Winter Monsoon Circulation of the Northern Arabian Sea and Somali Current

JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 105, NO. C3, PAGES 6359-6376, MARCH 15, 2000

Winter monsoon circulation of the northern Arabian Sea and Somali Current

Friedrich A. Schott and Jfirgen Fischer Institut fiir Meereskunde, Kiel, Germany

Abstract. The winter monsoon circulation in the northern inflow region of the Somali Current is discussedon the basis of an array of moored acoustic Doppler current profiler and current meter stations deployed during 1995-1996 and a ship survey carried out in January 1998. It is found that the westward inflow into the Somali Current regime occurs essentially south of 11øN and that this inflow bifurcates at the Somali coast, with the southward branch supplying the equatorward Somali Current and the northward one returning into the northwestern Arabian Sea. This northward branch partially suppliesa shallow outflow through the Socotra Passagebetween the African continent and the banks of Socotra and partially feeds into eastward recirculation directly along the southern slopes of Socotra. Underneath this shallow surface flow, southwestward undercurrent flows are observed. Undercurrent inflow from the Gulf of Aden through the Socotra Passageoccurs between 100 and 1000 rn, with its current core at 700-800 rn, and is clearlymarked by the Red Sea Water (RSW) salinity maximum. The observations suggestthat the maximum RSW inflow out of the Gulf of Aden occurs during the winter monsoon seasonand uses the Socotra Passageas its main route into the Indian Ocean. Westward undercurrentinflow into the Somali Current regime is also observedsouth of Socotra, but this flow lacks the RSW salinity maximum. Off the Arabian peninsula, eastward boundary flow is observedin the upper 800 m with a compensatingwestward flow to the south. The observedcirculation pattern is qualitatively compared with recent high-resolution numerical model studies and is found to be in basic agreement.

1. Introduction At the eastern entry of the Arabian Sea the Mon- soon Current south of Sri Lanka flows westward during Very few observationshave been carriedout during the winter monsoon as the Northeast Monsoon Current the Northeast Monsoon in the northern Somali Cur- (NMC) [Schottet al., 1994]. Their mooredobservations rent regime and interior northern Arabian Sea. As is yielded a mean westwardNMC transport of 11 Sv, con- knownfrom climatologicalship drift data [e.g.,Moli- centrated in the upper 100 m and within •100 km of nari et al., 1990], the near-equatorialsurface circula- the Sri Lanka topography.Where this current goesaf- tion of the westernIndian Oceanduring Novemberto ter passingthe 80øE meridian is not clear. Ship drift Februaryis characterizedby the southwardflowing So- data suggest a low-latitude westward continuation of mali Current,which encounters the northwardflowing the NMC but also northwestward flow along the In- East African Coast Current at 2ø-4øS and out of this dian subcontinent[Swallow and Fieux, 1982;Molinari confluencethe eastwardflowing South Equatorial Coun- et al., 1990]. Water massdistributions [Wyrtki, 1971] tercurrentemerges [Swallow et al., 1991].However, the support a throughflow of low-salinity Bay of Bengal Somali Current in winter doesnot appear to follow the Water into the Arabian Sea during this time period. coast all the way from the Horn of Africa to south of Donguy and Meyers [1995], from the analysisof the the equator. Instead, there appears to be inflow into World OceanCirculation Experiment (WOCE) expend- the coast in the 5ø-10ølatitude band with southward ablebathythermograph (XBT) lines,also supported the flow from there to beyond the equator and northward existenceof a westwardlow-latitude geostrophicflow as flowin the northernSomali Current regime [e.g., $chott the extension of the NMC. et al., 1990;Molinari et al., 1990]. West of India, an anticyclonic circulation feature, the LaccadiveHigh, existsin the late phaseof the Northeast Monsoon and appears subsequentlyto begin propagat- Copyright2000 by theAmerican Geophysical Union. ing westward[Bruce et al., 1994]. Modelstudies suggest Papernumber 1999JC900312. its origin to be remote forcing from the Bay of Bengal 0148-0227/00/1999JC900312509.00 by coastal trapped waves, which in turn, radiate Rossby

6359 6360 SCHOTT AND FISCHER: WINTER MONSOON CIRCULATION

22øN

Sonne 128

7.1.-29.1.1998

20øN ß CTD

o XBT • ICM 7 ooOOOO5 OO 18øN

16øN

14øN

12øN.•

10øN !•l•iit

8Ol,q -,,

6ON 50OE

Figure 1. Topographyof the ArabianSea, positions of the mooringswest and southof Socotra (diamonds),and ship track ofR/V $onnesurvey SO 128of January1998 marked by conductivity- temperature-depth(CTD) stations(solid circles) and expendablebathythermograph (XBTs) (opencircles). Individual sections are labeledA-D for easierrecognition in the text.

waves[e.g., McCreary et al., 1993;$hankar and $heyte, but the physical causeof this undercurrent and its con- 1997]. For the higherArabian Sea latitudesDonguy nection to neighboring latitudes are still unclear. At and Meyers[1995] suggest a cyclonciccirculation of or- 5øN the current structure in the moored observations der 10 Sv for the winter monsoonseason, with eastward of Quadfaseland Schott[1983] did not showsuch an transport in the 4ø-8ø N latitude rangeacross the inte- undercurrentstructure. For the 6ø-10øNlatitude range rior of the Arabian Sea and a correspondingwestward Quadfaseland Schott[1983] derived a southwardun- inflowinto the northernSomali Current regime. How- dercurrentunderneath the northwardflowing surface ever,their XBT linesfrom Perth, Australia,to the Per- boundarycurrent from analysisof earlierhydrographic sianGulf and to the Red Sea,respectively, leave a wide observations.Offshore in the northernregime, where range of the interior Arabian Sea uncovered. the Great Whirl spinsaround during the Southwest The subsurface structure of the Somali Current dur- Monsoon[Swallow et al., 1983; Fischeret al., 1996; ing the NortheastMonsoon also still posesseveral prob- Schottet al., 1997],remnants of this largeanticyclonic lems that require explanations. At the southern end gyre may linger far into the winter monsoon season the mooredtime seriesof Schottet al. [1990]showed a [Bruceet al., 1981]hidden underneath the developing northward cross-equatorialundercurrent in winter un- near-surface winter monsoon circulation. derneaththe shallowsouthward surface Somali Current, In this presentation the winter monsoon circulation SCHOTT AND FISCHER:WINTER MONSOONCIRCULATION 6361 of the northern Somali Current regime is discussedon K19 K18 K17 K16 K15 K14 K13 K12 the basis of two sets of observations: first, on a ship- [m] board survey with R/V $onne with water mass and 1000 current profiling observationscarried out in January 1998 and, second,on the recordsof the WOCE moored 2000 array ICM7 (Figure 1) placedin the offshoreSomali Current, south of the island of Socotra, and in the pas- 3OOO sagebetween Socotra and the African mainland (called Socotra Passagein the following) during 1995-1996. 4000 The shipboard study covered the 8øN section by hy- drography/tracerand acousticDoppler currentprofiler 5OOO (ADCP) observationsfrom the Maldiveswestward to 53øE and then went on to investigate the area around 7øN 8øN 9øN 10øN 11øN Socotra, the Socotra Passage,and the meridional section from Socotraup to the Arabian peninsula(Fig- Figure 2. Topography along 54øE and current me- ure 1). ter distribution on moorings K12-K19. Circles indicate positionsof rotor current meters (RCMs) that yielded For the interpretation of the observed current pat- completerecords over the winter 1995-1996 period, and terns we will also draw on recent model simulations acousticDoppler current profilers(ADCPs) are shown based on forcing by realistic wind fields of the obser- by symbols with acoustic beams. vational time period. We will mainly use output of the high-resolutionParallel Ocean Climate Model (POCM) of Semtnerand Chervin[1992] as alreadyanalyzed for the Indian Ocean overturning circulation by Garter- CPs on top, locatedat nominally250 m depth, which nicht and $chott [1997]. We also inspectedresults of all deliveredgood data. In the passage,mooring K10 the multilayer reduced gravity model run by J. Kindle on the westernside also was equippedwith an ADCP of the Naval ResearchLaboratory, Stennis Space Cen- (Figure3). ter, and of the Miami isopycnicmodel, run by O. Es- Besides ADCPs, conventional rotor current meters enkov of Rosenstiel School of Marine and Atmospheric (RCMs) of type Aanderaawere used (Figure 2). The Sciences(RSMAS), Universityof Miami. measurements from the ADCPs were interpolated to constant depths on the basis of backscatter distance measurementsfrom the surface[Visbeck and Fischer, 2. Observations 1995],thus eliminating the mooringmotion effects, while During April 1995 to October 1996 mooredcurrent the depthsof the RCM time seriesused in the following observations were carried out across the northern So- are their mean depths. The current observationlevel mali Current regimeas part of WOCE. Becauseof the closest to the surface for all ADCPs that could be used unresolved clearance situation for the Somali exclusive was the 25 m level. The level above was contaminated economiczone (EEZ), the array couldnot be deployed, becauseof interferenceby the direct sidelobeechos from as originally intended, normal to the coast and conti- the sea surface with the returns from the inclined main nental slopeacross the northern boundary current. In- beams at shallower depths. stead, a line of ADCP and current meter mooringswas The shipboardobservations on R/V $onneconsisted deployedon the continentalslope south of the island of conductivity-temperature-depth-O2(CTDO2) mea- of Socotra, with two mooringsalso coveringthe Soco- surements,current profiling by shipboardand lowered tra Passagelocated between the African continentand ADCP (LADCP), and, at five stationsalong the moor- the island of Abd al Kuri (Figure 1). This passage, ing line, currentprofiling by Pegasusprofiler. For the with a maximum depth of -•1050 m, had already been determination of current vectors out of the shipboard found to be crucial for the summer monsoon circulation ADCP registrations,a differential Global Positioning in an earliership survey[Fischer et al., 1996]. It was System(GPS) wasused for positiondetermination, and found then, and subsequentlyconfirmed, for the sum- the Ashtech GPS antenna was used for improving the mer monsooncirculation of 1995 [Schottet al., 1997] ship's heading. that the exchangeof the northernSomali Current with For the current sectionsand transportsin the follow- the interior Arabian Sea occurred dominantly through ing, the shipboardADCP and LADCP currentswere this passageand northwardaround Socotra rather than mergedto a joint productand objectivelyinterpolated directly eastwardin the 8ø-10ø latitude range. with mapping scalesthat varied, accordingto station The moored array was deployedduring March 1995 spacing,from 5 to 30 km and vertically from 20 m by R/V Meteorand retrievedby a charteredvessel in in the upper 300 m coveredby the shipboardADCP October 1996. The mooringsalong the 54øE line and to 100 m at depth. The merged current profileswere the instrumentcoverage are shownin Figure 2. Six of correctedfor barotropic tidal constituents. Along the the eightmoorings along that line carried150 kHz AD- mooring line and in the Socotra Passagethese could 6362 SCHOTT AND FISCHER: WINTER MONSOON CIRCULATION

200

4OO

600

800

1000

5.:?E......

1200 ....."": .... 51.5øE 52øE 52.5øE 53OE 53.5OE Figure 3. Topographyof theislands of Socotraand Abd al Kuri andthe SocotraPassage with instrumentdistributions on passage moorings K10 andKll (seeinset for locationof section).

be determined from the moored current records (M. there is a southwestwardundercurrent along the slope Dengler, personalcommunication, 1998). Away from of Socotra(Figure 1). This undercurrentshould be the regionsof observedcurrent time series,model tides de- source of the coastal southward undercurrent identified terminedby Le Provostet al. [1998](FES95.2 - Model) earlier off Sorealiafrom hydrographicpressure gradients were used. Generally, correctionsdue to the tides were by Quadfaseland $chott [1983]. smallfor the shipboardADCP/LADCP transportsused Time seriesof the two moorings,K10 and Kll, in the in the following. Since the currents are variable at sea- Socotra Passageshow that there is a continuousnorth- sonaland shorter timescales,current coresin this region ward near-surfaceoutflow into the Gulf of Aden, which are not typically associatedwith water mass coresthat doesnot havea clearmonsoonal modulation (Figure 5). would usually allow for an interpretation of longer-term At the 700 m level, there is still dominantly outflow mean circulations. Instead, water mass distributions on on the eastern side, but on the western side we ob- isopycnalsare marked by high variability due to isolated serve southward inflow into the Somali Basin most of lenses,making contouringdifficult. the time. This inflow is particularly strong during the winter monsoonphase. This deep passageis the main 3. Winter Monsoon Currents and inflowroute of the Red SeaWater (RSW) into the west- Transports From Moored Array ernIndian Ocean, as will also)0ecome apparent from the 1995-1996 ship section observationstaken in January 1998. Vector maps for the monsoon seasoncurrent means, Time seriesof the 40-hour low-passedcurrent vectors November 1, 1995, to Feburary 15, 1996, are shown in at station K12 are shown in Figure 4 to demonstrate Figure 6 for different depth levelsand are summarized the monsoonal cycle in the northern Somali Current with their standard deviations in Tables i and 2. regime. The strongestsignals are associatedwith the While the mean near-surface flow during the win- Great Whirl (GW) developmentsduring the summer ter monsoonis eastwardjust southof Socotra(station monsoonsof 1995 and 1996, where the GW was found K12), it is westwardfor all other stationsbetween 7øand to be much more northerly and stronger developedin 11øN south of Socotra down to the 200 m level and it is the firstthan in the secondyear [Schott et al., 1997].For northwardthrough the SocotraPassage (stations K10 the winter monsoon1995-1996, the interestingresult is and Kll). This suggeststhat the near-surfacewest- that there is a net northeastwardflow in the top 200 m ward inflow bifurcates near the coast into a northward at this station south of Socotra, but that below 200 m, branch off Sorealia which then suppliesthe northward SCHOTT AND FISCHER: WINTER MONSOON CIRCULATION 6363

K12 ----

50 m

100 m

150 m

400 m

700m .•l/,, K13 :

: 1995 1996 A M J J A S O N D J F M A M J J A S O Figure 4. Vectortime series of the 40-hourlow-passed currents at the northernboundary station,K12, during April 1995 to October1996. Currents are rotated for bettervisualization; see direction arrow at upper right.

K10 k:•l•i,'1 ,,i,• . 1•1,' -'• --- _ 100m

700m T : ß 900m .

. •...... ,.-...... • ..• ::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::

700m .... -

A M d d A S O N D d F M A M d d A S O

Figure 5. Vector time seriesof the 40-hour low-passedcurrents for severaldepths at stations K10 and Kll in the Socotra Passage;northward is up. 6364 SCHOTT AND FISCHER: WINTER MONSOON CIRCULATION

14øN

12øN

10øN

8øN

6øN 50øE

14øN 14øNI'•"•' "' l 20cm/s 12øN J 400mdepth

10øN

8ON

6ON 6øN8øN •'••,•••.- '" ... 50øE 50øE 52øE 54øE 56øE 5• •øE

12øN 12øN

10øN 10øN

8øN 8øN

6øN 6ON 50øE 50OE

Figure 6. Current vector meansfrom the moored array for the winter monsoonperiod, November 1, 1995, to February 15, 1996, for different depth levels.

upper layer throughflowthrough the SocotraPassage The currentsfrom mooringobservations normal to and the eastwardboundary current south of Socotra. the array (K12-K19)are shownboth for an annualmean At the 400-1000 m level the boundarycurrent di- (Figure7a) andfor the wintermonsoon season (Fig- rectly southof Socotrais southwestward,with monsoon ure 7b). The annualmean (Figure 7a) is dominated seasonmeans of 18 cm s-• at the 700 m level of station by the large summer monsooncurrents and showsthe K12 and 8 cm s-• at 1000m at K13, farther offshore structure of the Great Whirl with eastward flow north (Figure4 and Table2). The passageflow at the 700- of 9øN (stationsK12-K16) and westwardrecirculation 900 m levelis southwardalong the westernslope, while to the south (K16-K19). The winter monsoonmean currents farther offshore than K13 are weak. As men- (Figure7b) shows the small eastward core just south of tioned above, these two undercurrents should combine Socotrawith an estimatedtransport of only0.8 Sv and to supply the southward coastal undercurrent off north- the deep-reachingwestward undercurrent underneath, ern Somaliafound by Ouad/aseland $chott [1983]. carrying • 6 Sv. The mean westward flow in the sur- SCHOTT AND FISCHER: WINTER MONSOON CIRCULATION 6365

Table 1. Mean Velocities and Standard Deviation tbr the Annual Mean and for the Northeast Monsoon period in the Socotra Passage.

Nov. 1 to Feb. 15, cgs April 1995 to April 1996, cgs Type Depth,

MooringK10 11ø51'N,51ø51'E ADCP 50 -14.23 32.04 38.26 23.48 -17.83 25.56 33.69 28.49 ADCP 100 -7.38 21.11 20.22 23.14 -16.70 18.40 24.18 23.46 ADCP 200 3.07 12.49 -2.07 15.47 -6.38 12.50 9.21 18.33 RCM 725 11.69 8.78 -15.05 6.09 4.55 9.99 -6.60 9.81 RCM 900 2.05 3.55 -2.42 2.26 1.69 2.93 -1.32 1.94 MooringKll 12ø03•N,52ø04•E RCM 104 -12.36 23.95 23.99 9.27 -23.34 21.21 26.00 11.78 RCM 206 -1.92 10.85 4.77 5.19 -10.65 13.64 9.65 7.58 RCM 400 -1.04 5.07 1.27 2.53 -2.59 5.10 1.81 2.59 RCM 725 -6.61 8.19 2.44 4.10 -6.20 6.98 2.83 3.85

face layer between 7øand lløN amounts to 11.4 Sv on tially suppliesa throughflow through the Socotra Pas- the basis of the mooring data points. The time series sage and partially circulates northeastward along the show,however, that this mean circulation(Figure 7) southeasternslope of Socotra. This part of the pattern is superimposedby vigorousvariability (Figure 4) with is in agreement with the upper layer winter monsoon the standarddeviations larger than the means(Table 2) mean circulation found in the moored array currents suchthat shorter-periodmeans will showquite different (Figure 6). Part of the flow that turns southwardin spatial patterns. the ship survey recirculates acrossthe 8øN line between First estimatesof time seriesfor SocotraPassage 54 ø and 56øE. This structure is different from the win- transports derived from K10 and Kll data were pre- ter monsoon mean flow, obtained from the moored sta- sentedby Schott et al. [1997]and showedvariations tion data (Figures6 and 7), wherewestward upper-layer from 0-5 Sv during the 1995-1996winter monsoon.Re- inflow occurred in the whole latitudinal band from 7øN analysisof the data includingextrapolation toward the to 11øN. Given the large observedcurrent variability in topographyyielded even larger variationsranging from the period range of a few weeks, these differencesare 3 Sv southwardto 7 Sv northwardduring the winter not surprising. monsoonperiod. The winter monsoonmean (averaged East and north of Socotra, intense mesoscaleeddies overthe periodNovember 1, 1995,to February15, 1996) are found, and it is not clear how the eastward out- yieldeda top to bottom transportof 2.2 Sv northward, flow from the Somali Current that is attached to the whichwas dominated by the near-surface(upper 200 m) coast in section B south of Socotra continuesits way flowamounting to 3.8 Sv. Belowthe surfacelayer, there past this eddy field. At the deeper level the westward is net southwardtransport of 1.5 Sv coveringboth the zonalinflow is still presentat 9ø-11øNalong 54øE (Fig- PersianGulf Water (PGW) layer (200-500m) and the ure 8b). At this levelwe haveinflow through the Soco- RSW layer below 500 m, with most of the flow con- tra Passage,while the circulation features observed in centratedon the westernpart of the passage;the flow the northeasternpart of the surveyfor the near-surface at the easternside of the passagewas comparatively currents generally continue down to this level. weak. The southward flow of RSW varied from 0.2 to Alongthe northernboundary, off Oman, no consis- 0.8 Sv throughoutthe winter monsoonseason at a mean tent boundary current is found in the two sections nor- transport of 0.5 Sv southward. mal to the coast,apparently because of strongeddy activity. While at the 59øE section,coastal currents off Arabia are eastward,the flow near the coastat 57øE is 4. Circulation During the Ship Survey westward(Figure 8). of January 1998 Salinity sectionsfor the upper 800 m are shownin 4.1. Larger-Scale Connections Plate1 forfour sections (marked A-D in Figure1). The subduction of Arabian Sea surface water can be seen Vector plots from the shipboard ADCP are shown in in the sections;maximum surface-mixed layer salinity Figure 8 for the near-surface(20-50 m) (Figure 8a) occursin the sectionsC and D northeastof Socotra,and and for the 150-300 m layer (Figure 8b). The surface the southwardsubsurface penetration of this maximum flow shows westward inflow into the Somali Current in is seen in section B, south of Socotra. the 9ø-11øN latitude band that bifurcates southward At the RSW level, around the densitylevel and northward.The northernbranch of Figure 7a par- 27.2 [e.g., Quadfasel,1982], maximumsalinities are 6366 SCHOTT AND FISCHER: WINTER MONSOON CIRCULATION

Table 2. Mean Velocities and Standard Deviation for the Annual Mean and for the Northeast Monsoon period South of Socotra.

Nov. 1 to Feb. 15, cgs April 1995 to April 1996, cgs Type Depth, m • cr• • cry •, eYu • O'v

MooringK12 11o24' N, 53o00' E ADCP 50 10.57 32.87 7.00 18.3 22.55 63.37 18.64 33.41 ADCP 100 9.91 32.23 5.48 15.14 15.16 45.93 16.08 25.33 ADCP 200 -0.02 19.79 -1.62 6.20 5.27 21.18 4.88 10.19 RCM 400 -10.41 12.30 -3.93 6.31 3.02 19.55 1.33 7.54 RCM 700 -15.33 10.53 -9.28 6.03 -5.37 16.59 -4.33 5.80 MooringK13 11000' N, 53013' E RCM 150 3.90 25.73 -5.06 21.18 11.55 30.68 8.32 20.07 RCM 200 -0.92 15.81 -3.83 13.20 5.77 16.61 4.51 1] .54 RCM 400 -2.95 6.60 -2.25 3.66 1.51 6.59 0.89 4.63 RCM 700 -2.63 4.25 -0.45 1.85 0.15 5.47 0.43 3.21 RCM 1000 -9.14 6.38 -4.21 3.68 -3.91 14.88 -1.77 7.75 RCM 2000 -0.23 7.61 1.19 -0.60 -0.19 10.30 0.35 4.30 RCM 3000 -1.78 3.27 -0.60 3.25 -1.06 2.55 -0.40 2.29 MooringKIJ 10036' N, 53025' E ADCP 50 -5.11 34.63 -12.97 31.36 31.67 71.25 10.04 33.35 ADCP 100 -4.75 35.22 -14.34 30.25 18.52 48.38 8.72 31.77 ADCP 200 -4.77 19.28 -9.35 17.69 5.83 20.23 5.70 18.57 ADCP 400 -1.37 11.30 -4.42 11.93 3.74 11.74 2.24 12.12 RCM 700 0.82 9.36 -0.91 9.06 4.51 11.39 3.19 12.90 RCM 1000 -1.36 7.49 -0.04 4.26 3.69 10.26 2.78 10.75 RCM 4000 -2.63 4.25 -3.27 3.03 -1.31 3.07 -1.41 2.81 MooringK15 10012' N, 53038' E RCM 100 -20.87 24.60 -12.03 34.34 7.96 39 8O -4.27 25.95 RCM 200 -14.51 16.49 -6.01 18.70 0.54 20 05 0.83 16.51 RCM 400 -2.40 9.36 -0.59 12.60 0.08 10 05 1.14 12.21 RCM 1000 0.09 3.97 0.64 5.62 1.95 4 44 0.77 5.26 RCM 2000 0.09 2.45 -3.02 2.63 0.48 247 -0.15 3.43 RCM 3500 -1.83 5.75 -2.80 4.35 -1.56 4 04 -0.11 4.69 MooringK16 9ø40' N, 53ø56' E ADCP 50 -12.24 37.04 -1.76 52.49 5.18 37 00 -5.01 40.35 ADCP 100 -8.03 30.87 -4.71 46.66 5.22 30 91 -4.82 34.13 ADCP 200 1.99 15.83 -2.34 25.02 4.98 16 32 -2.70 18.30 ADCP 400 0.77 5.12 -1.90 8.06 0.68 532 0.89 7.33 RCM 700 0.86 3.13 1.05 7.36 0.48 4 29 -1.62 6.21 RCM 2000 4.74 3.92 O.O2 5.42 2.04 5 O5 -0.56 5.24 RCM 4000 1.82 2.62 1.21 2.28 0.23 216 0.46 1.60 RCM 4500 2.25 7.26 3.37 6.08 -0.03 5 33 0.88 4.25 MooringK17 8ø45' N, 7ø50' E ADCP 50 -20.42 19.25 -1 33 36.79 -16.23 43 46 -27.92 49.77 ADCP 100 -16.56 15.96 -5 90 0.25 -20.36 39 74 -24.96 43.33 ADCP 200 -6.54 5.90 025 19.71 -12.88 19 18 -8.92 24.00 RCM 400 -1.51 2.58 0 76 8.68 -3.41 4 46 -0.77 6.45 RCM 700 0.15 2.97 1 89 7.73 -1.85 374 -0.72 6.25 RCM 2000 -0.28 2.20 124 2.71 -0.54 2 66 0.06 3.67 MooringK18 7ø50' N, 53ø54' E ADCP 50 -14.47 25.26 -8.28 13.94 -25.07 47.59 -24.53 41.41 ADCP 100 -11.19 21.14 -8.36 13.42 -27.45 42.27 -17.11 31.90 ADCP 400 4.44 13.22 0.62 16.04 -4.92 13.56 -1.40 13.39 RCM 700 1.59 4.03 0.11 6.58 -1.49 4.45 -0.74 5.34 RCM 2000 0.42 1.78 0.66 2.70 -0.09 2.12 0.59 2.66 MooringK19 07000' N, 53025' E ADCP 50 -16.95 17.99 -2.99 13.53 -28.88 41.81 -6.61 47.01 ADCP 100 -10.97 12.75 -6.38 9.01 -27.22 38.33 -3.68 37.32 ADCP 200 1.38 6.27 -3.85 14.05 -6.72 13.62 -1.74 20.22 ADCP 400 3.62 5.36 -1.56 16.10 -0.58 9.96 -1.08 16.79 SCHOTT AND FISCHER: WINTER MONSOON CIRCULATION 6367

(a) ,19 K1 K17 K16 K15 K14 3 K12 [m]

250 -6 500•"

750

1000

2000

3000

4000 April 95 - March 96

5000

7øN 8øN 9øN 10øN 11øN

(b) 9 K1 K17 K16 K15 K14 3 K12 [rn]

250

500

750

1000

2000

3000

4000 I Nov 95 - 15 Feb 96

5000

7ON 8ON 9ON 10ON 11øN

Figure 7. Currentsand transports.normalto the mooringline (mooringsK12-K19) southof Socotra:(a) annualmean (April 1995to March1996) and (b) wintermonsoon (November 1995 to February1996). Transportnumbers (in Sverdrups)for boxesare marked.HeaYy numbers in boxesare net transportsfor the boxes;numbers below them are eastward(left) and westward (right) contributionsto box transports. Eastwardcurrents are shaded;the contourinterval 2 cm s-• . The verticalcoordinate is stretchedfor the upper1000 m.

found alongsection A, runningfrom Abd al Kuri to salinity, is also distinguishableby its higher 02 values the African continental slope. The salinity maximum (>0.3 mLL -i) from the backgroundwater (<0.2 mL is concentratedon the westernslope, indicating the in- L -1) of the northernArabian Sea. Alongthe 59øEsec- flow of RSW into the Indian Oceanthrough the Socotra tion (Figure 1), the near-coastalsalinities and oxygens Passage.High salinitiesat ae=27.2 are alsoobserved are both low, suggestingthat the secondarymaximum alongthe northernborder at 57øE(section D) but not in section D is an isolated lense rather than a coherent southof Socotra(section B). The RSW,besides its high RSW penetration along the northern border. 6368 SCHOTT AND FISCHER: WINTER MONSOON CIRCULATION

4.2. Passage Currents The shipboard ADCP and LADCP currents that Jan. 1998 were determined from the six stations acrossthat pas- mean flow sage, are shown in Figure 9. They are correctedwith 20m - 50m the tidal constituentsdetermined from mooringsK10 and Kll, but usingthe Le Provostet al. [1998]tidal model yields rather similar results. The near-surface outflow regime during the ship survey is shallow, only -•100 m thick, and yields a northward outflow transport of 2.2 Sv above ae=25.4 compared to 3.8 Sv for

. : the near-surfacelayer as the winter monsoonmean estimate from the current meter records. Below it, there is southward inflow, but it is restricted to the western side of the passage,and its deep current core of more than 30 cm s-• is associatedwith the RSW salinity maximum of Plate 1. The inflow transport into the Somali Basin below ae=25.4 totals 5.4 Sv, while the lower part on the eastern side shows a weak outflow of 1.0 Sv. i i This winter monsoon current structure is similar to

48 ø 50 ø 52 ø 54 ø 56 ø 58 ø 60 ø 62øE that of summer1993 reported by Fischeret al. [1996], where the inflow was also concentrated on the western channel side. The mooring data of Figure 5 also indi-

i cate this cross-channelstructure to be representative of (b1 the year-round structure on which strong variability is Jan. 1998 superimposed. Compared to the net northward trans- , i : mean flow i port of 2.2 Sv from the moored stations for the winter

i 150m - 300m , i monsoon 1995-1996 period mean, this net southward throughflowof 2.2 Sv through the Socotra Passagedur-

i : ing the time of the ship sectionhas to be explained in _t_ _ terms of the significant current variability observedin

i • i i i the mooredcurrents (Figure 5). At depth, however,the

.4 i-. large RSW currentcore of January 1998 (Figure 9) con- firms that the 1995 winter monsoon maximum observed ':SOC i at the 700 m level of station K10 is not randomly occur- ing but rather suggestsa winter monsoonRSW inflow ß maximum in agreementwith the observationsof Murray and Johns[1997] at the exit of the Red Sea.

4.3. Ship Section Along the Mooring Line

i , i i The shipboardADCP currentsalong this section(B

: : in Figure 1) were correctedfor tides on the basis of our own current meter record from stations K12-K19 48 ø 50 ø 52 ø 54 ø 56 ø 58 ø 60 ø 62øE (M. Dengler, personalcommunication, 1998), and for Figure 8. Shipboard ADCP currents during January comparisonthey were also corrected by using the tidal 14-27, 1998,for two depth ranges:(a) 20-50 m and (b) 150-300 m. modelconstituents of Le Provostet al. [1998],and the corrections to the transports were found to be small, as is illustrated by the following example. The tidal aliasing of the shipboard measurementsbased on the Le Provostet al. [1998]model tides accumulatesto an At ae=26.6, correspondingto the densitylevel of the apparent westward inflow of 0.5 Sv acrossthe 54øE sec- Persian Gulf outflow at 250-300 m depth, the maxi- tion (Figure 10) for the layerabove a•=27.3 (the RSW mumvalues combined with low (<0.1 mLL -z) 02 val- layer) and latitude range 8ø-12øN;there is no net dif- ues are found in the 60øE section; higher values are ference for that section between the estimation based found east of Socotra than along the northern border. on the moored records and those from the Le Provost The distributions at both levels ae=27.2 and 26.6 show et al. [1998]model, althoughthe detailsdiffer. How- howstrongly the northernArabian Seais dominatedby ever, we tried to remove this effect by subtracting the mesoscale water mass features. estimate of the barotropic tides to reduce further this SCHOTT AND FISCHER: WINTER MONSOON CIRCULATION 6369

'•35.[••=••'•._• 357

200...... •'-- -254-35 6..• _-•-- 200 ------35.6

6O0

800 8OO B

51.85øE 51.9øE 51.95øE 52øE 52.05øE 8øN 9øN 10øN 11øN 12øN

2OO

400

600

8O0 8OO

55øE 55.5ø E 56øE 13øN 14øN 15øN 16øN 17øN

Plate 1. Salinity sectionsfor upper 800 m: (a) SocotraPassage section (stations 49-57, section A), (b) 53øEsection south of Socotra(stations 30-46, section B), (c) 12øNsection east of Socotra (stations58-68, sectionC), and (d) 57øE sectionsouth of Oman (stations. 68-82, sectionD). Also marked are isopycnalsere= 25.4 and 26.9; for sectionlocations and station numbers,see Figure 1.

10% effect. The merged shipboardand LADCP cur- 700 m, that of the northeastward return branch south rentsalong the 54øEsection (stations 30-46, Figure 10) off Socotra is 1.8 Sv, and the transport of the west- show the westwardinflow into the boundary regime in ward undercurrent acrossthe box marked in Figure 10 the 8.5ø-10.5øN latitude range and the northeastward is 2.2 Sv between500 and 2000 m. The mooringsalso recirculation, which was already discussedin conjunc- showed a monsoon seasonwestward mean in this depth tion with Figure 7, where it was shown that this in- range (Figure 7b), which thereforeappears to be a ro- flow and recirculationreaches down to 500 m, but with bust winter monsoonfeature. Overall, comparing Fig- its intensive part limited to the layer above ere=25.0, ures 7 and 10 showsgeneral agreement among the two in correspondensewith the northward passageoutflow, flow structures north of •9øN. South of 9øN as men- which wasalso limited to the upper 100 m (Figure 9). tioned already earlier, the mean surface flow from the Underneath the northeastward recirculation currents moored currents continuesto be westward, while east- off Socotra we find deep-reachingwestward flow toward recirculation is observedin the ship survey situ- ward Somalia that supposedly feeds into the south- ation of January 1998. The relatively low salinity at ward coastal undercurrent off Somalia described ear- the RSW core layer of the westward undercurrent in lier [Quadfaseland Schott,1983]. Its currentmaximum the 54øE section(B in Plate 1) suggeststhat the RSW is quite deep during the ship survey, at 1000-2000 m. insertioninto the Somali Basin occursalong the Somali The westward inflow transport is •11 Sv in the upper coast past the passageand not around Socotra. 6370 SCHOTT AND FISCHER: WINTER MONSOON CIRCULATION

[m]

--- 25.4

250

- 26.9__.

500

750

section A Socotra Passage 51.7øE 51.8øE 51.9øE 52øE 52•.1OE

Figure 9. Currentsand transportsparallel to SocotraPassage axis (sectionA). Northward is positive(shaded); the contourinterval is 20 cm s-• . Alsomarked are isopycnalscre=25.4 and 26.9. Net transports for the three layers separatedby the two isopycnalsare given in Sverdrups (heavynumbers). Light numbersbelow are northward(left) and southward(right) box transport contributions.

4.4. Flow Between $ocotra and the Arabian model constituents. Figure 11 showsthe continuation Peninsula of the shelfedge flow of -•2 Sv off easternSocotra, which Currentsand transportsare shownin Figure 11 for crossedthe 54øE sectionsouth of Socotra. Offshore,the the 12øN section and in Figure 12 for the 57øE sec- reversing currents nearly compensate above eye-25.0, tion, both from merged shipboard and LADCP data where good resolution in the ADCP data exists, while and correctedwith the Le Provostet al. [1998]tidal the transport differencewithin the cr•--25.0-27.3 layer

e32

Sv [m]

250

5,0 Sv •i:•0! Sv 500

- 27. 750

SV 1 sv 30(

41 5000- section B, 53øE 8øN 8.5øN 9øN 9.5øN 10øN 10T.5oN 11•ON

Figure 10. As for Figure 9, but for section B, •53øE, south of Socotra. Positive currents (shaded)and transportsare eastward.The verticalcoordinate is stretchedfor the upper 1000m. SCHOTT AND FISCHER: WINTER MONSOON CIRCULATION 6371

[m]

250- -6J• Sv 0.5Sv•-7,4 Sv

273- 751

section C, 12øN

r T•T-- 54.8øE 55øE 55.2øE 55.4øE 55.6øE 55.8øE 56øE 56.2øE 56.4øE Figure 11. Asfor Figures9 and10 but for 12øNsection (C) eastof Socotra.Positive currents (shaded)and transportsare northward. between the northward and the southward transports with a transportof 11.6 Sv aboveao=27.3, of which maypartially be dueto the deepfeatures not beingre- 2.4 Sv are contributedby the surface-intensifiedflow solvedby the LADCP profiles.The salinitiesin Plate 1 abovea0=25.0. High salinitiesat the RSW level near confirmthat this sectionis not a supply route of RSW the northerncoast (Plate 1) indicatethat occasional into the Indian Ocean, at least not during the winter RSW spreadingmay occurin this direction,but it is monsoon.Offshore, high salinitiesnear ao=26.6 along interestingto notethat no PGW coreis detectedat the the 12øNsection (Plate 1) indicatethe presence of PGW a0=26.6 level. In the 13ø-16øNband the flow is west- of which remnants are still seen in the Socotra Passage. ward with 2.4 Sv abovea0=25.0 and a total of 10.5 Sv Alongthe 57øEsection (Figure 12) an eastward bound- abovea0=27.3. The net eastwardflow north of 13øN ary currentoff the southcoast of Omanis observed in the upper 700 m is near zero.

[m]

250.

500

750 - - - 27.3...... •:.'•' - 27.3

section D, 57øE 13ON13.5 ø 14ø 14.5ø 15ø 15.5ø 16ø 16'.5o i•ON 17.5ø Figure 12. Asfor Figures9 and10 but for 57øEsection (D) southof Oman. 6372 SCHOTT AND FISCHER: WINTER MONSOON CIRCULATION

22øN

20øN

18øN

16øN

14øN

10øN12øN

8øN

6øN

50øE 52øE 54øE 56øE 58øE 60øE

Figure 13. Schematiccirculation of. the northernSomali Current regime during winter monsoon. (a) upper layer (a0<25.0) with subductionout of the layer whichis indicatedby a circlewith a crossin it, and (b) intermediatelayer (25.0

5. Discussion and Conclusions 5.1. Upper Layer Transport Budget The winter monsoon circulation of the northwest In- The water massesindicate subductionout of the up- dian Ocean was observed with moored current meter per layer into the thermoclinelayer for the regionbe- stations during the winter of 1995-1996 and with a tweenSocotra and the Arabian Peninsula(Figure 13). shipboard survey in January 1998. A complex situa- Becauseof the large short-period variability obvious tion of westward inflow, exchangethrough the Soco- from the mooredtime series(Figure 13), it is, of course, tra Passagebetween the island of Abd al Kuri and unreasonableto expect a closed budget for the box the African continent, and northwestward recirculation enclosedby the combinedsections A, C, and D (Fig- emerged.Figure 13 summarizesschematically the flow ure 13), but we will checkon it here anyway. For the patterns for the upper and intermediate layers on the layer above ao=25.0 we obtain a northwardthrough- basis of moored and shipboard observations.The west- flowof 2.2 Sv throughthe SocotraPassage into the box, ward inflow in the near surface layer in the latitude a further small net inflow of 1.2 Sv throughthe 12øN range 8.5ø-10.5øN passedpartially northward through section(C), and a smallnet outflowof 1.0 Sv through the Socotra Passageand recirculated partially eastward the 57ø E section(D), i.e., a total of 2.4 Sv net inflow along the southern slope of Socotra. Part of this in- into the box. This number has to be comparedwith the flow also servedto supply the southwardcoastal Somali sum of the absolutetransports through the box bound- Current, althoughno observationscould be taken in the ariesfor the top layer, which is 16.8 Sv, i.e., a reasonable coastal region to quantify this fraction. balance, where on top one has to considerthat waters At the subsurfacelevel, below •100 m, a southward are transformed within the box out of the ao-25.0 cat- undercurrent inflow through the Socotra Passagewas egory into denser water mass classes,requiring a net observed that was concentrated on the African slope convergencein this layer. and had its current maximum at the salinity maximum In the ao=25.0-27.3 layer a net southward Socotra core of the RSW. The moored current observations in Passageoutflow of 2.2 Sv is nearlybalanced by the com- the passagesuggest that maximum RSW inflow into the bined4.2 Sv inflowthrough the 12øNsection (C) minus Indian Ocean through this passageoccurs during the 1.0 Sv outflowthrough the 57øE section(D), leaving winter monsoon, in agreement with the winter maxi- a small net divergenceof 0.8 Sv. This has to be com- mum observedby Murray and Johns[1997] in the Red pared with the total absolutetransports through the Sea outflow. Westward undercurrent flow was also ob- box boundariesin this layer of 40.5 Sv, i.e., a very rea- served below the eastward surfaceoutflow just south of sonableresidual. Overall then the box budgetsfor the Socotra. This undercurrent did not carry high-salinity upper 700 m appear to be reasonablywell closed,but RSW properties, thus indicating that the main RSW the agreementof upper layer convergenceversus lower inflow route into the Indian Ocean is the Socotra Pas- layer divergencewith the above water massconversion sage. argument can, of course, be coincidental if one consid- SCHOTT AND FISCHER: WINTER MONSOON CIRCULATION 6373

ers the large variability observed(e.g., Figure 5) over a $heyte, 1997; Vinayachandranand Yamagata,1998]). period of more than a week that it took to carry out the Their most active phase appears to be at and after the observations around that box and if one considers the end of the winter monsoon; that is, they may not in- likely errors in the ADCP and LADCP measurements, terfere much with the Sverdrup setup in the early and whichare of the orderof a few cm s-• ([Fischerand full-blown phase of the winter monsoonperiod. Visbeck,1993]). 5.3. Nonseasonal Variability 5.2. Response to Winter Monsoon There are two kinds of variability to be considered Wind Stress in conjunction with the interpretation of our moored A qualitative argument for westward inflow into the and shipboard observations: intraseasonal variations of westernboundary regime of the Arabian Sea in the 5ø- a period of a few weeks that may have causednonsyn- l0 ø latitude range might be derived from the climato- optic distortionsduring our ship survey and interannual logical wind stressforcing. The wind stresscurl in win- variability that may be related to global teleconnection ter is cyclonicover the interior of the Arabian Sea south patterns. As far as interannual variability is concerned, of the axis of the Northeast Monsoon shifting to anticy- it would, of course,be interestingto know how representative the winter monsoon circulations of 1995-1996 and clonic north of the Northeast Monsoon wind maximum off the Arabian Peninsula. The correspondingSverdrup 1997-1998actually were of the averagepatterns. Inspec- transport, assuming quasi-stationary response,requires tion of the Florida States University (FSU) wind stress a zonal inflow of •020 Sv between •-5øand 9øN (using fields([Leglet et al., 1989])for the Arabian Sea shows Hellerman and Rosenstein[1983] winds) into the So- fairly small (order 20%) year to year variationsof the mali Current regime, resulting in a boundary current winter monsoonstresses. By contrast, in the equatorial divergencein that latitude band, which is in agreement zone, wind stressanomalies are of the same magnitude with the northward upper layer flow in our observa- as the seasonalmeans themselves, and large interannual tional area. variations of the Equatorial Jet and Undercurrent have It is, however, questionable to what degree quasi- beenreported [e.g., Reppin et al., 1999].More observations from the northwest Indian Ocean need to become stationary response to Sverdrup dynamics is applica- available before firm conclusions as to the interannual ble in the Arabian Sea. The responsemechanism is by barotropic and baroclinic Rossby waves, and what the differencesamong the winter monsoonsreported on here will be possible. time lag between curt var'mbility and western boundary circulation is depends on the distance the Rossby Regarding the energetic current fluctuations in the waves have to travel to the western boundary. Because period range of a few weeks(Figures 4 and 5) work is of the low propagation speedsof the baroclinic waves, still in progressas to their interpretation. One such it was generally assumedthat there could be no quasi- study has determined that a large fraction of this vari- stationary responsein the off-equatorial oceans at the ance during the summer monsoonsof !995 and 1996 was seasonaltimescale. On the other hand, observationsof made by long baroclinic Rossbywaves that were corre- the past decadein the subtropical North Atlantic did, in lated with wind stress curl variability over the interior fact, show([Lee et al., 1996])that eastof the Bahamasa of the Arabian Sea (D. Quadfasel,personal communi- strong seasonalboundary current cycle exists, in agree- cation, 1999). These were, however,fluctuations super- ment with modelstudies ([BSning et al., 1991]),which imposed on the existing Great Whirl circulation, not those potentially responsiblefor its onset at the begin- is in fairly close correspondenceto a quasi-stationary Sverdrup response. This was attributed to the fact that ning of the summer monsoon. Another ongoingstudy a large fraction of the responsewas generated in the investigatesthe instabilities of the Great Whirl as a po- western basin by barotropic and first-mode baroclinic tential energy sourceof shorter-periodvariability in the waves with small time lags to the western boundary. northern Somali Current regime. These results will be reported elsewhere. Since our location here is close to the equator and the basin is small, quasi-stationaryresponse might, to a substantial degree, be accomplished. 5.4. Model Comparison Comparedto the North Atlantic case,a further com- A numberof studieswith high-resolutionnumerical plicationin the westernIndian Oceanmight be remote modelshave been carried out in recentyears to inves- forcingby wavesfrom the Indian subcontinent([Mc- tigate the seasonaland interannualvariability of the Crearyet al., 1993]). As mentionedin the beginning, Indian Ocean, includingthe Arabian Sea and its ex- theseare presumablyboundary waves traveling around changewith the Indian Oceanat large. Surfaceveloc- the Bay of Bengal and then around Sri Lanka into the ities out of the high-resolutionmodel of $emtner and Arabian Sea,where they radiate Rossbywaves. On the Chervin[1992] that wereanaylzed by [Garternichtand basis of model studies it is claimed that these bound- Schott,1997, Figure 2] showeda wintermonsoon cir- ary wavesare responsiblefor the LaccadiveHigh dur- culation pattern in the northwest Arabian Sea similar ing the late phaseof the wintermonsoon ([Shankar and to what was described here on the basis of our observa- 6374 SCHOTT AND FISCHER: WINTER MONSOON CIRCULATION

25øN

(a)

20øN

15øN

10øN

x 50 cm/s

5øN

Eq.

5os

10øS 40øE 45øE 50øE 55øE 60øE 65øE 70øE 75øE 80øE

25øN

20øN

15øN

10øN

5øN

Eq.

5øS

10øS 40øE

Figure 14. Monthly mean surfacecurrents for January 1996 from the Parallel Ocean Climate Model4 (POCM-4) (B. Semtner,personal communication, 1999) driven by dailyEuropean Centre for Medium-RangeWeather Forecasts (ECMWF) operationalsurface winds and fluxesfor the (a) surfacelayer and (b) undercurrentlayer, 360-510 m depth. SCHOTT AND FISCHER: WINTER MONSOON CIRCULATION 6375 tions. An updatedversion of that model,now drivenby they mergeinto the SomaliCurrent, providing Somali daily operationalEuropean Centre for Medium-Range undercurrentinflow from a large distance. Out of the Wheather Forecasts(ECMWF) winds and fluxes, is southward Somali undercurrent an eastward flow across shown in Figure 14. the Arabian Sea emergesthat can be traced over to In- Although presentedas monthly mean currents,de- dia. Overall, a complicatedcirculation pattern results, rived from a time serieswith three-day resolution, this in generalagreement with the observationsalong the model circulation is much more structured than that in eastern and western margins. the earlier version,which was driven by monthly-mean In othermodels the general patterns may look differ- winds rather than daily winds as in the present ver- ent. We were shownresults from the multilayer reduced sion. Yet the generalfeatures are similar to the earlier gravitymodel of the Naval ResearchLaboratory, (J. version. The westward surface flow is most intense at Kindle,personal communication, 1999) which is driven low latitudes with lit•tle indication of a large-scale cy- by the actual Fleet Numerical OceanographyCenter cloniccirculation around the ArabianSea (Figure 14a). (FNOC) windsfor 1997-1998,and of the Miami isopyc- At the western end, three inflow branchessupply the nicmodel (O. Esenkov,personal communication, 1999). Somali Current flow along the coast. Both varied in many respectsfrom Figure 14, but the The main inflow, in agreementwith the aboveSver- main features observed for the northern Somali Cur- drup dynamicsargument, flows westwardin the 4ø- rent regime and reproducedin Figure 14 are present, 7øN latitude range. It feedsa coastaldivergence with suchas generalwestward inflow in 6ø-10ølatitudeband equatorwardcurrents south of 6øN and a northward and divergenceat the coast,surface outflow through the boundarycurrent off northernSomalia. A secondinflow SocotraPassage and northeastwardrecirculation south branch encountersthe coast just north of the equator, of Socotra,and undercurrentinflow through the Soco- enhancingthe cross-equatorialSomali Current trans- tra Passageand westwardundercurrent south of Soco- port. The northwardSomali Current branchis further tra. Theseappear therefore to be robustfeatures of the strengthenedby a third inflow branchin the 9ø-11øN winter monsoon circulation, but their physics require range. This northwardboundary current exits mainly additional studies. throughthe SocotraPassage into the Gulf of Aden. In addition,the northerninflow branch supplies northeast- Acknowledgments. We thank C. Lichtenberg for his help in the current and transport analysis, M. Hamann for ward near-surface recirculation around Socotra. assistancewith the graphs and critical reading of the revi- In the southeastern Arabian Sea the NMC flows west- sion, and J. Reppin for his assistancewith the model fields. ward betweenthe equator and 6øN, and part of it con- Financial support by the Federal Minister of Science,Edu- tinuesto supply the main inflow branch of the Somali cation and Research(BMBF) was receivedunder contracts Current, while part circulatesanticyclonically around 03F0157A and 03G0128A as part of the German WOCE a large dynamic high east of 65øE and then recircu- program. lates northward toward the Indian coast. This ridge resemblesthe LaccadiveHigh of Bruceet al. [1994]and References is obtained in several model simulations already men- BSning,C. W., R. DSscher,and R. G. Budich,Seasonal tioned. A spinofffrom that flow followsthe Indian coast transportvariation in the westernsubtropical North At- northwestwardas one, if weak, branch of a cycloniccir- lantic: Experimentswith an eddy-resolvingmodel, J. culationregime, in agreementwith winter monsoonob- Phys. Oceanogr.,21, 1272-1288, 1991. Bruce,J. G., M. Fieux, and J. Gonella,A noteon the contin- servations[Sherye, 1991]. The modelshows only weak uanceof the Somalieddy after the cessationof the South- currentsalong the northern boundary,which are west- west monsoon, Oceanol. Acta, J, 7-9, 1981. ward off the NE Arabian peninsula and vanish in an Bruce, J. G., D. R. Johnson,and J. C. Kindle, Evidence eddy field farther west. The northeastwardmodel re- for eddy formation in the eastern Arabian Sea during circulation past Socotra also vanishesinto that eddy the northeast monsoon,J. Geophys.Res., 99, 7652-7664, 1994. field. Donguy,J., and G. Meyers,Observations of geostrophic At the 360-510 m level the model flow shows a south- transportvariability in the westerntropical Indian Ocean, ward undercurrent through the Socotra Passage and Deep Sea Res., Part I, •2, 1007-1028, 1995. aroundSocotra (Figure 14b), in agreementwith obser- Fischer,J., and M. Visbeck,Deep velocityprofiling with vations. Of[ India a southward undercurrent occurs in self-containedADCPs, J. Atmos. OceanicTechnol., 10, 764-773, 1993. the model, again in agreementwith the observationsof Fischer,J., F. Schott,and L. Stramma,Currents and trans- Sherye[1991]. This undercurrentdevelops westward af- portsof the Great Whirl-SocotraGyre systemduring the ter passingthe Maledives and mergesin a current core summer monsoon,August 1993, J. Geophys.Res., 101, that feeds into the Somali Current just north of the 3573-3587, 1996. equator. Garternicht,U., and F. Schott, Heat fluxesof the Indian Oceanfrom a global eddy-resolvingmodel, J. Geophys. The southwardundercurrent throughflow through the Res., 102, 21,147-21,159, 1997. SocotraPassage and westwardundercurrent inflow south Hellerman, S., and M. Rosenstein,Normal monthly wind of Socotra are linked upstream to a westward boundary stress over the world ocean with error estimates, J. Phys. current along the Arabian Peninsula, and downstream Oceanogr.,13, 1093-1105, 1983. 6376 SCHOTT AND FISCHER: WINTER MONSOON CIRCULATION

Lee, T. N., W. E. Johns, R. J. Zantopp, and E. R. Fillen- Summer monsoon responseof the northern Somali Cur- baum, Moored observationsof western boundary current rent, Geophys.Res. Lett., 2•, 2565-2568, 1997. variability and thermohaline circulation at 26.5øN in the Semtner,A. J., and R. M. Chervin, Ocean generalcircula- subtropical North Atlantic, J. Phys. ¸ceanogr., 26, 962- tion from a global eddy-resolvingmodel, J. Geophys.Res., 983, 1996. 97, 5493-5550, 1992. Leglet, D. M., I. Navon, and J. J. OBrian, Objective analy- Shankar,D., and S. R. Sheyte, On the dynamicsof the Lak- sis of pseudostressover the Indian Ocean using a direct- shadweephigh and low in the southeasternArabian Sea, minimization approach, Mon. Weather Rev., 26,709-720, J. Geophys.Res., 102, 12,551-12,562,1997. 1989. Sherye,S. R., The coastalcurrent off westernIndia during Le Provost, C., F. Myard, J.-M. Molines, M. Genco, and the Northeast Monsoon,paper presentedIAPSO confer- F. Rabilloud, A hydrodynamic oceantide model improved ence,Int. Assoc.for Phys. Sci. of the Oceans,Vienna, by assimilating a satellite altimeter-derived data set, J. 1991. Geophys. Res., 103, 5513-5529, 1998. Swallow,J. C., and M. Fieux, Historicalevidence for two McCreary, J.P. J., P. K. Kundu, and R. L. Molinari, gyresin the SomaliCurrent, J. Mar. Res., JO, 747-755, A numerical investigation of dynamics, thermodynamics 1982. and mixed-layer processesin the Indian Ocean, Prog. in Swallow,J. C., R. L. Molinari,J. G. Bruce,O. B. Brown, ¸ceanogr., 31, 181-244, 1993. and R. H. Evans,Development of near-surfaceflow pat- Molinari, R. L., D. Olson, and G. Reverdin, Surface cur- tern and water mass distribution in the Somali Basin rent distributions in the tropical Indian Ocean derived in responseto the southwestmonsoon of 1979,J. Phys. from compilations of surface buoy trajectories, J. Geo- Oceanogr.,13, 1398-1415, 1983. phys. Res., 95, 7217-7238, 1990. Swallow,J. C., F. Schott, and M. Fieux, Structureand Murray, S. P., and W. Johns, Direct observationsof seasonal transport of the East African Coastal Current, J. Geo- exchange through the Bab el Mandab Strait, Geophys. phys.Res., 96, 22,254-22,267, 1991. Res. Lett., 2• , 2557-2560, 1997. Vinayachandran,P. N., and T. Yamagata, Monsoonre- Quadfasel, D., Low-frequency variability of the 20øC sponseof the sea around Sri Lanka: Generation of thermal isotherm topography in the western equatorial Indian domesand anticyclonicvortices, J. Phys.Oceanogr., 28, Ocean, J. Geophys.Res., 37, 1990-1996, 1982. 1946-1960, 1998. Quadfasel, D., and F. Schott, Southward subsurface flow Visbeck,M., andJ. Fischer,Sea surface conditions remotely below the Somali Current, J. Geophys. Res., 33, 5973- sensedby upward-looking ADCPs., J. Atmos. Oceanic 5979, 1983. TechnoI., 12, 141-149, 1995. Reppin, J., F. Schott, J. Fischer, and D. Quadfasel, Equa- Wyrtki, K., OceanographicAtlas of the InternationalIndian torial currents and transports in the upper central Indian OceanEzpedition, 531pp., Natl. Sci.Found. Washington Ocean: Annual cycle and interannual variability, J. Geo- D.C., 1971. phys. Res., 10•, 15,495-15,514, 1999. Schott, F., J. C. Swallow, and M. Fieux, The Somali Current at the equator: annual cycle of currents and transports in the upper 1000 m and connection to neighbouring lati- F. A. Schott, and J. Fischer,Institut fiir Meereskundean tudes, Deep Sea Res., Part A, 37, 1825-1848, 1990. der Universitiit Kiel, DiisternbrookerWeg 20, 24105 Kiel, Schott, F., J. Reppin, D. Quadfasel, and J. Fischer, Cur- Germany,(iõ[email protected]; [email protected]) rents and transports of the Monsoon Current south of Sri Lanka, J. Geophys. Res., 99, 25,127-25,141, 1994. (ReceivedMarch 9, 1999; revisedOctober 7, 1999; Schott, F., J. Fischer, U. Garternicht, and D. Quadfasel, acceptedOctober 29, 1999.)