Seasonal Circulation in the South Indian Ocean

GEOPHYSICAL RESEARCH LETTERS, VOL. 24, NO. 22, PAGES 2773-2776, NOVEMBER 15, 1997
Seasonal circulation in the South Indian Ocean
Amy Ffield
Lamont-DohertyEarthObservatory,Palisades,
New York
John Toole
WoodsHole Oceanographic
Institution,WoodsHole, Massachusetts
Doug Wilson
NOAA AtlanticOceanographic
andMeteorolog.ical
Laboratory,
Miami, Florida
Abstract. Two World Ocean Circulation Experiment
hydrographic
cruisesin MarchandJune1995,alongwith TopexPoseidonaltimeterdataandNationalMeteorologicalCenterwind
data, estimatedan averageAgulhasCurrent volume transportin
the upper 1000 m of 65 Sv. Effortsto estimatethe seasonalityof
the AgulhasCurrenthaveproducedconflictingresults[Pearce
dataareusedto estimate
seasonal
changes
in theSouthIndian and Grfindlingh, 1982] in part due to different criteria and
Ocean subtropicalgyre. Mean annual curvesderived from the
altimeterand wind data reveal strengthening
of the anticyclonic
gyre in March and September, and weakening in June and
December. The seasonalchangescorrespond
to variationsin the
wind field south of 30 øS at the equinoxesand solstices. In
addition,the wind-drivengyre is furthernorthin July, andfurther
southin March. These variationsin strengthand locationof the
SouthIndian Ocean gyre may influenceinter-oceantransports
south of Africa. Despite the inferred mean annual seasonal
variationsin the South Indian Ocean gyre, volume transports
methods.
As part of the World OceanCirculationExperiment,high
quality,highresolutionhydrographic
surveysof theIndianOcean
wereobtained
in 1995with repetition
of certainsections
to
capture the variability. In March 1995 the NOAA Sh!p
MALCOLM BALDRIGEobtained"repeat"
hyd,rographic
data
acrossthezonalI5W section,nominallyalong33 øS,crossing
the
AgulhasCurrentsouthof Durbanandcontinuingeastwardto 55
øE (Figure1). In June1995,theRV Knorroccupied
the same
sectionobtaining"one-time"hydrographic
coverageof the I5W
estimatedin 1995 from the hydrographic
dataare closeto mean line. In this'letter,the two I5W cruises
alongwith Topexvalues. Apparently, a mesoscaleeddy in March disruptsthe PoseidonaltimeterdataandNationalMeteorological
Centerwind
strongerfall gyre, whereasin June the weaker winter gyre is dataareusedto estimateseasonal
changes
in thesubtropic•al
gyre
delayedby 1 month.
of the South Indian Ocean.
Circulation Inferred From Hydrographic Data
Introduction
Wind analysesfrom the EuropeanCentre for Medium Range
WeatherForecasts(ECMWF) reveal the anticyclonicforcing in
the South
Indian
Ocean.
The
annual
mean
wind
stress is
westward between 10 and 35 øS, and eastward between 35 and 60
Two potential temperature sectionsacrossthe southwestern
Indian Oceanreveal downwardslopingisothermsfrom the coast
of SouthAfrica to 32 øE associatedwith the strongsouthward
flowing AgulhasCurrent(Figure2). Eastof 32 øE the isotherms
generallyslopeupwardsin June indicatingpossiblenorthward
øS [Trenberthet al., 1990]. The anticyclonicforcingdrivesthe interior flow into the Indian Ocean. However in March the
subtropicalgyre of the SouthIndian Oceandefinedby the strong
Agulhas westernboundarycurrentflowing southnear the coast
25'
30'
35'
40'
45'
50'
55'
60'
of Africa, and relatively weak northward interior flow east to
I
I
I
I
I
I
I
I
!
I
I
I
I
I
I
I
I
I
1
....
ii:•.-'..:5!•:.-2-.
!•i•:
ß
.:i•:.--"•
Australia[Toole and Warren, 1993]. Away from the equatorthe
ß..:....
-.:•:.--.::
-.:::s•::
ß
======================
I
RMS
:.::...::,:,•....
ß
............
.
.........
,,
..............
.1
-25'
cud of wind stressis usedin the Sverdruprelationto describethe
-.>-•:"
•
.-.-:.:-'..:•:•:-'.:•5:•5•:•:ß:•:•B•:.-':.•:::•5".-:•:•-':'-•:•:B•:•
I
cm
.................................................................
1
( )
:-":"
•
...................................
depth-averaged,wind-driven, ocean circulation. The 1980 to
! 986 ECMWF windsforce an annualmeanSverdruptransportin
theSouthIndianOceanreaching
100Sv(1 Sv= 1 x 106m3/s)at
the tip of Africa [Trenberth et al., 1990]. The integrated
Sverdrup transportvalue is assumedto be closed by a western
boundarycurrent,in this casethe AgulhasCurrent.
At the southerntip of Africa, Gordonet al. [ 1987] estimated
the baroclinicAgulhasCurrent transportrelative to the seafloor
25'
30'
35'
40'
45'
50'
55'
60'
to be 95 Sv, closeto the amountneededto balancethe Sverdrup
circulation. Much of this flow is locatedin the upper ocean. Figure 1. Root-mean-squared
seasurfaceheightanomalies
in
Strammaand Lutjeharms[1997], usinghistoricalhydrographic the southwest
IndianOceansmoothed
by a 5ø by 5ø gaussian
filter. The contourinterval is 6 cm. CTD (circles)and XBT
(crosses)stationlocationsfor the March 1995 cruiseare shown,
Copyright
1997bytheAmerican
Geophysical
Union.
alongwitha subset
of thecriss-crossing
tracksOfthealtimeter
data. (CTD station locations for the June 1995 cruise are at both
circle and crosslocations.)
Papernumber97GL01253.
0094-8534/97/97GL-01253505.00
2773
2774
FFIELD ET AL.: SEASONAL CIRCULATION IN THE SOUTH INDIAN OCEAN
b) June 1995
a) March 1995
.0 •
'
....
'
' • ............
.'
• ....
! ....
I ....
transportsare integratedfor the upper 1000 m (only near the
Madagasca.r
Ridgeat 44 øE doesthe bottomtopography
rise
I
slightlyabove 1000 m).
?•::•i[
'"'"'""•'••••:.....
.••••
,, 18
The cumulativegeostropic
transports
(Figure3) quantifythe
100
:•
!••Jl•l'•
-•i.---:.-
2OO
200
300
300
400
4OO
500
500
6OO -
600
700
700
'
........
I ....
35
I ....
40
45
I
....
I ....
35
50
I ....
40
I ....
45
I
50
.flow patternsimplied by the potentialtemperaturesections
(Figure2). In JunetheAgulhasCurrentflowsswiftlysouthward
with61 Sv in theupper1000m, similarto thehistorical
average
valueof 64 Sv reportedby StrammaandLutjeharms[1997]. In
March the eddy obscuresthe offshoreedge of the Agulhas
Current.Integrated
to 32 øE - thewesternedgeof thecyclonic
eddy- the cumulative.transport
is 7 Sv lessin March(54 Sv) than
in June (61 Sv). However, when integratedto 36.7 øE - the
easternedgeof thecycloniceddy- thecumulative
transport
is 3
Sv morein March (60 Sv) thanin June(57 Sv). In the interior,
between
32 and50 øE,theflowis weakwith0 Svnettransport
in
March,and4. Sv net northwardtransport
in June. Thesenet
interiorflows are not significantlydifferentfromeachother.
Longitude(o East)
Figure 2. Potentialtemperaturesectionsnominallyalong 33 øS
for the a) March 1995 and b) June 1995 cruises. The contour
interval is 1 øC. CTD and XBT data are usedin a), and CTD data
are'used in b).
a) MeanAnnualAverage
0.00
15
situationis more complicatedwith the isothermsbelow 200 db
revealing a cyclonic mesoscaleeddy centeredat 34.5 øE, and
possiblenorthwardthen southwardflows centeredat 42 øE. Both
thesefeaturesare centeredover shallow 1000 m topography:the
MozambiqueRidge at 35 øE and the MadagascarRidge at 44 øE.
Cyclonic eddies have been previously observedin this region
[Griindlinghet al., 1991; Griindlingh, 1988].
To compare volume transports across the zonal sections,
geostrophic velocities are estimated. Rather than attempt to
determine the absolute transport across the sections, the
calculationsare performedin a way to insuregoodcomparisons
between the different occupations. In this way baroclinic
transportscan be directly comparedto eachother,and errorsdue
to bottom topography,stationspacing,and small velocitiesare
reduced: 1) the integrationstartsslightly offshoreAfrica where
the bottom depth is greater than 1000 m, 2) only stations
occupiedduringbothcruisesare included,3) a level of no motion
at 2000 dbarsor the deepestcommondepthis selected,and 4) the
•
0.25
0.50
0.75
1.00
10
E
v
•
o
I
<t:
-5
U• -10
-15
95.00
30
95.25
b) 1995
95.50 95.75
96.00
20
0
o
-20
-30
-40
-60
e---•---30
35
'-'
Figure 4. Sea surfaceheightanomaly(SSHA) differenceacross
the "Agulhas"(31.1 øS, 30.4 øE - 32.2 øS, 32.5 øE, .solidline),
"• +'"-June1995
40
45
50
LongitudeEast (o)
Figure3. Cumulative
geostrophic
transport
in theupper1000m
usinga subsetof the March 1995 (opencircles)and June1995
(soliddiamonds)
stationdatanominallyalong33 øS.
"Indian" (32.2 øS, 32.5 øE - 33.2 øS, 112.4 øE, thick dashedline),
and "SouthwestIndian" (32.2 øS; 32.5 øE - 33.2 øS, 50.0 øE, thin
dashedline). In a) the mean annual average SSHA differences
are smoothedby a 6 monthgaussianfilter, with starsmarkingthe
solsticesand equinoxes,and in b) the 1995 SSHA differencesare
smoothedby a 2 monthgaussianfilter with the occurrences
of the
March(opencircle)andJune(soliddiamond)
cruises
lndicated.
FFIELD ET AL.' SEASONAL CIRCULATION IN THE SOUTH INDIAN OCEAN
However, the energetic +/- 20 Sv flows centered over the
MozambiqueRidge at 35 øE andthe MadagascarRidge at 44 øE
in March contrastwith the +/- 5 Sv amplitudeflows observedin
June.
Circulation
Inferred
From Altimeter
Data
Fully processedTopex/Poseidonsea level heightdata were
obtained from the NASA
Ocean Altimeter
Pathfinder
Data
Projectfor the period,October1992 throughApril 1996. The
data are averagedto 0.2 degreesspatialresolutionalong each
2775
clearly reveal a seasonalcirculation in the South Indian Ocean
(Figure zt). On Figure 4 negative values imply anomalous
southward flow (or reduced northward flow if the mean flow is
larger and northward) and positive values imply anomalous
northward flow (or reduced southward flow if the mean flow is
larger and southward). The seasonalcirculation is maximum
with stronger southward "Agulhas" and stronger northward
"Indian"flows impliedduringMarch and September(starson
Figure 4a). In other words a strongergyre. In contrastduring
June and December, the oppositesituationis observed: weaker
southward
"Agulhas"
andweaker
northward
"Indian"
flows.The
satellitetrack and the threeyear, 1993 to 1996, meansubtracted. same geasonalityis found in the mean-annualcurve constructed
Consequently,
the contributionof the geoidandthe meanocean for the "Southwest Indian".
This strengtheningand weakeningof the SouthIndian Ocean
circulationto the sealevel heightvaluesis effectivelyremoved.
The remaining data are time series of sea surface height gyre may be driven by seasonalvariationsin wind, becausethe
correspondto equinoxes(March and September),
anomalies(SSHA) at approximately10 day intervals. The root- strengthenings
mean-square(RMS) of the SSHA data illustratesthe high and weakeningscorrespondto solstices(June and December).
variability(> 12 cm)of thesouthwestern
IndianOceanassociated The monthlymeanpositionof the centerof the subtropicalhigh
withtheAgulhasReturnCurrent,theeastern
edgeof theAgulhas reaches maximum southerly positions in March (35 øS) and
September
(32 øS),andmaximumnortherlypositions
in June(30
Current,andnumerousmesoscaleeddies(Figure 1).
(30 øS)[VanLoon,1971,Fig.7]. At 90 øE,
The slopeof the seasurface- whichcanbe calculatedusing Os)andNovember
SSHA data - is proportionalto geostrophicflow at the sea 50 øS, the annualcycle of mean zonal wind stressalso reveals
in March(3.5dyncm-2)andSeptember
(3.7dyncmsurface.Gcopotential
topography
at theseasurface,500 m, and maximums
in May (2.6dyncm-2)andDecember
(1.9dyn
1000m all revealan anticyclonicgyrein the SouthIndianOcean 2), andminimums
etal., 1989,Fig.23]. Onereport
of geostrophic
westof 80 øE [Wyrtki,1971,plates388, 389, 390]. Thissuggests cm-2)[Trenberth
thatcirculationrevealedby SSHA dataat the seasurfacemirrors volume transport of the Agulhas Current relative to 1000 m
deepercirculationpatterns. In orderto directlycomparethe suggeststhis seasonality:23 Sv in January-February,38 Sv in
SSHA datato theresultsof thetwo hydrographic
sections,SSHA April, 32 Sv in July-August-September, 27 Sv in Octoberdataintersecting
the sectionsare examined.The largestRMS November. However, when the geostrophic velocities are
variability along the gectionsis at 32.5 øE, which is also referenced to observed surface currents the seasonality
[Griindlingh,
1980,averages
compiled
fromhisFig.
approximatelythe transitionpoint betweenthe southward disappeared
flowing AgulhasCurrentand the eventualnorthwardinterior 6].
Despite the mean annual seasonality, at the times of the
flow to the east. To investigatethe variability of the current
flows, the differencesin SSHA acrossthreelongitudeintervals hydrographic cruises in March and June, 1995, the SSHA
areestimated:the "Agulhas"
(31.1øS,30.• øE to 32.2 øS,32.5 differencesare all within 3 cm of the mean (Figure4b). In March
of thegyreappears
disrupted
by
øE),the "Southwest
Indian"(32.2øS,32.5 øEto 33.2øS,50.0 øE; 1995theexpectedstrengthening
eastern
edgeof thehydrographic
sections),
andthe"Indian"(32.2 the mesoscaleeddy crossingthe section, whose circular SSHA
øS,32.5 øE to 33.2 øS, 11•.• øE,easternedgeof the SouthIndian depressionsignaturecan be tracedfrom March 9 to April 20. In
Ocean).
Mean annual curves constructed for the SSHA differences
a) Average Sverdrup Transport
0.00
75
'•' 70
between
0.25
34 and 45 øS
0,50
0.75
1.00
June 1995 the expected weakening of the gyre seems to be
delayed 1 month.
b) Latitude of Maximum Transport
0.00
36
0.25
0.50
0,75
1.00
38
o
n.
65
i-.. 60
E
o
55
Figure
5. Ina)themean
annual
curve
oftheaverage
Sverdrup
transport
between
34and45øSalong
32.5øEisdrawn
fortheSouth
IndianOceangyre,with starsmarkingthe solstices
andequinoxes.In b) themeanannualcurveof the latitudeof maximumSverdrup
transportalong 32.5 øE is drawn, along with the 1995 curve showingthe occurrences
of the March (open circle) and June (solid
diamond)
cruises.Thedataaresmoothed
bya 6 monthgaussian
filter,andthemeans
areindicated
by dashed
lines.
2776
FFIELD ET AL.: SEASONAL CIRCULATION
Circulation
Inferred
From
Wind
Data
The data-assimilating, operational, T-62 model wind field
from the National Meteorological Center is used with a drag
coefficient of 1.4, to compute the Sverdruprelation at -1.5
degree spatial resolution and weekly averaged temporal
resolution. The Sverdrup circulation is estimatedevery week
from 1991 through 1995, by integrating volume transports
calculated from the Sverdrup relation westward from the
Australian and New Zealand's coasts between 20 and 50 øS.
The AgulhasReturnCurrentis locatedbetween34 and45 øS,
and feeds the northward
interior circulation
of the South Indian
gyre. The meanannualcurveof the averageSverdruptransport
between 34 and 45 øS, at 32.5 øE is constructed for the South
indian gyre (Figure5a). The averageAgulhasReturnCurrentis
estimated to be 65 Sv. Seasonally, the wind-driven Agulhas
Return Current follows the same semi-annual seasonality
revealedby the altimeterdata. For example,in boththe altimeter
data and the wind data, March (70 Sv) and September(70 Sv)
correspondto maximumgyres,and June(55 Sv) and December
(65 Sv) to minimum gyres (stars on Figures 4a and 5a). The
centerof the subtropicalanticycloniccirculationalsovarieswith
season(Figure5b). The latitudeof maximumSverdruptransport
along32.5 øE is furthestnorthin July (38 øS) andfurthestsouth
in March (43 øS). This meridional movement of the gyre is
partiallyresolvedby historicalhydrographic
sections
crossing30
øS, 65 øE, that show a shift in the circulation southwardbetween
January and March, and northward during June and July
[Strammaand Lutjeharms,1997].
In March 1995 the position of the wind-driven SouthIndian
Ocean gyre is anomalously north relative to the mean annual
curve (Figure 5b). Whereas, in June 1995 the northward
movementof the gyre appearsdelayedone month.
Discussion
and Conclusions
Altimeter and wind data usedto estimateseasonalchangesin
the anticyclonic subtropical gyre of the South Indian Ocean
reveal a strongergyre in March and September,and a weaker
gyre in June and December. The Sverdrup transport of the
Agulhas Return Current suggeststhat seasonalvariationsrange
between +5 to -10 Sv about the 65 Sv mean. In addition, the
meridionalpositionof the gyre is furthestnorthin July (38 øS),
and furthest south (43 øS) in March. The meridional movement
and strength of the wind-driven gyre may have important
implications for inter-oceantransports,by possiblyinfluencing
the water
mass
sources
drawn
into
the South
Indian
Ocean
subtropicalgyre and by modulating volume transportssouthof
Africa.
OCEAN
in the southwestIndian Ocean. For example, even thoughthe
integratedtransportvalues acrossthe hydrographicsectionsare
similar in March and June, energetic +/- 20 Sv flows were
observedin March centeredover the MozambiqueRidge at 35 øE
and the Madagascar Ridge at 44 øE. No notable transport
features were seen there in June.
Theresponse
of thesouthwestern
Indian
Ocean
toseasonal
variation in wind forci.ng is clearly complicatedby the major
ridge systemsof this basinthat attaindepthsshallowerthan 1000
m. Onlybarotropic
wavemotions
propagate
sufficientiy
fastto
transmitinformation about seasonallyvarying wind stressacross
mid-latitude oceans.
Baroclinic waves are too slow.
Yet,
barotropic motions can be reflected and/or scattered into
baroclinic modes at tall bathymetry. Better understandingof
theseprocessesmay help explain the high SSH variability of the
southwesternIndian Ocean, and shed light on the seasonal
transport field of the Agulhas Current and its associated
interbasinexchanges.
Acknowledgments. This work was supportedby the National
Oceanic and Atmospheric Administration grant NA66GP0267. The
Lamont-DohertyEarth Observatorycontributionnumberis 5648.
References
Gordon,A. L., J. R. E. Lutjeharms,and M. L. Griindlingh,Stratification
and circulationat the AgulhasRetroflection,Deep-SeaRes., 34, 565599, 1987.
Griindlingh, M. L., On the volume transport of the Agulhas Current,
Deep-SeaRes., 27, 557-563, 1980.
Griindlingh,M. L., Review of cycloniceddiesof the MozambiqueRidge
Current,S. Aft'. J. Mar. Sci., 6, 193-206, 1988.
Griindlingh, M. L., R. A. Carter, and R. C. Stanton,Circulation and water
properties of the southwest Indian Ocean, Spring 1987 Prog.
Oceanog.,28,305-342, 1991.
Pearce,A. F., and M. L. Griindlingh, Is there a seasonalvariationin the
AgulhasCurrent?,J. Mar. Res.,40, 177-184, 1982.
Stramma,L., andJ. R. E. Lutjeharms,
The flow field of the subtropical
gyre of the SouthIndian Ocean,J. Geophys.Res., 102, 5513-5530,
1997.
Toole, J. M., and B. A. Warren, A hydrographicsectionacrossthe
subtropicalSouthIndian Ocean,Deep-SeaRes.,40, 1973-2019,1993.
Trenberth,K. E., W. G. Large,andJ. G. Olson,The meanannualcyclein
globaloceanwind stress,J. Phys.Oceanogr.,20, 1742-1760,1990.
Trenberth,K. E., J. G. Olson,and W. G. Large, A globaloceanwind
stressclimatology based on ECMWF analyses,NCAR Tech. Note,
NCAR/FN-338+STR, 93 pp., Nat. Center for Atmos. Res., Boulder,
CO, 1989.
Van Loon, H., A half-yearly variationof the circumpolarsurfacedrift in
the SouthernHemisphere,Tellus,23, 511-516, 1971.
Wyrtki, K., Oceanographicatlas of the International Indian Ocean
Expedition,531 pp., Nat. Sci. Found.,Washington,D.C., 1971.
A. Ffield, Lamont-DohertyEarth Observatory,Palisades,NY 10964-
Despite the inferred seasonal changes, baroclinic volume
transportsestimatedfrom the March and June1995 hydrographic
data are close to the historical
IN THE SOUTH INDIAN
mean.
In March
1995 a mesoscale
eddy disruptsthe expectedstrongerfall anticycloniccirculation
and the South Indian gyre movesunusuallyfar north during the
same time interval. In June 1995, the expectedweaker winter
anticyclonic circulation and northwardmovementof the gyre is
delayed by 1 month. These results mask the far more
complicatedreality indicatedby the large sea level height RMS
8000. (email: affield@ldeo.columbia.edu)
J. M. Toole, Mail Stop 21, Departmentof PhysicalOceanography,
Woods Hole Oceanographic Institution, Woods Hole, MA 02543.
(email:jtoole@whoi.edu)
W. D. Wilson, NOAA Atlantic Oceanographicand Meteorological
Laboratory, 4301 RickenbackerCauseway,Miami, FL 33149. (email:
wilson@aoml .noaa.gov)
(Received
March13, 1997;accepted
April 18, 1997.)