sea ice motion

can be mapped from successive images. Optical

systems provide extremely high resolution,

down to the level of leads in the ice. However,

optical sensors are limited by the frequent presence

of clouds at polar latitudes. Thus microwave

sensors provide comprehensive and

routine coverage of sea ice cover and concentration

(Gloersen et al., 1992). In addition the microwave

imagery can also be used to estimate sea ice

parameters such as thickness and ice age.

The original microwave imager used for sea

ice was the SMMR, which was launched in

1978. The Special Sensor Microwave Imager

(SSM/I) was developed by Hughes Aircraft

and has been in operation on DMSP satellites

since 1987. The SSM/I was a follow-on to the

SMMR and was used to observe a wide variety

of atmospheric conditions of interest to the military.

The SSM/I was the first “total power radiometer”

and has been copied in subsequent

microwave radiometers. The earliest passive

microwave systems proved to have such weak

signals relative to the large electronic “noise”

components of the system that it was necessary

to have some reference to extract the signal. In

1946 an engineer named Dicke invented

a passive microwave system that switched the

sensor between Earth’s surface and an internal

reference target to measure the surface relative

to the known internal target. This became

known as the “Dicke radiometer” and is still

the predominant design of a passive microwave

system. The SSM/I took advantage of the new

low-noise amplifiers that overcame the problem

of internal instrument noise and was built as

a “total power” passive microwave instrument.

Because of the value of SSM/I data for

research, the DMSP has released the data as

quickly as possible. The data are archived by

the National Snow and Ice Data Center. The

SSM/I data are processed for ice concentration,

ice edge, atmospheric water vapor, atmospheric

liquid water, and wind speed. Three of the four

frequencies on the SSM/I (19.7 37 85 GHz) are

dual polarized (vertical and horizontal) while

the 22 GHz frequency has only a vertical polarization.

Both polarization and frequency differences

are employed in SSM/I algorithms for

various parameters. None of the frequencies

are low enough to properly sense SST, although

some attempts were made. There were also

some methods developed to compute wind

direction for monthly composites. Ice concentration

algorithms have been based on polarization

differences in the NASA “team” algorithm

while a competing “bootstrap” ice concentration

algorithm was developed based on frequency

differences. Later the SSM/I channel brightness

temperatures have been used to compute allweather

ice motion for both polar regions. Ice

maps using SSMI have been produced routinely

for many years (Figure S16.51), and are incorporated

in operational ice analyses.

Twenty-two years of sea ice motion from

a combination of two successive passive microwave

imagers is summarized for both polar

regions in Fowler et al. (2004; see Sections 12.4

and 13.7 for the Arctic and Southern Ocean).

The motion was computed using the maximum

cross correlation (MCC) method applied to

SMMR and SSM/I passive microwave data

(Emery, Fowler, & Maslanik, 1995).

The SAR flying on Canada’s RADARSAT

satellite is also important for sea ice observations.

The primary mission of this satellite is to map

SATELLITES 71

and monitor Arctic sea ice for ship and shore

operations in the Arctic. These data have also

been used for land surface and Southern Ocean

sea ice applications. Scatterometers are also

used for tracking sea ice in the Southern Ocean.

S16.9.8. The Coastal Zone Color

Scanner and SeaWiFS

Chlorophyll content in the ocean’s surface

layer is directly related to the color at the sea

surface (Section 3.8), which can be sensed by

satellites. Chlorophyll content is related to

primary productivity. The Coastal Zone Color

Scanner (CZCS) was a pioneering instrument

supplying surface imagery in a number of

narrow visible bands, which could then be separated

into the individual color components. The

CZCS was launched in 1978 on the NASA satellite

NIMBUS-7 with an expected two-year lifetime,

but continued to provide useful image

data for 8 years, through mid-1986. In spite of

some problems with sensor drifts, the CZCS

produced some very valuable images of the

ocean’s productivity, yielding the first global

maps of ocean primary productivity, with 1 km

spatial resolution (seasonal maps in Figure 4.28).

The follow-on to the successful CZCS was the

WIde Field of view System (WiFS) built by the

former Hughes Aircraft Santa Barbara Research

Systems (now Raytheon). This instrument was

integrated with a small satellite system developed

by the Orbital Sciences Corporation, which

was then called SeaWiFS. NASA arranged with

Orbital Sciences to purchase data from this

system, but the satellite and its operation

FIGURE S16.51 (a) Sea ice concentration for January 4,

2004 (red e low; dark blue e high) from routine analysis

based on the SSMI passive microwave radiometer. Source:

From NOAA Marine Modeling and Analysis Branch (2004). (b)

Operational sea ice analysis, based on a combination of

satellite and in situ observations, for the same week for

a portion of the Barents Sea. The “egg” codes in the right

column are described on the NATICE Web site. Source: From

National Ice Center (2004).

72


belonged to and were handled by Orbital

Sciences Corporation. This NASA data purchase

applied primarily to the GAC data from Sea-

WiFS (as with the AVHRR 4 km GAC resolution)

and some limited amount of 1 km resolution

direct readout SeaWiFS data particularly for

calibration studies. Orbital Sciences sold the

rest of the 1 km data to interested users around

the world.

SeaWiFS was launched in 1997 and continues

to provide data thirteen years later. Various sites

were granted licenses to collect the direct

readout data for subsequent transfer to Orbital

Sciences Corporation. Initially for six months

after launch all of these data were free to any

user capable of receiving them. After this time

a license was required to receive and process

SeaWIFS data. This system continues to provide

valuable ocean color imagery to a wide range of

investigations. There continue to be challenges

regarding the accurate retrieval of chlorophyll

and ocean productivity particularly for what is

known as case 2 waters, which are markedly

productive coastal regions. For the weaker

productivity of the open ocean, or case 1 waters,

the algorithms all seem to successfully agree.

The launch and operation of MODIS on

NASA’s Terra (morning) and Aqua (afternoon)

satellites in 1999 and 2002 has provided ocean

color channels that have also been used successfully

for the computation of chlorophyll in the

open ocean and coastal waters (Section 4.8).

The lack of agreement of ocean color algorithms

in coastal waters is emphasized by the fact that

three different MODIS chlorophyll algorithms

are being used for case 2 waters. With the availability

of MODIS data, the future of the SeaWIFS

instrument is not clear. The MODIS instruments

is expected to be replaced by the Visible Infrared

Imaging Radiometer Suite (VIIRS), which is to

fly on the next generation of U.S. polar-orbiting

operational weather satellites formerly known

as the National Polar orbiting Operational Environmental

Satellite System (NPOESS), and now

the Joint Polar Satellite System; launch is now

scheduled for 2014 and the mission should

continue for 12 years.

Numerous other color instruments are also

on satellites, prompted by the great utility of

ocean color measurements. Several European

and Japanese color sensors were launched in

1996 and 1997, at about the same time as Sea-

WIFS. ESA’s Envisat, launched in 2002, includes

an ocean color instrument called the MEdium

Resolution Imaging Spectrometer (MERIS).

S16.9.9. Sea Surface Height: Satellite

Altimetry

One of the most important developments in

satellite oceanography is the radar altimetric

satellite. Altimeters measure the distance from

the satellite to Earth’s surface. Two major products

from these observations include: (1) maps

of surface height associated with meso- to

large-scale geostrophic circulation and associated

with changes in sea level due to thermal

expansion and changes in mass, and (2) maps

of significant wave height. If the shape of

Earth’s geoid is known, then the altimeter

measurement can be processed to yield seasurface

height. If the geoid is not precisely

known, then the altimeter measurements still

provide very accurate measurements of seasurface

height variation, since the geoid does

not change in time. With corrections to the

signal described in the next section, the accuracy

of the most recent altimeters in detecting seasurface

height is 1e2 cm.

The first test of radar altimetry on SEASAT

(Section S16.9.4) showed the possibilities for

monitoring the sea surface topography and the

estimation of surface geostrophic currents associated

with gradients of this surface topography.

The next altimeter was on the U.S. Navy’s GEO-

SAT satellite, which was launched in March

1985. After a classified geodetic mapping

mission, it was moved to the orbit previously

occupied by SEASAT and continued collecting

data until 1989 with an accuracy of 5 to 10 cm.

SATELLITES 73

In 1992 a new class of altimetric satellite,

called TOPEX/Poseidon (T/P: Figure S16.52),

was launched, with altimeters with much

greater accuracy than GEOSAT or SEASAT (~2

cm). It was designed for 3 years, but collected

high quality data for 12 years, ending in

2004. T/P carried both an American (NASA)

wave-tube altimeter and a French (CNES)

solid-state altimeter. These two altimeters

shared the same antenna and thus could not

be operated in parallel. The TOPEX altimeter

was operated about 80% of the time. In addition

to its altimeters, T/P carried a radiometer to

measure atmospheric water content for corrections

to the altimeter measurement. It also

carried very precise navigation instruments.

The high accuracy of TOPEX/Poseidon

ushered in the new era of quantitative study of

the ocean’s global eddy variability and was

pivotal for development of ocean state estimation

(combination of observations with an ocean

general circulation model). Interpretation of

other oceanographic data sets was greatly

enhanced by the availability of altimetric data.

The T/P satellite was followed by the launch

in December 2001 of the Jason-1 altimeter satellite

(Figure S16.52b). Jason-1 carries the

Poseidon-2 altimeter, which is the successor to

the Poseidon altimeter on T/P. After nine months

during which Jason-1 and T/P tracked each

other about 90 seconds apart for inter-calibration,

T/P was shifted to a new orbit halfway between

the 110 km cross-track separation of Jason-1’s 10-

day repeat orbit, which is the old track of T/P.

Called the “tandem mission,” this pair of parallel

satellites made it possible to better resolve the

mesoscale ocean surface circulation until the

end of the T/P lifetime in 2004. Jason-2 was

launched as a follow-on to Jason-1 in June 2008,

with the next generation Poseidon-3 altimeter.

Altimetric spatial resolution will be improved

with a Wide Swath Ocean Altimeter, which will

provide a two-dimensional swath of altimetric

measurements 15 km apart over a 250 km swath.

ESA launched its first altimeter mission,

ERS-1, in 1991. Its second altimeter mission,

ERS-2, was launched in 1995, with nearly identical

instrumentation except for an additional

FIGURE S16.52 Artist’s renderings of (a) TOPEX/Poseidon altimeter satellite and (b) Jason-1 altimeter satellite. Source:

From NASA/JPL-Caltech (2004a).

74


ozone sensor. The third ESA mission with a radar

altimeter, Envisat, was launched in 2002 to

continue the ERS time series. ERS-1 was retired

in 2000. The ERS accuracy is about 5 cm. Both

ERS satellites include an SAR for radar images

of Earth’s surface, a scatterometer for winds,

and radiometers for atmospheric water content.

The ERS orbits are often in a 35-day repeat cycle,

enabling higher spatial resolution than T/P, but

coarser temporal resolution at a given location.

Thus ERS and T/P complement each other,

together providing temporal and spatial

coverage. Data sets blended from T/P (now

Jason-1 and -2) and ERS altimeter data are

proving to be the most accurate in estimating

sea-surface height changes (Figure S16.53) associated

with changes in geostrophic circulation

and with changes in ocean heat content or

mass. Sorting out the source of variations

requires merging these data with in situ profile

data.

The next major goal for satellite altimetry is to

measure Earth’s geoid accurately enough that

the absolute sea-surface height can be determined

from the altimeters. Thus the geoid

(mean gravity field) must be mapped at the

same spatial scales as the oceanographic

phenomena of interest. The Gravity Recovery

and Climate Experiment (GRACE) mission

was launched in March 2002 to produce just

this map. Satellite gravity missions such as

GRACE measure the deflection of the satellite

due to the underlying gravity field. GRACE

consists of twin satellites orbiting close to each

other. Sensors on the satellites very accurately

measure the distance between them. Variations

in the gravity field are sensed by changes in

the distance between the GRACE satellites.

The satellites are also precisely navigated with

GPS. After extensive processing, a map of the

geoid is produced and resolved to 200 km

(Figure S16.54). With the advent of GRACE, it

is expected that absolute sea-surface topographies

will become routine. It should be noted

that the energetic boundary currents and mesoscale

eddies have spatial scales that are smaller

than the 200 km resolved by GRACE.

Aside from its mission to improve the accuracy

of altimetric sea-surface height observations,

GRACE has been pivotal in detecting

changes in ice sheet mass in Greenland and Antarctica.

The shrinkage in both hemispheres,

with the Antarctic record entirely resulting

from GRACE, reflects global change.

The continued success of the various altimeter

missions is allowing the physical oceanography

FIGURE S16.53 Mean sea level

anomaly (cm) from merged ERS

and T/P data (1992e1997). (Courtesy

of P. LeTraon/Envisat Data

Products.)

SATELLITES 75

wave height on the surface of the ocean. Many

experiments have been carried out to verify this

assertion.

FIGURE S16.54 Gravity anomaly map (mGal where 1

mGal ¼ 10 5 m/s 2 ) from the GRACE mission with 363 days

of observations (Gravity Model 02). Source: From NASA/

University of Texas (2004).

community to make significant developments in

tide modeling/monitoring, assessing variability

in the mean circulation, mapping global eddy

energy and planetary waves, and monitoring El

Niño events. Altimetric data are used widely in

data-assimilating models. These models then

provide a diagnosis of ocean circulation and

large-scale properties including heat content.

Altimeter data alone provide surprisingly good

constraints on the models, although in situ

profiling of temperature and salinity structure

is also needed for accuracy.

S16.9.10. Wave Height and Direction

In addition to mapping sea-surface height, the

other primary mission for the altimeters is to

provide maps of wave height and direction

(Section 8.3.2 in Chapter 8). Mean wave-height

estimates accurate to 1 m or 25% of the actual

wave height are possible from RADAR altimeter

backscatter. This measurement is made possible

by looking at the waveform returned to the satellite

from the altimeter reflection. The slope with

which it returns is a function of the significant

S16.9.11. Wind Speed and Direction:

Scatterometry

Satellite instruments can measure wind

vectors over the ocean through radar backscatter,

a technique called scatterometry. Scatterometers

are active radio frequency instruments. The scatterometer

on SEASAT used a two-stick antenna

configuration, making it possible to resolve the

wind direction within a 180 degree directional

ambiguity, which is then resolved by a knowledge

of the overall atmospheric pressure pattern. The

SEASAT scatterometer was an outstanding

success, pointing the way toward future measurements

of ocean wind speed and direction.

The first opportunity to fly a scatterometer

after the short-lived SEASAT (Section S16.9.4)

was the first Japanese ADEOS mission in 1996,

which included an instrument called NSCAT.

Unfortunately this satellite had a massive power

failure six months after it started operation. A

replacement, stand-alone satellite, QuikSCAT,

was launched quickly thereafter in 1999. The scatterometer

on QuikSCAT is known as SeaWinds.

Another SeaWinds scatterometer was launched

on the Japanese ADEOS-II mission in 2002. Sea-

Winds uses a conically scanning antenna rather

than fixed beam antennas. It measures the ocean

wind vector with an accuracy of 2 m/sec and

20 degrees (Figure S16.55).

QuikSCAT data from both satellites are

provided through NASA’s PO.DAAC (Section

S16.10).

Scatterometers are also now flying on the European

ERS-2 and Envisat satellites. Launched

before QuikSCAT, ERS-2 is a “fan-beam” antenna

system capable of resolving two components of

the wind vector. Launched after QuickSCAT,

Envisat carries a similar scatterometer.

Surface wind speed can also be inferred from

microwave brightness values in terms of the

76


FIGURE S16.55

(2004b).

Pacific wind speed and direction from NSCAT (September 21, 1996). Source: From NASA/JPL-Caltech

change in emissivity due to the surface roughness.

Accuracies are around 2.6 m/sec. The

scattering cross-section of a nadir RADAR

altimeter return also provides an estimate of

wind speed at the ocean’s surface, accurate to

around 2.0 m/sec.

S16.9.12. Other Satellite Sensors and

Derived Products

Other sensor systems flying on the meteorological

satellites are also useful, particularly in

the study of airesea interaction. A list of some

of the directly observed or derived quantities

important for physical oceanography include:

1. Radiant energy components are estimated to

about 2 W/m 2 from both the visible and

infrared radiances. Visible radiances are used

to estimate the instantaneous effects of clouds

on solar insolation to correct for the total

amount of incoming radiation from sun and

sky. Infrared imagery can be used to compute

outgoing longwave radiation (Q b ).

2. Rainfall over the ocean can be inferred from

the presence of highly reflective clouds seen

in both geostationary and polar-orbiting

satellite imagery. This provides a fairly crude

estimate because there is no definite

relationship known between the amount of

highly reflective cloud present and the level

of rainfall experienced. Correlations between

reflective cloud and in situ observations of

rainfall have suggested excellent correlations

in the tropical regions, but such studies have

not been as successful in higher latitude

regions. Rainfall can also be estimated

directly from microwave radiances as cloud

liquid water. TRMM integrates an onboard

active radar with passive microwave

instruments to estimate rainfall and rainfall

rates over the tropical ocean.

3. Atmospheric water vapor can be directly

measured as a vertical integral by microwave

channels or can be computed from a moisture

SATELLITES 77

profile derived from primarily infrared

channels. The TIROS Operational Vertical

Sounder (TOVS) on the NOAA polar-orbiting

weather satellites uses a combination of

infrared and microwave channels to measure

atmospheric moisture and temperature

profiles. Both of the microwave and infrared

methods produce atmospheric water vapor

values accurate to around 2 g/cm 3 ; the

microwave data are cloud independent while

the infrared sensors are limited by the

amount of cloud cover. Since 1999 the

weather satellites carry a profiling

radiometer called the Advanced Microwave

Sensor Unit (AMSU). As part of NASA’s EOS

program the Atmospheric Infrared Radiation

Spectrometer (AIRS) flies on the afternoon

Aqua satellite. AIRS has a large number of

narrow infrared channels and is capable of

observing very highly resolved atmospheric

temperature and water vapor profiles.

4. Upwelling events in the sea can be located

and monitored by both their surface thermal

signatures and their expression by increased

primary productivity in ocean color imagery.

One must be careful to separate the in situ

effects such as plankton blooms and heating

and cooling from changes due to upwelling

alone.

5. Currents can be estimated from the

displacement of features in sequential

imagery. The first studies used the visual

method of feature tracking (Vastano &

Borders, 1984) while subsequent efforts

computed the maximum cross-correlation

location between images to estimate advective

displacements between the images (Emery et

al., 1986). This same procedure can be applied

to sequences of visible satellite images to

compute the motion of sea ice (Ninnis, Emery,

&Collins,1986). Applied to highly resolved

sea ice SAR imagery, this method produces

very detailed maps of the ice displacements

(Collins & Emery, 1988). Applied to the cloudindependent

SMMR and the new SSM/I data,

this method can be used to produce allweather

images of sea ice displacement. The

same technique can be applied to sequences of

ocean color imagery to estimate surface

advection (Bowen et al., 2002; Wilkin, Bowen,

&Emery,2002). It is important to remember

that all of these measurements are from

remotely sensing platforms and thus are not

direct observations of a variable. Thus

considerable validation with “ground-truth”

data is needed to determine the levels of

accuracy and reliability possible with the

remote measurements. Still, the advantages of

synoptic sampling of large regions make it

worthwhile to understand and use these data.

We must remember in this validation exercise

that the satellite does not often sense the same

thing that we measure in situ. For example,

infrared satellite sensors are only capable of

viewing the micron thick “skin” of the ocean

while the in situ ship and buoy SST

measurements represent temperatures from 1

to 5 m in depth. In addition a satellite infrared

measurement represents the temperature

measured over 1 km square of the ocean

surface (its “footprint”) while the ship or buoy

SST is for a single “spot” in the ocean. This is

true of most other comparisons that we make

between satellite and in situ measurements,

becoming an even greater problem when

dealing with passive microwave systems with

larger footprints. An example is that a moored

buoy measurement of wind speed represents

a single spot (or the minimum area of a couple

of meters) while the passive microwave spot

size may range from 12 to 50 km.

S16.9.13. Satellite Communications and

Navigation

S16.9.13.1. Satellite Communication

Satellites are important for communications

with autonomous instruments and for navigation,

as well as for remote sensing. The French

system Argos has provided communications

78


with instrumented platforms such as surface

drifters and pop-up floats for many years. Argos

is capable of accurately (1 km) locating the

buoy in the ocean using the Doppler shift of

the transmitted signal; the Argos system can

also receive data from the buoys at a limited

rate of up to 32 data words per transmission.

Other satellites with higher data transmission

rates, such as the GOES satellites, are gradually

coming into use.

Recently constellations of small polar-orbiting

satellites have been set up to provide global satellite

telecommunication. Best known is Iridium

System, originally conceived of and built by

Motorola Corporation. Intended to provide

global telecommunications with a series of satellite

shells with up to 80 satellites, the system cost

greatly exceeded company estimates. The high

cost of the ground units and the communication

charges soon led to bankruptcy. The U.S. military

purchased the system and now operates it at

a profit, providing global wide bandwidth telecommunication

for military and commercial

users. Due to the polar orbit, connectivity is not

latitude dependent. (Geostationary systems can

communicate easily with the lower latitudes but

have problems poleward of 60 latitude.) The

system can be adapted to transfer more data

from buoys and floats than is possible with the

Argos system. For these applications it is critical

that the in situ platform be equipped with a GPS

receiver for accurate geographic location.

S16.9.13.2. Satellite Navigation

Determining the location of ships has been

another important function of satellites since

the early 1970s. The earliest system, NAVSTAR,

used a single shell of polar-orbiting satellites to

determine the ship’s location when the satellite

passed overhead, based on the Doppler shift of

the radio signal from the ship to the satellite.

NAVSTAR was supported and operated by the

U.S. Navy but was available to ships from all

nations. Commercial receiving units quickly

developed into low-cost systems for accurate

positioning. Coupled with good estimates of

ship’s speed and heading, NAVSTAR provided

an excellent means for mapping the course of

a vessel at sea. Used in conjunction with

a shorter range system, such as LORAN (short

range radio navigation based on beacons

installed in coastal regions), the satellite navigation

system provided a very precise check of

ship geolocation. At low latitudes, satellite position

fixes were possible every couple of hours

depending on the number of satellites in operation.

At higher latitudes, where the orbits of the

polar-orbiting satellites nearly overlap, fixes

were much more frequent.

The widely used GPS replaced NAVSTAR

beginning in 1993. GPS was developed and is

operated by the U.S. Department of Defense

(DoD), which provides very precise, accurate

geographic and time information. Prior to

2000, GPS was operated in two modes: higher

accuracy for military users (Precise Positioning

Service, PPS) and a degraded signal (Standard

Positioning Service, SPS) for all other users.

After 2000, SPS was discontinued and all signals

now are of the higher accuracy.

To provide continuous access to users, GPS

uses six shells of navigational satellites (Figure

S16.56) to at least three satellites simultaneously,

and usually uses five to eight (Dana, 1999,

accessed 2009). The minimum system has 24

satellites, although often there are more as new

ones are launched as replacements. The satellites

are in daily-repeating orbits. Accuracy in

both the horizontal and vertical directions is

100 m for SPS and 22 m for PPS. Thus GPS

can be used for aircraft navigation to report altitude

as well as geographic location. GPS can

also be used to accurately determine the time

(200 ns for PPS).

GPS signals are processed in a GPS receiver,

enabling the receiver to compute position,

velocity, and time. Receivers are made for

aircraft, ships, vehicles, and for individuals.

“Differential” GPS (DGPS) is a method of

greatly improving SPS accuracy to up to +10 cm

DATA ARCHIVES AND DATA CENTERS 79

(a)

(b)

GPS Nominal Constellation

24 satellites in 6 orbital planes, 4 satellites in each plane

20,200 km altitude, 55° inclination

(after Dana, 1998)

FIGURE S16.56 (a) GPS satellite orbits and (b) GPS contacts with receivers. Source: After Dana (1999).

by using GPS reference stations with precisely

known positions. These stations broadcast

information on the error of the satellite locations.

DGPS was originally introduced to reduce the

large errors of the SPS system (prior to 2000),

but it improves on even the more precise PPS.

The Russian global navigation system (GLO-

NASS) consists of 21 satellites in 3 shells. It

became operable in the mid-1990s but then fell

into disrepair. ESA plans its own network of

global navigation satellites called Galileo. It

will include 30 satellites in 3 shells and will be

interoperable with GPS, and is intended to

be more accurate than GPS. It is planned to be

operational in 2013.

S16.10. DATA ARCHIVES AND

DATA CENTERS

Oceanographic data are archived in various

data centers. Most countries have a central

data archive. The primary international repositories

for in situ data are the three World Data

Centers for Oceanography located in the United

States, Russia, and China, under the umbrella of

the International Council for Science (ICSU).

The World Data Center in the United States is

NOAA’s National Oceanographic Data Center

(NODC), located in Silver Springs, Maryland.

Sea ice information, including satellite information,

is archived in the United States at the

National Snow and Ice Data Center.

Satellite data are so voluminous that many

different data centers have been set up to

archive and disseminate these data. NASA’s

EOS includes Data Active Archive Centers

(DAACs). PO.DAAC (http://podaac.jpl.nasa.

gov/) is located at NASA’s Jet Propulsion Laboratory

(JPL) in Pasadena, California. In conjunction

with the French space agency (CNES),

PO.DAAC processes, archives and disseminates

all altimetry data and data products. PO.DAAC

also handles NASA scatterometer data products

80


and SST products derived from infrared satellite

imagery as well as some other smaller data sets.

NOAA provides its satellite data through its

NOAA Satellites and Information (NOAASIS)

Web site. The final archive of all U.S. weather

satellite data as well as for NASA EOS data is

NOAA’s National Climatic Data Center

(NCDC), located in Asheville, North Carolina.

The “satellite active archive” at NCDC provides

wide access to environmental satellite data. ESA

provides access to satellite SST and ocean color

data through its Ionia Web site.

The Global Ocean Observing System (GOOS;

http://www.ioc-goos.org/) is part of the Global

Earth Observing System of Systems (GEOSS).

GOOS provides a framework for different types

of oceanographic data centralization and products.

This differs from data archiving, which is

mainly covered by the oceanographic data

centers such as those listed above.

References

Aanderaa Instruments, 2000. Recording current meter

RCM7 and RMC8. Aanderaa Instruments, <http://

www.aanderaa.com> (accessed 5.13.04). (Link to

RCM7-8 information is no longer active 6.2.09.)

Argo 3000, 2007. Argo 3000. <http://www-argo.ucsd.edu/

FrArgo_3000.html> (accessed 2.18.09).

Avsic, T., Send, U., Skarsoullis, E., 2005. Six years of

tomography observations in the central Labrador Sea.

J. Acoust. Soc. Am. 107, 28e34.

Beardsley, R., 1987. A Comparison of the Vector-Averaging

Current Meter and New Edgerton. Germeshausen, and

Grier, Inc., Vector-Measuring Current Meter on a Surface

Mooring in Coastal Ocean Dynamics Experiment 1.

J. Geophys. Res. 92, 1845e1859.


Nobre, P., Campos, E., Yu, L., Planton, S., Busalacchi, A.,

Moura, A.D., Servain, J., Trotte, J., 2008. The Pirata

Program: History, accomplishments, and future directions.

Bull. Am. Meteor. Soc. 89, 1111e1125.

Bowen, M.M., Emery, W.J., Wilkin, J.L., Tildesley, P.L.,

Barton, I.J., Knewtson, R., 2002. Multiyear surface

currents from sequential thermal imagery using the

maximum cross-correlation technique. J. Atm. Ocean.

Tech. 19, 1665e1676.

Carlson, R.E., 2011. The Secchi Disk. Secchi Dip-In. http://

www.secchidipin.org/secchi.htm (accessed 3.23.11).

Cartwright, D.E., 1999. Tides: A Scientific History. Cambridge

University Press, UK, 292 pp.

Cheney, R.E., Marsh, J.G., Beckley, B.D., 1983. Global

mesoscale variability from collinear tracks of SEASAT

altimeter data. J. Geophys. Res. 88, 4343e4354.

Collins, M.J., Emery, W.J., 1988. A computational method for

estimating sea ice motion in sequential SEASAT

synthetic aperture radar imagery by matched filtering.

J. Geophys. Res. 93, 9241e9251.

Cornuelle, B., Wunsch, C., Behringer, D., Birdsall, T.,

Brown, M., Heinmiller, R., Knox, R., Metzger, K., Munk, W.,

Spiesberger, J., Spindel, R., Webb, D., Worcester, P., 1985.

Tomographic maps of the ocean mesoscale. Part 1: Pure

acoustics. J. Phys. Oceanogr. 15, 133e152.

Dana, P.H., 1999. The Geographer’s Craft Project. Department

of Geography, The University of Colorado at

Boulder, developed 1994, copyright 1999. <http://www.

colorado.edu/geography/gcraft/notes/gps/gps.html >

(accessed 2.18.09).

Davis, R.E., Killworth, P.D., Blundell, J.R., 1996. Comparison

of autonomous Lagrangian circulation explorer and fine

resolution Antarctic model results in the South Atlantic.


Dietrich, G., Kalle, K., Krauss, W., Siedler, G., 1980. General

Oceanography. Ulrich Roll, S., Ulrich Roll, H.U., (Trans.),

second ed. Wiley, New York (Wiley-Interscience), 626 pp.

Dushaw, B.D., 2002. Acoustic Thermometry of Ocean

Climate (ATOC). Applied Physics Laboratory, University

of Washington. <http://staff.washington.edu/

dushaw/atoc.html> (accessed July, 2008).

Dushaw, B.D., 2003. Acoustic thermometry in the North

Pacific. CLIVAR Exchanges 26, 5 pp.

Emery, W., Fowler, C., Maslanik, J., 1995. Satellite remote

sensing of ice motion. In: Ikeda, M., Dobson, F.W. (Eds.),

Oceanographic Applications of Remote Sensing. CRC

Press, Boca Raton, Florida, 492 pp.

Emery, W.J., Cherkauer, K., Shannon, B., Reynolds, R.W.,

1997. Hull mounted bulk sea surface temperature

measurements from volunteer observing ships. J. Atm.

Ocean. Tech. 14, 1237e1251.

Emery, W.J., Thomas, A.C., Collins, M.J., Crawford, W.R.,

Mackas, D.L., 1986. An objective method for

computing advective surface velocities from sequential

infrared satellite images. J. Geophys. Res. 91, 12,

865e12,878.

Emery, W.J., Thomson, R.E., 2001. Data Analysis Methods in

Physical Oceanography, second ed. Elsevier, Amsterdam,

638 pp.

Esterson, G.L., 1957. The induction conductivity indicator:

a new method for conductivity measurement at sea.

Chesapeake Bay Institute Technical Report, 57-3, 183 pp.

Faraday, M., 1832. The Bakerian Lecture d Experimental

researches in electricity. Second Series. Sec. 5. Terrestrial

REFERENCES 81

magneto-electric induction. Philos. T. Roy. Soc. London

Part I, 163e177.

Flosadottir, A., 2004. Research cables around the world.

NOAA/PMEL. <http://www.pmel.noaa.gov/wbcurrents>

(accessed 5.13.04).

Folland, C.K., Parker, D.E., 1995. Correction of instrumental

biases in historical sea surface temperature data. Q.J.

Roy. Meteor. Soc. 121, 319e367.

Fowler, C., Emery, W.J., Maslanik, J.A., 2004. Satellitederived

evolution of Arctic sea ice age: October 1978 to

March 2003. IEEE Geosci. Rem. Sens. Lett. 1, 71e74.

Fu, L-L., Holt, B., 1982. SEASAT views oceans and sea ice

with synthetic aperture radar. Jet Propulsion Laboratory

publication 81e120. Pasadena, CA, 200 pp.

General Oceanics, 2009. MK3C/WOCE CTD. <http://

www.generaloceanics.com/genocean/mk3c.htm>

(accessed 5.29.09).

Gloersen, P., Campbell, W.J., Cavalieri, D.J., Comiso, J.C.,

Parkinson, C.L., Zwally, H.J., 1992. Arctic and Antarctic

Sea Ice, 1978e1987: Satellite Passive-Microwave Observations

and Analysis, Spec. Publ., 511, 290 pp. NASA,

Washington, D. C.

Guildline, 2009. Guildline8400BDatasheet pdf. Guildline

Instruments. <http://www.guildline.com/Datasheet/

Guildline8400BDatasheet.pdf> (accessed 6.1.09).

Hamon, B.V., 1955. A temperature-salinity-depth recorder.

Conseil Permanent International pour l’Exploration de

la Mer. J. Conseil 21, 22e73.

Hamon, B.V., Brown, N.L., 1958. A temperature-chlorinitydepth

recorder for use at sea. J. Sci. Instrum. 35, 452e458.

Howe, B., Chereskin, T.K., 2007. Oceanographic Measurements.

In: Tropea, C., Yarin, A.L., Foss, J.F. (Eds.),

Springer Handbook of Experimental Fluid Mechanics.

Springer, Berlin, pp. 1179e1271.

Howe, B., Worcester, P., Spindel, R., 1987. Ocean Acoustic

Tomography: Mesoscale Velocity. J. Geophys. Res. 92,

3785e3805.

ICPC, 2007. International Cable Protection Committee.

<http://www.iscpc.org/cabledb/Scientific_Cable_db.

htm> (accessed 6.1.09).

InterOcean Systems, 2011. S4 current meter family. Inter-

Ocean Systems, Inc. <http://www.interoceansystems.

com/s4main.htm> (accessed 3.21.11).

Larsen, J.C., Sanford, R.B., 1985. Florida Current volume

transports from voltage measurements. Science 227,

302e304.


slow mixing across the pycnocline from an open-ocean

tracer-release experiment. Nature 364, 701e703.

Lewis, E.L., 1980. The Practical Salinity Scale 1978 and its

antecedents. IEEE. J. Oceanic Eng. OE-5, 3e8.

Longuet-Higgins, M.S., Stern, M.E., Stommel, H.M., 1954.

The electrical field induced by ocean currents and

waves, with applications to the method of towed electrodes.

Papers in Phys. Oceanogr. and Met. MIT and

Woods Hole Oceanogr. Inst. 13 (1), 1e37. <http://hdl.

handle.net/1912/1064>.

Mariano, A.J., Ryan, E.H., Perkins, B.D., Smithers, S., 1995.

The Mariano Global Surface Velocity Analysis 1.0. USCG

Report CG-D-34e95, 55 pp. <http://oceancurrents.

rsmas.miami.edu/index.html>(accessed 3.4.09).

Marine Physical Lab, SIO, 2009. Floating Instrument Platform-

FLIP. Marine Physical Lab, SIO, University of

California, San Diego. 2003e2009. <http://www-mpl.

ucsd.edu/resources/flip.intro.html> (accessed 5.29.09).

McPhaden, M.J., Meyers, G., Ando, K., Masumoto, Y.,

Murty, V.S.N., Ravichandran, M., Syamsudin, F.,

Vialar, J., Yu, L., Yu, 2009. W., RAMA: The research

moored array for African-Asian-Australian monsoon

analysis and prediction. Bull. Am. Meteorol. Soc. 90,

459e480.

Morawitz, W.M.L., Sutton, P.J., Worcester, P.F.,

Cornuelle, B.D., Lynch, J.F., Pawlowicz, R., 1996. Threedimensional

observations of a deep convective chimney

in the Greenland Sea during winter 1988/1989. J. Phys.

Oceanogr. 26, 2316e2343.

Munk, W., Worcester., P., Wunsch, C., 1995. Ocean

Acoustic Tomography. Cambridge Monographs on

Mechanics. Cambridge University Press, Cambridge,

UK, 447 pp.

Munk, W., Wunsch, C., 1979. Ocean acoustic tomography:

a scheme for large scale monitoring. Deep-Sea Res. 26,

123e161.

NASA GSFC, 2004. Remote Sensing Tutorial. NASA Goddard

Space Flight Center. <http://rst.gsfc.nasa.gov>


NASA/JPL-Caltech, 2004a. Ocean surface topography from

space d Missions. NASA Jet Propulsion Laboratory.

<http://topex-www.jpl.nasa.gov/mission/mission.html>;

(accessed 5.11.04).

NASA/JPL-Caltech, 2004b. WINDS home page. NASA Jet

Propulsion Laboratory. <http://winds.jpl.nasa.gov/>

(accessed 5.04).

NASA/University of Texas, 2004. GRACE gravity model d

gravity recovery and climate experiment gravity model.

University of Texas at Austin Center for Space Research.

<http://www.csr.utexas.edu/grace/gravity/> (accessed

5.04).

National Ice Center, 2004. <http://www.natice.noaa.gov>

(accessed 5.11.04).

Nerem, R.S., 2009. Global mean sea level tide gauges.

University of Colorado at Boulder. <http://sealevel.

colorado.edu/tidegauges.php> (accessed 6.1.09).

Neumann, G., Pierson, W.J., 1966. Principles of Physical

Oceanography. Prentice-Hall, Englewood Cliffs, N.J.

545 pp.

82


Ninnis, R.M., Emery, W.J., Collins, M.J., 1986. Automated

extraction of pack ice motion from advanced very high

resolution radiometer imagery. J. Geophys. Res. 91,

10725e10734.

NOAA Global Drifter Program, 2009. The Global Drifter

Program. NOAA AOML GDP. <http://www.aoml.

noaa.gov/phod/dac/gdp_information.html> (accessed

6.1.09).

NOAA Marine Modeling and Analysis Branch, 2004.

MMAB sea ice analysis page. NOAA, Marine Modeling

and Analysis Branch. <http://polar.ncep.noaa.gov/

seaice/Analyses.html> (accessed 5.11.04).

NOAA NDBC, 2008. Moored buoy program. NOAA

National Data Buoy Center. <http://www.ndbc.noaa.

gov/mooredbuoy.shtml> (accessed 2009).

NOAA NESDIS, 2009. Geostationary Satellite Server,

NOAA Satellite and Information Service. <http://www.

goes.noaa.gov> (accessed 6.2.09).

NOAA PMEL, 2009a. Impacts of El Niño and benefits of El

Niño prediction. <http://www.pmel.noaa.gov/tao/

elnino/impacts.html> (accessed 3.26.09).

NOAA PMEL, 2009b. The TAO project. TAO Project Office,

NOAA Pacific Marine Environmental Laboratory.

<http://www.pmel.noaa.gov/tao/> (accessed 6.1.09).

NOAA UOTC, 2009. The expendable Bathythermograph

(XBT). NOAA Upper Ocean Thermal Center. <http://

www.aoml.noaa.gov/goos/uot/xbt-what-is.php>

(accessed 6.1.09).

Ocean World, 2009. Buoys, tide gauges, Nansen bottles and

the like. Ocean World. <http://oceanworld.tamu.edu/

students/satellites/satellite2.htm>, (accessed 6.1.09).

Paroscientific, Inc., 2009. Digiquartz High Pressure

Sensor Design. <http://www.paroscientific.com/pdf/

DQAdvantage.pdf>. (accessed 5.29.09).

Pinkel, R., 1979. Observation of strongly nonlinear motion

in the open sea using a range-gated Doppler sonar. J.


Preisendorfer, R.W., 1986. Secchi disk science: Visual optics

of natural waters. Limn. Oceanogr. 31, 909e926.

Reynolds, R.W., 1988. A real-time global sea surface

temperature analysis. J. Clim. 1, 75e87.

Reynolds, R.W., Smith, T.M., 1994. Improved global sea

surface temperature analyses using optimum interpolation.

J. Clim. 7, 929e948.

Reynolds, R.W., Smith, T.M., 1995. A high-resolution global

sea surface temperature climatology. J. Clim. 8, 1571e1583.

Richardson, M.J., Gardner, W.D., 1997. Tools of the trade.

Quarterdeck 5(1), Texas A&M University Department

of Oceanography. <http://oceanography.tamu.edu/

Quarterdeck/QD5.1/qdhome-5.1.html> (accessed 2.19.09).

Rossby, T., 1969. On monitoring depth variations of the

main thermocline acoustically. J. Geophys. Res. 74,

5542e5546.

Rossby, T., 2007. Evolution of Lagrangian methods in

oceanography. In: Griffa, A., Kirwan, A.D.,

Mariano, A.J., Özgökmen, T., Rossby, H.T. (Eds.),

Lagrangian analysis and prediction of coastal and ocean

dynamics. Cambridge University Press, pp. 1e39.

Rossby, T., Dorson, D., Fontaine, J., 1986. The RAFOS

system. J. Atm. Ocean. Tech. 3, 672e679.

Rossby, T., Webb, D., 1970. Observing abyssal motion by

tracking Swallow floats in the SOFAR channel. Deep-Sea

Res. 17, 359e365.

Rowe, F., Young, J., 1979. An ocean current profiler using

Doppler sonar. IEEE Proc. Oceans 79, 292e297.

Scripps Institution of Oceanography, 2009. R/V Melville

Photos. Scripps Institution of Oceanography, UCSD.

<http://shipsked.ucsd.edu/Ships/Melville/(accessed>

6.1.09).

Sea-Bird Electronics, Inc., 2009a. SEACAT Thermosalinograph.

<http://www.seabird.com/products/ThermoS.

htm> (accessed 5.29.09).

Sea-Bird Electronics, Inc. 2009b. Profilers. <http://www.

seabird.com/products/profilers.htm> (accessed 5.29.09).

Secchi Dip-In, 2011. Secchi disk. Secchi Dip-In, Department

of Biological Sciences, Kent State University. <http://

www.secchidipin.org/secchi.htm> (accessed 3.23.11).

Send, U., Schott, F., Gaillard, F., Desaubies, Y., 1995.

Observation of a deep convection regime with acoustic

tomography. J. Geophys. Res. 100, 6927e6941.

Stewart, R.H., 1985. Methods of Satellite Oceanography.

University of California. Press, Berkeley.

Strickland, J.D.H., Parsons, T.R., 1972. A Practical Handbook

of Sea-Water Analysis (second ed.). Fish. Res. Bd.

Can. Bull. 167, 311 pp.

Swallow, J.C., 1955. A neutral-buoyancy float for measuring

deep currents. Deep-Sea Res. 3 (1), 93e104.

TAO, 2009. The TAO project. TAO Project Office, NOAA

Pacific Marine Environmental Laboratory. <http://

www.pmel.noaa.gov/tao/> (accessed 6.1.09).

Teledyne RD Instruments, 2011. Teledyne RDI marine

measurements. Teledyne RDI. <http://www.

rdinstruments.com/sen.aspx> (accessed 3.21.11).

TriOS, 2009. TriOS optical sensors dRamses. TriOS Gmbh,

<http://www.trios.de> (accessed 2.19.09).

Tyler, J.E., 1968. The Secchi disc. Limnology and Oceanogr.

13, 1e6.

UNESCO, 1981. The Practical Salinity Scale 1978 and the

International Equation of State of Seawater 1980. Tech.

Pap Mar., Sci. 36, 25 pp.

University of Rhode Island Graduate School of Oceanography,

2009. RAFOS float group home page. <http://

www.po.gso.uri.edu/rafos/> (accessed 5/29/09).

U.S. Argo Program, 2009. Argo home page. Scripps Institution

of Oceanography. University of California, San Diego.

<http://www-argo.ucsd.edu/> (accessed 02.18.09).

REFERENCES 83

USGS, 2005. Instrumentation. U.S. Department of the Interior,

U.S. Geological Survey. <http://pubs.usgs.gov/

dds/dds74/WEBPAGES/instrumentation.html>

(accessed 6.1.09).

Vastano, A.C., Borders, S.E., 1984. Sea surface motion over

an anticyclonic eddy on the Oyashio Front. Rem. Sens.

Environ. 16, 87e90.

Von Arx, W.S., 1950. An electromagnetic method for

measuring the velocities of ocean currents from a ship

under way. Pap. Phys. Oceanogr. Meteor. MIT and

Woods Hole Oceanogr. Inst. 11 (3), 1e62.

Watts, D.R., Rossby, H.T., 1977. Measuring dynamic heights

with inverted echo sounders: Results from MODE. J.


Weller, R.A., Davis, R.E., 1980. A vector measuring current

meter. Deep-Sea Res. 27, 565e582.

Wertheim, G.K., 1954. Studies of the electrical potential

between Key West, Florida and Havana, Cuba. Trans.

Am. Geophys. Union 35, 872e882.

WHOI Image Galleries, 2009. Image of the Day d October 6,

2006. Woods Hole Oceanographic Institution. <http://

www.whoi.edu/page.do?pid¼10897&i¼588&x¼184>

(accessed 6.1.09).

Wilkin, J.L., Bowen, M., Emery, W.J., 2002. Mapping

mesoscale currents by optimal interpolation of satellite

radiometer and altimeter data. Ocean Dynam. 52,

95e103.

WMO JCOMM, 2009. The Global Sea Level Observing System

(GLOSS), World Meteorological Organisation, Joint Technical

Commission for Oceanography and Marine Meteorology.

<http://www.gloss-sealevel.org/> (accessed

5.29.09).

WOCE, 2009. Drifters. World Ocean Circulation Experiment

Global Data Resource. <http://woce.nodc.noaa.gov/

wdiu/index.htm> (accessed 6.1.09).

Young, F.B., Gerard, H., Jevons, W., 1920. On electrical

disturbances due to tides and waves. Philos. Mag. Series

6 (40), 149e159.

Zhang, H.-M., Prater, M., Rossby, T., 2001. Isopycnal

Lagrangian statistics from the North Atlantic Current

RAFOS float observations. J. Geophys. Res. 106,

13817e13836.

Color Plates

FIGURE 1.1 (a) Sea surface temperature from a satellite advanced very high resolution radiometer (AVHRR) instrument

(Otis Brown, personal communication, 2009).

FIGURE 2.4 Seafloor topography for a portion of (a) the fast-spreading EPR and (b) the slow-spreading MAR. Note the

ridge at the EPR spreading center and rift valley at the MAR spreading center. (Sandwell, personal communication, 2009.)

(b)

FIGURE 4.1 (b) Satellite infrared sea surface temperature ( C; nighttime only), averaged to 50 km and 1 week, for

January 3, 2008. White is sea ice. (See Figure S4.1 from the online supplementary material for an austral winter image from

July 3, 2008). Source: From NOAA NESDIS (2009b).

FIGURE 4.4 Mixed layer depth in (a) January and (b) July, based on a temperature difference of 0.2 C from the nearsurface

temperature. Source: From deBoyer Montégut et al. (2004). (c) Averaged maximum mixed layer depth, using the 5

deepest mixed layers in 1 1 bins from the Argo profiling float data set (2000e2009) and fitting the mixed layer structure

as in Holte and Talley (2009).

(a)

0

1000

2000

3000

1

0

1

2

3

4

5

10

3

15 20

4

10

5

3

4

5

(b)

0

1000

2000

3000

34.7

34.3

34.7

34.5

34.9

37

36

35

36

34.9

4000

0

1

2

2

4000

34.9

5000

5000

0

1

Atlantic

Atlantic

θ

Salinity 34.7

34.7 34.7

6000

6000

0 2000 4000 6000 8000 10000 12000 14000 km 0 2000 4000 6000 8000 10000 12000 14000 km

60°S 40°tS 20°S 0° 20°N 40°N 60°N

60°S 40°tS 20°S 0° 20°N 40°N 60°N

(c)

0

1000

2000

3000

4000

5000

27.8

46.1

46

46.1

27.5

27

60°S 40°tS 20°S 0° 20°N 40°N 60°N

27.8

26

45.7

45.8 45.8

45.9

46

46

0

1000

2000

3000

4000

5000

200

220

240

180

180

220

200

220

240

220

140

200 180

200

Atlantic

Atlantic

σ θ and σ 4

46.1

Oxygen

6000

6000

0 2000 4000 6000 8000 10000 12000 14000 km 0 2000 4000 6000 8000 10000 12000 14000 km

(d)

240

260

80

240

220

240

260

220

240

240260

60°S 40°tS 20°S 0° 20°N 40°N 60°N

FIGURE 4.11 (a) Potential temperature ( C), (b) salinity (psu), (c) potential density s q (top) and potential density s 4 (bottom) (kg m 3 ), and (d)

oxygen (mmol/kg) in the Atlantic Ocean at longitude 20 to 25 W. Data from the World Ocean Circulation Experiment.

260

280

(a)

0

1000

2000

3000

4000

5000

Pacific

θ

6000

0 2000 4000 6000 8000 10000 12000

(c)

0

1000

1

2 2

1

1.5

5

10

15 20

3 3

60°S 40°S 20°S 0° 20°N 40°N

27

27.5

1

26

5

4

1.5

2

km

(b)

1000

0

1000

0

34.6

2000 34.73

3000

4000

34.4

34.65

34.7

34.68

34.5

34.6

34.4

34.65

34.7

5000

Pacific

Salinity

34.7

6000

0 2000 4000 6000 8000 10000 12000

(d)

34.7

34.6

34.6

34.65

34.5

34.4

34.3

34.68

60°S 40°S 20°S 0° 20°N 40°N

180

260

220

180

160

40

100

80

40

34

km

2000

3000

4000

46.05

45.95

45.85

45.9

45.8

45.75

5000

45.95

Pacific

σ θ and σ 4 45.95

45.9

6000

0 2000 4000 6000 8000 10000 12000

60°S 40°S 20°S 0° 20°N 40°N

45.9

45.7

45.8

45.85

2000

3000

4000

200

160

180

80

100

160

5000

180

Pacific

Oxygen

6000

km 0 2000 4000 6000 8000 10000 12000

140

160

180

60°S 40°S 20°S 0° 20°N 40°N

120

140

160

140

km

FIGURE 4.12 (a) Potential temperature ( C), (b) salinity (psu), (c) potential density s q (top) and potential density s 4

(bottom; kg m 3 ), and (d) oxygen (mmol/kg) in the Pacific Ocean at longitude 150 W. Data from the World Ocean Circulation

Experiment.

(a)

0

1000

2000

2

2

3

10

2

15 20 25

5

4

3

(b)

0

1000

2000

34.73

34.3

34.6

34.7

34.5

35.4

34.6

34.9

35

34.8

3000

1

3000

34.7

34.73

34.73

4000

0

1

4000

5000

0

1000

2000

3000

4000

27.8

46.1

46.0

46

45.9

0

Indian

Indian

q

Salinity

6000

6000

0

2000 4000 6000 8000 km 0 2000 4000 6000 8000 km

60°S 50°S 40°S 30°S 20°S 10°S 0° 10°N 60°S 50°S 40°S 30°S 20°S 10°S 0° 10°N

(c)

27

46.1

5000

46

Indian

s q and s 4

6000

0 2000 4000 6000 8000 km

60°S 50°S 40°S 30°S 20°S 10°S 0° 10°N

27.5

45.8

45.9

26

27

24

5000

(d)

0

1000

2000

3000

4000

5000

6000

200

0

220

Indian

Oxygen

220

240

180

40 80 0

200

200

180

180

240

40 80 0

34.7

220

200

200

180

120

100 80

0 2000 4000 6000 8000 km

60°S 50°S 40°S 30°S 20°S 10°S 0° 10°N

FIGURE 4.13 (a) Potential temperature ( C), (b) salinity (psu), (c) potential density s q (top) and potential density s 4 (bottom; kg m 3 ), and (d) oxygen

(mmol/kg) in the Indian Ocean at longitude 95 E. Data from the World Ocean Circulation Experiment.

180

140

140

80

120

160

80 40 0

40

100

(a)

0

1000

2000

3000

4000

130

100

5000

30

5000

Atlantic

120

Atlantic

20

Nitrate

6000

6000

Silticate

0 2000 4000 6000 8000 10000 12000 14000 km 0 2000 4000 6000 8000 10000 12000 14000 km

(c)

0

1000

2000

3000

4000

32.5

32.5

32.5

32.5

32.5

5000

Pacific

35

Nitrate

6000

0 2000 4000 6000 8000 10000 12000 km

60°S 40°S 20°S 0° 20°N 40°N

(e)

0

1000

2000

3000

35

30

30

32.5

25

32.5

25

4000

32.5

-10

20 30 1

10 20 30 1

32.5 -10

20 30 1

5000

Indian

Nitrate

6000

0 2000 4000 6000 8000 km

60°S 50°S 40°S 30°S 20°S 10°S 0° 10°N

20

20

35

35

32.5 30

25

20

60°S 40°S 20°S 0° 20°N 40°N 60°N

10 1

20 35

10

20

40

32.5

35

1

32.5

30

30

35

32.5

35

32.5

37.5

10

42.5

1

20

40 42.5

35

1

20

30

40

37.5

35

10

15

37.5

0

1000

2000

3000

4000

0

1000

2000

3000

4000

5000

6000

0

1000

2000

3000

4000

120

120

130

120

100

Pacific

Silicate

80

120

60

80

5

10

60°S 40°S 20°S 0° 20°N 40°N 60°N

100

5

10

80

40

60

120

130

130

10 20 40 60 80 50

50

100 120 40 80 60

5

5000

Indian

Silicate

6000

0 2000 4000 6000 8000 km

60°S 50°S 40°S 30°S 20°S 10°S 0° 10°N

40

60

140

0 2000 4000 6000 8000 10000 12000 km

60°S 40°S 20°S 0° 20°N 40°N

130

50

80

100

120

130

60

80

120

100

5

10

20

40

80

100

120

140

20

80

100

130

150

150

130

40

40

160

150

20

10

20

170

5 20

40

FIGURE 4.22 Nitrate (mmol/kg) and dissolved silica (mmol/kg) for the Atlantic Ocean (a, b), the Pacific Ocean (c, d), and

the Indian Ocean (e, f). Note that the horizontal axes for each ocean differ. Data from the World Ocean Circulation

Experiment.

(b)

(d)

(f)

140

160

10

5

20

FIGURE 4.28 Global images of chlorophyll derived from the Coastal Zone Color Scanner (CZCS). Global phytoplankton

concentrations change seasonally, as revealed by these three-month “climatological” composites for all months between

November 1978eJune 1986 during which the CZCS collected data: JanuaryeMarch (upper left), AprileJune (upper right),

JulyeSeptember (lower left), and OctobereDecember (lower right). Note the “blooming” of phytoplankton over the entire

North Atlantic with the advent of Northern Hemisphere spring, and seasonal increases in equatorial phytoplankton

concentrations in both Atlantic and Pacific Oceans and off the western coasts of Africa and Peru. See Figure S4.2 from the

online supplementary material for maps showing the similarity between particulate organic carbon (POC) and chlorophyll.

Source: From NASA (2009a).

FIGURE 4.29 Euphotic zone depth (m) from the Aqua MODIS satellite, 9 km resolution, monthly composite for

September 2007. (Black over oceans is cloud cover that could not be removed in the monthly composite.) See Figure S4.3

from the online supplementary material for the related map of photosynthetically available radiation (PAR). Source: From

NASA (2009b).

(a)

40˚

20˚

0˚

20˚

40˚

60˚

-100

-50

50

100

150

-100 -150

50

0

150

100

0

60˚

180˚

80˚N

-50

120˚W 60˚W 0˚ 60˚E 120˚E 180˚

80˚N

-50

-50

0

-50

0

50 50

150

50

0

100 150

0

-100

-150

50

-50

0

-100

-50

150

150 200

100

100

100 100

0

50

-50

-100

-50

60˚

60˚

0.4 Sv

Northern

40˚

40˚

20˚

–1.0 Sv

0˚ Subtropics/

Tropics

20˚

0.6 Sv

Southern

80˚S

180˚ 120˚W 60˚W 0˚ 60˚E 120˚E

0.3 Sv –0.1 Sv

–0.2 Sv

Pacific

Atlantic

Indian

FIGURE 5.4 (a) Net evaporation and precipitation (E P) (cm/yr) based on climatological annual mean data (1979e2005)

from the National Center for Environmental Prediction. Net precipitation is negative (blue), net evaporation is positive (red).

Overlain: freshwater transport divergences (Sverdrups or 1 10 9 kg/sec) based on ocean velocity and salinity observations.

Source: After Talley (2008).

180˚

80˚S

FIGURE 5.9 Outgoing Longwave Radiation (OLR) for Sept. 15eDec. 13, 2010. Source: From NOAA ESRL (2010).

(a)

180 90 W 0 90 E 180

180 90 W 0 90 E 180

60 N

Short

Long

100

100

<

wave

-50

30 N

-50

> wave

200

<

0

> > >

200

>

200

30 S

200

>

-50

100

-50

<

60 S

30 S -100 -150 -100

0 -15

(c)

(d)

60 N

Latent -15 Sensible

30 N -100

0

60 S

(b)

60 N

-50

> 30 N

0

30 S

60 S

60 N

-15 30 N

0

30 S

60 S

180 90 W 0 90 E 180

180 90 W 0 90 E 180

–200 –150 –100 –50 0 50 100 150 200

Mean heat fluxes (W/m 2 ) (SOC)

FIGURE 5.11 Annual average heat fluxes (W/m 2 ). (a) Shortwave heat flux Q s . (b) Longwave (back radiation) heat flux

Q b . (c) Evaporative (latent) heat flux Q e . (d) Sensible heat flux Q h . Positive (yellows and reds): heat gain by the sea. Negative

(blues): heat loss by the sea. Contour intervals are 50 W/m 2 in (a) and (c), 25 W/m 2 in (b), and 15 W/m 2 in (d). Data are from

the National Oceanography Centre, Southampton (NOCS) climatology (Grist and Josey, 2003).

180

120 W

60 W

0 60 E

80 N 80 N

-100

60 N 60 N

-50

0 0

-0.1

-50 0.6

40 N -150 0.8

40 N

-0.2

-100

0.8

1.2

20 N 0 20 N

0.7 0

1.2

0 100 50

50

0

50

0.6

0

0

20 S 0

-0.4

20 S

0

0.3

-1.3

-50 -50

40 S 0.1

0.5

-50 0

40 S

0

0

0

0

60 S 60 S

120 E

80 S 80 S

180 120 W 60 W 0 60 E 120 E 180

180

–200 –150 –100 –50 0 50 100 150 200

Annual mean net heat flux (W/m 2 ) (NOCS, 2003)

FIGURE 5.12 Annual average net heat flux (W/m 2 ). Positive: heat gain by the sea. Negative: heat loss by the sea. Data

are from the NOCS climatology (Grist and Josey, 2003). Superimposed numbers and arrows are the meridional heat

transports (PW) calculated from ocean velocities and temperatures, based on Bryden and Imawaki (2001) and Talley (2003).

Positive transports are northward. The online supplement to Chapter 5 (Figure S5.8) includes another version of the annual

mean heat flux, from Large and Yeager (2009).

(a)

0˚

60˚

120˚

180˚

240˚

300˚

0˚

40˚

60˚

Annual mean

0.1 N/m

60˚

40˚

20˚

20˚

0˚

0˚

–20˚ –20˚

–40˚ –40˚

–60˚ –60˚

(b)

40˚

60˚

0˚

60˚

February

120˚

180˚

240˚

300˚

0.1 N/m

0˚

60˚

40˚

20˚

20˚

0˚

0˚

–20˚ –20˚

–40˚ –40˚

–60˚ –60˚

(c)

40˚

60˚

0˚ 60˚ 120˚ 180˚ 240˚ 300˚ 0˚

August

0.1 N/m

60˚

40˚

20˚

20˚

0˚

0˚

–20˚ –20˚

–40˚ –40˚

–60˚ –60˚

0˚ 60˚ 120˚ 180˚ 240˚ 300˚ 0˚

–0.20 –0.16 –0.12 –0.08 –0.04 0 0.04 0.08 0.12 0.16 0.20

Zonal

wind

stress

FIGURE 5.16 Mean wind stress (arrows) and zonal wind stress (color shading) (N/m 2 ): (a) annual mean, (b) February,

and (c) August, from the NCEP reanalysis 1968e1996 (Kalnay et al., 1996).

FIGURE 5.16 (d) Mean wind stress curl based on 25 km resolution QuikSCAT satellite winds (1999e2003). Downward

Ekman pumping (Chapter 7) is negative (blues) in the Northern Hemisphere and positive (reds) in the Southern Hemisphere.

Source: From Chelton et al. (2004).

FIGURE 6.4 Different types of surfaces for mapping. The Mediterranean Water salinity maximum illustrated using:

(a) a standard depth surface (1200 m); (b) an isopycnal surface (potential density s 1 ¼ 32.2 kg/m 3 relative to 1000 dbar,

s q ~ 26.62 kg/m 3 relative to 0 dbar, and neutral density ~ 26.76 kg/m 3 ); (c) at the salinity maximum of the Mediterranean

Water and North Atlantic Deep Water (white areas are where there is no deep salinity maximum); and (d) data locations

used to construct these maps.

FIGURE 6.5 Objective mapping of velocity data, combining density and ADCP velocity measurements. California

Current: absolute surface streamfunction and velocity vectors in April, 1999, using the method from Chereskin and Trunnell

(1996). Source: From Calcofi ADCP (2008).

FIGURE 6.10 Examples of frequency-wavenumber spectra. (a) Equatorial waves (Kelvin and Rossby) from SSH

anomalies, compared with theoretical dispersion relations (curves). Source: From Shinoda et al. (2009)

Stn 308 307 306 305

0

Stn 308 307 306 305

0

2500

2500

5000

AAIW

Stn 308 307 306 305

0

2500

5000

WSDW

ABOVE 0.90

0.75 - 0.90

0.50 - 0.75

0.25 - 0.50

0.10 - 0.25

BELOW 0.10

5000

UCDW

FIGURE 6.17 Example of optimum multiparameter (OMP) water mass analysis. Southwestern Atlantic about 36 S,

showing the fraction of three different water masses. Antarctic Intermediate Water, AAIW; Upper Circumpolar Deep Water,

UCDW; and Weddell Sea Deep Water, WSDW. Source: From Maamaatuaiahutapu et al. (1992).

-3 log(K) m 2 /s 2

1000

2000

-4

z (m)

4000

5000

Mozambique Plateau

Madagascar Plateau

SW Indian Ridge

6000

0 2000 4000 6000 8000

r (km)

SE Indian Ridge

Ninety-East Ridge

Diamantina FZ

Perth

-5

-6

GM IW

FIGURE 7.2 Observed diapycnal diffusivity (m 2 /s 2 ) along 32 S in the Indian Ocean, which is representative of other

ocean transects of diffusivity. See Figure S7.4 for diffusivity profiles. Ó American Meteorological Society. Reprinted with

permission. Source: From Kunze et al. (2006).

FIGURE 7.8 Ekman response. Average wind vectors (blue) and average ageostrophic current at 15 m depth (red). The

current is calculated from 7 years of surface drifters drogued at 15 m, with the geostrophic current based on average density

data from Levitus et al. (1994a) removed. (No arrows were plotted within 5 degrees of the equator because the Coriolis force

is small there.) Ó American Meteorological Society. Reprinted with permission. Source: From Ralph and Niiler (1999).

(c)

FIGURE 8.2 (c) Directional wave spectrum (spectral density) from the NE Pacific (station 46006, 40 53’ N 137 27’ W, May

16, 2009. In (c), wave periods are from about 25 sec at the center of the ring to 4 sec at the outer ring. Blue is low energy,

purple is high. Direction of the waves is the same as direction relative to the center of the circle. Gray arrow in center

indicates wind direction. “Hs” indicates significant wave height. Source: From NOAA Wavewatch III (2009).

FIGURE 8.3 (a) Significant wave height (m) and (b) peak wave period (s) and direction (vectors) for one day (May 16,

2009). Source: From NOAA Wavewatch III (2009).

FIGURE 8.7 Sumatra Tsunami (December 26, 2004). (c) Global reach: simulated maximum sea-surface height and arrival

time (hours after earthquake) of wave front. Source: From Titov et al. (2005).

(b)

0

100

0.06

0.05

Depth (m)

200

300

400

0.04

0.03

0.02

(u¢) 2 +(v¢) 2 (m 2 s −2 )

500

0.01

600

−150 −100 −50 0 50 100 150

Distance (km)

FIGURE 8.11 (b) Velocity variance (variability) observed along a section crossing the Hawaiian Ridge, which is located

just below the bottom of the figure at 0 km; the black rays are the (group velocity) paths expected for an internal wave with

frequency equal to the M 2 tide; distance (m) is from the center of the ridge. Source: From Cole, Rudnick, Hodges, & Martin

(2009).

0

FIGURE 8.11 (c) Breaking

internal solitary wave, over

the continental shelf off

Oregon. The image shows

acoustic backscatter: reds

indicate more scatter and are

related to higher turbulence

levels. Ó American Meteorological

Society. Reprinted

with permission. Source:

From Moum et al. (2003).

(c)

FIGURE 8.15 Tidal effects on Georges Bank. (c) Chlorophyll a concentration (mg/m 3 ) on October 8, 1997, from the

SeaWiFS satellite. Source: From Sosik (2003).

FIGURE 9.4 Sea surface temperature from the GOES satellite. (a) Gulf of Mexico showing the Loop Current beginning to

form an eddy. (b) Gulf Stream, showing meander at the Charleston Bump and downstream shingling. Black contours are

isobaths (100, 500, 700, 1000 m). Source: From Legeckis, Brown and Chang (2002).

Depth [m]

0

500

1000

1500

2000

2500

3000

3500

4

2.

16

15

6 7 10

5

3.8 4

3.6

3.4

3

2.4

2.2

20

18

34.76

34.96

5

34.92

36.9

36.5 36.6

35.08 35.1

35.06

35.04

35.02

35

34.98

34.94

120 140

160

190

200

210

260

220

200

210

160

200

255

260

255

260

265

0

500

1000

0

1500

270

2000

2500

3000

65

4

14 67890123

3500

4000

4500

5000

5500

6000

6500

South America

8

3.

Puerto Rico

1.8

.8 3 3.2 3.4

1.6

2

North America

South America

34.93 34.91

Puerto Rico

34.84

34.9

34.86

34.88

1.4

Pot.

6000

Temp.

Salinity

Oxygen

6500

15°N 20°N 25°N 30°N 35°N 40°N 15°N 20°N 25°N 30°N 35°N 40°N 15°N 20°N 25°N 30°N 35°N 40°N

FIGURE 9.7 Subtropical North Atlantic at 66 W in August 1997. (a) Potential temperature ( C), (b) salinity, and (c) oxygen (mmol/kg). (World Ocean

Circulation Experiment section A22.)

North America

South America

220

Puerto Rico

230 250

235

240 245

265

255

260

260

North America

4000

4500

5000

5500

(a)

FIGURE 9.8 Canary Current System. (a) SST (satellite AVHRR image) on August 27, 1998. Source: From Pelegrí et al. (2005).

FIGURE 9.9 North Atlantic Current and Labrador Current at the Grand Banks. (a) SST (AVHRR) on October 12, 2008,

showing cold Labrador Current moving southward along the edge of the Grand Banks. Source: From Johns Hopkins APL

Ocean Remote Sensing (1996).

(a)

0

–100

0

–10

–25

–50

–75

–100

–150

depth in meters

–200

–300

–200

–250

–300

–400

–400

–500

–60

–40

–20

longitude

Min = –0.83 m/s Max = 0.97

0

–500

Contour 0.1 m/s

–1 0 1

FIGURE 9.11 Tropical current structures. (a) Eastward velocity along the equator, from a data assimilation. Source: From

Bourlès et al. (2008).

(c)

66°W 60°W

54°W

48°W 42°W

32°S

36°S

40°S

44°S

48°S

Jul 14 1994

5 10 15 20

FIGURE 9.12 (c) Infrared satellite image of the Brazil-Malvinas confluence. Black lines are current vectors at moorings, at

approximately 200 m depth. Light curve is the 1000 m isobath. Source: From Vivier and Provost (1999).

(c)

20°S

Poleward Flow

30°S

40°S

Ben g u el a

Cu

r r en t

Sout h At lant ic

Current

Ag u

W a l v

C ap e

lh as Ext ension

i d g e

i s R

B asi n

SOUTHERN

AFRICA

Cape

Town

A g u l h a s Cu

Agulhas Return Current

r r e n t

So u t h

A t l a n t i c

Cu r r e n t

10°W

0°

10°E 20°E 30°E 40°E

FIGURE 9.13 Benguela Current and Agulhas retroflection. (a, b) AVHRR SST monthly composite for July (winter) and

December (summer) 2005. Source: From UCT Oceanography Department (2009). (c) Schematic of Agulhas retroflection and

eddies, with flow directions in the intermediate water layer. Gray-shaded rings are the Agulhas anticyclones. Dashed rings

are cyclones that are generated in the Agulhas. Source: From Richardson (2007).

(a)

FIGURE 9.15 Schematics of deep circulation. (a) NSOW (blue), LSW (white dashed), and upper ocean (red, orange, and

yellow) in the northern North Atlantic. Source: From Schott and Brandt (2007).

FIGURE 9.15 (b) Deep circulation pathways emphasizing DWBCs (solid) and their recirculations (dashed). Red: NSOW.

Brown: NADW. Blue: AABW. (M.S. McCartney, personal communication, 2009.)

(a)

0

1000

2000

North Atlantic 24°N


27.3

Antarctic Intermediate Water and Mediterranean Water

27.74

Labrador Sea Water

36.96

3000


4000

5000

45.91

AABW

6000

0 1000 2000 3000 4000 5000 6000

–80 –75 –70 –65 –60 –55 –50 –45 –40 –35 –30 –25 –20 –15

(c)

0

1000

32.0 34.0 34.5 34.7 34.8 34.9 35.0 35.5 36.0 36.5 37.0

South Atlantic 32°S


26.2

Lower thermocline

26.9


27.4

2000

North Atlantic Deep Water/Circumpolar Deep Water

3000

4000

45.86 45.86

45.88

AABW

5000

(c) South Atlantic 32°S

6000

0 1000 2000 3000 4000 5000 6000

–50 –45 –40 –35 –30 –25 –20 –15 –10 –5 0 5 10 15

FIGURE 9.16 Salinity (color and white contours) and isopycnals (black contours) at (a) 24 N in 1981 and (c) 32˚S in 1959/

1972. After Talley (2008), based on Reid (1994) velocities.

FIGURE 9.18 Potential temperature ( C) versus salinity for (a) full water column, and (b) water colder than 10 C.

(c) Potential temperature versus oxygen for full water column. (d) Station location map. Colors indicate latitude range.

Contours are potential density referenced to 0 dbar. Data are from the World Ocean Circulation Experiment (1988e1997).

(a)

0

500

1000

3

1500

2000

2500

1.8

3000 2

3500

0 500 km

(c)

0

500

1000

1500

2000

2500

3000

2.2

2

3.6

2.4

2.2

2

300

290

1.6

295

295

3.2

3

2.9

2.8

4

3.8

2.8

2.6

290

3.4

300

295

295

285

290

295

9

8

9

8

9

8

295

29

3500

0 500 km

3

4

3.2

2.9

2.8

58°W 55° 51°W

2

4

5

5

2

2.2

2.4

00

295

3.2

0 500 1000 1500 km

35°W 30° 25° 20° 15° 10°W

285

2.8

2

1.6 1.8

280

3

2.8

2.6

295

285

4

290

3.8

3.6

275

3

2.

270

240

260

280

5

7

3.4

3

6

260

9

8

230

265

8

(b)

230 1000

240

245

255

1500

250

265

5

275

2000

270

24 23

22 21 20 19

18

17 16 14 12 10

8

275

2

265 2500 24

260

0

500

1000

1500

2000

2500

Labrador Sea

3500

0 500 km

0

500

3000

3

3

Labrador

34.9

3000

34.82

34.84

34.84

27.9

27.92

Oxygen σ

285

255

3500

θ

0 500 1000 1500 km 0 500 km 0 500 1000 1500 km

34.86

34.88

34.9

34.84

34.84

34.91

58°W 55° 51°W

8

Greenland

34.88

34.9

34.88

34.88

Irminger Sea

34.9

34.86

34.9

34.95

34.91

Reykjanes Ridge

35.2

35.1

34.92

34.92

34.93

34.94

34.96

34.9

34.91

34.92

96

34.94

34.95

34.9134.95

Potential temperature Salinity

(d)

27.9

27.

27.74

27.76

27.78

27.5

27.6

27.7

27.72

27.8

27.82

27.84

27.88

0 500 1000 1500 km

27.86

35

34.9

Iceland

Basin

35.3

Rockall Plateau

35.4

35.2

Rockall Trough

35°W 30° 25° 20° 15° 10°W

27.6

27.8

27.84 27.82

7.9

27.88

27.76

27.8

27.5

27.82

27.84

27.9

27.7

27.72

27.74

27.9

27.4

27.3

27.88

0

0

500

1000

1500

2000

2500

3000

3500

500

1000

1500

2000

2500

3000

3500

58°W 55° 51°W

35°W 30° 25° 20° 15° 10°W

58°W 55° 51°W

35°W 30° 25° 20° 15° 10°W

FIGURE 9.20 Subpolar North Atlantic at about 55 N from May to June, 1997. (a) Potential temperature ( C), (b) salinity,

(c) oxygen, and (d) potential density (s q ) in the Labrador Sea (left side) and from Greenland to Ireland (right side). (World

Ocean Circulation Experiment sections AR7W and A24)

FIGURE 9.21 Labrador Sea Water. (b) Chlorofluorocarbon-11 (pmol/kg) in the upper LSW layer, at s q ~ 27.71 kg/m 3 .

Source: From Schott et al. (2009) and from Kieke et al. (2006).

(a)

Depth [m]

80°W 75°W 70°W 65°W 60°W 55°W 50°W 45°W 40°W 35°W 30°W 25°W 20°W

0

25

20

20

15

15

500

10

10

1000

1500

2000

2500

3000

3500

Florida Strait

5

4.4

5

4.4

4

4

3.4

3.4

3

3

2.4 2.4

(b)

0

500

1000

1500

2000

2500

3000

3500

80°W 75°W 70°W 65°W 60°W 55°W 50°W 45°W 40°W 35°W 30°W 25°W 20°W

36.50 36.00

35.70 35.60

35.50

35.30

35.20

Florida Strait

35.20

35.08

35.04

35.00

34.98

34.96

34.95

34.94

34.92

34.92

34.91

4000

4500

5000

5500

(c)

Depth [m]

2

1.6

Distance [km]

2

6000

6000

0 500 1000 1500 2000 2500 3000 3500 4000 4500 5000 5500 6000 6500 0 500 1000 1500 2000 2500 3000 3500 4000 4500 5000 5500 6000 6500

80°W 75°W 70°W 65°W 60°W 55°W 50°W 45°W 40°W 35°W 30°W 25°W 20°W

0

500

1000

1500

2000

2500

3000

3500

4000

4500

5000

5500

6000

Bahamas

Florida Strait

Bahamas

265

240

260

265

250

250

190

180

Mid-Atlantic

Ridge

160

0 500 1000 1500 2000 2500 3000 3500 4000 4500 5000 5500 6000 6500

Distance [km]

160

Mid-Atlantic

Ridge

200

240

220

230

235

140

>

245

245 245

>

Africa

Pot.

temp.

Africa

Oxygen

4000

4500

5000

5500

(d)

0

500

1000

1500

2000

2500

3000

3500

4000

4500

5000

5500

6000

Bahamas

34.90

34.88

34.86

Distance [km]

80°W 75°W 70°W 65°W 60°W 55°W 50°W 45°W 40°W 35°W 30°W 25°W 20°W

Florida Strait

Bahamas

>

>

0.5

0.2

0.6

0.4

2.0

0.5

0.05

<

0.2

0.005

0.02

0.01

2.0

1.0

0.1

Mid-Atlantic

Ridge

Mid-Atlantic

Ridge

0 500 1000 1500 2000 2500 3000 3500 4000 4500 5000 5500 6000 6500

Distance [km]

0.05

0.01

0.005

>

>

0.005

>

0.005

>

0.01

34.90

0.01

0.005

Africa

Salinity

Africa

CFC-11

FIGURE 9.22 Subtropical North Atlantic at 24 N from July to August 1992. (a) Potential temperature ( C), (b) salinity,

(c) oxygen (mmol/kg), and (d) CFC-11 (pmol/kg) at 24 N. (World Ocean Circulation Experiment section A05). Adapted: From

WOCE Atlantic Ocean Atlas, Jancke, Gouretski, and Koltermann (2011).

FIGURE 10.4 Kuroshio velocity structure. (d) Eastward velocity of the Kuroshio Extension at 152 30’E [red (blue)

indicates eastward (westward) flow]. Source: From Yoshikawa et al. (2004).

50°

130° 120°W

CC

SUMMER

40°

SCE

30°

N

50°

CC

DC

WINTER

40°

30°

N

SCC

50°

40°

DC

EARLY

SPRING

CC

30°

N

CC = California Current

DC = Davidson Current

SCC = So. California

Countercurrent

SCE = So. California Eddy

130° 120°W

FIGURE 10.5 (a) Schematic of the surface currents in the CCS in different seasons. Source: From Hickey (1998). (b) Mean

seasonal cycle of satellite-derived surface temperature (color) and altimetric height, showing the geostrophic surface

circulation. Source: From Strub and James (2000, 2009).

FIGURE 10.24 Tropical instability waves. SST from the Tropical Rainfall Mapping Mission (TRMM) Microwave Imager

(TMI) for two successive 10-day periods in August 1998, after establishment of the cold tongue during a La Niña. A more

complete time series (June 1eAugust 30, 1998) is reproduced in Figure S10.20 on the textbook Web site. Source: From Remote

Sensing Systems (2004).

FIGURE 10.25 Zonal wind speed and SST in the equatorial Pacific to illustrate the annual cycle. Positive wind speed is

toward the east. Climatological means in February and August and an expanded time series for 2000e2007 are shown in

Figure S10.21 on the textbook Web site, to emphasize the seasonal cycle. Source: From TAO Project Office (2009a).

(a)

La Niña Conditions

(b)

Normal Conditions

(c)

El Niño Conditions

Convective

Circulation

Equator

Equator

Equator

Thermocline

Thermocline

Thermocline

120°E 80°W

120°E 80°W

120°E 80°W

FIGURE 10.27 (a) La Niña, (b) normal, and (c) El Niño conditions. Source: From NOAA PMEL (2009b).

FIGURE 10.28 (a) Correlation of monthly SST anomalies with the ENSO Nino3.4 index, averaged from 1948 to 2007. The

index is positive during the El Niño phase, so the signs shown are representative of this phase. (Data and graphical interface

from NOAA ESRL, 2009b.)

FIGURE 10.29 Potential T-S curves for selected stations (inset map). Acronyms: NPCW, North Pacific Central Water;

SPCW, South Pacific Central Water; NPSTUW, North Pacific Subtropical Underwater; SPSTUW, South Pacific Subtropical

Underwater; NPSTMW, North Pacific Subtropical Mode Water; SPSTMW, South Pacific Subtropical Mode Water; NPIW,

North Pacific Intermediate Water; AAIW, Antarctic Intermediate Water; DtW, Dichothermal Water; MtW, Mesothermal

Water; CCS, California Current System waters; and PCCS, Peru-Chile Current System Waters. Mean T-S curves are shown

for every 10 degrees square in Figure S10.45 on the textbook Web site.

FIGURE 10.33 (a, c) Salinity and (b, d) oxygen (mmol/kg) at neutral densities 26.75 kg/m 3 and 27.3 kg/m 3 , characteristic

of NPIW and AAIW, respectively. In the Southern Ocean, white at 26.75 kg/m 3 shows the isopycnal outcrops; the gray curve

in (c) and (d) is the winter outcrop. Depth of the surfaces is shown in the WOCE Pacific Ocean Atlas. Source: From WOCE

Pacific Ocean Atlas, Talley (2007).

FIGURE 10.34 Dense water formation in the Okhotsk Sea. (a) Bottom potential temperature in September, 1999, and

mean velocity vectors at the two moorings. Source: From Shcherbina, Talley, and Rudnick (2003, 2004).

FIGURE 11.3 Somali Current regime during the Southwest Monsoon (August/September, 1995). Source: From Schott and

McCreary (2001).

FIGURE 11.5

(2007a).

SST in July 2003 (Southwest Monsoon), from the MODIS satellite. Source: From NASA Goddard Earth Sciences

FIGURE 11.11 Indonesian Archipelago and Throughflow with transports (Sv). Lower panel summarizes transport

above and below 680 m (Makassar Strait sill depth). Source: From Gordon (2005).

FIGURE 11.12 (a, b) Red Sea Overflow Water: salinity with potential density contours overlaid on sections in the Gulf of

Aden in FebruaryeMarch, 2001. North is on the left. Source: From Bower et al. (2005). Ó American Meteorological Society.

Reprinted with permission. (c) Red Sea outflow in the Gulf of Aden: climatological salinity on the isopycnal s q ¼ 27.20 kg/m 3 .

Source: From Bower, Hunt, and Price (2000).

(a) (b) (c)

N

N

N

N

S

S

S

S

S

S

S

669

714

69 9

68 4

75 9

744

729

8 04

774

7 89

E E E E

Latitude

20°N

9°N

2°S

1 3°S

2 4°S

3 5°S


25

20

15

10

5

0

22

23

24

25

26

27

AAIW

ITF

Central Water

LCDW

Equatorial Water

28

SAMW

STUW

RSOW

34. 5 35. 0 35. 5 36. 0 36. 5

Salinity

Arabian Sea Surface

PGW


25

20

15

10

5

0

0 40 80 120 160 200 240

Oxygen ( mol/kg)

FIGURE 11.18 (a) Station locations, (b) potential temperature ( C) d salinity and (c) potential temperature ( C) d

oxygen (mmol/kg) for the Indian Ocean along 60 E. After the WOCE Indian Ocean Atlas, Talley (2011).

(a)

X

X

X

X

X

FIGURE 12.10 Circulation schematics. (a) Subsurface Atlantic and intermediate layers of the Arctic Ocean and the

Nordic Seas. Convection sites in the Greenland and Iceland Seas, and in the Irminger and Labrador Seas are also shown

(light blue), as is a collection point for brine-rejected waters from the Barents Sea. Source: From Rudels et al. (2010).

FIGURE 12.13 (a) Schematic circulation of summer Bering Strait Water (blue) and Alaskan Coastal Water (red) during

the positive phase of the Arctic Oscillation (Chapter S15 on the textbook Web site). (b) Temperature ( C) of the shallow

temperature maximum layer, which lies between 50 and 100 m depth, in the Canadian Basin. Source: From Steele et al.

(2004).

(a)

210˚

180˚

150˚

(b)

0

500

MaB NP

CaB

AmB

NaB GrS IcS

WSC

270˚

240˚

NP

CaB

MaB

AmB

120˚

90˚

Pressure

1000

1500

2000

2500

2000

2500

3000

WSC

IcS

AmB

GrS

NP

CaB

MaB

NaB

NaB

3000

3500

300˚

330˚

GrS

IcS

NAC

0˚

WSC

30˚

60˚

3500

4000

4500

4000

4500

34.89 34.90 34.91 34.92 34.93 34.94 34.95 34.96

30 31 32 33 34 35

Salinity

Pressure

(c)

0

500

1000

1500

2000

2500

3000

3500

4000

4500

GrS

CaB

MaB

NP

GrS

NaB

AmB

2000

2500

3000

3500

4000

NaB

AmB

WSC

GrS

WSC

IcS

AmB

NaB

IcS

CaB

MaB

NP

4500

–1.2 –1.1 –1.0 –0.9 –0.8 –0.7 –0.6 –0.5


–2 –1 0 1 2 3 4 5 6 7


(d)


2.5

2.0

1.5

1.0

0.5

0.0

–0.5

–0.5

–0.6

–0.7

–0.8

–0.9

–1.0

–1.1

46.38

IcS

WSC NaB

46.4

–1.0

IcS

CaB GrS

–1.5

MaB NP NaB

AmB

–2.0

30 31 32 33 34 35

Salinity

25

GrS

CaB

26

AmB

46.44

46.48

NP

MaB

–1.2

34.89 34.90 34.91 34.92 34.93 34.94 34.95 34.96

27

IcS WSC

FIGURE 12.17 (a) Station map (1994 and 2001), (b) salinity, (c) potential temperature ( C), and (d) potential temperaturesalinity.

Acronyms: CaB, Canada Basin; MaB, Makarov Basin; NP, North Pole; AmB, Amundsen Basin; NaB, Nansen Basin;

WSC, West Spitsbergen Current; GrS, Greenland Sea; IcS, Iceland Sea; and NAC, Norwegian Atlantic Current. Expanded from

Timmermans and Garrett (2006).

28

FIGURE 12.21 Arctic ice ages: (a) 2004 and (b) cross-section of ice age classes (right) as a function of time (Hovmöller

diagram), extending along the transect across the Arctic from the Canadian Archipelago to the Kara Sea shown in (a). Source:

Extended from Fowler et al. (2004).

FIGURE 13.3 Properties at 50 m depth. (a) Potential temperature ( C), (b) salinity. Source: From WOCE Southern Ocean

Atlas, Orsi and Whitworth (2005).

54°S

25 cm/s

120

56°S

100

58°S

60°S

80

60

40

streamfunction, cm

62°S

20

64°S

0

66°W

68°W

56°W

58°W

60°W

62°W

64°W

FIGURE 13.9 Mean currents in the Drake Passage, averaged over 30e300 m depth, from 128 ADCP crossings over 5

years. Strong currents from north to south are the Subantarctic Front (56 S), the Polar Front (59 S), and the Southern ACC

Front (62 S). After Lenn, Chereskin, and Sprintall (2008).

25

20

Atlantic Ocean 20° to 25°W

Blue: south of 51°S

Purple: 51°S to 32°S

Red: 32°S to 1°N

Orange: 1°N to 63°N

23

23.5

24

24.5


15

10

25

25.5

26

60 N

27

28

5

26.5

30 N

28.5

29

0

30 S

0

60 S

90 W 60 W 30 W 0 30 E

FIGURE 13.14

33 34 35 36 37

Salinity

Potential temperature-salinity diagram in the Weddell Sea and Atlantic Ocean.

(a)

(b)

(c)

(d)

FIGURE 13.15 Properties along a Lower Circumpolar Deep Water isopycnal (neutral density 28.05 kg m 3 ), corresponding

roughly to the salinity maximum core. (a) Potential temperature ( C), (b) salinity, (c) depth (m), (d) oxygen (mmol/kg).

Source: From WOCE Southern Ocean Atlas, Orsi and Whitworth (2005).

(a)

(b)

(c)

(d)

FIGURE 13.16 Properties on an Antarctic Bottom Water isopycnal (neutral density 28.27 kg m 3 ). (a) Potential

temperature and (b) salinity. Bottom properties (depths greater than 3500 m): (c) potential temperature and (d) salinity.

Source: From WOCE Southern Ocean Atlas, Orsi and Whitworth (2005).

FIGURE 13.20

(2008).

Antarctic latent heat polynyas: sea ice production, averaged over 1992e2001. Source: From Tamura et al.

Gulf

Stream

System

Equator

Labrador

Current

North Atlantic

Current

NECC

NEUC

North Brazil

Current System

Brazil

Current

System

East Greenland

Current

NEC

Norwegian

Atlantic Current

SEC

Canary

Current

System

Subtropical

Gyre

Atlantic Equatorial

Current System

Subtropical

Gyre

Benguela

Current

System

40

N

Somali

Current

System

Indonesian

Throughflow

Subtropical

Gyre

SEC

Kuroshio

System

40S

East Kamchatka

Current

Leeuwin

Current

Oyashio

East Australian

Current System

Bering

Strait

Subpolar Gyre

Subtropical Gyre

NEC

Subtropical Gyre

Beaufort

Gyre

Alaska

Gyre

Pacific Equatorial/Tropical Current System

North Pacific

Current

California

Current

System

SEC

NEUC

NECC

Peru-Chile

Current

System

Equator

Malvinas

Current

Agulhas

Current

System

Weddell

Sea

Gyre

Antarctic

Circumpolar

Current System

Subtropical Gyres

Equatorial and Tropical Circulations

Intergyre and/or Interbasin Exchanges

Polar & Subpolar Current Systems

Ross Sea

Gyre

FIGURE 14.1

Surface circulation schematic. Modified from Schmitz (1996b).

FIGURE 14.2 (a) Surface dynamic topography (dyn cm), with 10 cm contour intervals, and (b) surface velocity

streamlines, including both geostrophic and Ekman components; color is the mean speed in cm/sec. Source: From Maximenko

et al. (2009).

FIGURE 14.6 Net transports (Sv) in isopycnal layers across closed hydrographic sections (1 Sv ¼ 1 10 6 m 3 /sec). (a)

Three calculations from different sources are superimposed, each using three isopycnal layers (see header). Circles between

sections indicate upwelling (arrow head) and downwelling (arrow tail) into and out of the layer defined by the circle color.

Source: From Maltrud and McClean (2005), combining results from their POP model run, Ganachaud and Wunsch (2000), and

Schmitz (1995).

FIGURE 14.7 Modeled upwelling across the isopycnal 27.625 kg/m 3 , which represents upwelling from the NADW layer.

Source: From Kuhlbrodt et al. (2007); adapted from Döös and Coward (1997).

FIGURE 14.11

Global overturning circulation schematics. (a) The NADW and AABW global cells and the NPIW cell.

(c)

Southern Ocean



SAMW, AAIW

Low, mid-latitude upper ocean waters

LCDW

UCDW

Pacific-Indian

upwelling &

diffusion

PDW/IDW

Antarctica

AABW

formation

(brine

rejection)

NADW

PDW/IDW

formation

(diffusion)

NADW

formation

(convection)

AABW

FIGURE 14.11 (b) Overturn from a Southern Ocean perspective. Source: After Gordon (1991), Schmitz (1996b), and Lumpkin

and Speer (2007). (c) Two-dimensional schematic of the interconnected NADW, IDW, PDW, and AABW cells. The schematics

do not accurately depict locations of sinking or the broad geographic scale of upwelling. Colors: surface water (purple),

intermediate and Southern Ocean mode water (red), PDW/IDW/UCDW (orange), NADW (green), AABW (blue). See

Figure S14.1 on the textbook Web site for a complete set of diagrams. Source: From Talley (2011).

40˚

60˚

80˚N

Labrador Sea

Water 27.8 σ θ

60˚W 0˚ 60˚E 120˚E 180˚ 120˚W

Mediterranean Water

28.0 σ θ

North Pacific

Intermediate Water

27.0 σ θ

20˚

0˚

Red Sea

Water

27.7 σ θ

20˚

40˚

60˚


Water 27.1 σ θ

80˚S

FIGURE 14.13 Low- and high-salinity intermediate waters. AAIW (dark green), NPIW (light green), LSW (dark blue),

MW (orange in Atlantic), RSW (orange in Indian). Light blue in Pacific: overlap of AAIW and NPIW. Light blue in Indian:

overlap of AAIW and RSW. Cross-hatching: mixing sites that are particularly significant for the water mass. Red dots

indicate the primary formation site of each water mass; fainter dots mark the straits connecting the Mediterranean and Red

Seas to the open ocean. The approximate potential density of formation is listed. Source: After Talley (2008).

FIGURE 14.16 Eddy kinetic energy (cm 2 s 2 ) from surface drifters. Source: From NOAA AOML PHOD (2009).

A complementary figure based on satellite altimetry (from Ducet, Le Traon, & Reverdin, 2000) is reproduced in Figure S14.6c on

the textbook Web site.

FIGURE 14.17 (a) Horizontal eddy diffusivity (m 2 /sec) at the sea surface (color) with mean velocity vectors, based on

surface drifter observations. Source: From Zhurbas and Oh (2004). (b) Eddy diffusivity ellipses at 900 m based on subsurface

float velocities. Colors indicate different scales (see figure headers). Source: From Davis (2005). The Atlantic surface map and

Indian 900 m map from the same sources are reproduced in Chapter S14 (Figures S14.7 and S14.8) on the textbook Web site.

FIGURE 14.18 Surface-height anomalies at 24 degrees latitude in each ocean, from a satellite altimeter. Source: From Fu

and Chelton (2001).

FIGURE 14.21 Tracks of coherent cyclonic and anticyclonic eddies with lifetimes of more than 4 weeks, based on altimetric

SSH, color coded by a “nonlinearity parameter,” which is the ratio of velocity within the eddy compared with the

eddy propagation speed. White areas indicate no eddies or trajectories within 10 degrees latitude of the equator. Source: From

Chelton et al. (2007).

FIGURE 14.22 Near-inertial motion. (a) Average inertial current speeds (cm/sec), based on surface drifters. Source: From

Chaigneau et al. (2008). (b) Rotary power spectra in 2.5 degree latitude bins in the Pacific Ocean. The solid curve is the inertial

frequency at each latitude; the dashed curve is twice the inertial frequency. Negative frequencies rotate counterclockwise

and positive frequencies rotate clockwise. Source: From Elipot and Lumpkin (2008).

Descriptive Physical Oceanography (6º Edi.) - L. D. Talley - G. L. Pickard - W. Emery - J. H. Swift

Create successful ePaper yourself

Delete template?

Save as template?