total of 10 pages only may be xeroxed - Memorial University's Digital ...

CENTRE FOR NEWFOUNDLAND STUDIES
TOTAL OF 10 PAGES ONLY
MAY BE XEROXED
(Wi+houtAWs Pomiuim)

Ocean Surface Current Estimation
Using a Long-Range, Single-Station,
High-Frequency Ground Wave Radar
@Ken Hickey, B.Sc.
A THESIS SUBMITTED TO THE SCHOOL OF GRADUATE
STUDIES IN PARTIAL FULFILLMENT OF THE
REQUIREMENTS FOR THE DEGREE OF
MASTER OF ENGINEERING
FACULTY OF ENGEEEmG AND APPLIED SCIENCE
LIE.MORIAL UNIVERSITY OF Nl?\VF'OUSIlI.A\U

Abstract
Esrlm=r#oo of ocean surf- currents kon a long-range, arngle srnrlon, narrow-beam,
'ugh bquency (HF) ground nave dar (GWR) wrcm s pmeoied. Thre sprem,
located a: Cane . Race. . Crdouniiand -, a a hmlrenor . modulated -~ --- -- ~ramn~nt~i n,nr#n- r ~ ~-- ~ -- -
uous wave dar that ooerstes in the lower HF band betweetwee 5 and 8 Mhz. It hs a nominal range capability of 200 km over a 120' sector h m 61' to 181' (True).
Even though its prlmary purpose is for ofshore tazget surwillance, it can be easily
m n h d for the monitoring of oceanic surface conditions such as currents and
waves.
An experiment was performed during the fall of 1992 to test the current measuring
capability of this experimental system. This HF GWR can monitor projections of the
surface current field in azimuthal and range increments of ~ppmxhately 4' and 400
m, respectively. These projections or radial surface current components are extracted
from the kt-order contributions of the radar Doppler spectra and compared with
the estimates derived born the pmitional tr& of three Accurate Surface Tracker
drifters. The comparison demonstrates the ability of the radar to estimate radial
uurents to within one standard deviation of both current measuring techniques.
This has been demonstrated with simulated as well as actual data.
An algorithm is also presented to estimate the tangential current components
asuming the current is uniform about the location of the drifter velocity estimate.
This algorithm was tested with simulated radar data and the analysis suggests the
ermr of the tangential component to be within one standard deviation of the rada
and drifter error estimates. However, in the comparimn using the real rsdar data
these error8 were more than 2 standard deviations iager than the e m estimated
by the simuiations. These deviations have been attributed to a number of factors
such as possible beamforming errors or nan-stationary menm over the radar beam
dweU periods. Hoaiewr, since the slmulatians strongly demonstrate the potential for
mapping the vector Nrrent field without the need of a second site, the results from
this thesis are very encouraging for further development in this area.

In memory of my loving parents who passed away in March and April of 1995
May God bless them

Acknowledgements
Quite a number of individual effortr were requLed to make thb small contribution
to the engineering mmmunity pmsihle.
I would &st Ue to thank the l od investment community who pmvided the
funding to develop the Cape Race radar faeilityvia Northem Fads Systems Limited.
Without this hacking, the wrk presented here would not he pmible.
I wo?lld, semndly, like to thank the engineers who designed, bufit and managed
the development of the radar system. These individusls indude Dr. Rafaat Khan
who designed the radar waveform and the signal proesing chain. Bob Davis and Dr.
Saoudy Saoudy who designed the antenna system, Fabian Hartery who paimtalungiy
maintained it. Bdan Gamberg and Wayne Pearwn who wrote the system software,
and Barry Dave for his patient management of the project.
Dr. Jim Helhig of the Department of Fisheries and Ocem. who supervised the
experiment and processed the drifter data, is gmtefully admmledged an weU an the
captain and mew of the HMCS Skeem who deployed the buoys.
I would also iike to t hd the system m anw of Memorial's central mmputing
facilih: Randy Dodge, who gave me access M the VAXfVMS system for s ohe
development purpaee4 This wa made increasingly difficult as this system, which
contained all the pro-ing software used to generate the early results from thb
thesis, was no bngez being supported by the university during the period in early
1995 when the VAX/VMS system wa~ being replaced by a UNM platform.
Upon my departure fmm CCORe in late 1995, the data and software fmm the
VAX system was transferred to the computing fadity at the engineering faculty.
Therefore, I wuld Ue to thank Dave PCP, thesystem manager of the Faculty of En-
gineering's mmputimg facility, for helping me with the transition from the VAX/VMS
platform to a UNIX system. This was a slow pm- since the Qe and software for-

mats had to be modified for ace- by the new system.
Upon my arrival at Raytheon Canada in 1996 the data Ues were reformatted
again and forther sftm r aites were required. Therefore, I w dd Like to thank
Randy H o d of Raytheon Canada for helpinp to trader all th- to a
Raytheon's PC envimnment and rawriting the Fortran mde in MATLAB such that
I could lkbh the data procersing requirements hr this thesis.
Eric Gill, who has taken the time to proohad and make suggestions regarding the
presentation of this document, is &o gratefully a d m ~ L It ~ was . a pleasure to
work with him over the past several years in this 848 ~mce his enthusiastic approach
to his research has provided me with much inspiration.
I dsc me thanh to my partner and best friend. Roianda Ryao. Her ability to
how when and how to mdre me laugh during difficult periock in this academic en-
deawr is fondly appreciated. Her iwe and support were paramount to the completion
of this work.
Finally, I would like to thaoldully acknowledge my supervisor. Dr. John Welsh
for his unmwMg support over Ihe past 10 -. In conjunction w~th Dr. Welsh. I
would dsc like to thank the Faculty of Enaeering for giving me the opportunity to
undertake this task and MLighten myself and the HF radar mmmunity in the topic I
haw chosen ss a thesis Lr my master's degree.

Contents
Abstract
List of Figures
Lit of Tables ix
List of Symbols xi
1 Introduction 1
1 1 1mporta~ee of SLufaee Currents .................... 4
1.2 Conventional Methods for Measuring Currents ........... 5
1.3 The HF Fadm Technique ..................... 5
1.3.1 Advantages ........................... 6
1.3.2 Two Station Method ...................... 7
1.3.3 One Station Appmach ..................... 9
1.4 Literature Review ............................ 10
1.5 Smpe of Thepis ............................. 16
2 Techniqme
2.1 Introduction . .....
2.2 Radial Surface Currats
2.2.1 Theory ......

2.2.2 Mearurwent ........................... 20
2.3 Vector Surface Currents ......................... 23
2.3.1 General Problem Fombtion .................. 23
2.3 2 Solution .............................. 27
3 Experimental Data 29
3.1 Introduction ................................ 29
3.2 The Drifter Component ......................... 31
3.3 The Radar Component ......................... 33
3.3.1 Radar System .......................... 33
3.3.2 Operations ............................ 35
3.3.3 System Performanee ...................... 37
4 Data Analysis 38
4.1 Introduction ............................. 38
4.2 Drifter Data ............................... 39
4 3 Radar Data ............................... 39
4.4 Siulatioos ................................ 42
4.5 Direct Radial Radar/Drifter Comparison ................ 44
4.6 Vector Surface CurrenG Estim~tion ................. 46
4.6.1 Introduction ............................ 46
4.6.2 Sdection of Algorithmn Parameters ............... 47
4.6.3 Algorithmn Constraints ..................... 47
4.6.4 Simpli6catiorm ........................... 50
4.6.5 Data Reduction .......................... 52
4.6.6 Processing ............................. 53
5 Results and Discusdon 54
5.1 Introduction ................................ 54

5.2 Direct Radial Surface Current Comp~~ison ............... 55
5.2.1 Simulations ........................... 55
5.2.2 RealData ............................. 56
5.3 VRtor Surface Currents ......................... 62
5.3.1 Simulations ............................ 62
5.3.1.1 Radial Component ................... 62
5.3.1.2 Wential Component ................. 63
5.3.2 Real Data ............................. 64
5.3.2.1 RAld Component ................... 65
5.3.2.2 Tangential Component ................. 67
5.4 Discussion ................................. 73
6 Conclusions 75
6.1 Direct Radial Currents ....................... 75
6.2 Vector Currents .............................. 76
6.3 Discussion ................................. i7
6.4 Future Developments ........................... D
References 82
A Chmnology of the Experiment 86
B NRSL Cment Velocity Data Tapes 90
B.O.l Tape Format ........................... 90
B.0.2 Fie format ............................ 91
B.0.3 Reading The File ......................... 91
C Catalogue of the NRSL Radial Surface Current Data 95
D Cmss-Reference Table for Radial Surface Current Data 98

List of Figures
1.1 Radial component of true -tor current as monitored by the radar
along three Merent directions .....................
1.2 Geometry for hstation tdmique ...................
2.1 Radar Spectrum (sold Line) hm 6.75 MHz operation. An underly-
mg current with a radial mmponent of approximately 100 cm/s and
moving away from the radar induces a Doppler shift of approximately
0.045 Hz on thespectrum o M with no underlying mrrent (dashed
fine). ................................
2.2 Geometry of region S where a radial current component is being men-
m a d along the direction Zi .......................
2.3 Region Q is subdivided into m x n sub-regions, where each subregion,
rj, is monitored by the radar to give a radial mmponent v, along the
$ direction and the it%equi-distant cell h im the radar. .......
3.1 The positions of the three Accurate Surf- nadier drifters as recorded
by the satellite system. This raw positional data wzs subsequently wed
to estimate the drifter velocitie. . ...................
3.2 An enhaoeed view of the three AST drifter tr& with the numbers
indicating days of the experiment on which the radar mlleded d ace
mrrentdz%ta. ...............................
3.3 Layout of Cape Race radar site .....................

3.4 System architecture of Cape Raee radar . . . . . . . . . . . . . . . . 35
3.5 Coverage Regjon off Cape Raee . . . . . . . . . . . . . . . . . . . . . 36
4.1 Radial oxrent map processed hm r h data collected on Oct 21,1992
at i1:53. NST. Note the wi~d-~wn surface current Bow is predomi-
nantly landward in tbis enample. . . . . . . . . . . . . . . . . . . . . 42
4.2 Typical polar sectors used for vector current estimation . . . . . . 52
5.1 Scattergrams d simulated radial surface currents of drifter vem radar
dimat- for 30 dB we. (a) , (b) and (e) correspond to simulatioo.
horn data generated horn Buoys 1, 2 and 3, respenively. . . . . . . . 55
5.2 Seattergrams of rdalsurfaeeeeenta of driierve- rada atiiates
wing ieal data; (a) . (h) and (c) correspond to Buoys 1, 2 a d 3
respectidy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57
5.3 Comparison of currents estimated horn Buoy 3 (solid Line) and error-
barred radar data (*) . . . . . . . . . . . . . . . . . . . . . . . . . . 58
5.4 Comparison of cvrrenta estimated from Buoys 1 (dotted tine) aod 2
(dashed Line) and the ermr-barred radar data (*). . . . . . . . . . 59
5.5 Wind velocities for position 45' 14' Nand 5Z0 18' W during the q er-
imental period . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60
5.6 Scattergrams of simulated data results for -torments for azimuth
e t s of (a) 12. (b) W (c) 36'. . . . . . . . . . . . . . . . . . . . . 63
5.7 Scattergram of vector current estimates using real radar and drifter
daca for azimuth exteats of (a) 12' (b) 2W (e) 360. Note the reduction
m the standard error of the tangential component as the azimuthal
extent is increased. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65

5.8 Scattergrams of direct radialcomponent comparison using (a) the orig-
inal rdar data and (b) the bema-em& radar data. Note the similar
scatter b m n the two sets. . . . . . . . . . . . . . . . . . . . . . . 71
5.9 Sattergrams of vector component cornpariron using (a) the origio~l
radar data and (b) the beam-ored radar data. Note the inereme in
the variability of the tangential component estimates . . . . . . . . . 72
B.l &cord structure for binary radial current hles. Each record bas a
apacity for s Full 360 degrees of azimuth where the index in the record
reference the azimuth . . . . . . . . . . . . . . . . . . . . . . . . . . 92

List of Tables
5.1 Summary of statistical mmparison of simulated cment components
using an SNR of 30 dB. The mean and standard deviations of the
three datasets are denoted by - and u, respetiwly, and the standard
deviation of the regression of y on z is Labelled s. ........... 56
5.2 Summary of statistical cornparkon of radial w ent components using
real data with (1) 10 dB SNR and (2) 30 dB SNR The m- and
standard deviations of the three datasets are denoted by - and o,
respeetiwly, and the standard deviation of the re-ion of y on z a
labelled s. ............................... 57
5.3 Summary of statistical comparison of radial and tangential current
components computed horn the simulations. The mean and standard
deviations of the three datasets are denoted by - and 0, respectively,
and the standard deviation of the regression of y on 2 is labelled s.
Cases (1),(2) and (3) correspond to szimuthal extents d lZO, 20' and
3k respectively This table summarizes the statistics avaciated with
the scattergrams of Fiwe 5.6 ..................... 64

5.4 Summary d atatiitiurl comparison of radial and tangential -nt
components computed fmm the actual radar data. The mean and
standard deviations of the three datasets are denoted by - and a,
respenively, and the standard deviation of the re-ion of y on z is
labelled s. This table summarizes the statistics of the scattergrams
presented in Fme 5.7 . . . . . . . . . . . . . . . . . . . . . . . . .

List of Symbols
d : Distance irom radar sit- to origin of *station geometry
? : Surf- current vector
K : X component of surface current vector
V, : Y component of s& current vector
< : Wait -or dong direction from radar site 1 to peeition of ?
% : Unit -or dong direction fmm radar site 2 to pasition of I?
V, : Pmjenioo of I? along <
L5 : Projection of ? dong 6
L : Orrao wavelength
Radar waveMh
Velaclty of Bragg wave
Acceleration due to gravity
speed of l&t
Bragg Doppler frequency
Dopper shift induced by currents
Radial mmponent of current within a radar scattering eel1
Length of Linear array
Badwidth of .uswform t rdtted by radar
At : Single beam dwell period
Ar : Effective radar range resolution
A8 : Effective radar receive beewidth
S : Region of uniform surface cment

Q : Poiar sector which is a subset of fegion S
6 : Unit wetor along -tion of centre of receive beam j
-
v, : Radial component of V dong Cj
AR : hdid went of region Q
ae : ~ n span d of region ~ Q
rn : Number of effective range cells comprising region Q
n : Number of effective receive be- mmprising region Q
B, : Angled unit vector 4
i, g : Urut vector along x- aod y-axes, respectiwiy
u, : Radiai current component in i'" and g" effective range eeU and
-
ceceive beam, respectively, of region 4
a : Vector of all radial current projections within region Q
e' : Vector of all error tern in
0, : Vector of cosines of beam direction angle3 within region Q
-
0~ : Vector of sines of b- direction angler within region Q
F : Resuency range in Dopplec spectrum rumed for computing sipd noise Boor

Chapter 1
Introduction
The monitoring of o w surface currents is an important coastal management prob-
lem in the ares of ashore oil development. Gsheties research and maaward appli-
cations. O'i spill trajectories from tankers and offihare fields are lmgely determined
by the motion of the surface layer. The tr-port of nutrients and Bh larvae require
knowledge of water mwement for fisheries preservation research. Timely and am-
rate surface ment data is & required for search and rmue models to optimize
the size of search region zones when traekhg drifting vessels. However, the surface
current measurements obtained by conventionsl devi- such as furrent meters and
hiners may not give the required spatii or temporal detail to be optimal for the
ahwe purposes. In addition. the lo*ties mciated with deploying these devices is
difficult. t i consuming and very expensive.
An almtiw to these in sihr tffhniqum is provided by H&h Requency (HF)
radar using the ground wave mode of propagation. The HF hand ranges fmm about
3 to 30 MHz, and at these bequendes the relatively high conductiviQ of the sea
water enables the radar tr-mission to extend well beyond the line-of-sight range
experienced by traditional microwave season. Because of the HF band's dekametric
wavelengths, the dominant interanion is with the energy carrying long-wavelength
gravity waves. Thm waves are usually of mneern to &shore developers for structural
de@ purpcees. Howwer, the analysis of -uttered radio warn fmm these ocean

wae is also useful for the estimation of surf- layer physical mography and, in
particular, surface currents.
Scattermg of eiectromsgoetie w ae hm rough surfaces is a classical problem in
radio science, and the scattering of surf- or pund waves hm the sea is a special
-. If ones m e s that theom hurfaeeem be represented by a threedimensional
Fourier tramform, then the scattered wave energy can be modelled. Using this a p
proach, the theory of ground wave radar, h m an oceanopphic perspective, has
been well establhed. Barrick [I] formulated the ht and second order radar cr-
sections of the acean surf-. Mare recently, Walsh and Donnelly [2] have addressed
hmdamental eieekonmpetic scattering and propagation problems relating to HF
ground wave radar (GWR) operation in an oeean environment. The latter work hz
led to the development of new radar crcswenions of the ocean surf- to thrd aider
(Walsh et el. (31). These efforts have provided the bais for the fmherance of GWR
techlogy and haw led to signi6eant ad-- in the use of HF Doppler radar.
for the remote sensing of various oceanic surface parame-. In pwticular, narmw-
and broad-beam arrays, operating in the ground wave mode, have been sue-fully
employed in the determination d oeean eUTTMb, surface winds, and directional and
noodirectional wave parameters (Hickey and Khan 141, Hall and Walsh [S], Gill and
Walsh [6]). Such initiatives have been spurred on by the the fact that HF radars, as
compared to in rttudevices such as current meters or wave buoys, are able to provide
comprehensive synoptic oeeanopphic information over a v w 1- area.
HF radar surface current measurements are representative of the total current
(i.e. the interactive sum of g-tropic, tidal, wave, and wind generated components).
AU HF radars indirectly measure the component of the average current lying along
the ~adar Line of sight. For example, if the radar is directed at a patch of the oeean
surface, ~t will o b the component of the surface current along the direction of
obenatim. This is illustrated in Figure 1.1 where a uniform current is Rowing

southward. In t k example the same current projects &ent components dong the
three radar beam directions. T k components or projections are referred to a. the
radial components of the surf- current. By varying the op-tmg frequency of the
radar, the mean current for a variable depth may also be detammed (Barrick et d
[7l, S tem and Joy [a]). If two or more radars map the radial current field from
their common cowrage area., the vector surface current field can be estimated.
Fiwe 1.1: Radial component of true -tor current a. monitored by the radar along
three merent directions

1.1 Importance of Surface Currents
Besides wa-, of all the oceanographic parameters that may dect o&ihore oper-
ations, currents are by far the mmt important (Hodgins [9]). The total Bow at a
panidar pmition ean he regarded as the superpmition of current components kom
different forcing mechanism. These currents are highly variable. Besides being driven
by gemtropic forces and tides, they are strongly idueneed by the local surface wind
and wave fields.
Currents near the surf- are t he which are associated with the mixed layer of
the ocean. On the Grand Banks reeon off Newfoundland, for example, this layer
ranges in depth kom 25 to 100 metres helm the ocean surface. Since currents in thip
Layer are responsible for the transportstion of any materials in the layer, being able
to measure them is of great importance in the management of coastal area. The
following are a few examples of the use of this type of information.
. TYaddng of oil and other dce-borne poliutantts may he more easily performed.
. The data can he used by rd-time ice and iceberg drift prediction models.
. Search and -cue teams can me the data to estimate the trajectory of a drifting
vessel.
The safe transfer of people and equipment at sea should be performed when
svrfhce current conditions are ideal, such as when net current Bows me small.
. A hmledge of local surf- current velocities is wential for the safe mooring
of Boating structures.
. Predictive models for extreme currents rely on Long-term time series measure
ments. A database for these measmements, both spatially and temporally, can
now he developed.

1.2 Conventional Methods for Measuring Cur-
rent s
Fixed point estimates of the surf- went haw usually been performed by moored
emnt meters (Beardsley et o l PO], Hdpem Ill], WeUer and Davis [lZ]). Since these
dwices must be moored at depths exceeding 20 metres, they provide little indication
of the current at the surf-. Also, b~ause of energetic high-frequency motions in the
surface layers, data produced by these -nt meters are subject to various fo- of
contamination depending on the type of instrument, the mooring codguration and
the type of environmental co%iditioGxperienced (Smith [131).
Large-scde current circulation has conventionally been mwured by the tracldng
of floating objects. Drogued or undmgued drifters are d-ed to follow the ws-
ter such that they are only rmoimdy inaueneed by wind. Since there drift- are
tracked by aircraft, radar, wsek or ~atellite, using this technique to determine the
surface current field be quite expensive and very ticonsuming (Diemand et
a1 [14],Petrie and Benor [IS]). Mar-, the spatial distribution of driftw changes
with time ar they p a through the region of intewt and/or avoid areas of net diwr-
gence. This is important, for example, in fisheries applications for the understanding
of plankton trampart.
1.3 The HF Radar Technique
Cmmbie [26] deduced experimentally that the current beneath the surface waMs of
a patch of ocean could be measured indirectly by spectrally andHng the returned
radar signal from that region. This was Later mn6nned, theoretically. by Barrick
[I] and since Crombie's initial inwstigatiorw, goad quality m-nts of surf-
currents have been obtained in North America, Europe and Australia using tm
station system (Lipa and Barrick [lq, Prandle and Ryder [IS], Jaoopaul et 01. (191).

1.3.1 Advantages
There are numerous admntw in using HE' radar technology for orcurrent m-e-
ment. Many of these are attributed to the unique allweather capability pmvided by
the techno lag^ for long-term synoptic m n t mapping. The fouooing advantages
are especially noteworthy.
. Measurement of currents near the ocean surface is diffidt using conventional
systems and the formation of heresolution ment mapa has been virtually
impmsible. HF radar techniques awrmme these dldties when two stations
are used to survey their common coverage areas.
Extraction of current patterns using conventional methods -ot be performed
in near red-time. Therefore, these methods are not useful for any real-time
response such as would be desM in the ~mplementation of oil-spill eounter-
measures.
A databare of surface currents can now be constructed to provide a better
understandingof circulation patteras in regions of exploration a d development,
with a goal to ultimately understand how operations may be dected in these
repiom.
Manwent requirements are dramatically reduced, as the overhead associated
with eonwntianal methods is eliminated by virtue of the fact that the currents
are remotely sensed h m a laod-based station far removed kom the region of
interest.
. The pmspect of having continuous surface current data should provide the
impetus to correlate currents with their short-term driving forcer such as winds,
wave and tides. Potentially, such a correlation could be a means of using this
data to mesure thase driving forcer indirectly. Since these form &ct ice and

iceberg motion, a better relationship between currents and their eficts can be
obtained (Dinsmore [201).
Icebergs, bergy bits, and growlers may attain high veioclties under the iduenee
of surf- currents. Therefore, detailed current records of the Grand Banlis area,
for example, may aid in the design of any platformed structures that may reside
in the region.
Real-time surfarr current records can aid in the prediction of the hehaviour of
ice over short t m intervals. Even though the general drift of ice over a few
days can be reasonably well defined, its short term m o m may not be.
Since the use of HF radar technology alloars surface currents to be mapped over
large spatial and temporal extents, it b dear that this tffhnology offers superior and
more emoomically fessibie capabilities than conventional in aihr devices.
1.3.2 Two Station Method
Oman surface currents have been mapped by HF radar since the 1970's using short
range (less than 60 km) broad-beam and nurow-beam systems. However, these short
range systems were designed to m eme near

are not &om an orthogonal reference frame. Hoarem, it can be shown
that j can be obtained from the foUoaring tdomtion
where d is the distance from the origin to either site arhile r, and rz are the distance
&om (x,y) to both sites, respmtiveLy This equation con- the +site rsdid data
(V,,V2) into a Cartesian vector (V,,Vv). It may be noted from this tramformation
Figure 1.2: Geometry for +station technique
that if y is small or x is large, the determination of Vv becomes unstable. Along
the Line between the two radar stations which is referred to as the bascline, the

magnitude of the radid component is the m e from both site. T k is commonly
referred to a~ the baseline instability. Ak, at rangen which are much Larger than the
site separation, the same radial component is sensed h m both sites. Therefore, the
spacing of these stations must be d eMed fmm the single station range and the
required mverage. However, if a two-station setup is logistically difficult to deploy
an alternative appro& may be needed.
1.3.3 One Station Approach
If the surface current field can he estimated using only one station, then the problems
esociated with the *station appmaeh can be avoided. First of dl, the geometric
constraints no longer apply since there are no baseline or range dependent insta-
bilities. Also, hardware casts and manpower demands are minimized. Inter-site
communication m also not necessary. This is especially vsefvl in regioor, where only
one site can be constructed due to geographic constraints.
The surface current vector &mated from the radial data is based on the w
sumptlon of a uniform current about the region where a point estimate is required.
If this is the ease the radial components within this region wiu simply be dierent
projections of this current. By fitting the data in a lesst-~quarer sense we can obtain
the original vector current. The quality of the fit can be determined by analysing the
variance of the errozs from the estimated and actual pmjeetions. For example, if the
standard error is small as compared to the computed current magnitude, then the
vector current is more likely to be d i m in the region. As well, an average current
estimate may also be obtained if detailed current information is not required, for
example in the estimation of the mean current vector over a large area for mescale
-nt studis. H~we~wee~we, this is dependent on the required application.

1.4 Literature Review
The origm of surface current meamcements uslng HF radar can be traced to Crombie
[I61 when he analysed the Doppler hueney shi of a 13.56 MHz radio transmission
which .uar reflected born the ocean surface. He observed a large peak in the radar
spectra at 0.38 Hz. His explanation of tbk phenomena was that the sea waves act as
diffraction gratings and "Bragg scat& is observed. This is the same phenomenon
which is responsible for the scattering of X-rays in crystals and light rays born halc-
grams. Under g ibn wind conditions, ocean wave of varying lengths are generated.
Of the warn, those which have a wavelength L and are travelling to and away
from the tr-mitter will re8eCt back a co~tmetive sC& when L = XI2 where X
is the radar wavelength. Since we !am the velocity of the sea wave, the Doppler
shiR of this enhanced sipal can be easily computed. From Crombie's calculations he
determined the Doppler bequeney to be approximately 0.37 Hz, which arar in dose
agreement with his experimental result.
The theory explaining Crombie's observations was not developed until some yem
bter. Wait [Zl] observed that since the amplitude of the scatted signal is quite large,
the reflection or scsttemg coefficient of the sea must be large rn well. He, Wait [221.
described a simple but quantitative theory far the resomat type of reflections which
are seen in the Doppler spectra at HF by considering the ocean surface as "gently
ripplea' In tbk treatment, he considers a formulation for the signal received bom
an o-n patch which has been irradiated with energy fmm a vertical electric dipole
when the antenna is tr-mitting pulses. From his mult it is evident that the most
sipnificant response oc- when the radar wavelength is twice the wavelength of the
oeean (vdm which are w-ed to be r - m i b l This is in -meat
with Crombie's hypothesis in 1955 1161.
The problem of kst-arder HF radio wave scattering bom a moving target such
as the sea surface was addressed by Barrick [I]. Based on mrk by Peake [23] and

Rice 1241, Barrick [I] developed an expression for the average badacattered signal
power density spectrum. This qr-ion, when integrated with respect to frequency
clearly shows that the ocean surface pmduees scatter by the Bragg mechanism, thus
supporting Crombie's hypothesis in 1955[16] and Wait's fornulation [22].
Since HF rdections From the sea surface are scattered mnsrmctively fmm ocean
wavelengths, pmportional to the radar 1~~velength, it WBS envisioned by Crombie that
a mdti-Frequency HF system mdd be wed as a swwave spectrometer. That is, the
amplitudes of the dominant spectral mmponents may be related to the waveheight of
the gravity wa- whose wavelengths interact coastructively with the radar wave. In
the 1960's, experiments were performed at Eigjn Air Force Base on the Florida coast
where an ionspheric sounder, modi6ed to radiate in the horizontal direction, war
used to mUeet the sea echo rdections in the 2 to 10 MHz band. The intent of these
experiments ara~ to mmpare the theoretical and measured Doppler shifts wer the
Lower HF band (Cromhie 1251). However, there were diserepaneies between the simple
theory that Cromhie propaned in 1955 and the actual measurements. The &st-order
spectral components were slightly shined fmm their theoreticai dues, which resuited
in a velocity discrepancy of approximately 40 cm/s. Since this war comparable with
the hstorieal magnitudes of the ocean surf- -rents off the Florida mart, he
armired that ocean surfme currents, induced by waves, winds and tides, amounted
for thir shift. That is, the whole water surface moves due to underlying currents.
Another experiment was performed at Jupiter Inlet, Florida, which utilized two
spaced receiving antennae for direction-6ndiog purposes to determine the directions
in which the sea waves were travelling. By considering the coherence function of
the amals received at the two antema=, Cmmbie 1261 showed that mnstrunive
backscatter sign& were received over a wide range of azimuths and their Doppler
shifts indicated that the current was Bowing nearly perpendicular to the mast and
in the same diretion as the wind.

Fmha experiments were conducted at a HF M t y on Sao Clemente Island dur-
ing the early 1970's to monitor -rents. Thir installation used a phaped-array to
generate a narrow beam of approximately 15O along two s e a 4 directions. Dur-
ing one experiment the current E m obserwd by the radar was consistent with the
wind conditions, in that the current was approximately 3.5% of the wind speed (Wu
[27]). A heuristic appmaeh was used to explain the radar measured component of
the current (Barrick et ol. [q). In mother experiment drifting buoys, which auld
monitor the current at varying depths, were used to compare the radar and drifter-
induced currents (Stewart and Joy [a]). These eqerimerimts strongly suggested the
HF technique was a valid means of measuring ocean surface currents.
In 1975 , NOAA's Wave Propagation Laboratory developed a HF short range a ys
tem called the Coastal Ocean Dynamics Applications Radar (CODAR). This system
muld monitor surfaee currents, to a range of 70 km, fmm two shored-based stations
separsted by approximately 20 Irm. Eaeh station could measure the radial mmpo-
Dent of the surface current field of their common coverage areas. Hence, the vector
component muld be constructed from geometric mns~derations. CODAR operation
was based on Crombie's M ement interfernmeter system in that a small aperture
antenna using a direction-bding technique, developed by Munk d d. in 1963 [28],
was used to resolve the direetim of a signal om 360' with a four-element square
array. Since this small aperture srjtem was compact and portable the real estate
requirements. which are needed to form a narrow beam using a linear phased array,
were not necessary These systems were used in over 20 experiments by NOAA and
other agencies in order to wrify the technique. Rerults of these investigations re-
port msldmum magnitude and directiad errom of 15 em/s and lo0, respectively. A
sample of these investigations can be found in Holbmok and Rweh [29], Janopaul
et d. [19], Lawence and Smith PO] and Hodgins [31]. Other gmups in Europe and
Australia also verified the HF radar technique using a mmbination of short and long

ange broad and narrow-ham systems (Collar et ol. [32], Venn et 01. [33], Esen et
d. [34], Prandle [35]. Hemn et ol. [36]).
This aforementioned HE ~ology was transferred to Canada in the early 1980'3
as part of a t&bed pm- to study the feasibility of using HF radar for the wer-
the-horizondetectionof ice (Butt et 01. [3q, Butt andHickey 1381). The researchwas
initiated by Dr. John Walsh of The Faculty of %neering at Memorial University
of Newfoundland (MUN) and was comprised not ody d ice detection research. but
aLFo research in the detection of ships, h-Bying ahaft, and o-ogrgrphic dace
parameter phenomenon such as nurents and waves. The 61% use of thir new tech-
nology for the purpose of e mnt surveillanee in Canada wss in the winter of 1983.
Two stations, one situated at Cappabayden, Newfoundland (NF), and the other at
Cape Race, NF, were used to eoUect HF radar data b m the iee-infested mtea off
the Ballard Bank. The sumeeaful initial trial was the Lst instance whereby current
maps w e obtained from radar data that was contaminated by ice refieetions
(Butt et oL 1371). Many mare successful trials followed and the system wa5 eontin-
"ally upgraded throughout the 1980's to help &date Walsh's models, via ice, ship.
wave, and current trials, using various narrow- and broad-beam receiving antenna
configurations (W&h et ol. [39], GU and Walsh 161, W&h and Srivastava [40]).
In the late 1980's the research &rt in this area, at MUN, was at a matvie
stage. A local compaoy, Northern Radar Systems Limited (NRSL), was formed to
commercialize the reulta fmm the previous ten years d resea&. The hrst project
for this new company was the complete eomtruction of a multi-purpose, Long range.
shore based gmund wave radar system at Cape Race, NF. This steerable narrow-
beam system wss deigned to provide continuous surveillance of a 120' sector of ocean
encompassing the greater Gcand B&, induding the Hibernia and Terra Nm ail
fields, out to a madmum range of 400 b.
One of the iatended data products b m this new lo~-rangesystern is the pzodue-

tioo of merit maps. However, it is nee- to demonstrate that a single long-mnge
station can pmvide adequate radial current estimates. One of the goah of this thesis
is to report on this capability, since experimental d t s from such an exercise are
not documented in the literature. As well, the new drifters I& in the experiment
are minimally iduend by wind, which har deeted other part radarjdriRer com-
parisons using short-mge bmad-beam systems. Therefore, a unique opportunity is
now available to present a more r&tie comparison between the eurrent components
hom both techniques.
Testation vector current mapping is aLw plausible provided other stations are
built which are strategically located to shere a common coverage zme. The suc~sful
demonstration of this capability would permit the radar to map the total current
&or field. For example, the dats kom Cape Race could be mupled with data from
a system at Cape Bonavista to routinely map the currents on the Grand Banks. This
reglon, which b already horn a~ one of the world's best fishing grouods, b now
bang developed because d its rich oil rere-. Reliable and timely surface current
data 1s certainly usehl For further ofihore development planning. However, surface
c~reot data in this r@on is sparee and has typically b- mnfined to -rent meter
records at points of drilling activity. Smee there are no detailed current remrds from
this area, the radar data wodd provide valuable information in the study OF the
physical oeeaooqaphy of the &e layer of the Grand Banks. This is even more
mmpelling considering the recent muapse of the eod sta* and the apparent iack of
understanding of the relationship of currents to this problem.
Since two stations are nody required to estimate vector currents, the faeility
at Cape Race provides an opportunity to test the feasibility of using a sge-station,
long-range radar to remotely estimate the total went field. This has not been
documented in the literature for a long-range system. Inwstigations io this area
have been confined oniy to one short-rage srjtem (Lipa and Barrick [17]). However.

the shon-r- radar is quite diEerent from the Cape Race fmiity in that it uses
a ermed-loop antetee. system for broad-beam reeption. Lipa and Barrick 1171 for-
mulated an expression for the radial ewent components as a Linear hmunetion of the
current magnitude and a non-linear function of direnion. To solve there functions, a
closed-form solution is used to estimate the -rent magnitude while the non-linear
portion of the formulation e linearized and treated to a least sqvares solution. Erron
using thia formulation were found to range from 4 to 20 -1s in magnitude and 12-
to 30' in angle. Horn, there no sea truthing for the renrlts presented and
consequently the above error statements could not be validated.
Another technique using a single station radar for War current estimation is the
spaced antenna (SA) technique which has been used for atmospheric radm in the
determination of wind velocities sine 1950 (Brigp et of. 1411). In 1989, this method
was applied to a singlestation shorLrange, narrow-beam, HE system for nurent
mesuremeats (May [42]). In that method, it can be sboam that the bdcattered
electric field is translated aerass the receiving antenna syjtem at a velocity of 2V
where V is the tangential current velocity in the scattering area. Therefore, the cr-
correlation hetion of the signal detected by two aatenna separated by a dintam d
will have a peak at a time lag of d/2V. This technique could not be implemented using
the experimental data in this thesri became the raw data was irreversibly proceased
to yield radial current maps. Hawever, the radar could be m-ed to test such a
technique by processing the baseband signal in the above m er. For this care, the
oridnal Linesr array would need ta be partitioned into two subarrays. In aoy event,
there is no evidence in the literature of any sea-truthing experiments which could
verify this method.

1.5 Scope of Thesis
The purpme of this thesis is twofold. The &-st objectwe is to validate the ability
of the Cape Rae radar to monitor the radial current 6eld. This system is the &st
of its kind in North America and it was not svbjsted to any previom validation
experiments in this area. Hence, the need for an investiffatin of this kind is imper-
ative to test and further promote the technique. The second objective is to examine
the passihility of using the radial data to estimate total nurent vectors. This has
not been prwiously attempted using a long-range -em. Therefore, an experiment
of this nature pmvide. a valuable dataset for testing algorithmme to promote this
capability.
In the -tor current method presented in this thesis, the radial components are
formulated as a system of linear equations which were derived directly from the
tem geometry. The u n l m in ~ this system are the current components which can
be determined from a closed-form solution using a simple re-on analysis. This
alleviates the errors introduced in linearizing the system of equations derived by the
method presented in Lipa and Barrick [17]. The end result is a much simpler proea
dure which can be esrily implemented as a real-time software module in a practical
radar sptem.
The general techaiques used to pro- the radar data are d-ibed in Chapter 2.
The elanrid technique used to spectrally analyze the radar data will be summarized,
with a description of the current estimation method med to yield the radial nurents.
This will be followed by the formulation of the vector current aigarithmn and its
soiution.
The experiment is discussed in Chapter 3. Normally, this would be dealt with as
an appendix. H-wr, since this is essentially an experimental thesis where the data
for mod4 validation was provided by the Northern Radar Cape Race facility from
October 20 to November 21, 1992, it is felt that a description of the same is deserving
16

of a chapter. Thig dso provides an appropriate intermission from the radar theory,
es it discusses the experiment &om a practical standpoint before the data analysis
pmcedures are presented in the foliowing chapter.
Chapter 4 presents the data analysis pmcedm used for the drifter and radar
data and discusses the possible emm arrociated 4th both techniques. The data
simulations, which were designed to alleviate the complexities arsociated with typical
radar simulations, are ah dLmmed here.
The results from the data analysis procedures a e discussed in Chapter 5. These
mulu provide a mmparisan of the simulated and the actual radarfdriner current
estimate.
Condusions from this study d be presented in Chapter 6. Findy, some ideas
for future deveio~ments in this sea are considered.

Chapter 2
Technique
2.1 Introduction
The radar coverage zone he partitioned into a polar grid of radar scattering
cells, eaeh of which 1s specified by the radar's range and azimuthal mlution. A HF
Doppler spectrum from each of these ceb is charmterized by two dominant pe&
spaced symmetrically about the carrier. The frequencies of these peeks are used by all
HF GWRwtems for the m-uement of ocean surface m t s since they indirectly
trace the underlying average surfme -rent component. These camponents, or radial
current projections, can hs estimated for all eeUs in the radar ewerage zone to yeld
a grid d the radial current field. How these datagrids are generated will be discussed
m the following sections.
For the estimation of wetor currents, a technique will he presented which as-
sumes that the current is uniform about the estimation point. In this ease adjacent
azimuthal radial current components will he different projections of this mrrent. A
geometric linear equation describes this relationship for one radar cell. A bet of these
equations can be generated for dl cells which fall within the assumed uniform c mnt
ream. These equations can then he subjected to a linear regression analysis to yield
an estimate of the vector current components.

2.2 Radial Surface Currents
2.2.1 Theory
The motion of the ocean we.- is seen by the radar ss a translation of the frequency
of the reeelved echo signal fmm that of the carrier. In kping with the Bragg scatter-
ring phenomenon, the dominant received radiation is resonant scattering h m ocean
waves trading to-& or away horn the radar and having s wavelength one half
that of the transmitted sf@. In this mode, the radar em- ao ocean wave
spstmmeter since it predominantly observes only these wavetrains. This interaction
induces on the radar Doppler frequency spectrum symmetrically s p d peaks whoae
positions are dictated by the pbase velocities of the advancing and receding wave
trains, respectively. In the HF commuaih: these are commonly referred to as Bragg
lines. In the absence of currents, these lines have well defined Doppler frequencies
If a current exists beneath the surf- waver, the radar sees this as a translation of
the co-ordinate frame of the scattering wave trains. This imparts another frequency
shift on the Bragg lines wbieh is proportional to the radial component of the current
velocity dong the radar receiving drection'.
The following short mathematical explanation of the above pmcess dl be pre-
sented next since it mneisely summarize. the above parsgraphs. For ao excellent
introduction on fundamental radar principles the reader should refer to the text
"Radar Principles" by Nadav Levanon(431.
It is well esttabhshed tbt for p~grazingineidence, the velocity, K, of the Brauwave
ss derived hom the deep water gravity wave dispersion relation is given by
I~he bmk of the mathematical briustions of the a h mnap~s and tbessartemg
a not be preented here sinsin thhh h beaed eaed f"dSmenta1 radar prhcip1ei. Inned, only the
prhcipal equations gmwnhg Dopplm ahiffa which o-te kom mwiog tars* be marldeed

where X is the radar waveIMpth and 9 is the acceleration due to pavity The come-
sponding Bmgg hquency, fa, induced on the Doppler spectnun due to the motLon
of the wavetrain in the ahsetme of =face eurrents is given by
f b = G (2.2)
which at 6.75 MHz is 0.2651 Hz. The edstence of surface currents shifts the dominant
peaks h m their theoretid dues as in show by the example presented in Fme 2 1. The difference between the frequency of the theoretical Bragg wave and the one
determined h m the radar data is a measure of the average radial component of the
sdxe current to a depth of appmximately X/8= m (Stewart and Joy [8]). This is
approximately 2 m for a radar operating at 6.75 m.
Since the radial current measurement consists of determining the shift Af from
its theoreti4 value of fa, h m the above equations it can be shm that the radid
component v, of the surface current within a radar scattering cell is
far an ideal HF radar system.
2.2.2 Measurement
To remotely memure ocean surface currents h m a HF system, the azimuthal r-
iution pmded by the receiving antema beam and the radar's r- resolution are
used to isolate the radar return from a resolution di on the ocean surf-. The
radar echo fluctuation from this cell is indirectly sampled at the waveform repetition
frequency and the resulting time series of samples are subjected to a k t Fourier
m-form (FFT). T~LI generates the Doppler frequency spectrum.
Rom Seetian 2.2.1, it is clear that the accuracy of the current measurement is
dependent on the muracy of the computation of the frequency of the two Bragg

Figure 2.1: Radw Spectrum (solid line) bm 6.75 MHz operation. An underlying
current with a radial component of approximately 100 em/s and moving away from
the radar induces a Dopplershift of apprmdmately 0.045 Hz on the spectrumobserved
vlth no underlying m n t (dashed line).
pe& Since the frequency computation is performed by using a standard soft-e
FFT implementation, the frequency rerolution of the spectrum is the reciprocal of the
time interval, in seconds, over which the data from each cell is sampled. This tune
interval or coherent integration time (CIT) restricts the radar's ability to "pinpoint'
the tiue Doppim bequency using the periodogram appro&. Hence,the actual peak
may be 'somewhere between' two adjacent kqnency bins. The estimation of the
'mcnn' or centre h.rl\nency of a 3pectral pePk rs performed wrrh the mrmnonly urm
>It u ab-""d ,hat rdsl men, dcwuolu u d
by LDB~~YLR~OLR~O
onon uu,&m,,
' mlu3arwl ru the c uml raalutrou LNVI&~ by th. FM

'centroid' definition, while an estimate of the ermr in this calculation is provided by
the FFT resolution. Thus, a radial current estimate, as well as an estimate of its
error. is obtained h m each rada spectrum.
To illustrate the above mncepts, an example i. presented using the system param-
eters of the Cape Race radar, namely t hw associated with the radar's spatial and
Doppler renolutiom. The azimuthal resolution ofa hear array is approximately AID
radiaas where A is the radar wavelength and D is the Length of the array (Cob [MI).
Since the receive array at Cape Race. circa 1992, wa. 866 m in leogth and the centre
radar frequency m8 6.75 Mhz, the approximate radar beamwidth is 3 degreesP. The
radar's range miutlon is determined by the bandwidth, f, of the transmit wavefo~m
and is giwn by c/(2f) where e is the velociCy of light. Since the radar transmusion
bandwidth used to generate the experimental data of thk report aras 375 ldIz the
range re~olution for the experiment is approximately 4M1 m. Therefore, the cell size
at ranges of 50, 100 and 200 1rm are approximately 1, 2 and 4 square km, respstively.
To determine the approximate current frequency resolution we need to refer to the
wawform mpetition mterval which, for the expeMlental data, was apprmrimately 0.6
sands. 'Qpicall~ 512 mmples were collected at this sampling interval, per radar
beam, to yield a CIT of approximately 300 semnds. Smce the frequency resolution is
the reciprocal of this time interval, the resultant Doppler resolution is apprmdmately
,003 Hertz. The mrrespondiog current velocity resolution, using Equation 2.3. is
approximately 7 em/s. In other mrds, the current estimate is approximate to within
13.5 wn/s of the FET estimate. The azimuth associated with this estimate is ob
tained by the appropriate phasing of the received signal at each antenna. Therefore,
to ewer a selected azimuthal extent in the radar ewerage zone, we can dwell in a
direction for a time intenral, At, then 'shift dimion and dwell' until the region is
coveted. We can t hdre generate radial velocity estimates at range aod andar
'This that each el-t ia ia ktmplc SemoT. aoa*ew, the beamwidth thtuauy vari~a
from appmdmately 2 8 to 6.0" beoava of the sut-amay pat-

intervals of Ar and A.9, where thee parameters are the nominal radar range resolu-
tion and heamwidth, respectively. (See Khan cf d [45] for a complete description of
the dwign OF the Cape Hace system).
2.3 Vector Surface Currents
2.3.1 General Problem Formulation
A won. S, of raoge resolution Ar and azimuth resolution A.9 in the cowrage area
of a HF radar is considered. It is assumed that a uniform aurfaee cumnt, v, exiin
S duing the dwell time for this region. This gwmetry is sh- in Figure 2.2
where the radar h situated at the origin.
Let the unit vector Z, point in the direction of the centre of a radar receive beam
such that it makes an angle 8, with respect to the x-ivds of the eo-ordinate systan.
Let the uniform current vector, which projeeffi itself along the beam drection, be
denoted by ?. Hence, at the pwition (r,8,) in S, a radial current measurement d
be monitored by the radar.
In thL4 wordinate system. ? h
where i: aod p are voit -ton dong the x- and y- A, respectively and VV. and
V, are the components OF ?. Since we can represent the unit vector in the radial
direction perpendicular to Cij a~
and the radial component of ? along ths direction as
it foUows that
+
VJ=V.$, (2.6)
V, = v, CS(B,) + v,*in(e,) (2.7)
23

Figure 2.2: Geometry of region Swhere a rad~al cumeat component is being measured
dong the direction *>
which is the projection of !? along the direction of i*
Next, a subset of region 5, say Q, is considered. The total radial and angular
exteas of Q are AR and AB, e~peetivel~. Further, the region Q is subdivided into
subsecton of range extent AT and angular extent A0 such that

Figure 2.3: Wan Q is mbdivided into rn x n subregions, where each subregion,
ij, is mmitmed by the radar to give a radial mmponent u,- dong the $ direction
and the tt"udistant cell from the radar.
where n be- interset the region and A#, is the beamwidth of be- number I.
This geometry is shown in Figure 2.3.
Let them range cells within 4 be indexed from 0 to m - 1 and let a rage cell
within this range be indexed by i. Let's mnsider the radial current from the tjLh eeu,
where j is the index of the beam which makes an angle 8, with the x--. Sice
this current obtained from an actual radar system consists of an error mmponent,
e,,, equation (2.7) may be remtten a~

Let eo be the smallest angle a beam in Q m h with the x-:-ads. Therefore, iin
-
be- intersect this region lue have for n radial projections, w. v,. . . . , v.-, of v in
the it" range sector of Q
Over the entii region, Q, there are m sets of such equations arhieh may be cast

The abwe system of equations is swerely over-deterrmned since we haw 2 un-
lmowns in m x n equations. Howewr, the solution of such a system is well horn
and will be dkussed in the following sectLon.
2.3.2 Solution
Consider the system of equations which were derived fmm the previous section. If
we let the radial m n t components be mitten ss
"ID

Tb is an example of a multiple regression model with two reg-rs. The model
is non-linear in the two regressor a, and m2, but it is linear in the unkn-
repression coefficients V, and V,. If are mume the error terms, s, in the model have
zero mean and mnstant variance, and that the errors are uncomelated, we can proceed
to use standard resesion teehniquea to estimate the s h e
current pluameters V,
and V..

Chapter 3
Experimental Data
3.1 Introduction
The prime objective of thia Raesnb is to investigate the surface went sensing
capabilities of a long-range, singlastation, HF GWK This has been documented
with tw-station, short-range HF installations, as described in the Literature review
of Chap- 1. However, it ha. not been documented using a single Long-range system.
A single long-range station, for example, with a range and azimuthal coverage of
approximately 250 Irm and 120D, hss the potential to map an ma of apprdmhtely
66,000 square kilometers. Therefore, agencies respooslhle for monitoring the exelusive
economic zozoes of coastal nations may hod data products from such a system to be
extremely informative, timely, and emt-effetive. As a renult of shrinkhg national
budgets for such semi-, joint venturer involving many nations have evolved. For
example, the experiment upon which this thesis is bssed was part of a larger study
by the Canadian sod U.S. Coast Guards, a. well s~ the Canadian Department of
Fisheries and Oceans (DM)). Their primary goal is to improw the drln predictions
of search and rescue models. Presently the search and reme models la& adequate
surface current data for the estimation of Jearch regions. Therefore, the information
provlded by a HF system muld enhance the vsefulners of theae models.
Another objective of HE radar experiments is for hheries r-ch since o-ts
play amleinnutrient tr-port. The DFO- interested in theabilityofHF GWRto

estimate ocean surface axrents since the late 1980's when a portable HF system anu
deployed to map currents in Conception Bay, NF. This system, the CODAR which
w introduced in Chapter 1, consists of two stations which memure the components
of the surface current Beld radial kom each station to a range of approximately 30
km in polar increments of 1.2 km and 5O. In this was the -tor eurwt &Id may
be calculated fmm strictly geometric considerations by using the radid current pids
produced fmm each station (Barrik cf a1[46]). Since surface eurrene play a major
roie in the dispersion or 'flmhing out' of fish larwie contained in these inlets, this
system auld provide a clears picture, both temporally and spatially, of the current
circulation patterns which DFU bad di5eulty in obtambg via in dtt methods. DFO
was only capable of monitoring Eulerian ments and a few Lagra.gian measure
ments with &drifters. This was a wry time consuming exercise thst provlded sparse
current vectors as compared to the radar technique. which could provide a map of
the CircuLation patteras an an hourly basis. Since many of the standard instruments,
such as current meten, must be deployed to obtam the memurement scale that DFO
required, the CODAR system w a ideally suited for their data requirements. For ex-
ample, a gtid of current data could he constructed with a resolut~on of appmxhe.tely
1.2 km wer a 2000 square km zone.
The DFO m & interested in the -rent pattern over the Grand Banks since
these currents &ct nutrient circulation in this rich fishing zone. Since the portable
system d~d not haw the capability for these long range measurements, an alternative
system was needed. This new system anu built in 1990 at Cape Race, by NRSL. 6om
local gownment and inwstor hmding. It was designed primarily as a ship detection
radar system but it was & capable of mapping currents. The radar has been
demonttrated to have a nominal range of 200 km over a 120' %tor, which entailed a
large portion of the the Grand B ah. Even though this system could measure only
the radial components of the current fieid it was envisioned that other systems would

e built at other eoastd sites such that the vector field may be mnstructed between
stations with dud coverage. Hence, it was adwtageous for DFO to bemme involvd
at this stage of the program dwelopment since the data pmdvets provided by these
modem radar system mold aid in future SJheries research, especially in light of the
collapse of the md bhery in Newfoundland.
3.2 The Drifter Component
Theexperiment commenced on October 20,1992 with the deployment of three Seimae
Limited of Canada Accurate Suriaee naeker (AST) buoys an the Southern Grand
B& (Figure 3.1).
Figure 3.1: The positions of the three Accurate Surface Rder drifters ar recorded
by the satell~te system. This raw paitiand data luss subequently used to estimate
the drifter velocities.

Drifter pasition. were obtained from the ARGOS satellite system, in near real
time, app~teiy 10 times My. The 0.5m diameter by 1.0-m long barrel-shaped
AST b- were dwigned to ae-ately track the Mi of the top metre of the ocean.
Although these buoys are not b e d , they me designed to fill with water on &ploy-
ment and are bottom weighted so that they remain upright. Extensive triaLs during
the Canadian Atlantk Storms Project, as doevmented by Anderson and Shaw [47l,
prod that they have excellent drift characteristics. At 6.75 MHz the radar provides
an aver* current wloeiv for about the top 2 m of the water mlumn (Stewart and
Jay [S]). Thus. the buoys and the radar sampled esrentially the same oceanographic
repime.
--*--++'
13. 5; *
Drifter 1
1: Drifter21 Drifter 3
Figure 3.2: An enhaneed view of the three AST drifter tr& with the numbers
indicating days of the experiment on which the radar mlleeted surface current data

Since the mean currents within mast of the r dar coverage zone arr weak (0 20
cmls, Petrie and Anderson 14811, and since the surf- Bow in the autumn storm sea-
son is generally dominated by directly luind-driwn currents, the bum were deployed
in the centre of the -rage zone along the broadside look diretion of the riuiar
(120DT). at ranges of 75, 125 and 175 Irm. Unfortunately. a tropieal stonn stalled
over the Grand Bank shortly after deployment and its steady northeasterly winds
quickly dmve the inner two buoys towards the edge of the radar emage. Buoy 1.
deployed at a range of 75 km, stayed within view fewer than 5 days and returned
only 6 radial velocity intercomparisons; Buoy 2, deployed st 125 h, survived for 12
days and provided 23 mmparisons; fortunately Buoy 3, deployed at 175 km, remained
within range for mmt of the experiment and yielded 48 useful comparisons.
3.3 The Radar Component
3.3.1 Radar System
The radar sptem, designed, owned and operated by NRSL, is located on a 2.5 km
strip of land along the mast at Cape Race, as s hm in Figure 3.3.
The HF GWR site eonnistn of receive and transrmt antennas, a receive/mntral
and acmmodations building, and a transmit building The transmit antenna is a
log-periodic dipole array antenna with a half-power beamwidth of 120'(over perfect
ground), a peak power ratlng of 10 KW, and an average power of appmdmately
1 KW. The transmitted waveform is a frequency modulated interrupted continous
wave (FMICW) where the carrier is swept owr 375 kHz to obtaio a range reolution
of approximately 400 m. An operating frequency range bet- 5 and 8 MHz
adable, but the two frequencies used in this experiment were 6.25 and 6.95 MHz.
A blockdiagram ofthesystem architectwe is shovn in Figure 3.4. The &element
phased array IS partitioned into ten Celement subarrays which eao be electronically
phd, via analog beamforming units, to form a subarray pattern. Insteed of the

TX and RX Sites
0 Pond
Figure 3.3: Layout of Cap Rae radar site
approximate 114 element spacing of the original may, the effective antenna element
separation of the 10 ehaonel subarray system is now approximately 21. Subsequent
beamforming in software of these 10 channels can steer the resultant main lobe to any
direction within the sub-array pattern. Subeequently only ten receivers, as opposed
to 40, are required to perform the waveform demodulation aod analog-to-digital eon-
version. The pr-prowsor unit performs the range &scrimination, via the FFT, on
a pukeper-pulse basis using 12 inter-leaved digital signal promsing units. Doppler
proceing b performed on a VAX 11/785 system far subsequent analysis and display.
The receiving antema is an 866 metre phased array consisting of 40 planar
diamond-shaped monopoles. The beamwidth increases from appmximately 3.0' broad-
side to 6', with 61P steering from broadside. The total angular ccwerage, which in-

Figure 3.4: System architecture of Cape Race radar
cludes a large portion of the Grand B& as shorn in Figure 3.5, is from a bearing
of 61' to 181' (True). The interested reader b referred to Khan et el. [45] for a more
complete description of the radar system
The radar began tradring the buoys on October 21, 1992, and ws tvrned off on
November 22, 1992. Typically. one to three radar measurements were made daily
between the hours of 09:00 and 17:W local tie (Figure 3.2). However. data were
rendered unuseable on day 2 (Julian day 296) due to data collection software erro1.j
and on days 22-25 (Julian days 316 - 319) when hardware beam forming problems
were encountered. (See Appenb A for a chmnology of the experiment).
Sin- the surface currents of the Grand B& region sre normally less than 50
cm/s, the pmitions of the drifters m e not erpeted to vary appredably in a daily

Fire 3.5: Cowrage Region off Cape Race
period. As a redt of t b, and to reduce the data volume, the radar usually only
seaooed a region in thevicinity of thedrifcer positions. For example, the h s rffarded
positions each day were relayed to the radar operator from DFO and were used to
select a mimimum 6Wh-in-range by We-in-uimuth sectors centred on each buoy
Within the secton the radar aehiewd an approximate 40Wm by 4- resolution, which
translaw to 400 m by 5.2 Irm at 75 km range, or 4CO m by 12.2 km at 175 Irm
raoge. In addition, larger data~ets consisting of current memmments of the total
radar coverage area m e surveyed daily to measure the larger-s& surface current

spatid variability and to check system performance. These seetor extents are noted
in Appendix C.
As a data g dty check the radar aras continually monitored on a per data ruo
basis. Plots of the retuned radar signal were regularly produced to inspect range
periormance. On-line spectral pro=-ing was also performed to determine the quality
of the Bragg components. Also, ss they beearneeam available, most radial surface current
maps m e inspected for obvious errors.
Field personnel were stationed at the r& site for the duration of the experiment
so that hardware problems muid be quickly addressed.
3.3.3 System Performance
There arar one hardware failure noted during the experimental period. An inspenion
of the wtw on the morning of the 1 5 of ~ November ~ (Julian day 320) revealed that
2 of the 10 beamforming units were dektive. The defective o ts were immediately
replaced. It was Later determined these units were causing problems on November
llth (Julian day 316), as revealed by a degradation in the data quality tom that
day to the day of unit replacement. Other data I-, though very minimal, were
mainly due to minor software pmblems that occurred during the initial stage. d the
prow. (See Appendix A, for the experimental chronology).
The maximum rsdar range was obswed to be approximately 300 kilometerr, with
a minimum range of apprmdmately 200 Irm. This was sufficient for an experiment of
thk nature since the drifter positions never exceeded 2W km.

Chapter 4
Data Analysis
4.1 Introduction
The driRer and radar data analysis procedures are hrieEy d-ibed in this chapter.
The drifter analysis prod- will describe hoar the current veloeites are derived
kom the positionai data recorded by DFO fmm the ARGOS satellite system. How-
ever, since this tffbnique has been reported elsewhere (Helhig 1491). it will only he
briefly discussed. Simihiy, for the radar description, only the high-lwei pr-sing
techniques will he presented since the fundamental details of the radar system dsip",
and its implementation, have been previously reported (Khan et 01. 14.51).
A discussion of the radar data pm-iag methods will be partitioned into tmo
sections. The Grst section dele with the pmcedsing of the radar time series data
after it hap been heamformed into a grid of rangeazimuth him. The radial current
algorithm pme- the time series end generates a polar grid of radial current
speeds fmm the resultant spectra. This b the hiodamental analysis step since thjs
data, when compared to simiiar data from another m r , illustrates the radar's
ability to sense the radial component of the current. The -nd section de& with
the vector current processing data d p b . This aoalysis entail, the selection of the
primary algorithmn parameters, some pramsing mnsiderations, data reduction of
the polar grid to match the algorithm parameters, md the h al pm-ing step
which yields the orthogond current components. The current estimates prwided by
38

oth of these techniques ean then be compared to the dts from thedrifteranalysis.
Both simulated and actual data will be subjected to the algorithm.
4.2 Drifter Data
Weighted splines were fitted to the northerly aod easterly errordinates of each driner
track to form an evenly spaced time series, with t he hour inten*, of position and
velocity rn described by Helbig (491. Weights were chosen to he inversely proportional
to the &mated quality of each pasitional Ex. s supplied by Senice Arga. The
data were not low-pass atered and extensive simulations by Helbig (491 indicated the
rdtant rpeeds exhibited standard dwiations of about 5-7 m/s.
The gist of the drifter velocity calculation is to aerentiate the splined-fitted
pasitiaodd&a. It should be noted that although thespline 6ttio~ anddiEere~tI~ting
a intended to smooth irregularities from the drifter tracks, the deriwd velocities are
not eqwvalent to the radar results since they mntaio iaiormation at scds which may
be different than thase mewred by the radar. The drifter estimate is Lagrandan
in nature. glving an average due over a portion of a tradr. Howwer, the radar
meslue is Euleriao, being ao aver* estimate over the radar resolution cell, which
may he several square Idlometem. Therefore, the drifter and radar estimates may
never truly equate unless the current is uniform wer a region that is large, with
respect to the dimensions associated with both estimates. Hoover, this is the ody
Imown reliable technique to ted the radar cwent estimate since currents near the
s h e are extremely =cult to monitor.
4.3 Radar Data
Data from each radar beam direction was -pled at a rate of appmximately 1.7
Hz, as described in Section 2.2.2. This unambigously displays the spectral mntent
of the signal to apprmdmately f 0.8 Hz. which is su!Xcient for current measurement
39

since at 6.75 MHz the Bragg ~uencie. are apprmdmately +0.25 Hz. The nominal
dwell time, or time to collect a series of mnsecutiw eeha pukes, was approximately
3W seconds. Therefore, appmdmately 512 samples were obtained and spectrally
processed to give an FFT bin resolution of apprmdmately 0.0033 Hz. This yielded a
current veimity resolution of appmdmately 7 on/sec.
During any dwell period. 8ignals at the receivem are combined to yield a eon-
structive signal from one coarse direction only (See Section 3.3.1). The radar was
configured to yield 31 coarse beams, each separated by appro-tely C. Therefore,
s selected extent in azimuth may be covered by dwelling on a coarse beam for a
time inted, say At, at one &rune azimuth and then 'shifting' the beam direction
and 'dwelling' until the required szimuthal axtent is cwered. Directions between the
marse beams can be synthesized by h e beam steering, in -hare, the coarse beam
data to generate t he samples at range and angular intervals of Ar and A6, rep-
tively, where AT is the radar range resolution and A6 is the nominal radar beamwidth.
This h e beam steering calculation is the spatial equivalent of the frequency estima-
tion problem in that the freqvency resolution can be improved by either inmesing
the sampling interval, coUedig more samples, or both. In the beamforming ease,
the number of directions is increased by chmsing an appropriate angular ln-ent
and introdveing new phase factan to &steer the coarse beam data. This angular
increment is d y some integral factor of the beamwidth (Collin [44]).$
Time samples from each range-azimuth 4 were spectrally piocensed using an
FFT routine to yield the Doppler inf~nnation.~ Sine the frequency placements of
"(rmfo- '(rmg. LDcrv May wll not k &wed here mm I, u . -dud pracnluc m
radu rlgd pro-q 1% ad mlc, bomn. r b all sr

the dominant, or 9-, peaks of the spectrum have to he isolated, the data analpis
method h one of peak detection. These peak frequency pmitions were determined
from the spectra of eaeh mgaazimuth "all "sing the f0ll0~ steps:
1. A 150 emls window was placed about *fb such that the two frequencies wizh
greatest spectral po- muld be ~solated. These two peaks are the signals of interest
since they will typically be due to the Bragg waves if there h su5tient waw energy
for both components. If the radar frequency is 6.75 MHz then thjs corresponds to a
Doppler fqueney window of +0.0675 Ha about the Bragg huency. This window,
called the signal window, rvss ehmen to emre the detection of the largest expected
meat veloeitles in the experimental area (Pettie and Anderson [@I);
2. A noise Boor was calculated from the average power of all sign& in the region
0.5 Hz < Iq c 1.0 Hz where F is the Doppler frequency range. Because there are
potentially tan, Bragg sign& per radar spectrq one from the negative and pmltive
halves of the Doppler spectrum, this computation result9 in two signal and two noise
pmr measurements, one from either side of the spectrum;
3. The onasidd spectrum with the greatest signal-to-nohe ratio (SNR), of at
Least 10 dB, wan chaaen es the candidate spectrum for the current component esti-
mation; and
4. A moment calculation wa. performed by weighting the frequency components
in the signal windoar of the canmdate spectrum with the SNR dues to estimate the
central peak frequency. The SNR dues for each point used m the weighting had to
survive the minimum SNR threshold of 10 dB.
Thespectra from eafh radar cell ws subjected to the ahow pmcedure to generate
polar pi& of radial current data for each radar dateset. A sample plot of one of these
grids in shown in Figme 4.1. For display purposes each point on the grid is averaged
over a polar seetar of 20 km in range and 10' in azimuth.
period A m , Ine~pLIUlblel~lYIm hm thedata analysis may '81- mvwtlng thia mption. 41

-so
I
-59 \ -52O -5t0
I
Longtaude (Degrees East)
Figure 4.1: Radial current map processed h m radar data collected on Oft 21.1992
at 11:53, NST. Note the wind-driven surface current Bow is predominantly Landmd
in this example.
4.4 Simulations
A simulation study was undertaken to estimate an approximate emr bound on the
radar and drifter current eomparisona, hssed on the estimated errors of both tech-
niques. This m performed for the direct radial and the vector current cornparisom
to: (1) test the s o h e modules used in the data analysis procedures; and (2) to
aid in the analysis of the actual radar data in the event that anomalous results were
obtained.
For the direct radial component comparison (see Section 2.2.2) these erron can be
42

modelled, since they are known quantities. For example, the estimated radar error
per radar cell is ir h3.5 -/s while the drifter error ia J +6 em/s (Helbig 1491).
Hence, a comparison could be simulated and used as a benchmark for the aetual
radial component comparison. in this amy the statistice from s comparison of the
simulated and the real data should agree if the errors were properly modelled. The
Maor went algorithm could a h be tested using a similar scenario. Since thin
algorithm eaumes a uniform current about a test pwltion an the oeean surface, the
generation of this ment d d be simulated from any current which was ssumed
to prevail at the point, and in the vidoity of the point. Morewer, by extending the
reglon over which the benchmark c m t is mumed to exist, both the radial and the
vector current ulgoritbmn can be tested with the same simulated datasets.
The radar test data, which emulateda timesequence of actual radial cunent grid..
=re simulated arsuming the existence of a spatially uniform current fieid about a
position of comparison in the cowrage zme from each radial map. This was deemed
adequate for the dire2 radial component comparison since these positions would be
representative of ao actual radar/drifter test point. In other words, the same radial
component should essentially be estimated using either the radar or driner current
estimate. For the application of the veetor current algorithmn this may not be the
-since the a ctd current data may deviate from the uniform tea current aimate,
as the distance from the test position in=-es. Hoarever, for the simulations, good
agreement is eqected since the asumption d current uniformity rs enforced by the
p"ailing test current. By using thin ~pproaeh the statistiw of the simulation results,
and the actual datg should be similar if the actual current was indeed uniform. This
may also provide some insight in detecting nan-uniform current regions.
Since the driRer data is considered to be the sea-truthed dataset for this eompari-
Jon, emrents estimated from this data were chosen as the test or benchmark current.
For both the radar and drifter simulated datasets, the actual drifter estimate at the

midpoint of a radar data collection, in time, was taken to be the tent current. How-
ever, the ermrs associated with this drifter estimated e mnt are born to exhibit
Gausian standard errors from 5- 7 m/s. To simulate this error range, random nor-
mal deviates torn a uniform dktribution wer this interval were added to the deer
went components. For example, if the output from a random uoifocm number
generator on the above emor interval wa9 5.5 a/s, then this would be the standard
error of the Gaussian random variable. This is the thrust of the current simulat~ons
for the drifter components. Far the radar simulation the henchmsrk current
held constat for the duration of an actual mdar data collection. The current was
projmted on each of the radar beam directions used to process the actual radar data.
This step yielded a set of noiseless radar radial velocity data. Since it is has been
shown that suface current radial velocities are Gausian random variables (Barrick
and Snider [51]), the noiseless data was smeared with a random normal deviate. The
standard error of this random variable cuas chcsen to be 112 the current resolution
of an FFT frequency bin, or approximately 3.5 un/s (See Section's 2.2.2 and 4.3).
AU the radar datasets are used in this manner to generate a slmuldted version of the
anual radar data
4.5 Direct Radial Radar/Drifter Comparison
At each point of comparison a radar estimate obtained by averaging the radial
components withinaradius of 2.4 Ian from the location of thedrifter current estimate.
Since the radar lange resolution was 0.4 km, apprmimately 12 range elk along the
rdal direction were used in the averagirg prows. Ebr example, at ranges of 50,
1W and 150 !am an ardength of 4.8 Ian, subtends angles of approximately 60, 3*
and 2'. respectively. Therefore. at 50 and 150 Lrms approximately 24 and 12 points,
rerpectively, were used to compute the mean radial c-t from a region with a
surface area of approximately 20 square Irms.
44

Sine the drifter estimates m e mnstrdned to three hour intervals, a linear interpolation
of the drifter estimate - reqvjred if the radar time did not mincide +th
one of these intervals.
To illustrate the dect d SNR on the comparison, the rdar data we- pmcesed
using two SNR values of 10 and 30 dB.= The 10 dB case wss studied to determine if
this minim- SNR esse was sofF,tient for radial current estimation. In mntrast to
this, the 30 dB ease - maridered to determine if the data mmparison is Lnprowd
by using a relatively high SNR criteria. Since all three driner datasets are^ used in
this comparison, this yielded a total of six radial mmponent mmpackons. These were
mbsequmtly analysed, using a simple linear regression, to estimate the mrrelation
between the segtruthed radial nrrrents estimated fmm the drifter data and those
estimates pmvided by the radar technique.
*Sa S&n 4.3 h a dirdn of hov them SNR valuer vae ofputed frofro the radar e r a .
45

4.6 Vector Surface Current Estimation
4.6.1 Introduction
The performance of the venor surface current estimation algorithm will h t be ad-
dressed using the knom test radial current data as described in Section 44. T he
simulated fields of radial m t grid. will form the basic test vectors for the algo-
ritbmn. Ideally the test data set should mimic the aetual data with simulations based
on some surface m n t model parameten. However, this poses a problem since the
ectual data is a hetion of the esseessetially unlmoam m n t -e. The only lmoam
quantities ue the cment estimates *om the three drifter tr&. Therefore the ex-
tent of the test data for the algorithm" is Lwed on the current sirnulatiom from the
known driner velocities.
The purpose of the simulat~ona was for so- testing and not for establishing
a robust radar signal simulation. This could be quite complex for a HX system if
the uncartsinties propaffatlng throughout the radar system were calculated, starting
from the raw voitagez measured at the antenna system to the output radar spectra
(Lipa and Barrick [17]. Lipa and Barrik [SZ]). Since a robust radar signal s~mulation
is beyond the scope of thh thesis, for the sake of simplicity uncertainties w e sunu-
iatd in the resultant spectra as an error in the radial current estimate. As already
mentioned in Section 4.4, the radial mmponent of the reference current estimated
from the drifter data wa.s projected along each radar beam ditectioo used in the at-
tual radar datasets to simulate a uniform current fieid far each set. Error estimates
vere computed from lmowing the Gaussian statisti- of the radar current estimate.
The slmvLated radial cment data were then p r o d by the algorithm to crudely
determine its stability for varying algorithm parameters. Fiy, the dgorithm
ws~ subjected to the actual radar data with a subsequent comparison to the actual
drifter wlocities.
46

4.6.2 Selection of Algorithmn Parameters
For a single vector current estimate, radial current speeds tom a polar -tor of radar
data are required, namely a Q-type sector as discused in Senion 2.3.1. Homer,
determining the optimal size of thLn sector by ~peeifying the values of AR and A0
has been ao arbitrary aspect of the procedure at this stage. The emphaek has heen
on algorithmn mrrectness as o ppd to optirmzation. Lo other words, algorithmn
performance in the presence of a uoifouoifm current with added noise is the primary
objective of this exercise. At this stage of the aigorithm development this b the
simplest approach, since we wish to provide a comparative analysis between the radar
and drifter methods with the m ption the drifter estimate is representative of the
current in a Q-type senor in the vicimty of this estimate. However, it is pwsible
that the oceanographic regimes about the drifter estimate aod the polar radar sector
may not be the same. Since the algorithm requires, as input, radar data from a
region wheh may be larger than the w on represented by the drifter estimate, it is
unclear how the effeceets of a non-uniform current may bias the results. In the worst
case scenario it is hoped the current estimate is indicative of the mean current in the
region of interest. The problems associated with a conparison of this nature will be
discused in the following section.
4.6.3 Algorithmn Constraints
To perform a dehitive comparison between the radar and drifter estimates, the
oceanographic regimes ssrociated with both estimates should coincide. This presents
s problem whm the polar sector used by the vector current algorithmn issignificantly
larger than the region in the vicinity of a drifter estimate. To apply the algorithmn
we are assurmog the current b uniform in the sector. However, this may be true for
only a small region about the drifter pwition. Ideally. this region Jhould span same
porton of the drifter t rd. It should aLso be as smd as possible, such that the radar

vector current estimate is not distorted by radial components far removed h m the
position of the driiter estimate.
As the polar sector size is reduced the coneem noted in the abow parapaph do
not exist or are not ss prominent. H m r , there are other issues to consider when
this situation arise and there are a function of the radar's spatial parameters. For
example, the anpllar extent of the seetor, A@, used by the vector m ent algorithm.
will be affected by the number of be- required to span it. This may adversely
dect the tangential component estimate if the ratio of this parameter to the radar
beamwidth is not greater than two. This is apparent fram inspecting the general
formulation equatioas in Section 2.3.1, sioee at leas two b- must span the region
for the solution to he unique Thk &ect ia even more pronounced if the spectral
resolution, and henee the current resalution, is too mme to sense the radial current
change from beam to beam.
The other extreme case is when the region size becorner too large. In thjs csse.
ermrs m estimating the t-ntial current may be due to u nhm current miation
throughout the region. In other words, the point estimate provided by the drifter
data does not adequately repreent the mean evrrent of the region.
For a consistent comparison each radar ertimate should be computed using data
from a constant region size for each radarfdrifter estimate. This will amid any hiss
as a result of variable spatial aver- in a comparison dataset. Far erample, the
radar ediate from a sector of a hundred square !dometers wi!l consist d more
range-azimuth cells, and hence more regression points, than an -timate from a few
square kilometers. The validity of such a comparison holds only if the same c mnt
e*sts owl both regions.
Due to the inherent polar nature of the radar data, the radial point density is
a function of range in that fewer points are aMtilable per unit area cs the range is
increased. This is B result of the beamwidth of a directional antenna, since the are
48

subtended by the b- increases in sbb with range eerulting in a lager radar ceU size
per 6xd range resolution and beamwidth. This would not beapmbi-if the antenna
beamwidth a.as a 'pencil' beam, since it could be atexed in any diretion to obtain
the desired radial current density per unit area. However, with a ked beamwidth
h m an wtual system only one rsdid current speed b estimated fmm each range-
azimuth cell, and this cell glm in size with increasing range. At ranges of 50, 100
and 200 km, for example, a beamwidth of 5' subten*, in azimuth, a distance of 4.4,
8.7 and 17.4 h respectively. If the radial extent of the input data is 5 km and data
h m three adjacent be- are uJed to estimate the current components, the region
sizes corresponding to there ranges and beamwidth will be 66. 130 and 260 square
!dometers respectively. In order to maintam constant-area sectors, an alternative
approach would be to permit the radial width of the cell to proportionately decrease
as the range b increased. However, the shape of the seetars will &bit a range
dependency in that the sectors becomes 'thinner' as the range is increased. As well,
there is still no guarantee that current uniformity will prevail even in this situation
since the region, instead of being modi6ed radially, will be extended in a lsteral sense.
Therefore. when comparing e4timates at different ranges, this aspect of the current
processing scheme will ultimately have to be addreased.
The azimuth extent used by the Maor current algorithmn must take into eonsid-
erstion the receive antenna heamwidth in the direction of the location of the estimate.
For example, an noted in Section 3.2 1, the radar beamwidth of the Cape Race sys
tem is appmxbately 6' when the beam is steered 6W from broadside. Therefore,
when performing an estimate near the azimuthal extremes of the radar coverage zone
care m m he taken to mure that this extent is Large enough to at least cwer two
be- in the estimate sector. Moreom, if the current resolution of the datanets is
coarse, an compared to the mean nurents of the average area, additional care must
be taken such that the h g e in radial current over azimuth is adequate to s e a

the tangentid component. For -pie, the adar data used here were processed to I
yield a current resolution of 8pprmdmately 7 mjs. while the mean m ent owr the
experimental zone k on the order of 20 -1s. Therefore, it may be difficult to mea-
sure the change in the radial current from adjacent be-, aod hence the tangential
current component, if the &age rn lea thao the nominal error of the radial current
messurement. An alternative in this ease is to in- the slmuthal extent of the
region and proceed with the essmnption of current uniformity
In summary, from the above discurnion there is a elear tradeoff betMen the radar
parameters of beamwidth aod current resoittian, an well an the selection of the opt&
ma1 region size to be used for wetor current estimation based on these parameters.
This vdl aLso depend on the variability of the current regimes being monitored and
the magnitudes of the currents. For example the technique may provide better ec
timates when cur-& are large, relative to the current resolution of the system. it
may &o provide reliable mean current etimates in the event of extreme -nt
variability in the estimation region if the size chasen is large enough to 'force' a mean
estimate from the region. AU in dl, m y of tbe paints noted here wiU have to be in-
vestigated further to determine their etfeets on the reliability of the estimate provided
by the aigorithmn.
4.6.4 Simplifications
At t h stage in the algorithm development, it is unclear how the mam algorithmn
parameters, AR and AB, sect the solution. It may vltimately require some 'good-
ness of fit' testing while varying the algorithmn parameters to minimize the solution
in a least squares sense with same optimal set. Since this may detract from the fun-
damental issue of establishing the potential of the algorithm for current estimation,
optimal parameter selection will not be dealt with. Hence, for the 6 of simplicity,
key issues related to the algorithmn fundament& will be a ddrd here. These will
50

simplib the analysis by converting this multi-parameter model to a simpler and more
intuitive form.
To begin this SimpWcation process, consider the general formulation equations
presented in Section 2 3.1 In this formulation can be equated to V, the radial
component in the sector, while Vv becomes K, the tangential component. This can
ea~iiy be shown by an appropriate axis rotation. By setting the radial mmpment
to an arbitrary mnstant it is clear that the tangential component is a function of
the changing radial components me& by the radar. Even though this may be
intmtive. this step re& the key parameter in the tangential estimation process-
the linear change in the radial -t varlation over azimuth. Hence, using this
approach, an inwtigation of the effect of this variable only will he considered. This
simpliFxation allows us to consider the e kt of the variation of azimuth only As the
radial anent is held constant the effect of that variable should be minimized with
respect to the azimuthal variation.
With this simplification, the variability of results is studied using three azimuthal
extents. The range inuement war held constant at the arclength used for the direct
radial component averaging, 4.8 km, while the angular extents arere chosen ar IZ',
2(P and 36" (i.e. an odd integral number of nominal radar beamwidthe.). For each
care the point dens,@ is preserved, in that radial data irom each radar cell contains
only one radial component at any range while the sector area is increasing with range.
In ather wrds, for each angular extent the same -her of radial components d
be used to estimate the vector current at each estimation point. However, this may
not be the optimal approach, as mentioned in the previous seetion, since this does
not preserve area. At range of 50 and ZOO 00, the area ewerage is approximately
50 and 2W square kilometar, respmtively, for the 12- ease. Far the 36" case this
corresponds to are- of 150 and 6W square kilometers. An flustration of the size of
these regions with respect to the radar -rage area is show in Figure 4.2.

4.6.5 Data Reduction
The experimmtal data were processed to yleid a polar grid resolution of 1" and
400 meters along the azimuth and radial directions, respectively. This data was
mbsequently averaged to yield a resultant aid resolution of 4" by 1.2 km. This
step was performed, krtly, to adhere to the nominal rsdar beamwidth (i.e. 4O) and,
secondly, to reduce the variability of the radial current estimate over range. These
stem were alno necessary to reduce the computational load on the software module
which computer the current components. These new values cornpond ta the the
parameters A0 aod AT, respectively, as discussed in Chapter 1. In this ease, however,
A? is the range miution of the new grid (ic. 1.2 km). The eorrsponding values
52

for then beams is therefore 3, 5 and 9 for the lZO, 20° and 3V c-, respectively.
The number of deetiw rsnge elk, m, is 4.
4.6.6 Processing
The output hom the data reduction step was processed for ortor nureots using the
three region sizes dehed in the prwious mtion. The central position of tbs region
coincided with the position where a drifter estimate computed. Hence, for each
drifter position used in the cornpaikon. 12,213 and 36 input data points were awilable
for the regreion algorithm. Each of the points, which comprised the sectors, also
had to meet the SNR criteria of at least 10 dB to be considered as valid. If this
resalted in any missing data points in a seetor, or 'holes', which did not meet the
SNR criteria, the estimate computd from the sector was not used in the comparison.
A regression analysis arss then applied to the drifter and radar computed current
components in order to study the correlation between the two estimates.

Chapter 5
Results and Discussion
5.1 Introduction
The r dts from the application of the data analysis procedures outlined in Chapter
4 are p-ted and discussed. T he are partitioned as two distinct sections, one for
the direct radial and one for the -tar current comparisons.
The results from the direct radial surface current comparison are presented in the
Fmt section of this chapter. This k the fundamental mmpariwn since it illustrates
the radar's ability to directly monitor the radial component of the surf- current m
the vicmity of a drifter estimate. The results presented are from the analysis of the
three drifter/radar datssets, both simulated and aetuai.
Next, the results from the vector surface current algorithm are presented. How-
ever, it was not deemed n- to provide a cornparkon from all three drifter
datsetr, as was done in the direct radial current comparison, since comparisons
showing both current components, with varying algorithm parameters. muld only
clutter the thesis with redundant graphs and tables. Since an illustration of the p+
tential of the algorithm for vector surface m mt estimation is one of the gods of
this demonstration, data &om only buoy 3 arar used in the cornpariaon study, ss it
provided the most radarfdrifter comparisons.
For ewe of presentation, the comparisons are presented as seattergr- and ar
soclated tables illustrating the statistics of the m mwn. The statistics assodated
54

with the graphs are from a linear reeiin used to summarize the mults of the anal-
pin. Since the drifter and radar current estimates are not strictly ~luident random
variables (See Section 4.2) these statistid tables are only intended to illustrate any
gram inconsistencies in the dysis results.
A brief discussion on the mults from the comparisons, including possible anoma-
lies in the inventigation, will $Jc be presented.
5.2 Direct Radial Surface Current Comparison
5.2.1 Simulations
One set of the results fmm the simulations, in the form of seattergr-, is shown in
Figure 5.1. In these scattergrams the minimum SNR criteria for the simulated radar
data ara~ 30 dB. In other words. only Brag spectral components which met this
criteria were used m the mean radar estimate (See Sections 4 3 and 4.5). A summary
Simdtsd Radial VMaW (-1 - .%dar
Figure 5.1: Scattergrams of simulated radial surf- currents of drifter verm radar
estimates for 30 dB case. (4 , (h) and (e) mrrespond to simulations from data
generated from Buoys I, 2 and 3, respectively.
of the statlstid analysis of the radial component comparisons from the simulations
is shown in Table 5 1.

Table 5.1: Summary of statistical comparison of simuiated current components using
an SNR of 30 dB. The mean and standard dwiationsaf the three datmets are demoted
by - and s, respectively, a d the standard dwiation of the regression of y on r rs
The mmbined rerulu &om the simulations indicate that all measurements fall
within hpprcoiimateiy +6 em/s of each other, which is within the estimated emom of
the buoy and radar me-rements. For the mmbied e m . 83% of the total variation
in the radial components is explained by the regression.
5.2.2 Real Data
Fimres 5.3 and 5.4 display the time series of the buoy radial velocity, tagether with
the radar point estimates and their standard errors. Considering the variation of the
drifter-den& current mmponents over time it is quite remarkable that the radar 30
clmly imitated this trend.
Asampie of theresult*, in the formofscatterpams, illustrating the p erfomof
the radar wing an SNR value of 30 dB, is sh- in Figure 5.2. A statistical summary
of the regression for each individual uw in shown in Table 5.2 for both SNR atudia.
The scatter between radar- and buoy-derived estimates of the radial surface ment
within the estimated error bounds of both twhdquer. This is in claw agreement
with the simulated results, thereby giving credibility to the simulation technique.

Fiere 5.2: Scattergrams of radid surface meats d driller versus radar estimates
mhgreal data; (a) , (b) and (c) mmpond to Buoys 1, 2 a d 3 respectively.
Table 5.2: Summary of statistical mmparisan d radial current mmponenb using real
data with (1) 10 dB SNR and (2) 30 dB SNR. The mean and standard deviations of
the three datasets are denoted by ' and a, respectively, and the standard deviation
of the rrgression of y on s is labelled s

The combined result. ~h nearly dl meagur-ntt fall aithin tppr~xi-tely
46.5 cmfs of each other, which is within the estimated standard deviation of the
buoy and radar meas.uements. The correlation eaadent, for the same, indicates
that 76% of the total variation in the radialcomprmeots is e x p M by the weaion.
4 " " " " '
n p s w z e n a s m n
OoydOm."mem
Figure 5.3: Comparison d ments est~mated from Buoy 3 (did be) and errorbarred
radar data (*)
The dght d81ences in the standard m r and the correlation coeffidents in the
simulated aod anual results can be attributed to the way the radar and drifter sense
the current, as described in Seetion 4.2. It is apparent from the results presented in
58

Table 5.2 that the SNR eases of 10 dB and 30 dB are statistically equivalent. Hence,
an SNR af 10 dB may be &dent for the estimation of the radial currene from
radar data coueeted during day-time operation. Since these dues obtained
Figure 5.4 Comparison of currenh estimated tom Buoy* 1 (dotted line) and 2
(duhed Lbe) and the error-barred radar data (*).
wer a wide variety of oceanic conditions, as is el- tom the wind data appearing in
Figure 5.5, it is apparent that the r dts were independent of the sea state mnditionr
experienced.
Prior invenigationr have provided evidence of discrepancies in radar-derived cur-
59

Figure 5.5. Wind '.eiocitles for position 4 6 14' N and ST IS' W during the experi-
mental period
rest. when wind a d tide are in opposite dirsct~oo. (Lawrenee and Smith (301) and
when cmeet shear, coincident anth wind reversals, occurs (Essen et ol. (341). How-
ever, in this c- there arss no evidence that these moditiorw were present when
the buoys were being interrogated by the satellite. If they did edrt, they were not
observed in the eomparjson. Thus, anomalies ohserved by Lamence and Smith (301
and E m et al. (341 may have resulted b m the fact thu, in the experiments cited,
current meters provided gromdtruthing. The nurent meters m e moored at depths
well below the &/4a law, to which the radar measurement. are sensitive, whde
the drifters used in the wriment related to this thesis rerponded to about the top
metre of ocean. hother factor which might be expected to cause discrepancies in
60

m-ments deliwred by radar and buoys b their temporal sampling merence. A
relatively d l CIT, typically six minutes, b used to obtain a point radar current
messurement. The co-po*ding drlFter estimate, which is obtained fmm spiine fit-
ting and dkrete time merentiating the satellite deduced drifter tr&, may r d t
m a 'smwther' result. This may be more pronounced since apprmdmately 10 paei-
tiom per day were typically recorded kom the ARGOS satellite system. Here again,
however, there a no obvious indication of results being a cted signiheantly by thk
possibility One explanation for this obse-tion wss that the current regime was
uniform for a period of several horn, or many radar data collection inted.

5.3 Vector Surface Currents
Current components estimated bm the track of drifter buoy 3 awe compared to
those estimated from the vector m n t algorithm. Three cases were studied to
illustrate the performanee of the vector current algorithm. Each rase used a constant
angular spsn of 12'. 209, and 36', respectively, while the other two parameters, Ar
and AR, wereset to 1.2 and4.8 km, respectively (Seesections 2.3.1,4.6.4. and 4.6.5).
Simulated and actual data are compared, using both current components, to yield 12
intercomparisons, or sLx for eaeh component.
5.3.1 Simulations
The results, in the form of scattergrams illustrating the performanee of the vector
current algorithm in estimating bath current components using simulated data, are
shown in Figure 5.6.
tions.
Table 5.3 shows a summary of the statistid analysis of the results of the simula-
5.3.1.1 Radial Component
The algorithm produces estimates which are comp-ble wth those obtained fmm
the direct radial component simulations. The remewion erron are of the same order
of magnitude, with dues ranging fmm 4.54 to 6.65 cm/s. In the worst case, 77% of
the totd variation in the radial components is explained by the regression.
The d ts, from the three test caser. tend to suggest the mean radial component
computed from the algorithm is not signiscantly afiffted by the almuthal variation
of the sector. This may be miely due to the small spatial radial current variability
in the regioas selected for praceJing. In other words, the mean radial component
in the @on should not be vey different from the drifter-computed estimate sine
current uniformihr we 'forced' into each -or of radial current data, input to the
62

Figure 5.6: Scattergrams of simulated data resulU for vector currents for azimuth
extents of (a) 12' (b) 21P (e) 36"
algorithm.
5.3.1.2 Tangential Component
There is a high degree of correlation mthin each comparison for the three easer
studied, and the statisties are wry similar as well. However, it is observed that an
increase in the azimuthal extent of the setor d t s in ao increase in the correlation
co&eient ar this extent is inered from 29" to 36O. In other words, a better 6t
to the data is obtained in the mamaurn angular extent ease. One -on for this
observation can be attributed to an inoreme in the range of the regenrion variable,
Ae, which leads to a more stable estimate of the current component ar the range of

Table 5.3: Summary of statistical comparison of radial and tangential current mmponents
computed h m the simulations. The mean and standard deviations of the
three datasets are denoted by - and m, respectiwly, and the standard deviation of the
regression of y on 2 is labelled 8. Caes (1),(2) and (3) correspond to azimuthal m-
tents of 12'. 20' and 36' respect~veiy. TU table summarizes the statistics associated
with the scattergrams of Figure 5.6
the variable increases (Montgomery and Peek 1531). An analow here is the differme
in the slope ermr of a line estimated from a smd number of 'closely spazed. points
as oppmed to s large number which are 'widely spaced'. L the ammt we, 92% of
the total variation seen in the tangential components is explained by the regression.
5.3.2 Real Data
Figure 5.7 displays the results of the paformanee of the algorithm, using real radar
data, in the form of scattergr-. The graphs illmtrate that the emr in the tangen-
tial component estimate may be dependent on the angular extent of the sector. This
was also observed in the simulated data examples. However, the variability of the
scatter about the unity slope line for the tangential component estimates are quite
different from the simulated examples.

Figure 5.7: Scattergrams of -tor merit estimates using real radar and drifter data
for azimuth extents of (a) lZO (b) 2LP (c) 36O. Note the reduction in the standard
error of the tangentid component as the azimuthel extent is increased.
A statistical su-q of the results d the mual data comparison is presented in
Table 5.4.
5.3.2.1 Radial Component
The scattergrams ~ugged that the statisti- from all three cases, as shown in Fig-
ure 5.7, are not si&-tly difkent. This is also evidence3 from the eorrelat~on
mefficients, which range from 0.80 to 0.84, and the standard errom, which range
from 6.94 to 7.22 m /s.
Thestaodarderrorsof thersdialeomponentsestimated hornthe dgorithmarenot
65

Table 5.4: Summary of statistid comparison d radial and tangential w ent mm-
ponents computed fmm the actual radar data. The mean and standard deviations of
the three datasete are denoted by - and o, mpectiwly, and the standard deviation
of the regrension of y on z is labelled s. This table summarizes the statistics of the
scattergrams presented in Figure 5.7
s~gnificantly Uerent from the errors observed m the simulations, with a maximum
mean eeeenee of approximately 1 cm/s. This is a h evident from inspecting the
scattergrams tor both c-. Howeyer, the mrrelation coefficients are Lower and dier,
in the worst ease, by approximately 10 % tom those of the simulations. The poorer
fit, of the real data case, may be attributed to the differences between the radar
and drifter current estimation mothods, as expleilained in Seetian 4.2, or due to non-
uniform currents. The results indieate that at i& 64 %of the variation in the radial
components is explained by the regression. The mmponding quantihr for the wetor
current simulations is 77 %
The wultn also suggest that the rdid component mimates are not signifieaotiy
&eaed by the change in angular extents. This tends to suggest that the result may be
useful inestimating the mean radial components of sectors of iubitrary but reasonable
66

size. In other words, the algorithm may give reliable radial estimates for a variety of
polar extents and these estimates may provide a measure for the aetual mean radial
current. Howewr, since the aetual mean current was uhm in each seetor, this
mnclusion, bmed on the mdts fmm this small sample set, is only hypothetical.
An error of approximately 7 -1s is observed far the radial component estimate
obtsioed here while the direct radial component estimation error is apprmdmately
6 5 em/s. Even though one may not consider these difference to he statistically
si@-t, the slightly larger error in the wetor algorithm mnlt is probably due to
a radal estimate which is computed using radial data from variable region size, ar
mmpared to the k d re@n of averaging used in the direct comparison method. As
well, the drifter estimate may he aerent h m the estimate mmputed here since the
latter is a function of radial current data which are siightly distamed from the track.
Of note, the radial component estimate from the -tor cmrent method may be
used ss a measure of the quality of the tangential component estimate. For exampie,
one would intuitively expect the estimate fmm this method to be positively mrdated
with the direct radial component measurement when the two methods are applied to
data from the same oceanic d on. Without this inherent correlation, the tangential
estimate computed h m the -tor current technique may be suspect. Inother wards,
the resuit may be due to radial current estimates withio the sector that exhibited
extreme variability or noo-linear hehaviour, with respect to magnitude, over azimuth.
5.3.2.2 Tangential Component
The correlation co&eients, far the tkee case studied, increase along with the w
imuthsl extent of each sector and range from 0.41 to 0.74. Coincident with this
ohtion is a derrease in the staodard error of the tangential component estimate
as the azimuth extent increases. Howwer, the mean d ue of this regression error.
of approximately 18 em/$, is relatively large as mmpared to the standard dwia-
67

tions of the man drier and radar dues. Moreover, since this staodard ermr is
appmximately three timer larger than the mrresponding ermr term obserwd in the
simulations, there is no strong evidenee to suggest that a good positive correlation
exists between the tangential components estimated h m the radar data and the
corresponding drifter estimates. H m , it may he noted that the scatter in the
pbts h m these cases are not ar randomly distributed ar the preceding deduction
mmts. Rom observing Figure 5.7, one notes that the points are ewnly distributed
about the line of unity slope, indicating a positive wmiation, albeit with a large
regression standard error relative to the standard devistions of the simulated radar
and drifter tangential curnnt estimates. There may he a number of ieasoas for this
result. First of all, the Large errors may indicate the current - not uniform over
the setors: a scenario that w not d e d in this thesis. In this case, the resuits
may be unpredictable since the nature of the current regime is unImo-. However,
if this the case one may expmt the radial current ertimater h m this tffhnique
and the direct radial component cornparkon to be urnorrelated. Tbis certainly
not the ease since the erron for both comparisons m e on the order of 6.5 emjs,
which is within the m r bounds of the radar and drier technique. Therefore, a
contradiction is apparent with respfft to this rdt. Semndly, it is plausible that
the mean drifter tangenti Components were not adequately reprenented by the paint
drifter estimate. In other words, the actual tangential component exhibits a greater
variabiihr bath temporally and spatially, than the radial component. This is evident
fmm Table 5.3 where the tangential component standard errors are approximately
twice that of the radial's. Another explanation for the suspect tangentlrrl component
comparison may be due to pcmible system problems which were undetected during
the experimental period. For example, the beamforming accuracy of the system was
nwer thoroughly investigated during the program period. The beamforming sys-

tem, which is a programmable fsture of the radar, 6- the radar it's directional
receiving capabilities. Appropriate pharing of the reaived signal, at each antenna
channel, is required for this purpose in order to achieve the required beamwidth and
directivity Any errors in this arpgt d the pmwsing may be due to malfunctioning
of the beamforming units3, sofkvare problem, or inadequate calibration data. How-
ever, these heamforming ermrs may not be severe enough to mmtaminate the results
of the direct radial component comparison. To understand why this is passibie, en
explanation k provided, as well ar a simulation, to demonstrate the effects of beam
direction errors on the radar's direct radial component estimate. To begin, one may
debe a beam direction error, as a function of the radar radial current estimate and
ermr, to be given by
where
A* = sin-'(V-/K)
V-, = Radial velocity error of radar
V, = Mean rsdial velocity estimate of radar
This ermr may not significantly eLct the direct radial mmponeot mmparison if the
actual radial components do not vary apprrciably over the beamerror range, Aor. For
example, let's consider the ease wheee V- is 3 4 0 emfs, the FFT resolution emr
of the experimental radar data. If V, is the expected maximum radial vel~ity of the
current to be observed by any beam, then an angular be- emr of 3 i6' is obtained
when this velocity is set to 40 m/s, or the maximum radial current o b s d during
t* d thi. upabiliw; and (2) mod ezeamat aM obserwd in the mdar1-r radial current
c~psd3mswhishm~ pcfomedwiodicdlythro~t thew-entalperi Lother-ds.
there was no appnnnr need to check the *st- in the h t pk.
JThistypeofrnaifvM~on~ noteddvdng thewerimentd p e E ~ m , sy.-eualpe~tttt
-only ~erhrmed when obvlolu deworation of aystem psfomma a- whish occvrnd
When more than 1 unit - deffftlffft. Thae lyper of fadm are noted in Appendix A.

the experimental trials. In other wrds, a beam angle error of this magnitude may
not significantly effect the results of the direct radial velocity comparlson since the
'sloppiness' of the beam is not detected in the end result (i.e. the radar estimate is still
within the error limits of the radar and driler current estimate techniques). Further,
this beam enor is a 'best ea~e scenario' since currents are typically on the order of
20 emls in the region where most of the radar estimates &om the experimerimtal data
used here where obtained which is in the area of the Avalon Channel. In other words,
ifment mwtudes are not srneantly larger than the radar current error, these
dimtional errors may be difficult to isolate upon inspection of a radarldrifter radial
component comparison. The masking of these beam errors will be hrrther illustrated
by a simulation. In the simulation the beam mom are esumed to be Gaussian, with
zero mean and having a standard deviation of 6'. This beam direction 'noise' wa4
added to the actual r& I& data by shifting the baing of the radar estimate, by
this emr angle term, to generate s beam-red data set. A direct radial component
comparison wss then performed between the radial components of drifter buoy 3 and
the beam-ermred data. This example utilized the same analysis procedure which was
used in Seetion 4.5. The results are shown in the scattergrams of Figure 5.8. It is
evident. from inspeetion, that the difference in the scatter is not significantly a%eeted
by the beam errors used here. Therefore, for the current regime studied here, it is
quite plausible to obtain a goad comparlson for the direct radial component estimate
even with significant beam direction errors.
The effect of besmbnoing enors on the tangential component estimate may not
give rise to ermrs of the magnitudes which were observed for the radial component.
The errors are compounded by the fact that more than one beam of data is required
to calculate the vector component estimate. Sice the &ect of this type of error on
the estimate from the vector current algorithm is not b orn, asimulation is presented
which illustrates the magnitudes of this kind of error. The algorithm parameters used

Radii vetocify [cws) - Radar
Figure 5 8: Scattergrams of direct redid component comparison using (a) the orienai
radar data and (b) the beam-errored radar data. Note the similar scatter between
the twa set..
in the simulation are equivalent to the parameters used in the case 1 study of Section
5.3. The input data is the simulated beam-errared data which -generated for the
simulation desenption of the previous parwaph. The outputs from the simulat~on
are the radial and tangential component. estimatedfrom the vector current algantbm.
A comparison is shown in Figure 5.9 where the actual data case is compared to
the simulated result. Even though the radial componente are estimated to mtthin
re-n for both eases, the tangential components exhibit sidcaotly more variability.
As illustrated by this example, it is plausible that beam direction emam may be
responsible for the resulte observed in the actud tangential component mmparison.
Another problem that may affet the radarfdriner comparison is that the current
iesolution of both measurement W ques may be too mane for adequate sensing of
theeurrent~ monitored in this experiment. Inother words, thecurrents are di5cult to
m-ue with any preeislon since they are not signibantly Larger than the radar and
71

-40
4 - 2 0 0 20.0
40 -20 0 20 60
~ediai vhty imhl- Radar sirnuratad -&at velw (cWs1- Radar
-60
-1 W
-lw -50 0
SirnulaledTengenfial Veb5ty CWs) - Radar
Flgur 5 9 Scnrrermnmsof Mtor mmponent mmparrson usmg fa) rheorlpnd radar
data and (b) the benmcnorcd radar data Note the mcree I" rhc mulab llry of the
t~lmd~nrlal componeor esrlrcares
drifter current errors. fir example, kom Table 5.2 it is obserwd that the standard
deviations of the radid c m t s e~tMated by the radar and driftem throughout the
experimental period, a. aod nv, respeaively, are appmxhateiy twice that of the
regression error of the comparison, s. Therefore, since the eurmt msgnitude are
not signis-tly larger thao the merit velocity resolution of the radar system or
the drifter error, conclusions regarding the quality of the radar mrect radial current
estimates may be premature. More-, when mupled with the above be^^
issue, it is not quite dear just how well the radar or drifters perform in monitoring
'I2

ocean surface m nts where the current regime consists of evrrents which are not
significantly larger than the s-or errors. It is evident that future erperunents must
be wefvily planned when testing the m ent measuring capabilities of a HF radar
system operating in the gmvnd wave mode4.
5.4 Discussion
The redid components estimated from the vector current algorithm, ming simulated
and actual data, strongly agree with the drifter wmpvted dimaten. The standard
deviations are within the lmown errors of both techniques and the results are well
correlated.
The redid components computed from the direct radar estimate and the -ti-
mate from the vector current algorithm correspond well with the drifter computed
estimate. The standard deviations are within the knom ermis of both techniques
and the resuits are well correlated.
Even though the tangential component estimates were highly correlated using
simulated data, they were not very well correlated using the actual data with, at
best, only 55% of the variability in the mmparison explained by the regression. This
may be due to extreme variability in the current field in the regions of interest (Petrie
and Anderson [48j) or to b e d m errors. The latter pmibiity was discussed in
the previous section. If beamforming problems did exist they were unknown during
the experiment, with the exception of the major hardware problems noted durlng the
later phs~er of the exe~cise. Since beam integrity is critical for the adequate testing
4Sincethemsgutuder aftheanorroftheradar~irmtrdtPngntialNnent mmp0"er.e indicate
thst the -&sent& radar data may bo $wee, the i n q d~ e r may mnder why no other
HF radar datasea - tetpd with the -1 UFM~ pmpminginprh-. m i~radar data bm ~o
othn current modwring wmimentr are avnilable hm the Cape Raee lysmn. The 6mt dataret
war mkld d- the of 1991 to mnp nnd study the raal current pattern. *timaled by
the radar. E m . there uar no aca-uutbing made 8MilabLF br that mdre to p-*e data w
tert the wetm cvnent algorithm. The 0th- datast uas mktd in the fu of 1996. Even though
sesAruthulg driftus - deployed d- tbat penad, the eimuthd remluticm of tLe radar data
..aa roo erne to pmde adequate test data for the algorithm.

of the -tor current algorithm, huther experiments should be performed and this
parameter -fully monitored.

Chapter 6
Conclusions
6.1 Direct Radial Currents
The radial &e current components, estimated kom the drifter tracks of three AST
drifter buon. were compared to those estimated h m the hrder components of
the spectra procd h m data collected at the long-range HF fdty at Cape Race.
NF, duriog the fdl of 1992. Each radar estimate m computed h m the mean of all
the radar cvrrent measurements that arere within a cirde of radins 2.5 !an centred
about the pcsition of each drifter estimate. The standard deviation of the drifter
aod radar estimates were appr-ately 6 and 4 em/s, respectively. This comparison
produced standard deviations between the two which were within the ermr b o d
of bath techniques, or approximately 6.5 m/s using a linear regression analysis.
This m alna demonstrated with simulations, albeit with a slightly smaller error of
approximately 6 em/=.
The foil-g points are also noteworthy kom the andysk
. A lo dB SNR far the Bragg companenU of the Doppler spectra han been shown
to be su5tient for day-time radar operation;
. The standard ermr of the radar estimate does not exhibit any dependency on
range or azimuth; and
. The -Its were shown to be independent of the sea date conditions witnessed

during the trial period, when 4nd speeds from 10 - 90 km/hour were experi-
enced.
6.2 Vector Currents
C-ts were assumed to be miform over three polar extents, eaeh spaaoed in
the radial direction by 4.8 Lrm and in azimuth by 12'. 20" and 36'. respectidy.
The radar mmponents estimated h m all three eases were compared to the drifter
estimated components from one AST buoy. Using simulated and actual radar data
for these cases, the results demonstrate the algorithm can estimate the radial went
components to within 7 cm/s, or approximately within one standard deviation of
both W gues The result for the radial component case may be independent of
the azimuthal extent since the correlation coefficients were not significantly different
for all t he c-. However, this may &o be due to mean radial components in the
regions that exhibited tittle relative variability.
The tangential component errors for the simulations were well behaved, exhibiting
a nominal standard error of apprmd-tely 6 emls. Howem, for the actual data the
errors were too large to consider the results usefui br the cases examined. ThL may
be due to non-uniform currents, non-stationary currents wer the beam switching
interval, as mentioned in Section 4.3, or possible sensor resolution and beamfarming
problems which were discussed in Section 5.3. The standard deviations kom the
actual data results indicate the tangential component m r to he apprmtimately 2 5
times the rdal component m r, or appmximately 18 emls. However, since the
simuhtions produced a more rwonbble error of appmxhately 7 m/s, theadltional
error may be due to the above mentioned anomalies and their sources, which were
discwsed in Section 5.3.2.2. These will have to be addressed in future studies using
a more eontmlled experiment.
Though convincing results using actual data m e not obtained for the radar tan-
76

gential component estimate, wen with the observed errors the technique may stili be
useiul for the estimation of the mean tangential component over large regions. The
results for the cmes considered produced a large relative error, but it is apparent from
the scattergrams and the correlation eoaeients that the statistical results indicate
a better 6t as the azimuthal extent is inere&. Better results may certainly he
obtained if the outstanding issues mentioned in the abwe parwaph were resolved
Horn, the validity of this estimate may ultimately depend on the current variabil-
ity in the reeon, which hap not been eonsidered here. If the W que ean pmvide
a mean current estimate capability for Large oceanic sectors, it is still quite a usefui
result since this parameter can't he estimated, in real-time, by other systems.
ID condwion, heeavse of the following unlm-:
Degree of current uniformity about drifter estimate;
. Possible non-tionary currents within beam switching intervals; and
Possible beamforming prohi-
the performance of the vector current algorithm has not been rigorously established.
Horn, the slmulatioos suggest this parameter can be easily estimated to within
one standard deviation of the drifter and radar velocity distributions under uniform
current conditions.
6.3 Discussion
Since the -tor current algorithm sense. the radial current change over azimuth
to estimate the other orthogonal current component, the extent of this azimuth in
relation to the radar beamwidth is an important parameter. Obviously, this extent
should resect the extent of the uniform nurent won. Ciearly then, as this extent
decreases, the radar pedormance is dependent on the 'narrmess' of the receiving
77

eam for spatial selectivity. Optimally, each beam synthesized by the radar should
isaiate a differrot mdu scattering area, or 'pi-piece' from the radar coverage zone
such that the number of adjaeent be-, and hence the region size, is kept to a
midmum If the extent is too large, the risk of using current data from a non-uniform
region increases.
The resolving paruer of any radar syaem is dependent on its ability to distinguish
between multiple targets within a radar scattering cell. Howew, for currents there
is only one fundamental target, that being ocean .naves that are matched to the
radar carrier via the Bragg phenomena. The mean current is indirectly seared
by thir mechanism for each radar eeU. Theretore, the cell size is a measure of the
spatial resolution of a current meemring HF radar. Since the sector used for wetor
went stimation consists of anumber of these cells, from a least squares perspective,
as this number increases better estimates of the mean current may be expected.
Further, the size of tlus number is a function of the radar system parameters Jueh as
antenna hemwidth and radar bandwidth, which de6ne the are and radial extents,
respectively, of the cell. Therefmore, in the design phase of a current measuiog radar
system, these parameters must he carefully selected, based on the knowledge of the
current @mes which are to be monitored.
For the estimation of wetor currents at long rangen it b n-sary to synthesize
very narrow beams. This may not be practical at eoartal sites because at HF they
requM antenna arrays which are hundreds of meters in length. Hence, for a practicd
system, if the beamwidth is not adequate to estimate 6ne scale current measurements
at long ranges it may be necmary to provide a second station to sense the current
6dd from another direction. Howewr, if a 6naresolutian onestation system, which
overcomes the aperture Limitations of a linear array, mn be developed, the method
presented here may provide a framework for a praetieal real-time w ent monitoring
system.
78

One of the goals of this thesis has been to fiubtt~te the potential of using a sing&
station radar to estimate both radial and tangential current mmponents. Thk is
certainly pcssihle as the simulations have demomrated. However, further studies will
need to he attempted, with more extensive simulations and gmvndtruthing exerto
validate the technique with actual data.
-
Even though the technique developed here may not dupiieate the accuracy ab tained from a *site system it may stii be sficient for search and rescue purposes,
or for oil spill mntinpncy planning where surface transport may be required
as opposed to heresolution point measurements.
6.4 Future Developments
The most important radar parameter for the estimation of ocean currents is the
Doppler frequency resolution of the system, since this dictates the resolving p ow of
the radar with respect to current magnitude dewtion. To improve on thk resolution
the examination of modem speetral estimation techniques may have to be considered
(Hickey et ol. [54]). These techniques may be used to obtain a better estimate of the
displaced Bragg frequencies and hence the radial current. Such techniques usually
require fewer samples than the FFT p- speetd estimate, thereby allowing a scan
of the radar coverage area to he performed in a mare timely manner than presently
possible. For example, a full scan of the mwage area requires appmdmately one
hour d data mumtian to yield the velocity resolution pmvided by the experimental
data presented here. If only 32 time samples for e d beam were required, ar oppd to 512 to obtain aspect& estimate, the same zone muld be scanned in approximately
10 minutes. This is very advantageous, especially in the area of search and r-e or
oil4piU traddng, where timely and regvlar surfwe current updates are required. It is
also important for singlestation wetor current estimation since the sea is more Likely
to remain stationary from beam to beam, with respeet to its statistical properties,
79

ar tbis scan time is reduced.
Other algorithms may be evaluated for total current wetors, such as those using
the equation of mntinuity (Rid and Leise [55]). In addition, these may be incor-
porated into the present algorithms mnstraints or as additional variables in a Least
~qn- ~pproach to the solution.
Aoother approach to the solut'ion could be attempted by wing an iterati~ tech-
nique. It has been shown that the radial mmponent can be calculated directly from
the radar data. Homww, by varying the tangentid mmponent in the formulatiom
developed in Chapter 2, a tangential estimate which minimizen the error, with respect
to the known constraints on the radial component, may be computed. This mdd
lead to another estite of the tangential component. In the present formulation the
re@m approach reduce8 the m eX error, thus neglecting the mean knom radial
component.
The quality d the measured ocean surface parameters har to he of prime impor-
tance. A level of mnhdence has to he assigned to each estimate such that a user
can quantify the error in the result. Statistical tdmiques must be incorporated in
the radar measvrements such that the user has some degree of certainty in the men-
sured quantity The uncertainties should propagate throughout the analysis, from the
initial ante- voltage. to the resultant meat map. Since the parameters to be
estimated are random variables with known statisti-, techniques which are available
to q-tify these errors should he used (Lipa and Barriek (171).
Future experiments must eosure that beam integrity is preserved, such that errors
incurred here are within some angular tolerance I d . This Level should he a -ion
of the error in the spectral estimator used to calculate the Bragg frequencies. This is
e p d y important if her resolution radial current measurements are made, since
the mrrelatlon betmen adjacent beam directions should be kept to a minimum to
enhance spatial selectiv~ty.
80

Future work dl also deal with the fine tuoiog of the vector current tehique
presented here, via modem statistical proeed-. The simple analysis procedure used
here may be adequate for demonstration pup- but a more robust solution must
be devised and areas to be examined may include, but are not limited to, residual
dysis, detetion and treatment of outlien, r&ed model fitting, aod wdighted I&
squares solutions (Montgomery and Peck [53]).
The subject of the vonitity of the went field has not been addrd here. For
example, 'can eddy uurents be extracted u se the method or some variation of the
method presented here?" In other words, W the fundamentals of the algorithm
pemut the estimation of a non-uniiorm field where the rdal components are con-
stant but 4, aam many beams?" A minimum eapabilihr, with mpect to radar
perfonnanee, would be the identification of the location of tbene current features.
More extenswe simulations should be used to study the effect of system parame-
ters. For example, the fundamental wauefonn used by the radar should be properly
simulated such that variations in the Brag hqu~cy from cell to cell may be better
mddstood In this study the Brag frequency wa4 assumed to be the centre &eqvency
of the transmispim band. However, this aversimpl&es what may be actually
occurring since the radar bandwidth - actually 375 liHz. Beam pattern simulations
should aLso be incorporated so a~ to study the &st d side lobes, which may eon-
taminate the data when the predominirnt wind direction coincides with one of these
lobs.
To test the algorithm far the estiaation of the mean current field ao experiment
should be devised to deploy m y closely-spaced drifters, such that the actual mean
current can be estimated. A positive r 4t in this case may still be useful for the
estimation of temp& mesc-s& current variations.

References
[ll D. Barrick "Remote sensing of sea state by radar" in Remote Sming of the
%osph& (V. Den, ed.), ch. 12, pp. 146, ~assidptopto, DC: U.S. Gowment
Printing Office, 1972.
fl 3. Walsh and R. DonneUy "A consolidated approach to two-body electromagnetic
scattering problems! Phys. Revim A, MI. 36, no. 9, pp. 4474485, 1987.
[4] K. Hickey and R. Khan "Results of the 1991 CODAR surfaee current experiment,"
Contract Report'92-CS. Department of Supply and Service, 1992.
[SI R. KoweU and J. Walsh "Measurement of ocean wavespectra wing m o w beam
HF radar," IEEE J. 0honic Eng., vol. 18, no. 3, pp. 296305, 1993.
161 E. Gi aad 3. WAh, "Fxtrsnionofoeean wave parameters from HF backscatter
reeeived by a four-dement may: Analysis and application," IEEE J Oeeonic
Eng., vol. 17, no. 4, pp. 376386, 1992.
[7] D. Barrick! 3. Headrick, R Bogie, and D. Crombie, "Sea badecatter at HF:
Interpretat3on and utilization of the echo," Pmceedings IEEE, vol. 62, pp. 673-
680, June 1974.
181 R. St- and J. JOY, "HF radio measurements of surface currents,'. D ~ ~ ~ . S
Rwcoich, mi. 21, pp. 1039-1049, 1974.
191 D. Hodgiar "Oceanographic lofarmation of the eastern canadian oBhore " Data
Repon 82-;, Seamdt Marine Research, St. Jobds, Newfoundland, 1984.
[lo] R. Beardsiey, J. Smtt, and W. Boimurt, "CMICE 76: A current met= inter-
cornparism experiment conducted off Long Island in February-March 1976,'
Techoical Report 62, Woods Hole Oceanographic Institution, 1977
[Ill D. Hdpern. Air Sea Intemctian New Yark: Plenum Pub. Co., 1980. M. Kline
(ed.).
1131 P E Srmrh. 'An lntermmparlson of nm-rurface meawrvmenri with ~rucleraa
and AMF VACM m n r mew- In strong rrdal currents Data Rcporr 82-
3, Departmeor of Fisherrs and Oeeans Onnn Sc~cna add Suivers Arlanrc.
Redford lrur~rure of Ocramgrophy, 19~i

[I4 D. Diemand, E. Faemer, and J. Barrie. "Ocean drifter studies of svrface cur-
rents along the coasts of Newfoundland and Labrador and m the Beaufort Sea"
Data Report 823. Centre for Cold 0- Resource Engineering, St. ~oho.;,
Newiouodland, 1982.
1151 B. Petrie and A. benor. "An analysis of satellite-tracked drifter obsmtions
collected in the grand banks @on " Tech. Rep. 39. Canad~an Technical Report
of Hydrography and O m ~cienc&, 1984.
[I61 D. Crombie, "Doppler spectrum of se. echo 3 13.56 Me./s.." Nature, vol. 175,
pp. 681-682, 1955.
(171 B. Lipa and D Benick. 'kt-squares methods for the extraction of surface
-nts from CODAR nansed-loop data: Application at ARSLOE,'. IEEE J.
Oceonzc Eng., wl. 8, no. 4, pp. 226253, 1983.
[I81 D. Prandle and D. Ryder, "Measurement of surface currents in Liwpool Bay
by High-Frequency radar." Natm. ~01. 315. pp. 128-131, 1985.
[I91 M. Janapaul J. P. Broehe Maiatre H. Esren C. Bianchet G. Grau and E. Mittelstaedt,
c~domparisn oi me-hents of ;=a ments b HF rahar and conventlond
means." Int. J. of Remote Sewing, vol. 3, pp. 4OW22, 1982.
[20l R. P. Diiore 'Ice and its drift into the North Atlantic Ocean " in 1.C.N.A.F
Spec. Publ. No. '8 Synp. on Enmmnmmtal conditions in the ~orihwest Atlantic,
(Dartmouth. N.S.). 1972.
1211 J. Wait "Concerning the theory of scatter of HF radio ground waver " in Elmtmmo&ctic
Pmbing in Geophy~lu (J. Wait, ed.), Boulder, colordo: Golem
Pm. 1971.
1221 J. Wait. "Theory of BF ground wave backscatter from seawaves," J Geophys.
Res., "01. 71, no. 20, pp. 4839-4842, 1966.
[23] W. H Peake, "Theory of radar &urn fmm terrain," in 1959 IRE Int. Con".
Rec., pp. 27-41. 1959.
[24] S. &ce. Reflection of elechomopehe waves fmm slightly mugh surfaces. New
York: New Y0rL:Interseience. 1950.
[251 D. Crombie, "Backscatter of HF radio waves fmm the *a," Elmtmmapetzc
Pm6mng m Gmphysiu, pp. 131-162, 1971. J.R Wait (ed.)
[261 D. Crombie. "Resonant backscatter fmm the sea and its application to physical
oceanography" in Oeeow '72 Conf k.. pp. 173-179, DEEE Publ. No. 72CHO
6 6 OCC, ~ 1972.
I271 J. Wu, 'Wind-induced drift CUIIUL~S," J. Fluid Mcd., wl.68, pp. 4970, 1975.
[28] W. H. Munk 6. Miller F. Snodgrm and N. Barber "Direnional recording of
swell .+om dikant stor&; ~hil. lio&. ROY. sot.. ~ i 255, . no. 1062, 1963.
1291 J. Holbrook and A. Risch. "A comparison of near-surface CODAR and VACM
measurements in the Strait of Juan de Fuca," J. Geqhys. Res., vol. 86,
pp. 10908-10912, 1981.

[30] D. hmence and P Smith, "Evaluation of HF gmund-wave radar on the east
coast of Caoada," IEEE J Oceanic Eng., w1. 11, no. 2, pp. 24G250, 1986.
1311 D. Hod@ns. "Evaluation of the coastal ocean dynamia radar for surface current
me.mrement," contract report, Seaconsult Marine Research, Vancouver, B.C.,
1989.
1321 P. G. Collar.. D. Eeeles. M. J. Howarth. and N. Millard. "An intercomoarison
of HF radar obsemtions of d ace currents with moored current meter data
and dispiaeement rates of amustieally tracked droqued Boats," in Fourth Ocean
Data Caf-ce, Soczefy for Uvdmter Technology, (London, Eng.,), 1985.
[331 J. F.Veno, G. T. Mardell. J. Howmth, andC. 6. Holm- "Current measurement
by long lane HF ground-wave radar," in Pmc. of I~AARSS '88 Sympaaium
(Edinbmgh. Scotiand). 1988.
[34] H. ken, E. MitteIstaedt, and F. Sehirmer, "On near+hore mrfaee current mea-
surement by means of radar," OCch. Hydmg. Z., "01. 34, pp. 1-14, 1981.
PSI D. Prandle, "Messwing m nts at the sea surface by HF radar (OSCR)," J. of
the Soc. fm Undmuoter T&nolagy, wl. 11, pp. 24-27, 1985.
[36] M. Heron. F. Dexter, and B. Mcgann, 'Tarameters of the air-sea interface by
high-frequency ground-wave HF Doppler radar." Aurt. J. Mar Frrshur Re*.,
wl 36, pp. 655-670, 1985.
P'/l K. Butt, P. Jeans, and K. Hi*, "HF radar technology transfer project- imple
mentation and testing of a *site capability," Contract Report 8bl1, Centre
for Cold Ocean Resources Engineering, St. John's, Newfoundiand. 1983.
1381 K. Butt and K. Hickey, "HF radar technology transfer pmj& single site data
a&v~s;,C;~~;t$~~~ ,a9;: Centre for Cold Ocean Resources Engineering,
P9] J. W&h, B. Dawe, and S. Srivastswi, "Remote sensing of icebere by ground
wave doppler radar," IEEE J Oceonsc Eng.. MI. OE. no. 11, pp. 27G284. 1986.
1401 J. W&h and S. Srivadava, 'LRough surface propagation and scatter with applications
to ground "8.- remote sensing in an ocean envlranment " in Pme.
AGARD (NATO) 40th EPP Spectolist meettng, Rome, Italy, pp. i3.1-23.15,
1987.
[41] B. Briggs 6. Phillip md D. Shinn The analysis of obsmtions on s p d
receivers df the fading Afradio waves,"'~mc. Roy. Soe. Lando~ wl. 863, pp. 106-
121, 1950.
1421 P. May, "Single station ocean current vector measurement :application of the
spaced antenoa (SA) technique," Geophys. Res. Ldtel-S, w1. 16, no. 9, pp. 999-
1002, 1989.
[43] N. Le~oon. Rodor mntiplea. New York: John Wdq and Sons, 1988.
144) R. Collin, Antmno and Radiowave Pmpogotion. New York: McGraw-Hill, INe.,
1985.
[45] R Khan B. Gamberg, D. Power, J. W&h, B. Dam, W. Pearsoarso, and D. Mian
"Target hetection and traddng with a high frequency ground wave radar," IEEE
1. Ocennie Eng., wl. 19, pp. 54M48, October 1994.

(461 D. Barriek, M. Evans, and B. Weber, "Oceansurface -eats mapped by radar,"
Science. voi. 198, pp. 13S144. 1977. 1
[47] C. Anderson and R. Shaw, The Canadian Atlantic Starms Project (CASP):
ao wemiear," in Pme. of the Intmatirmal Workshop on Wove .Yzndcoshng and
F d i v , (HaMax, NS), pp. 3-13, ESRF, 1987.
1481 B. Petrie and C. Andmon, "Cmulation on the Newfoundland continental shelf"
Atmosphere-Ow% wl. 21, no. 2, pp. 201-226, 1983.
[49] J. Helbig, "On the we of splines for &racting velocities *om drining bvoy
trade," J Atmospheric Ocurnic teehnol., wl. 229, pp. 1617, 1994.
[SO] R. Davis, "Beam forming br forty element Hf gmundwave radar antenna," in
Pmc. NECEC 90, St John's. NF, CA, May 1990.
[Sl] D. Bamiek and J. Snider. 'The statistics of HF sea-& dappler spectra," IEEE
linns. Anteme Pmmot., wl. AP-25, pp. 1S28. 1977
[52] 8. Lipa and D. Barrick, "Extraction of sea state from HF radar sea do: Mathematical
thwry and modeling," Rodio Seenee, wl. 21, no. 1. pp. 81-100, 1986
[53] D. Mant me and E. Peck Intmduetion to Lznw Regroadon annlysm. New
YOI~: J~E wYq and sons, i982.
[54] K. Hickey, R. Khan, and J. Waish, "Parametric estimation of ocean surface
currents with HF radar," IEEE J. Oceanic Eng , wl. 20. April 1995.
[55] A. Rid and J. Leise,n"A note on using continuity to extend HF radar surhace-
current meauremen*, J. Geaphys. Res., wl. 86, pp. 1108%-11090, 1981.

Appendix A
Chronology of the Experiment
DATE
21/10/92
TIME
(LOCAL)
10:W
EVENT
Data mlleetion
be*
Hardwarde problem
Hard- problem
corrected
SoRware problem
#1
Software problem
#2
Comments
nansmitter 'tripped'. Thb
oenmed after I deleted the
data mUenion proms from the queue
and re-submitted b.
The transmitter is fine now, for
no apparent rwon.
Array processing mion of
interpretation oftw ware failed due
to s 'divide by zero' error at line
212 of the main program. Will
the Vax sohe until
the problem an be found.
Raw Data 6le only contains 32
time sales points for the &st
hardware beam. There must be an error
in the data collection software
as a result of recent changes in
thk module to aceornodate this
experiment. Raar data will be backed
UP

DATE (LOCAL) EVENT Comments
shorn 6 2j to 6 95 \$HZ There a&$
rrmficanr amounl of mrcr'ereoce
II /I
System change Operating at 6.95 MHz.
data This may not effect the current
pm-ing capability of the system
but as BJ precautionary measure the
operating equency will be changed
back to 6.25 MHz.
Operaranp, hequcncy chaxgd bxk
to 6 25 \IHz Spectral perrodrlry
ha- d~wppeared
A 'bug' found in data collection
softw8_re. Yesterday's processed
data may be aBected. The perlodid@
of the spectral data noted at 15:OO
today may be caused by th.
Processed data starts at 76 lon
iaptead of 44. This eLns
data 6le CURRAD29692.001 and
... 002. As a result, one of the
'driftem', at a range of 64 !an,
may not have been suneyed.

TIME
DATE (LOCAL) EVENT Comments
I
24/10/92 11 18:00 /I S o h e problem Pro-ed data f~m data
11 mllected at 13:OO and 16.00 may
be cormpt.
Beam number 28 was used today
for the Grst time, which resulted
in an indexing problem that
sects the dkectionality
associated with the data.
28/10/92
30/10/92
14/11/92
0930
10:30
16:00
0940
W55
Power dovn
Power re~to~d
Software problem
Hardware problem
'Floating divide by zero' from VAX
current s ohe praeeaing
data file D31. Data collection and
may prowing software testing mas
performed during the same period. There
may be a problem in using both
1 may processor. simultaneously.
nammitter 'tripped'
Hardware problem Transmitter 'tripped' again.
See problem d 21-Oct-92 as it
occurred under the same circumstances.
In the future, nro
NFSL$EXE:DEALLOCATGAI.LSFTW
to prevent tbis problem fmm I occurring. It has something to do with
tbc pre-pocesur 1 1 properly ~
uurraluirrg itsell llpun sysrem
I _IYYD ~ IYLC I the cc-~mer
I

TIME
- DATE (LOCAL) EVENT Comments
-=--
04/11/92 21:OO Hardware change Center frequency changed to 6.95
MHz far navy frigate trial.
1 06/11/92 /( 09:OO I(
1230
20:30
I I
6.25. from 6.96 MHz. The inspected
radial current velocity data displayed
a variability which did not seem
plausible.
Hardware problem Cleuit breaker of semnd receiver
rack had 'tripped'and this resulted
in 5 missing ch-els of data.
This may have Sected the 16:50
data run today.
System down 11 Power d m in Trewey I
System up
Hardware problem
Po- up again
Chaanel#lO data may be corrupt.
Its effects may not be very dismatie.

Appendix B
NRSL Current Velocity Data
Tapes
Each data tape consists of radial velocity data p med *om time series coilmed
at Cape Race, NF, during the period Oct. 21 to Nw. 20, 1992. The pmcessiog was
performed on a DEC Vax 8530 with VMS V5.1 a d FORTRAN V5.5-98.
B.O.l Tape Format
Eaeh file on the tape k identified by the foiloaring literal:
where
MMMNN - a 5 digit integer indicating the
where
Julia date of the data colletion
MMM - Julian day
NN - Year
PPP - a 3 digit integer indicating
the dataset number for the day.
For example, the 6ie CURRAD31992.W contains data pm-ed from the 4th
data set colleoted an the 31gth day of 1992.

The tape was mated by using the DEC utility BACKUP.
The fallowing exampler will illustrate how the radial surfgee current data is re
tried &om a magnetic tape uaing the VMS operating system.
Example: Retried of all the current data on tape.
$ ALL MUAO:
$ MOU MUAO:/FOR
$ BACKUP MUAO:JAN131992.BCK/LABEL=CUR * - *
Example. Retrkve a subs& of the data OD. tape.
S ALL MUAO:
$ MOU MLTAO:/FOR
S BACKUP MUAO:JANl31991.BCK/LABEL=CUR/SELECT=(
CURRAD31992.003 r - r
B.0.2 File format
Each 6le is of a dirrct oeeess type containing NRANGx2+1 vnformatted records
where NRANG is the number of400 meter ~rncesred range rings. Each record (except
the header) consists of NANG REAL*4 d ue mmponding to NANG degrees of
azimuth. The structure of this 6le is shorn in Egwe B.1.
B.0.3 Reading The File
The following standard FOWRAN code can be used to meens the data.
OPEN(UNIT=LZTN,T(PE='OLD',NAME='CURRAD31992.OO11,ACCESS='DIREC
+FORM='uNF0RivfATPEDD)
READ (LUN,l)DA~OG,BORSIT,RANGE,NANGCNFFT,NFIELDDLBEAR
. .I

T",. NO*
ST-INOEX* WNO
Radr Cunnt Data
ws
m a bin ah
--...-.-- ---*-,,-
-m.-.-
..m-----
..- .e"-.- ----m..-- -*ll-m ----.---- -.,* - a..r-.,
btaNMbY
~~i
Figure B.1: Remrd structure for binary radial current 6 l~.
Eacb record hsr a ca-
pacity for a full 360 degrees of azimuth where the index in the reeord reference the
azimuth

Where
DATLOG - Array of 32x3 bytes representing i"p
logistic informstion such as site
loeation and time of data collection
BORSIT - Receiw antenna boresiteis 'BORSIT degrees
elockarise from Th~e Nonh.(REAL*4)
RANGE - Array of 3 (4-byte) red value
where
RANGE(1) - Distance to middle of k t range
annulus (km)
RANGE(2) - Distance to middle of Lsst range
mulul (h)
RANGE(3) - Rage annulus width (km)
NFET - FFT size used to generate file (INTEGERe4)
NFIELD - Number of fields in data file format.
Far thi. data set NFIELD=2 (one for the radial
velocities estimate md one for the SNR values).
(INTEGER-4)
LEEAR - The last bearing processed. This bearmg
indicates where the processed data begins .
i.e. RAD(i) is from bearing LBEAR, RAD(2) is
tom bearing LEEAR+l. RAD(3) is tom bearing
LEEAR+P, ete.
Each remaining reeord can be accessed as shoaa in the following example
Let
NRANG - Number of range snnulli p r o d
NANG - Number of angular bins (max. of 360)

RAD(360,NRANG) - (Real-4) array of radial velocity valuer
SNR(360,NRANG) - ... and their SNR valuer.
The Wrtran code is a~ follows:
STANG=-LEEAR+IIO
STlNDEx=MAX(l,~(STANG,360))
DO IRANG=l,NRANG
LREC=IRANG+l
READ(LUN 'IREC)(RAD(I,IRANG),I=ST~EX,ST~Ex+NANG)
IREC=TT(ANG+NRANC+I
READ(LUN 'IREC)(SNR(I,IRANG),IsST~DEX,STmEX+NANG)
END DO
Notes:
By the way. NRANG = (RANGE(2)-RANGE(l))/RANCE(3).
The variable BORSIT m y not be correct in the data 6le header. It should be the
value 121. In any ease, it doesn't matter what value is there. The proper borerite
value is incorporated in the data reeord structure.

Appendix C
Catalogue of the NRSL Radial
Surface Current Data
Date
(1992)
Zl/iO
22/10
23/10
24/10
25/10
26/10
27/10
28/10
29/10
30/10
Time
(Lod)
1153
10:lO
14:17
la16
1308
1534
20:47
1031
11:12
15:18
la24
1300
17:44
09:31
10:47
10:40
13:31
0900
1309
09:30
12:W
16:06
Filename
CURRAD.
29492.001
29592.001
29592 002
29692.001
29692.002
29692.W3
29692.004
29792.001
29892.001
29892.002
29992.001
29992.002
29992.003
30092.W1
30092.002
30192.001
30192.002
30292.001
30292 002
30392.001
30392.002
30392.003
Starting
range
(Lrm)
10.4
50.4
50.4
76.0
76.0
44.4
12.0
52.0
40.0
44.0
40.0
40.0
40.0
40.0
40.0
40.0
40.0
40.0
80.0
400
80.0
40.0
Ending
range
(h)
300.4
200.4
200.4
170.4
170.4
170.4
152.4
173.2
241.2
170.4
161.2
161.2
161.2
161.2
199.6
161.2
199.6
161.2
220.4
151.6
151.6
279.6
Initial
Bearing
'('Rue)
60
96
84
104
104
104
104
104
95
104
104
104
104
104
94
95
95
95
95
77
104
56
Final
Bearing
O(nue)
155
165
165
176
176
I76
176
176
180
180
180
180
180
180
165
176
176
176
176
118
176
176
FFT
size
512
512
512
512
512
512
512
512
512
512
512
512
512
512
512
512
512
512
512
512
512
512

Date Time Filename Starting Ending Initial Final FFT
1 (Igg2) 1 1
1 cmD- 1
;= 1 ;= 1 ?&?,
1 1 size

Date
(1992)
11/11
12/11
13/11
14/11
15/11
16/11
17/11
18/11
19/11
20/11
Time
(Local)
W:OO
la38
12:W
1500
09:OO
1200
18.30
09.00
12:OO
1500
18.00
0900
12:OO
14:43
17:23
09:OO
12:W
1500
18:OO
09:31
12:OO
15:OO
18:OO
09:OO
12:OO
1500
18:OO
06:OO
09:OO
12:OO
15:OO
0933
12:W
1500
0700
10:OO
13:OO
Filename
cuRRnD.
31592.001
31592.002
31592.003
31592.004
31692.001
31692.W2
31692.003
31792.001
31792.002
31792 003
31792.004
31892.001
31892.002
31892.003
31892.004
31992.001
31992.002
31992 003
31992.004
32092.001
32092.002
32092.003
32092.004
32192.001
32192 002
32192.003
32192.004
32292.001
32292.042
32292.W3
32292.W4
32392.001
32392.002
32392.003
32492.001
32492.W2
32492.W3
Starting
range
(Ian)
80.0
103.0
80.0
80.0
80.0
120 0
80.0
80 0
120.0
80.0
80.0
80.0
120.0
10.4
80.0
80.0
120.0
80.0
80.0
80.0
120.0
80.0
80.0
80.0
120.0
80.0
80.0
80.0
120.0
80.0
80.0
80.0
120.0
80.0
80.0
80.0
120.0
Ending
range
(Icm)
239.6
240.4
239.6
239.6
239.6
2M.4
239.6
239.6
260.4
239.6
239.6
239.6
260.4
203.6
239.6
239.6
260.4
239.6
239.6
239.6
260.4
239.6
239.6
239.6
260.4
239.6
239.6
239.6
260.4
239.6
239.6
239.6
260.4
239.6
239.6
239.6
260.4
Initial
Bearing
.(True)
64
104
104
105
105
105
105
105
105
105
105
105
105
57
105
105
105
105
105
105
105
105
105
105
105
105
105
105
105
105
105
105
105
105
105
105
105
Final
Bearing
.(nu=)
180
176
176
179
179
179
179
179
179
179
179
179
179
176
179
179
179
179
179
179
179
179
179
179
179
179
179
179
179
179
179
179
179
179
179
179
179
FFT
size
512
512
512
512
512
512
512
512
512
512
512
512
512
512
512
512
512
512
512
512
512
512
512
512
512
512
512
512
512
512
512
512
512
512
512
512
512

Appendix D
Cross-Reference Table for Radial
Surface Current Data
I I Time I l FFT I
Filename
CURRAD29492.001
CURRAD29592.001
CURRAD29592.002
CUF3%4D29692.001
CUFiFAD29692.002

Time I FFT 1
- Date
01/11/92
02/11/92
03/11/92
04/11/92
05/11/92
06/11/92
07/11/92
08/11/92
09/11/92
10/11/92
11/11/92

Date
12/11/92
13/11/92
14/11/92
15/11/92
16/11/92
17/11/92
18/11/92
19/11/92
ZO/11/92
,
Time
(Local)
09:OO
1200
la30
W:W
12:W
15:W
18:OO
09:OO
12:W
14:43
1E23
W:W
1200
1500
18:OO
0931
1200
1500
18:W
09:W
12:W
15:OO
18:OO
06.00
09:OO
1200
15:OO
09:33
1200
1500
0E00
10:OO
13:OO ,
Filename
CURRAD31692.001
CURRAD31692.002
CURRAD31692.003
CURRAD31792.001
CURRAD31792.002
CURRAD31792.003
CURRAD31792.004
CURRAD31892.001
CURRAD31892.002
CURRAD31892.003
CURRAD31892.004
CURRAD31992.001
CURRAD31992.002
CURRAD31992.003
CURRAD31992.004
CURRAD32092.001
CURRAD32092.002
CURRAD32092.003
CURRAD32092.004
CURRAD32192.001
CURRAD32192.002
CURRAD32192.003
CURRAD32192.004
CllRRAD32292.001
CURRAD32292.002
CURRAD32292.003
CURRAD32292.004
CURRAD32392.001
CURRAD32392.002
CURRAD32392.003
CURRAD32492.001
CURRAD32492.002
CURRAD32492.003 ,
FFT
size
512
512
512
512
512
512
512
512
512
512
512
512
512
512
512
512
512
512
512
512
512
512
512
512
512
512
512
512
512
512
512
512
512 ,