Xia ZHAO,Dongliang YUAN,Guang YANG,Hui ZHOU,and Jing WANG

1Key Laboratory ofOcean Circulation andWaves,and Institute ofOceanology,Chinese Academy ofSciences,Qingdao 266071,China

2Laboratory forOcean Dynamicsand Climate,Qingdao National Laboratory forMarine Science and Technology, Qingdao 266235,China

3CenterforOcean and Climate Research,First InstituteofOceanography,State Oceanic Administration,Qingdao 266061,China

Role of the Oceanic Channel in the Relationshipsbetween the Basin/DipoleM ode of SST Anomalies in the Tropical Indian Ocean and ENSO Transition

Xia ZHAO∗1,2,Dongliang YUAN1,2,Guang YANG3,Hui ZHOU1,2,and Jing WANG1,2

1Key Laboratory ofOcean Circulation andWaves,and Institute ofOceanology,Chinese Academy ofSciences,Qingdao 266071,China

2Laboratory forOcean Dynamicsand Climate,Qingdao National Laboratory forMarine Science and Technology, Qingdao 266235,China

3CenterforOcean and Climate Research,First InstituteofOceanography,State Oceanic Administration,Qingdao 266061,China

The relationships between the tropical Indian Ocean basin(IOB)/dipole(IOD)mode of SST anomalies(SSTAs)and ENSO phase transition during the follow ing year are exam ined and compared in observations for the period 1958–2008. Both partial correlation analysis and com posite analysis show that both the positive(negative)phase of the IOB and IOD (independentof each other)in the tropical Indian Ocean are possible contributors to the ElNi˜no(La Ni˜na)decay and phase transition to LaNi˜na(ElNi˜no)aboutoneyear later.However,the influenceon ENSO transition induced by the IOB isstronger than that by the IOD.The SSTAs in the equatorial central-eastern Pacific in the com ing year originate from subsurface temperature anomalies in the equatorial eastern Indian and western Pacific Ocean,induced by the IOB and IOD through eastw ard and upw ard propagation tomeet the surface.During this process,how ever the contribution of the oceanic channel processbetween the tropical Indian and Pacific oceans is totally di ff erent for the IOB and IOD.For the IOD,the influence of the Indonesian Throughflow transportanomaliescould propagate to theeastern Pacific to induce theENSO transition.For the IOB,the impactof theoceanic channelstaysand disappears in thewestern Pacificw ithoutpropagation to theeastern Pacific.

Indian Ocean SSTAs,dipolemode,basinmode,ENSO transition,oceanic channel

1.Introduction

ENSO is themost pronounced interannual variability in the tropics,contributing significantly to climate fluctuations in many regions of the globe.Several causalmechanisms of ENSO oscillation have been suggested,such as the delay oscillator(Suarez and Schopf,1988),the rechargeoscillator(Jin,1997a,1997b),and theWestern Pacific Oscillator (Weisberg and Wang,1997).Thesemechanismsemphasize processes in the Pacific Ocean basin,but they do notconsider the cross-basin linkage between the tropical Pacific and Indian oceans.The Indian Ocean’s variabilitym ay a ff ect that in the Pacific Ocean,althoughmostattention has been paid to the impactof the Pacific on the Indian Ocean(e.g.,Ding and Li,2012).There is evidence that SST variability in the Indian Ocean canmodulate ENSO variability either through atmosphericw ind variability(Yasunari,1985,1987;Yu etal., 2002;Behera and Yamagata,2003;Wu and Kirtman,2004; Kug and Kang,2006;M cPhaden,2008;Xie et al.,2009; Izumo etal.,2010;Luo et al.,2010;Du etal.,2013)or Indonesian Throughflow(ITF)variability(Wyrtki,1987;Wajsow icz and Schneider,2001;Yuan et al.,2011,2013).This provideshope for enhanced ENSO prediction skill(Izumo et al.,2010;Luo etal.,2010).

It iswell known that the SST anomalies(SSTAs)in the tropical Indian Ocean often co-occurw ith ENSO variation. Thus,studies tend to focuson the synchronous influence of the Indian Ocean on the Pacific(e.g.,Annamalaietal.,2005, 2010).On the other hand,several studies have provided evidence that the Indian Ocean acts as a negative feedback mechanism to ENSO(Kug and Kang,2006;Kug etal.,2006; Ohba and Ueda,2007;Izumo etal.,2010;Yuan etal.,2011, 2013;Zhou et al.,2015).The suggestion is that the Indian Ocean SSTAs during El Ni˜no(La Ni˜na)lead to a relatively faster ENSO term ination and phase transition to La Ni˜na(El Ni˜no).Such a lagged impactof the tropical Indian Ocean on the Pacific is the focusof thepresentpaper.

Themost dominantSST variations in the tropical Indian Ocean are the basin-w idewarming/coolingmode(hereafter,the IOB)(e.g.,Yang etal.,2007)and the dipolemode(hereafter,the IOD)(Sajietal.,1999;Websteretal.,1999),which can beobtained viaEOFanalysisof the tropical Indian Ocean SST variability,i.e.,EOF-1 and EOF-2,respectively(Fig.1). The IOB involvesa nearuniform warming(cooling)over the entire basin.It reaches itsmaximum after themature phase of ENSO.The IOD involvesaweak warming(cooling)over thewestern Indian Ocean and a strong cooling(warm ing)in theeasto ff Java and Sumatra(Sajietal.,1999;Webster etal., 1999).The IOD peaks in fall then quick ly recedes.Results show thatmuch of the IOB is caused by ENSO-induced surfaceheat flux anomalies(Klein etal.,1999).However,there are di ff erentviews regarding the IOD,w ith some suggesting it is independent of ENSO(e.g.,Sajiand Yamagata,2003; Yamagataetal.,2003)and others indicating itto beprimarily forced by ENSO(Xie et al.,2002;Annamalaiet al.,2005; Ohba and Ueda,2005).Regardlessof theoriginsof the IOB and IOD,they significantly a ff ectENSO evolution or transitionwhen the Indian Ocean SSTAsappear.

Fig.1.The(a)firstand(b)second EOFmodes,with(c)their PCs,of themonthly SSTAs in the tropical Indian Ocean,from the ERSST data for the period 1958–2008.(d)Standard deviation of the PC of the firstmode(PC1)and the PC of the second mode(PC2)as a function of calendarmonth.The red(black) curves in(c)and(d)denote PC1(PC2).

For the IOD,Izumo etal.(2010)and Yuan etal.(2011, 2013)investigated its influence on the follow ing year’s El Ni˜no,and they indicated thata positive(negative)phase of the IOD tends to co-occur w ith El Ni˜no(La Ni˜na)and favors La Ni˜na(El Ni˜no)about 1 year later.The form er study concentrated on the im pact of the atmospheric bridge between the Indian and Pacific oceans over the period 1981–2009.On the other hand,Yuan etal.(2011,2013)focused on the oceanic channel and assigned a key role to the ITF for the period 1990–2009.The positive(negative)phase of the IOD isaccompaniedbyanegative(positive)SSH anomalies(SSHAs)in the tropical eastern Indian Ocean,which forcesupwelling(downwelling)Kelvinwavesat theequator. Theupwellinganomaliespropagateeastward from the Indian Ocean into thewestern equatorial Pacific Ocean through the channels of the seas of Indonesia.They force an enhanced (weakened)ITF to transportmore(less)warm water from the upper-equatorialPacific Ocean to the Indian Ocean.This anomalous transport produces thermocline depth anomalies and cold(warm)subsurface temperature anomalies in the western equatorial Pacific Ocean.Then,these temperature anomaliespropagate to the eastern equatorial Pacific,inducing a LaNi˜na(ElNi˜no)in the follow ing year.Thus,theirobservationalandnumericalmodeling resultssuggested thatthe ITF playsan important role in propagating the Indian Ocean anom alies into the equatorial Pacific Ocean,and this ocean channelbetween the two basins is important for the evolution and predictability of ENSO.

For the IOB,Kug and Kang(2006)and Kug etal.(2006) revealed thatpositivephases tend to favor phase transition of ElNi˜no to LaNi˜na throughanomalouseasterly in thewestern Pacific,paying attention to the atmospheric teleconnection over the period 1958–2000.However,it is unclearwhether themodulationof theoceanic channelbetween the Indianand Pacific oceansplaysa role in the relationship between Indian Ocean SSTAsand ENSO transition.

As reviewed above,the relationships between the IOB/ IOD over the tropical Indian Ocean and ENSO evolution have previously been examined somewhat in isolation and through using dataw ith di ff erent time periods.We note that the impacts of the IOB and IOD have not been discussed and compared together through observationaldiagnostics.In particular,IOD events are always followed by IOB events, butpreviousstudiesdo notconsider thee ff ectbetween these twomodes.Moreover,numerical simulations are inconsistentw ith observational results.For instance,using a CGCM, Ohba and Ueda(2007)indicated that during boreal w inter the IOB enhances surface easterliesover the equatorialwestern Pacific during them ature-decay phase of El Ni˜no,which induces a transition to a La Ni˜na phase through upwelling Kelvinwaves.However,theirmodel experimentdid not reproduce the significant impact of the IOD on the Pacific. Thus,the firstpurpose of thepresentpaper is to compare therelationshipsbetween the IOB/IOD and ENSO phase transition using observationaldata,afterexcluding thee ff ectof the preceding IOD and follow ing IOB through partial correlation analysis and composite analysis.In addition,for the IOB, previousstudies focused only on the atmospheric bridgeprocess,neglecting the roleof theoceanic channelprocessin the relationships w ith ENSO transition.Therefore,the second purpose of the presentpaper is to discusswhether the relative role of the oceanic channel is sim ilar for these two leading SSTAsm odes.

The presentstudy investigates the impactof the external forcing of the tropical Indian Ocean on the Pacific.We attempt to reveal and compare the relationships between the twomajor SSTAs patterns in the tropical Indian Ocean and thedecay and phase transition of ENSO in the follow ing year using observationaldata from 1958 to 2008,aswell as discuss the dynamics associated w ith the oceanic channel,especially after separating the e ff ectof the IOD and IOB.Our com parison of the impacts of the IOB and IOD w ill provide a m ore comprehensive understanding of the role of the Indian Ocean in ENSO variability,and confirm previousmodel results.Note that the present paper isnot to discusswhether ornotENSO triggers the SSTAsin the tropical IndianOcean. Weonly study the impactsof the IndianOcean on thePacific, and consider how they influence ENSO decay and transition in the follow ing yearwhen the Indian Ocean SSTAs patterns appear.

2.Data

The latestversion of ERSST,i.e.,ERSST.v3b,isobtained from http://www.esrl.noaa.gov/psd/.The field ison a2°×2°global latitude–longitudegrid(Xueetal.,2003;Smith etal., 2008).The other SST data used in this study are from the HADISSTdataset(Rayneretal.,2003),compiledona1°×1°latitude–longitudegrid.

The subsurface temperature,salinity,sea level and horizontalvelocity dataareobtained from themonthlymeansof theSODA(version 2.1.6)productfor1958–2008(Carton and Giese,2008).This dataset is available at a 0.5°×0.5°horizontal resolution and has 40 vertical levelsw ith 10-m spacing near the surface.Xie et al.(2002)indicated that SODA agrees well w ith expendable bathythermograph data;plus, many previousstudieshaveused SODA data to calculate ITF transport(e.g.,England and Huang,2005;Potem ra,2005; Tillinger and Gordon,2009).Another reanalysis dataset used is the ECMWF’s ORAS4(Ocean Reanalysis System 4).These data(Mogensen et al.,2012;Balmaseda et al., 2013)areavailable ata horizontal resolution of 1°×1°,w ith 42 vertical levels and an approximate 10-m level thickness in the upper 200 m.Compared w ith conventional observational datasets,these two reanalysis datasetsprovidea longer time record(1958–2008)to investigate the relationship between the ITFand Indo-Pacific variability on the interannual to decadal timescales.

The analysis period for this study is51 years,from January 1958 to December 2008,because the SODA data are only available from 1958 to 2008.The annual cycle of each variable is removed by subtracting themonthlymean ateach grid point.And the anomaliesused in the present paper are detrended after removing theannual cycle.

3.Results

3.1.Relationshipsbetween the IOB/IOD and ENSO transition

The IOB and IOD over the tropical Indian Ocean are derived from EOF-1 and EOF-2 of the ERSST data,respectively(Figs.1a and 1b).EOF-1 and EOF-2 account for32% and 14%of the total variance,respectively.Note that the anom alies used in this study are detrended after removing the annual cycle.Through analyzing the principal com ponent(PC)of EOF(Fig.1c),it can be seen that the seasonal variability of the IOB(PC1)and IOD(PC2)hasan obvious phase-locking feature(Fig.1d).The IOB is strongest in borealw inter(February to March).Themaximum of the IOD occurs in fall(July to October)in the ERSST data,which show samaximum in September to October in theHADISST data(not shown).In order to investigate the influences of the IOB and IOD on the temporal evolution of ENSO,PC1 during February–March is chosen to define the borealw inter IOB index(WBI),and we use PC2 during September–October as the boreal fall IOD index(FDI).Note that differentmonthsare used to accord theseasonaldependenceof the IOB and IOD.Moreover,the follow ing resultsassociated w ith theWBIand FDIare not sensitive to themonthsused for the definition of these two indexes.PC1(PC2)during January–March(August–October)showssimilar results.

We denote the ENSO-developing year as year 0,the decaying year as year 1,and the follow ing year as year 2.The top panelsof Fig.2 give the lagged correlationsbetween the WBI/FD Iand the equatorial(2°S–2°N)SSTAs from September(year 0)to June(year 2)using the ERSST dataset.The longitude–time sections of the lagged correlation are used to describe the influenceof the IOB and IOD in the tropical IndianOcean on the temporalevolution of ENSO,separately.It canbeseen thattheWBIisclosely correlatedw ith the following equatorialeastern Pacific SSTAs.In borealw inter(year 0),the correlation is positive in the Indian Ocean,show ing the e ff ect of a positive IOB during the peak phase.In addition,positive correlation in the equatorial eastern Pacific indicates thata positive IOB tends to co-occurw ith ElNi˜no. In the follow ing spring to summer(year 1),the positive correlation in the Indian Ocean begins to weaken,whichmeans that the IOB decays.In them eantime,the positive correlation in the equatorial eastern Pacific diminishesand rapidly changes to an opposite sign,suggesting thata positive IOB of the Indian Ocean leads to a relatively faster termination of ElNi˜no and phase transition.The La Ni˜na then begins from July(year 1)and persists until the next year’s spring.The situation issimilar for the FDI,except foranegative correlation over theeastern IndianOcean andwestern Pacific Oceanduring the firstw inter(year 0),show ing the e ff ectof a peak positive IOD phase.Fornegativephasesof the IOB and IOD, the conditions are the same butw ith anomalies of opposite sign.The relationship between theWBI/FDIand ENSO evolution isalso obtained from theHADISST dataset.Theabove resultsshow thatboth the IOB and IOD in the tropical Indian Ocean possibly feed back negatively on ENSO evolution.

Because IOD eventsare always followed by IOB events, onemay imagine that the correlation between the tropical Indian SSTAs and the ENSO phase transition during the follow ingyear isinduced only by the IOD.Furthermore,weuse partial correlation(Cohen and Cohen,1983)to remove the partial influence of the preceding IOD and follow ing IOB, respectively.Thismethod has been used inmany previous studies(e.g.,Sajiand Yamagata,2003;Yu etal.,2005;Kug and Kang,2006;Izumo etal.,2014).Asshown in thebottom panelsof Fig.2,in the follow ingyear,thesignificantnegative lag correlations in the equatorial central-eastern Pacific still exist for both the IOB and IOD after excluding each other’s e ff ects.Therefore,the lagged teleconnection between the IOB/IOD and ENSO transition is not interdependent.Compared to the IOB,however,thenegative lagged correlation in the equatorial eastern Pacific during summer tow inter(year 1)isobviously weaker for the IOD.

Fig.2.Left-hand panels:lagged correlationsbetween theWBIand equatorial SSTAs(2°S–2°N)from September(year 0)to June(year 2).Rightpanels:lagged correlationsbetween the FDIand equatorial SSTAs.In the bottom panels the partial influenceof the IOD(left-hand panel)and IOB(right-hand panel)on theequatorial SSTAsis removed using the partial correlation.The color shading indicates positive(warm color)and negative correlations(cold color)above the 90%significance level.

Fig.3.Time seriesof theWBI(red),FDI(blue)and Ni˜no3.4SSTAsduring December–February(black).

One may doubt that the above lagged correlation m ay not imp ly a cause–e ff ect relationship.To clarify this problem,compositeanalysis is furtherconducted.Figure3 shows the time series of the WBI,FDIand Ni˜no3.4 SSTAs during December–February.The Ni˜no3.4 SSTAs are averaged over(5°S–5°N,170°–120°W).Itcan be seen thatsome positive(negative)IOB and IOD eventsconcurw ith ElNi˜no(La Ni˜na)events.In addition,in the follow ing year,the tropi-cal Pacific ismostly in its cold(warm)phase.Similarly,it is necessary to separate the e ff ect of the IOD and IOB due to some of IOD events coinciding w ith IOB events.Ultimately,eight independent IOB eventsover the period 1958–2008,each greater than one standard deviation and notcoincidentw ith IOB events(positive events:1958,1968,1969, 1987;negative events:1967,1970,1983,1985),are chosen to carry out the composite analysis.For the composite IOD, there are 14 independenteventsgreater than one standard deviation and not coincidentw ith IOB events(positive events: 1961,1967,1977,1978,1986,1994,2006;negative events: 1958,1960,1971,1974,1988,1996,1998).

Fig.4.Composites of tropical SSTAs for the di ff erencebetween independent positiveand negative IOB events(left-hand panels)and the di ff erencebetween independentpositive and negative IOD events(right-hand panels)in di ff erentseasonsover the period 1958–2008:(a)fall(year 0)[SON(0)];(b)w inter(year 0)[D(0)JF(1)];(c)spring(year 1)[MAM(1)];(d)summer (year 1)[JJA(1)];(e)fall(year 1)[SON(1)];(f)w inter(year 1)[D(1)JF(2)];and(g)spring(year 2)[MAM(2)].Color shading indicates thatthe positive(negative)SSTAs(°C)greater(less)than 0.2.

Figure 4 shows the composites of the tropical SSTAs for the di ff erence between independent positive and negative IOB events(left panels)and the di ff erence between independent positive and negative IOD events(right panels) from the boreal fall(year 0)season[SON(0),September–November(0)]through the spring(year2)season[MAM(2), March–M ay(2)].For the IOB(left panels of Fig.4),the SSTAsare positive in the Indian Ocean in thew inter(year 0) [D(0)JF(1),Decem ber(0)–February(1)],which is accompanied by warming in the central-eastern Pacific.However,thepositive SSTAs in the central-eastern Pacific become weak in thecomingspring(year1)[MAM(1),March–May(1)]and changes to negativein the follow ingsummer(year1)[JJA(1), June–August(1)].These negative SSTAs develop in the follow ing fall(year1)andw inter(year1)seasons.For the IOD (rightpanelsof Fig.4),the SSTAsarenegativein the tropical eastern Indian Ocean and positive in thewestern and central Indian Ocean in the late fall(year 0),which reflects the impactof the peak positive IOD phase.A t the same tim e,a significant teleconnection isdemonstrated by thepositive SSTAs in the eastern Pacific and the negative values in thewestern Pacific.The SSTAsover theequatorialcentral-eastern Pacific change to an opposite sign in the follow ing spring and persists in the follow ing summer tow inter seasons.The SSTA’s evolution showssomedi ff erencesbetween the IOB and IOD. The anomalies in the central-eastern Pacific during the second year induced by the IOB are obviously stronger than those induced by the IOD,and the former is located in the equatorial central-eastern Pacific while the latter is situated in theeastern Pacific.Thus,sim ilar to the resultof the lagged correlation analysis,composite analysis supportsour conclusion.

In previous studies,the relationships between the IOB/ IOD and ENSO evolution have been examined in isolation. In particular,the e ff ect between these two modes has not been considered.Moreover,themodel experimentof Ohba and Ueda(2007)did not reproduce this significant connection between the IOD and the tropicalcentral-eastern Pacific. Our result,using partial correlation and composite analysis and the sam e data,indicates thatboth the IOB and IOD could lead to ENSO phase transition.The di ff erence is that the influence induced by the IOB is stronger than thatby the IOD.

3.2.The relative role ofthe oceanic channel

Izumo et al.(2014)indicated that a positive IOD in fall and a positive IOB in w inter both promote a transition of ENSO events during the coming year,through the w ind anomalies over the western Pacific promoted by both IOD and IOB events(e.g.,Kug and Kang,2006;Ohba and Ueda,2007).In addition to the atmospheric bridge,some studies have indicated an influence of the tropical Indian on ENSO transition throughmodulation of the ITF(Sprintall et al.,2000;Wij ff els and Meyers,2004;Kandaga etal.,2009; Drushka et al.,2010;Yuan et al.,2011,2013).Yuan et al. (2011,2013)suggested that the IOD could induce ENSO phase transition through transport variations of the ITF,using observationaldataand numericalexperiments.However, the role of the ITF in the relationship between the IOB and SSTAs evolution in the equatorial Pacific has not been discussed.Therefore,it is necessary to compare the relative roles of the oceanic channel process in the relationship between the IOB/IOD and ENSO evolution in the follow ing year.

From theaboveanalysisin section 3.1,itcan beseen that the SSTAs in the equatorial central-eastern Pacific about 1 year later do not come from the horizontal propagation of the SST signal elsewhere.For the IOD,Yuan et al.(2013) indicated that lagged correlations in the subsurface temperature in a verticalsection of theequatorial Pacific Ocean suggesteastward propagation of the upwelling anomalies from the Indian Ocean into the equatorial Pacific Ocean through the seas of Indonesia.This result seems to suggest that the ocean channel connection between the two basins is important for the evolution and predictability of ENSO.Follow ing the work of Yuan et al.(2013),Fig.5 show s the partial correlationsbetween theWBI(FDI)and subsurface temperature anom alies in a vertical section of the equatorial Indian and Pacific oceans from the SODA data,in which the partial influenceof the IOD(IOB)on theequatorialSSTAs is removed using the partial correlation.Sim ilar to the resultof the IOD (right-hand panels),for the IOB(left-hand panels)the cold SSTAs in the equatorial central-eastern Pacific in the following year come from cold subsurface temperature anomalies in the equatorial eastern Indian and western Pacific oceans, which propagateeastward and upward along the thermocline to arriveat the surface.The result from the ECMWFORAS4 data(not shown)is consistentw ith that of the SODA data. Furthermore,composites of tropical subsurface temperature anomalies for the di ff erence between independent positive and negative IOB(IOD)eventsalso support the resultof the partialcorrelation analysis(Fig.6).

However,Kug and Kang(2006)and Ohba and Ueda (2007)indicated that the generation and propagation of subsurface temperature anomalies in the Pacific possibly comes from the atmospheric bridge process associated w ith the IOB/IOD.Anomalouseasterlies(westerlies)in theequatorial western Pacific during thematurephaseofElNi˜no(La Ni˜na), exciting the equatorial oceanic Kelvin wave,p lays a role in ENSO transition.Therefore,in addition to the oceanic channel,Figs.5 and 6 also include the e ff ectof the atmospheric bridgeprocess related to the IOB/IOD.It isnecessary to clarify the relative role of the oceanic channelprocess in the relationship between the IOB/IOD and ENSO transition.

Figure 7 shows the time series of themonthly transport anomalies of the ITF dealing w ith the low-pass-filter.The anomalies of the ITF volume transport are defined as the depth-integrated northward velocity in reference to the 729-m level of no motion through a zonal chokepointsection at 8.25°S,based on the SODA data.The tim e series are filtered by a Gaussian filterusing a cuto ff period at13months. Negativevalues indicate transports from the Pacific to the Indian Ocean.England and Huang(2004)suggested that the ITF from the SODA reanalysis data is a reasonably accurate reconstruction of theobserved ITF,and the ITF transport anomalies derived from the geostrophic flow show broadly similar behavior.The ECMWFORAS4 data show a similar temporalevolution.

Fig.5.Left-hand panels:partial correlations between theWBIand subsurface temperature anomalies in the Indian–Pacific equatorial vertical section from SODA data in di ff erent seasons,in which the partial influence of the IOD on the equatorial SSTAs is removed using the partial correlation.Right-hand panels:as in the left-hand panels,but for the FDI,in which the partial influenceof the IOB is removed.Color shading indicatespositiveand negative correlationsabove the90%significance level.

Fig.6.Composites of tropical subsurface temperature anomalies(°C)in the Indian–Pacific equatorial vertical section from the SODA data for the di ff erence between independent positiveand negative IOB events(left-hand panels)and the di ff erence between independent positiveand negative IOD events(right-hand panels).Color shading indicates thatthe positive(negative) subsurface temperatureanomalies greater(less)than 0.5.

Figure 8 show s the lagged correlations between the ITF transport anomalies during the peak phase of the IOB/IOD and subsurface temperature anom alies in the equatorial Indian and Pacific verticalsection during the follow ing seasons from the SODA data.Clearly,for the IOD(right-hand panels),the cold subsurface temperature anomalies in the eastern Indian Ocean andwestern Pacific Ocean,associatedw iththe ITF transports anomalies,could propagate eastward and upward along the thermocline to arrive at the surface in the central-eastern Pacific during the follow ingw inter(year 1). For the IOB(left-hand panels),however,thenegativeanomalies only stay in the tropicalwestern Pacific and the seasof Indonesiainw inter(year0),whichweakens in the follow ing summerand disappearsin the follow ing fall(year1).The ITF transportanomaliesduring thepeak phaseof the IOB arenot significantly correlated w ith subsurface temperature anomalies in the central-eastern Pacific in the follow ing summ er (year1)tow inter(year1).Thus,the ITF transportanomalies related to the IODm ightinduce the SSHAsand SSTAs in the central-eastern Pacific during the follow ing seasons through the eastward propagation of upwelling Kelvin waves.However,the ITF influence induced by the IOB only stays and disappears in the tropicalwestern Pacific w ithout propagation to theequatorialeastern Pacific to a ff ect the temperature in the cold tongue.The result from theECMWFORAS4 data isconsistentw ith thatof the SODA data(notshown).

Fig.7.Low-pass-filtered time series of themonthly ITF transportanomalies from the SODA data(top panel)and ECMWFORAS4 data(bottom panel).

To clearly show the contribution of the ITF transport anomalies induced by the IOB/IOD on ENSO phase transition,we repeat the com positesof tropical subsurface tem peratureanomalies in Fig.6 after removing the influenceof ITF transport anomalies during the peak phase of the IOD/IOB (Fig.9).For the IOB(left-hand panels),the influence of the ITFanomalieson the equatorial Pacific subsurface temperatureisvery small,because thePacific subsurface temperature anomaliesstillhavesimilarvaluesafter removing the ITFsignal.For the IOD(right-hand panels),thenegativesubsurface temperatureanomaliesareweakened in theequatorialPacific in the follow ing seasons,especially in the eastern Pacific,after the removal of the ITF signal induced by the IOD.This suggests thattheoceanic channelbetween the tropical Indian and Pacific oceans contributes to thedynamicsof the lagged teleconnection between the IOD and the ENSO evolution in the follow ing year,but the ITF transportanomalies is ineffective as a link between the IOB signals and ENSO phase transition.

4.Discussion

This paper only investigates the impactof external forcing from the tropical Indian Ocean.The internal dynam ics over the tropical Pacific isstillvital during ENSO evolution. The external forcing from the Indian Ocean may reinforce localprocesses in the Pacific,which could a ff ectENSO evolution in the com ing year.It iswell known thatENSO hasa typical period of roughly 3–7 years and a biennial(～2 yr) component.However,the quasi-biennial oscillation is not strong.A lthough lagged correlations between the Nino3.4 SSTAs index during them aturephaseof ENSO and the equatorial Pacific SSTAs reverse to being negativeduring the follow ing summer(year 1)tow inter(year 1),the negativevaluesarenotsignificant(notshown).However,because of the influence of the tropical Indian Ocean,the ENSO phase reversal seems to complete in one year after themature phase of ENSO(Fig.2).This suggests that the ENSO cycle shows enhanced variability for periodsof about2 years.Also,previousmodeling studies indicate that ENSO’s biennial com-ponentsignificantly increasesin the Indo-Pacific run,ascompared to simulationsonly including Pacific coupling(e.g.,Yu, 2005).Izumoetal.(2014)also indicated thatthe interactions between ENSO,the IOD and the IOB operate on a biennial timescale.

Fig.8.Left-hand panels:lagged correlations between the ITF transportanomalies during the peak phaseof the IOB(February to March)and subsurface temperature anomalies in the Indian–Pacific equatorial vertical section from the SODA data in differentseasons.Right-hand panels:as in the left-hand panels,but for the ITF transportanomalies during the peak phaseof the IOD(September to October).Color shading indicatespositiveand negative correlationsabove the90%significance level.

Fig.9.As in Fig.6,butw ith the influence of the ITF transportanomalies removed.

Many recentstudies(e.g.,Ohbaand Ueda,2009;Ohbaet al.,2010;Okumura et al.,2011;Dommenget et al.,2013; Ohba,2013)show that the transition process of ENSO is asymmetric.Ohba andWatanabe(2012)showed asymmetric im pacts of the IOB on ENSO transition between the warmingand coolingphaseof ENSO.Moreover,itis found thatthe asymmetric impacton ElNi˜no and La Ni˜naalso exist for the IOD whenwe separate the analysis for the positiveand negative phase of the IOD.The details and relative rolesof the atmospheric bridgeand oceanic channelw illbe discussed in another paper.

The relationship between the tropical Indian Ocean SSTAs and ENSO transition is influenced by multiple factors.Many studies have documented interdecadal variability in ENSO or the Indian Ocean SSTAs.Izum o etal.(2014)exp lored the interdecadal robustnessof the influenceof the IOD and Pacific rechargeon the follow ing year’sElNi˜no over the period 1872–2008,w ith a focus on the atmospheric bridge process.Yuan etal.(2013)concluded that the dynam icsare not related to the atmospheric bridge over the period 1990–2009,which supports the ITF connection.Thus,the relative rolesof the atmospheric bridge and oceanic channelmay be di ff erent in this lagged remote relationship over di ff erentperiods.In addition to the FDI,the 1-year lagged correlations of theWBIw ith the follow ing year’s ENSO also possess significant interdecadal variability(not shown).Therefore,the lagged relationship between the tropical Indian Ocean SSTAs and ENSO transition on interdecadal timescales,and the associated dynamicalprocesses,stillneed further investigation. In addition,tropical Atlanticwarm ing during ElNi˜no isalso clear,and some of the di ff erences between the two Indian Ocean SSTAsmodesmightoriginate from the Atlantic SST variability.Therefore,the role of Atlantic SST variability in Indian Ocean SST variability needs further investigation.

Using a novelmethod based on information flow,Liang (2014)discussed the relationship between the IOD and ENSO.The study indicated that the IOD and ENSO aremutually causal,but the causality isasymmetric:the IOD functions tomake ENSO more uncertain,while ENSO tends to stabilize the IOD.Thisnew method hasnotbeen applied to investigate the influence of the IOB on ENSO.Itwould be of interest to discuss the relationship between the IOB and ENSO,including ENSO’sphase transition.

5.Conclusion

The tropical Indian and Pacific oceans can a ff ectone another.Rather than discussing whether or not ENSO triggers the SSTAs in the tropical Indian Ocean,thepresentpaper focuseson thee ff ectsof the leadingmodesof the SSTAs in the tropical IndianOceanon thevariabilityof ENSO about1 year later,which isnotasynchronousinfluence.The SSTAs in the tropical Indian Ocean have twomajormodes:the IOB and IOD.Theiroccurrence influences thevariation in the tropical Pacific Ocean.

However,the influence on ENSO transition associated w ith the IOB and IOD has previously only been exam ined in isolation and using data w ith di ff erent time periods.In particular,whilst IOD events are always followed by IOB events,previous studies have not considered the e ff ect between these two modes.M oreover,the model experiment of Ohba and Ueda(2007)did not reproduce this significant connection between the IOD and the tropical central-eastern Pacific.Thepresentstudy compares the relationship between the IOB/IOD and ENSO transition in the follow ing year using observationaldata for 1958–2008,through partial correlation analysis and composite analysis to separate the e ff ect between the preceding IOD and follow ing IOB.Our results indicate that both the positive(negative)phase of the IOB and IOD(independent of each other)in the tropical Indian Ocean are possible contributors to the El Ni˜no(La Ni˜na)decay and phase transition to La Ni˜na(El Ni˜no)about 1 year later.The di ff erence is that the influence induced by the IOB isstronger than thatby the IOD.

The cold(warm)SSTAs in the equatorial central-eastern Pacific in thecomingyearoriginate from cold(warm)subsurface temperature anomalies in the equatorial eastern Indian and western Pacific Ocean.This signal propagateseastward and upward along the thermocline to arrive at the surface. Somestudieshaveindicated thatthegeneration and propagation of subsurface tem perature anomaliesm ight com e from themodulation of the oceanic channel,in addition to the atmospheric bridge.However,for the IOB,previous studies only focused on theatmospheric bridgeprocess,w ithoutdiscussion of the roleof theoceanic channelprocessin the relationshipsw ith ENSO transition.Our resultsshow thatthe relativecontributionsof theoceanic channelaredi ff erent for the IOB and IOD.For the IOD,the Kelvinwaves induced by the IOD could penetrate from the Indian Ocean into thewestern Pacific through the ITF,which furtherpropagateeastwardand upward to induce ENSO decay and transition in the following year.However,for the IOB,the associated Kelvin waves propagate to the equatorialwestern Pacific w ithoutpropagation to theeastern Pacific,which disappear in thewestern Pacific during the coming summer–fall.The indication is that the ITF transportanomaliesare ine ff ectiveasa link between the IOB signalsand ENSO phase transition.However,why the Kelvinwavesinduced by the IOB areunable to propagate further to the eastern Pacific remainsan open question.

Furthermore,it can be seen from Fig.10 that the atmospheric bridge process plays a dominant role in the lagged teleconnection between the IOB and ENSO phase transition. For the IOB,significant easterly anomalies over the western Pacific occur during D(0)JF(1),which sustain to the follow ing summer[JJA(1)].Through thebaroclinic atmosphere Kelvinwave,the IOB in the Indian Ocean can a ff ect the developmentof the anomalouseasterlies in thewestern Pacific (e.g.,Xie etal.,2009;Du etal.,2013),which then influenceEl Ni˜no’s transition.However,the easterly anomalies over the western Pacific associated w ith the IOD are very weak throughout.Thus,the e ff ectof the IOD on ENSO transition through the atmospheric bridge isnotdom inant.

Fig.10.Composites of tropicalwind anomalies(m s-1)at850 hPa in di ff erentseasons for the di ff erence between independent positiveand negative IOB events(left-hand panels)and the di ff erence between independent positive and negative IOD events (right-hand panels).

Acknow ledgements.Thiswork was jointly supported by the Strategic Priority Research Program of theChineseAcademy of Sciences(GrantNo.XDA11010102),the NSFC(GrantNos.41375094 and 41406028),the“973”project(Grant No.2012CB956000), and the NSFC–Shandong Joint Fund for Marine Science Research Centers(Grant No.U1406401).The latest version of ERSST, ERSST.v3b,is available at http://www.esrl.noaa.gov/psd/,and the HADISST data at http://www.meto ffi ce.gov.uk/hadobs/index.htm l, w ithout charge.The SODA data are freely available from http:// iridl.ldeo.columbia.edu/SOURCES/.CARTON-GIESE/.SODA,and ECMWFORAS4 from http://www.ecmw f.int/en/research/climatereanalysis/ocean-reanalysis.

Annamalai,H.,S.Kida,and J.Hafner,2010:Potential impactof the tropical Indian Ocean-Indonesian Seas on El Ni˜no char-acteristics.J.Climate,23,3933–3952.

Annamalai,H.,S.P.Xie,J.P.M cCreary,and R.M urtugudde, 2005:Impactof Indian Ocean sea surface tem perature on developing ElNi˜no.J.Climate,18,302–319.

Balmaseda,M.A.,K.Mogensen,and A.Weaver,2013:Evaluation of the ECMWFocean reanalysissystem ORAS4.Quart. J.Roy.Meteor.Soc,139,1131–1161.

Behera,S.K.,and T.Yamagata,2003:Influence of the Indian Ocean dipole on the southern oscillation.J.Meteor.Soc. Japan,81,169–177.

Carton,J.A.,and B.S.Giese,2008:A reanalysisofocean climate using Simple Ocean Data Assimilation(SODA).Mon.Wea. Rev.,136,2999–3017.

Cohen,J.,and P.Cohen,1983:Applied Multiple Regression/Correlation Analysis for the Behavioral Sciences.Law rence Erlbaum Associates,545 pp.

Ding,R.Q.,and J.P.Li,2012:Influencesof ENSO teleconnection on the persistence of sea surface temperature in the tropical IndianOcean.J.Climate,25,8177–8195,doi:10.1175/JCLID-11-00739.1.

Dommenget,D.,T.Bayr,and C.Frauen,2013:Analysis of the non-linearity in the pattern and time evolution of El Ni˜no Southern Oscillation.Climate Dyn.,40,2825–2847.

Drushka,K.,J.Sprintall,S.T.Gille,and I.Brodjonegoro,2010: Verticalstructureof Kelvinwaves in the Indonesian Throughflow exitpassages.J.Phys.Oceanogr.,40,1965–1987.

Du,Y.,S.-P.X ie,Y.-L.Yang,X.-T.Zheng,L.Liu,and G.Huang, 2013:Indian Ocean variability in the CMIP5multimodelensemble:The basin mode.J.Climate,26,7240–7266,doi: 10.1175/JCLI-D-12-00678.1.

England,M.H.,and F.Huang,2005:On the interannualvariability of the Indonesian Throughflow and its linkagew ith ENSO.J. Climate,18,1435–1444.

Izumo,T.,and Coauthors,2010:Influence of the state of the Indian Ocean Dipole on following year’s ElNi˜no.Nature Geoscience,3,168–172

Izumo,T.,M.Lengaigne,J.Vialard,J.-J.Luo,T.Yamagata,G. M adec,2014:Influence of the Indian Ocean Dipole and Pacific recharge on the follow ing year’s El Ni˜no:Interdecadal robustness.ClimateDyn.,42,291–310,doi:10.1007/s00382-012-1628-1.

Jin,F.-F.,1997a:An equatorial ocean recharge paradigm for ENSO.Part I:Conceptualmodel.J.Atmos.Sci.,54,811–829.

Jin,F.-F.,1997b:An equatorial ocean recharge paradigm for ENSO.Part II:A stripped-down coupled model.J.Atmos. Sci.,54,830–847.

Kandaga,P.,A.L.Gordon,J.Sprintall,and R.D.Susanto,2009: Intraseasonal variability in the Makassar Strait thermocline. J.Mar.Res.,67,757–777.

K lein,S.A.,B.J.Soden,and N.-C.Lau,1999:Remoteseasurface temperature variations during ENSO:Evidence fora tropical atmospheric bridge.J.Climate,12,917–932.

Kug,J.-S.,and I.-S.Kang,2006:Interactive feedback between ENSO and the Indian Ocean.J.Climate,19,1784–1801.

Kug,J.-S.,T.Li,S.-I.An,I.-S.Kang,J.-J.Luo,S.Masson,and T. Yamagata,2006:Role of the ENSO-Indian Ocean coupling on ENSO variability in a coupled GCM.Geophys.Res.Lett., 33,L09710,doi:10.1029/2005GL024916.

Liang,X.S.,2014:Unraveling the cause-e ff ect relation between time series.PhysicalReview E,90,052150.

Luo,J.-J.,R.C.Zhang,S.K.Behera,Y.Masumoto,F.-F.Jin,R. Lukas,and T.Yamagata,2010:Interaction between El Ni˜no and extreme Indian Ocean Dipole.J.Climate,23,726–742.

M cPhaden,M.J.,2008:Evolution of the 2006-2007 ElNi˜no:The role of intraseasonal to interannual time scale dynam ics.Advances in Geosciences,14,219–230.

Mogensen,K.,M.A lonso Balmaseda,and A.Weaver,2012:The NEMOVAR ocean data assimilation system as implemented in the ECMWF ocean analysis for System 4.ECMWFTechnicalMemorandum 668,59 pp.

Ohba,M.,2013:Im portant factors for long-term change in ENSO transitivity.International Journal of Climatology,33,1495–1509.

Ohba,M.,and H.Ueda,2005:Basin-w idewarming in the equatorial Indian Ocean associatedwith ElNi˜no.SOLA,1,89–92, doi:10.2151/sola.2005-024.

Ohba,M.,and H.Ueda,2007:An im pact of SST anomalies in the Indian Ocean in acceleration of the El Ni˜no to La Ni˜na transition.J.Meteor.Soc.Japan,85,335–348.

Ohba,M.,and H.Ueda,2009:Role of nonlinear atmospheric response to SSTon theasymmetric transition processof ENSO. J.Climate,22,177–192.

Ohba,M.,and M.Watanabe,2012:Role of the Indo-Pacific interbasin coupling in predicting asymmetric ENSO transition and duration.J.Climate,25,3321–3335.

Ohba,M.,D.Nohara,and H.Ueda,2010:Simulation of asymmetric ENSO transition in WCRP CM IP3 multimodel experiments.J.Climate,23,6051–6067,doi:10.1175/2010 JCLI3608.1.

Okumura,Y.M.,M.Ohba,C.Deser,and H.Ueda,2011:A proposedmechanism for theasymmetric duration of ElNi˜noand La Ni˜na.J.Climate,24,3822–3829.

Potemra,J.T.,2005:Indonesian Throughflow transport variability estimated from satellite altimetry.Oceanography,18,98–107.

Rayner,N.A.,D.E.Parker,E.B.Horton,C.K.Folland,L.V. Alexander,D.P.Rowell,E.C.Kent,and A.Kaplan,2003: Globalanalysesof sea surface temperature,sea ice,and night marine air temperature since the late nineteenth century.J. Geophys.Res.,108,4407,doi:10.1029/2002JD 002670.

Saji,N.H.,and T.Yamagata,2003:Structure of SST and surface wind variability during Indian Ocean dipole mode events: COADSobservations.J.Climate,16,2735–2751.

Saji,N.H.,B.N.Goswami,P.N.Vinayachandran,and T.Yamagata,1999:A dipolemode in the tropical Indian Ocean. Nature,401,360–363.

Smith,T.M.,R.W.Reynolds,T.C.Peterson,and J.Lawrimore,2008:Improvements to NOAA’s historicalmerged land-ocean surface temperatureanalysis(1880–2006).J.Climate,21,2283–2296.

Sprintall,J.,A.L.Gordon,R.Murtugudde,and R.D.Susanto, 2000:A sem iannual Indian Ocean forced Kelvin wave observed in the Indonesian seas in May 1997.J.Geophys.Res., 105,17 217–17 230.

Suarez,M.J.,and P.S.Schopf,1988:A delayed action oscillator for ENSO.J.Atmos.Sci.,45,3283–3287.

Tillinger,D.,and A.L.Gordon,2009:Fifty years of the Indonesian Throughflow.J.Climate,22,6342–6355.

Wajsow icz,R.C.,and E.K.Schneider,2001:The Indonesian throughflow’s e ff ect on global climate determ ined from the COLA coupled climate system.J.Climate,14,3029–3042.

Webster,P.J.,A.M.Moore,J.P.Loschnigg,and R.R..Leben, 1999:Coupled ocean-atmosphere dynam ics in the Indian Ocean during 1997–98.Nature,401,356–360.

Weisberg,R.H.,and C.Z.Wang,1997:A western Pacific oscillator paradigm for the El Ni˜no-Southern Oscillation.Geophys. Res.Lett.,24,779–782.

Wij ff els,S.E.,and G.Meyers,2004:An intersection of oceanic waveguides:Variability in the Indonesian Throughflow region.J.Phys.Oceanogr.,34,1232–1253.

Wu,R.G.,and B.P.Kirtman,2004:Understanding the impactsof the Indian Ocean on ENSO variability in a coupled GCM.J. Climate,17,4019–4031.

Wyrtki,K.,1987:Indonesian through flow and theassociated pressuregradient.J.Geophys.Res.,92,12 941–12 946.

Xie,S.-P.,H.Annamalai,F.A.Schott,and J.P.M cCreary Jr, 2002:Structure andmechanisms of South Indian Ocean climate variability.J.Climate,15,864–878.

Xie,S.-P.,K.M.Hu,J.Hafner,H.Tokinaga,Y.Du,G.Huang, and T.Sampe,2009:Indian Ocean capacitor e ff ecton Indowestern Pacific climate during thesummer following ElNi˜no. J.Climate,22,730–747.

Xue,Y.,T.M.Smith,and R.W.Reynolds,2003:Interdecadal changesof30-yr SST normalsduring 1871–2000.J.Climate, 16,1601–1612.

Yamagata,T.,S.K.Behera,S.A.Rao,Z.Y.Guan,K.Ashok,and H.N.Saji,2003:Comments on“Dipoles,temperature gradients,and tropical climate anomalies”.Bull.Amer.Meteor. Soc.,84,1418–1422.

Yang,J.L.,Q.Y.Liu,S.-P.Xie,Z.Y.Liu,and L.X.Wu, 2007:Impact of the Indian Ocean SST basin mode on the Asian summermonsoon.Geophys.Res.Lett.,34,L02708, doi:10.1029/2006GL028571.

Yasunari,T.,1985:Zonally propagatingmodesof theglobaleast– west circulation associated with the Southern Oscillation.J. Meteor.Soc.Japan,63,1013–1029.

Yasunari,T.,1987:G lobal structure of the El Ni˜no/Southern oscillation.Part II.Time evolution.J.Meteor.Soc.Japan,65, 81–102.

Yu,J.-Y.,C.R.Mechoso,J.C.McWilliams,and A.Arakawa, 2002:Impactsof the Indian Ocean on the ENSO cycle.Geophys.Res.Lett.,29,46-1–6-4,doi:10.1029/2001GL014098.

Yu,J.-Y.,2005:Enhancement of ENSO’s persistence barrier by biennial variability in a coupled atmosphere-ocean general circulationmodel.Geophys.Res.Lett.,32,L13707,doi: 10.1029/2005GL023406.

Yu,W.D.,B.Q.Xiang,L.Liu,and N.Liu,2005:Understanding the origins of interannual thermocline variations in the tropical Indian Ocean.Geophys.Res.Lett.,32,L24706,doi: 10.1029/2005GL024327.

Yuan,D.L.,and Coauthors,2011:Forcing of the Indian Ocean dipole on the interannual variations of the tropical Pacific Ocean:Roles of the Indonesian throughflow.J.Climate,24, 3593–3608.

Yuan,D.L.,H.Zhou,and X.Zhao,2013:Interannual climate variability over the tropical Pacific Ocean induced by the Indian Ocean dipole through the Indonesian Throughflow.J. Climate,26,2845–2861,doi:10.1175/JCLI-D-12-00117.1.

Zhou,Q.,W.S.Duan,M.Mu,and R.Feng,2015:Influence of positive and negative Indian Ocean dipoles on ENSO via the Indonesian Throughflow:Results from sensitivity experiments.Adv.Atmos.Sci.,32,783–793,doi:10.1007/s00376-014-4141-0.

:Zhao,X.,D.L.Yuan,G.Yang,H.Zhou,and J.Wang,2016:Role of the oceanic channel in the relationships betw een the basin/dipole mode of SST anomalies in the tropical Indian Ocean and ENSO transition.Adv.Atmos.Sci.,33(12),1386–1400,

10.1007/s00376-016-6048-4.

(Received 26 February 2016;revised 24 June2016;accepted 18 July 2016)

∗Corresponding author:Xia ZHAO

Email:zhaoxia@qdio.ac.cn

©Institute ofAtmospheric Physics/Chinese Academy of Sciences,and Science Press and Springer-Verlag Berlin Heidelberg 2016