JOURNAL OF GEOPHYSICAL
RESEARCH, VOL. 101, NO. C10, PAGES 22,495-22,512, OCTOBER 15, 1996
Local ocean response to a multiphase westerly wind
burst
1. Dynamic response
W. D. $myth
Collegeof Oceanic and Atmospheric Sciences,Oregon State University, Corvallis
D. Hebert
Graduate School of Oceanography,University of Rhode Island, Narragansett
J. N. Mourn
Collegeof Oceanic and Atmospheric Sciences,Oregon State University, Corvallis
Abstract.
The dynamic responseto a westerly wind burst which occurred
during the Coupled Ocean AtmosphereResponseExperiment in the warm pool
of the equatorial Pacific Ocean is describedusing velocity, hydrography,and
microstructure measurements. Turbulent fluxes distributed momentum input from
the wind over • near-surface l•yer of v•ri•ble thickness. Coriolis •nd pressure
gradient terms combinedto induce a wavelikeresponsewhosefrequencywas close
to the local inertial frequency. Wind stressvariations on near-inertial timescales
interfered both constructivelyand destructivelywith the wave response,exerting
considerable
influence
on the observed currents.
1. Introduction
(TOGA-COARE).Oneoftheprimarygoalsof COARE
was[Webster
andLukas,1992,pp. 1394-1395]"to de-
Westerlywindbursts(WWBs) in the westernequato- scribe and understand the ocean'sresponseto combined
buoyancyand wind stressforcing in the western Pacificwarm pool region." Understandingthe response
to
WWBs is clearly an important aspectof this objective,
rial Pacific Ocean are potentially important in triggering and sustainingE1 Nifio-SouthernOscillationevents.
WWBs also provide the most energeticmechanismfor
maintainingthe mixedlayerin the westernPacificwarm
and it is fortunate that such an event occurred during
pool [LukasandLindstrom,1991].In additionto accel- the intensiveobservationperiod of COARE.
eratingsurfacecurrentstowardthe east,WWBs tend to
In SHM2, we discussthe impact of the WWBs on
generatewestwardcurrentsnear the thermoclinewhich, the thermohaline structure of the upper ocean. This
combinedwith the surface current, create strong verti- paperwill coversomedynamicalaspectsof the ocean's
cal shear at depth [McPhadenet al., 1988, 1992; Del- response
to the WWBs, primarilyvia the evaluationof
croix et al., 1993; Zhang, 1995] and therebygenerate terms in the equationsgoverningzonaland meridional
intense mixing well below the surface. To date, ob-
momentum. A discussionof observations and process-
servations of WWBs
ing methods(section2) is followedby an overviewof
the observations
(section3). In section4, wedescribe
a
have not included microstructure
measurements,so the role of turbulent mixing in mod-
ulating the ocean'sresponseto the surfaceforcinghad near-inertial oscillation which appearsto be a response
to be inferredindirectly[e.g.,McPhadenet al., 1988]. to the wind forcing. Turbulentmomentumfluxesare
In this paper and in a companionpaper [Srnythet al., discussed
in section5, as a preludeto our interpretation
this issue](hereinafterreferredto as SHM2), we de- of the local dynamicsin termsof zonaland meridional
scribe the ocean's responseto the first such event for momentumbudgets(section6). A summaryof these
which detailed observations,including microstructure observationsis given in section7.
data, are available[Mournand Caldwell,1994].
Our observations were made as part of the ocean
componentof the Tropical Ocean-GlobalAtmosphereCoupled Ocean Atmosphere Response Experiment
2. Experimental Details
Paper number 96JC02005.
The observations
discussed
hereweremadefrom R/V
Moana Wave,whichheldstationat 1ø45•S,156øEfrom
December20, 1992, to January 12, 1993. This was the
approximatelocationof a surfacemooring(whichin-
0148-0227/96/96JC-02005509.00
cluded sensorsto measure surface meteorologicalpa-
Copyright 1996 by the American Geophysical Union.
22,495
22,496
SMYTH ET AL.' LOCAL RESPONSE TO A WESTERLY WIND BURST, 1
shipwasdrivensidewaysinto the platform,destroying
conductivity)deployed
to providelonger-term
measure- it and resulting in a dramatic nighttime recoveryon
mentsthroughoutthe COARE intensiveobservation
pe- Christmas Eve.
riod [WellerandAnderson,1996].Duringthe time we
rameters and subsurface currents, temperature, and
remained on station, we stayed within 10 km of this
3. Overview
mooringat all timesto ensurethat we had nearlycoinIn this section, we present an overview of the data
gathered during the cruise, reservingdetailed analysis
CHAMELEON profiler[Mournet al., 1995],deployed and interpretationfor later sections.In cruise-averaged
from the stern of Moana Wave. Profiles to 250 m depth profiles,potential temperature 0, salinity S, and potenwere made every 6-10 min. TemperatureT, salinity $, tial densityer0 were nearly constantoverthe upper 60 m
cident measurements for comparison.
Microstructure
measurements were made from the
potentialdensityer0,Brunt-V•iis/ilafrequencyN, turbulent kineticenergy(TKE) dissipation
rate e, and temperaturevariancedissipationrate X werecomputedusingthesedata. Horizontalcurrents
weremeasured
using
(Figurela). Belowthis level,a layerof strongvertical
gradients(in all three quantities)extendedto ~120 m.
currentprofiler(ADCP), with 8-m pulselengthand 4-
The apparent thickness of this pycnoclineis increased
by the averagingprocess,sincethe layer was displaced
vertically through several tens of meters by internal
wave motions, primarily the semidiurnal tide. Instan-
m bin width. Hourly averagedcurrentsare accurate to
taneously,
pycnocline
thickness
(definedby N 2 > 10-4
a 150-kHz RD InstrumentsshipboardacousticDoppler
within •1 cm s-•, the primary sourceof uncertainty s-2; seeFigure5a) wastypicallyof the orderof 20-
beingthe highverticalshear[e.g.,Lien et al., 1994]. 40 m. Below the pycnocline,salinity remainedroughly
The squaredverticalshear$h2 and gradientRichard- constantwith depth (in the mean),whiletemperature
sonnumber(Ri -
N2/Sh2) werecomputed
from continuedto decrease. In the cruise-averagedprofiles,
hourly averagedprofilesof velocity and density. Vertical derivativeswere approximatedby centereddifferencesbetween4-m depth bins. Measurementsof surface
meteorology(includingeddy-correlation
flux measure-
we seeno indicationof the "barrierlayer" (i.e., a thin
measurements,heat and buoyancyfluxes and the surface stresswere estimated on hourly averagesusing the
to temperature differences,so we expect salt stratification to play an important role in governingthe response
COARE version2 algorithm[Fairall et al., 1996].
to surfaceforcing(SHM2).
layer of stable haline stratification in a regionof negligible thermal stratification, thus a "barrier" to the ver-
tical mixing of heat) found by Lukasand Lindstrom
ments)weremadefrom Moana Waveby C. Fairalland [1991]. However,densitystratificationdue to salinity
G. Youngand kindlyprovidedfor our use.Fromthese differencesis generallyof similar magnitudeto that due
Our preferred method of operation was to steam
slowly into the wind, making as little way over land
as possiblewhile maintaining ship steeringcapability.
This provided unobstructed measurementsof surface
meteorologyfor all of the mast-mountedsensorson the
bow. To stay within 10 km of the surface mooring,
however,we periodically had to make downwindruns.
While this means of deploymenthas been straightforward in many other locations,the strongand intermittent squallsthat dominatedthe surfaceforcing during
the wind burst created considerableship-handling difficulties. The wind signatureof line squallsresultedin
a dramatic increasein ship-basedobservedwind from
The cruise-averagedzonal current had a complex
structurein the upper 200 m (Figure lb). The average current was directed slightly to the left of the local
wind, whichwas dominated during the cruiseby a westerly wind burst. Near the depth of the thermoclinewas
a westward current which may have been the southern
extension of the South Equatorial Current appearing
beneath the shallower wind-driven currents or may have
been part of the dynamic responseto the wind burst
[McPhadenat al., 1988]. Belowthis wasthe southern
edgeof the EquatorialUndercurrent(EUC). Meridional
velocitieswere small in the mean, althoughimportant
dynamically on shorter timescales,as will be argued in
background
(5 m s-•, for example)to morethan 20 section6. The time integral of the 14-m current during
m s-• in a periodof severalminutes. Oncethe peak the cruisewas 700 km to the east, 200 km to the north.
wind had passed,there followedturbulent gustsfrom The integratedzonal current observedhereis about one
all directions. This caused us two serious problems. fifth of the warm-pool displacementobservedduring the
While deploying CHAMELEON from the stern, the 1986-1987E1Nifio event[McPhadenandPicaut,1990].
The cruise-averaged
N 2 profilewasdominated,
asexsquallfront wind sometimesforcedthe ship backward;
(Figin one squall in which instantaneouswind speedsof 34 pected,by strongstratificationin the pycnocline
m s- • wereobserved,
the shipactuallywasforcedback- ure lc). Mean stratificationwas relativelyweak near
(N 2 < 5 x 10-5s-2 for depthz > -50 m)
ward over CHAMELEON, leavingit stuckbeneaththe the surface
ship. Thanks to quick thinking and perseveranceby but increasedsharply in the upper 15 m. Enhanced
somemembersof our group, we were able to recoverit stratification near the surface appears to have been as(1) hawith only minor damage. We had also deployeda sur- sociatedwith the followingtwo mechanisms:
face layer, instrumented platform alongsidethe ship. line stratificationresultingfrom intenserainfalland (2)
However, in a squall with especiallyenergeticturbu- thermal stratification due to the combination of weak
lentgustsin whichthewindincreased
by 20m s-• and windsand surfacewarmingwhichdominatedduringthe
changeddirectionby 90o in a periodof about 15 s, the later stagesof the cruise(comparewith SHM2; Figure
SMYTH ET AL.: LOCAL RESPONSE TO A WESTERLY WIND BURST, 1
22,497
o (øo)
]6
20
24
28
s(psu)
34
0
i
!i
,,
-5o
;:
,,,,
36
,
-0.4
I
v(r/s)
0
0.4 10-s 10-4 10-'3 10-1
i
101
0
(a)',
(b)
,;,
.
:,
.,,,,,
\\
,
'
1
,,
"
,
(c)
,:,\\
':,
"
-
• -100
-100
-150
ß
-150
-200
'I
-200
22 24 26
oo(kg/rn
3)
-0.4
0
0.4 10-s 10-4 10-• 10--910-•10-710
-e
U(m/s)
Sh• (s-2)
Log•o
• (W/kg)
Figure 1. Cruise-averaged
profiles
of (a) potential
temperature
T (dashed
curves),
salinity$
(dotted
curves),
potential
density
•r,(solid
curves),
(b)zonal
velocity
V (solid
curves),
mcridion•l
velocity
U (d•shcd
curves),
(c)squ•rcd
sh½•r
$h• (solid
curves),
Brunt-Vi/is•l•
frequency
N•
(d•shcd
curves),
(d)Richardson
number
Ri (d•shcd
curves),
•nddissipation
rate• (solid
curves).
Eachpairofcurves
represents
the95%bootstrap
confidence
limits[E/ronandTibshirani,
1993]
onthemeanprofile.($h2 andRi werecomputed
using
hourlyaveraged
data,thenaveraged
over
the cruise.)
2). Themeansquare
verticalshearSh• waslargest
in December20 to January4 (inclusive)as "the" wind
the upper thermocline,but stratificationin that region burst and to the shorter windy periodswithin that inwas sufficient to make the mean gradient Richardson terval as "phases1, 2, and 3" of the wind burst. The
number(Ri = N2/Sh•) large(Figureld). Associatedthree intervalsshownby solidbandsat the top of Figwith the increase in mean Richardson
number was an
order-of-magnitudedecreasein TKE dissipationrate e
ure 2a will be used in later calculations to represent
phases1, 2, and 3. The gray-shaded
bandsindicatein-
acrossthe thermocline. Above the thermocline, dissipa- terveningperiodsof relativecalm. The periodof milder
tion rateswereof the orderof 10-7 W kg-•, increasing weatherfollowingJanuary4 will be calledthe "recovery
gradually
towardthesurface.
Above-o15m, eincreasedphase." As is shownby the arrowson Figure2a, the
windswerepredominantlyfrom the west,thougha sigmore sharply.
Wind stress was dominated by a period of intense nificantnortherly componentappearedtowardthe end
winds,whichlastedroughlyfromDecember20 to Jan- of phase 3.
Significant
rainfall(Figure2b) occurredduringthe
uary4 (Figure2a). Windrecords
fromthe nearbyImproved
Meteorological
Instrument
(IMET) buoy[Weller followingfour intervals:(1) throughoutthe first phase
ofphase
2, (3)
and Andersen,1996, Figure 6] showthat windswere ofthewindburst,(2) nearthebeginning
between
phases
2
and
3,
and
(4)
at
the
beginning
of
the
weakfor severaldaysprior to our arrivalon December
recovery
phase.
Most
precipitation
was
concentrated
in
20, althougha moderatewind episodeoccurreddurintense
squalls
of
15-30
min
duration
(as
described
in
ing December
13-16. Strongwindsappeared
in three
2). Duringthe firsthalfof the windburst,the
distinct intervals,each lasting between3 and 5 days. section
surfaceheatflux waspredominantly
upward(out of the
We will refer to the wind event which extended from
22,498
SMYTH ET AL.: LOCAL RESPONSE TO A WESTERLY WIND BURST, I
December 1992
20
0.4
(a)
22
.......
24
26
::::::::::::::::::::::::::::::::::::
...........
January 1993
28
30
2
:..............
:-:.:.:.:-•...........
:---:-:
...............
:<.
4
6
8
10
12
,:.h.:.•-•.
:-:--.•.-::::½r::½•:•:.-.
.,, • •'-"---;
................
..•
.........
;.....
,• ,,•,• • , .'
0.3
•0.1
0
i
i
i
i
J
i
i
(b)
i
(c)
500 ....
o
i
i
j
' ....
........
(d)30
[
[
i
' ....
' ....
[ ....
] , , ,
....
• -500
-1000
i
•.•:•.•,...:....._.•.•
..
•L_
•'•;•½'•"*•*•½•4
.••.......
ß •::
•::,
......
29 a:•a•:,•..
-•'•'•
........
• •a•:;:½:':'-':•-•"½•,•.
':.•5'
•:.':"
'•.-'"•
.............
"'.
...............
':....•J•...•..
','••._••_.....
25 •I•
241 ....
355
•"'
/ /
• ....
I
360
365
[ ....
...:"
'''
.... .............••;•-,
a;:
..... .
"• •
[ • , , , ] , ,
370
375
Year Day 1992 (GMT)
•J•u•e 2. (•) Sgu•ed m•kude
o[ t•e wJ•dstress.A•mwsi•dJc•ted•i]• •em•ed stress;
to the right denotese•tward. The intervalsindicated by the solid and gray-shadedbands at the
top are used in the text to approximate ph•es 1, 2, and 3 of the wind burst and the adjacent
periodsof relativecalm. (b) Rainfall(hourlydenotedby circles,daily denotedby shading).(c)
Total surfaceheat flux (hourlyaveragedenotedby gray,daily averagedenotedby solidcurve).
(d) Seasurfacetemperature(solidcurve)and air temperature(dashedcurve). The boundaries
of the shadedregion indicate 2•hour running means.
3b and 3c). The near-surface
halinestructurefeatured
(Figure2c). During the secondhalf of the wind burst, freshpoolswhichwereoften(thoughnot always)co-
ocean),
withdailyaveraged
values
reaching
200W m-2
the daily averagedflux was closeto zero. The surface
flux was predominantly downward in the recovery period. The atmospherewas generallycoolerthan the sea
surface;the averagetemperaturedifferentialwas1.25øC
incident with local rainfall. Below the main halocline,
salinity generallyincreasedto a maximum between 100
and 150 m (in the vicinityof the EUC). Short-period
oscillationsin T, S, and a were due primarily to the
(Figure2d). Seasurfacetemperature(SST) decreasedsemidiurnal tide.
Horizontalcurrents(Figure 4) were dominatedby
duringthe windburst(the net coolingwas~iøC), then
increasedrapidly during the recoveryphase,exhibiting a strong semidiurnal oscillation plus fluctuations on
a strongdiurnal variation. Subsurfaceprocesses
which longer timescales. Zonal accelerationsin the surface
influencethe responseof SST to surfaceforcing are dis- layer generallyfollowedthe wind stress;the surfacecurcussed in detail in SHM2.
rent increasedfrom near zero to speedsin excessof 0.7
Coolingassociatedwith the wind burst was evident m s-1 eastward near the end of the wind burst. Below
downto ~60 m (Figure3a). In the latter stagesof the the surfacelayer (i.e., at 100 m), westwardflow oscilwind burst, the top of the thermocline descendedfrom lated in speedwith a period of 10-12 days. Below that
60 to ~90 m. Followingthe cessationof strong winds, was a layer of eastward flow associatedwith the EUC.
the thermocline shallowed rapidly and restratification The meridional current exhibited little or no direct relawas apparent in the surface layer. Similar behavior tionship with the wind stress. Near the surface,we obwas evidentin the salinity and densityfields(Figures served two periods of northward velocity, each of which
SMYTH ET AL- LOCAL RESPONSE TO A WESTERLY WIND BURST, 1
December1992
20
22
:•:•.•:......:•,!
24
,
26
! , .,
I
January1993
28
,
22,499
t
30
,
I
2
,
,
I
4
•
I
6
,
I
8
•
10
I
•
I
12
•
!
.:.:j..•
.....
.:• -:-..••:•
......... ..........
.• •........
.../.•.• ..•
............
,.,• ..:•......:,,.•
.•:•:.•
......
-.• .•.=•.:...•............
• •.....•........
•..........•] 400
(a)
-50
-100
-150
-2OO
-250
28.5
29
29.5
40
20
(b)
-50
-100
-150
-2O0
-250
-1
-2
......
:•:'"•:::':•'"'
"•'":::'::'m!ii'.•!:•:i"
.........................
0
i•":
.................
•,>.'?'-•'
ß'• l•:i•..::•.:-:•i•i'":•i'•'•:!'"':•'•"•:•':"•i"•
!:•.:•'-:-'•-•.---':•i:':•
:'•..:
.....•,',',.
•::i•
,:,•
..........
.•::.•.::,::.
...........
i.•.-,.•:
:::::':•-::.. -.::4....
(c)
.:i::i.i::.-i:..-.':•:•:•
•:<•:•::.-..:• •:•:..-:•:.-'iii.•iiii.•:
• ' .-'.'-•-•i•i:: .'::-:•..<•::•.>
i:.-':i'i:::::i:'•'.!:-•:.
.<•.!:iii:
,•.•'•:•i:....'i•'•
............
i.i•.<•,':..'•':-.
-•::•...:".•....
..........................
•'•:•
•":':'•""•:•"
"'
•-•-:•'•-•'-"•l
'":'
:•;•'•:"•
.............
':::::'
.........................
•:-:•-•
'ø';f.."•t
::::•'":
•:::':.'
:•:•-:.:
•"
...............
:-":•.
'•.
:•"•..."•
-
.
-oo
25
-150
25
-200
-250
355
360
365
Year Day 1992 (GMT)
370
375
oe(kg/m
3)
21.4
21.5
......
.:
::."::•
'•-•'•:.:....'•!•:•i•i•
21.6
21.7
Figure 3. (a) Temperature.Valuesabove28.5øCare shadedat 0.05øCintervals.'The contour
intervalis 1øC,with an addedcontourat 28.5øC.(b) Salinity. Contourintervalis 0.2 practical
salinityunits(psu). (c) Potentialdensity.Valuesbelow21.7 kg m-3 are shadedat 0.025kg
m-3 intervals.Otherwise,the contourintervalis 0.5 kg m-3. Profilerdata wereaveragedinto
1-hour, 4-m bins before contouring. For reference,the hourly surface heat flux, precipitation,
and buoyancyflux are includedin Figures3a, 3b, and 3c (top), respectively.
22,500
SMYTH ET AL.: LOCAL RESPONSE TO A WESTERLY
December1992
20
22
24
26
WIND BURST, I
January1993
28
30
2
4
6
8
10
12
, I , , I , ! , I , I , ! ' •}0.2
,,,..,••--'•:;•i••
...........
..'•1••••1•_..._
_;.._•
JL_
• _.IL_•,,,,_,,_..
,,...-•
•- ---'-'
- 0
•
'
0
_
'•-•.;.'-%::)......
,.•:•:,
t::'•
•:,-•]-••:e'•
•ß ".:'•:•,1{•
•
,' ½
20
22
24
26
28
30
December 1992
2
4
6
8
-100
N
iii]l
10
12
January1993
-0.8
-0.4
0
0.4
0.8
Figure 4. (a) Zonalcurrent.(b) Meridionalcurrent.Contoursrepresent
-I-10,-I-20,... cm s-•.
Dashedcontoursindicatenegativevalues. Currentdata are from shipboardacousticDoppler
currentprofilersand are averaged
into 1-hour,4-m bins. For reference,
hourlyaveragedzonal
andmeridional
windstresses
areshownin Figures4a and4b (top), respectively.
appeared following a period of strong eastwardflow. ing valueswell in excess
of 10-3s-2 (N = 18 cph)
Analysisof the momentumbudget(seesection6) will (Figure 5). Squaredshearof similar magnitudewas
showthat this accelerationwas the northwardveering seen at the base of the surface layer during the final
of the wind-driven eastward currents due to the Coriolis
daysof the wind burst (January1-6, 70-100 m). The
force. Below ~50 m, the meridional flow exhibited an
~10-day oscillationprominent in Figure 4 was also evoscillatorypattern, with upwardphasepropagation,an identin the shearandin N 2. The diurnalmixingcycle
amplitudeof ~0.2 m s-:t, anda periodsomewhat
in ex- [e.g., Mournet al., 1989; $chudlichand Price, 1992]
cessof 10 days(asis alsoevidentin the zonalcurrent). was evident in the upper 70 m during periodsof strong
This wavelike feature is discussed in detail in the next
wind. The white-coloredregionsnear the surfaceindisection.
A region of intense stratification was evident in the
cateN 2 < 10-•s-2 (N = 0.57cph). In the nocturnal
mixed layer, Ri frequentlydecreased
below1•4 for pethermoclineduringthe recoveryphase,with N 2 reach- riods of severalhours. We alsoobservedregionsof low
SMYTH ET AL- LOCAL RESPONSE TO A WESTERLY WIND BURST, 1
December 1992
20
22
I
I
I
24
I
I
26
I
I
Januaw 1993
28
I
22,501
I
30
I
I
2
I
I
I
4
I
I
6
I
I
8
I
10
I
I
12
I
I
•'•i!•::•i
"' .-".:•:•i•:.
'::•"-'•*•
......•.•:::•'•::;..'..'
.i...':"
'.....'
• :•i•i. :'/.';•:.•..,.,-.•,:.:.'.•
',i....•l/..•.•i!:::
•' •:: ,'q"•:
':-."' ":½•".., '..........;...'..,::•:'.
:,.....•
......
.',::':;'•
.......
-::..-•
.......
-. ::•:'U:
'•.::.
' .......
! 0
(a)
•?'.•5;:-.-:".;!..'i..:;g'!'":.:.:'"':
.......
'....•i'•i•
..;i;;!i:;:.:•.:;•.`.•5i•:•!;•;•:..!•.•{`•:..::::•{:...•:.:•:•::ii•!•!i•!•.:•:•...?•
.......:,ii':':'
':!i'i.............
•::,'•
........
,•;•:•:::;.½::?:•..
':•::::';:::..-;::'-;:.':'.'.-Y,
li':':,i•r:•
• ' •.'"•,?-:i::!
.:-'-.-'•'•:..-'.•
::'"
......
... .......
•-•..
:..•...
...........
•'•.:•.:'.-.
....
:....
.......
:'...•:."...•::•'.::•--•:-:,•::•':::',':'::'"'.,
'"
'. ','.
..........
'.'•:•:•:::.':,..•.•.'":::'
-.•...•."'
.'•..".'.'•
'•ß....
".:.:...:'.'::".'",-..'"
••.•
....
,;"'.
......
i•!.:•-'-::':',.:
.....
•'-:":<',':
:--:'.::':.,::':,:,.,.,:•:..:-.,.•-..,:::,-'"
::?
.•..:•..
......
.e:':
::.5':......:'•
.,::'•-.'..!.,.
......
.........
;.......
'"•?•
.:•:.,..:
.....
..:,:•:•
......
-::,..
-....'
.......
ß
ß
'•.:-:-'..''•.'..
..........
ß
...........
ß
'.....
••:.:.•
'•.....
. ....
..........
'?'&"...'
....
.•
:•'..'
5•":
.......
'.:•.:.:'
....
::
..:"'•
.....
• ...........
,..,
•.,:...,:.
........
ß
..............
,.¾,:.::.:•.:.:.
It-..
.•....
.........
.:.?.:....
::.
.............
.....:.:
....
%......,......,.•
:•.::•.,...•:,.:..,..
.......
....:!::.•....•::,•:
.."':.:•
:..
.:...
....
:...•:,.
:•
......
....
......:.,:::..::.:
.....
.•........
• •
• •
•
'•"• '.....:•:'.•: '....•
';:?.'b:g•
.•5'":•'&'•;:
:•.•:
•:•'•:•
......
•:; "•:'"**•'{ {s
::;• .
'•-;
....
•:•'
.......
:•'•;
' •.'?
'.......
..:•,
,-•."'•.
•.•*•'
,."' :.:
:.'
.....
:''.'• • •".:'
.•.'
.:•..
'....;:}•
,•.';':'•
"•'
..":
'•'
• •:'•
-200
'
,
.......
i'
,
I
'
'
' '
' '
I'"
'"'
' '
'
'
I
....
I
'
'
--
•0
(b)
-50
-lOO
-150
-200
-250
-6
-5
-4
-3
o
(c)
-
-50
-100
-150
_ -200
_
_
355
360
365
370
-250
375
Year Day 1992 (GMT)
l":•:'--':---"•....:•...-::•i:;
-•-•,•....:•
,•: :;:•::-'•..,:..'-:•i:..
:..: t
0
1
2
3
4
5
Figure 5. Time-depthsections
of (a) squared
Brunt-V•iis•ila
frequency,
(b) squared
magnitude
oftheverticalshear,and(c) gradient
Richardson
number
Ri - N2/(U• + V•). Thescalebelow
Figure5b appliesto both Figures5a and 5b. Verticalderivatives
wereapproximated
by 8-m
finitedifferences
averaged
over
4
m
(i.e.,
centered
differences
between
4-m
bins).
Black
contours
indicatethe values10-6 , 10-5 , and 10-4 s-2. Whitecontours
indicatethe values10-•'• and
10-• s-2. The white contoursin Figure5c enclose
regionsin whichRi < 1/4.
22,502
SMYTH ET AL.' LOCAL RESPONSE TO A WESTERLY WIND BURST, 1
December1992
20
22
24
26
January1993
28
30
2
4
6
8
10
12
0.4
, I ., I , I , I , I , , I , I , I , ! , I , t
%
0
...............
•!
..........
:....'".!
':'•
........
'. :- • f' ':'': ,'•':'"'
•:' -...-'. -..., !i..• '• • .•. •"ß' •: I
.:.•:..., ....
.,
.•.:.. • ß •.• . •. '-.: .
.
,•:•
;
. • •. .
. ....
;½½
'"•'•.:½'.:
'•'• .."''.....i:'
...:::.•.•
:•. ...•; •
• -,
?:";
":t,•: :: /'i. • '.' '• ....
:.:.•..'::'•'
• '•';•
;. ,; ,•':• ;'• • •, •.:':'.?
:'.::•':'•)'.•"
..:.•
•:'.'•• ,
5•'i•:1 •"•, •'. ;, • ,,.;;.,•,•
'•'-
.,?"
,:'•*.•'
•,,•.:•,.*•
• :•?.'"•:'
,•"•:.
,,,., •,,., • , •.,•..
ß • •1,
.., ,...,...
.•.,
•!....;,1•.•,•
..,-.•.
, t.1,,""•,",•','
, ,. .t.•....
,• .•..,.
. ,•'•"
,
N
• ,'•,
' t, •,,•
•"•,F.•
',,.;"
",."'•,•.',;'
•I'•'W?•:(,
•'•",';•
,',...
'•.,.•"•
,
.• t.•.•,, , ,,
,..I,.l , • ,)
. .,, ,I.,• -•
, ,'
•-;.,
,
'
,,.t
.. ..,,
, •, •[t'
;
I
'• '
'
I
•.•", ....
I
I
•; '
I
.,
,
I
' F;; "
ß
,
I
,.
-.
•
; ,
I
'• •
•
I
•
•
ß
,,.
,
I
,
I
;
•
.
I
•'",
ß
.......
I ,
I
-•'• •
I •
.,...,,
,
,
.'I,
'
.
,
I
,
"'•r•
,
'
¾••¾•:;
.....•'- •-•"'• ' ",.
•,.
. 0
-250
-
.,;h,'
-
_
)'
...
•-•oo•
•
"
L-150
,
-
....
355
360
365
,
,
-
370
Year
Day
1992
(GMT)
-250
375
•
-9
•!i•'•=
-8
-7
-6
-5
Figure 6. (a) Turbulentkineticenergydissipation
rate e. (b) Temperaturevariancedissipation
rate X. For reference,the wind stressmagnitudeis given(top).
Ri near the base of the surfacelayer during and after 4. Near-Inertial
Oscillations
the wind burst; this was a consequence
of the high shear
at the base of the wind-driven current. Deeper regions
Eriksen[1993]hascomputedthe linearresponse
to an
of low Ri were associated with the shear zones on the
idealized,translating,westerlywind burst, and the re-
upper and lower flanks of the EUC.
Near the surface, both e and X showeda strong diurnal variation whoseamplitude and depth penetration
sults of that calculationprovide a usefulcontextwithin
which to interpret our observations.Eriksenfoundthat
the dynamic responseto a wind burst centeredon the
was clearly modulatedby the wind stress(Figure 6). equatorhastwoparts:(1) a Yoshida
jet and(2) a wake
During periods of strong wind, regionsof particularly consistingof gravity and mixed Rossby-gravity
waves.
strongturbulenceoccurredeacheveningat the baseof The jet tendsto dominatethe response
near the equathe mixed layer as it descended. Turbulence was of- tor, while the wave wake dominates farther to the north
ten, thoughnot always,associatedwith low Richardson and south. (This relationshipis somewhatmorecomnumber(seeFigure5c). The relationshipis evident,for plicated when the wind burst is asymmetricabout the
example,
in theturbulent
layernear70 m, wheree in- equator.) The dominantfrequencyof the wavewakeis
creasedto a maximumon January7, severaldaysafter relativelyinsensitiveto wind burst geometryand time
the diurnal mixing cycleassociatedwith the wind burst dependence
and appearsinsteadas a slowlyvarying
had subsided.
function of latitude closelyrelated to the local Cori-
SMYTH ET AL.' LOCAL RESPONSE TO A WESTERLY WIND BURST, 1
olis frequency.At the latitude of our station (1ø45•S),
IUl (m/s)
0
the inertial period is 16.3 days and the dominant period
of the wave wake is predictedto be 12 days [Eriksen,
1993,Figure7].
Horizontalcurrents(Figure4) exhibiteda clearsemidiurnal signal, as well as oscillationsat other frequencies. The zonal currents divided naturally into three
depth intervals.
1. From the surfaceto a depth of-,,40-70 m, the zonal
current remained closeto its near-surfacevalue, which
was clearly associatedwith the zonal surface stress. A
temporary acceleration appeared in the wake of each
phase of the wind burst. However, the magnitude of
the current responsefollowingeach interval of strong
winds exhibited no obviousrelation to the strength of
the wind. Phase i drove a zonal current of magnitude
0.05
0.1
22,503
IVl (m/s)
0.15
0
0.05
0.1
0.15
-5O
-lOO
-15o
0.3-0.4 m s-i, while phase2 led to a muchsmaller
acceleration. Phase 3 generated a pronouncedacceler-
-2oo
ation to speedsin excessof 0.6 rn s-1. The reasonfor
these discrepancieswill become clear when we examine
the momentum budgets. Meridional currentsexhibited
northwardaccelerations
(in the directionof the Coriolis force)followingthe eastwardacceleration
associated
-25o
with each phaseof the wind burst. The meridional current, like the zonal current, remained similar to its sur- •'igure 8. Depth dependence
of the Fouriermodehavface value in this layer. Unlike the zonal current, the ing frequencyf = 0.0833 cpd for zonal and meridional
meridional flow exhibited no direct relationship with the currents. Shaded curve is amplitude; solid curve is
wind stress.
phase.
2. A second layer extended from ,-,70 to ,-,120 m.
It was distinguished by predominantly westward cur3. The third layer was dominated by the EUC. In
rent whose amplitude was maximum during two periboth the secondand third layers,the meridionalflow
ods, roughly 10 days apart.
was dominatedby an oscillatorypattern with period of
i
i
i
I
I
I
I
I
order 10 daysand ascendingphasepropagation,upon
which was superimposedthe ubiquitous semidiurnal
103
tide.
10
2
computedfor each 4-m depth bin, and the resulting
spectrawereaveragedtogether(Figure7). The dom-
Rotary power spectra for the horizontal currents were
inant peak in the velocity spectra is consistentwith
the near-inertialperiodpredictedby lineartheory(al-
• 101
though the length of the time seriesis insufficientto
allow us to distinguishwith certaintybetweenthe predicted 12-day period and 16.3 days, the local inertial
period). The senseof the near-inertialoscillationwas
anticyclonic,which Eriksen's[1993] model also predicts. There is a striking contrast betweenthe zonal
Figure 7. Averagerotary powerspectrafor horizontal andmeridionalcurrentsat the 12-dayperiod(Figure8).
currentsbetween14 and 210 m. Spectrawerecomputed The zonal current changedsignabruptly at z = -80 m
(with cosine-bell
windowing)
for individual4-m depth and again at z = -180 m. In contrast, the meridional
bins, then averaged.Pairs of curvesindicate 95% boot- oscillationexhibiteda smoothupwardphasepropagastrapconfidence
limits [Elton and Tibshirani,1993]on tion belowthe top of the pycnocline.Thesecontrasting
the depth average.The error bar at top right indicates
the spreadin the 95% uncertaintylimits (4 timesthe behaviorsare also apparent in the time-depth sections
standarddeviation)for a singleestimateof the power shown in Figure 4 and were probably a consequence
spectrum,assuming25 degreesof freedom. Solid curves of the strongverticalshearin the meanzonalflow (see
with
denoteanticyclonic(counterclockwise)
rotation. Dot- Figure1). Upwardphasepropagationis consistent
that
ted curvesdenotecyclonicrotation. The vertical line in- downwardenergypropagationand hencesuggests
dicatesthe 12-dayperiodpredictedby Eriksen's[1993] the wave was driven by surfaceforcing. Thus the freo
I
Frequency(cpcO
linear model.
quency, the senseof the oscillation, and the direction
22,504
SMYTH ET AL.- LOCAL RESPONSE TO A WESTERLY WIND BURST, 1
Year Day 1992
355
360
365
375
370
0.4
ß
,,,I,,t
o
o
_
-
-50
•- -100
-150
-200
_ -250
20
22
24
26
28
30
2
4
December 1992
6
8
12
10
January 1993
•i•ii•ii•i•E-.':
"•"•---":•'?.•.-:;•,•!•i
......
'......
-4
-3
-2
-1
Figure9. Time-depth
section
oftheturbulent
viscosity
K,•. Thedashed
lineindicates
thedepth
of thediurnalmixedlayer[Smythet al.,thisissue].Forreference,
thewindstress
magnitude
is
given(top).
of phasepropagationall suggestthat the near-inertial tain an expressionfor the turbulent viscosityin terms
oscillationwas a local expressionof the large-scalere- of the dissipation rate, namely,
sponse to the wind burst.
5. Turbulent
i
•
I•rn
---1--/•f•h2'
Fluxes
(3)
Near the surface, Km exhibited the same diurnal cycle
Turbulent
fluxes were estimated from measured val-
uesof • and vertical shear usingthe dissipationmethod
[e.g.,Dillon et al., 1989]. The methodis basedon the
I
25O
I
I
I
-3
-2
TKE equationfor stationary,homogeneous
turbulence,
namely,
P + Jb- ½.
(1)
200
-
150
-
100
-
Thefirsttermontheleft-hand
sideof(1)represents
the
rate of TKE productionby turbulent Reynoldsstress,
•' - -p(u/w/, v/wl), interactingwith the meanvertical
shear:
P=-.
•'
0U
p
Oz'
50-
0
and is parameterizedas K.•Sh 2, whereI(.• is the tur-
-6
bulent eddy viscosity. The second term on the left-
handsideof (1) represents
the creation(or, moreusually in a stably stratifiedfluid, destruction)of TKE
via the turbulent buoyancy flux. The flux Richardson
number/•/ representsthe fractionof shearproduction
that is spent increasingpotential energy via negative
buoyancyflux; that is, R• = -Jb/P. A frequentassumptionis that the valueof Ry is constant(the validity of this is questionedby Gatgertand Moum[1995]
and Moum and Gatgert[1995]). In this work, we set
/t/= 0.17 (thisis equivalentto settingthe mixingefficiency,r = J•/½ = /•//(1-/•/),
to 0.2). Using(2)
and the definitionsof K,• and/•, we solve(1) to ob-
-1
LogloKrn
Below DML
Arithmetic
Geometric
Median
Mode
mean
mean
Within DML
4.67 x 10-'3 5.73 x 10-2
6.36 x 10-4 2.88 x 10-2
5.01 x 10-4 3.16 x 10-2
2.19 x 10-4 7.24 x 10-2
Figure 10. Histogramsfor K,• in the layer-70 m <
z < -18 m. The thick (thin) curve corresponds
to
valuesobserved
within (below)the diurnalmixedlayer.
Vertical lines (near top) indicatethe geometricmean
of each subsample. Also given are various measuresof
centraltendencyfor measurements
made below(left)
and within (right) the diurnalmixedlayer(DML).
SMYTH ET AL ß LOCAL RESPONSETO A WESTERLY WIND BURST, 1
Stress
(N/rn
2)
0
-0.15
0
Stress
(N/rn
2)
0.15
-0.15
0
,
Stress
(N/rn
2)
0.15
-0.15
V
V
(c)
(b)
(a)
-20
0.15
0
I
I
•7V
22,505
_
-40
-60
-80
calm
WWB
phase I
-100
i
0
I
I
'
I
'
'
I
I
I
I
i
(f)
(e)
(d)
-20
-40
1
-60
-80
calm
WVVB
calm
phase III
-100
0
.
i
I
i
I
I
,
,
,
,
i
i
i
I
!
i
i
I
(h)
(g)
-20
-40
-60
-80
tire
ntire
recovery
period
-100
i
-0.15
i
I
i
i
i
o
0.15
-0.15
Stress(N/rn2)
I cruise
i
i
0
Stress
(N/rn
2)
0.15
-0.15
,
I
'
'
o
0.15
Stress(N/rn2)
Figure 11. Time-averaged
verticalprofiles
of theturbulent
Reynolds
stress.
Triangles
nearthe
surfacedenotethe surfacewind stress,with the zonal(meridional)components
denotedby the
solid(dotted)curves
andthesolid(open)triangles.
Thespread
between
thetriangles
andthat
betweenthe thin outer curvesindicate95% confidence
limits on the time average:(a) t - days
355,358;(b) t- days358,359.5;(c)t- days359.5,362;(d) t- days362,365;(e) t- days
365,369;(f) t- days369,371;(g)t355,371;(h) t371,378; (i) t 355, 378. WWB is westerlywind burst.
McPhaden
et al. [1988]fitas did e (Figure9). The largestvaluesof Km wereob- m2 s-1. For comparison,
servednear the surfaceon windy nights. The large dissipationrates associatedwith the deepcycleof turbulencedid not generallyinvolvelargeturbulentdiffusivities, becausethe shearin that regionwasintense.The
ted a current profile observedat the equator soon after
a wind burst to a simple, constant Km model for the
equilibrium current generatedby a steadywesterlywind
and therebyobtainedthe estimateKm~ 10-2 m2 s-1.
regimeof largestKm coincided
with the diurnalmixed The assumptionthat Km is constant, however,is clearly
layer,the depthof whichis definedby a densityincrease naive. The histogramsof valuesof Km aboveand below
of 0.01kg m-3 fromthesurface
value(SHM2). The ge- the baseof the diurnalmixedlayer(Figure10) suggest
ometric(arithmetic)
meanofKmin theupper70rn(de- separate physical regimes. Each histogram is roughly
finedin section
4 aslayer1) was1.2x 10-3 (1.5x 10-2) lognormal. Within the diurnal mixed layer, the geomet-
22,506
SMYTH ET AL.: LOCAL RESPONSE TO A WESTERLY WIND BURST, 1
ric meanof K,• is 2.9 x 10-2 m2 S-I; belowthat layer, pendencewhich was closerto linear during Tropic Heat
it is 6.4x 10-4 m2 s-•. (Notethat ourestimates
of the 2 (TI-I2)(Figure 12, triangles). Both of thesesetsof
measurementswere made farther east along the equator, near 140øW; the former were in relatively constant
windy
conditions,and the latter were in light and varito the noiselevel(seeFigure5b; the errorin hourlyavarithmetic mean of K,• are sensitiveto the largest values observed. These values occur when the shear is close
erages
of $h2 is •0 10-6s-2) andtherefore
involvelarge
able winds.
The mean wind was from
the east in each
relative errors. Large valuesof K,• are alsooften found case. In all three cases,the dissipation method yields
in regimeswherethe flow is staticallyunstable,sothat estimates of the time-averagedReynolds stress,which
the assumption
of constantRf in (3) is of questionableappear consistentwith the requirement of continuity of
stress at the surface. The shape of the COARE provalidity.)
The Reynolds stress is parameterized as r = files resemblesthe TI-I2 results more than TH1, despite
pKr•3U/•3z. During all phasesof the cruise,the Rey- the fact that the latter experiment, like ours, was carnoldsstresscomponentswere nearly zero belowthe top ried out in conditionsof strong wind. While the depth
of the pycnocline(Figure 11). Abovethis depth, the dependenceof the COARE stressprofiles is similar to
stresscomponentsincreasedin a mannerthat suggestsa that observedin TH2, magnitudes are closerto TH1.
goodmatch with the surfacestress,althoughour stress Becauseof the extreme depth attenuation of turbulent
estimates
become
highlyvariablein the 18-and22-m stressfound during TH1, it was hypothesizedthat a
depth bins due to the low observedshear. The corre- processother than turbulence was responsiblefor redisspondencebetweenthe wind stressand shallowturbu- tributing momentum vertically over the upper 50-100 m
lent stressesis a test of the accuracy of the estimates in order to match the depth dependenceof the zonal
of K,• discussed
above. The resultssuggestthat while pressuregradient and thus closethe zonal momentum
hourly estimatesmay involvelargeerrors,averagesover budget. Internal gravity waveswere thought to be a
a few days are probablycorrectto within a factor of prime candidate, and, indeed, these have been observed
2. DuringTropicHeat 1 (TI-I1), Dillon et al. [1989] to be especiallyintenseat this location [Mournet al.,
and greatermagfound that the zonal componentof the Reynoldsstress 1992]. The weakerdepthdependence
increasedexponentiallytowardthe surface(Figure 12, nitude of the momentum transport due to turbulence
circles),whereasHebert½tal. [1991]founddepthde- in COARE suggeststhat turbulence is especiallyefficient at vertically transporting momentum in the upper 50 m. Continued analyseswill focus on the role of
internal gravity wavesbelow 50 m during COARE.
Ixj
10--4
I
10-3
, , ,, .... I
t
-50
10-2
........
I
10-1
.....
'
.
6. Analysis of the Momentum
Balance
Further insight into the dynamic response to the
wind burst may be gained via analysis of zonal and
meridional momentumbudgets. These budgetsare constructed as follows. The usual equations for horizontal momentum conservationare integrated betweentwo
depths, z• and z2, to obtain
,:/':'
OU
= fV+ *-x(z2)
,-x(z•)
+R•'
c•t
hpo
hpo
'
(4a)
c3V
_ -fU+ ,-y(z2)
,¾(z•)
t-R•; (4b)
c3t
hpo
hpo
Here U, V is the horizontal current; f is the local Cori-
olis frequency;,%,,-yis the Reynoldsstress;and the
overbar indicates a depth averageover the layer whose
lower and upper boundariesare z• and z2, respectively.
The h = z2- z• is the layerthickness.The background
densityPois giventhe value1023kg m-3. Foreachof
Figure 12. Zonal componentof the Reynoldsstress, (4a) and (4b), the first termon the right-handsiderepaveragedover the durationof the wind burst (shaded resentsthe Coriolis force. The secondterm represents
curve). The width of the shadedcurvespansthe 95%
confidencelimits on the time mean (from hourly val- the turbulent momentumflux at the top of the layer
ues). Circlesand trianglesrepresentzonalstresses
from
Tropic Heat i and 2, respectively.(Valuesfrom the
Tropic Heat experimentshave been multiplied by -1
(the windstressif z2 = 0), whilethe third denotes
the
turbulent momentumflux at the bottom of the layer.
The remainingtermsin the horizontalmomentumequa-
to permitlogscaling.)Symbolsat the surfaceindicate tions(primarily,pressure
gradientforces,verticalfluxes
the zonal wind stress.
dueto internalwaves,andlateraladvection)cannotbe
SMYTH ET AL.- LOCAL RESPONSE TO A WESTERLY WIND BURST, 1
22,507
Layer 1' 0 - 70m
Year Day 1992
355
360
365
375
370
_,•
0.4' '.•........•
'..•'...•.......•
' ..........
.... ....' '•
I
I
I
I
i
i
0.2
2
(a)
-1
1
(b)
-2
(c)
-2
3
(d)
-fu
R
-2
2O
22
24
26
28
December 1992
30
2
4
6
8
January 1993
10
12
Figure13. Themomentum
budget
forthelayer-70 m < z < 0 m (layer1). (a) Cumulative
(i.e.,timeintegrated)
zonal
acceleration
since
thebeginning
ofthestation
(shaded
curve)
and
cumulative
zonalmomentum
inputfromthe surface(dashedcurve). (b) Residual(takento
represent
thezonalpressure
gradient
force)(shaded
curve),
zonalCoriolis
acceleration
(solid
curve),
andcumulative
zonal
acceleration
duetoturbulent
mixing
(dashed
curve).
(c)Same
asFigure13a,butin themeridional
direction.
(d) Same
asFigure
13b,butin themeridional
direction.Low-passfilter cutofffrequencyis 1.5 cpd.
estimated reliably from our data and are instead rep-
locity, namely,
resentedas a residualforceR•, P•, whichis invokedto
account for the observed acceleration.
The residual also
,x(d,v) -
' o(v,
V)dt,
containsany errorswhich are presentin our estimates
where overbarsindicate the vertical averagebetweenz•
of the other forcesappearing in the budget.
in (4a) and(4b). The cumulative
The cumulativemomentumbudgetis obtainedby in- andz2whichappears
in orderto alloweasyvisualassesstegratingin time from the beginningof the observation budgetis employed
contribution
ofeachterm.
period to obtain net changesin verticallyaveragedve- mentof thetime-integrated
22,508
SMYTH ET AL.: LOCAL RESPONSETO A WESTERLYWIND BURST,1
Layer la: 0 - 35m
Year Day 1992
355
360
o.4j,
365
370
...,., ,,, ....
375
. ....
,,,
(a)
(b)
o
•-1
-2
(c) o
<1-1
-2
3
(d)
2
t•
1
<• 0
-1
-2
20
22
24
26
28
December 1992
30
2
4
6
8
10
12
January 1993
Figure 14. Sameas Figure 13, but for -35 m < z < 0 m.
the pressuregradientforce, in which casethe appearanceof the residualat the end of phase1 suggeststhat
the eastwardflow generatedan adversepressuregradient in the near-surface
region. Zhang[1995]has simulated the responseto WWB-like forcingand has noted
facelayerdominatedby wind-drivencurrentsandiden- the developmentof an adversepressuregradientwhose
tified in section4 as layer i (Figure13). Duringthe magnitudeis similarto that of our residual. During
firstphaseof the windburst(t = days355-358),zonal phase2, the actionof the Coriolisforcegreatlyreduced
Fluctuationsdue to the diurnal mixing cycle and the
semidiurnaltide were removedby meansof a low-pass
filter with cutoff frequencyequal to 0.75 cpd.
We evaluate the terms in the momentum equations
for severallayers, beginningwith the 70-m-thicksur-
momentum transferred from the wind accounts almost
the effect of the eastward wind stress. During phases1
identicallyfor the observedacceleration
of the'currents and 2 and in the interveningperiod of calm winds,the
in this layer. At the endof phase1, an immediatede- meridional current exhibited a steady northward accelcelerationof the eastward flow was driven primarily by
eration which was driven by the Coriolis force and op-
the residualforce. FollowingMcPhadenet aL [1988], posedby both the surfacewind stressand the residual.
wehypothesize
that the residualis mainlya measure
of It is the Coriolisturning of this northwardcurrentwhich
SMYTH ET AL.- LOCAL RESPONSE TO A WESTERLY WIND BURST, 1
22,509
Layer 2:80 - 120m
Year Day 1992
-•
355
360
365
370
375
I
•- 0.2
i
i
]
I
....
ß
I
,
,
,
__ •-:':'•:•4.2•
_•
'--
-0.5
....
0.5
I
'
05
;.
i
I
....
I
....
I
....
I
'
'
, , , , I , t , , I , •, [ I , , , , I , ,
o
-0.5
i
,
,
i
i
i
.............
•::!iiii•
•:---.'•ji•i?..•i:ii
i•j..ii;•{i•"-•'"'"'-"'•,
-••••••i
i:iiii
f
i
i
I
[
i
i
i
i
i
i
,
i
0.5
(c)
.•. o
-0.5
1.5
(d)
-fu
-1.5
20
22
24
26
28
December 1992
30
2
4
6
8
10
12
January 1993
Figure 15. Sameas Figure13, but for layer2 (-120 m < z < -80 m) and cutofffrequencyof
0.25 cpd.
accounts for the attenuated responseof the zonal cur- phase 1, the near-surfaceregion was closeto a state of
rent to wind forcingduring phase2. It appears,then, motionlessequilibriumwhen phase3 began(although
that the eastward acceleration which arose due to the
Coriolis turning of the southward meridional current
zonal wind during phase1 generateda responsein the accelerated the zonal flow slightly during days 363Coriolisand pressuregradientterms whichopposedthe 365). The responseto the wind Stressduring phase
wind-driven zonal accelerationduring phase2. In other 3 was therefore not greatly complicatedby interference
words,phase2 arrivedout of phasewith the wavewake effects, and the observed eastward acceleration of the
currents
(~0.2 m s-1 fromday365to day
generatedby phase1, and the resultingeastwardaccel- near-surface
368) closelyparallelsthe momentuminput from the
eration was reduced correspondingly.
Because of the destructive
interference
between the
wind. We can now advance an explanation for the out-
wind stressduring phase 2 and the wave wake from of-proportion responseof the surface currents to the
22,510
SMYTH ET AL.- LOCAL RESPONSE TO A WESTERLY WIND BURST, 1
Layer 3:130 - 210m
y½.arnay 1992
•-
355
360
'----0.2
,
ß
v
,
,
,
I
370
375
' ' ' .... ' .... '"
•
05
(a)•
365
,
,
,
,
I
,
,
,
,
I
,
,
,
,
I
,
,
0
-0.5
0.5
fv
11
.....................
,.,:,:,:,:,,
,,
-0.5
R
r
_F
-'1;210
0.5
(c)
---
0
-0.5
t,,,,I,,,,I,,,,I,,,,I,,
1;130
i................
•..... '•"•
""
'"'"""
'"'"
•,::•
f
/
'
'
'
'
I
'
'
'
'
I
'
'
'
'
I
'
'
'
'
I
'
'
'
3
(d)
2
-fu
1
ß-'1;210
R
-2
-3
20
22
24
26
28
30
2
December 1992
4
6
8
10
12
January 1993
Figure 16. SameasFigure13,but for layer3 (-210 m < z < -130 m) andcutofffrequency
of
0.25 cpd.
separatephasesof the WWB; the responseto phase 1 force arosewhich opposedthe Coriolis force and mainwas truncated by an adversezonal pressuregradient, tained the eastward jet in a state of approximate equiwhile the responseto phase 2 was attenuated by the libriumduringthe remainderof the station.The eastCoriolis force. The eastward accelerationin responseto ward residual force which is evident during days 372phase3 wasalmostperfectlyconsistentwith the amount 375 coincidedwith the appearanceof the long-livedjet
of wind forcing. This eastward accelerationled to a following the wind burst. During days 374 and 375,
strongnorthwardCoriolisforce,which acceleratedthe both the zonal residual and the Coriolis force decreased
meridionalcurrent in oppositionto the northerlywind. as the northward meridional current decelerated. The
As the strongeastwardwind subsided(days369-371), near balance betweenCoriolis and residual forcesis apthe Coriolis force associated with the meridional current
parent in the meridional direction, as well as in the
acted to decelerate the near-surface zonal flow.
Dur-
ing the succeeding
few days, however,a zonal residual
zonal.
As we expect based on the Reynolds stressprofiles
SMYTH ET AL.' LOCAL RESPONSETO A WESTERLY WIND BURST, 1
(Figure 11), the turbulentmomentumflux at z -
22,511
0.25 cpd in order to more clearly display the dynam-
-70 m played a negligiblerole in governingthe mean ics at subdiurnal frequencies. The meridional momencurrentsin layer I (Figure 13). However,it playeda tum budget suggeststhat undercurrent was held in a
crucial role in distributingthe momentuminput at the state of approximate geostrophicequilibrium between
surface, as well as that generated at particular depths the southward pressuregradient force and the northby other forces, throughout the layer. This is illus- ward Coriolis force. The zonal budget showsthat the
trated in Figure 14, which showsthe cumulativemo- undercurrent was acceleratedduring the first half of the
mentum budget for the layer betweenthe surfaceand wind burst, then deceleratedover severaldaysfollowing
z - -35 m. Here the distribution of wind stress, Coriolis, and residualforcesis qualitatively similar to that
seen in the 70-m layer, but the turbulent momentum
flux across the surface z - -35 m now plays a central role. During phase1 of the wind burst, only about
one half of the eastward momentum input at the surface appearedin the current above35 m; the remainder
was transferred to the deeper region by the turbulent
stress. Throughout the remainder of the wind burst
the end of the wind burst, by a combination of Coriolis
and residual forces. Again, turbulent momentum fluxes
are negligible at these depths.
7. Discussion
A three-phase westerly wind burst occurred in the
Western PacificWarm Pool betweenDecember20, 1992,
(days358-370),the zonalturbulentstressactedcontin- and January4, 1993. Nearly 500 mm of rain fell during
uouslyto transfereastwardmomentumto the regionbelow z- -35 m. In the meridional direction, the northward Coriolis-driven acceleration was amplified by the
action of the turbulent
stress. This was because the re-
gionbelowz - -35 m wasunaffectedby the southward
surface stress and therefore respondedmore rapidly to
the northward Coriolis accelerationthan did the region
nearer the surface. Subsequently,northward momentum was transferred from the lower to [he upper layer
via the turbulent
stress.
We turn next to [he momentum budget for the layer
between 80 and 120 m depth. Our purpose is to describe the dynamicsof the westwardjet that is appar-
ent in this layer (seeFigure4). A weakwestwardjet
this period, and SST dropped by 1øC. The upper 50 m
of the water column was well mixed from the outset,
remained so during the wind burst, but quickly restratified when the heavy weather subsided.
Our purpose here has been to learn what we can
about the ocean's response to the wind burst from
the one-dimensionaldata gathered aboard the Moana
Wave. A fuller understandingmust await analysisof
basin-scaledata and model output. Our interpretations
of one-dimensionaldata in terms of horizontal pressure
gradients, for example, are speculative at this point.
Nevertheless, we have found it possible to construct a
hypothetical scenario which is consistentwith both the
one-dimensional
data
and our current
theoretical
un-
(~0.1 m s-•) wasin placeat thebeginning
ofourcruise derstandingof equatorialdynamics,and we presentthis
andhad apparentlybeenpresentsinceleg 1 [Wijesekera scenarioas a guide for future analyses.
The dynamic responseto the wind burst was domiand Gregg,1996].Duringour cruise,the velocityof the
jet variedby ~0.3 m s-• on a timescale
of ~10 days. nated by wavelike motions in the near-inertial band. In
Since these accelerationsare clearly associatedwith the addition, an eastwardsurfacejet with averagespeed40
for at least1 weekfollowing
the cesnear-inertial oscillation discussedabove, we have low- cm s-• persisted
pass filtered the terms in the cumulativemomentum sation of strong winds. Despite the low latitude of the
budgetwith a cutofffrequencyof 0.25 cpd in orderto COARE region(• • 2øS), the Coriolisforceplayeda
more clearly displaythe low-frequencydynamics. The crucial role in the dynamics. Analysis of the momentum
initial accelerationsin this layer are to the west and the
budget for the upper 70 m has shownhow the response
north (Figure 15). Theseaccelerations
are drivenby of the surface currentsto wind forcing is controlledby
the Coriolis force and the meridional residual force, re-
interference
spectively.The meridionalbalanceduring the remainder of the wind burst suggeststhat the westwardjet
is held in approximategeostrophicequilibriumbetween
the oscillatory wake left by previous wind stress. This
interference effect greatly attenuated the responseto
phasesI and 2, but it amplified and prolongedthe responseto phase 3.
Mixing played a crucial role in distributing momentum above the thermocline, as has been shown via the
momentum budget for the upper 35 m. If the balance
the southward Coriolis force and a northward pressure
gradientforce. Both the Coriolisandthe residualterms
oscillate about this equilibrium with a period of 10-12
between the instantaneous
wind stress and
days,but the oscillationin the residualleadsthat in the
(Coriolisversuspressuregradient)beCoriolisforce by 900. The resultingimbalancebetween we hypothesize
the Coriolis force and the residual drives the wavelike low that depth holds,then it alsois not necessary
for inmeridional accelerationsvisible in Figure 4. The zonal ternal wavesto transport significantmomentum, as has
accelerationsassociatedwith the "reversingjets" seen beenfound farther east alongthe equator. The magniin Figure4 are alsodrivenby a wavelikealternationof tude of K,• in the diurnal mixed layer is approximately
Coriolis and residual forces.
(within a factorof 3) that obtainedby McPhadenet al.
The momentum budget for the layer between z - [1988]in their constant-Kmodel of WWB-generated
130 m and z - 210 m is shown in Figure 16. Again, we flow. Below this layer, however, K,• is more than a
have low-passfiltered the results with cutoff frequency factor of 10 smaller.
22,$12
SMYTH ET AL.: LOCAL RESPONSETO A WESTERLY WIND BURST, 1
The westwardjet centeredat 100 rn (layer 2) appeared initially during leg 1, coincidingwith a weak
wind event which was observedon days330-333 [Wijesekeraand Gregg,1996]. During the Decemberwind
burst, this current exhibited strong oscillations. The
initial acceleration was to the north and appeared as
part of the residualin our one-dimensionalmomentum
budget. This accelerationmight be attributable to a
southwardpressuregradient, though it is unclearhow
such a gradient could have originated. Coriolisturning of the resulting northward current drove the initial intensification of the westwardjet on days 357 and
358. Further accelerationsappear as out-of-phaseos-
cillationsin the residual(pressure
gradient)forceand
the Coriolis force. In the layer containing the EUC
Lukas, R., and E. Lindstrom, The mixed layer of the western
equatorial Pacific ocean, J. Geophys. Res., 96, 3343-3357,
1991.
McPhaden, M.J., and J. Picaut, E1 Nifio-Southern Oscillation displacementsof the western equatorial Pacific warm
pool, Science, œ50, 1385-1388, 1990.
McPhaden, M.J., H.P. Frietag, S.P. Hayes, B.A. Taft, Z.
Chen, and K. Wyrtki, The responseof the equatorial Pacific Ocean to a westerly wind burst in May 1986, J. Geophys. Res., 93, 10,589-10,603, 1988.
McPhaden, M.J., F. Bahr, Y. du Penhoat, E. Firing, S.P.
Hayes, P.P. Niiler, P. L. Richardson, and J.M. Toole, The
responseof the western equatorial Pacific Ocean to westerly wind bursts during November 1989 to January 1990,
J. Geophys. Res., 97, 14,289-14,303, 1992.
Moum, J.N., and D.R. Caldwell, Experiment explores the
dynamics of ocean mixing, Eos Trans. A GU, 75, 489-
(layer 3; see Figure 16), the strongestaccelerations
490,495,
1994.
J.N., and A.E. Gatgert, Mixing efficienciesin stratiby far were the meridionalCoriolis and residualforces Moum,
fied fluids: Data from the main thermochne
and a turbu-
which maintainedthe EUC in approximate(presumably geostrophic)
equilibrium.However,oscillations
in
the subinertial band were present in this depth range
as well.
Acknowledgments.
We are grateful to Chris Fairall
and GeorgeYoung for sharingmeteorologicaldata and computed fluxes with us; to Mike Gregg for allowing us to use
his CTD for daily comparisonswith CHAMELEON temperature and conductivity profiles; to Doug Caldwell and
to the anonymousreviewersfor commentson early versions
of the manuscript; to Mike Neeley-Brown, Ray Kreth, Ed
and Colleen Llewellyn, Sue Young, Chaojiao Sun, Bill Freer,
and Walt Waldorf for engineeringand at-seasupportfor this
project; and to our families for allowing us to be at sea during Christmas. This work was supported by the National
ScienceFoundation(OCE 9110552).
lent tidal front, paper presented at 11th Symposium on
Turbulence and Diffusion, Am. Meteorol. Soc., Boston,
Mass., 1995.
Moum, J.N., D.R. Caldwell, and C.A. Paulson, Mixing in
the equatorial surfacelayer and thermochne, J. Geophys.
Res., 9•, 2005-2021, 1989.
Mourn, J.N., M.J. McPhaden, D. Hebert, H. Peters, C.A.
Paulson, and D.R. Caldwell, Internal waves, dynamic instabihties and turbulence in the equatorial thermocline:
An introduction to three papers in this issue, J. Phys.
Oceanogr., œœ,1357-1359, 1992.
Moum, J.N., M.C. Gregg, R.-C. Lien, and M.-E. Cart, Comparison of turbulent kinetic energy dissipationrates from
two ocean microstructure profilers, J. Atmos. Oceanic
Technol., 1œ, 346-366, 1995.
Schudlich, R.R., and J.F. Price, Diurnal cycles of current,
temperature and turbulent dissipation in a model of the
equatorial upper ocean, J. Geophys. Res., 97, 5409-5422,
1992.
Smyth, W.D., D. Hebert, and J.N. Moum, Local ocean responseto a multiphase westerly wind burst, 2, Thermal
and freshwater responses,J. Geophys. Res., this issue.
Delcroix, T., G. Eldin, M.J. McPhaden, and A. Morliere,
The effect of westerly windbursts upon the western equa- Webster, P.J., and R. Lukas, TOGA COARE: The coupled
ocean-atmosphereresponseexperiment, Bull. Am. Metetorial Pacific ocean, February-April 1991, J. Geophys.
orol. Soc., 73, 1377-1415, 1992.
Res., 98, 16,379-16,385, 1993.
Weller, R.A., and S.P. Andersen, Temporal variability and
Dillon, T.M., J.N. Mourn, T.K. Chereskin, and D.R. Caldmean values of the surfacemeteorologyand air-sea fluxes
well, Zonal momentum balance at the equator, J. Phys.
in
the western equatorial Pacific warm pool during the
Oceanogr., I9, 561-570, 1989.
TOGA Coupled Ocean-Atmosphere Response ExperiEffort, B., and R.J. Tibshirani, An Introduction to the Bootment, J. Clim., in press, 1996.
strap, Chapman and Hall, New York, 1993.
Eriksen, C.C., Equatorial ocean responseto rapidly trans- Wijesekera, H.W., and M.C. Gregg, Mixing in the western
equatorial Pacific during the first leg of TOGA-COARE,
lating windbursts, J. Phys. Oceanogr.,23, 1208-1230,
J. Geophys. Res., 101, 977-998, 1996.
1993.
References
Fairall, C., E.F. Bradley, D.P. Rogers,J.B. Edson, and G.S.
Young,The TOGA COARE bulk flux algorithm, J. Geophys. Res., IOI, 3747-3764, 1996.
Zhang, Q., The oceanicresponseto atmosphericforcingin
the western equatorial Pacific, Ph.D. dissertation, Univ.
of R. I., Narragansett, 1995.
Gargett, A.E., and J.N. Moum, Mixing efficiencies
in turbulent tidal fronts: Results from direct and indirect measure-
ments of density flux, J. Phys. Oceanogr.,25, 2583-2608,
1995.
D. Hebert, Graduate Schoolof Oceanography,University
of Rhode Island, Narragansett, RI 02882.
J. N. Mourn and W. D. Smyth, College of Oceanic and
AtmosphericSciences,OregonState University, 104 Ocean
Hebert, D., J.N. Mourn, C.A. Paulson,D.R. Caldwell, T.K.
Chereskin,and M.J. McPhaden, The role of the turbulent AdministrationBuilding, Corvallis,OR 97331. (e-mail:
stressdivergencein the equatorial Pacific zonal momen- [email protected])
tum balance, J. Geophys. Res., 96, 7127-7136, 1991.
Lien, R.-C., M.J. McPhaden, and D. Hebert, Intercomparison of ADCP measurementsat 0ø, 140øW, J. Atmos.
Oceanic Technol., 11, 1334-1349, 1994.
(ReceivedMay 30, 1995;revisedMay 10, 1996;
acceptedMay 15, 1996.)