University of South Florida Scholar Commons

Marine Science Faculty Publications College of Marine Science

1989

The Wind‐Driven Seasonal Circulation in the Southern Tropical Indian Ocean

Karen E. Woodberry

Mark E. Luther Florida State University, [email protected]

James J. O'Brien

Follow this and additional works at: https://scholarcommons.usf.edu/msc_facpub

Part of the Life Sciences Commons

Scholar Commons Citation Woodberry, Karen E.; Luther, Mark E.; and O'Brien, James J., "The Wind‐Driven Seasonal Circulation in the Southern Tropical Indian Ocean" (1989). Marine Science Faculty Publications. 499. https://scholarcommons.usf.edu/msc_facpub/499

This Article is brought to you for free and open access by the College of Marine Science at Scholar Commons. It has been accepted for inclusion in Marine Science Faculty Publications by an authorized administrator of Scholar Commons. For more information, please contact [email protected]. JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 94, NO. C12, PAGES 17,985-18,002, DECEMBER 15, 1989

The Wind-Driven SeasonalCirculation in the SouthemTropical Indian Ocean

KARENE. WOODBERRY1,MARK E. LUTHERAND JAMES J. O'BRIEN

MesoscaleAir-Sea Interaction Group, Florida State University, Tallahassee

A numerical model of the Indian Ocean, driven by climatological monthly mean winds, realistically simulates the major features of the large scale upper ocean circulation observed in the southem hemisphereand equatorial regions. The principal feature in the tropical Indian Ocean is a basin-wide clockwise southemhemisphere (cyclonic) gyre comprisedof the South Equatorial Current to the south, the the South Equatorial Countercurrentto the north, and the East African Coastal Current in the west. Rossbywaves propagatewestward in the shearzone between the South Equatorial Current and the South Equatorial Countercurrent,and are obstructedand partially reflected by the banks along the Seychelles- Maudflus Ridge (60øE). A region of high eddy activity northwestof Madagascaris an extension of the tropical gyre and is a tropical analog to the Gulf Stream recirculation region. Oscillations in meddional transportat the equatorhave westwardphase speedand eastwardgroup velocity and are the result of mixed Rossby-gravity (Yanai) waves forced by oscillations in the highly nonlinear wesrum boundary current region. Oscillationswith 40- to 50-day periods are seen in most currents. These oscillationscannot be atmosphericallyforced, as the shortestperiod in the mean monthly wind forcing is 60 days. Mean transports in the wesrum basin agree with observations. Small (2 Sv) mean throughflow from the Pacific to the Indian Ocean at the eastem open boundary is due to wind-forced Indian Ocean dynamicsalone and is within the range of observationsof throughflowfrom the Pacific.

1. INTRODUCTION Lamb, 1979], one might expect variability in the wind- driven southern hemisphere circulation to be an important The expansivesouthern hemisphere oceans have beenless consideration. Indeed, Anderson and Moore [1979] show studied than the smaller, more well-travelled oceans of the that changesin the southernhemisphere trade winds with the Northern Hemisphere. Large areas of the southern monsoonscan affect the Somali Current north of the equator. hemisphereoceans are relativelyunexplored, resulting in a Knowledge of the southernIndian Ocean is also necessaryfor scarcityof both oceanographicand meteorological data. The Indian Ocean is a predominantlysouthern hemisphere ocean understandinglarger scale phenomena. Madden and Julian [1972] postulated an Indian Ocean source for observed40- to anddata coverage is correspondinglythin. The effortsof the 50-day atmospheric oscillations. Barnett [1983] International Indian Ocean Expedition in 1962-1965 investigatedthe interactionbetween Indian Ocean and Pacific increasedthe oceanographicdata availableby a factorof 5 Ocean winds and found strong coupling at interannual time but emphasizedthe equatorialzone over southernregions scales and a strong connection to E1 Nifio events in the [Wyrtki,1971]. More recentefforts such as the FGGE (First Pacific. Nicholls [1984] found Indonesian sea surface GARPGlobal Experiment) Indian Ocean Experiment (INDEX) temperature(SST) anomaliesto lead SouthernOscillation and [see Schott, 1983; Swallow et al., 1983] and individual Pacific SST changesby about a seasonand suggestedthat studies of Schott et al. [1988] and Swallow et al. [1988] interaction between ocean and atmospherein the Indonesia- have addedto our knowledgeof the westernportion of the north Australia region may be a direct link between basin, but on the whole, observationaldata remain sparse. Indonesian SST and E1 Nifio - Southern Oscillation (ENSO) We attemptherein to increaseour understandingof this events. As ENSO events are often viewed as perturbationsof oceanthrough the use of a realisticwind-driven numerical the seasonalcycle, it is important to understandthe seasonal model by comparing model fields with available cycle of the circulation. observationaldata. Where agreementis found,if,; modelcan Numerical models have proven useful in studying the be usedto interpretthe data in a wider context. responseof the ocean to changingwind, helping to integrate The Indian Ocean is interesting most obviously because and interpret widely scattered observations. A of the seasonal monsoon reversals in winds and currents comprehensive review of modelling efforts in the tropical alongthe SomaliCoast. Extensiveobservational effort has Indian Ocean is given by Knox and Anderson [1985], in been focusedon the northwestregion, with relatively little Knox [1987] and in Luther [1987]. Modelling of the Indian attentionpaid to the southernsubtropical circulation. A Ocean at the Florida State University, Tallahassee, has foundationis thereforelacking for understandingthe effects concentratedon the northwest (Somali Current) region of the of the southernhemisphere on the Somali Current and the Indian Ocean. We have recently extended the domain of the equatorialregimes. Since much of thevariability in the wind Luther and O'Brien [1985] model to cover the Indian Ocean fields is found in the southernhemisphere [Hasienrath and to 25øS. In this paper we describe the model simulation of the seasonal circulation in the southern hemisphere and 1Nowat the Centerfor Atmospheric Theory and Analysis, compareit with available observationsand theory. University of Colorado,Boulder The dominant feature of the observedsouthern hemisphere circulation is the subtropicalgyre with the westward flowing Copyright1989 by the AmericanGeophysical Union South Equatorial Current (SEC) at 12øN as its northern Paper number 89JC01511. boundary(Figure 1 [Diiing, 1970; Pickard and Emery, 1982]). 0148-0227/90/89JC-01511505.00 The SEC splits at the coast of Madagascarinto northward and

17,985 17,986 WOODBERRY ET AL.' SEASONAL CIRCULATION IN THE INDIAN OCEAN

20' 30* 40' 50' 60' 70' 80' 90' I00' I I O' 30' 120' 130' 140' 150•0.

SURFACE CIRCULATION

20' JANUARY 20'

NAUTICAL MILES CM/SEC PER DAY

6 - 18 • 12 -- 35 ( 5 *--- ( IO

NUMBERS ON MAP ARE NAUTICAL MILES PER DAY

30'

40'

60'

70' •0ø 20' 40' 60 ø 80' I00' 120' 140ø

20' 30' 40' 50' 60' 70' 80' 90' I00' I I O' 30' 120' 130' 140' 150•0.

SURFACE CIRCULATION

JULY 20'

NAUTICAL MILES CM / SEC PER DAY

>..6 - 2018 ••--- 1•2--3540

NUMBERSON MAP ARE NAUTICAL MILES PER DAY

30'

40'

50 ,0'

60'

70' 70' 20' 40' 60' 80' I00' 120' 140'

Fig. 1. Observedlarge scalesurface circulation during (a) NE and (b) SW monsoonsfrom an analysisof ship drift data (courtesyof K. Wyrtki). WOODBERRYET AL.: SEASONALCIRCULATION IN THE INDIAN OCEAN 17,987

southward branches. During the northeast monsoon 25øN

(Novemberto March), the northwardbranch rounds the tip of 20øN Madagascarand continuesnorth along the African coast as the East African Coastal Current (EACC) to meet the 15øN I• SOCOTRA southwardSomali Current. They turn offshore together at 10øN about 3øS [$chott, 1983; $chott et al., 1988] and form the 5ON eastwardSouth Equatorial Countercurrent (SECC). During the I MALDIVES EQ southwest monsoon (May to September), the EACC feeds SEYCHELLES into the northward Somali Current and return flow is via the 5oS SouthwestMonsoon Current from the northernhemisphere. 10oS The model simulationreproduces this southernhemisphere 5 •AYAMAL•DE tropical gyre defined by the westwardSEC to the south, the 15øS 11 • NAZAIRETH eastward SECC to the north, and the EACC in the west. The 20oS MADAGASCARmtß BAN12 westernpart of the gyre is a region of high eddy activity. OPEN 25oS The model flows split aroundthe islandsin the domain with 50o transportsthat compare well with observations. Available observational data from the equatorial and Fig. 2. Model geometryfollowing the 200 m depthcontour. Land boundariesare shaded. This studyfocuses on the region southof southernIndian Ocean are concentratedin a few regions. 5øN. Islandsin this regionare identifiedin the text. Also indicated Wyrtki [1973] and Knox [1976] documentedthe zonal jets are the numberedsections used for the transportcalculations shown and current reversals at the equator during monsoon in Figures 4-10 and 12-13. transitions. Luyten and Roemmich [1982] showed a semiannual period dominating the zonal currents near the African coast, with a dominant 26-day period in the include the Laccadive and Maldive islands, the Chagos meridional velocities at 200 m. The latter were shown to be Archipelago,and the banks aroundSocotra and along the mixed Rossby-gravity, or Yanai, waves generated at the Seychelles-MauritiusRidge. These are all treated as land western boundary [Kindle and Thompson,1989; Moore and (closed)boundaries. Along the Seychelles-MauritiusRidge, McCreary, 1989]. South of the equator,several studies have the major banks and island groupsare the Seychelles(5øS, found 40- to 50-day oscillations in the alongshorecurrents 55øE), Saya de Malha Bank (10øS, 63øE), and NazarethBank between0 ø and 5øS [Mertz and Mysak, 1984; Mysak and (15øS, 63øE). Thesebanks are generallyless than 30-40 m Mertz, 1984] and off the northern tip of Madagascar deep and are dotted with reefs and small islands. They [Quadfasel and Swallow, 1986; Schott et al., 1988]. Schott present significant barriers to the upper level flow in the et al. [1988] and Swallow et al. [1988] present transport region and are therefore treated as closed boundaries. calculations for the boundary currents along eastern The model is forcedby the monthlymean climatological Madagascarand questionthe existenceof a seasonalcycle. winds of Hellerman and Rosenstein [1983]. The Global Variousmethods have beenused to estimatethe throughflow Marine Sums wind speed climatology from the National from the Pacific into the Indian Ocean through the Climate Data Center is used to computeand divide out an Indonesianislands [Godfrey and Golding, 1981; Piola and average drag coefficient from the Hellerman-Rosenstein Gordon, 1984; Fine, 1985; Gordon, 1986]. In addition to stressdata, recoveringvalues of pseudostress,WW, where W these observations,we compare our model results with the is the vector wind velocity and W is its magnitude, as transportsand oscillationsin anotherreduced gravity model describedin Leglet et al. [1989]. This is done for reasonsof [Schottet al., 1988; Kindle and Thompson,1989]. consistencywith other wind data products. The 2ø by 2ø In the following pageswe first describethe model and the gridded data set is interpolated to the model grid (0.2ø seasonalcycle of the windsused to force it. We thenpresent between like grid points) using a bicubic spline the resultsof the model simulationand highlightagreement interpolatingprocedure. We assumethat eachmonthly mean with observations,looking first at the large-scalecirculation representsthe value at the middle of the respective month and then at the areas of interest noted above. We conclude and interpolatelinearly between them to obtain a pseudo- with an overview of the wind-driven circulation and a stressdata set covering a full annualcycle at the model time discussionof the strengthsand weaknessesof the present step of 20 min. We convert the pseudostressfields back to model. wind stressusing the bulk aerodynamicformula X=pa CD WW 2. MODEL where Pa is the density of air and C D is a constantdrag The mod½lused is that describedby Luther and O'Brien coefficient. The drag coefficientthus becomes a parameterof [1985], Luther et al. [1985], and $immorkvet al. [1988]. For the model rather than of the wind analysis. For the results the model domain, realistic geometry of the Indian Ocean presentedhere, Pa = 1.2kg m -3 andC D = 1.5x 10-3. basin is used from 35øE to 120øE and from 25øS to 26øN as The model is integratedfrom rest startingat 000 UT on shown in Figure 2. The boundary conditions at all solid December 16 with an initial value of model upper layer (land) boundariesare the no-slip conditions: u=v=0. Most thickness(ULT) set at H0=200 m. For the resultspresented of the southernboundary along 25øS and a portion of the here, we set the kinematic eddy viscosityA equal to 750 eastern boundary from 10øS to 20øS are open boundaries. m2 s-1 andthe reducedgravity g' to 0.03m s-2. For The boundary condition applied there is the Sommcrf½ld simplicity, the model year has 360 days, with 30 days in radiation condition describedby Camerlengo and O'Brien each month. The model is integratedfor 20 years, with the [1980]. The 200-m depth contour defines the mod½l land annual wind cycle repeated year after year. After boundaries. The shallow banks and islands in the domain approximately6 to 8 years of integration,the model has 17,988 WOODBERRYETAL.: SEASONAL CIRCULATION IN THE INDIAN OCEAN settled into a steady, repeating seasonalcycle, with only strengthen and become more southerly. As in coastal small differences from one year to the next occurring in regions, thesepatterns intensify throughJuly and then decay regionswith highly nonlinearflows [Luther and O'Brien, throughSeptember. During the October transitionfrom SW 1989]. Resultsfrom the tenth year are presentedhere. to NE monsoon, the changing winds in the Arabian Sea convergenear the southerntip of India, forming a westerly 3. WINDS maximum at 5øN, 80ø-85øE, and near-zero wind stress curl is again found along the equator. Figure 3 showsthe wind and curl fields over the model The trade winds in the southern hemisphere are domain at the height of the northeast and southwest southeasterly throughout the year. In January they are monsoonsfrom the climatologicalwinds of Hellerman and strongestin the southeast,turning easterly at about 60øE, Rosenstein [1983] that are used to force the model. For a and convergingover Madagascarwith the NE monsoonwinds complete descriptionof the atmosphericfields over this crossing the equator (Figure 3a). Southerly winds at the region, see the excellentatlas by Hastenrath and Lamb coast of Australia diverge to contributeboth to the SE trades [1979]. The seasonalcycle of winds over the Indian Ocean and to westerly winds at the easternboundary of the model• is dominated by the monsoon reversals in the northern Near zero wind stress curl is found across the basin near hemisphere. Differential heating drives oscillations in 20øS, with negativecurl to the north and positive curl to the northern and southern hemisphere pressuresystems causing south (Figure 3b). The tradesstrengthen in April, reaching seasonal changes in the winds. During the northern the African Coast, and in May they extend acrossthe equator hemispherewinter, the northeastmonsoon is characterized with the onset of the SW monsoon, where they feed the by a high pressure center that exists over central Asia, Findlater Jet. drivingnortheasterly winds acrossthe northernIndian Ocean into a troughnear the equator. (We use the meteorological 4. MODEL CmCUL•TION FIELDS convention of describing the wind by the direction from which it is blowing, while we use the oceanographic Plate 1 showsthe model fields for February and August for conventionof namingthe directiontoward which a currentis the model domain from the tenth year of the integration. flowing;thus, a westwardcurrent flows in the samedirection (Plate 1 is shownhere in black and white. The color version as an easterly wind.) During the southwestmonsoon of can be found in the separatecolor sectionin this issue.) The northernhemisphere summer, a high-pressurecenter develops upper layer thickness,which mimics dynamic topography, over Madagascarand the MascareneBasin with concurrent and the major currentsin the model are in good qualitative intensification of the southeast tradewinds. The trades blow agreementwith the climatologiesof dynamicheight and 20ø acrossthe equator,are redirectedtoward the northeastby the isothermdepth of Wyrtki [1971], with the ship drift data of African highlands,and form a strongatmospheric jet, often Cutler and Swallow [1984], and with the climatology of called the Findlaterjet [Findlater, 1971], which blows into a mixed layer depth of Rao et al. [1989]. The monsoon troughsituated over northernIndia. reversals in the Somali Current, the formation of the great The northeast (NE) monsoon lasts from November whirl in the (boreal) summer Somali Current, and the throughMarch with strongestnortheasterly (southwestward) formation of the eddy field off of the Arabian Peninsulaare winds in January along the Somali coast to about 15øS well represented in the model and are consistent with (Figures3a and3b). The northeasterlywinds are associatedprevious simulationsfrom an earlier version of the model with negativecurl along the coast and positivecurl on the describedby Luther and O'Brien[1985], Luther et al. [1985] eastern side of the jet axis. Patterns of stress and curl and Simmons et al. [1988]. We concentrate here on the remain similar in February through March, although the seasonal circulation in the southern tropical Indian Ocean strengthof both decreasesas the NE monsoondecays. By from 20øS to 5øN. The primary feature of the circulation in April the windsnorth of Madagascarare southeasterlyas the this region is a basin-wide tropical wind-driven gyre, defined trades extend to the African coast. by the South Equatorial Current to the south, the South The onset of the SW monsoon begins in May with Equatorial Countercurrentto the north, and the East African southerlywinds curving along the entireAfrican coast. Curl Coastal Current as its western boundary current. This patternshave changedsign north of 10øS. Thesestress and tropical gyre is in many respects similar to a mid-latitude curl patternsintensify dramaticallyduring May throughJuly wind-driven gyre, responding to the basin-wide wind stress (Figures 3c and 3d). A sharpcurl gradientover the Arabian curl distribution, with a western boundary current Sea follows the axis of the southwesterly Findlater jet recirculation region characterizedby intense eddy formation. [Findlater, 1971]. The SW monsoondecays in August and It differs from a mid-latitude gyre in its proximity to the September with the transition from SW to NE monsoon equatorial wave guide and in the extreme seasonal wind occurringin October. The NE monsoonintensifies through variations with the changing monsoons. Because of the the end of the year with reversalsin direction of the wind seasonalcycle in the winds, the northern limit of this gyre stress and the sign of the curl patterns along the African migrates between the northern and southernhemispheres, so coast. that the western boundary current crossesthe equator during In the equatorialregion, westerly winds predominateaway the southwest monsoon and feeds the northwestward Somali from the coast with relatively strong westerly maxima Current. The presenceof the equatorial wave guide allows present during monsoon transition months of April and energy to propagate away from the western boundary in the October. During the height of the NE monsoon,equatorial form of equatorial Kelvin waves and mixed Rossby-gravity winds have a northerly component changing to westerly waves. These waves are not available in a mid-latitude south of the equator. Curl is positive in a 10-15 ø band western boundary region. The low latitude of this gyre along the equator. In April wind stresscurl is near zero all allows energy input by the winds across the interior to along the equator, becoming negative by May as winds propagatewestward much more rapidly. WOODBERRYET AL.' SEASONALCIRCULATION IN THE INDIAN OCEAN 17,989

ILl L.LJ

t•

z z z z z 0 LO 0 LO LO 0 LD 0 LF) CXJ LO o LO 0 C• .-• .-• .'• .-• N N N

z z o9 LID L• L• N 17,990 WOODBERRYET AL.: SEASONALCIRCUI•TION IN THE INDIAN OCEAN

O. $ !'4/5 FEB t6, YR tO

: . 25N

20N

15N

tON

SN

EQ

55

iss

20S

25S .. 3sE •0E •tSE S0E SSE 60E 6SE 70E 7SE 80E 8SE g0E gSE 100E 105E I tOE t lSE

1$0 152 1G• 1715 188 200 212 22• 226 2•f:• 261) 27• 28• 296 308

0.5 M/'$ RUG 16, YRI0 25N

,...: i.-'•½½.-.'-'•a:;:m•;-.;;•½.:•-?:4..'..;:•.. 20N

15N

•:½... -.... q:•: ..--•:N•;.-:.:::•:•>A•>-:.:.,:• • • .:.;•:::•:m•.•;:;•.•;•:..•::.•:..:•':::. . ... tON

5N •; <':- :,:.• • •:•:::•!?:•:;•'••:•..•::::::.::•:?::.•:..•.. • .•i•:;•!::.':.....•: • •..• ...... :.:. EQ

55 ' :":'""• ...... : ::::::'4•.•:::: ...... •.';...... ::...... :•::.;':•. •...... • ::•":..:•: :::::.::::.:.?• ,,:•:: :::...... :...... •,.:.....:.....:-•:::::::'::::::-:: •-.-j•..• % :;•[•..;;..:"'•: ..' :•::;: ß......

.... .:...:...... tSS ..• ...... :::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::q•:.:•::::

20S ];; .....h•.. ,.. ::•,• .:?.. ..:•.:•::.,..•.....:.•:.:: ::..... ,,:,. : 25S ...... 3SE ,•0E •tSE S0E SSE 60E 6SE 70E 7SE 80E 85E õ0E õSE 100E 10SE I t0E i lSE

b 14:0 1$• ! 6• 176 188 200 212 22• 2:•6 2•8 260 272 284: 296 •08

Plate 1. Model upper layer thickness and velocity from the tenth year of integration. Arrows indicate upper layer velocity,with only one arrow showheach 1.6ø in latitudeand longitude. Arrowsrepresenting velocities greater than 1.0 m s-1 are truncated,and those less than 0.05 m s-1 are supressedfor clarityof display. Scalearrow is shownat upperleft. (a) February. (b) August. (The color versionand a completedescription of this figure can be found in the separatecolor sectionin this issue.)

In this section, we describethis tropical gyre beg'tuning evidencewherever possible. Becausewe are discussingflows with the westward flow in the SEC, then the western in both northern and southernhemispheres, we will refer to boundaryregion and finally the equatorialcurrents and the the senseof rotation of a flow as either clockwise(cyclonic SECC. The model fields are compared with observational in the southernhemisphere, anticyclonic in the northern) or WOODBERRY ET AL.: SEASONAL CIRCULATION IN THE INDIAN OCEAN 17,991 counterclockwise (anticyclonic in the southern hemisphere, model domain through the southern open boundary, and a cyclonic in the Northern). previously unnamed northward current which we call the Northeast Madagascar Current (NMC). The latitude of this 4.1 The South Equatorial Current splitting is consistentwith the observed split at 17øS from Schott et al. [1988]. Transportsin the EMC range from 9 to The SEC is observed as a broad westward flow between 12 Sv, with a mean of 10.3 (Figure 6). The NMC, with a 8øS and about 20øS, with transportsbetween 33 and 39 Sv mean transport of 0.6 Sv (Figure 7), combines with the [Wyrtki, 1971; Godfrey and Golding, 1981; Pickard and northern branch of the SEC at 13øS. The combined current Emery, 1982]. The model reproduces this flow as a flows around Cape Amber at the northern tip of Madagascar, meandering westward current between 10ø and 20øS, with with mean transport of 16.9 Sv (Figure 8), and then maximumtransport between 12 ø and 15øS (Plate 2). (Plate 2 continues westward to the African coast, where it feeds the is shown here in black and white. The color version can be EACC. A portion of this current (0.5 Sv in the mean; Figure found in the separatecolor sectionin this issue.) East of the 9) flows to the south through the Mozambique Channel Seychelles-MauritiusRidge (SMR), the annualmean westward along the African coast and out the southernboundary, as in transportin the model SEC at 63øE between8øS and 23øS is the atlas of Wyrtki [1971] and in the observations of 24.3 Sv, ranging between 23 and 25.5 Sv (Figure 4). The Lutjeharms et al. [1981]. Weak counterclockwiseeddies form SEC separatesthe clockwise tropical gyre, which is driven in the northern part of the channel, similar to those by the cyclonic (negative) wind stresscurl distributionto the describedby Saetre and da Silva [1984]. north, from a counterclockwisesubtropical gyre to the south, Schott et al. [1988] present transport calculations from which is driven by the anticyclonic (positive) wind stress current meter moorings at 12øS and at 23øS. Swallow et al. curl distribution there. This subtropical gyre straddles the [1988] estimated mean geostrophic transports from southern open boundary of the model, so that only its hydrographicsections in the same regions. The moorings at northern half is represented. The SEC roughly follows the Cape Amber were locatedbetween 1løS and 12øS. Transports line of zero wind stress curl. were calculated parallel to the main current axis of 313øT. The atlas of Wyrtki [1971] shows seasonalvariability in Schott et al. [1988] calculated a value of 10.8 Sv for the top the latitudinal range of the SEC, as pointed out by Schott 200 m, noting that their horizontal length scales do not [1983]. In boreal summer, the SEC is shown between about include all of the current through the region. The model 12ø and 22øS, while in winter it is shown as narrower and transportin theupper 200 m at 49øEbetween 10øS and 12 øS located a few degrees farther to the south, between has a mean value of 16.9 Sv. There is some recirculation approximately15 ø and 20øS. Similar seasonalvariability is included in this transect that occurs to the north of 10øS also seen in the climatology of mixed layer depth of Rao et which is due to clockwise circulation in the lee of the al. [1989]. Inspection of Plates 1 and 2 shows similar Farquhar group of islands and shoals at 10øS, 51øE, to the variability in the model to the east of 60øE. This variability north and east of Cape Amber. Model transport shows a is generatedto the east of 100øE by the annual cycle in the seasonalvariation of 3 to 4 Sv (Figure 8). The transport wind stresscurl and propagateswestward as a Rossbywave at time series from the moorings shows no measurableseasonal approximately0.1 m s-1. Plate3 showsa time-longitudesignal between 150 m and 1100 m depth. Schott et al. section of model ULT through the SEC along 12øS. The [1988] note a negligible seasonalcycle in this current in the westwardpropagation of thesefeatures can be seenfrom their model of Kindle and Thompson [1989]. Swallow et al. origin to the east of 100øE. Similar westward propagating [1988] conclude on the basis of shallow hydrography and features are seen in sea surface topographyfrom GEOSAT historical ship drifts north of Madagascar that there may altimetry data (C. Perigaud, personalcommunication, 1989), indeed be a seasonal variation of +2 Sv with a maximum in and in model simulationsby Perigaud and Delecluse [1989] August-Septemberand a m'mimumin January-February. The and by Kindle and Thompson [1989]. This Rossby wave signal that they see appearsto be trapped in the surfacelayer energy is partially blocked by the banks along the above the moored array analyzed by Schott et al. The model Seychelles-MauritiusRidge at 60øE. In March through June, transect at Cape Amber shows a seasonal cycle with a small clockwiseeddy forms to the east of the SMR between maximum transport of 19 Sv in July-September and 10ø and 12øS, most likely due to nonlinear interactions minimum transportof 14.8 Sv in November-February. among the incoming long Rossby waves and the reflected The current axis for the secondarray of currentmeters was short Rossby waves [see Pedlosky, 1987]. By late July, this 25øT at 23øS across the EMC. The model transect at 22øS eddy has become small enough in diameter that it is advected has a mean transport of 10.3 Sv (Figure 6). Schott et al. throughthe gap in the SMR at 12ø to 13.5øS and continues [1988] and Swallow et al. [1988] find transport values toward the west. This advectiveoscillation appearsto be the throughthis sectionof 7.0 Sv and 7.4 Sv, respectively. The source of the 70-day oscillationsseen in the model transport seasonal cycle from the model bears no resemblance to the at the east coast of Madagascar. seasonal cycle which Schott et al. [1988] describe for 0- to The SEC is diverted by the southernportion of the SMR 1100- m transport for this section. This current passes between 13.5ø and 17øS (the Nazareth Bank and the Cargados through the southernopen boundary of the model and forms Carajos Shoals). In the mean, 10.9 Sv of the flow passes part of the western boundary current for the subtropicalgyre throughthe gap between 12ø and 13.5øS(Figure 5) and on to to the south of 15øS. Since this gyre is not resolved the coast of Madagascar,while a substantialportion flows to completelyby the model, it is not surprisingthat the model the south around the SMR and then toward the west between fails to faithfully reproducethe EMC. 17ø and 19øS. This southernbranch of the SEC splits again The reduced-gravity model of Kindle and Thompson at the east coast of Madagascar into the southward flowing [1989] ( see also Schott et al. [1988]) is also forced by the East Madagascar Current (EMC), which flows out of the Hellerman-Rosensteinwind climatology and producesflows 17,992 WOODBERRYET AL.:SEASONAL CIRCULATION IN THE INDIAN OCEAN

o. 5 M/5

E0

5S

i5S

205

•Og •Sg 50g

"1•0 150 160 ! 70 180 190

O.5

EO

5S

105

15S

55E 60E 65E 70E 75E 80E b :ZOO Z 10 ZZO Z:•O Z•O 250 2:60 Z?O ZSO

Plate 2. Southwesternarea of the model domain (36øE to 80øE, 22øS to 4øN). Arrows are as in Plate 1. (a) January. (b) April. (c)July. (d) October. The Rossbywave between 8ø and 12øSis blocked by the Seychelles-MauritiusRidge in Januarythrough April. An eddy forms to the east of the SMR and is advectedthrough the gap at 11ø to 13øSin early July. Intense eddy activity is seenin the westernboundary current recirculation region to the north of Madagascarand to the west of the Seychellesthroughout the year. Mixed Rossby-gravity(or Yanai) waves are particularly evident in the equatorial region during the monsoontransition months of April and October. (The color version and a complete descriptionof this figure can be found in the separatecolor sectionin this issue.) WOODBERRYET AL.: SEASONALCIRCULATION IN THE INDIAN OCEAN 17,993

o. 5 M/5

EQ

5S

...... ß, .,...'.;:•? ?:m,- • ::JS•;•:.•::•' '"•: 47.•. . ••:'!:.•? '•:',..i.•,..•,½iiœ,.,• ...... • :.:;:;•:•,•::::.

10S ...

-:.: ßB•,=. • .:•* ' ...... f•:•:•?:.-:::•::.--:::::::•:•:•.:.;......

15S

...... ;...... ' •;:e•::- '...... '...... • ;•"•e•:•4.½• .*'• "½•'....'f•E', ..' ...... ':::"::-'• --

gO5 i•:::::::::::::::::::::' '"'•'•:'q•iS½•:?E;?:• ;• ;;': :::j;.... ; ...... ':..'"...... "". :.;.•::::: ...... -.,:::*'*'*:'•:•..:;:,'...... ,.-.•;,•.';5•,•.•,"•"•';. .•',.,-,•::•:.•=. ';• ..;"-• ...... •'•..•,/.•:', ,•......

55E 60E 65E 70E 75E 80E c 190 ZOO Z:10 ZZO 230 Z•O ZSO ZBO ZTO Z80

0.5 H/S OCT [6, TRiO

-'...... *:*- ... . •: •. ..•...... --•:• • • .-• •.• :*•:.:• .... ;•:. ::, ..• .• ..,.• -..• ...:• • •-•,,:: ..• • ...... • ß . . -.•-•.•-:•. • . .":.•.• .•-..• • "..-•.• .:• .• :.:..•.,:•. •: _• • .• : EQ • .....::..:'•..*//%.•,.::•'-• .,':.::.:•-'.;•.:,-"[ ...•.•;;:..."-• .• • .• • •'.:;.-• .• ..• ..• • • ½ . * ..

. .:.-f:A. ß T•;: ,..,- ..... ,-.:-?(..::½• :.•:..- - - :..-:..-.--•.... -:•:.•:•.-...... -:•% :::...... •*-'.,,:-- ....:...... :- ...... :......

55 . '.:•/"• . • '- '•••.•••: - '"- .;• ,,•:*'.:• -.•.: :*:' ,:- .:-,. , ,.:" ..... %.'. -.... ß ..... '-::,...... •. .• -:'• • • •...... • *': .• • .7 ß . .•:-. • -•: :: ';,•.'"I" • ..•--'"•;"' :•* *,•:" -".•' *: *• ? ...... '"'•; :.;.•: ,: C' '•,,•':*:--•" :". '*•;•" :-• *• • * -! ...... :;...... ß ....- .. ß...... :- ß' ...... •::-•,...... •,,.::;,•½-•" * .... '•;•'•.- .::½•?':•,..½,---•:• ...;. '••4':•:..- --.... '" •:'½?•-..-•?•:':'-• '...... :'...... •,•..•*:• ½•":•". ....•-." :"...''", ½•; .. ;:: " ...... '...',-.- *" •...-.•':•'•:•'•*",' ',-• ',-•'•:• . .... *'' '"'½,•,,::::.:-.":'•"•--,:.'-.'• •*-m:.,.:•iq4•"':--..•½•.:•-•:::• :½ 105 ; '"....:...... :..-'L;..:': ::'"-.---,---.. -..-...... ::-.. '*-*' ...... - .•--:::, - •;,...... **-...... ,':..•$• • '::; ..'•:..... ::.:,-,-' • '.':-•.... ":c•,•:.'.::• .,•U ...... c•,-,":.--•, '....'.'•.•* '....:"•. •,.,.' * ..a•**'•'•C, •,,":•,•.. . -*•...... ::•: . *' .•*',,• ' ,.:

: ,,•...... -----:**,,-"-"...... -' ..,:,;:•';.'::.: '";:;::...... ,•?a* .•:",,..• •.','-•...• ...... •::,:•.' ..... •...... %--.-.,-,,,,,,•..•. .: '•.:;'" ...a•-•".:;•/,:.:;....'"...... g•::'.';;::: .::•.'...... ::::':'•...:•,•*•;:.• *::,,...•;•.'.' ...... ::• .....'.*: :::•'". ,..•'... •: "* ,. '.;..•.....;:.•'...... :;..•: ,....•':", 15'5

...:.,,...;¾:,•:::••,• ...... ;.':•',,:-•:"•¾.•:,-:-,'a?, .,., '-:-::,'.:,::-' .:".:•f":-" ' .;-'""% "'"?•'•'""' :•..'-'-,,½•;,7'"'•:,,½• .:,:::•,,•,•?'•-•-,,::---•;.

205

d

Plate 2. (continued) which are very similar to thosedescribed here. Transport of a dragcoefficient of 1.5x 10-3, andthe projection of the values from that model compare well with those calculated wind stressonto a layer with meanthickness of only 200 m over the entire depth (0-1100 m) of the currentmeter array in the presentmodel calculation(as describedin section2) but are larger than those in the presentmodel calculation. accountfor the differencein model transports. Kindle and Thompson [1989] use the full Hellerman- Rosensteinstresses to drive their model, projectedonto an 4.2 East African Coastal Current upper layer that has a mean value of 275 m. The conversion Most of the flow from the SEC reachingthe coast of of the Hellerman-Rosensteinstresses to pseudostress,the use Africa turnsnorthward as a strongwestern boundary current, 17,994 WOODBERRYET AL.: SEASONALCIRCULATION IN THE INDIAN OCEAN

Fig. 4 [Diiing and Schott, 1978; Schott, 1983]. The convergence

-23 between the EACC and the southward Somali Current moves A to about 4øS in February-March with a clockwise eddy to the south and a counterclockwise (anticyclonic) eddy to the north. Although the northeasterliesare relaxing, this strong -24 counterclockwisecirculation remains through most of March before weakening and slowly giving way to northward flow at the coast. By late March, the weakening of the southward winter Somali Current has allowed the EACC to push northward, apparently through the inertial mechanism proposedby Anderson and Moore [1979]. The winds along the coast become southerly by mid-April and then enhance the northward inertial flow of the EACC as in the

-26 i i i i , i . 0 60 120 180 240 300 360 observationsof Leetmaa [1972]. Strong northward flow is

Day found along the African coast south of the equator by June and persiststhrough July and August. Fig. 4. Zonal transport in the SEC across63øE between 8øS and The model solutions (Plates 2a-2d) show strong 23øS (section 2). Time is in days after December 16, year 9. Transportis in sverdrups(1 Sv = 106m 3 s-l). Mean transportis recirculation and eddy formation south of the equator in the 24.3 Sv to the west. Dashed line is instantaneous transport region of the EACC. Comparisonof the mean transportin computed at 6-day intervals Solid line has been smoothedwith a the EACC at 8.2øS (12.8 Sv), into the Mozambique Channel five-point (1 month) filter. (0.5 Sv) and that in the zonal current at Cape Amber (16.9 Sv) suggeststhat 3.6 Sv is recirculatedin the EACC south of that latitude, while comparison of the transport past Cape the East African Coastal Current (Plates 1 and 2). The mean Amber with that farther to the east and south suggestsa transport in the model EACC at 8.2øS, 40 ø to 43øE, is 12.8 recirculation of 5.4 Sv from the north. Maximum eddy Sv to the north, with a range of 8.4 Sv in February to 17.7 kinetic energy (EKE) for the year is found near 3øS, 45øE Sv in May (Figure 10). The EACC is a tropical analog to a (Figure 11). High EKE is found in the EACC south to about mid-latitude western boundary current, driven by the wind 9øS. The strong shearsin the region result in eddies being stresscurl field over the interior of the basin. During the NE formed through barotropic instability. These ringlike eddies monsoon this current flows in conflict with the monsoon circulate in a clockwise fashion in this approximately10 ø by winds. Plate 2a shows the EACC being met by the 10ø region, often merging and reinforcing each other. southward flowing Somali Current early in the year and Counterclockwise eddies are generated just west of the turning offshore at 2ø-3øS in agreement with observations Seychellesnear 5øS. They propagatewestward to the African

36O

33O

3OO

270

2zt0

2t0

t80

iS0

t20

9O

60

30

: .

35E tl0E q•E 511E 55E 60E 65E 70E 75E 80E 85E ':JOE FSU CY205 07/89 t;HE TERS •

t6tt 17lB 176 182 188 lõq: 2"00 206, 212 218 Z2,q: 2:30 236 2q:2:

Plate 3. Time versuslongitude contours of model upper layer thickness(ULT) along 12øS. Time is in days after December16, year 9. The Rossbywave generatedeast of 100øEis blockedby the Seychelles-MauritiusRidge at 60øE An eddy shed throughthe gap in the ridge is seento propagatewestward until it is absorbedinto the northwardcurrent at the east coastof Madagascarat 49øE (the brown vertical strip). (The color versionand a completedescription of this figure can be found in the separatecolor sectionin this issue.) WOODBERRYET AL.: SEASONALCIRCULATION IN THE INDIAN OCEAN 17,995

Fig. 5 Fig. 8

-14

-15 -lO

-16 -11

-17

-12 -18

-13 -19

-20 -14 i i i i i i i i i i 60 120 180 240 300 360 0 60 120 180 240 300 360 Day Day Fig. 5. Zonal transportin the SEC across61øE between 13.4øSand Fig. 8. Transport past Cape Amber across 49øE between 10ø and 11.4øS, through the gap between the Nazareth and Saya de Malha 12øS (section 5) as in Figure 4. Mean transportis 16.9 Sv to the Banks along the Seychelles-MauritiusRidge (section3) as in Figure west. Dashed line is instantaneoustransport computed at 6-day 4. Mean transport is 10.9 Sv to the west. Dashed line is intervals Solid line has been smoothedwith a five-point (1 month) instantaneoustransport computed at 6-day intervals Solid line has filter. been smoothedwith a five-point (1 month) filter.

Fig. 9 Fig. 6

'• -10 •' •' ! '•" 1i '" li 1\ • , -'V -";•. -2 / •øa. ! ...... " ' '"i • i\ / • -11 -4

-12 -6 i i i i i ' 0 60 120 180 240 300 360 -13 o 1,o Day Day Fig. 9. Transport in the Mozambique Channel across 12øS between Fig. 6. Transport in the EMC across22øS between 48ø and 50øE 41ø and 48øE (section 10) as in Figure 4. Mean transportis 0.5 Sv (section4) as in Figure 4. Mean transportis 10.3 Sv to the south. to the south. Dashed line is instantaneoustransport computedat 6- Dashed line is instantaneoustransport computed at 6-day intervals day intervals Solid line has been smoothed with a five-point (1 Solid line has been smoothedwith a five-point (1 month) filter. month) filter.

Fig. 7 Fig. 10

'• 14

i- 12

V 'J -4 ! , i ß ß i i i i i 0 60 1;•0 180 2•,0 300 360 60 120 180 240 300 360 Day Day

Fig. 7. Transportacross 15øS between 50 ø and 53øE (section11) as Fig. 10. Transportin the EACC across8.2øS between 40 ø and 43øE in Figure 4. Mean transportis 0.6 Sv to the north. Dashed line is (section6) as in Figure 4. Mean transportis 12.8 Sv to the north. instantaneoustransport computed at 6-day intervals Solid line has Dashed line is instantaneoustransport computed at 6-day intervals been smoothedwith a five-point (1 month) filter. Solid line has been smoothedwith a five-point (1 month) filter. 17,996 WOODBERRYET AL.: SEASONALCIRCULATION IN THE INDIAN OCEAN

- o

A - 55 o lO

- 105

- 155 \¾ L.,:

-lO i i i ß I - 205 60 120 180 2•0 300 360 1 Day •OE qSE 50E 55E 60E 8 E 7 E 75E 80E

Fig. 11. Mean annual eddy kinetic energy (EKE) in the equatorial region near the African coast, 40 ø - 80øE, 5øN - 12øS. EKE is concentratedat the coast south of the equator with a maximum of A 280x 10-4 m2 s-2 near3øS. Zonalrelative maxima are separated by approximately250 km. Labelsare in unitsof 10-4 m2 s-2 with a contour interval of 1.0 x 10-3 m 2 s-2. o lO coast, cutting through the clockwise circulations, and are absorbedinto the EACC. Beginning in the summer, these eddiesform west of the Seychellesat 40- to 60-day intervals throughthe remainderof the year. Kindle and Thompson

[1989] report eddy formation in this region and in this -lO period band in a similar model of the Indian Ocean. 60i 1 •l0 180i 240i 300i 360 A strong clockwiseeddy is found throughoutthe year to Day the south of the separation of the EACC from the coast. Fig. 12. Transportin the Somali Current as in Figure 4, (a) between During the SW monsoon,this eddy oscillateson a clockwise 46.4øE, 2.6øN and 48øE, IøN (section7; mean transportis 0.8 Sv to the north)and (b) between 48.6øE, 5N and 49.6øE, 4øN (section 8; path which edges acrossthe equator. Not all of the EACC mean transport is 2.6 Sv to the north). Dashed line is instantaneous flow recirculates in this eddy; outflow from this eddy transportcomputed at 6-day intervals. Solid line has been smoothed continues northward in the Somali Current which at 4 ø to 5øN with a five-point (1 month) filter. reaches a peak transportof 20.5 Sv in August (Figure 12). This eddy is interpretedas the southerngyre of the so-called two-gyre systemobserved along the Somali coast during the migration of the southerngyre. The southernhemisphere SW monsoon [Bruce, 1973; Swallow and Fieux, 1982; winds were anomalouslystrong in those 7 years, driving a Swallow et al., 1983; Schott, 1983; Luther and O'Brien, stronger EACC transport; however, that simulation did not 1985; Luther et al., 1985]. Salinity observationsearly in include fully the effects of the southernhemisphere trade the SW monsoon show the water from the EACC to winds, which also affect the transport in the EACC. recirculatesouth of the equator,while the equatorialcurrents Andersonand Moore [1979] find that the separationlatitude are fed by water from north of the southerngyre [Swallow et for a cross-equatorialinertial jet movescloser to the equator al., 1983]. The clockwise great whirl forms between4 ø and as the transportof the jet increases. This is supportedby 6øN , while the southern gyre is observedsouth of 2ø-3øN. the resultsof the 23-year simulation. The formation of the southern gyre may be caused by The climatological wind forcing used in this simulation alongshore winds near the equator without requiring the favors the blocking situation. The clockwise eddy in the forcing of the SE trades, as shown by McCreary and Kundu EACC separation moves northward as the $W monsoon winds strengthenin May and June to become the southern [1988], but Anderson and Moore [1979] point out that the gyre. As it approachesthe equatorit has the wrong sense curl of the wind near the equator can be important in of rotationfor the northernhemisphere. A counterclockwise determiningthe latitude at which the current tums offshore. eddy forms on the other side of the equator through the Observationsshow the southerngyre to leave the equator, offshore advection of boundary layer potential vorticity move northward and coalesce with the great whirl in late aroundthe southerngyre (Plate 2c). Both eddiesare cyclonic August and September [Bruce, 1973; Brown et al., 1980; in their respectivehemispheres. The net effect of this two Evans and Brown, 1981; Schott, 1983). McCreary and eddy systemis that the northern eddy blocks the northward Kundu [1988] conclude from their model results that the movementof the southerngyre as in the mechanismof Cox migration of the southern gyre is due to the ocean's not [1979]. In this model simulation,the cross-equatorialeddy being in equilibrium with peak monsoon winds. This pair begins to move south with the transition to the NE migration has been simulatedwithout the SE trades [Luther et monsoon in October-November and southward flow across al., 1985] and was seen in 14 years of a 23-year simulation the equatoris reestablished(Plate 2d). forced by the winds of Cadet and Diehl [1984]. In the other 7 years, during which the southern gyre and great whirl were 4.3 Equatorial Currents clearly present, the transport in the EACC was higher than that in years with migration, and a counterclockwise eddy The model equatorialcurrents reverse four timesduring the developednorth of the equatorwhich preventedthe northward year, producingthe eastwardequatorial jets as documentedby WOODBERRYET AL.: SEASONALCIRCULATION IN THE INDIAN OCEAN 17,997

Wyrtki [1973]. He noted the appearanceof the jet during 12øS. In the western portion of the basin, eastward return monsoon transitions in April-May and in September- flow to the SECC meanders around the banks and islands Octoberwith strongestflow between60øE and 90øE. O'Brien south of about 4øS when equatorial flow is westward (Plate and Hurlburr [1974] showed that these jets were transient 2a). When equatorial currentsare eastward,they also feed responsesto the onset of along-equatorwinds. Except in into the SECC. There are two main tributaries to the SECC. August, the reversals in the model equatorial currents The most continuous flow is north of the Seychelles and propagate from east to west as reported from buoy ChagosArchipelago which feeds the SECC farther to the east trajectories by Reverdin et al. [1983] and Reverdin and near 80øE. This tributary is stronger during the SW Luyten [1986] and from a currentmeter array at 40øE to 62øE monsoonthan during the NE monsoon, as it receives input by Luyten and Roeromich[1982]. The following chronology from the equatorial and Arabian Sea regions during the refers to flow east of about55øE, away from the eddy activity (northern)summer months. Plate 2 showsa secondtributary near the coast. West of the Maidives and Gan (73.5øE), to the south of the Seychellesand the Chagos Archipelago. model geometryincludes numerous islands which affect the This flow feedsthe SECC along the northernedge of Saya de flow; this area is more active in terms of eddies and Malha Bank and is modulated by the annual Rossby wave oscillations. Because of these natural distinctions, the train. The flow from the SECC into the SEC occurs in Maldive longitude will be used to split the basin into west westwardmigrating bands to the east of the lows associated and east for the following discussion. with the annual Rossby wave train. In December and January the flow west of the Maidives and Gan is westwardwith weak flow in the east. In February, 4.5. Other Currents weak equatorial flow is found across the basin (Plate 1). Although the major currentsof the Indian Ocean basin are Westerly winds in March begin to drive eastwardflow in the reproduced by the model, in localized regions the model eastern basin, while outflow from the EACC into the SECC circulationsdo not agree well with observations. Along the begins to move toward the equator in the west. Eastward western Australian coast, the model produces only a weak flow is seen between 60 ø and 80øE in mid-April, with weak, poleward flowing Leeuwin Current [Godfrey and Ridgway, confusedflow farther to the east. The meanderingSECC is 1985; Thompson, 1984] during December through June. locatedjust southof the equatorto the west of 60øE, where it This feature propagates offshore as a Rossby wave and is remains through May. By mid-June, westward flow prevails replaced by equatorward flow in August through October. east of 65øE, while eastward flow is found near the western This is most likely a limitation of the model, which lacks boundary, fed by the meanderingoutflow from the EACC. thermodynamicforcing, throughflow from the Pacific, and By mid-July, westwardflow is seen acrossmost of the basin, vertical resolution. Studies of the Leeuwin Current have with strongestflow between 60ø and 80øE. Equatorial flow looked at the roles of thermohalineforcing [McCreary et al., becomesweak and confusedby mid-August acrossthe basin. 1986; Weaver and Middleton, 1989] and remote forcing by Eastward flow begins near the western boundary in late Pacific throughflow[Godfrey and Golding, 1981; Godfrey and August and early September,again in the meanderingoutflow Ridgway, 1985; Kundu and McCreary, 1986], both of which from the EACC, and progresseseastward, covering the entire appear to play a role but are missing from this model. The equatorial region by mid-October. During this part of the model also simulates only a weak seasonallyreversing South year, the eastward equatorial jet forms a part of the SECC. Java Current [Wyrtki, 1961] and likely misses other Westward flow again appears first in the east during localized circulations as well. In all of these regions, other November and covers the basin to 55øE in December. forcing mechanismsin addition to the wind forcing used in The model response in the eastern basin is asymmetrical this study are likely of importance. between the northern and southern hemispheres due to the asymmetryof the Indonesian coast. The eastwardjets are 5. DISCUSSION associatedwith downwelling Kelvin waves at the equator which impinge on the NW-SE slanted boundary. The The southern hemisphere gyre is the response of the ocean to the mean basin-wide wind stress curl distribution reflected equatorialRossby waves are asymmetricaland their effects on the near-eastern boundary region will be and exhibits most of the features of a classical mid-latitude asymmetrical. In numerical experiments involving the gyre, modified by its proximity to the equatorial waveguide reflection of Kelvin waves at an asymmetrical eastern and by the large seasonalvariability in the wind fields. The boundary,the reflectedRossby waves become symmetrical as forcing over the ocean interior is therefore an important they move into the ocean interior (T. Jensen, personal determinant of the western boundary flow. The negative communication, 1988). Part of the incoming energy (cyclonic) wind stress curl over the interior of the southern propagatespoleward as coastal Kelvin waves which is also hemisphere tropical gyre causes upward Ekman pumping, an asymmetric response. The southern hemispherecoastal stretching vortex tubes and requiring that they move Kelvin wavesPropagate out the easternopen boundary. The poleward to conserve potential vorticity as in classical northern hemispherecoastal Kelvin waves follow the closed Sverdrup dynamics. Equatorward flow at the western coastline of the Bay of Bengal, round the southern tip of boundaryprovides for mass conservationand restorationof India, and enter the Arabian Sea. potential vorticity to match that of the interior flow. The recirculation region near the western boundary south of the equator, like that associatedwith the Gulf Stream, provides 4.4 SouthEquatorial Countercurrent additional time in the boundary current for fluid parcels to The South Equatorial Countercurrentis not seen as a acquire the necessary potential vorticity to reenter the continuous current across the basin in the model. Rather, it interior flow. has several tributaries and is affected by the reversing The trade winds of the Indian Ocean are significantly equatorialcurrents and by the train of Rossbywaves at 6ø- different from those of the Atlantic and Pacific. The 17,998 WOODBERRYET AL.: SEASONAL CIRCULATION IN THEINDIAN OCEAN southernhemisphere southeasterly trades are found farther Sverdrup flow is generally broad and toward the west- south than in the other oceansand exhibit a strong annual northwest across the the basin from 25øS to 10øS. With the cycle. Trade winds are found in the northernhemisphere throughflowimposed, the Sverdrupcalculation yields a more only during the NE monsoon, and the reversals of the narrow, zonal flow for the SEC, as in observed dynamic monsoon winds have no counterpart in areal extent or height from Wyrtki [197I]. In our model simulation, the intensityover other oceans. In comparison,the Pacific and SEC appearsas a meandering,zonal current at the observed Atlantic are relatively symmetricallyforced north and south latitude, without external imposition of an Indo-Pacific of the equator, although there is also some seasonal throughflow. The only throughflowin the model is that of variability in their trades. These oceansdisplay major trade 2 Sv through the open boundary required by the internal wind-driven gyres in both hemispheres. There is a large Indian Ocean dynamics. The Sverdrupcalculation of course gyre in the southerntropical Indian Ocean but no counterpart ignoresthe time dependentcomponent of the flow field. In in the northern hemisphere. During the SW monsoon,this fact, Godfrey and Golding [1981] attribute differences gyre extendsacross the equator,encompassing the flow from between their computed mass transport function and the the SW Monsoon Current in the Arabian Sea. observeddynamic topography to eddiesin the SEC. They also attributenoise in the hydrographicdata along 110øE to 5.1 lndo-Pacific Throughflow eddy activity. Their Sverdruptransport was also computed on a coarse grid (5ø in both directions). The implication At the easternboundary of the model, the mean transport here is that the structureof the SEC doesnot dependcrucially is 2 Sv westward into the Indian Ocean. This is the on the imposition of an Indo-Pacific throughflow,although transport required by internal Indian Ocean dynamics as details of the flow in the eastern basin most likely are forced by the winds over the ocean basin. The westward affected by the throughflow. direction of the observed throughflow indicates a gradient between the Pacific and Indian Ocean basins [Wyrtki, 1987] which is not simulated in this model, since the model 5.2 Annual Signal considers the wind-forced Indian Ocean alone. Estimates The seasonalcycle in the southernhemisphere trades from various observationalstudies yield an averagewestward generatesan annual Rossby wave in the ocean to the east of throughflowof 9.2 Sv [Gordon, 1986]. Although the model 100øE (Figure 4). The effects of this annual Rossby wave underestimatesthe volume of the throughflow, the seasonal signal are seen in the transport of the SEC, in the flow cycle seen in the model transport (Figure 13) is in phase through the eastern boundary, and in the flow patterns off with that seen in ocean general circulation models (GCM) northwest Australia. Although partially blocked by the which include both Indian and Pacific Oceans [Kindle et al., Seychelles-MauritiusRidge, this Rossby wave is responsible 1987; Schott et al., 1988]. Westward flow is maximum in for the seasonalcycles in the transportof the EMC and NMC July through September, reaching values of 8 Sv, with along the coast of Madagascar. An eddy formed to the east minimum values of less than 0.1 Sv in December and January of this ridge through nonlinear interactionsin the reflected and a brief burst of eastwardtransport in October. Rossby wave packet breaks through the gap in the ridge as it Godfrey and Golding [1981] calculatethe mass transport decays, causing 70-day period oscillations at the northeast stream function from the Sverdrup relation from the equator coast of Madagascar. to 40øS, with and without an observed value of Indo-Pacific throughflowimposed at the easternboundary. They find that the inclusion of a throughflow from the Pacific into the 5.3 50-Day Oscillations Indian Ocean on the order of 10 Sv is essential to obtain a Several studies have noted a low-frequency variability in realistic South Equatorial Current. Without the impositionof the currents of the western Indian Ocean. Mysak and Mertz observed flow at the eastern boundary, they find that the [1984] found a 40- to 60-day oscillation in the longshore currents at the African coast between the equator and 5øS. Quadfaseland Swallow [1986] reported50-day oscillationsin currentmeter recordsoff the northerntip of Madagascarand similar oscillations in surface currents along a transect at 11øS west of Madagascar. It was suggestedthat these oceanic oscillationswere being forced by the 40- to 50-day oscillation reported in the tropical atmosphereby Madden and Julian [1972] and found in winds over the westernIndian Ocean in 1976 and 1979 by Mertz and Mysak [1984]. Current meter data reported by Schott et al. [1988] show transportvariations in the 40- to 55-day period band, which accountedfor over 40% of the total transport variance. The oscillations were also present in the ship measurementsof -8 Swallow et al. [1988]. 0 3•0 360 Day In agreementwith observations,but in conflict with the suggested forcing mechanism, the model fields contain Fig. 13. Transportthrough the easternopen boundaryacross 115øE oscillations in the same 40- to 50-day period band, even between9 ø and 20øS (section 1) as in Figure 4. Mean transportis 2 thoughmean monthly winds are used as forcing. The region Sv to the west. Dashed line is instantaneoustransport computed at 6-day intervals. Solid line has been smoothedwith a five-point (1 of eddy activity to the north of Madagascarand along the month) filter. east African coast is dominated by variability in the 40- to WOODBERRY ET AL.: SEASONAL CIRCULATION IN THE INDIAN OCEAN 17,999

50-day period band (Figures8, 10, and 12). The EACC has It is still unclear why the period of 50 days is especiallyprominent oscillations with a period of about50 preferentiallyselected. It may be that 40- to 50-daysis a days(Figure 10). Oscillationsin this periodrange are found natural period of oscillationof the system,one that is not in the model solutionsof Kindle and Thompson [1989] and necessarilylimited to the surfacecurrents as in the model. It are in agreementwith observations[Schott et al., 1988]. is interesting to note that the core of the Equatorial With 60 days as the shortestperiod resolved in the wind Undercurrentin the Pacific oscillatesabout the equator with a forcing, the model and observationalvariability can not be periodof about40 days [Moore and Philander, 1977] and due to atmosphericforcing as has been suggestedby others, that 50-day oscillations are seen in the tropical Brazil since the model has no forcing at those periods. The Current (T. Lee and W. Johns, personal communication, presenceof oscillationsin the model responsewith periods 1988). It appearsthat oscillationsin this period band are shorter than 60 days requires another mechanism. These ubiquitousin the tropical oceans. It is possiblethat these model results support the conclusions of Kindle and oceanic oscillations force those in the atmosphere, rather Thompson [1989] and of Schott et al. [1988] that the than the converse; indeed, Madden and Julian [1972] oscillations are due to internal instabilities in the ocean. postulatethat the 30- to 60-day atmosphericoscillations Horizontal shear (barotropic) instability is the only may be generatedsomewhere over the Indian Ocean through dynamicalinstability mechanismin this model. The strong convection processes. We leave this for further horizontalshears throughout the region lead to the formation investigations. of eddiesthrough this mechanismat a period of 40- to 50 days. 5.4 28-Day Oscillations These oscillations are absent from the zonal flow in the SEC farther to the east. Sections across the SEC at 63øE, in Linear wave theory identifies several discrete modes of the gap in the Seychelles-MauritiusRidge, and along the east oscillationat the equator [Moore and Philander, 1977]. The coast of Madagascar show only annual and semiannual dispersion relation derived from the linear momentum oscillations(Figures 4, 5 and 7). Following the first split equationshas solutions which correspondto high-frequency of the SEC at 60øE, both the northern and southern branches gravity waves, low-frequency Rossby waves, equatorially of the flow contain meridional oscillations associated with trappedKelvin waves, and the specialcase of mixed Rossby- the annual Rossby wave. The 70-day oscillationsin the gravity (Yanai) waves. These mixed waves behave like meridional transportsof the EMC and in the zonal transport Rossby waves at low frequency and like gravity waves at pastCape Amber are advectiveeffects due to the eddy that is high frequency. Only Kelvin waves and mixed Rossby- trappedto the east of the Seychelles-MauritiusRidge and is gravity waves are allowed at periods between about a week shedthrough the gap in the ridge at 60øE, 10ø-12øE. Kindle and about a month [McPhaden, 1982; Moore and McCreary, found transport variations with periods of 70-90 days 1989]. associatedwith westward eddy propagationpast Cape Amber Plate 4 shows a time-longitude section of meridional [Schott et al., 1988]. transportacross the equator. Waves are generatedseveral

360

330

3OO

Z70

Zz}O

ZlO

180

150

IZO

90

60

30

50E 55E 60E 65E ?OE • eOE 6• 90E

...... "'••:?•.•....•.g?:•}:•:•:•?•½•. •:?.:•.i•i:•.:!!?•..:-:•-• ...... : ...... -70 -60 -50 -•0 -30 -ZO -10 O 10 •O 30 •0 50

Hate 4. Time versuslongitude contours of mcridionaltransport across the equator.Time is in daysafter December 16, year 9. Packetsof waveenergy with westwardphase and eastwardgroup propagation can be seenemanating from the westernboundary region. Thesewaves have a periodof 28 daysand are interpretedas mixedRossby-gravity (Yanai) wavesgenerated by instabilitiesin the westernboundary region. Transport is in sverdrups(1 Sv = 106m 3 s'l). Contourinterval is 5 Sv. (Thecolor version and a completedescription of thisfigure can be foundin the separatecolor section in this issue.) 18,000 WOODBERRYET AL.:SEASONAL CIRCULATION IN THEINDIAN OCEAN times during the year near the coast out of the eddy activity current of this tropical gyre. Outflow from this region which has been described there. The westward phase speed meanderseastward through various tributaries to the SECC (about20 cm s'l) andeastward group velocity (about 24 cm and then into the Sverdruplike interior, where it returns to s'l) of thesewaves is evident.The oscillationsare visible the SEC. in the velocity fields of the model along the African coast A Rossbywave is generatedannually in the easternbasin and along the equator(Figure 12; Plate 2). Zonal wavelength and propagateswestward in the shear zone betweenthe SEC is 500-650 kin, increasing slightly in the east. and SECC. The annual signal of this wave is found in Measurementseast of 60øE are indicative of mixed Rossby- branchesof the SEC through the banks at 60øE and in the gravity waves. The waves decay eastwardand are partially currentsalong the east coast of Madagscar. The Seychelles- blocked by the Maldive Islands at 73ø-74øE, but are seen Mauritius Ridge causespartial reflectionof the Rossbywave, across the basin. and an eddy forms due to nonlinear effects in the reflected The dominant period in meridional velocity at 55øE is wave packet. The eddy moves through the break in the found by Fourier analysisto be 40 to 50 days with a second banks early in the year and continues westward into the peak near 28 days. West of 55øE the semiannualsignal active western boundary region. becomes very strong, but the same two peaks are still The seasonalcycle of the monsoonsis most evident in visible at 50øE. The dominant period is 33 days at 60øE the western boundary region and along the equator. Strong with a smallerpeak near 26 days. At 65øE and 70øE there is horizontal shearsnear the African coast between Madagascar a single peak at 28 days, suggestinga mixed Rossby-gravity and the equator cause the formation of clockwise and wave. Luyten and Roemmich [1982] observed a dominant counterclockwise eddies. The Somali Current is southward 26-day period in a compositemeridional velocity at 200 m during the NE monsoon, meeting the EACC and turning from moored current meters between 47øE and 62øE. offshore at 2ø-3øS. The clockwise circulation to the south of Meandersin the equatorialjet with a period of 25 days were this offshore flow develops into the "southern gyre" reportedfrom buoy trajectoriesby Reverdin et al. [1983] and observed during the SW monsoon when currents are Reverdin and Luyten [1986]. These oscillations are northward along the coast. The gyre is blocked from its attributed to mixed Rossby-gravity waves. O'Neill [1984] often observed northward migration by a counterclockwise also found characteristicsof mixed Rossby-gravitywaves in eddy which forms in the northern hemisphere. The an array of moorings at 53øE. O'Neill notes that incoming counterclockwise circulation contributes to the southward energy is likely to be reflected as mixed Rossby-gravity reversal of the Somali Current with the return of the NW waves owing to the NE-SW angle of the African coast at the monsoon. equator. Kindle and Thompson [1989] report 27-day Equatorial currents reverse four times annually, with equatorial oscillations being generated in their model by the westward flow found near the height of both monsoonsand "instability of the eastward flow" away from the coast. eastward transport found during monsoon transitions. Moore and McCreary [1989] show, for the stratification of Maximum zonal velocities are found in the western basin the Indian Ocean, that only Kelvin and mixed Rossby with reversals starting generally in the east. Eastward gravity waves are generatedat a western equatorialboundary equatorial currents contribute to the SECC which is fed by for periods less than 30 days, while for periodsnear 60 days, several tributaries around the islands in the domain. short Rossbywaves in addition to these waves can propagate We have noted regions where the model has some energy eastward. The present model results indicate the difficulty, primarily at the easternboundary in regard to the presenceof only mixed Rossby-gravitywaves at 26- to 28- Leeuwin Current, the Java Current, and the Pacific day periods. The energy at 40 to 50 days appearsas a result throughflow. Additional forcing mechanismsappear to be of the eddy activity to the south, as no eastwardpropagation required for accurate modelling of these circulations. of this signal is found. Inclusion of throughflow from the Pacific is not required in order to simulate the SEC-SECC systemin agreementwith its observed location.

6. SUMMAlaY • CONCLUSIONS The model results are limited by the accuracyof the wind forcing, as wind observationsfrom the southernhemisphere A numerical model of the Indian Ocean, driven by the are very limited. The open boundary conditions are also a climatological monthly mean winds of Hellerman and problem, as flow throughthe boundariesis determinedsolely Rosenstein [1983], simulatesthe major featuresof the large- by physics interior to the model domain in a region where scale upper ocean circulationin the southernhemisphere and there is a significantinput from the south and the east. The equatorial regions. Major currents (SEC, SECC, EACC, placementof the southernboundary at 25øS may be a factor Somali Current) are found in their observedlocations. Mean in the simulatedcirculation to the east of Madagascarand in transports in the western basin are comparable to the Mozambique Channel, as no continuity of flow is observations. The principal feature of the circulationin the allowed aroundthe southerntip of Madagascarin the model. southernhemisphere tropical Indian Ocean is a basin-wide It is possibleto reproducemany of the large-scalefeatures clockwise(cyclonic) gyre comprisedof the SEC in the south, of the southernIndian Ocean with wind forcing alone. The the SECC to the north, and the EACC in the west. The results of this study provide encouragementfor further model westernboundary region of this tropical gyre is a sourceof development. Verification of improvements will require energetic eddy activity that is generated through shear increasesin observationaldata, while improved models will instabilities, and it is a tropical analog of a western enable better interpretation of the available data. A boundarycurrent recirculationregion of a subtropicalgyre. combined effort of modellets and observationalists is The southerngyre of the (northern)summer Somali Current therefore required for further understandingof the circulation is seen as a northward extension of the western boundary of the southerntropical Indian Ocean. WOODBERRY ET AL.: SEASONAL CIRCULATION IN THE INDIAN OCEAN 18,001

Acknowledgments. The authors express their appreciation to Knox, R. A., On a long series of measurement of Indian Ocean David M. Leglet, Alan C. Davis and James D. Merritt for their equatorialcurrents near Addu Atoll, Deep Sea Res., 23, 211-221, computationalexpertise, and to Rita Kuj)per for documentprocessing 1976. assistance. Thanks also to Jon Alquist, Sharon Nicholson, Klaus Knox, R. A., The Indian Ocean: Interaction with the monsoons, in Wyrtki and John Kindle, as well as two anonymousreviewers, for Monsoons, editedby J. Fein and P. Stephens,pp. 365-398, John their insightful commentsand suggestions. This work was partially Wiley, New York, 1987. supportedby a NASA Traineeship in Physical Oceanography and Knox, R. A., and D. L. T. Anderson, Recent advancesin the study of Meteorology. Additional support was provided by the Office of the low-latitudeocean circulation,Prog. Oceanogr.,14, 259-318, Naval Research,the Institute for Naval Oceanography,and the Naval 1985. Ocean Research and Development Activity. The wind analysis is Kundu, P. K., and J.P. McCreary, Jr., On the dynamics of the supportedby the NOAA TOGA Project Office. Partial supportwas throughflow from the Pacific into the Indian Ocean, J. Phys. provided by the Florida State University through time granted on its Oceanogr., 16, 2191-2198, 1986. Cyber 205 supercomputer.This is contributionnumber 284 of the Leetmaa, A., The response of the Somali Current to the southwest Geophysical Fluid Dynamics Institute and number 89-100 of the monsoonof 1970, Deep Sea Res., 20, 319-325, 1972. SupercomputerComputations Research Institute of the Florida State Legler, D. M., I. M. Navon, and J. J. O'Brien, Objective analysisof University. pseudostressover the Indian Ocean using a direct-minimization approach,Mon. Weather Rev., 117, 709-720, 1989. REFERENCES Luther, M. E., Indian Ocean modelling, in Further Progress in Equatorial Oceanography,edited by E. Katz andJ. Witte, pp. 303- Anderson, D. L. T., and D. W. Moore, Cross-equatorialinertial jets 316, Nova University Press, Dania, Fla., 1987. with special relevance to very remote forcing of the Somali Luther, M. E., and J. J. O'Brien, Modelling the variability in the Current,Deep Sea Res.,26, 1-22, 1979. Somali Current, in Mesoscale/Synoptic Coherent Structures in Barnett, T., Interaction of the monsoon and Pacific trade wind Geophysical Turbulence, edited by J. C. J. Nihoul and B. M. systemat interannualtime scales,I, The equatorialzone, Mon. Jamart, pp. 373-386, Elsevier, New York, 1989. Weather Rev., 111, 756-773, 1983. Luther, M. E., and J. J. O'Brien, A model of the seasonal circulation Brown, O. B., J. G. Bruce, and R. H. Evans, Evolution of sea surface in the Arabian Sea forced by observed winds, Prog. Oceanogr., temperaturein the Somali Basin during the southwestmonsoon of 14, 353-385, 1985. 1979, Science, 209, 595-597, 1980. Luther, M. E., J. J. O'Brien, and A. H. Meng, Morphology of the Bruce, J. G., Large-scalevariations of the Somali Current during the Somali Current systemduring the southwestmonsoon, in Coupled southwestmonsoon, 1970, Deep Sea Res., 20, 837-846, 1973. Ocean-AtmosphereModels, edited by J. C. J. Nihoul, pp. 405- Cadet, D. L., and B.C. Diehl, Interannual variability of surface 437, Elsevier, New York, 1985. fields over the Indian Ocean during recent decades,Mon. Weather Lutjeharms, J. R., N. D. Bang, and C. P. Dugan, Characteristicsof Rev., 112, 1921-1935, 1984. the currents east and south of Madagascar, Deep Sea Res., 28, Camerlengo,A. L., and J. J. O'Brien, Open boundary conditionsin 879-900, 1981. rotatingfluids, J. Comput.Phys., 35, 12-35, 1980. Luyten, J. R., and D. H. Roemmich, Equatorial currents at semi- Cox, M.D., A numerical study of Somali Current eddies, J. Phys. annual period in the Indian Ocean, J. Phys. Oceanogr., 12, 406- Oceanogr., 9, 311-326, 1979. 413, 1982. Cutler, A. N., and J. C. Swallow, Surface currents of the Indian Madden, R. A., and P. R. Julian, Description of global-scale Ocean (to 25øS, 100øE): compiledfrom historicaldata archivedby circulationcells in the tropics with a 40-50 day period, J. Atmos. the Meteorological Office, Bracknell, UK, Rep. 187, 8pp, 36 Sci., 29, 1109-1123, 1972. charts, Inst. of Oceanogr.Sci., Wormley, England, 1984. McCreary, J.P., and P. K. Kundu, A numerical investigationof the D/iing, W., The Monsoon Regime of the Currents in the Indian Somali Current during the Southwest Monsoon, J. Mar. Res., 46, Ocean, 68 pp., East-WestCenter Press, Honolulu, Hawaii, 1970. 25-58, 1988. D/iing, W., and F. Schott, Measurementsin the sourceregion of the McCreary, J.P., S. R. Sherye, and P. K. Kundu, Thermohaline Somali Current during the monsoonreversal, J. Phys. Oceanogr., forcing of eastern boundary currents: With application to the 8, 278-289, 1978. circulation off the west coast of Austraha, J. Mar. Res., 44, 71- Evans, R. H., and O. B. Brown, Propagationof thermal fronts in the 92, 1986.•' Somali Current system,Deep Sea Res., 28, 521-527, 1981. McPhaden, M. J., Variability in the central equatorial Indian Ocean, Fine, R., Direct evidenceusing tritium data for throughflow from the I, Ocean dynamics, J. Mar. Res., 40, 157-176, 1982. Pacific into the Indian Ocean, Nature, 315, 478-480, 1985. Mertz, G. J., and L. A. Mysak, Evidencefor a 40-60 day oscillation Findlater, J., Mean monthly airflow at low levels over the western over the westem Indian Ocean during 1976 and 1979, Mon. Indian Ocean, Geophys. Mem., 115, 53 pp., H. M. S. O., Weather Rev., 112, 383-386, 1984. London, 1971. Moore, D. W., and J.P. McCreary, Excitation of intermediate- Godfrey,J. S., and T. J. Golding, The Sverdruprelation in the Indian frequency equatorial waves at a western boundary: With Ocean, and the effect of Pacific-Indian Ocean throughflow on application to observationsfrom the Indian Ocean, J. Geophys. Indian Ocean circulation and on the East Australian Current, J. Res., in press, 1989. Phys. Oceanogr., 11, 771-779, 1981. Moore, D. W., and S. G. H. Philander, Modeling of the tropical Godfrey, J. S., and K. R. Ridgway, The large scale environmentof oceanic circulation, in The Sea: Ideas and Observations on the poleward-flowing Leeuwin Current, Western Australia: Progress in the Study of the Seas, vol. 6, edited by E. Goldberg, Longshore steric height gradients, wind stress and geostrophic I. McCave, J. O'Brien, and J. Steele, pp. 319-361, John Wiley, flow, J. Phys. Oceanogr., 15, 481-495, 1985. New York, 1977. Gordon, A. L., Interocean exchange of thermocline water, J. Mysak, L. A., and G. J. Menz, A 40- to 60-day oscillationin the Geophys.Res., 91, 5037-5046, 1986. source region of the Somah Current during 1976, J. Geophys. Hastenrath,S., and P. J. Lamb, Climate Atlas of the Indian Ocean, I, Res., 89, 711-715, 1984. Surface Climate and AtmosphericCirculation, 19 pp., 97 charts, Nicholls, N., The Southern Oscillation and Indonesian sea surface University of Wisconsin Press, Madison, 1979. temperatures, Mon. Weather Rev., 112, 424-432, 1984. Hellerman, S., and M. Rosenstein,Normal monthly wind stressover O'Brien, J. J., and H. E. Hurlburr, Equatorialjet in the Indian Ocean: the world ocean with error estimates, J. Phys. Oceanogr., 13, Theory, Science, 184, 1075-1077, 1974. 1093-1104, 1983. O'Neill, K., Equatorialvelocity profiles, I, Meridional component,J. Kindle, J. C., and J. D. Thompson,The 26- and 50-day oscillations Phys. Oceanogr., 14, 1829-1841, 1984. in the westem Indian Ocean: Model results,J. Geophys.Res., 94, Pedlosky, J., Geophysical Fluid Dynamics, 624 pp., Springer- 4721-4736, 1989. Verlag, New York, 1987. Kindle, J. C., G. W. Hebum, and R. C. Rhodes, An estimate of the Perigaud, C., and P. Delecluse, Simulationsof dynamic topography Pacific to Indian Ocean throughflow from a global numerical in the northwesternIndian Ocean with input of Seasat altimeter model,in Further Progress in Equatorial Oceanography,edited by and scatterometerdata, Ocean-Air Interactions, 1, 289-309, 1989. E. Katz and J. Witte, pp. 317-322, Nova Univiversity Press, Pickard, G. L., and W. J. Emery, Descriptive Physical Dania, Fla., 1987. Oceanography, 249 pp., Pergamon,New York, 1982. 18,002 WOODBERRYET AL.: SEASONALCIRCULATION IN THE INDIAN OCEAN

Piola, A. R., and A. L. Gordon, Pacific and Indian Ocean upper layer distribution in the Somali Basin in response to the southwest salinity budget, J. Phys. Oceanogr., 14, 747-753, 1984. monsoonof 1979, J. Phys. Oceanogr., 13, 1398-1415, 1983. Quadfasel, D. R., and J. C. Swallow, Evidence for 50-day period Swallow, J. C., M. Fietax, and F. Schott, The boundarycurrents east planetary waves in the South Equatorial Current of the Indian and north of Madagascar, 1, Geostrophiccurrents and transports, Ocean, Deep Sea Res., 33, 1307-1312, 1986. J. Geophys.Res., 93, 4951-4962, 1988. Rao, R. R. , R. L Molinari, and J. F. Festa, Evolution of the Thompson,R. O. R. Y., Observationsof the Leeuwin Current off climatologicalnear-surface thermal structureof the tropical Indian western Australia, J. Phys. Oceanogr., 14, 623-628, 1984. Ocean, 1, Description of mean monthly mixed layer depth, and Weaver, A. J., and J. H. Middleton, On the dynamicsof the Leeuwin sea surfacetemperature, surface current, and surfacemeteorological Current, J. Phys. Oceanogr., 19, 626-648, 1989. fields, J. Geophys.Res., 94, 10,801-10,816, 1989. Wyrtld, K., Physical oceanographyof the SoutheastAsian waters, Reverdin, G., and J. Luyten, Near-surfacemeanders in the equatorial Scientific Results of the Maritime Investigations of the South Indian Ocean,J. Phys. Oceanogr.,16, 1088-1100, 1986. China Sea and Gulf of Thailand 1959-1961, NAGA Rep. 2, 195 Reverdin, G., M. Fieux, and J. Luyten, Free drifting buoy pp., Scripps Inst. of Oceanogr., La Jolla, Calif., 1961. measurements in the Indian Ocean equatorial jet, in Wyrtld, K., Oceanographic atlas of the International Indian Ocean Hydrodynamicsof the Equatorial Ocean, editedby J. Nihoul, pp. Expedition, 531 pp., National Science Foundation, Washington, 99-120, Elsevier, New York, 1983. D.C., 1971. Saetre, R., and A. J. da Silva, The circulation of the Mozambique Wyrtki, K., An equatorial jet in the Indian Ocean, Science, 181, Channel, Deep Sea Res., 31, 485-508, 1984. 262-264, 1973. Schott, F., Monsoon responseof the Somali Current and associated Wyrtki, K., Indonesian through flow and the associatedpressure upwelling, Prog. Oceanogr., 12, 357-382, 1983. gradient, J. Geophys.Res., 92, 12,941-12,946, 1987. Schott, F., M. Fieux, J. Kindle, J. Swallow, and R. Zantopp, The boundary currents east and north of Madagascar, 2, Direct measurements and model comparisons, J. Geophys. Res., 93, 4963-4974, 1988. M. E. Luther and J. J. O'Brien, Mesoscale Air-Sea Simmons, R. C., M. E. Luther, J. J. O'Brien, and D. M. Leglet, Interaction Group, Mail Stop B-174, Florida State University, Verification of a numerical ocean model of the Arabian Sea, J. Tallahassee, FL 32306 Geophys. Res., 93, 15,437-15,453, 1988. K. E. Woodberry, CATA, Box 391, University of Colorado, Swallow, J. C., and M. Fieux, Historical evidence for two gyres in Boulder, CO 80309 the Somali Current, J. Mar. Res., 40, suppl., 747-755, 1982. Swallow, J. C., R. L. Molinari, J. G. Bruce, O. B. Brown, and R. H. (ReceivedApril 13, 1988; Evans, Developmentof near-surfaceflow patternsand water mass acceptedJuly 6, 1989.) WOODBERRY ET AL.: SEASONAL CIRCULATION IN THE INDIAN OCEAN 18,241

$ I .

EO-

55-

105 -

155 -

I I I 11

0 5 • UC, 6. N

ISS

1 "SE 4 E 4' '• S G 6S• • E ' 8 •: • 1'• 0 I '• b

Plate 1 [Woodberryet al.]. Model upper layer thicknessand velocity from the tenth year of integration. Arrows indicateupper layer velocity, with only one arrow showneach 1.6ø in latitude and longitude. Arrows representing velocitiesgreater than 1.0 m s'1 aretruncated, and those less than 0.05 m s'1 are supressedfor clarityof display. Scale arrow is shownat upperleft. Color indicatesULT value in meters,with the color scaleshown at the bottom. Brown indicatesland values. (a) February. (b) August. 18,242 WOODBERRY ET AL.: SEASONAL CIRCULATION IN THE INDIAN OCEAN

5S -

10

15

I I I I I

o. M 5 R [5. lO

5S

b

Plate 2 [Woodberry et al.]. Southwesternarea of the model domain (36øE to 80øE, 22øS to 4øN). Color and arrows are as in Plate 1. (a) January. (b) April. (c) July. (d) October. The Rossbywave between8 ø and 12øSis blockedby the Seychelles-MauritiusRidge in Januarythrough April. An eddy formsto the eastof the SMR and is advectedthrough the gap at 11ø to 13øS in early July. Intense eddy activity is seenin the westernboundary current recirculationregion to the north of Madagascarand to the west of the Seychellesthroughout the year. Mixed Rossby-gravity(or Yanai) waves are particularly evident in the equatorialregion during the monsoontransition months of April and October. WOODBERRYET AL.' SEASONALCIRCULATION IN THE INDIAN OCEAN 18,243

$ I I

I i I I !

L i' I

55

d I I I

Plate 2 (continued) 18,244 WOODBERRYET AL.: SEASONALCIRCULATION IN THE INDIAN OCEAN

30

330-

3 -

70-

2-t -

210 -

I -

150

120

6O

t It tl F C 5 O7 rE

i 16 I i I• -• i -- --! i

Plate 3 [Woodberryet al.]. Time versuslongitude contours of model upper layer thickness(ULT) along 12øS Time is in days after December 16, year 9. Color indicates ULT value in meters, with the color scale shown at the bottom. Brown indicatesland values. The Rossbywave generatedeast of 100øEis blockedby the Seychelles-MaurifiusRidge at 60øE. An eddy shed throughthe gap in the ridge is seen to propagatewestward until it is absorbedinto the northward current at the east coast of Madagascarat 49øE (the brown vertical strip).

360

330

30O

Z70

Z•O

zlo

160

150

120

Plate4 [Woodberryet al.]. Time versuslongitude contours of meridionaltransport across the equator. Time is in days after December16, year 9. Color indicatestransport value, with greento red denotingnorthward transport and yellow to blue indicatingsouthward transport. Brown indicates land values. Packetsof wave energywith westwardphase and eastwardgroup propagation can be seenemanating from the westernboundary region. Thesewaves have a periodof 28 days and are interpretedas mixed Rossby-gravity(Yanai) wavesgenerated by instabilitiesin the westernboundary region. Transportis in sverdmps(1 Sv = 106m 3 s-l). Contourinterval is 5 Sv.