JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 102, NO. C5, PAGES 10,391-10,422, MAY 15, 1997
Seasonal variations of thermocline circulation and ventilation in the Indian Ocean
Yuzhu You
Center for Climate SystemResearch, University of Tokyo, Tokyo Laboratoired'Oc•anographie Physique, Museum National d'Histoire Naturelie, Paris
Abstract. Two seasonalhydrographic data sets, including temperature, salinity, dissolved oxygen,and nutrients, are used in a mixingmodel which combines cluster analysis with optimum multiparameteranalysis to determinethe spreadingand mixing of the thermoclinewaters in the Indian Ocean.The mixingmodel comprises a systemof four majorsource water masses, which were identifiedin the thermoclinethrough cluster analysis. They are IndianCentral Water (ICW), North IndianCentral Water (NICW) interpretedas agedICW, AustralasianMediterranean Water (AAMW), andRed SeaWater (RSW)/PersianGulf Water (PGW). The mixingratios of these watermasses are quantifiedand mapped on four isopycnalsurfaces which spanthe thermocline from 150 to 600 m in the northernIndian Ocean, on two meridionalsections along 60øE and 90øE, andon two zonalsections along 10øS and 6øN. The mixingratios and pathways of the thermocline watermasses show large seasonal variations, particularly in the upper400-500 m of the thermocline.The mostprominent signal of seasonalvariation occurs in the Somali Current,the westernboundary current, which appears only duringthe SW (summer)monsoon. The northward spreadingof ICW intothe equatorial and northern Indian Ocean is by way of the SomaliCurrent centeredat 300-400 m on the c•0=26.7isopycnal surface during the summermonsoon and of the EquatorialCountercurrent during the NE (winter) monsoon.More ICW carriedinto the northern IndianOcean during the summermonsoon is seenclearly in the zonalsection along 6øN. NICW spreadssouthward through the westernIndian Ocean and is strongerduring the wintermonsoon. AAMW appearsin bothseasons but is slightlystronger during the summerin theupper thermocline.The westward flow of AAMW is by wayof theSouth Equatorial Current and slightly bendsto the northon the c•0=26.7isopycnal surface during the summermonsoon, indicative of its contributionto thewestern boundary current. Outflow of RSW/PGWseems effectively blocked by the continuationof strongnorthward jet of the SomaliCurrent along the westernArabian Sea duringthe summer,giving a rathersmall contribution of only up to 20% in the ArabianSea. A schematicsummer and winter thermocline circulation emerges from this study. Both hydrography andwater- massmixing ratios suggest that the contributionof the waterfrom the SouthIndian Oceanand from the Indo-Pacificthroughflow controls the circulationand ventilationin the westernboundary region during the summer.However, during the winterthe wateris carriedinto theeastern boundary by theEquatorial Countercurrent and leaks into the eastern Bay of Bengal, fromwhere the water is advectedinto the northwestern Indian Ocean by theNorth Equatorial Current.The so-calledEast Madagascar Current as a southwardflow occursonly during the summer,as is suggestedby bothhydrography and water-mass mixing patterns from this paper. Duringthe winter (austral summer) the current seems reversal to a northwardflow alongeast of Madagascar,somewhat symmetrical to the SomaliCurrent in the north.
1. Introduction Indian Ocean can reach as deep as 500 m, while Wyrtki [1973] suggestedthat impact of the southwestmonsoon is noticeableas The main thermoclinelayer of the oceanis in generaltreated as deep as 1000 m at the Somali Current region. Tchernia [1980] a layer that is not subjectto seasonalvariability, becauseit is proposedan even deeper effect of seasonalvariation on water situatedwell below the seasonalsurface layer. Betweenthese two massproperties down to 2000 m. Probablythe most profound layersthere is a densitytransition which Defant [ 1961] calledthe effect of seasonal monsoons is on the Somali Current, which "barrierlayer." However, since climatology of the IndianOcean is appearsonly duringthe summer.The greatdifferences among the dominatedby two distinct reversal monsoons,the southwest(or estimatedtransports from the Pacific to the Indian Ocean, ranging summer) monsoon and the northeast (or winter) monsoon, from 1.7 Sv [Wyrtki, 1961] to 18 Sv [Cox, 1975, after Gordon, seasonalvariations throughout the main thermoclineof the Indian 1986]. The latter also suggeststhat strongseasonal variation must Ocean are significantand cannotbe ignored. Colborn [1975] play a role in thermoclinestructures. found that monsoonal effect on thermal structure of the northern Several factors, including distinct seasonalmonsoons and the effects of a complicated geography, the northern Indian Ocean Copyright1997 by the AmericanGeophysical Union. being blockedby the Eurasianlandmass, the lack of a Subtropical ConvergenceZone, and an interoceanicthroughflow including its Papernumber 96JC03600. distinct hydrological front, lead to the formation of a unique 0148-0227/97/96JC-03600509.00 thermoclinewater- massstructure in the Indian Ocean.Sverdrup 10,391 10,392 YOU: INDIAN OCEAN SEASONAL THERMOCLINE CIRCULATION et al. [1942] identified three water massesand discussedtheir basiclarge-scale structure of the thermoclinein the Indian Ocean formation mechanisms in the main thermocline of the Indian north of 40øS are listed in Table 1, which also includes a Ocean: (1) Red Sea Water (RSW), formed in the Red Sea; (2) compendium of the same water masseswith different names Indian Central Water (ICW, also called SubtropicalSubsurface accordingto variousauthors. A relatively recentgeneralized Water by Warren et al. [1966] and SubantarcticMode Water by IndianOcean temperature - salinity(T-S) diagramof Emeryand McCartney [1977]) formed in latitudes 40ø-45øSduring late Meincke [1986] is shownin Figure 1. There is only one water winter by convectiveoverturning [Wyrtki, 1973; Colborn, 1975]; massin Table 1 thatwe havenot mentionedyet, theNorth Indian and (3) Indian Equatorial Water (IEW), formed in the western Central Water (NICW), which will be discussedlater. equatorial region through some unspecified mechanism. A closer examination of these water masses, based on their Marhayer [1975] added two more water massesto the Indian regionsof formation,suggests that the main thermoclineof the Ocean north of 40øS:Bengal Bay Water (BBW), which is only IndianOcean north of 40øSis actuallydominated by only three found in the surface layer, and Timor Sea Water (called source water masses, ICW, AAMW, and RSW/PGW. $harma Australasian Mediterranean Water (AAMW) by You and [1976], Quadfasel and Schott [1982], and You and Tomczak Tornczak[1993]), which originates in the deep basins of the [1993] have arguedthat the IEW of Sverdrupet al. [1942] is Indonesianarchipelago. The latter was missedby Sverdrupet al. actuallya mixtureof water massesfrom the northernand southern [1942] becausethey lacked data for the easternIndian Ocean Indian Ocean and from the Pacific. The BBW definedby [Colborn, 1975]. In a recentreview, Emery and Meincke [1986] Marhayer [1975] is the resultof excessprecipitation and river included anotherwater mass,Arabian Sea Water (ASW), which is runoff over evaporationresulting in a low salinity(between 32 influenced by outflow of Persian Gulf Water (PGW) and is and 33 psu). This low salinity causesit to be isolatedfrom the distinguishedfrom RSW. These water masseswhich form the water in the main thermoclineby a sharphalocline located
Table 1. An Inventory of Water Mass Definitionsin the Thermoclineof the Indian Ocean
Water Mass Study T-S Characteristics Notes T, øC S, psu
Indian Central Water (ICW) a Sverdrupet aI. [1942] 7.00- 16.00 34.70- 35.70 Mamayev [ 1975] 16.00 35.60 Emery and Meincke [1986] 8.00- 25.00 34.60- 35.80 Warren et aI. [ 1966] 7.00- 15.00 34.70- 35.50 McCartney [ 1977] Formedby late winter convectiveoverturning at 40ø-45øS
Australasian Mediterranean You and Tomczak[1993] 5.55- 14.89 34.52- 34.77 Water (AAMW) b Mamayev [1975] 25.00 34.50 Emery and Meincke [1986] 8.00- 23.00 34.40- 35.00 Wyrtki [1961] After Warren [ 1981] Rochford[ 1966]
North Indian Central Water Gordon [1986] Defined as thermocline (NICW) water below the surface water at the Bay of Bengal. Youand Tomczak[1993] 7.80- 15.72 34.84- 35.10 The same as Gordon but definedas aged ICW
Red Sea Water (RSW) Sverdrupet aI. [1942] >7.00 >35.00 Definedas oxygenminimum core. Mamayev [ 1975] 23.00 40.00 Defined as intermediate water.
Bengal Bay Water (BBW) Mamayev [1975] 25.00 33.80 Defined as surface low- salinity water. Emery and Meincke [ 1986] 25.00- 29.00 28.00- 35.00 Defined as surface low- salinity water.
Arabian Sea Water (ASW) Emery and Meincke [1986] 24.00- 30.00 35.50- 36.80
Indian EquatorialWater Sverdrupet aI. [1942] 5.00- 16.00 34.90- 35.20 (IEW) Mamayev [1975] 25.00 35.30 Emery and Meincke [1986] 8.00- 23.00 34.60- 35.00
a. CalledSouth Indian Subtropical Water by Mamayev[ 1975],South Indian Central Water by Emeryand Meincke[ 1986], SubantarcticMode Waterby McCartney[1977], and SubtropicalSubsurface Water by Warrenet aI. [1966]. b. CalledTimor Sea Water by Mamayev[1975], Banda Intermediate Water by Rochford[1966], Banda Sea Water by Wyrtki [1961], and IndonesianUpper Water by Emery and Meincke [1986]. YOU: INDIAN OCEAN SEASONAL THERMOCLINE CIRCULATION 10,393
BBW defined as an independentsource water massbecause it lacks a ' ' /• ' ! ' / commonformation process. They found that when they did not ..,./ include NICW in their mixing model they derived a totally /I _-?'- different water - masscontribution which couldnot be interpreted •5 - c?/ physically. Using an isopycnal-mixing-dominatedsurface with three conservative tracers (temperature, salinity, and initial phosphate)they showedthat NICW is indeedthe agedICW (see
< lO You and Tomczak[1993] and their Figure A2). Gordon [1986] used NICW to symbolically represent the North Indian thermocline water. Thus from Table 1 a total of four water ,,wF ! masses, ICW, NICW, AAMW, and RSW/PGW, are chosen to • 5 - 1NIL.'"'> define thermocline circulation and ventilation in the Indian Ocean.The characteristicsof thesewater massesin T-S diagrams / / and in other property-propertydiagrams of temperatureagainst o oxygen and nutrients are shown in Figure 8. A detailed I, CDW • l • •• •:• descriptionof the definitionsof thesewater masseswill be given 34 35 36 in section 4. Among these source water masses,RSW/PGW S A LINITY contributes the least. Because the northern Indian Ocean does not Figure 1. A generalizedT-S diagramfor the main watermasses have a SubtropicalConvergence Zone, RSW/PGW has limited in the Indian Ocean. The water masses are abbreviated as follow: sources.In addition,as NICW is considereda water massof aged BBW, BengalBay Water;ASW, ArabianSea Water; IEW, Indian ICW, the thermocline water of the northern Indian Ocean has to EquatorialWater; IUW, IndonesianUpper Water; SICW, South be replenishedfrom the tropicsor even farther south.Warren et Indian Central Water; AAIW, Antarctic Intermediate Water; IIW, IndonesianIntermediate Water; RSPGIW, Red Sea- PersianGulf al. [1966] seem to be the first to put forward the idea that the IntermediateWater; and CDW, CircumpolarDeep Water. The thermoclineof the northernIndian Ocean may be ventilatedfrom figureis adoptedfrom Emery and Meincke[1986, their original the south. Colborn [1975] noted that there is some degree of Figure5, permittedby Gauthier-Villars Editeur]. mixing between BBW and AAMW in the Bay of Bengal and pointed out that some diapycnal mixing must occur in the northernIndian Ocean. Toole and Warren [1993] suggestedthat the deep and bottom waters could be largely converted to between 50 and 100 m. Therefore, BBW defined by Mamayev thermocline and surface waters. Recently, You [1996] found [1975] is clearly restrictedto only the surfacelayer (also see strong dianeutral upwelling in the northern Indian Ocean and Figure 1) and will be excludedfrom the discussionof the main downwelling in the southernIndian Ocean, and concludedthat the thermoclinewater in this paper. AAMW, also known as the dianeutral mixing plays a vital role in achieving water mass conversion and thermocline ventilation in the northern Indian Banda IntermediateWater accordingto Rochford [1966], is a Ocean. majorwater mass of the world oceanand, as part of the warm water routeof North Atlantic Deep Water (NADW) returnflow, In an annual mean situation, You and Tomczak [ 1993] noted that the movement of ICW toward the northern Indian Ocean was playsa vitalrole in Gordon's[1986] global model of thermocline circulation. AAMW showsas a tongueof low salinityand high evident on the c•0=26.7isopycnal surface (see their Figures 9a, silicateextending westward from Indonesiain the International 13a and 15a), which is consistentwith Warren et al. [1966, p. Indian Ocean Expedition(IIOP) atlas [Wyrtki, 1971]. ICW is 842] who statethat "northward-movingSubtropical Subsurface formed at the surface in late winter at latitudes 40ø-45øS. As a Water (namely ICW) appears to penetrate well north of the result of subductionprocesses and subsequentmixing, the equator."You and Tomczak[1993] proposeda transitionpath of propertyrelationships of ICW are nearly linear [Tomczakand ICW: subductionin the SouthIndian Ocean at about 40ø-45øS, Large, 1989]. ASW is actuallya mixtureof PGW andRSW [You advection with the subtropicalgyre, and exit to the northern and Tomczak,1993]. Accordingto Sharma [1976] the effect of Indian Ocean as part of western boundary currents. The PGW on thermocline water in the Arabian Sea is relatively transformationof ICW to NICW is a very slow process, smaller than that of RSW. You and Tomczak [1993] found, from accompaniedby a significant fall in oxygen values due to an annual mean data set, that the influence of RSW in the upper consumptionby oxidationof organicmatter and a corresponding thermocline is mainly restricted to the Arabian Sea, where it increaseof nutrientconcentrations. The thermoclinein theBay of contributesup to 30% to the thermoclinewater. In contrast,Olson Bengal containsthe oldestcentral water - NICW. The conclusion et al. [1993], usingchlorofluorocarbon (CFC) data,found an even was drawnin favor of propertydistribution along the transition smaller contribution of RSW, less than 10%. path of ICW on the isopycnal-mixing-dominatedsurface, From the preceding brief discussionof water mass structure, o0=26.7.ICW andNICW haveidentical temperature and salinity three water masses, IEW, BBW and ASW, can be excluded here but differentoxygen and nutrientlevels. You [1996] could trace in view of their formation sources.The discussionin this paper thiscentral water formation on the surfacewith mappeddensity concentrates on main thermocline water itself. RSW and PGW are ratioRp ( Rp = or0z/ [•Sz)' whereot= - p-1/) 0/ 30 isthe consideredas a joint sourcewater mass(even thoughPGW is not appropriatethermal expansion coefficient, [•= p- 130/ 3Sis the listed in Table 1), since they have similar water mass halinecontraction coefficient, and Oz, and Sz are vertical gradients characteristics.Recently, You and Tomczak[1993] introduceda of potentialtemperature and salinity. He founda steadyincrease virtual water mass,NICW, consistingof agedICW from the main of Rp from2 at thecentral water of subtropicsto 20 nearthe Bay thermoclinelayer in the Bay of Bengalbelow BBW. They argued of Bengal along the path. The westwardflow of AAMW across that the introduction of this water mass is required by their the Indian Oceanwas also clearly shownin You and Tomczak's optimum multiparameter (OMP) analysis, although it is not [1993] mixing model results. However, the reversed seasonal 10,394 YOU: INDIAN OCEAN SEASONAL THERMOCLINE CIRCULATION monsoons mentioned at the beginning are known to have a 40ø E 60ø 80ø 100ø 120ø 140ø E profoundeffect on the Indian Oceancirculation and watermass I I I I I I , I I , , station dmtnbution ,' structure. The extent of this monsoonal effect on thermocline o 20 N- summer (• --20øN circulation and ventilation in their mixing model is not known, sinceYou and Tomczakused only an annualmean data set. _ Under the influence of two distinct monsoon regimes, the transition of ICW to the northern Indian Ocean may undergo o-• strongseasonal changes. The throughflowof AAMW transported into the AgulhasCurrent should also be affectedby the monsoon reversals, with stronger westward flow during the southwest 20ø •:e;.- ?b ..... •"--:3': ...... •3•• '- •oø monsoonin the upper thermocline,because South Equatorial Current reaches its maximum during the summer monsoon. ß7 ::. :....?..;.:; - During the winter monsoonthe SouthEquatorial Current does not 40øS 40 S belong to the monsoon gyre [Wyrtki, 1973]. Wyrtki [1973] suggestsa much strongerimpact on the water massesbelow the thermocline by the southwest monsoon compared with the 40ø E 60ø 80ø t 00ø 120ø 140ø E northeastmonsoon. He pointedout that the depthof circulationis greatest during the southwest monsoon. This affects the 40ø E 60ø 80ø 100ø 120ø 140øE movement of water masses far below the thermocline. The Somali Current formation causesa strongbaroclinic adjustmentof the structurewhich is especiallypronounced in the upper400 m and noticeableas deep as 1000 m. Since the main thermoclinewater of the northernIndian Oceanis locatedin the 150 to 600 m depth range, well within the depthof monsoonalinfluence, the seasonal changesof main thermoclinewater massesin responseto the monsoonal changes have to be considered. The thermocline circulationand ventilationstudied in this paperare determinedby examiningwater massmixing amongthe four water mass of the system:ICW, NICW, AAMW, and RSW/PGW. Therefore the seasonaleffect on thesewater masses,as definedthrough source water types, will be quantified. Afterward, calculation of the mixing ratios among those sourcewater massesis carriedout to accountfor a seasonalsignal. In section2, two data sets,which representtwo monsoon seasons,and the methodsused in this Figure 2. Distributionof hydrographicstations on the {•0=25.7 paper are briefly described.The distributionof hydrographic isopycnalsurface, during (a) the southwestmonsoon (May- parametersincluding pressure,potential temperature,salinity, October)and (b) the northeastmonsoon (November-April). Up to dissolvedoxygen, phosphate, and silicateon individualisopycnal five parameterswere measuredat each station: temperature, surfaces for the two seasons is shown in section 3. The definition salinity,oxygen, phosphate, and silicate. of the sourcewater typesfor the two seasonsis discussedin detail in section4. The resultsof the water massmixing analysesfor the two seasons are presented in section 5. The summary and Thesehistorical hydrographic data include (1) theNatiohal discussion in section 6, based on a schematic of summer and OceanographicData Center (NODC) data throughAustralian winterthermocline circulation, conclude the paper. OceanographicData Center (AODC) and preparedfrom the Universityof Sydney[You and Tomczak,1993], (2) the archived 2. Data and Methods data availablein the Center for Meteorologyand Physical Oceanography,Massachusetts Institute of Technology(Y. You, Because of the limited available data resourcesat that time, You an unpublishedmanuscript submitted to InternationalWOCE and Tomczak [1993] used a single data set of historical News Letter, 1996), and (3) somerecently completed cruises hydrography,covering all seasonsand years,to studythe annual includingCharles Darwin cruise29 (CD29) datafrom the Woods mean of thermocline circulation and ventilation in the Indian Hole OceanographicInstitution [Toole and Warren, 1993], the Oceannorth of 40øS. Sincethen therehave been opportunities to MonsoonArabian Sea Investigations(MASAI) data from the accessadditional data from both historical and update cruises Universityof Maimi [Olson et al., 1993] and the FrenchIndien obtained for this ocean basin. The combined total data set has GeochimieOcean (INDIGO) data [Jamouset al., 1992]. These been synopticallydivided into two seasonsaccording to their two seasonaldata setscontain physical-chemical observations measuring time, the southwest monsoon (May-October) and recordedat discretedepth levels including temperature (T, degree northeastmonsoon (November-April) (hereinafter"summer" and Celsius),salinity (S, practicalsalinity units (psu)), oxygen (02, "winter" refer to the northern hemispheresummer and winter millilitersper liter), phosphate(PO4, microgram atom per liter), seasons).As a result,there is reasonablygood coverage for both nitrate(NO3, microgramatom per liter), andsilicate, (H4SiO4, seasons(see Figures 2a and 2b) on the uppermostisopycnal microgram atom per liter). Since most stationsdo not contain surface,{•0=25.7, with a total of 1451 stations,for summerand nitrate observations, these data are not used. All the stations were 1651 for winter. This is reduced to 1299 for summer and 1468 for thenlinearly interpolated for four selectedisopycnal surfaces, winter on the lowermostisopycnal surface, (•0=27.1. The summer {•0=25.7,{•0=26.7, {•0=26.9 and {•0=27.1,as usedby Youand data are not distributedas well as the winter data (Figure 2). Tomczak[ 1993]. YOU:INDIAN OCEANSEASONAL THERMOCLINE CIRCULATION 10,395
A water massis definedhere as a body of water with a common required to adequatelyrepresent the water massesin the Indian formationhistory for all of its containedelements. The permanent Ocean thermoclineand their parametervalues. The methodcan thermocline contains several such water masses, which can generallybe appliedto find naturalgroupings, or clusters,in a penetrateone anotheras a resultof mixing.As watermasses are data set with all parametersrepresented by one matrix. Hencethe physicalentities of finite volumn,they can be mathematically permanentthermocline can be classifiedinto severalgroups, and describedby functionalrelationships between their characteristic with our oceanographicknowledge, the sourcewater types can be propertiesand a set of standarddeviations. A water type is identifiedby parametercombinations or pointsin parameterspace definedas a point in parameterspace; it is only a mathematical (the clustercenter) in the contextof water massanalysis. By constructand does not occupy any volume in parameterspace. applyingcluster analysis on a numberof isopycnalsurfaces in The functionalrelationship of a water massrepresented by its succession,it is possibleto derivea seriesof sourcewater types characteristicproperties can then be describedby an infiniteset of acrossthe density range that spansthe thermoclinelayer and watertypes, which are calledsource water types [Tornczak and approximatesthe parameterrelationships of the unknownwater Large, 1989].In practice,if the functionalrelationship between massesto an anticipatedaccuracy. Readers are referred to You all parametersof a watermass is linear,the watermass can then and To•nczak[1993] for the detailsof the applicationof cluster be describedby a minimumof two sourcewater types. analysis. The variables used in this study are considered to be The sourcewater types determinedas the clustercenters on conservative(following You andTornczak [1993]). This is a individual isopycnalsurfaces are then used to derive the "real" reasonableassumption if the analysisis restrictedto mixingin the source water types through regressionanalysis in property- upperkilometer of the oceanand over horizontaldistances of the property diagrams. The intersectionpoints of the isopycnal order of hundreds of kilometers. On that scale, the effects of surfaces and the fitted lines are then chosen as the "real" source advection and turbulent diffusion clearly outweigh any water types for OMP analysis.Having thereby establishedthe biochemical effects [You and Tornczak, 1993]. However, the source water types and thus the water masses from cluster present analysis is concerned with circulation features on an analysis,one can then calculatemixing proportionswith OMP oceanic scale where biochemical oxygen consumption and analysis.Here is the brief review of the method:it solvesa linear nutrientgain cannotbe ignored.To copewith theseeffects in the systemof mixingequations for eachdata point, in whichall water frame work of OMP analysis, the virtual water mass NICW, massesare representedthrough water types [Tomczak and Large, introducedin section1, is usedto representaged Indian Central 1989; You and Tomczak,1993] and the linear systemfor any Water. As was mentioned above, this "water mass" is defined water samplecan be written as only for the purposeof OMP analysisand not in the senseof our definition above. The main idea here is that when ICW transits to GXg xv R, (1) the northern Indian Ocean through advection and diffusion epipycnally (or epineutrally) or diapycnally (or dianeutrally) whereG is a matrixcontaining the parametervalues which define along the isopycnalsurfaces (or neutral surfaces[see You, 1996]), the sourcewater types; x v is a vectorcontaining the parameter ICW imprintsits age. It is reflectedin the changeof oxygenand valuesfor the sample (the observations); Xgis a vectorcontaining nutrient contentson its paths. In other words, ICW ages as its the relativecontributions, or mixing ratios,of the sourcewater oxygen value decreasesfrom its formationregion to the Bay of typesto the sample;and R is a vectorcontaining residuals. The Bengal.This is becauseICW carriesoxygen-rich water from the water massconservation equation is expressedas southern Indian Ocean to the northern Indian Ocean and its M oxygen is consumedby oxidation of organic matter during its Y_,x =1, gk advectiontoward the Bay of Bengal.The sameidea is appliedin k=l (2) this paper.NICW is introducedto representsuch aged ICW. It is defined in the Bay of Bengal and is distinguishedfrom BBW, whereM is thetotal number of sourcewater types. Equations (1) which is defined only for the surfacelow-salinity water by and(2) form a linearmixing model of all watermasses. Solving Marnayev[1975] andEmery and Meincke[1986] (seeFigure 1). equation(1) by minimizingthe sumof the squaredresiduals leads You and Tornczak[1993] applied a mixing model which to the determination of the minimum of combinescluster analysis with OMP analysis to study the thermocline circulation and ventilation in the Indian Ocean. The T T T methodof OMP analysis,first proposedby Tornczak[1981] and w W(Gx-Xv) furtherdeveloped by Mackaset al. [1987], Tornczakand Large 2 [1989], and You and Tornczak [1993], uses nutrients and dissolvedoxygen, along with temperatureand salinity,to derivea =j= %1Wjm 21n i=E 1G..x l• gi -x vj ' (3) distribution of water masses that matches the observed distributionof propertiesin an objectivelydefinable best fit. The where m is the number of parameters,n is the total number of techniquewas initially developedfor a frontal mixing situation stations,and W representsthe weight attributedto the various (mixingof two watermasses) and later used for thatpurpose. You parameters,reflecting differencesin measurementaccuracy, and Tornczak[ 1993] extendedthe method to an ocean-basin-wide degree of conservativeness,and other processeswhich may mixing of the Indian Ocean thermocline.They used cluster rendersome parameters less reliable than others. These weights analysisfor an objectivedefinition of water typesas a precursor are derived from to OMP analysis.They appliedthe methodon isopycnalsurfaces which are more alignedwith the probablemixing paths,giving W =02./•5. meaningful results of thermocline circulation and ventilation. J I j max' (4) FollowingYou and Tomczak[1993], "clusteranalysis" is applied in this paper to determinethe numberof sourcewater types wheregj is thestandard deviation of parameterj over the entire 10,396 YOU:INDIAN OCEAN SEASONAL THERMOCLINE CIRCULATION
data set (a measurefor the ability of parameterj to resolve 30 40 50 60 70 80 90 100 110 120 130 30 30 differencesin watermass content), defined as a 20
(7J= •- i = Gji- GJ (5) •o
and•Jjrnax is the largest variations for the water masses. Gj is the -10 meanof Gjigiven by
n -20 -20
= 1 ,•.,Gji. (6) Gj •'i=1 -30 -30 Two conservativeparameters, temperature (beneath the mixed -40 -40 layer) and salinity, usually attain the largestweight. This largest 30 40 50 60 70 80 90 100 110 120 130 weight is also allocatedto the mass conservationequation (2), Longitude
since the method of weight calculation describedabove is not 30 40 50 60 70 80 90 100 110 120 130 applicableto massconservation. 30 30 As was mentioned above, five parameters were found to provideuseful information for the presentstudy. This restrictsthe 20 20 number of source water types that can be handled by OMP analysisto a maximum of six, and we mustask whethersix water types are sufficient to representall water massesin the region. The earlier discussionestablished the presenceof three source water masssystems, ICW, AAMW, and RSW/PGW, in the main -10 thermocline of the Indian Ocean. Later, an additional virtual -20 water mass, NICW, was introducedto representthe aged ICW. -20 This is a total of four water masses,but OMP analysiscan handle -30 -30 only three in this particular case. However, as was mentioned above, RSW/PGW provide a relatively small contributionto the -40 -40 thermocline circulation and ventilation, between about 30% 30 40 50 60 70 80 90 100 110 120 130 annually to the Arabian Sea Water estimatedby Youand Tomczak Longitude [1993] and about 10% estimatedby Olson et al. [1993] through CFC data analysis.Therefore, with the limitation of the methodto 30 40 50 60 70 80 90 100 110 120 130 handlethe numberof sourcewater types,RSW/PGW is excluded 30 30 from OMP analysis, although by parallel calculation in the b westernIndian Oceanupper thermoclineits relative contribution 20 could still be estimatedby excludingAAMW. This is becausein the uppermostthermocline, AAMW showsjet-like flow apparent in the easternIndian Ocean, while relatively small amountof its water reachesthe westernIndian Ocean. Consequently,for the western Indian Ocean the contribution by RSW/PGW can be -10 approximatedthrough the calculation of mixing among ICW,
NICW and RSW/PGW. Readers are referred to You and Tomczak -20 -20 [ 1993] for this particularapproach. With five parametersand six source water types, OMP analysisdoes not degenerateinto a -30 -30 uniquely determined system of equations [Tomczak, 1981] becausethe solutions, which are subject to the nonnegativity -40 30 40 50 60 70 80 90 100 110 120 130 constraint,have to be determinedthrough residual minimization. Longitude
30 40 50 60 70 80 90 100 110 120 130 30 30 3. Seasonal Variations of Parameter Distribution
on the Isopycnal Surfaces 20 20
Four isopycnal surfaces((50=25.7, 26.7, 26.9 and 27.1) were mappedto spanthe main thermoclinefrom about 150 to 800 m on average. Each pair of surfacesis about 150 - 200 m apart. The 0 distance between the same set of surfaces in the southern Indian '• -lO -10
-20 -20 Figure 3. Distributionof hydrographicproperties on the(50=25.7 isopycnalsurface, for (left) the summermonsoon and (right) the -30 wintermonsoon: (a) pressure(P, dbar),(b) potentialtemperature -30 (0, degreeCelsius), (c) salinity (S, psu), (d) oxygen (02, o -40 -40 millilitersper liter), (e) phosphate(PO4, microgramatom pe• 30 40 50 60 70 80 90 100 110 120 130 liter), and(f) silicate(H4SiO 4, microgramatom per liter). Longitude YOU: INDIAN OCEAN SEASONAL THERMOCLINE CIRCULATION 10,39'7
30 40 50 60 70 80 90 100 110 120 130 30 40 50 60 70 80 90 100 110 120 130 30 30 30 30
20 20 20 20
10 10 10 10
0 0 o 0 0
-10 • -10 -10
-20 -20 -20 -20
-30 -30 -30 -30
4O 40 -40 30 40 50 60 70 80 90 100 110 120 130 30 40 50 60 70 80 90 100 110 120 130 Longitude Longitude 30 40 50 60 70 80 90 100 110 120 130 30 40 50 60 70 80 90 100 •110 120 130 30 30 30 30
20 20 20 20
10 10 10 10
0 0 o 0 0
-10 • -10 -10
-20 -20 -20 -20
-30 -30 -30 -30
-40 -40 -40 40 30 40 50 60 70 80 90 100 110 120 130 30 40 50 60 70 80 90 100 110 120 130 Longitude Longitude
30 40 50 60 70 80 90 100 110 120 130 30 40 50 60 70 80 90 100 110 120 130 30 30 30 30
20 20 20 20
lO lO 10 10
o o 0
-lO • -lO -10
-20 -20 -20 -20
-30 -30 -30 -30
-40 -40 30 40 50 60 70 80 90 100 110 120 130 30 40 50 60 70 80 90 100 110 120 130 Longitude Longitude 30 40 50 60 70 80 90 100 110 120 130 30 40 50 60 70 80 90 100 110 120 130 30 30 30 30
20 20 20 20
lO lO lO lO
0 ! o o o
-lO ,• -lO -lO
-20 -20 -20 -20
-30 -30 -30 -30
40 40 -40 40 30 40 50 60 70 80 90 100 110 120 130 30 40 50 60 70 80 90 100 110 120 130 Longitude Longitude Figure 3. (continued) 10,398 YOU:INDIAN OCEAN SEASONAL THERMOCLINE CIRCULATION
30 40 50 60 70 80 90 100 110 120 130 30 40 50 60 70 80 90 100 110 120 130 30 30 30 30
20 20 20 20
10 10 10
0 0 o
40 '• 40 -lO
-20 -20 -20 -20
-30 -30 -30 -30
-40 -40 -40 -40 30 40 50 60 70 80 90 100 110 120 130 30 40 50 60 70 80 90 100 110 120 130 Longitude Longitude 30 40 50 60 70 80 90 100 110 120 130 30 40 50 60 70 80 90 100 110 120 130 30 30 30 30
20 20 20 20
10 10 10
0 0 • 0 o 0
-10 • -lO -10 • -10
-20 -20 -20 -20
-30 -30 -30 -30
40 40 -40 -40 30 40 50 60 70 80 90 100 110 120 130 30 40 50 60 70 80 90 100 110 120 130 Longitude Longitude
30 40 50 60 70 80 90 100 110 120 130 30 40 50 60 70 80 90 100 110 120 130 30 30 30 30 d 20 20 20
lO lO 10 10
0 o 0 0
'•1 -•o 2 -•o -10
-20 -20 -20 -20
-30 -30 -30 -30
40 40 -40 -40 30 40 50 60 70 80 90 100 110 120 130 30 40 50 60 70 80 90 100 110 120 130 Longimde Longitude 30 40 50 60 70 80 90 100 110 120 130 30 40 50 60 70 80 90 100 110 120 130 30 30 30 30
20 20 20 10 10 lO /,•.•,•lO o o
40 'g -•o -10
-20 -20 -20 -20
-30 -30 -30 -30
-40 -40 -40 -40 30 40 50 60 70 80 90 100 110 120 130 30 40 50 60 70 80 90 100 110 120 130 Longitude Longitude Figure4. Sameas Figure 3 buton the {•o=26.7 isopycnal surface. YOU: INDIAN OCEAN SEASONAL THERMOCLINE CIRCULATION 10,399
30 50 60 70 80 90 100 110 120 130 Ocean is about 50 m wider than in the northern Indian Ocean. 30 30 Pressure,potential temperature, salinity, oxygen,phosphate, and
20 silicate for the summer monsoon (the left panels) are directly comparedwith the sameparameters of the winter monsoon(the rightpanels) in Figures3-6. Seasonaldifferences in all parameters are seenclearly. The uppermostisopycnal surface, os=25.7, lies in the upper thermoclineand thereforeexperiences the greatest
-10 seasonalchange. In Figure 3a during both summerand winter monsoonsthere is a distinct high-pressureband between about -20 -20 10øSand 25øS,east-west tilting with five latitudinaldegrees more to the southin the westernIndian Ocean.This high-pressureband -30 -30 is slightly wider and extends more westward to the eastern African coastduring the summermonsoon (Figure 3a, left panel). -40 -40 Oceanthermocline deepening in this latitudinalband suggests the 30 40 50 60 70 80 90 100 110 120 130 Longitude throughflow path of AAMW, which is indicated by low- 30 40 50 60 70 80 90 100 110 120 130 temperature,low-salinity water flowing in from the Indonesian 30 30 seas,seen as the tonguesin Figures3b and 3c. North of the high- pressureband in the equatorialand northernIndian Ocean for the 20 20 summer in Figure 3a, there is a large low-pressure tongue extending from the Somali coast to the northeastIndian Ocean accompanied by relatively uniform temperature and salinity o distributions(left panelsof Figures3b and 3c). A significant pressuredifference between the two seasonsis -lO seenin the subtropicalocean south of 30øS(Figure 3a) wherethe thermocline shallowsbetween 50 and 100 m during the winter -20 -20 monsoon. The right panel of Figure 3a reflects the effect of
-30 -30 northwardlow-temperature, low-salinity water from the Southern Ocean as seenin the right panelsof Figures3b and 3c southof
..40 -40 32øS. Seasonal signals are apparent in the temperature and 30 50 60 70 80 90 lOO 11o 12o 13o salinity fields of Figures 3b and 3c, indicative of the signalsfor Longitude the Somali Current. The 17.0øC isotherm and 35.3 isohaline clearly show a northwardextension along the Somali coastduring 30 40 50 60 70 80 90 100 110 120 130 30 30 the southwest monsoon, with a return southward during the northeast monsoon. These contour lines migrate about 5 to 8 20 20 latitudinaldegrees between the two seasons. The seasonaldifferences of temperatureand salinity are also apparent in the eastern Indian equatorial region, where same contours propagate more eastward in the summer than in the winter. Note the 16.5øC isotherms and 35.0 isohalines at the
-10 equatorin Figures 3b and 3c. A slightly strongersummer signal for AAMW is more obvious in the salinity contoursin the left -20 -20 panel of Figure 3c where a low-salinity tongue extends more westward to the eastern African coast (compare the 35.2 -30 -30 isohaline).In fact, a strongersouthwest monsoon, compared to the northeast monsoon, stands out by the more uniformly -40 4O 30 40 50 60 70 80 90 100 110 120 130 distributedtemperature and salinity fields in the subtropicaland Longitude Somali Currentregions.
30 40 50 60 70 80 90 100 110 120 130 The temperature and salinity contours are different in the 30 30 subtropical ocean during the northeast monsoon in the right panelsof Figures3b and 3c and are significantlyaffected by the 20 austral summerseason. The low-temperature,low-salinity water southof 32øSindicates most likely SouthernOcean surfacewater moving to north,initiated by ice melting and precipitationduring 0 australsummer. Colborn [1975] suggestedthat duringthe austral summer, a weaker thermocline formation occurs near the • -10 -10 Subtropical Convergence Zone, while during the northern hemisphere summer the surface water could be mixed to a -20 -20 relatively great depth.It is quite clear from the fight panelsof Figures 3b and 3c that the weaker thermocline formation is -30 -30 causedby invasionof the coolerand fresherwater from the south.
-40 Sverdrupet al. [1942, pp. 695 ] wereprobably the first to notethe 30 50 60 70 80 90 100 110 120 130 impact of seasonal change on ocean currents in the southern Longitude Indian Ocean, statingthat "betweenSouth Africa and Australia Figure 4. (continued) the currentis directedin generalfrom westto east.In the southern 10,400 YOU: INDIAN OCEAN SEASONAL THERMOCLINE CIRCULATION
30 40 50 60 70 80 90 100 110 120 130 30 40 50 60 70 80 90 100 110 120 130 30 30 30 30
20 20 20 20
lO lO lO ] lO
• 0 o 0 -10 • -10 -10
-20 -20 -20 -20 ..
-30 -3O -30 -3O o -40 -40 -40 , -40 30 40 50 60 70 80 90 100 110 120 130 30 40 50 60 70 80 90 100 110 120 130 Longitude Longitude 30 40 50 60 70 80 90 100 110 120 130 30 40 50 60 70 80 90 100 110 120 130 30 30 30 30
20 20 20
lO lO lO
0 o 0
-10 -10 '•,..1 -lO
-20 -20 -20 -20
-30 -30 -30 -30
-40 -40 -40 -40 30 40 50 60 70 80 90 100 110 120 130 30 40 50 60 70 80 90 100 110 120 130 Longitude Longitude
30 40 50 60 70 80 90 100 110 120 130 30 40 50 60 70 80 90 100 110 120 130 30 30 30 30 d 20 20 20 20
lO lO
0 0 o 0
'• -10 -10 • -10 -10
-20 -20 -20 -20
-30 -30 -30 -30
¸ -40 -40 -40 40 30 40 50 60 70 80 90 100 110 120 130 30 40 50 60 70 80 90 100 110 120 130 Longitude Longitude 30 40 50 60 70 80 90 100 110 120 130 30 40 50 60 70 80 90 100 110 120 130 30 30 30 30
20 20 20 20
lO ] lO lO
0 0 o 0
-10 • -10 -10
-20 -20 -20 -20
-30 -30 -30 -30
o 0 o 40 -40 -40 -40 30 40 50 60 70 80 90 100 110 120 130 30 40 50 60 70 80 90 100 110 120 130 Longitude Longitude Figure 5. Sameas Figure 3 but on theo•=26.9 isopycnalsurface YOU: INDIAN OCEAN SEASONAL THERMOCLINE CIRCULATION 10,401
30 50 60 70 80 90 100 110 120 130 summer the current bends north before reaching the Australia 30 30 continent." These early studies suggestthat the Subtropical
20 ConvergenceZone probablymigrates northward for somedegrees during the australsummer and so does the subtropicalgyre, at 10 10 leastin its southernbounds. The fight panelsof Figures3b and 3c suggest such a seasonal shift probably for about 5 latitude 0 degrees.Seasonal changes certainly complicate the gyre dym/mics as well. When Wyrtki [ 1973, pp. 26 ] was inspectingthe dynamic -10 topographyfor two seasons,he foundthat "thesubtropical gyre of
-20 -2O the southern Indian Ocean is developed somewhat differently from the correspondinggyres in the other oceans."These suggest -30 -30 that one shouldbe careful while calculatinggeostrophic flows. Seasonal changeshave to be consideredeven in the southern
-40 -40 Indian Ocean.The seasonalsignal is apparentin the Mozambique 30 40 50 60 70 80 90 100 110 120 130 Channel and eastof Madagascar.In the MozambiqueChannel the Longitude 30 50 60 70 80 90 100 110 120 130 35.3-psu contour line in Figure 3c extendsfarther south during 30 30 the summer. A striking seasonal difference is found east of Madagascar.Contour lines extendsouthward dufing the summer 20 20 but reverse to the north during the winter (note the northward stretchingof the35.3 isohalineto northof Madagascarin thefight panel).This seasonalreversal is alsorevealed in otherproperty distributionsand on the lower isopycnalsurfaces. As we know, these seasonalchanges have important implicationsfor the -10 circulationin the southwestIndian Ocean, particularly in respect to the AgulhasCurrent. -20 Large seasonalsignals are alsoreflected in oxygenand nutfient
-30 -30 fields. Like temperatureand salinity,one can seein Figures3d, 3e, and3f thatthe contourlines of oxygen(2.5 mL/L), phosphate
-40 -40 (1.4 gg - atom/L) and silicate(20 gg- atom/L), apparentlyextend 30 40 50 60 70 80 90 100 110 120 130 more northward along the Somali coast during the southwest Longitude monsoon(left panels)than duringthe northeastmonsoon (right panels).It is notedthat oxygenand phosphatedo not showclear 30 40 50 60 70 80 90 100 110 120 130 30 30 signalsfor AAMW becauseICW is too strong,while silicate marksthe summerextension of AAMW clearly.The left panelof 20 Figure3f indicatesa relativelylow silicateband extending from the SomaliCurrent region to the easternIndian equatorial region, separatingthe higher-silicatewater in the Bay of Bengalfrom that
o of the AAMW tongue.This clearlysuggests that the high-silicate water in the Bay of Bengal is not from the same source as • -lO -10 AAMW. The separationis not so obvious during the winter (Figure3f, fight panel),implying that the communicationbetween -20 -20 AAMW and the water in the Bay of Bengal only might occur during the northeastmonsoon. -30 -30 You and Tomczak[1993] found that the c•0=26.7isopycnal surfaceis morecharacterized by isopycnalmixing. They implied -40 -40 30 40 50 60 70 80 90 100 110 120 130 that the ventilationwater in the northernIndian Oceanoriginated Longitude from the SouthernOcean and is mostlikely advectedalong this 30 50 60 70 80 90 100 110 120 130 30 30 surface.This isopycnalsurface lies at aboutthe 400 m depthlevel (Figure 4a) and is characterizedby almostthe sametemperature
20 20 and salinity values for both the summer and winter monsoons (Figures4b and4c) betweenthe formationregion of ICW at 40øS to the subtropicalgyre, and the westernboundary region to the Bay of Bengal. The northward extension of contour lines in o temperature(12.5øC) and salinity (35.2 psu) fields during the
'• -10 -10 southwestmonsoon is seenclearly alongthe Somali coastin the left panelsof Figures4b and4c. As seenabove, the throughflow -20 -20 water of AAMW is indicatedby the low-temperatureand low- salinitywater tongues in Figures4b and4c. A northwardsummer -30 -30 migrationof about5 degreesin the contourlines of oxygenand nutrientsis seenclearly in the left panelsof Figures4d, 4e, and4f -40 -40 30 50 60 70 80 90 100 110 120 130 at the SomaliCurrent region. The high-silicatetongue in the left Longitude panelof Figure4f, associatedwith AAMW and separatedby a Figure 5. (continued) relative silicate minimum to the north, extendsfarther westward 10,402 YOU: INDIAN OCEAN SEASONAL THERMOCLINE CIRCULATION
30 40 50 60 70 80 90 100 110 120 130 30 40 50 60 70 80 90 100 110 120 130 30 30 30 30
20 20 20 20
10 10 10 10
0 0 o 0 0
'g -20 -1o• -1o -10
-20 -20 -20 -2O
-30 -30 -30 -3O
30 40 50 60 70 80 90 100 110 120 130 30 40 50 60 70 80 90 100 110 120 130 Longimde Longimde 30 40 50 60 70 80 90 100 110 120 130 30 40 50 60 70 80 90 100 110 120 130 3O 30 30
20 20 20 20
1o 1o 1o
0 o 0 0 -10 •-1-10 -10
-20 -20 -20 -20
-30 -30 -30 -30
-4O -40 -40 -40 30 40 50 60 70 80 90 100 110 120 130 30 40 50 60 70 80 90 100 110 120 130 Longitude Longitude
30 40 50 60 70 80 90 100 110 120 130 30 40 50 60 70 80 90 100 110 120 130 30 3O 3O b d 20 20 20
1o 1o 1o
0 o 0 0 -1o • -1o -1o
-20 -20 -20
-3O -30 -30 -30
40 40 40 40 30 40 50 60 70 80 90 100 110 120 130 30 40 50 60 70 80 90 100 110 120 130 Longimde Longimde 30 40 50 60 70 80 90 100 110 120 130 30 40 50 60 70 80 90 100 110 120 130 30 30 30 30
20 20 20 20
10 1o 1o 1o
0 0 o 0 0
• -1o -10 • -10 -10
-20 -20 -20 -20
-30 -30 -30 -30
40 40 -40 I , • , • , I •', • -40 30 40 50 60 70 80 90 100 110 120 130 30 40 50 60 70 80 90 100 110 120 130 Longitude Longitude Figure6. Sameas Figure 3 buton theC•o=27.1 isopycnal surface. YOU: INDIAN OCEAN SEASONAL THERMOCLINE CIRCULATION 10,403
30 40 50 70 80 90 100 110 120 130 than in the winter (right panel), suggestinga relatively larger 30 30 summersignal.
20 20 Though one might argue that the proposedtransition path of ICW by You and Tomczak[1993] is reasonablysupported by the temperatureand salinityfields as seenin Figures4b and 4c, this path is not obviousin the oxygen and nutrient fields. This has 0 alreadybeen describedbefore in a conceptof water massaging.
-10 Oxygen in ICW startsto decreaseafter it last comesin contact • -10 with the atmospherein the southernIndian Oceanand advectsto
-20 -20 the northern Indian Ocean because of the biochemical oxygen consumption and corresponding nutrient gain during the -30 -30 remineralization processes. Colborn [1975] suggestedthe
¸ usefulnessof oxygen in the Indian Ocean as an indicatorfor the -40 relativeage of ICW. Becausethere is a stronghydrological front 30 40 50 60 70 80 90 100 110 120 130 Longitude at 15øS east of 50øE formed by the throughflow (AAMW), the westernboundary (west of 50øE) becomesonly one possiblepath 30 40 50 60 70 80 90 lOO 11o 12o 130 30 30 for ICW transiting to the north. You and Tomczak [1993] proposedthat northwardtransition of ICW occursin the western 20 20 boundarycurrent and that the Bay of Bengalcontains the oldest ICW. This is further supportedby the property distributionson 10 10 the 00=26.7 isopycnalsurface in Figure 4. Except for the afore mentionedtemperature and salinity distributionsand the stronger northwardtransition of oxygen and nutrientsduring the summer -10 in the Somali Current region, one notesthat there is a ratherclear transition path of ICW to the Bay of Bengal east of Sri Lanka. -20 -20 The left panels of Figures 4d, 4e and 4f show a consistent northward extension of oxygen (1.5, 1.0 and 0.5 mL/L), -30 -30 phosphate(2.0 and 2.2 gg - atom/L) and silicate (30 and 35 gg - atom/L) contour lines east of Sri Lanka during the southwest -40-- -40 30 40 50 60 70 80 90 100 110 120 130 monsoon that is not so evident during the northeastmonsoon in Longitude the fight panels.This will be further supportedby the water mass mixing ratio distribution discussedlater, in section 5. In contrast 30 40 50 60 70 80 90 100 110 120 130 to the southwardflow eastof Madagascarduring the summer(left 30 30 panels of Figures 4d to 4f), northwardreversal during the winter is again seenclearly in the oxygen (5.0 and 4.5 mL/L), phosphate 20 20 (1.0 and 1.2 gg - atom/L) and silicate (10 and 15 [tg - atom/L) contour lines in the fight panels of Figures 4d, 4e, and 4f. This situation consistently occurs even in the deeper surfaces, for example,the right panels of Figures5c (S=34.8psu), 5d (02=5.0 mL/L), 5f (H4SiO=15gg - atom/L),6b (T=7.0øC),6c (S=34.6 -10 psu),6d (02=4.5 and4.0 mL/L), and6f (H4SIO4=30,35 and40
-20 -20 gg- atom/L). Wyrtki [1973] suggestedthat the pronouncedseasonal variations
-30 -30 occurred primarily only in the upper 400 m. The 00=26.7 isopycnal surfacediscussed above is just located at this level. At
-40 -40 greater densitiesthe seasonaleffect on property fields becomes 30 40 50 60 70 80 90 100 110 120 130 weaker. Figure 5 showsthe 00=26.9 isopycnalsurface, which has Longitude an average depth of about 600 rn for both seasonsin Figure 5a. 30 40 50 60 70 80 90 100 110 120 130 30 30 Seasonalsignals can still be clearly identified along the Somali coastin the summertemperature (11.5 øC) and salinity (35.3 psu) 20 20 fields in the left panels of Figures 5b and 5c, where at least 2 degreesof latitudinal shift are evident.In contrastwith the upper two isopycnal surfaces, the temperature and salinity fields at 00=26.9 do not show any structurefor AAMW during the two seasons.This suggestsAAMW to be significantlyweaker on this
-10 level. However, AAMW can still be identified in the silicate field in Figure 5f, where AAMW is characterizedby slightly higher -20 -20 silicate and distinguished from the high silicate of NICW by relatively lower silicate in between.The seasonalsignal doesnot -30 -30 disappeareven at the deepest00=27.1 isopycnalsurface in Figure 6. The differenceof seasonalsignal can still be well identified, for -40 -40 example, the summer temperature (11.0øC), salinity (35.4 psu) 30 40 50 60 70 80 90 100 110 120 130 Longitude and oxygen (1.0 mL/L) in the left panelsof Figures6b, 6c, and 6d Figure 6. (continued) in contrastto the right panelsduring the winter. 10,404 YOU: INDIAN OCEAN SEASONALTHERMOCLINE CIRCULATION
The initial sourcewater types derived through cluster analysis on ClusterAfor the summer• Cluster B for the winter each isopycnal surfaceare then plotted in property-property diagrams: temperature against salinity, temperature against oxygen,and temperatureagainst nutrients, as shownin Figure 8. We note that clusteranalysis presents a simplified water mass structure.The calculationwas carriedout for all four isopycnal surfacesduring the two seasons.Consequently, four cluster centers, or initial source water types, were obtained for each Overlapped area sourcewater mass. Figure 8 showsthat in all casesexcept for one A: Cluster center of cluster A (the 0 - H4SiO4 relationshipfor ICW in Figure8d), theproperty- B: Cluster center of cluster B propertyrelationships defined by the initial sourcewater types are (a) closeto linear. Basedon thesealready nearly linear initial source water types, the closestlinear representationsare determined ClusterAfor the summer • throughregression analysis, which are shownas the straightlines luster B for the w•nter in Figure 8. For a comparisonbetween the two seasons,results areshown in theleft panelsfor thesummer monsoon and the right panels for the winter monsoon.The four isopycnal surfaces appliedhave beenindicated in the 0-S diagramin Figure 8a. A glimpseat thesemapped source water typescontains already informationsabout seasonalsignals. For example,the largest seasonalvariation is foundin RSW/PGW watertypes (see Figure • •• for,nit,al source water type 8). The final "real" sourcewater types for mixing model were dehmtions of the two seasons obtainedas intersectionsof regressionlines and the specified C: Cluster center for both cluster A and B isopycnal surfaces.They are indicated as the open circles in Figure 8. The procedurealso yields a variancefor eachwater massused Figure 7. Illustrationof how a fixed samplingarea is takenin the todetermine •ax in equation(4). In fact,the standard deviation overlappedregion of two clustersfrom two differentseasons for was calculatedfor each parameterof each sourcewater massin accurate determinationof seasonalsignals in the initial source orderto calculatethe weights.Because the silicatevalue for ICW water types:(a) two clustercenters of two seasons(A andB), and more than doublesfrom the surfacec•0=26.9 to c•0=27.1and (b) one clustercenter (C) for two seasons. thereforethe 0-H4SiO 4 relationshipof ICW cannotbe represented by a single linear relationship,the procedurewas applied separatelyto the densitysurfaces above and below the c•0=26.9 4. Definition of Source Water Types and the densitysurface. The "real" sourcewater typesneeded for OMP Mixing Model analysiswere taken from the intersectionpoints of regression lines and isopycnalsurfaces as the valuesfound on the highest As mentionedabove, the sourcewater typesdetermined as the (c•0=25.7)and lowest (c•0=27.1)density surfacesand on the cluster centers on individual isopycnal surfaceswere used as densitysurface (c•0=26.9) at the breakpoint of the 0-H4SiO4 initial guessesfor the final determinationof the "real" source relationshipof ICW. They are tabulated,together with weights water types. The data in the respectiveregions were used to derived from the data and water mass variances, in Table 2 and determine the cluster centers (i.e., points in parameter space Table 3 for the two monsoon seasons. The seasonal differences which give the smallestsum of the distancesto all otherpoints in for eachsource water type can be examinedby comparingthe two their respectivecluster). However, since we are studyingthe tables.Of all the sourcewater types, RSW/PGW has the strongest seasonal variations of thermocline circulation and ventilation, an seasonalvariation, NICW has the next strongestseasonal accuratedetermination of the initial sourcewater typesis crucial variation, and ICW has the weakest seasonal variation at the to the mixingmodel. Although the respectiveregional shifting of sourceregion. 'I•is is becausethe ICW sourceregion is notunder the clusters on an isopycnal surface (presumablycaused by mosoonaleffect. Sincewe have only six linear equationsin (1) seasonalchanges) cannot affect the definitionof the samesource and (2), correspondingto five parametersand one water mass watertype very much,it canchange the locationof clustercenters conservationequation, our watermass mixing model can handle a to somedegree in different seasons.Therefore seasonal signals maximumof only six sourcewater types. With a minimumof two derived from uncommon cluster centers can no longer be source water types, required to define one sourcewater mass, compared.To derive a more accurateseasonal signal in initial mixing amongthe six sourcewater types on the highestand sourcewater types,a fixed areawas takenin the overlappingtwo lowest densitysurfaces for ICW, NICW, and AAMW constitutes clustersof thetwo seasonsfor initialsource water type definitions the mixingmodel. It is usedto derivemixing ratios as discussed of both the summer monsoon and the winter monsoon. This is in the followingsection. Meanwhile, the relativecontribution of representedschematically in Figure 7. In the caseof Figure 7a, RSW/PGWwill alsobe estimatedon the uppermostisopycnal since two clusters cover different areas in two different seasons, surface in the model and as a residual on the other surfaces. their centersare then not locatedin the sameplace, A and B. This 5. Results problemcan be solvedin the caseof Figure 7b by taking a fixed area in the overlappingregion of two clustersof two seasons,so 5. 1. Isopycnal Surfaces that two clusters have a common cluster center C. The difference betweenany parameter measured at C duringthe two seasons is With the limitation of only six availableequations, the mixing then consideredto be the true seasonalsignal for the sourcewater model comprisesthree water masses.It outputsmixing ratiosfor types,and theseseasonal signals can be comparedto eachother. each water mass (i.e., the sum of the contributions from its two YOU: INDIAN OCEAN SEASONAL THERMOCLINE CIRCULATION 10,405
•30 ! i ! i ! i i I i i ! i ! i , i 28 summer * :lCW summer 26 oe = 25.7 sudace • + :AAMW z•: NICW 24 V: RSW 22. . O: water type 20
18 E16 .9surface - 12 [' /;4,'"3' ':cw
- /// -I
2
0 I • I • I • I I I I 33 34 35 36 37 38 39 40 2 3 4 5 S (psu) O 2 (ml/L)
' i , i ' i ' i ' i ' i , 30 , , , , , i , i , i , i • * :ICW - winter / '1 winter +: AAMW 2826 . oe=25.7sudce• • • A: NICW V: RSW 24 0: water type 22
20
18
16
14
12
10 8• //,• +:AAMW :
4 /// ,':.STM. -
2
0 I I • I I I • I J I , [ 33 34 35 36 37 38 39 40 0 1 2 3 4 5 S (psu) O 2 (ml/L) Figure8. Thedefinition of sourcewater types in property-propertyplots as derived from cluster analysis, for (left) the summermonsoon and (right) the wintermonsoon: (a) potentialtemperature-salinity diagrams, (b) potential temperature-oxygendiagrams, (c) potentialtemperature-phosphate diagrams, and (d) potentialtemperature-silicate diagrams.
source water types); the sum of all mixing ratios for all three The property distributionswere analyzed in section 3. They water massesadds to 100% plus a residual.Figures 9-12 showthe showed large seasonal signals on this uppermost isopycnal contributions of the source water masses to the thermocline water surface. Figure 9 containsfurther evidence on the mixing ratio circulationand ventilationderived from the mixing model for the distributionsof the sourcewater masses.Percentage contour lines summermonsoon (left panels),in contrastto the winter monsoon for ICW during the southwestmonsoon in the left panel of Figure (right panels). Figure 9 showsthe mixing ratios on the c;0=25.7 9a clearly show northwardextension along the Somali coast(see isopycnalsurface, which experiencesthe largestseasonal change the 60% contour line), indicative of more ICW in the Somali among the four isopycnal surfaces.Three water masses,ICW, Currentflowing to the north.At least 10% more ICW is carriedto NICW, and AAMW, are includedin the mixing model, and their the eastern equatorial Indian region during the summer than mixing proportionsare shownin Figures 9a, 9b, and 9c for both during the winter (compareleft and right panelsof Figure 9a; also the southwest and the northeast monsoons. The relative see the 20 and 30% contour lines). In contrast, there is no contribution of RSW/PGW is also estimatedon this isopycnal northwardstretching of the percentagecontour lines seenin the surface (Figure 9d) by including ICW, NICW, and RSW/PGW Somali Current region during the northeastmonsoon. Instead the but excludingAAMW in the westernIndian Ocean for the upper right panel of Figure 9a shows a southward extension. The thermoclinefrom the mixing model [ Youand Tornczak,1993]. northward excursion of 20 and 10% ICW contour lines east of Sri 10,406 YOU: INDIAN OCEAN SEASONAL THERMOCLINE CIRCULATION
+ 'AAMW 26 •,: NICW •, • +:AAMW _ 2428 '• summer V.* 'ICWRSW 2 O ßwater type 22
20 2
18 • 1
•'16 1
• 14 1
. 12 • 1
.
lO 1
.
8
6
4
2
0 1.0 2.0 3.0 4.0 0 10 20 30 40 50 60 70 80 90 100 PO4 (pg-at/L) H4SiO 4 (I.Lg-at/L)
3O ' ' ' ' i , [ , [ i , i ! , i , , , , i ! , , , i T , , ]li ! , ! ! i [ ! , ! i ! i [ i [ i , i ! i [ i ] i ! i ,
. 28 winter * 'ICW winter * 'ICW :AAMW + 'AAMW 26 - NICW A: NICW V:RSW V' RSW water type O ßwater type
18
•'16
8
6
4 4
2 2
I I I I I I I I ' I , I , I t I • I • 0 1.0 2.0 3.0 4.0 0 10 20 30 40 50 60 70 80 90 00 PO4 (gg-at/L) H4SiO 4 (I.Lg-at/L) Figure 8. (continued)
Lanka and westof the Bay of Bengal(left panel)suggests the imageof ICW, the NICW in theright panel of Figure9b andcan pathof ICW intothe Bay of Bengalduring the summer. However, be clearlyseen in the lower surfaces. duringthe winter, the 10, 20, and 30% ICW contourlines south of Anotherinteresting feature is foundin thesubtropics during the Sri Lankapoint to the west(right panel), suggesting an opposite winter. The 100% contour line of ICW contribution retreats flow. The 100% of ICW mixingratio in the subtropicalregion northwardfrom the southernboundary (Figure 9a, rightpanel). pointsto it as its sourceregion. The densemixing ratio contours This hasbeen described before in the parameterdistribution of found east of 70øE at 15øSindicate the locationof a strong section3 and is a consequenceof the northwardinvasion of low- hydrologicalfront resulting from mixing betweenICW and temperatureand low-salinitywater resulting from ice - melting AAMW inflow. andprecipitation during austral summer in the SouthernOcean. If A strikingfeature is foundeast of Madagascar.During the the southernboundary of the subtropicalgyre retreats north and summer, the water mass contributionof ICW (see 70, 80, and sodoes the SubtropicalConvergence Zone, this seasonal change 90% contourlines) extendstoward the south,suggesting a during winter might explain the occurrenceof the reversed southwardflow (Figure 9a, left panel). However,these contour northward flow east of Madagascar.The afore mentioned linesevidently extend north during the winter(Figure 9a, right seasonalfeature of ICW mixingratio distribution from Figure 9a panel).The 60% line is especiallynoticeable that extends to north suggestsa ratherclear picture of ICW spreadingnorthward from of Madagascar.This seasonalreverse is alsoevident in the mirror the subtropicsto theBay of Bengal.During summer, ICW hasto YOU: INDIAN OCEAN SEASONALTHERMOCLINE CIRCULATION 10,407
Table 2. SourceWater Type Definitionsand ParameterWeights for the SummerMonsoon
WaterType c•0 Potential Salinity,psu Oxygen,mL/L Phosphate,gg- Silicate,gg- Temperature,øC atom/L atom/L
ICW 25.7 18.15 35.62 5.94 0.06 1.57 ICW 26.9 9.21 34.79 5.05 1.29 8.06 ICW 27.1 6.56 34.55 4.79 1.66 21.39 NICW 25.7 15.60 34.84 0.07 2.13 21.44 NICW 27.1 9.29 35.05 0.52 2.84 58.59 AAMW 25.7 14.96 34.64 2.63 1.41 22.94 AAMW 27.1 6.73 34.56 2.30 2.41 80.08 RSW 25.7 27.60 39.21 4.19 0.12 1.07 (12.82) (12.82) (4.59) (0.85) (4.59)
* Parameterweights are given in parentheses.
find its way flowing to the north far off east of Madagascar indicate that a communicationbetween NICW in the Bay of becauseof the existingapparent southward flow alongeast of Bengal and AAMW is unlikely during the southwestmonsoon. Madagascarfrom the SouthEquatorial Current [Wyrtki, 1973]. We recognizea clear separationof the low silicate distribution ICW thentakes its waysinto the Bay of Bengaladvected by the betweenNICW and AAMW. However, propertyfields suggest South Equatorial Current from the west, the Somali Current, and that communication might be possible during the northeast the SouthwestMonsoon Current. During winter a northwardpath monsoon. This conjecture is confirmed by Figure 9c, which of ICW paralellsthe easternMadagascar, as is suggestedrather indicates a 20% mixture of AAMW flowing into the eastern clearly by both parameter distribution and water - mass Bengal Bay during the winter (right panel), but not during the contribution. Because the western boundary current off the summer(left panel). AAMW's spreadinginto the Bay of Bengal Somalicoast is replacedby the NorthEquatorial Current pointing is probablyby way of the EquatorialCountercurrent. Note that to the south, further northward flow of ICW has to be in the the mixing ratio pattern south of 30øS is due to invasion of EquatorialCountercurrent. Southern Ocean water which has the same characteristics as Since NICW is interpretedas an agedICW, the mixing ratio AAMW (see section 3), and shouldnot be regardedas a distribution of NICW can be used as a complementary contribution of AAMW. explanationof the ICW transitionpath. The extensionof NICW Figure 9d shows the relative contributionsof RSW/PGW on the from the Bay of Bengal to Madagascarpoints clearly to the c•0=25.7isopycnal surface during the two seasons.The way to western Indian Ocean in Figure 9b. The dense contoursin the estimate the relative contribution of RSW/PGW has been easternIndian Ocean suggestthat southwardflowing NICW is describedbefore. The seasonaleffect is documentedby the water blockedby a stronginflow of AAMW. The right panelof Figure massspreading patterns of RSW/PGWin theArabian Sea. During 9b showsa largeproportion of NICW flowingto the southduring the southwestmonsoon, RSW/PGW contributesonly 20% to the the northeastmonsoon. In contrast,a tongue of NICW 50% thermoclineventilation (left panel).The northwardstretching of contouroff Somali coastsuggests less NICW extendingto the the 10% and 20% contour lines in the western Arabian Sea has southduring the summer,leaving more spacefor northwardICW implicationsfor why RSW/PGW contributesso much less during to pass.AAMW flow is apparentlystrongest during the summer the southwestmonsoon. This is causedby a strong,jet-like (Figure 9c, left panel). In summerthe westwardtongue of its western boundary Somali Current penetrating the western mixing ratio extends as far as the eastern African coast line Arabian Sea. It effectively blocks the outflow of RSW/PGW. (Figure 9c, left panel). As mentioned before, property However,during the northeastmonsoon, the westernboundary distributions,particularly that of silicate,on this densitysurface currentin the regionis replacedby a southwardmonsoon current,
Table 3. SourceWater Type Definitions and Parameter Weights for the WinterMonsoon
WaterType o0 Potential Salinity,psu Oxygen,mL/L Phosphate,gg- Silicate,gg- Temperature,øC atom/L atom/L
ICW 25.7 18.17 35.62 5.84 0.05 1.13 ICW 26.9 9.39 34.80 4.98 1.36 8.87 ICW 27.1 6.77 34.56 4.72 1.76 22.49 NICW 25.7 15.70 34.85 0.04 2.12 23.34 NICW 27.1 9.17 35.03 0.54 2.67 58.49 AAMW 25.7 14.93 34.63 2.62 0.97 22.01 AAMW 27.1 6.90 34.59 2.31 1.78 58.67 RSW 25.7 19.23 35.97 1.67 1.85 14.12 (5.56) (5.56) (2.44) (0.67) (2.44)
* Parameterweights are given in parentheses. 30 40 50 60 70 80 90 100 110 120 130 30 40 50 60 70 80 90 100 110 120 130 3O 3O 30
20 20 20
10 10 10 • 10
0 0 o 0 0 -10 • -10 -10
-20 -20 -20 .. -20
-30 -30 -30 -30
-40 -40 -40 -40 30 40 50 60 70 80 90 100 110 120 130 30 40 50 60 70 80 90 100 110 120 130 Longimde Longimde 30 40 50 60 70 80 90 100 110 120 130 30 40 50 60 70 80 90 100 110 120 130 30 30 30 30
20 20 20 20
lO lO lO j lO
o o o 0
-10 • -10 -10 -20 t .. -20 -30 -30 -30 -30
-40 -40 -40 -40 30 40 50 60 70 80 90 100 110 120 130 30 40 50 60 70 80 90 100 110 120 130 Longitude Longitude
30 40 50 60 70 80 90 100 110 120 130 30 40 50 60 70 80 90 100 110 120 130 30 30 30 30 d 20 20 20 20
10 10 10
0 !o o
-•o '=•,..1 -10 -10
-20 -20 -20 -20
-30 -30 -30 -30
-40 -40 30 40 50 60 70 80 90 100 110 120 130 30 40 50 60 70 80 90 100 110 120 130 Longitude Longitude 30 40 50 60 70 80 90 100 110 120 130 30 40 50 60 70 80 90 100 110 120 130 30 30 30 30
20 20 20
10 10 10
0 o 0
-•0 '•• -10
-20 -20 -20 -20
-30 -30 -30 -30
-40 -40 -40 -40 30 40 50 60 70 80 90 100 110 120 130 30 40 50 60 70 80 90 100 110 120 130 Longitude Longitude Figure9. Watermass contributions on the 0o=25.7 isopycnal surface for (left)the summer monsoon and (fight) the winter monsoon:(a) the contributionof Indian Central Water (ICW), (b) the contributionof North Indian Central Water (NICW), (c) the contributionof AustralasianMediterranean Water (AAMW), and (d) the contributionof Red Sea Water (RSW)/Persian Gulf Water (PGW). YOU: INDIAN OCEAN SEASONAL THERMOCLINE CIRCULATION 10,409 the North Equatorial Current. Under northeasterlymonsoonal Bengal in both panelsof Figure 1l a, a zero contributionis given forcing, RSW/PGW contributesmuch more to the thermocline when the residualof 10% during the winter (Figure 1l d, right ventilation during the winter (right panel), but its major panel) is subtracted.However, with 5% of residualduring the contributionis still limited to the Arabian Sea region. summer(Figure 11d, left panel) one derivesan ICW contribution Water - massfeatures on the (•0=26.7isopycnal mixing surface of 5%. Therefore the seasonal difference of water- mass show unexpected patterns. ICW again indicates stronger contribution needs to be analyzed in combination with the northwardextension during the southwestmonsoon (Figure 10a, residualmap. Taking a 5% residualinto accountfor both seasons left panel).Contour lines of percentagemixing ratio extendnorth in Figure 11d, we still find a relatively strongersummer signal of CapeGuardafui, Somali. In the northeasternIndian Oceanthe (Figure 11a, left panel) in the northwardflow of ICW at northeast 30% contourline approachesSri Lanka more closelyto than in of Madagascar(note the 70 and 60% contourlines) compared to the winter situation. The path of NICW during the southwest the rightpanel of Figure11 d. However,we shouldkeep in mind monsoonis depictedin the easternIndian Ocean. At 10øSit turns that one usually assumesonly the surfacelayer to be controlled toward the west (Figure 10b, left panel); there is not sucha clear by the monsoonregime. This paper concernsthe circulation pathduring the winter(right panel). Interestingly, AAMW shows below the surfacedown to the main thermocline.Although the northwardbending during the southwestmonsoon (Figure 10c, mixing pattern of water - mass contributionsin the upper left panel).However, no suchtendency of the returnmovement is thermocline has suggesteda circulation more or less similar to apparentduring the northeastmonsoon (right panel), even though that in the surfacelayer, we may not assumethat thereexists no AAMW is driven to as far as 5øN. The distributionof the mixing countercurrent (opposite to the surface flow) in the lower ratio patternis more zonal. The seasonaldifference of AAMW thermocline.On the other hand, the strongernorthward flow of contributionis probablydue to this isopycnal-mixing-dominated ICW through the western boundaryhas to return to the south surface. Under wind forcing of the southwest monsoon, the rather than pile up in the northernIndian Ocean and wait for the epipycnaladvection of water - massmovement is certainlythe following oppositemonsoon season to take it back to the south. strongestamong the four isopycnalsurfaces. Thereforethe returnflow is expectedalong the easternboundary The northwardbending of AAMW seen in the left panel of or in the lower thermocline. Figure 10c is not surprisingif one looks back at the parameter The pathof southwardNICW is still seenclearly in the western distributionof section3 on the same surface.The left panels of Indian Ocean(Figures 1 lb and 12b).Although AAMW is weaker Figures4d, 4e, and 4f showclear northwardbending of oxygen in the lower thermocline, its westward cross-Indian Ocean flow (1.5 mL/L), phosphate(2.0 gg - atom/L) and silicate (30 gg- and contributionto the Agulhas Current is shown clearly in atom/L) contourlines southof Sri Lanka during the summer.One Figures 11c and 12c for both seasons. may askwhy we do not seethe northwardbending of AAMW on the upper isopycnal surface,where property distributionalso 5.2. Cross Sections shows northward extension. A simple answer is that above surface,00=25.7, is not an isopycnal-dominatedsurface, and You and Tomczak [1993] showed that ICW extends northward diapycnal/dianeutralmixing has to be invoked[You, 1996]. into the northernIndian Oceanwith a high proportionof mixing One common point can be drawn during the two seasonsin ratios even in an annual mean situation. They suggestthat the Figure 10c, namely, AAMW extendswestward enough to be northward movement of ICW during the summer monsoonhas another source water of the western boundary current. These thereforea much strongerimpact on the thermoclinecirculation resultsin Figure 10 also supportthe transitionpath for ICW on and ventilation. In other words, during the summer monsoon, this isopycnalsurface proposed by Youand Tomczak[1993]. The underthe strongsouthwesterly wind, a westernboundary current, residualmaps (the• sum of total contributionsof the sourcewater the Somali Current, developsand transportsa large volume of typesminus 100%) in Figure 10d, showrather small residualsin water to the equatorialand northernIndian Ocean along Somali most areasduring the two seasons,except the subtropicalregion coast. The Southwest Monsoon Current, which is a continuation and the Arabian Sea. The Indian Ocean subtropicalgyre region of the Somali Current, flows eastward and then turns to south has a residual of 10% in the ICW formation region, suggesting mergingwith the SouthEquatorial Current. The SouthEquatorial that the ICW contribution can be overestimatedby as much as Current, flowing westwardwith the throughflowof AAMW and 10%. Much larger residualsare found in the Arabian Sea. This northern subtropical gyre water, is enhanced, and most of its has a technical reason. RSW/PGW contributions were calculated water recirculates into the Somali Current. It forms a wind-driven as residuals from the mixing model. Nevertheless, we are cyclonic gyre in the equatorialIndian Ocean, similar to the North confidentwith suchan approachin our mixing model, sincethe Atlantic and North Pacific Ocean gyres. During the winter residual maps of Figure 10d show that omitting RSW/PGW monsoon, under the northeasterly wind, the Somali Current induces great errors. The large residuals are confined to the disappears and is replaced by a reverse current, the North Arabian Sea. EquatorialCurrent, which carriesa much smallervolume of water Below the {50=26.7 surface the seasonalsignals are weaker, southwardat a rathershallow depth [Wyrtki, 1973]. thoughthey do not disappear,as we have seenfrom the parameter To show clearer evidences of seasonal variation of thermocline distribution in section 3. Of course, the thermocline circulation circulation and ventilation in the Indian Ocean, four cross - couldbe expectedto be rathersluggish. This is especiallytrue in sections,two meridional sectionsalong 60øE and 90øE and two the northernIndian Ocean.Because seasonal signals on the lower zonal sectionsalong 10øS and 6øN, are mappedon the basis of two isopycnalsurfaces are not as apparentas thoseon the upper water-masscontribution already shown in Figures9-12. Figure 13 two surfaces,we have to analyzethe subtleseasonal differences is a meridional section at 60øE in the western Indian Ocean. A carefully.Moreover, because the residualduring the winter (the tongue-likestructure of consistentnorthward extension of ICW is right panelsof Figures 1l d and 12d) is relatively larger than found during the summerat 300-400 m (Figure13a, left panel). during the summer in the northern Indian Ocean, a spurious However, the structure is not so apparent during the winter, higher mixing ratio for the winter may sometimesoccur. For especially,south of 15øS(right panel). There is an evidencethat example,with an ICW mixing proportionof 10% in the Bay of more ICW moves to the north in the upper 500 m of the western 30 40 50 60 70 80 90 lO0 110 120 130 30 40 50 60 70 80 90 lO0 llO 120 130 30 30 30
2O 2O 2O 2O
10 10 10
0 ,u 0 ? -lO • -lO -10
-20 -2• -20 -20
-30 -30 -30 -30
-4O -4O -40 -4O 30 40 50 60 70 80 90 100 110 120 130 30 40 50 60 70 80 90 100 110 120 130 Longitude Longitude 30 40 50 60 70 80 90 100 110 120 130 30 40 50 60 70 80 90 100 110 120 130 30 30 30 30
20 20 20
lO lO lO
o ,g o o -10 • -10 -10
-20 -20 -20 -20
-30 -30 -30 -30
-40 -40 -40 40 30 40 50 60 70 80 90 100 110 120 130 30 40 50 60 70 80 90 100 110 120 130 Longitude Longitude
30 40 50 60 70 80 90 100 110 120 130 30 40 50 60 70 80 90 100 110 120 130 30 30 30 30 b d 20 20 20
10 10 10
o ,• o 0 -lO •{ -lO -10
-20 -20 -20
-30 -30 -30 -30
-40 -40 -40 30 40 50 60 70 80 90 100 110 120 130 30 40 50 60 70 80 90 100 110 120 130 Longitude Longitude
30 40 50 60 70 80 90 100 110 120 130 30 40 50 60 70 80 90 100 110 120 130 30 30 30
20 20 20
lO lO
0 o 0
-10 • -10 -10
-20 -20 -20 -20
-30 -30 -30 -30
-40 -40 -40 -40 30 40 50 60 70 80 90 100 110 120 130 30 40 50 60 70 80 90 100 110 120 130 Longitude Longitude Figure 10. Water masscontributions on the oo=26.7isopycnal surface (see Figure 9)' (a) the contributionof ICW, (b) the contributionof NICW, (c) the contributionof AAMW, and(d) the distributionof residuals. 30 40 50 60 70 80 90 100 110 120 130 30 40 50 60 70 80 90 100 110 120 130 30 --30 30 30
20 20 20 20
10 10 / 10
0 o 0 0 -10 • -10
-20 -20 -20 , . -20
-30 -30 -30 -30
-40 -40 -40 40 30 40 50 60 70 80 90 100 110 120 130 30 40 50 60 70 80 90 100 110 120 130 Longitude Longitude 30 40 50 60 70 80 90 100 110 120 130 30 40 50 60 70 80 90 100 110 120 130 30 30 30 30
20 20 20 20
10 10 10 10
0 0 o 0 0
-10 -10 • -10 -10
-20 -20 •. -20
-30 -30 -30 • -30
-40 -40 -40 -40 30 40 50 60 70 80 90 100 110 120 130 30 40 50 60 70 80 90 100 110 120 130 Longitude Longitude
30 40 50 60 70 80 90 100 110 120 130 30 40 50 60 70 80 90 100 110 120 130 30 30 30 •. 30 d t 20 20 20 20
10 10 10 10
0 0 o 0
, •-,•-,d•-• -10 • -10 -lO • -lO
-20 -20 -20 . -20
-30 -30 -30 ø/ -30
40-- 40 40.... •--•--,--+ , t --•'•---•--•--•-•- -40 30 40 50 60 70 80 90 100 110 120 130 30 40 50 60 70 80 90 100 110 120 130 Longimde Longimde 30 40 50 60 70 80 90 100 110 120 130 30 40 50 60 70 80 90 100 110 120 130 30 30 30 30
20 20 20 20
10 10 10 / 10
0 0 o 0 _• 0
• -10 -•0 • -lo
-20 -20 -20
-30 -30 -30 •o\ -30
-40 -40 -40 -- 30 40 50 60 70 80 90 100 110 120 130 30 40 50 60 70 80 90 100 110 120 130 Longitude Longitude Figure 11. Water masscontributions on the 0o=26.9isopycnal surface (see Figure 9): (a) the contributionof ICW, (b) the contributionof NICW, (c) the contributionof AAMW, and (d) the distributionof residuals. 30 40 50 60 70 80 90 100 110 120 130 30 40 50 60 70 80 90 100 110 120 130 30 30 30 30 ½ 20 20 20 20
lO - lO ? 10
0 o 0 0
-•o '•• -lO -10
-20 -20 -20 -20
-30 -30 -30 -30
-40 -40 -40 30 40 50 60 70 80 90 100 110 120 130 30 40 50 60 70 80 90 100 110 120 130 Longitude Longitude 30 40 50 60 70 80 90 100 110 120 130 30 40 50 60 70 80 90 100 110 120 130 30 30 30 30
20 20 20 20
10 10 10 10
0 o 0 0
-10 • -10 -10
-20 -20 -20 -20
-30 -30 -30 -30
-40 -40 -40 -40 30 40 50 60 70 80 90 100 110 120 130 30 40 50 60 70 80 90 100 110 120 130 Longitude Longitude
30 40 50 60 70 80 90 100 110 120 130 30 40 50 60 70 80 90 100 110 120 130 30 30 30 30 d 20 20 20 20
lO lO lO lO
0 o 0 0
-lO • -lO -lO
-20 -20 -20 -20
-30 -30 -30 -30
-40 -40 -40 -40 30 40 50 60 70 80 90 100 110 120 130 30 40 50 60 70 80 90 100 110 120 130 Longitude Longitude 30 40 50 60 70 80 90 100 110 120 130 30 40 50 60 70 80 90 100 110 120 130 30 30 30 30
20 20 20 20
10 10 10 10
• 0 0 o 0 0
• -10 ,.3 -lO • -lO -lO
-20 -20 -20 -20
-30 -30 -30 -30
-40 -40 -40 -40 30 40 50 60 70 80 90 100 110 120 130 30 40 50 60 70 80 90 100 110 120 130 Longitude Longitude Figure 12. Watermass contributions on the •=27.1 isopycnalsurface (see Figure 9)' (a) thecontribution of ICW, (b) the contributionof NICW, (c) thecontribution of AAMW, and(d) thedistribution of residuals. YOU: INDIAN OCEANSEASONAL THERMOCLINE CIRCULATION 10,413
-40 -35 -30 -25 -20 -15 -10 -5 0 5 10 15 20 25 30 Latitude
-50o
-700
ß40 -35 -30 -25 -20 -15 -10 -5 0 5 10 15 20 25 30 Latitude
-100 -
-300.
-soo- -700 • •' '•!• ••(:•.,' 'i•' --700--500
-900 i ._9oo
.... i .... • ...... i i .... [ .... i .... i , ,,, -35 -30 -25 -20 -15 -10 -5 0 5 10 15 20 25 30 Latitude
-300
• -50o -500
- -700
- -900
-40 -35 -30 -25 -20 -15 -10 -5 0 5 10 15 20 25 3 • Latitude Figure13. Watermass contributions in the western Indian Ocean along 60øE for (left)the summer monsoon and (fight)the winter monsoon: (a) thecontribution of ICW, (b) thecontribution of NICW, and (c) thecontribution of AAMW. 10,414 YOU: INDIAN OCEAN SEASONAL THERMOCLINE CIRCULATION
-lOO
-300 -300 / , -500
-7oo
-900
-40 -35 -30 -25 -20 -15 -11 -5 0 $ 10 15 20 25 30 Latitude
-500 '• -500 7OO • o -700
-900 -900
I .... I ' '''1 .... I .... I .... I .... I .... I .... I .... I' ' '' I .... I .... I .... I''''1 -40 -35 -30 -25 -20 -15 -10 -5 0 5 10 15 20 25 30 Latitude Figure 13. (continued)
Indian Ocean (note the difference in the 40 and 50% contour latitude. The 100% contributioneast of 110øE in Figure 15c lines, north of 10øS, between panels in Figure 13a). During indicates the source region of AAMW. Relatively stronger winter, more NICW flows to the south in the western Indian influenceduring summertime is seenin the upper300 m (Figure Ocean (Figure 13b, right panel). Relatively higher contributions 15c, left panel). Larger ICW percentagein the west, shownin of AAMW are found in the upper 300 m of the thermoclineat Figure15a, indicates the majorpathway of ICW. A moreapparent 10øS during the summer(Figure 13c, left panel). These results tongueof ICW mixing proportionin the left panelof Figure 15a supportWyrtkt's [ 1973] view that the summermonsoon has much suggestsa strongernorthward transition of ICW in the western stronger impact on thermocline water - mass than the winter boundaryduring the summermonsoon. The largestcontribution monsoonand pronouncedseasonal variation mainly occursin the of 70% near the African coast(Figure 15a, left panel) indicates upper400 m. the core of northward ICW at about 300 m. To show more In the eastern Indian Ocean at 90øE, ICW does not show an evidenceof the strongersummer signal, another zonal sectionis apparentnorthward tongue during the summer(Figure 14a, left chosenat 6øN (Figure 16). Thereforeone can examinethe flow panel). The rather densecontour lines which are more vertically patternboth in the west(at the Somalicoast), where a strongest distributed suggest a front there that effectively blocks a northwardflow is expected,and in the eastat Sri Lanka andin the northward penetrationof ICW. As in Figure 13a, this front is easternBay of Bengal,where flow in and out of the bay can be significantly weaker in the western Indian Ocean. There is compared.The left panel of Figure 16a showsthat duringthe evidencethat ICW is transportedto the Bay of Bengalfrom the summermonsoon, northward ICW is strongernot only at the west during the summer, since about 10% of the ICW westernboundary but also in thewestern Bay of Bengal.Note the contribution is found in the upper thermoclinenorth of 10øN 20 and 30% contourlines at 80øEin Figure16a. The higher (Figure 14a, left panel). This amountof ICW contributionis not contributionof 20-30%in the easternBay of Bengalimplies a seen during winter (right panel). The relatively stronger flow pointing southward as a return of the northward flow southwardtongue of NICW in the left panelof Figure 14b at 300 enteringfrom westof the bay. Interestingly,the left panelof m has some implication for the return flow of the western Figure 16c shows about a 10-20% AAMW contributionto the boundary current during the summer. A clearer pattern was westernboundary current at the Somalicoast during the summer. alreadyshown in the left panel of Figure 10b where a westward A similaramount of AAMW is foundsurrounding Sri Lanka.In return flow is depictedin the easternboundary region, merging contrast,an AAMW contributionof about20% evidentlyenters with the SouthEquatorial Current. AAMW is slightlystronger, the easternBay of Bengalduring the wintermonsoon (right especiallyin the upper 300 m, during the summer(Figure 14c, panel).Its path is probablyby way of the SouthEquatorial left panel). There seemsto be someleakage of AAMW into the Current and then the EquatorialCountercurrent, as described easternBay of Bengal.As describedbefore, this leakagemight before. occur when the water is advectedinto the easternequatorial regionby the EquatorialCountercurrent and prior to turningto a 6. Summary and Discussion southwardflow alongwestern Sumatra. Figure 15 shows a zonal sectionat 10øS.The westwardcross- Usingtwo seasonal hydrographic data sets in a mixingmodel of Indian Ocean flow of AAMW is approximatelycentered at this optimummultiparameter analysis combined with cluster analysis YOU: INDIAN OCEAN SEASONAL THERMOCLINE CIRCULATION 10,415
-too
-300
o$oo -700
.
-900
-35 -3o -25 -2o -15 -lo -5 o 5 lo 15 20 2.5 30 Latitude
• -700.
-900
-35 -30 -25 -20 -15 -t0 -5 0 5 10 15 20 2.5 30 Latitude
-tOO
-500 -500
-700 -700 -900 -900 -40 -35 -30 -25 -20 -15 -10 -5 0 5 lo 15 20 2.5 30 Latitude
-tOO-3oo • ( / •oo -300-lOO -5oo -500 -700 -700
-900 -900
-40 -35 -30 -25 -20 -15 -10 -5 0 5 t0 15 20 25 30 Latitude Figure14. Sameas Figure 13 butin theeastern Indian Ocean along 90øE. 10,416 YOU' INDIAN OCEAN SEASONAL THERMOCLINE CIRCULATION
-5oo
-700
4O -35 -30 -25 -20 -15 -lo -5 0 5 lo •5 20 25 30 Latitude
-40 -35 -30 -25 -20 -lS -10 -5 0 5 10 15 20 25 30 Latitude Figure 14. (continued)
has producedsome significant results in this paper.Quantified When thesenewly derivedresults are combinedwith previous water- massmixing proportionson selectedisopycnal surfaces studies, a clear schematic summer and winter thermocline showedlarge seasonalvariations of thermoclinecirculation and circulationin the IndianOcean is derived(Figure 17). Figure17a ventilationin theIndian Ocean, especially prominent in theupper showsthe summerthermocline circulation, and Figure 17b shows 400 m. Clusteranalysis became a usefultool for determiningthe its winterequivalent. Before interpreting the diagrams, we needto initialsource water types objectively. Fixing the sampling region discusssome broad conditions. First, currents and eddies advect in two overlappingclusters from two seasonsguaranteed the watersfrom their sourceregions and aremixed with otherwaters. accurateestimation of seasonalsignals in the initial sourcewater Therefore the nomenclature for water- mass and current should typedefinition and valid comparison of the signals.Water - mass be distinguished.In Figure17 we havedepicted names for water mixing proportionswhich were quantifiedand mappedon massesnear their sourceregions, while namesfor currentsare isopycnalsurfaces resulted in clearmixing patterns and spreading markedalong streamlines. Second, one usually names currents as paths for each water mass. Seasonalvariations of thermocline permanentflows withoutreversing their directionsduring the circulationand ventilation can thus be inferredfrom those mixing year. As we know, the IndianOcean represents an exception, patterns. sincethe reversalmonsoon regimes and their impact on currents However,the successfulapplication of the methodused in this are very profound.Those monsoonally controlled currents exert a paper still relies heavily upon skilled preconditioning.For profoundeffect throughoutthe thermocline.Therefore we retain example, without introducingthe virtual water- mass,North those frequently used names such as the "SouthwestMonsoon IpdianCentral Water, one wouldderive very differentmixing Current," "North Equatorial Current," "Equatorial patterns in Figures 9-12 which could not be interpreted Countercurrent",and for the westernboundary current, "Somali physically.Another difficulty occurswhen the numberof water Current." The East MadagascarCurrent [Tchernia, 1980; types to be solved for mixing exceedsthe numberof known Rochford,1966] is usuallydescribed as a southwardflow along equations.For example,we had four sourcewater masses, Indian theeastern Madagascar. However, results from this paper suggest Central Water, North Indian Central Water, Australasian a flow reversaltoward the north duringthe winter (austral MediterraneanWater andRed SeaWater/Persian Gulf Water,but summer).We thereforegive the nameonly to its southwardflow only six equations.The decisionwas madeto keep Red Sea during the northern summer, similar to the situation for the Water/Persian Gulf Water out of the model as residual. SomaliCurrent. Since the currents east and west of Indiachange Fortunately,such an approachdid not causea significantbias of dramatically with the monsoonreversal, we tend not to name the modelresults, since the large residual is confinedto onlythe those currents. Arabian Sea in the upper thermocline. Of course, such an Duringnorthern summer (Figure 17a), IndianCentral Water is approachmight haveeliminated the possibilityof Red SeaWater formednorthward along the subtropicalfront and carried to the spreadingto the southin the lower thermocline.In fact, thishas to subtropicsby the SouthIndian Ocean Current [Lutjeharms and be consideredwhen one or two additionalparameters were Van Ballegooyen,1984; Stramma, 1992]. Park et al. [1993] availablefor the modelanalysis. arguedto usea moreclassic name, the AgulhasReturn Current. Nevertheless,a picture of seasonalvariation of the Indian The SouthIndian Ocean Current has several northward paths. In thermoclinecirculation and ventilation emerges from our study. the southwestern Indian Ocean those northward flows form YOU: INDIAN OCEAN SEASONAL THERMOCLINE CIRCULATION 10,417
-lOO
-3oo
-5oo
-7oo
-9oo
40 50 60 70 80 90 lO0 110 120 130 Longitude
r , ] -100
•o -3
-500. m -700.
-900 -
40 50 60 70 80 90 100 110 120 130 Longitude
-lOO
! ,,,,•,• ...... , ...... , ...... , ...... , ...... , ...... ,,••,,,.-900 40 50 60 70 80 90 100 110 20 130 Longitude
• -500'
• -700-
-900 -
40 50 60 80 9O 100 110 120 130 Longitude Figure 15. Sameas Figure 13 but for a zonalsection along 10øS. 10,418 YOU: INDIAN OCEAN SEASONAL THERMOCLINE CIRCULATION
-lOO
-5oo
-700-
-900
40 50 60 70 80 90 100 110 120 130 Longitude '"-f'3øø • -5004••• -500 •-700 -700
i i 30 4o 50 60 70 80 90 100 110 120 130 Longitude Figure 15. (continued)
recirculation cells locally feeding the Agulhas Current. In the back into the western boundary current. It forms a cylonic gyre central basin, some branches of the South Indian Ocean Current similar to the North Atlantic and North Pacific [Wyrtki, 1973]. point to the north. Its easternmost extension is continued Some part of Indian Central Water in the South Equatorial northwardby the West Australia Current [Rochford, 1967]. East Current merges with the East Madagascar Current. It flows of the West Australia Current there should be a southward flow through Mozambique Channel as the Mozambique Current. In near the West Australia coast. There indeed exists a southward balancing the southward Mozambique Current, the West flow called the Leeuwin Current [Cresswelland Golding, 1980]. Madagascar Current [Tchernia, 1980] is a rather permanent However, the Leeuwin Current may be too shallow to have a northward flow, although it is largely an anticyclonic feature significant impact on the large-scale thermocline circulation [Wyrtki, 1973; Tchernia, 1980]. The returnflow of Indian Central below. Anyhow, we symbolically include a possiblesouthward Water complets its loop in the Agulhas Current, most of which flow there to balance the northward West Australia Current near returns back to the Indian Ocean as the South Indian Ocean the West Australia coastwithout naming it. Indian Central Water Current.A small part of it penetratesinto the Atlantic as the warm is carriedto the west by the SouthEquatorial Current. Some part water route. It compensatesfor the North Atlantic Deep Water of its water returns to the southwestIndian Ocean, forming a absorbedand advected by the Antarctic Circumpolar Current subtropicalgyre and supplyingthe Agulhas Current. Part of it is southof Africa [Gordon, 1986]. carried north of Madagascarin a path probably far offshore of The circulation of Australasian Mediterranean Water begins eastern Madagascar. It flows east of the southward East with a zonal flow westward crossing the Indian Ocean. Madagascar Current. Further northward progress of Indian AustralasianMediterranean Water is carriedwestward first by the Central Water occursby way of the East Africa CoastalCurrent Indonesian Throughflow (a current name, to be distinguished [Swallow et al., 1991]. Indian Central Water then enters the from the name Australasian Mediterranean Water used for water equatorial and northern Indian Ocean in the westernboundary mass) and then mergesinto the South Equatorial Current, which current, the Somali Current, and part is carried to the eastern bendssomewhat to the northunder stronger southwest monsoonal Indian Ocean by the Southwest Monsoon Current which is a forcing. Northeast of Madagascar, AustralasianMediterranean broad eastward flow. The remainder of the western boundary Water contributespart of its water to the AgulhasCurrent through currentflows farther to the north. It parallelsthe westernArabian the East MadagascarCurrent and the MozambiqueCurrent. The Sea in the East Arabian Current [Tomczakand Godfrey, 1994] remainder suppliesthe Somali Current throughthe East Africa and then turns southeastward,west of India, as part of the CoastalCurrent. Wyrtki [1973] suggestedthat abouttwo thirdsof Southwest Monsoon Current. Indian Central Water ends its the SouthEquatorial Current passes north of Madagascarand one northwardjourney enteringthe Bay of Bengal from west and east - third tums to the southalong the eastcoast of Madagascar(East of Sri Lanka. Madagascar Current). The northward path of Australasian The southward return route of Indian Central Water starts from Mediterranean Water resembles the Indian Central Water route. the easternBay of Bengal, southwardin the Sumatra-JavaCurrent We did not give a return route for Australasian Mediterranean [Rochford, 1967]. Indian Central Water returns westward, Water, since Australasian Mediterranean Water is a net inflow enhancingthe SouthEquatorial Current. A large part of it feeds water from the Pacific to the Indian Ocean. This is correct at least YOU: INDIAN OCEAN SEASONALTHERMOCLINE CIRCULATION 10,419
30 35 40 45 50 55 60 65 70 75 80 85 90 95 100 Longitude
--•..•o4'1•-3•
• -5•
-7•
30 35 40 45 so 55 60 65 70 7.5 80 85 90 95 100 Longitude
-100• • --• •-100
-900I.... I.... I.... I...... I' ' ' '•- .... I.... I' ' "l 'l q,.... I'' Qd•-900 30 35 40 45 50 55 60 65 70 75 80 85 90 95 100 Longitude
-100 . ' ' - -100
-500 oo• 400
400 • ' .i'• -700
-900 --900
30 35 40 45 50 55 60 65 70 7s 80 85 90 95 100 Longitude Figure 16. Sameas Figure 13 butfor a zonalsection along 6øN. 10,420 YOU: INDIAN OCEAN SEASONAL THERMOCLINE CIRCULATION
-lOO
-100 .-3oo lo -300 o..•c.• -5oo
-700
-700 ß -900 -900
30 35 40 45 50 55 60 65 70 75 80 85 90 95 lOO Longitude
--300
ß-500
7oo ß-700
ø900- ß-900 30I .... 35I 't , ,40 I .... 45I ...... 50 55I .... 60I .... 6I 5 .... 70 75 85 90 95 100 Longitude Figure 16. (continued) in depthrange of the thermocline.In the deeperlevel (below 1000 reverse northward. Therefore the East MadagascarCurrent m), Fieux et al. [1994] did find an opposite flow, but even vanishes during the northern winter. As we saw before, the averagingfrom the surfacedown to 1900 dbar,they still got a net westwardflow of AustralasianMediterranean Water is slightly transportof 18.6+7Sv (1 Sv=106m 3 s'l) flowinginto the Indian weakerduring the winterin theupper thermocline. One branch of Ocean from the Pacific. thesouthward flow of theSouth Equatorial Current seems to split During northern winter (Figure 17b), the South Indian Ocean far off East Madagascar.This in turn addssome weight to the Current bendsto the north and so does the subtropicalfront, a reversedflow east of Madagascar.A radical changein the factor alreadypointed out by Sverdrupet al. [1942], thoughthey thermoclinecirculation during the winter is thedisappearance of did not name this current.The northwardshift of the subtropical the SomaliCurrent, which is replacedby a southwardflow of the front during the winter causesthe flow east of Madagascarto North EquatorialCurrent. Indian Central Water then has to take
Equa tort al
Subtroptcal Gyre
).
Front Indtan Ocean
20 ø 40 ø 60 ø 80" 100 • E 120 ø
Figure 17. Schematic(a) summerand (b) winterthermocline circulation in the Indian Ocean. YOU: INDIAN OCEAN SEASONAL THERMOCLINE CIRCULATION 10,421
Equatortal
Current
Indtan Central Water SubtroptcaltGyre
South Indian Ocean Curre Sub tropical Fr b I I 20' 40' loo' E 120' Figure17. (continued) its way in the EquatorialCountercurrent. Results from this study the INDIGO data. Some archiveddata were collectedduring my visit to suggestsome leakage of the EquatorialCountercurrent into the MassachusettsInstitute of Technology.Some suggestionfrom T. Mtiller is acknowledgedabout the presentationof the final sketch, Figure 17, easternBay of Bengal.This leavesa possibilityfor the sourceto which was nicely drawn by A. Eisele. Constructivecomments from two feed the westwardNorth EquatorialCurrent from the eastand to anonymousreviewers, especially the secondreviewer, were very helpful form a loop. As describedbefore, this leakageoccurs when the in improvingthe presentationof the results. water is transportedinto the easternequatorial Indian Oceanby the Equatorial Countercurrent. Most of the Equatorial References Countercurrent then turns southward and becomes the Sumatra- Java Current. Nevertheless, about 10% of Indian Central Water Colbom,J. G., The ThermalStructure of the Indian Ocean, 173 pp., Univ. and 10-20% of Australasian Mediterranean Water leak into the of Hawaii Press, Honolulu, 1975. easternBay of Bengal in the upper thermoclineas seenin Figure Cox, M.D., A baroclinic numerical model of the world ocean: 16. Preliminary results,in Numerical Models of Ocean Circulation, 364 pp., Natl. Acad. Press,Washington, D.C., 1975. Wyrtki [1973] suggestedthat somelow-salinity water originates Cresswell, G. R., and T. J. Golding, Observationsof a south-flowing from the Bay of Bengal flowing westward during the northeast currentin the southeasternIndian Ocean, Deep - Sea Res.,part A, 27, monsoon.One branch continueswestward along 5øN, and the 449-466, 1980. other northwestwardalong the coastof India. Sheryeet al. [ 1991] Defant, A., Physical Oceanography, vol. I., 729 pp., Pergamon, Tarrytown, N.Y., 1961. also observedthis low-salinity water of the Bengal Bay origin off Emery, W. J., and J. Meincke, Global water masses:Summary and West India. In the Arabian Sea, Tomczak and Godfrey [1994] review, Oceanol. Acta, 9, 383-391, 1986. found a cyclonic eddy-like flow during the winter. Therefore the Fieux, M., C. Andri6, P. Delecluse, A. G. Ilahude, A. Kartavtseff, F. East Arabian Current shouldstill exist. It pointsto the north from Mantisi, R. Molcard, and J. C. Swallow, Measurements within the the Gulf of Aden. These studies further indicate a southwestward Pacific-IndianOceans throughflow region, Deep - Sea Res. Part I, 41, 1091-1130, 1994. flow across the central Arabian Sea. This conjecture is Gordon, A. L., Interocean exchangeof thermocline water, J. Geophys. commensuratewith the water- massspreading paths in the fight Res., 91, 5037-5046, 1986. panelsof Figures9a, 9d, and 10a. Jamous,D., L. M6mery, C. Andri6, P. Jean-Baptiste,and L. Merlivat, The The above schematic summer and winter thermocline distributionof helium 3 in the deepwestern and southernIndian Ocean, J. Geophys.Res., 97, 2243-2250, 1992. circulation is gathered from the parameter distributions and Lutjeharms,J. R. E., and R. C. Van Ballegooyen,Topographic control in patterns of large-scale water- mass mixing and spreadingon the AgulhasCurrent system, Deep - Sea Res.,31,1321-1337, 1984. selectedisopycnal surfaces.It has not yet been combinedfully Mackas,D. L., K. L. Denman,and A. F. Bennett,Least - squaresmultiple with observed velocity fields and transports, although basic traceranalysis of water masscomposition, J. Geophys.Res., 92, 2907- 2918, 1987. observed flow paths of currents have been inferred. The Mamayev, O. I., Temperature-SalinityAnalysis of World Ocean Waters, schematicsin Figure 17 thereforecan only be a guide to possible 174 pp., Elsevier,New York, 1975. flow paths of the thermocline waters of the Indian Ocean and McCartney, M. S., Subantarcticmode water, in A Voyage of Discovery, their seasonalchanges with the monsoonregime. George Deacon 70th AnniversaryVolume, edited by M. Angel, pp. 103-109, Pergamon,Tarrytown, N.Y., 1977. Olson, D. B., G. L. Hitchcock, R. A. Fine, and B. A. Warren, Acknowledgments. This studystarted at the Schoolof Earth Science, Maintenanceof the low-oxygenlayer in the centralArabian Sea, Deep Flinders University of South Australia, Adelaide, Australia and was - Sea Res., 40, 673-685, 1993. finished during my stay as a visiting investigator at Institut for Park, Y.-H., L. Gamberoni, and E. Charriaud, Frontal structure,water Meereskundean der Universit•it Kiel, Germany. I thank John Toole for masses,and circulation in the Crozet Basin, J. Geophys.Res., 98, the CD29 data, Donald Olson for the MASAI data and Daniel Jamous for 12,361-12,385, 1993. 10,422 YOU: INDIAN OCEAN SEASONAL THERMOCLINE CIRCULATION
Quadfasel,D. R., and F. Schott,Water-mass distribution at intermediate Toole, J. M., and B. A. Warren, A hydrographicsection acrossthe layersoff the Somali coastduring the onsetof the southwestmonsoon, subtropicalSouth Indian Ocean,Deep - SeaRes., 40, 1973-2019, 1993. J. Phys. Oceanogr.,12, 1358-1372, 1982. Warren, B. A., Transindianhydrographic section at lat. 18øS:Property Rochford, D. J., Distribution of Banda Intermediate Water in the Indian distributionsand circulation in the South Indian Ocean, Deep - Sea Ocean, Aust. J. Mar. Freshwater. Res., 17, 61-76, 1966. Res., Part A, 28, 759-788, 1981. Rochford,D. J., The phosphatelevels of the major surfacecurrents of the Warren, B. A., H. Stommel, and J. C. Swallow, Water masses and Indian Ocean, Aust. J. Mar. Freshwater. Res., 18, 1-22, 1967. patternsof flow in the Somali Basin duringthe southwestmonsoon of Sharma,G. S., Transequatorialmovement of water massesin the Indian 1964,Deep - SeaRes., 13, 825-860, 1966. Ocean, J. Mar. Res., 34, 143-154, 1976. Wyrtki, K., Scientificresults of marineinvestigations of the SouthChina Shetye, S. R., A.D. Gouveia, S.S. C. Shenoi,G. S. Michael, D. Sundar, Sea and the Gulf of Thailand 1959-1961. Naga Rep., 2, 195 pp., A.M. Almeida, and K. Santanam,The coastalcurrent off western India ScrippsInst. of Oceanogr.,Univ. of Calif., SanDiego, La Jolla, 1961. duringthe northeastmonsoon, Deep - SeaRes., 38, 1517-1529,1991. Wyrtki, K., OceanographicAtlas of the International Indian Ocean Stramma,L., The SouthIndian Ocean Current,J. Phys. Oceanogr.,22, Expedition,531 pp., Natl. Sci. Found.,Washington, D.C., 1971. 421-430, 1992. Wyrtki, K., Physicaloceanography of the Indian Ocean, in Ecological Sverdrup,H. U., M. W. Johnson,and R. H. Fleming, The Oceans,Their Studies:Analysis and Synthesis,vol. 3, editedby B. Zeitzschel,pp. 18- Physics, Chemistry and General Biology, 1087 pp., Prentice-Hall, 36, Springer-Verlag,New York, 1973. EnglewoodCliffs, N.J., 1942. You, Y., Dianeutralmixing in the thermoclineof the Indian Ocean,Deep Swallow, J. C., F. Schott,and M. Fieux, Stractureand transportof the - Sea Res., 43, 291-320, 1996. East African Coastal Current, J. Geophys.Res., 96, 22,245-22,257, You, Y., and M. Tomczak, Thermocline circulationand ventilation in the 1991. IndianOcean derived from watermass analysis, Deep-Sea Res. 40, 13- Tchernia, P., Descriptive Regional Oceanography,253 pp., Pergamon, 56, 1993. Tan3rtown,N.Y., 1980. Tomczak, M., A multiparameter extension of temperature/salinity diagramtechniques for the analysisof non-isopycnalmixing, Prog. Y. You, Center for Climate SystemResearch, University of Tokyo, 4-6- Oceanogr.,10, 147-171, 1981. 1 Komaba,Meguro-ku, Tokyo 153, Japan. Tomczak, M., and J. S. Godfrey, Regional Oceanography: An Introduction,422 pp., Pergamon,Tan3rtown, N.Y., 1994. Tomczak,M., and D. Large, Optimummultiparameter analysis of mixing (ReceivedJune 14, 1995;revised September 13, 1996; in the thermoclineof the east Indian Ocean,J. Geophys.Res., 94, acceptedOctober 17, 1996.) 16141-16149, 1989.