Dynamics of late autumn rainfall reduction over southeastern Australia

Dynamics of late autumn rainfall reduction over southeastern AustraliaWenju Cai,1,2and Tim Cowan1,2Received21February2008;revised4April2008;accepted10April2008;published13May2008.[1]During1950–2006,austral autumn rainfall over the south-eastern Australian state of Victoria decreased by about 40%from its long-term seasonal average.The dynamics of this decline are not understood.We show that the reduction predominantly occurs in May,with variations of rainfall linked to two sources of sea surface temperature(SST) variability.A higher SST in the Indonesian Throughflow region,which occurs as the Indo-Pacific system enters into La Nin˜a-like conditions,is associated with a rainfall increase over northern Victoria.An SST variability pattern with a positive anomaly in the central southern Indian Ocean(IO)and a negative anomaly in the subtropical IO are linked to atmospheric wave-trains,with a low pressure center over Victoria,is conducive to a rainfall increase.The weakening of the wave-trains since1950,and an increase (reduction)in El Nin˜o(La Nin˜a)events,contribute to the rainfall decline.Possible influences from climate change forcing are discussed.Citation:Cai,W.,and T.Cowan(2008), Dynamics of late autumn rainfall reduction over southeastern Australia,Geophys.Res.Lett.,35,L09708,doi:10.1029/ 2008GL033727.1.Introduction[2]There has been a decreasing trend in rainfall over much of southern and eastern Australia during the past 50years.These trends exhibit strong seasonality and regional differentiations: e.g.,in austral winter over the south-west Western Australia(SWWA),and summer over southern Queensland[Cai et al.,2007].On average,the south-east Australian state of Victoria(Figure1e)receives more rainfall in winter(June–August,200mm)and spring (September–November,178mm)than in autumn(March–May,158mm),however the reduction trend is greatest in autumn(Figure1).The autumn rainfall decline is dominated by May(Figures1c and1g),which makes up some50%of the total seasonal reduction.It is manifested through a series of consecutive low rainfall years,predominantly from1990 onwards.The rainfall reduction,experienced through much of south-east Australia(SEA,south of33°S,east of135°E), significantly reduces inflows to the Murray and Darling Rivers,Australia’s longest river system[Cai and Cowan, 2008].[3]The Victoria autumn rainfall reduction is similar to what has occurred over SWWA,where since the late1960s,winter rainfall has decreased15%[Cai et al.,2007].There, the reduction is caused by a southward shift in rainfall and pressure systems,induced by an upward trend of the southern annular mode(SAM)forced by multi-decadal variability,and anthropogenic forcing.We will show that as the SEA rainfall decline occurs in autumn,it is likely that it is driven by different mechanisms to the SWWA rainfall reduction.2.Data and Method[4]The observed rainfall data,subjected to extensive quality control,are from the Australian Bureau of Meteo-rology.Reanalysis data used include an updated version of the Global Sea Ice and Sea Surface Temperature data set [Rayner et al.,1996],and mean sea level pressure(MSLP), 500mb geopotential heights(Z500),and surface winds from the National Centers for Environmental Prediction[Kalnay et al.,1996].Monthly and seasonal anomalies are con-structed over the period1950–2006.We examine the tele-connection patterns between Victoria May rainfall and other circulation fields using detrended data.The evolution of these patterns is then invoked to explain the rainfall trend.3.Teleconnection Between Rainfall and SST Variability[5]In an attempt to identify a responsible dynamical forcing for SEA rainfall variability,we conduct empirical orthogonal function(EOF)analysis using a correlation matrix on detrended SST anomalies in the Indo-Pacific domain,but are unable to find a dominant mode that is coherent with SEA rainfall.We subsequently conduct correlation analysis between Victoria May rainfall and grid-point circulation fields(Figures2a and2d).[6]The Victoria May rainfall has little coherence with rainfall outside of SEA(Figure2d).It is linked with SST anomalies in three centers:southern midlatitude IO,sub-tropical IO,and the Banda Sea region of the Indonesian Throughflow(ITF)Passage(denoted as A,B and C respectively in Figure2a;see figure for definition).A higher SST in the ITF region is conducive to May rainfall across most of eastern Australia,particularly in the south-east (Figures2b and2e).The influence is conducted through the familiar northwest cloud-bands similar to those in the winter and spring seasons[Whetton,1988;Nicholls,1989]. In a cyclonic circulation pattern,high-moisture flows from the tropics veer westwards rotating northeastward over the Tasman Sea,east of SEA(Figure2b).Composite analysis using the SST index in the ITF region over positive events greater than one-standard deviation produces a similar rainfall pattern.GEOPHYSICAL RESEARCH LETTERS,VOL.35,L09708,doi:10.1029/2008GL033727,2008 1CSIRO Marine and Atmospheric Research,Aspendale,Victoria,Australia.2Wealth from Oceans National Research Flagship,CSIRO,North Ryde,New South Wales,Australia.Copyright2008by the American Geophysical Union.0094-8276/08/2008GL033727[7]There is no suggestion of any linkage with a zonally symmetric SAM-like pattern,as can be seen through corre-lation analysis of May rainfall with MSLP,which is also reflected in the wind pattern(vectors,Figure2a).This is further confirmed by correlating May grid-point rainfall with a well established SAM index[Marshall,2003],which produces no coherence of SAM with SEA rainfall(not shown).In association,analyses using residual anomalies after removing variances associated with the SAM yield almost identical results.[8]Given that SST variability in the ITF region shows no coherence with variations over the subtropical and midlat-itude IO centers(Figure2b),we will hereafter treat it as a separate source.We treat the two IO sources as an entity using a‘‘meridional IO SST index’’defined as the differ-ence in SST anomalies over these two centers(A–B).This index,at its positive phase,represents a heating to the south, producing downstream wave-trains with a cyclonic center over Victoria,as revealed by correlating the index with circulation fields(see section5).[9]Next we discuss what controls variability of SSTs in the ITF region and the southern IO using detrended data. We then examine whether the dynamics are consistent with the rainfall decline over SEA using raw data.4.ITF SST Impacts on May SEA Rainfall [10]SST variability in the ITF region is intimately involved in the cycle of the El Nin˜o Southern Oscillation (ENSO).Lag-correlation of the time series of May SST anomalies in the ITF region with grid-point SST and MSLP anomalies shows that two months prior(i.e.,in March)the Indo-Pacific system is either in a neutral or transition phase from an El Nin˜o event to a La Nin˜a.There is still a basin-scale warming in the IO,associated with negative MSLP anomalies(Figure3a).Thereafter,SSTs between the eastern IO and the western Pacific increase rapidly,as the region is a ‘‘choke point’’of the equatorial Indo-Pacific waveguides [Wijffels and Meyers,2004]associated with a decaying El Nin˜o or a developing La Nin˜a.By May,significantly low MSLP anomalies have developed over the ITF region (Figure3b),and a structure of wave-trains enacting from the ITF warming region is seen.The low MSLPs extend to SEA with the overall wave-train structure resembling the Pacific-South American(PSA)pattern(Figure3b).The PSA wave-trains grow into a traditional Southern Oscillation-like pattern by July(Figure3c),indicating development of a La Nin˜a-like phase.Thus through the SST teleconnection over the ITF center(Figures2b and2e),an increase in Victoria May rainfall is associated with a transition into La Nin˜a.[11]This teleconnection can be confirmed through direct correlation between a well-known ENSO index,the South-ern Oscillation Index(SOI)with grid-point rainfall in May (Figure3d).The SOI-SEA rainfall relationship in May is reminiscent of the correlation pattern between ITF SST and SEA rainfall(Figure2e),suggesting that the May SOI is dominated by the northern Australia-western Pacific center, which is linked with the SST variability over theITF Figure1.Time series of Victoria monthly rainfall anomalies for(a)March,(b)April,(c)May,and(d)the austral autumn season.An11year running average is also shown(thick black line).(e)–(h)The corresponding pattern of total trend since 1950(in mm/57yrs).Figure2.Map of correlation of Victoria May rainfall with(a)May SST(contour)and winds(vectors)and(d)May grid-point Australian rainfall.(b)and(e)Correlation of SST anomalies in the ITF region(denoted by C:125°E–130°E,2°S–7°S)with grid-point SST and with grid-point Australian rainfall.(c)and(f)Correlation of an SST index with grid-point SST and rainfall;the index is referred to as the meridional IO SST index,defined as SST anomalies averaged over a mid-latitude box(denoted by A:90°E–95°E,37°S–42°S)minus that averaged over a subtropical latitude box(denoted by B: 90°E–95°E,19°S–24°S).Areas within dashed lines in Figures1a–1c and areas shaded in Figures1d–1f indicate correlations that are significant at the95%confidence level.All analyses are conducted using detrended data.Figure3.Correlation of detrended May ITF SST with detrended MSLP in(a)March,(b)May,and(c)July.(d)Correlation of May SOI and Australian rainfall.(e)Correlation of May SOI and May grid-point MSLP(colour legend as same as Figure3a).(f)Time series of raw May SOI(with an11year running average shown as thick black line).Areas within dashed lines in Figures3a–3c and3e and colours greater than0.26in Figure3d indicate correlations that are significant at the95%confidence level.region.To confirm this,we correlate detrended May SOI with detrended grid-point MSLP.The correlation pattern (Figure3e)is indeed similar to the ITF SST-MSLP rela-tionship(Figure3b).Thus,before the pattern associated with SOI develops into a traditional east-west hemispheric pattern by July,wave-trains enacting from the western Pacific occur,influencing SEA rainfall.[12]Since the1976climate-shift,there have been more El Nin˜o events[Power and Smith,2007],which are more protracted in terms of duration,than La Nin˜a episodes.The protracted El Nin˜o events tend to show up as consecutive El Nin˜o Modoki events[Ashok et al.,2007].This disparity in event numbers is a contributing cause for the large number of consecutive years of anomalously low late autumn rainfall,as the Indo-Pacific system spends far less time in a La-Nin˜a development phase.This feature is manifested in the raw May SOI time series which has been trending down since the1980s,showing stronger negative values (Figure3f).Furthermore,May SST trends since1950shows that relative to warming in the ITF region,a faster rate occurs to the east,north of Papau New Guinea.These features are consistent with a global warming-induced weakening trend in the Walker Circulation during the late 20th and early21st century[Tanaka et al.,2004;Vecchi et al.,2006;Ashok et al.,2007].[13]An identical analysis for April highlights a similar evolution of anomalies but weaker in strength.This sug-gests that a mechanism operates in April similar to that in May,in part contributing to the April rainfall decline.5.Impacts from the Meridional IO SST Variability[14]We recall that the SST pattern associated with the meridional IO SST index is identified through a correlation with Victoria May rainfall(Figure2a).The setting of the wave-trains must generate circulation anomalies in SEA that are conducive to a rainfall increase.Midlatitude heating, like orographic forcing,generates wave-trains downstream [Frederiksen,1982;Hoskins and Karoly,1981].A correla-tion between the meridional IO SST index with MSLP anomalies(Figure4a)using detrended data clearly displays wave-trains with a wave-number4structure,curving pole-ward west of the Drake Passage.These wave-trains are barotropic in nature,as seen in the Z500anomalies(Figure 4b).One important feature is that for a positive phase of the meridional IO SST index,there is a low pressure center over much of southern Australia that provides favourable con-ditions for rainfall over Victoria.Factors that determine this spatial structure may include the magnitude and structure of the mean flow,and the dissipation of the flow and energy, and this needs further investigation.[15]Lag-correlation between the meridional IO SST index and circulation anomalies suggest that the pattern does not propagate eastward,but is quasi-stationary(not shown).It starts in April,peaks in May,decays in June,and disappears by July.The fact that it commences in April suggests that the mechanism for an April rainfall decline is linked to that in May.A detailed study is required to examine the associated dynamics,involving forcing an atmospheric general circulation model with a similar SST anomaly pattern,but this is beyond the scope of the present study.However,an imprint of a positive feedback mecha-nism is seen:positive MSLP centers are situated to the east of warm SST anomalies,generating poleward wind anomalies,which in turn reinforce the warm SST anomalies [Baines and Cai,2000].[16]Since1950,warming has proceeded throughout much of the southern IO,at vastly different rates for different locations.How has the warming affected the meridional IO SST index?We project raw SST anomalies onto the anomaly pattern associated with the index for each May in the domain of10°S–50°S,80°E-120°E to obtain a time series of the pattern regression coefficient(black-line,Figure4d).The time series displays a downward trend meaning that the IO warming pattern favours a Victoria May rainfall reduction. Further evidence comes from an application of EOF analysis to raw Z500anomalies in a domain of20°S–60°S.This second EOF mode(Figure4c)accounts for13%of the total variance,sharing a similar pattern to that of Figure4b.The associated time series(orange-line,Figure4d)trends down-wards,indicating a rising MSLP over Victoria that is not conducive to rainfall there.The two time series in Figure4d share significant coherence with a correlation of0.43,sug-gesting that a coupling between them exists.It is notclear Figure4.Correlation of the detrended meridional IO SSTindex with(a)detrended May MSLP and(b)Z500.(c)EOF2pattern of raw May Z500anomalies.(d)Time series of thepattern regression coefficient obtained by regressing rawSST data onto a SST anomaly pattern associated with anincrease in Victoria May rainfall(black-line)and a timeseries of Z500EOF2(orange-line).Areas confined bydashed lines in Figures4a–4b indicate correlations that aresignificant at the95%confidence level.whether the declining EOF2trend is forced by an IO warm-ing,or vice-versa.Further study is needed.6.Conclusions[17]Victoria May rainfall has linkages to sources of SST variability in the Indo-Pacific system.A higher SST in the ITF region,which occurs as the system enters into a La Nin˜a phase,is associated with a rainfall increase over the northern and central SEA,through an increase in northwest cloud-bands.A positive SST anomaly in the southern midlatitude IO in association with a negative anomaly in the subtropical latitudes,represented by a meridional IO SST index,is conducive to a rainfall increase over southern Victoria.The associated wave-trains exhibit a low pressure center over SEA.The decline of the meridional IO SST index since 1950,and the increase in prostrated El Nin˜o events accom-panied by a reduction in La Nin˜a episodes contribute to the rainfall decline.Recent studies have suggested that the weakening of the Walker Circulation and the changing ENSO characteristics is in part induced by a climate change forcing[e.g.,Power and Smith,2007;Vecchi et al.,2006]. There is a consensus that the long-term IO warming is at least in part due to global warming[Alory et al.,2007],and the fact that the weakening in the meridional IO SST index is linked to the long-term IO warming suggests a component of climate change is active in the rainfall reduction.A quantification of the relative importance of different forcing factors awaits further investigation.[18]Acknowledgments.W.Cai and T.Cowan are supported by the Australian Greenhouse Office,the Wealth from Oceans National Research Flagship,and the South East Australia Climate Initiative.We thank an Associate Editor and two reviewers for their helpful comments. ReferencesAlory,G.,S.Wijffels,and G.Meyers(2007),Observed temperature trends in the Indian Ocean over1960–1999and associated mechanisms,Geo-phys.Res.Lett.,34,L02606,doi:10.1029/2006GL028044.Ashok,K.,S.K.Behera,S.A.Rao,H.Weng,and T.Yamagata(2007),El Nin˜o Modoki and its possible teleconnection,J.Geophys.Res.,112, C11007,doi:10.1029/2006JC003798.Baines,P.G.,and W.Cai(2000),Analysis of an interactive instability mechanism for the Antarctic circumpolar wave,J.Clim.,13,1831–1844. Cai,W.,and T.Cowan(2008),Evidence of impacts from rising tempera-tures on inflows to the Murray-Darling Basin,Geophys.Res.Lett.,35, L07701,doi:10.1029/2008GL033390.Cai,W.J.,D.A.Jones,K.Harle,T.D.Cowan,S.B.Power,I.N.Smith, J.M.Arblaster,and D.J.Abbs(2007).Past climate change,in Climate Change in Australia,technical report,edited by K.B.Pearce,et al.,pp. 17–28,CSIRO Marine and Atmospheric Research,Aspendale,Victoria, Australia.Frederiksen,J.S.(1982),Eastward and westward flows over topography in nonlinear and linear barotropic models,J.Atmos.Sci.,39,2477–2489. Hoskins,B.J.,and D.J.Karoly(1981),The steady linear response of a spherical atmosphere to thermal and orographic forcing,J.Atmos.Sci., 38,1179–1196.Kalnay,E.,et al.(1996),The NCEP/NCAR40-year reanalysis project,Bull. Am.Meteorol.Soc.,3,437–471.Marshall,G.J.(2003),Trends in the Southern Annular Mode from obser-vations and reanalyses,J.Clim.,14,4134–4143.Nicholls,N.(1989),Sea surface temperatures and Australian winter rain-fall,J.Clim.,2,965–973.Power,S.B.,and I.N.Smith(2007),Weakening of the Walker Circulation and apparent dominance of El Nin˜o both reach record levels,but has ENSO really changed?,Geophys.Res.Lett.,34,L18702,doi:10.1029/ 2007GL030854.Rayner,N.A.,E.B.Horton,D.E.Parker,C.K.Folland,and R.B.Hackett (1996).Version2.2of the Global Sea-Ice and Sea Surface Temperature dataset,1903–1994,Climate Res.Tech.Note CRTN74,25pp.,Hadley Cet.,Met Off.,Bracknell,U.K.Tanaka,H.L.,N.Ishizaki,and A.Kitoh(2004),Trend and interannual variability of Walker,monsoon and Hadley circulations defined by velo-city potential in the upper troposphere,Tellus,Ser.A,56,250–269. Vecchi,G.A.,B.J.Soden,A.T.Wittenberg,I.A.Held,A.Leetma,and M.J.Harrison(2006),Weakening of the tropical atmospheric circulation due to anthropogenic forcing,Nature,419,73–76.Whetton,P.H.(1988),A synoptic climatological analysis of rainfall varia-bility in south-eastern Australia,Int.J.Climatol.,8,155–177. Wijffels,S.,and G.Meyers(2004),An intersection of oceanic wave guides: Variability in the Indonesian Throughflow region,J.Phys.Oceanogr.,34, 1232–1253.ÀÀÀÀÀÀÀÀÀÀÀÀÀÀÀÀÀÀÀÀÀÀW.Cai and T.Cowan,CSIRO Marine and Atmospheric Research,PMB1, Aspendale,Victoria,3195,Australia.(wenju.cai@csiro.au)。