IMPACTS OF THE TROPICAL PACIFIC COUPLED PROCESS ON THE INTERANNUAL VARIABILITY IN THE INDIAN OCEAN

Vol.16 No.3JOURNAL OF TROPICAL METEOROLOGYSeptember 2010Article ID: 1006-8775(2010) 03-0271-09IMPACTS OF THE TROPICAL PACIFIC COUPLED PROCESS ON THEINTERANNUAL VARIABILITY IN THE INDIAN OCEANFENG Jun-qiao (冯俊乔)" , BAI Xue-zhi (白学志)(1. Institute of Oceanology, Chinese Academy of Sciences, Qingdao 266071 China; 2. Key Laboratory ofOcean Circulation and Waves, Chinese Academy of Sciences, Qingdao 266071 China)Abstract: The basic features of climatology and interannual variations of tropical Pacific and IndianOceans were analyzed using a coupled general circulation model (CGCM), which was constituted with anintermediate 2.5-layer ocean model and atmosphere model ECHAM4. The CGCM well captures the spatialand temporal structure of the Pacific EI Nino-Southern Oscillation (ENSO) and the variability features inthe tropical Indian Ocean. The influence of Pacific air-sea coupled process on the Indian Ocean variabilitywas investigated carefully by conducting numerical experiments. Results show that the occurencefrequency of positive/negative Indian Ocean Dipole (IOD) event will decrease/increase with thepresence/absence of the coupled process in the Pacific Ocean. Further analysis demonstrated that theair-sea coupled process in the Pacific Ocean affects the IOD variability mainly by influencing the zonalgradient of thermocline via modulating the background sea surface wind.Key words: coupled GCM; IOD; tropical Pacific Ocean; ENSOCLC number: P732.6Document code: Adoi: 10.3969/j.issn. 1006-8775.2010.03.0091 INTRODUCTIONreduced cloud cover and increased solar radiation overthe eastern Indian Ocean. Therefore, the increase of netThe air-sea coupled system in the Indian Oceanheat flux results in warm SSTA in all regions but thehas long been considered a response to that of thesouthwest tropical Indian Ocean. Albeit with thePacific Ocean. Research on it is rare due to the lack ofinstantaneous atmospheric response， the SSTobservations. However, air-sea interactions in thevariability of the eastern Indian Ocean usually lagsIndian Ocean have attracted unprecedented interestbehind the ENSO variabilityl". Xie et al.!tsI alsosince the discovery of IOD phenomenon', which ispointed out that the two anticyclonic circulations oncharacterized by opposite signs of sea surfaceboth sides of the equator, produced by the equatorialtemperature anomaly (SSTA) in the western andanomalous easterly associated with ENSO， caneastern tropical Indian Ocean. Then, the debate hasemanate westward propagating Rossby wave, whichshifted the focus on the relationship between the Pacificwill lead to warm SSTA in the western Indian Ocean.and Indian Ocean variabilityl. Many_ scientistsSince it usually takes a few months for thebelieve that they are related to each other-/, i.e., thedownwelling Rossby wave to propagate to the west,ENSO variability can be affected by variations in thewarm SSTAs in the western and eastern Indian OceanIndian Ocean-15, and vice versa.are not synchronous. Using a full CGCM ofPrevious research studied the influence of ENSOSINTEX-F1 (Scale Interaction Experiment Frontieron the Indian Ocean by either analyzing observationsversion 1), Behera et al.7] studied the I0D event and)r performing numerical experiments. So far, mostits relationship with ENSO. Their results suggestedscientists believe that the ENSO event can significantlythat the prominent period of Dipole Mode Index (DMI)affect the Indian Ocean SST variability through bothis about 3.5- 4 years in the control run with a globalatmospheric and oceanic processeslt. 5. During theair-sea coupling. However, the main period of DMI ismature phase of EI Nino, the suppressing of thabout 2-year in the numerical experiment without theascending branch of the Walker circulation leads tcENSO process and the, IOD mode turms into the first中国煤化工YHCNMHGReceived date: 2009-1 1-23; revised date: 2010-05-28Foundation item: National Natural Science Foundation of China (40776013, 40306006); State Key BasicResearch Development Project (2006CB403603)Biography: FENG Jun-qiao, Ph.D., primarily undertaking research on air-sea interactions.E-mail for corresponding author: xuezhi.bai@gmail.com272Jourmal of Tropical MeteorologyVol.16mode, rather than the second mode as in the controlPacific_ Decouplerespectivelyhereafter. Bothrun.experiments were integrated for 40 years. TheThe ocean- atmosphere system in the Indo-Pacificclimatological fields were obtained by averaging theregion has a significant influence on the ambient, or34-year output over simulation year 7 to 40. Theeven global, weather and climate. So it is not only ofoutputs from year 11 to 40 were selected to analyze thesocial benefits but also very important in practice tointerannual variability.study the air-sea interaction and its mechanism in thisregion. Although a number of works regarding theffect of ENSO on the Indian Ocean have beenpublished, there is not yet specific agreement on thisMixed layerissue because of its complexity, and most of the resultsare based on the full CGCM. Here, in the present study,. Entrainment layerwe will further investigate the above scientific problemin the Indo-Pacific region by utilizing an intermediateThermocline layerCGCM, which has lower computational costs andmuch clearer physical processes in contrast to the fullycoupled model.The organization of this paper is as follows. InDeep sea resting layersection 2, the air-sea coupled model used in the studywill be briefly described. The model outputs areanalyzed and the results will be shown in section 3.Finally, section 4 comprises an overall summary andconclusions.Fig. 1. Schematic diagram of the 2.5-layer ocean modelstructure2 MODEL DESCRIPTION ANDEXPERIMENTS DESIGN3 RESULTSThe intermediate coupled model, used in the3.1 Experiment designpresent paper, was developed by University of Hawaii.Details of it are described in Wang et al."8] and Fu et3.1.1 HORIZONTAL FLOW FIELDSal.!9. Only a brief description is given here. TheThe CTRL-simulated annual mean SST, togetheratmospheric component is ECHAM4 from Max Planckwith its difference from the observation, is shown inInstitute for Meteorology (MPI-M), with a horizontalFig.2. It can be seen that the main SST features in theresolution of T42 (a triangular truncation at waveIndo-Pacific region are well captured by thenumber 42)， about 2.8*x2.8, and 19 levels in theintermediate CGCM, which indicates the skills of thevertical. A 2.5-layer ocean model (as shown in Fig. 1)model in reproducing the basic status of climateincluding a variable mixed layer, a variable :variability. The difference between observed SST andthermocline layer and a deep resting layer, was used asthat from CTRL is large in some regions, like, alongthe oceanic component in the CGCM. Its horizontalthe coast of the eastern Pacific where it is 1.5- 2°C.resolution is 0.5*x0.59. The ocean simulation region isHowever, the difference is lower than 0.5°C in the(30°S- 30°N，0°- 360°E). The atmosphere modelregions of the western Pacific warm pool and theexchanges information with the ocean model once aIndian Ocean, where the air-sea interaction is active.day. The former provides the latter with daily meanFigure 3 shows the distribution of thermoclinesurface winds and heat fluxes, and the latter sendsdepth and surface wind stress in the Indo-Pacific region.daily mean SST back to the former. The initialIt is ilustrated in the Fig. 3a that the deep/shallowatmospheric field is from the European Centre fothermocline in the western/eastern Pacific Ocean,Medium- Range W eather Forecasts (ECMWF) analysisassociated with the tropical trade wind, corresponds toon January 1, 1988. The initial ocean field is a steadythe Pacific warm pool/cold tongue (Fig. 2a), whichstate for January of 10-yr integrations of the oceanfurther in中国煤化主goud is simulatedmodel forced by observed climatological surface windsuccessful二he Indian summerand heat flux. Two numerical experiments weremonsoon,TYHC N M H Gbecormes strong inconducted. In the first experiment the global tropicalthe southeast Indian Ocean. Consequently, sea water inocean was fully coupled with the atmosphere, while thethe eastern Indian Ocean accumulates to the west andPacific Ocean was decoupled in the second one. Thedeepens the thermocline there. Meanwhile, the Somalitwo experiments will be refered to as CTRL andcurrent is intensified because of the enhanced windNo.3FENG Jun-qiao (冯俊乔) and BAI Xue -zhi (白学志)273forcing. As a result, the upwelling enhances, partlyThe results during the northeast monsoon season arecontributing to the western Indian Ocean SST cooling.generally opposite (Fig. 3b).10NEG10206080E100E120E 140E160E18160W140W120W 100W0W30N .20N+(tEC60E 80E120E 140E 160E180 160W 140W120W 100W 80WFig. 2. Simulated SST in CTRL experiment (a) and its difference from the observations (b) Unit: °C30N下201310N，EQoS-20S十(a)120E130E140E150E160E170E 180 170W 160w 150W 140W 130W 120w 110W 100w 90w 80w 70W20N-007040Q一805020S50EFig. 3. Thermocline depth (shaded, unit: m) and sea surface wind stress (vector,中国煤化工120ual mean in PacificOcean, (b) Indian Ocean in February, and (c) Indian Ocean in AugustHCNMHG3.1.2 INTERANNUAL VARIABILITYthe prominent period of ENSO simulated in CTRL isTime series of Nino3 index and its wavelet4 -7 years, closing to the observation.analysis are depicted in Fig. 4. We can see clearly that274Journal of Tropical MeteorologyVol.16a)2(2Year303E4C。(6)1"5-0.563264Period (Year)Fig. 4. (a) Nino3 index time series with 5-month running mean series in dashed line (unit: °C), and (b) its dominant periodfrom wavelet analysis with dashed line denoting 95% significant level20N一20N10N」10NQ一,EQ+，;0.0S一1 10S.0.1^20SOE 40E 50E60E70E80E90E100E110E120E30E40E50E60E70E80E 90E 100E 110E 120E4(d)152025.30354025....303:[(e)[(f)每2|.025)580.25中国煤化工816Fig. 5. First EOF mode associated with SSTA in CTRL: (a) space distribMYHC N M H Gelet analysis of timeseries (with dashed line denoting 95% significance level). (b), (d), (f) are the same as (a), (c), (e), respectively, but forthe second EOF mode.In order to investigate the interannual variability inthe Indian Ocean， EOF (Empirical OrthogonalNo.3FENG Jun-qiao (冯俊乔) and BAI Xue -zhi (白学志)275Function) analysis was carried out on the SSTA field.EOF analysis for Hadley SST over 1971-2000 wasResults are presented in Fig. 5. The spatial patternconducted. Results are shown in Fig. 6. The first andassociated with the first mode, which accounts for 31%the second EOF modes represent 30% and 13% of theof the total variance, exhibits a basin-wide union mode.total variance, respectively. The primary periods ofFor its corresponding time series, a main period iscorresponding time series are 3- -6 and 2 years withabout 5- -8 years (Fig. 5e)， consistent with that ofconfidence level exceeding 95%. By comparing Fig. 5ENSO, which suggests that the first mode is related towith Fig. 6, we can conclude that the CTRL experimentENSO. Thesecond substantialEOFmode,basically reproduces the interannual variability of therepresenting 7% of the variance for the SSTA, is theIndian Ocean, albeit with some differences both indipole mode which is characterized by opposite signs inENSO period and IOD pattern, which may havethe western and eastern Indian Ocean. Correspondingly,something to do with the resolution of the model.the time series derived from the wavelet analysis showAdditionally, the air-sea flux with monthly mean solaran interannual signal with a 2-year period.radiation may also contribute to the error in the modelTo compare the above results with the observation,integration.ON ta20N .(b.呢10N: EQ10S10S-20S .20SIOE 40E 50E 60E 70E 80E 90E 100E 110E 120E 30E 40E 50E 60E 70E 80E 90E 100E 110E 120E| (d)pw197519801985.19019952000 .1985. 19902000YearYe25(e) .2015包15詹1.豆1052528160.25 0.5Period (Year)Fig. 6. Same as Fig. 5, but for Hadley SsTIn a word, this intermediate coupled model canfrom CTRL.well simulate the seasonal and interannual variability3.2.1 INDIAN OCEAN VARIABILTY IN THE PACIFIC _DECOUPLEin the Indo-Pacific region.SSTA variability was also investigated by carrying3.2 Effects of Pacific coupled process on the Indianout EOF analysis. The first and the second EOF modesOceanrepresent中国煤化工total variance,respectivespatial pattern ofIn order to assess the influence of Pacific coupledthe primal.MYHCNMHG'basin-wide mode.process on the Indian Ocean variability, we designed aHowever，the wavelet analysis result of temporalPacific_ Decouple experiment, in which the oceancoefficients shows a dominant period of about 0.5- 1model in Pacific was decoupled from the atmosphereyear, suggesting that the prominent variation is not anmodel. This section will compare its outputs with those276Journal ofTropical MeteorologyVol.16interannual signal. The second EOF mode isspatial variability of IOD are close to those simulatedcharacteristic of IOD, with the associated time seriesin CTRL, indicating the occurrence of I0D eventsshowing about a 2-year period. Both the temporal andwithout the Pacifie coupled process.20N( Q,10NEQEQ-D10S0.5-1 10S0S十8影S (0420530E40ES0E60E70E8OE90E100110E10E30E4E50E60E70E8OE9OE100E110E120E(C).I(d)1520 2:3035Year8(e)s(f)60.2500.5(Year)Period (Year)Fig 7. Same as Fig 5 but for SSTA in Pacific_ Decouple experiment3.2.2 INFLUENCE ON THE loD PATTERNleast 3 months. A positive/negative IOD event isBased on the previous studies, we also selected thedefined by a positive/negative value of DMI.difference of SSTA between western and eastermFigures 8a & 8c show the DMI time seriesIndian Ocean to characterize the strength of IOD. Thesimulated in CTRL and Pacific Decouple experiments,SSTA was preprocessed as follows. The zonal mean ofrespectively. Wavelet analysis results are shown inSSTA was firstly extracted from the original SSTAFigs. 8b & 8d, with the dashed line denoting aseries at each grid point so as to get rid of the seasonalconfidence level of 95%. According to the identifyinginfluence of solar radiation on the whole basin. Thencriteria given above, the numbers of positive/negativethe Western Indian Ocean Index (WI) and EasternIODevents in CTRLandPacific_ DecoupleIndian Ocean Index (I) were calculated by theexperiments are 5/7 and 8/6, respectively, over 3average of SSTA in the regions (5°S- 5°N, 60°E -75°E)years of the model outputs. Composite patterns ofand (10°S- 0", 90°E-110*E), respectively. By carryingpositive events from the two experiments are shown inout the bandpass filter between 3 months and 8 years,Fig. 9. It can be demonstrated from Fig. 8 and Fig. 9we can obtain the interannual time series of WIl andthat the I0D phenomenon with mainly 2-year periodEII. The DMI, which evaluates the I0D strength, iscan be sin中国煤化工ents However, thedefined as the difference between WII and ElI. WeamplitudeYHCNMHG the IOD eventsadopted the occurrence criteria of IOD introduced bysimulatedlarger than that inSajl', i.e., signs of WIl and Ell should be oppositePacific_ Decouple, as suggested in Figs. 8a & 8b. Inwhich sustain for at least 3 months, and DMI isother words, when there are air-sea coupled processesrequired to exceed 0.5 standard deviation (STD) for at .in the Pacific Ocean, the strength of I0D can beNo.3FENG Jun-qiao (冯俊乔) and BAI Xue-zhi (白学志)277enhanced, and the number of positive/negative IODprobability of IOD, independent of the ENSO eventsevents decreases/increases as suggested by thesimulated in CTRL, is 59%.statistical analysis. In addition, the occurrence(a)(C)Aht20Year52535Year'40s(b).(d)品2.5Period (Year)Fig. 8. Time series of DMI simulated in (a) CTRL and (b) Pacific_ Decouple, with the dashed line denoting a 5-month runningmean (unit: °C). Prominent period of DMI using Wavelet analysis for (c) CTRL, (d) Pacific_ Decouple, with dashed linedenoting 95% significance level20N10NEQ{-0.4.10S10S .20S .0.6~叫20S4OE 5060E 70E80E90E100E110E 120EOE 5060E 70E 80E 90E 100E 110E 120EFig. 9. Composite SSTA pattern in the mature phase during a positive I0OD event simulated in (a) CTRL and (b) Pacific_ Decouple(Unit: °C)In order to further understand how the coupledsignificant thermocline difference between the west andprocess in the Pacific influences the IOD, the seasonalthe east (Fig. 10a). Therefore, the spatial pattern 01cycle of sea surface wind difference, thermocline depthannual mean is similar to that during the summer time.difference and SST difference between CTRL andThis background condition is in favor of the occurrencePacific_ Decouple in the equatorial Indian Ocean wereof negative IOD events and disadvantageous to that ofanalyzed. Results are displayed in Fig. 10. From apositive IOD events. Furthermore, the meridional windclimatological point of view, when the coupled processchange in some distinct regions also has someoccurs in the Pacific Ocean, the anomalous easterlycontribution. For instance, the anomalous southwardand westerly are popular during winter and summerwind along the Sumatra-Java coast can somewhat favormonsoon (Fig. 10a), respectively. In other words, thethe development of negative IOD events.coupled process in the Pacific can lead to strongsummer and winter monsoons in the Indian Ocean. As3.2.3 INFLU中国煤化工Na result, during summer and autumn, the thermocline isTheCNMHGd in CTRL and"THshallower in the west than in the east (Fig. 10b), andPacific_ Decoupie expriicnis are displayed in theSSTA is colder in the west than in the east (figures notupper and lower panels of Fig. 11, respectively, withshown). However, the anomalous easterly appearingthe shades denoting values exceeding 0.4. It is clearlyduring winter monsoons is too weak to induceseen from the figures that the SSTA variability is278Journal of Tropical MeteorologyVol.16significantly influenced by the Pacific air-sea coupledactivity in south tropical Indian Ocean 159S- 0° overprocess during most seasons, especially in the regionDecember- March is also remarkable. These factsnorth of 109S. Generally, from October to next May, afurther confirm that the IOD is stronger in CTRL, atime that covers the life-cycle of ENSO, the STD ofconclusion we have come to in previous sections.SSTA north of 109S is relatively larger in CTRL. The()TRL-Pocific_ Decouple zWS(5*S-0)(b)CTRL -Pocifli. Decouple 几L (5*S-0)DECT] DE-0.05--|NC-0.05-OCT|o0sEP 20.1-|sEP.|AUG-一10-0.05Ju-0.05. J0.05MAYH口>」 MI-10**”APR|APRAR<0.0 MAR 10.-FEB-0.05-.NDE50E 60E 70E 80E 900NOE 5OE 6E 70E BOE 9OoE 100(c)CTRL -Pacific_ Decouple(Aug.)(d)CTRL-Pacific_ _Decouple(Annual)ONτ20ON10EQECos05-Fig. 10. Seasonal cycle of difference between CTRL and Pacific_ Decouple in the equatorial region (5°S- 0°) of the tropical IndianOcean for (a) zonal wind stress (ZWS) (unit: Pa) and (b) thermocline layer depth (TL) (unit: m). The wind stress and thermoclinedepth difference between CTRL and Pacific_ Decouple of the tropical Indian Ocean in (c) summer and (d) annual mean (The vecoris wind stress, unit: Pa; the shades are thermocline depths, unit: m)(a) Annual mean(b) 0ct.- May(c) Jun. - Sep. .20N0N-(d) Annual mean(e) 0ct. - May(f) Jun. - Sep.0.3。40ES0E60E70E80E90E100110E120E40E50E6DE 7OE 8OE 90E100E110中国煤化工I108Fig. 11. Standard deviation of SSTA in CTRL for (a) annual mean, (b) OYHCN MH5tember, and that inPacific_ Decouple for (d) annual mean, (e) October- next May, (f) June- SeptembGalues that exceed 0.4(unit: °C)No.3FENG Jun-qiao (冯俊乔) and BAI Xue-zhi (白学志)2794 SUMMARY AND CONCLUSIONS[6] KARUMURI A, GUAN Z Y, YAMAGATA T. A look at therelationship between the ENSO and the Indian Ocean dipoleThis present work studied the variability in the[]. J. Meteor. Soc. Japan, 2003, 81(1): 41-56.tropical Indian and Pacific Oceans by employing an[7] WU R G KIRTMAN B. Understanding the impacts of theIndian Ocean on ENSO variability in a coupled GCM []. J.intermediate CGCM. The CGCM is capable ofClimate, 2004, 17: 4019-4031.reproducing the seasonal and interannual signals in the[8] SAJI N H, YAMAGATA1 T Possible impacts of IndianIndian and Pacific Oceans. It well captures the spatialOcean dipole mode events on global climate [J]. Climate Res.,pattern of ENSO and its irregular period feature. The2003, 25: 151-169.[9] KUG J s, KANG I s. Interactive feedback between the .IOD phenomenon is also simulated sucessfully. ByIndian Ocean and ENSO []. J. Climate, 2006, 19: 1 784-1carrying out numerical experiments, the influence 0801.the coupled process in the Pacific on the Indian Ocean[10] WU R G KIRTMAN B. Understanding the impacts of the .was investigated. It was found that, if the atmosphereand ocean couple together in the Pacific Ocean, theClimate, 2004, 17:4 019-4 031.occurrence of a positive/negative IOD event will be[11] ANNAMALAI H, XIE S P, MCCREARY J P, et al. Impactmore/less frequent. Further analysis suggested that theof Indian Ocean sea surface temperature on developing ElNino [J]. J. Climate, 2005, 18: 302-319.coupled process in the Pacific Ocean can affect the[12] YU J Y, MECHOSO C R, MCWILLIAMS J C, et al.zonal gradient of the tropical Indian Ocean thermoclineImpacts of the Indian Ocean on the ENSO cycle []. Geophys.depth through modulating sea surface wind, and henceRes. Lett, 2002, 29(8): 461-464.the IOD. Moreover, results also show that the Pacific13] KUG J S, KIRTMAN B P, KANG 1 s. Interactiveair-sea coupled process affects the variability north offeedback between ENSO and the Indian Ocean in an10°S in the Indian Ocean significantly during the lifeinteractive ensemble coupled model []. J. Climate, 2006, 19:6 371-6 381.cycle of ENSO over October- May, but insignificantly[14] LAU N C, NATH M J. Impact of ENSO on the variabilityduring June- September.of the Asian- Australian monsoons as simulated in GCMexperiments [J]. J. Climate, 2000, 13: 4287-4 309.REFERENCES:[15] XIE S P, ANNAMALAI H, SCHOTT F A, et al. Structureand mechanisms of south Indian Ocean climate variability [J].. [1] SAJI N H, GOSWAMI B N, VINAYACHANDRAN P N, etJ. Climate, 2002, 15: 864-878.al. A dipole mode in the tropical Indian Ocean [J]. Nature,[16] KLEIN S A, SODEN B J, LAU N G Remote sea surfacetemperature variation during ENSO: Evidence for a tropical1999, 401: 360-363.[2] LI Chong-yin, MU Ming quan. The dipole in the equatorialatmosphere bridge []. J. Climate, 1999, 12: 917-932.Indian Ocean and its impacts on climate []. Atmos. Sci, 2001,[17] BEHEAR S, LUOJ J, SEBAASTIEN M, et al. A CGCMstudy on the interaction between IOD and ENSO [J]. J.25 (4): 433-442.[3] CHAO Ji-ping, YUAN Shao-yu, CAI Yi. Large-scaleClimate, 2006, 19: 1 688-1 705.[18] WANG B, LI T M, CHANG P. An intermediate model ofair-sea interaction events in the tropical Indian Ocean [J]. J.the tropical Pacific Ocean [J]. J. Phys. Oceanogr, 1995, 25: 1Meteor., 2003, 61(2): 508-512.[4] FENG M, MEYERS G Interannual variability in the599-1 616.tropical Indian Ocean: A two-year time scale of IOD [J].[19] FU X H, WANG B. A Coupled modeling study of theseasonal cycle of the Pacific cold tongue, Part I: simulationDeep-Sea Res, 2003, 50: 2 263-2 284.[5] LIU Lin, YU Wei-dong, LIU Yu-guo. Interannual variabilityand sensitivity experiments [J]. J. Climate, 2001, 14: 756-779.of Indian Ocean dipole event and its connection with ENSO [凹].J. Zhanjiang Ocean Univ, 2006, 26(3): 50-55.Citation: FENG Jun-qiao and BAI Xue-zhi. Impacts of the tropical Pacific coupled process on the interannual variability in theIndian Ocean. J. Trop. Meteor, 2010, 16(3): 271-279.中国煤化工MHCNMH G