South American rainfall impacts associated with inter-El ...web.maths.unsw.edu.au/~matthew/Hill_SA_GRL.pdf · South American rainfall impacts associated with ... South American rainfall

South American rainfall impacts associated with inter-El Nino

variations

K. J. Hill,1 A. S. Taschetto,1 and M. H. England1

Received 25 July 2009; revised 25 August 2009; accepted 31 August 2009; published 1 October 2009.

[1] The impacts of inter-El Nino events on SouthAmerican circulation during austral summer areinvestigated using observations and an atmosphericgeneral circulation model (AGCM). The AGCM wasforced with sea surface temperature (SST) anomalies inthe tropical Pacific for the two El Nino events of 1997/1998(EN97) and 2002/2003 (EN02). The strong eastern PacificSST anomaly of EN97 resulted in a typical displacement ofthe Walker circulation, causing a decrease in precipitationacross the north of South America. A strengthened low-level jet (LLJ) east of the Andes during EN97 enhanced themoisture transport from low latitudes to the subtropics,leading to intensified precipitation over southeastern SouthAmerica. The simulated circulation in EN02 reveals aweakened LLJ and anomalous convergence of moistureover eastern South America, which can be attributed to adisplacement of the Pacific-South American (PSA) mode inresponse to the different location of the heat sources alongthe tropical Pacific Ocean. Citation: Hill, K. J., A. S.

Taschetto, and M. H. England (2009), South American rainfall

impacts associated with inter-El Nino variations, Geophys. Res.

Lett., 36, L19702, doi:10.1029/2009GL040164.

1. Introduction

[2] The tropical Pacific Ocean’s leading mode of vari-ability, namely the El Nino Southern Oscillation (ENSO),impacts South American rainfall via atmospheric telecon-nections [Ropelewski and Halpert, 1987; Grimm, 2003].Within the ENSO mode there is a level of inter-eventvariability, most notably characterised by variability in thelocation of peak sea surface temperature (SST) anomaliesalong the equatorial Pacific. In 1997, the near recordstrength El Nino (EN97) exhibited a ‘‘classic’’ structure(Figure 1a) and evolution, with the largest anomalieslocated in the eastern Pacific. In contrast, the 2002 El Nino(EN02) event was much weaker, with the largest SSTanomaly located further west in the central equatorialPacific (Figure 1b) and virtually no anomaly in the easternPacific. The mild EN02 event had its maximum anomalylocated near 160�W, extending across the dateline. Althoughthe warm anomalies during EN97 did extend as farwest as the anomalies seen in EN02, the SST anomaliesduring EN02 were larger over the Nino 4 region (Figures 1aand 1b). Despite the lower intensity in SST, some climate

impacts were stronger in the 2002 El Nino event; forexample, there were more severe droughts over Australia[Wang and Hendon, 2007] and South Africa [Reason andJagadheesha, 2005]. The difference in the position of theSST anomaly along the equatorial Pacific is thought to beresponsible for distinct rainfall responses associated withwarm ENSO events across Australia [Wang and Hendon,2007; Taschetto and England, 2009], driven by shifts in theWalker circulation [Ashok et al., 2007] and anomalousmoisture fluxes [Taschetto et al., 2009].[3] During the positive phase of ENSO, anomalous

warmer waters across the eastern Pacific modify the Walkercirculation by increasing convection in the east and reduc-ing it over the west and Indonesian region. Over SouthAmerica, enhanced subsidence east of the anomalous Walkercirculation suppresses rainfall across northeast Brazil[Ropelewski and Halpert, 1987]. ENSO also remotelyimpacts the South American extratropical climate via thequasi-stationary Rossby wave trains often referred to as thePacific South American (PSA) pattern [Karoly, 1989; Mo,2000]. El Nino events are commonly associated with pre-cipitation anomalies in a dipole-like pattern over SouthAmerica. Rainfall is reduced over southeastern Brazil andincreased to the south of this region [e.g., Grimm, 2003;Ropelewski and Halpert, 1987]. During El Nino events,warm moist air from the Amazon basin is transported fromthe tropical latitudes to the subtropics along an enhanced low-level jet (LLJ) east of the Andes. This configuration isconducive to rainfall over the La Plata basin and a weakeningof the South Atlantic Convergence Zone (SACZ) oversoutheastern Brazil [e.g., Herdies et al., 2002].[4] The SACZ is a common feature of the South American

climate seen in austral summer months (Dec–Feb; DJF). TheSACZ appears as a convective band characterised by theconvergence of warm, moist air extending from the AmazonBasin to the subtropical South Atlantic Ocean [Kodama,1992; Carvalho et al., 2002]. Recently, Silva and Ambrizzi[2006] showed observational evidence that the El Nino eventof 2002/2003 displaced the position of the South AmericanLLJ, resulting in less moisture transport to the south com-pared to the 1997/1998 event.[5] Motivated by the observational study of Silva and

Ambrizzi [2006] and with no previous model studies withsuch an experimental design available, we investigate herethe impact of the 1997 (EN97) and 2002 (EN02) El Ninoevents on the climate over South America, using not onlyobservations but also ACGM ensemble experiments. Wefocus our analyses on the South American monsoonseason, i.e., DJF. The SST anomalies for both the EN97and EN02 events were also at a peak during DJF, further

GEOPHYSICAL RESEARCH LETTERS, VOL. 36, L19702, doi:10.1029/2009GL040164, 2009ClickHere

for

FullArticle

1Climate Change Research Centre, University of New South Wales,Sydney, New South Wales, Australia.

Copyright 2009 by the American Geophysical Union.0094-8276/09/2009GL040164$05.00

L19702 1 of 5

supporting the choice of this season as the focus of thepresent study.

2. Data Analysis and Atmospheric Model

[6] In this study we analyse the rainfall data from theClimate Prediction Center Merged Analysis of Precipitation(CMAP) [Xie and Arkin, 1996] and SST from the HadleyCentre climatology (HADISST1) [Rayner et al., 2003]. Toassess the importance of inter-El Nino SST variability inmodulating South American rainfall, perturbation experi-ments were conducted using the National Centre for Atmo-spheric Research (NCAR) Community Atmosphere Model(CAM3) which has a T42 horizontal resolution (approxi-mately 2.8� latitude-longitude) with 26 hybrid sigma/pressure vertical levels. For a detailed description of thismodel see Collins et al. [2006].[7] A 50 year integration forced by the 12-month clima-

tology of SSTwas taken as the control experiment (CNTRL).The EN97 and EN02 ensemble sets consist of 50 integra-tions of the model run over 28 month periods; forced byobserved SST conditions during the periods September1996 – December 1998 and September 2001 – December2003, respectively. The anomalous monthly varying SSTperturbations were applied over the region 15�N to 15�S. Alinear weighted border was applied outside that region over a10� latitude band, to reduce spurious atmospheric circulationset up by unrealistic gradients at the edges of the SSTperturbation. These anomalies were then superimposed ontothe monthly varying climatology. The ensemble memberswere run under identical conditions, apart from a slightlydifferent atmospheric initial state, to provide an estimate ofthe internal variability in the atmosphere. A two-tailed

student t-test is used to assess where the experiments orobservations are statistically different from the control run orthe observed climatology.

3. Results

[8] To assess the different impact of EN97 and EN02events on precipitation over South America, the differencebetween these two events is shown (EN97 minus EN02) forthe observations (Figure 1c) and the mean of the ensembleset of El Nino experiments (Figure 1d). The observed andsimulated ensemble rainfall differences both show a similarspatial distribution, with the three main features from EN97and EN02 reproduced; namely (1) increased precipitationover the eastern equatorial Pacific in EN97 relative to EN02;(2) reduced rainfall over the northern half of South Americain EN97 relative to EN02; and (3) enhanced precipitationover the southeast of the continent in EN97 relative toEN02. The higher rainfall across the northwest coastline ofSouth America and over the east Pacific Ocean in the EN97case compared to the EN02 is the response to the large SSTanomaly located in the east Pacific during EN97, while thepeak anomaly is much more remotely located during EN02.Drier conditions over the northern half of the continent inEN97 is a response to the stronger Walker circulation anom-alies during the 1997/1998 case compared to 2002/2003. Theincrease in rainfall in the southeast during EN97 comparedto EN02 is associated with a shift in the position of theSACZ and a consequent enhanced southward moisturetransport via a strengthened LLJ during the canonical EN97.[9] The anomalous Walker circulation response in EN97

relative to EN02 is illustrated in Figure 2 by vectors ofanomalous vertical velocity and the divergent zonal wind

Figure 1. Sea surface temperature (SST) anomalies from the Hadley Centre (HADISST1) during DJF (a) EN97 and(b) EN02. The boxes in Figures 1a and 1b represent the Nino 4 region with the DJF SSTanomaly (�C) noted above left of thebox. Note that the EN97 and EN02 experiments only incorporate SST anomalies in the 15�S–15�N tropical region. Rainfalldifference during DJF between EN97 and EN02 (shown as EN97 minus EN02) for (c) the observations and (d) thesimulations. Units are in degrees Celsius (Figures 1a and 1b) and mm day�1 (Figures 1c and 1d) respectively. Colour shadedregions are statistically significant at 95% confidence level according to a two-tailed student t-test.

L19702 HILL ET AL.: SA RAINFALL AND INTER-EL NINO VARIATIONS L19702

2 of 5

component averaged over 10�S – 10�N. During the EN97experiments (Figure 2a), anomalous uplift is located between170�W and 90�W. Regions of descent are located over theSouth American continent and west of the date line. Incomparison, the anomalous region of uplift in the EN02 case(Figure 2b) is shifted west to the central Pacific Ocean, whileit is flanked by two weaker descending branches. This is adirect consequence of the maximum SST anomaly beinglocated further west in EN02 and there being virtually no SSTanomaly along the west coast of South America (Figure 1b).Although the location of the anomalous eastern subsidingbranch of the Walker cell remains unchanged between EN97and EN02, it is stronger in EN97 compared to EN02 around60�W (see Figure 2c).[10] The distribution of moisture flux and its convergence

at 850 hPa over the continent (Figures 2d–2f) reveals adifference in the low level moisture flux anomalies betweenthese two events. The EN97 ensemble (Figure 2d) has astrong and well organised northwesterly anomaly extendingfrom 10�S to 30�S. Under this region of strengthenedmoisture flux there is significant convergence in EN97 whichwould be conducive to unstable convective conditions lead-ing to higher rainfall, as seen in the ensemble experiments(Figure 1d). This northwesterly low-level moisture fluxanomaly is weaker over the southeast in the EN02 case(Figure 2e), wherein stronger moisture flux convergence islocated further north near 15�S. The difference between thesetwo El Nino events shows anomalous divergence of moisture

over northern South America and an intensification of thenortherly low-level moisture flux anomaly to the La PlataBasin in 1997/1998 relative to 2002/2003 (Figure 2f).[11] Following Ambrizzi and Hoskins [1997] we use

200 hPa meridional winds as an indicator for the ENSOteleconnection strength and pathways. When examiningthese meridional wind anomalies for the EN97 and EN02cases (Figure 3) we see wave trains characteristic of a PSApattern. During EN97 (Figure 3a), over 30�S and 50�W, thereis a strong negative (northerly) wind anomaly flanked diag-onally by positive (southerly) wind anomalies. During EN02(Figure 3b), the meridional wind anomalies are smaller inmagnitude over the same region. The wave train is alsoseen clearly in the difference plot of EN97-EN02 merid-ional wind anomalies (Figure 3c).[12] To assess the impact of a singular SST forcing

without the confounding influences of other remote factorsduring the 1997/1998 and 2002/2003 El Ninos, we examinedthe 200 hPa meridional winds from a set of idealisedensemble experiments. These were forced by a +1� SSTwarming superimposed on the seasonal cycle over the easternPacific (120�W–80�W, 10�S–10�N) and the central westernPacific around the date line (160�E–160�W, 10�S–10�N),hereafter referred to as the EPAC and CWPAC experiments,respectively. The meridional winds reveal a change in signaround 30�S, 50�W with a negative (northerly) anomaly inthe EPAC ensemble and a positive (southerly) anomaly inCWPAC (Figure S1 of the auxiliary material), suggesting a

Figure 2. (top) Vectors of vertical velocity (1 � 10�3 m s�1) and divergent zonal (m s�1) average wind componentanomalies during DJF averaged between 10�S–10�N for the ensemble set of El-Nino experiments (a) EN97, (b) EN02 and(c) difference between the two events (EN97–EN02). Shaded areas show positive (upward) vertical velocities. (bottom)Moisture flux and moisture convergence at 850 hPa during DJF for the ensemble set of El Nino experiments (d) EN97,(e) EN02 and (f) difference between the two ensembles (EN97–EN02). Units are kg/m/s for colour shading and the maximumvector length is 10 kg/m/s. In all plots black vector and colour shaded regions (Figures 2d–2f) are statistically significant at95% confidence level according to a two-tailed student t-test.

L19702 HILL ET AL.: SA RAINFALL AND INTER-EL NINO VARIATIONS L19702

3 of 5

phase shift of the PSA over the continent.1 These twowind configurations generate an anomalous anticyclonic(cyclonic) circulation over southeastern South Americafavouring (inhibiting) the moisture transport to the La Platabasin during EPAC (CWPac). Interestingly, when we examinethe low-level winds from the EN97 and EN02 experiments(Figures 3d and 3e) we find the same phase shift in thecirculation around 30�S, 50�W, as in the idealised sensitivityexperiments. This suggests that the location of peak tropicalSST anomaly could be a determining factor in setting theresponse phase of South America atmospheric circulation toEl Nino events.

4. Summary and Conclusions

[13] In order to assess the impact of inter-El Nino eventvariability on South American climate we perturbed theNCAR CAM3 model with observed 15�S–15�N SSTanomalies from the 1997/1998 and 2002/2003 El Ninoevents. The resulting ensemble set sufficiently reproducesthe major observed rainfall differences between the twoevents. The significantly higher rainfall during 1997/1998over the La Plata basin appears to be a direct result of theincreased available moisture transported to the subtropics viaan intensified South American LLJ. The widespread drierconditions across the northern half of the continent is shownto be a consequence of increased subsidence from theWalkercirculation in the EN97 case compared to EN02. Theseexperiments confirm that the location of the SST heat sourcemodifies the Walker circulation by changing the region ofatmospheric convection and uplift across the equatorialPacific. The more centrally located SST anomaly during

EN02 results in an upward anomaly across the dateline,while the EN97 case exhibits stronger convection in the east.During both EN97 and EN02 a secondary descending branchof the anomalous Walker circulation is located east of 90�W.These circulation responses to the position of the tropicalPacific SST anomalies impact the convection over SouthAmerica. Both cases show anomalous subsidence over thecontinent but with important differences in magnitude, whichcan explain the negative difference in precipitation across thenorth between EN97 and EN02 (Figure 1c). The shift in theregion of uplift is also seen in the 200 hPa velocity potentialand divergence (Figure S2), with a slightly stronger magni-tude of convergence over northern South America for EN97compared to EN02.[14] Investigation of the moisture flux and convergence

indicated that the main difference between these two eventswas a weakening of the LLJ due to anomalous northwardlow level winds east of the Andes in 2002/2003, that isgenerally present during canonical El Nino events such as1997/1998. Convergence of moisture and an increase inwinds into the region of the SACZ would be conducive toan increase in rainfall. This is shifted to the south in EN97and to the north in the EN02 ensemble set. This resultcorroborates the findings of Silva and Ambrizzi [2006] whoobserved a southward shift and strengthening of the LLJ inEN97, and a weaker and more northward located LLJ inEN02.[15] To further explain the change over southeastern

Brazil we investigated the teleconnection patterns usingupper-level meridional winds. The 200 hPa meridional windsin EN97 compared to EN02 showed a stronger northerlyanomaly over southeastern South America (Figure 3a),leading to enhanced rainfall over the La Plata Basin(Figure 1a) via intensified moisture transport into the region(Figure 2f). This reduction in the magnitude could explain

Figure 3. Meridional wind anomalies at (left) 200 hPa and (right) 850 hPa during DJF for (a and d) EN97, (b and e) EN02and (c and f) the difference between these two ensembles (EN97–EN02). Note that different color scales are used for the200 hPa and 850 hPa levels. Units are m s�1. For all plots, only the values with statistical significance greater then the 95%confidence level according to a two-tailed student t-test are shown.

1Auxiliary materials are available in the HTML. doi:10.1029/2009GL040164.

L19702 HILL ET AL.: SA RAINFALL AND INTER-EL NINO VARIATIONS L19702

4 of 5

the reduction in rainfall (Figure 1c) via weakening of thesouthward moisture transport into the region. On the otherhand, the meridional wind anomaly in 2002/2003 wassufficiently weak to keep the moisture transport to south-eastern Brazil.[16] It is worth noting that, apart from the location of the

maximum SST anomalies in the tropical Pacific, otherfactors can also influence the moisture availability andconvergence over South America, which then leads to rainfallanomalies over the La Plata Basin. One such factor is thegradient across the Subtropical South-Central Pacific(SSCP), as discussed by Vera et al. [2004], who showedthrough observations that the sign of the SST anomaly in theSSCP modified the strength of the teleconnection to south-eastern Brazil. Furthermore the amplitude and time evolu-tion of El Nino SST anomalies could also play a large rolein the variability of precipitation over this region. The presentstudy focused specifically on the influence of changes in thelocation of the maximum anomaly of the El Nino SSTsignature.[17] Further comparison with an idealised experiment set

was used to identify the mechanism responsible for theobserved climate shift in the LLJ and precipitation over theregion. The meridional winds from the idealised experimentsbased on simplified and localised SST warming showed asignificant difference between experiments forced by easternversus central-west Pacific SST anomalies, with a clearphase shift of the PSA apparent. This phase shift in theidealised experiments is possibly the main factor responsiblefor the different responses in South American atmosphericcirculation between the 1997/1998 and 2002/2003 El Ninoevents. The phase shift over southeastern South America wasalso seen in the low-level 850 hPa winds in the EN97 andEN02 experiments, suggesting that the PSA phase shift isresponsible for the shift in the moisture transport oversoutheastern Brazil between the two El Nino events. Insummary, the location of the maximum SSTanomaly appearsto be an important factor in controlling the amount ofmoisture transported into the SACZ region during El Ninoevents. This has important implications for seasonal andinterannual climate variability over South America.

[18] Acknowledgments. This study was supported by the AustralianResearch Council. The CMAP data provided by the NOAA/OAR/ESRLPSD, Boulder, CO, USA and the Hadley Centre HADISST1 climatologyare gratefully acknowledged. The model simulations were conducted on theAustralian National Computational Infrastructure (NCI) supercomputingplatforms.

ReferencesAmbrizzi, T., and B. J. Hoskins (1997), Stationary Rossby-wave propaga-tion in a baroclinic atmosphere, Q. J. R. Meteorol. Soc., 123, 919–928.

Ashok, K., S. K. Behera, S. A. Rao, H. Weng, and T. Yamagata (2007),El Nino Modoki and its possible teleconnection, J. Geophys. Res., 112,C11007, doi:10.1029/2006JC003798.

Carvalho, L. M., C. Jones, and B. Liebmann (2002), Extreme precipitationevents in southeastern South America and large-scale convective patternsin South Atlantic Convergence Zone, J. Clim., 15, 2377–2394.

Collins, W. D., et al. (2006), The Community Climate System Modelversion 3 (CCSM3), J. Clim., 19, 2122–2143.

Grimm, A.M. (2003), The El Nino impact on the summermonsoon in Brazil:Regional processes versus remote influences, J. Clim., 16, 263–280.

Herdies, D. L., A. da Silva, M. A. F. Silva Dias, and R. Nieto Ferreira(2002), Moisture budget of the bimodal pattern of the summer circulationover South America, J. Geophys. Res., 107(D20), 8075, doi:10.1029/2001JD000997.

Karoly, D. J. (1989), Southern Hemisphere circulation features associatedwith El-Nino Southern Oscillation, J. Clim., 2, 1239–1252.

Kodama, Y. M. (1992), Large-scale common features of subtropical pre-cipitation zones (the Baiu frontal zone, the SPCZ and SACZ). Part I:Conditions of subtropical frontal zones, J.Meteorol. Soc. Jpn., 70, 813–836.

Mo, K. (2000), Relationships between low-frequency variability in theSouthern Hemisphere and sea surface temperature anomalies, J. Clim.,13, 3599–3610.

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), Global analyses of seasurface temperature, sea ice, and night marine air temperature since thelate nineteenth century, J. Geophys. Res., 108(D14), 4407, doi:10.1029/2002JD002670.

Reason, C. J. C., and J. D. Jagadheesha (2005), A model investigation ofrecent ENSO impacts over southern Africa, Meteorol. Atmos. Phys., 89,181–205.

Ropelewski, C. F., and M. S. Halpert (1987), Global and regional scaleprecipitation patterns associated with the El Nino Southern Oscillation,Mon. Weather Rev., 115, 1606–1626.

Silva, G. M., and T. Ambrizzi (2006), Inter-El Nino variability and itsimpact on the South American low-level jet east of the Andes duringaustral summer—Two case studies, Adv. Geosci., 6, 283–286.

Taschetto, A. S., and M. H. England (2009), El Nino Modoki impacts onAustralian rainfall, J. Clim., 22, 3167–3174.

Taschetto, A. S., C. C. Ummenhofer, A. Sen Gupta, and M. H. England(2009), Effect of anomalous warming in the central Pacific on theAustralian monsoon, Geophys. Res. Lett., 36, L12704, doi:10.1029/2009GL038416.

Vera, C., G. Silvestri, V. Barros, and A. Carril (2004), Differences in ElNino response over the Southern Hemisphere, J. Clim., 17, 1741–1753.

Wang, G., and H. H. Hendon (2007), Sensitivity of Australian rainfall tointer-El Nino variations, J. Clim., 20, 4211–4226.

Xie, P., and P. A. Arkin (1996), Analyses of global monthly precipitationusing gauge observations, satellite estimates and numerical model pre-dictions, J. Clim., 9, 840–858.

�����������������������M. H. England, K. J. Hill, and A. S. Taschetto, Climate Change

Research Centre, University of New South Wales, Sydney, NSW 2052,Australia. ([email protected])

L19702 HILL ET AL.: SA RAINFALL AND INTER-EL NINO VARIATIONS L19702

5 of 5