01.08.2013 Views

Self - Induced Ekman Pumping Over Oceanic Mesoscale Eddies

Self - Induced Ekman Pumping Over Oceanic Mesoscale Eddies

Self - Induced Ekman Pumping Over Oceanic Mesoscale Eddies

SHOW MORE
SHOW LESS

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

<strong>Self</strong>-<strong>Induced</strong> <strong>Ekman</strong> <strong>Pumping</strong> over<strong>Oceanic</strong> <strong>Mesoscale</strong> <strong>Eddies</strong><br />

Peter Gaube* and Dudley Chelton<br />

Oregon State University<br />

• Two mechanisms for eddy self-induced <strong>Ekman</strong> pumping:<br />

1. From air-sea interaction as winds blow over eddy-induced perturbations of SST.<br />

- First shown for Gulf Stream rings by Park & Cornillon (2002) and Park et al. (2006).<br />

- The feedback effects of this SST-induced <strong>Ekman</strong> pumping on the ocean disrupt the<br />

coherent evolution of the mesoscale eddy field (Jin et al., 2009).<br />

2. From the influence of eddy surface currents on the relative wind, and therefore<br />

on the wind stress<br />

- The potential importance of the feedback effects of this surface current induced <strong>Ekman</strong><br />

pumping on the evolution of mesoscale eddies was first noted by Dewar & Flierl (1987).<br />

- First shown for Gulf Stream rings by Cornillon & Park (2001) and Park et al. (2006).<br />

• Both mechanisms tend to attenuate eddies.<br />

• The surface current effects generate upwelling within anticyclonies and downwelling<br />

within cyclones (Martin and Richards, 2001; McGillicuddy et al., 2007).<br />

* The results presented here are part of the PhD thesis by Peter Gaube


Procedure for Composite Averaging SST, Wind Speed<br />

and Wind Stress Curl in Eddy-Centric Coordinates:<br />

Synergy Between 4 Complimentary Satellite Datasets<br />

• Identify mesoscale eddies by altimetry from their SSH signatures.<br />

• Composite average the other satellite datasets in an “eddy-centric” translating<br />

reference frame with (∆x,∆y) coordinates relative to the eddy centroid normalized by<br />

the radius L s of maximum rotational speed at each location along its trajectory.<br />

- AMSR+AVHRR measurements of SST (Reynolds OI2 analyses).<br />

- QuikSCAT measurements of wind speed and wind stress.<br />

- SeaWiFS estimates of oceanic chlorophyll.<br />

• Because the dominant mechanism for eddy-induced SST variability is horizontal<br />

advection by the rotational velocity of the eddy, SST and wind speed must be<br />

composite averaged in a coordinate system that is rotated by an amount determined<br />

from the large-scale background SST gradient.


1. Eddy Influence on SST and Wind Speed


Global Composite Averages of SST in Eddy-Centric Coordinates<br />

Normalized normalized Distance distance from from Eddy eddy Centroid centroid<br />

mposites<br />

udes<br />

degrees C<br />

SST Eddy Composites<br />

SST Eddy Composites SST Eddy Composites<br />

Regions of Northward ∇T Midlatitudes<br />

Regions of Southward ∇T<br />

Clockwise Midlatitudes<br />

Rotating horizontally Clockwise normalized Rotating and rotatedMidlatitudes<br />

ed and rotatedCounterclockwise<br />

Rotating<br />

ard<br />

normalized distance from eddy<br />

horizontally normalized and rotated<br />

Northward<br />

normalized distance degrees Cfrom<br />

eddy centroid<br />

Southward<br />

Northward<br />

degrees C<br />

Counterclockwise Rotating<br />

Southward<br />

Normalized Distance from Eddy Centroid<br />

degrees C<br />

horizontally normalized and rotated<br />

Northward<br />

Southward<br />

degrees C<br />

Contour Interval<br />

is 0.05°C<br />

Southward


Global Composite Averages of Wind Speed in Eddy-Centric Coordinates<br />

Wind Speed Eddy Composites<br />

Wind Speed Eddy Composites Wind Speed Eddy Compos<br />

Regions of Northward ∇T Midlatitudes<br />

Regions of Southward ∇T<br />

Clockwise Midlatitudes<br />

Rotating horizontally Clockwise normalized Rotating and rotatedMidlatitudes<br />

Normalized normalized Distance distance from from Eddy eddy Centroid centroid<br />

y Composites<br />

udes<br />

m s -1<br />

ed and rotatedCounterclockwise<br />

Rotating<br />

ard<br />

normalized distance from eddy<br />

horizontally normalized and rotated<br />

Northward<br />

m s-1 normalized distance from eddy centroid<br />

Southward<br />

Northward<br />

m s -1<br />

Counterclockwise Rotating<br />

Southward<br />

Normalized Distance from Eddy Centroid<br />

m s -1<br />

horizontally normalized and rotated<br />

Northward<br />

Southward<br />

m s -1<br />

Contour Interval<br />

is 0.025 ms Southward<br />

-1


2. <strong>Ekman</strong> <strong>Pumping</strong> from Eddy-Related SST<br />

Influence on Wind Speed


The Influence <strong>Ekman</strong> <strong>Pumping</strong> of Wind from Direction Eddy Perturbations on SST-induced<br />

of SST<br />

for Winds from Various Directions <strong>Over</strong> a 1.0°C SST Anomaly at 30°S<br />

<strong>Ekman</strong> <strong>Pumping</strong><br />

cm day -1<br />

Contour Interval<br />

is 3 cm day -1<br />

1.0 deg<br />

C temp<br />

anomaly


The Influence <strong>Ekman</strong> <strong>Pumping</strong> of Wind from Direction Eddy Perturbations on SST-induced<br />

of SST<br />

for Winds from Various Directions <strong>Over</strong> a 0.5°C SST Anomaly at 30°S<br />

<strong>Ekman</strong> <strong>Pumping</strong><br />

cm day -1<br />

Contour Interval<br />

is 3 cm day -1<br />

0.5 deg<br />

C temp<br />

anomaly


3. <strong>Ekman</strong> <strong>Pumping</strong> from Eddy Surface Currents<br />

for an Idealized Gaussian Eddy


Eddy-<strong>Induced</strong> <strong>Ekman</strong> <strong>Pumping</strong><br />

<strong>Ekman</strong> <strong>Pumping</strong> from Eddy Surface Currents<br />

for an Idealized Anticyclonic Gaussian How it works: Eddy and Westerly Winds at 30°S<br />

• The surface circulation of an eddy can induce <strong>Ekman</strong> pumping in a uniform wind<br />

field. SSH Structure of<br />

Idealized Eddy Gaussian VorticityEddy<br />

<strong>Ekman</strong> <strong>Ekman</strong> <strong>Pumping</strong> <strong>Pumping</strong> Velocity<br />

l = −→ u a − −→ −→<br />

u rel u= o<br />

−→ u a − −→ u o<br />

−→<br />

τ = ρCD|urel| −→ u rel<br />

−→ u rel = −→ u a − −→ u o<br />

m s -1 per 100 km<br />

Contour Interval<br />

is 5 cm<br />

• 30°S<br />

• 20 cm amp.<br />

• 10 ms -1 wind<br />

• Max current<br />

40 cm s -1<br />

WE = 1<br />

ρf ∇×−→ τ<br />

cm day -1<br />

Contour Interval<br />

is 3 cm day -1


The Influence <strong>Ekman</strong> <strong>Pumping</strong> of Wind from Eddy Direction Surface on Currents Gaussian<br />

for an Idealized Anticyclonic Gaussian Eddy and Winds from Various Directions<br />

Eddy-induced at <strong>Ekman</strong> 30°S <strong>Pumping</strong><br />

cm day -1<br />

Contour Interval<br />

is 3 cm day -1


4. <strong>Ekman</strong> <strong>Pumping</strong> from the Combined Effects of SST<br />

Influence and Eddy Surface Currents<br />

for an Idealized Gaussian Eddy


<strong>Ekman</strong> The <strong>Pumping</strong> Influence from of Eddy Wind Surface Direction Currents on and Gaussian<br />

SST Combined<br />

for an Idealized Anticyclonic Gaussian Eddy and Winds from Various Directions<br />

Eddy-induced and SST-induced <strong>Ekman</strong> <strong>Pumping</strong><br />

over a 0.5°C SST Anomaly at 30°S<br />

cm day -1<br />

Contour Interval<br />

is 3 cm day -1<br />

0.5 deg<br />

C temp<br />

anomaly


m s -1 per<br />

How does this eddy-induced <strong>Ekman</strong> pumping compare<br />

with <strong>Ekman</strong> pumping associated with the large-scale wind field?<br />

Combined SST and Eddy <strong>Induced</strong><br />

−→<br />

u a − −→ a −<br />

u o<br />

−→ u o<br />

D|urel| −→ • 30°S<br />

u • 20 cm amp.<br />

rel<br />

• 10 ms-1 wind<br />

• Max current<br />

40 cm s-1 W from Currents and 0.5 E oC SST Anomaly<br />

<strong>Ekman</strong> <strong>Pumping</strong><br />

WE = 1<br />

ρf ∇×−→ τ<br />

cm day -1<br />

Normalized Distance Contour Interval<br />

is 3 cm day Risien and Chelton, 2008<br />

from Eddy Centroid<br />

• <strong>Ekman</strong> pumping due to eddy-induced SST anomalies is secondary to the<br />

upwelling induced by the eddy surface currents<br />

• Eddy-induced <strong>Ekman</strong> pumping is of the same order of magnitude as basin<br />

scale <strong>Ekman</strong> pumping<br />

-1<br />

Eddy-induced <strong>Ekman</strong> pumping is an Order-1 Perturbation<br />

of the <strong>Ekman</strong> pumping associated with the large-scale wind field.<br />

cm da<br />

0.5 deg


5. Is the previous <strong>Ekman</strong> pumping for an idealized<br />

Gaussian eddy observed in the QuikSCAT data?


5. Is the previous <strong>Ekman</strong> pumping for an idealized<br />

Gaussian eddy observed in the QuikSCAT data?


Composite Averages of Eddy-<strong>Induced</strong> Vorticity and<br />

<strong>Ekman</strong> <strong>Pumping</strong> Velocity in the Hawaiian Island Region<br />

Composite Averages of Current-<strong>Induced</strong> <strong>Ekman</strong> <strong>Pumping</strong><br />

Geostrophic Vorticity<br />

from Altimetry<br />

geostrophic vorticity current-induced pumping observed <strong>Ekman</strong> pumping<br />

region around the Hawaiian Islands in the Central Pacific<br />

m s-1 per 100 km<br />

cm day-1 anticyclones cm day cyclones<br />

-1<br />

contour interval<br />

2 cm<br />

Composite Averages of Current-<strong>Induced</strong> <strong>Ekman</strong> <strong>Pumping</strong><br />

Surface Current-<strong>Induced</strong><br />

<strong>Ekman</strong> Pumpming<br />

from Altimetry<br />

<strong>Ekman</strong> Pumpming<br />

from QuikSCAT<br />

region around the Hawaiian Islands in the Central Pacific<br />

cm day-1 anticyclones cyclones<br />

contour interval<br />

0.5 cm day -1<br />

contour interval<br />

0.5 cm day -1<br />

The regions of strong upwelling<br />

and downwelling are zonally<br />

elongated because the winds<br />

are predominantly nearly zonal<br />

(easterly) in this region.<br />

The somewhat weaker upwelling<br />

velocities in the QuikSCAT<br />

composites are consistent with<br />

small mislocations of the eddy<br />

centroids because of noise in<br />

the altimeter SSH fields.


Composite Averages of Eddy-<strong>Induced</strong> Vorticity and<br />

<strong>Ekman</strong> <strong>Pumping</strong> Velocity in the South Indian Ocean<br />

Composite Averages of Current-<strong>Induced</strong> <strong>Ekman</strong> <strong>Pumping</strong><br />

Geostrophic Vorticity<br />

from Altimetry<br />

geostrophic vorticity current-induced pumping observed <strong>Ekman</strong> pumping<br />

South Indian Ocean<br />

m s-1 per 100 km<br />

cm day-1 anticyclones cm day cyclones<br />

-1<br />

contour interval<br />

2 cm<br />

Composite Averages of Current-<strong>Induced</strong> <strong>Ekman</strong> <strong>Pumping</strong><br />

Surface Current-<strong>Induced</strong><br />

<strong>Ekman</strong> Pumpming<br />

from Altimetry<br />

<strong>Ekman</strong> Pumpming<br />

from QuikSCAT<br />

South Indian Ocean<br />

cm day-1 anticyclones cyclones<br />

contour interval<br />

0.5 cm day -1<br />

contour interval<br />

0.5 cm day -1<br />

The nearly circular regions of<br />

strong upwelling and downwelling<br />

are indicative of highly variable<br />

direction of the wind stress.<br />

The somewhat weaker upwelling<br />

velocities in the QuikSCAT<br />

composites are again consistent<br />

with small mislocations of the<br />

eddy centroids because of noise<br />

in the altimeter SSH fields.


6. Influence of <strong>Ekman</strong> <strong>Pumping</strong> on <strong>Oceanic</strong> Biology


<strong>Eddies</strong> Spawned from the Leeuwin Current<br />

a total of 734 anticyclonic and 818 cyclonic long-lived eddies were tracked from 2000-2008<br />

mg m -3


eddy scale<br />

Eddy eddy Radius scale<br />

AL.: EDDY-INDUCED EKMAN PUMPING X - 43<br />

eddy scale eddy scale<br />

<br />

mg m -3<br />

<br />

• Seasonally (May - September), we observed enhanced CHL at the cores of anticyclonic eddies.<br />

• Negative CHL anomalies<br />

eddy scale<br />

are a persistent feature of cyclonic eddies in this<br />

eddy<br />

region.<br />

scale<br />

<br />

contour interval<br />

1 cm day-1 Eddy-<strong>Induced</strong> <strong>Ekman</strong> <strong>Pumping</strong> Can can Sustain Phytoplankton Blooms<br />

filtered SeaWiFS chlorophyll with contours of QuikSCAT <strong>Ekman</strong> pumping<br />

contour interval<br />

1 cm day-1 Contour Interval<br />

1 cm day-1 Filtered wintertime SeaWiFS chlorophyll with contours of QuiksCAT <strong>Ekman</strong> pumping<br />

Eddy Radius<br />

mg m -3<br />

Eddy Radius<br />

• Seasonally (May - September), we observed enhanced CHL at the cores of anticyclonic eddies.<br />

• Negative CHL anomalies are a persistent feature of cyclonic eddies in this region.<br />

Eddy Radius


Conclusions<br />

1) Eddy-induced SST variability consists of horizontal advection by the<br />

azimuthal velocity of the eddies that is consistent with the coupling found<br />

previously between SST and wind speed along meandering SST fronts.<br />

2) The <strong>Ekman</strong> pumping associated with these SST anomalies over the<br />

eddy interiors is usually secondary to the <strong>Ekman</strong> pumping associated<br />

with eddy surface currents.<br />

• This <strong>Ekman</strong> pumping is clearly evident in the surface winds measured by QuikSCAT.<br />

• The small differences between the <strong>Ekman</strong> pumping velocity deduced from QuikSCAT data<br />

and the geostrophic velocity computed from altimeter data are consistent with noise in the<br />

SSH fields that results in small mislocations of the eddy centroids and hence misalignment of<br />

the <strong>Ekman</strong> pumping velocity fields.<br />

3) <strong>Ekman</strong> pumping over anticyclones appears to sustain blooms of<br />

phytoplankton within the cores of eddies in the South Indian Ocean<br />

during wintertime.<br />

• The reason that this is limited to wintertime is thought to be that the mixed layer is sufficiently<br />

deep to reach the nutricline, thus allowing the injection of nutrients ino the eddy interior where<br />

they can be utilized by phytoplankton trapped in the cores of anticyclones.

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!