GEOPHYSICAL RESEARCH LETTERS, VOL. 28, NO.

11, PAGES 2225-2227, JUNE 1, 2001

Subtropical Indian Ocean SST dipole events

and southern African rainfall
C.J.C. Reason Oceanography and EGS Depts., University of Cape Town, South Africa and Earth Sciences, University of Melbourne, Australia

Abstract.

An atmosphericgeneral circulation model ability of the large rural population to this, there is

(AGCM) is usedto examinehow the regional atmo- great local interestin identifyingregionalSST patterns
rainfall predictability. spheric circulation and rainfall over southern Africa that may provide someseasonal respond to a recently observed dipole in subtropical

seasurface temperature (SST) overthe SouthIndian Model and experiment description

Ocean. Observationssuggestthat when SST is warm to the south of Madagascarand cool off WesternAustralia, increasedsummer rains occur over large areas of southeastern Africa. The model resultssuggest that this SST pattern leadsto increased rainfall via enhanced convergence moisterthan averageair over the region. of Increasedevaporation occursover the warm pole in the
South West Indian Ocean and this moist air is advected

The model usedis the Melbourne University GCM, a 9 level spectral model at R21 horizontal resolutionwith

climatological SST forcingtakenfrom Reynolds [1988].

Further details of the model and its ability to represent southern African climate are given in Rocha and

Simmonds [1997].Three setsof experiments wereper-

formed with the GCM forcedby different SST anomalies towards Mozambique and eastern South Africa as a re- in the South Indian Ocean. In each set, an ensembleof sult of the low pressureanomaly generatedover this 11 integrationsstarting from differentinitial conditions for December i and ending on the following April 30 pole which strengthens onshore the flow. was derived. Set i consistedof a SST dipole anomaly with warm pole centred at 35S, 55E and cold pole Introduction centredat 25S,95E (similarlocations the BY obto

served composites). set 2, the centreof the warm In In a recentGRL letter, Beheraand Yamagata, [2001 polewasmoved 28S,42E (i.e. closer southern to to -BY] use UKMO GISST2.3b data[Raynet al., 1996] Africa) while in set 3, the cold anomalywasremoved et
to identify subtropicalSST dipole eventsin the South so that the SST anomaly consistedof the warm part of Indian Ocean that are phaselockedto the austral sumthe set 2 forcing. In each set, the extent of the Gaus-

mer. Correlations with Xie andArkin [1996] data sug-

gestthat when warm SST anomalies occurto the south maximum increased slightlyfrom the BY composite to of Madagascar and cooloff WesternAustralia, summer +2C so as to make the responsethat much clearer. rains over southeastern Africa are enhanced.BY sug- Examination of the UKMO GISST data indicated that gestthat the influenceon rainfall is related to a weakening of the maritime ITCZ overthe Indian Oceanand enhancedmoisturetransport towardssoutheastern Africa by strongersoutheasterlies. The objectivein this study is to use an atmospheric GCM to explorein more detail the atmosphericmechanisms behind the observed rainfall correlationswith the SST dipole. The GCM usedhas previouslybeen appliedto climate variability 30S studiesin the regionand there is confidence the roin

sian SST anomalies

was similar

to BY but the central

bustness its results[e.g. Rochaand $immonds, of 1997; Reason al., 1998]. et

Attention is paid to the model sensitivityto the location of the warm pole in the South West Indian Ocean in summer (JFM) 850 and to the presenceof the cold pole off Western Aus- Figure 1. Difference austral between ensemble the mean Set2 experof tralia. Given the significantinterannual variability in hPa winds
rainfall over southern Africa and the associated vulner-

imentsand the climatological ensemble. scale A difference vector shown. is Superimposed SSTanomaly isthe

Copyright 2001 by theAmerican Geophysical Union.

Paper number 2000GL012735.
0094-8276/01/2000GL012735505.00

positive (negative) anomalies) in theSet2 experiused

ments. The associated MSLP difference shows statisti-

(contour interval 0.25C, (dashed) solid contours denote

callysignificant %) lower (95 (higher) pressure the over warm(cool)pole.

2225

2226

REASON:

INDIAN

OCEAN

DIPOLE

EVENTS

AND SOUTHERN

AFRICAN

RAINFALL

WestIndianOcean(Figure1) evaporates moremoisture

over the warm pole and enhances mean summeronthe
shore advection of moisture over eastern South Africa

30S

and Mozambique (Figure 2). The divergence field indicates negative anomalies relativeconvergence) (or of
the low level moist airflow over the Mozambique re-

90E

gion(Figure3). The largescale velocity potential field suggests that there is relative enhancement the loof cal Walkercirculation overthe western Indian Ocean/

southernAfrican regionand relative weakening this of cellovernorthernAustralia/ Indonesia (Figure4). At hPa. Contour interval 1x 10 kg/kgandstippling200 hPa, there are anomalouslystrong easterliesover is -4 indicates significance the 95 % level. at

Figure 2..ks forFig. 1 except mixing at 850 ratio

low latitude Africa (not shown),a situationthat is knownfrom observations [Tyson,1986]to be associated with increased summer rainfall
Africa.

some of the six individual

austral summers that went

over southeastern

into the BY compositehad maxima of about 2C.

Results

Although the model is not ideally suited to simulate precipitation since its resolution does not capture

broadly similar except that the response over south- would like, this field is presentedsinceit is of much in(Figure 5). Consistent ern Africa is considerablyenhancedin Set 2 in which terestto the local community with the BY rainfall/SST correlations, the model inthe centre of the warm pole is just to the southwestof

the stronglocal topographic (e.g. Drakensberg) and SST (e.g. AgulhasCurrent) gradients well as one as The resultsfrom the Set 1 and Set 2 experiments are

indicates that mostof the Attention is thereforefocusedon Set 2 and on compar- latent heat flux (not shown) increased precipitation arises from enhanced evaporaison of these with the Set 3 results. Given that the dipolereaches maximumin February[BY], results in tion off the warm SST pole and relatively little from .eachcaseare presented the australsummer(JFM) either local land-surfaceevaporation or remote oceanic as
differencebetween the ensemblemean of SST perturFor Set 3, the SST forcing differsfrom Set 2 in that bation runs from an ensemble climatology.Gridpoint t the cold part of the dipole off Western Australia is retests are applied for statistical significance all fields of moved so that the forcing consistsonly of the warm at the 95% level. Consistentwith linear quasigeostrophic theory and anomaly east of South Africa. The pattern of the Set 3 previous GCM results [e.g.,Reason, 2001],a low (high) circulation changesover the South West Indian Ocean is very similar to Set 2; however,the magnitude is enpressure anomalywhich decays with heightis generated hanced. Consistentwith the strengthened onshore flow over, and downstream the warm (cold) SST pole. of, of moist air towardsthe region,the modelprecipitation The associatedcyclonic wind anomaly over the South anomaly over eastern South Africa and Mozambique is increased over that in the Set 2 experiments. Recall also that the Set 1 resultsin which the warm SST pole was
sources.

Madagascar (Figure1) ratherthanwellsoutheast this of island(Set 1). This resultis consistent with previous observational studies[e.g., Walker,1990; Jury et al., 1993]which showsignificant statisticallinks between SST in the Agulhas Current regionand local rainfall.

dicates enhanced

summer

rainfall

over coastal eastern

South Africa, Mozambique and Zimbabwe. Accumulated over the season,maximum rainfall anomaliesare of order 150-180 mm which is a significantfraction of the seasonaltotal for summer. Analysis of the surface

o
',

r
30E

'"x -. '!.':::i:!::;:.
60E 0 60E 120E

30S

Contour interval is 2.5 x 10-7 s- .

As for Fig. 2 Figure As Fig. except 3. for 2 ?vergence hPa. Figure 4.Contour interval exceptvelocitypotentialat at850 850 hPa. is lx 105 m2s-.

REASON: INDIAN OCEAN DIPOLE EVENTS AND SOUTHERN AFRICAN RAINFALL

2227

West Indian Ocean, increasedevaporation there, and

',\',i;. V",', ',q,,,..v ':,

,-...... -..::... 11,

,: ....."-.", ' ' .... ',,';dM ?/ %.::.:;:...y ',... ,,:'.y ,.,

-....

transport of a moistermarine airmasstowardssoutheastern Africa where relative convergence and uplift leadsto statisticallysignificant increased rainfall. Althoughthe modelgrid doesnot allowfine resolution of the significant topographic SST gradients and in the region, mechanisms the throughwhichthe increased
rainfall comes about are consistent with observations

30S

"--.,

] ,.

,;'2.', "

and provide furthersupportfor the [BY] results.Better understanding the linksbetween of rainfallvariability andregional SST patterns of greatimportance is for agriculture and the livelihoods the large rural popof
ulation in southeastern Africa and the results of this

30E

60E

study help in that goal. To improvethe representation Figure 5. As for Fig. 2 exceptprecipitation.Contour of local SST and topographicgradients,future work intervalis 0.2 mm/day. is aimed at applying a regional climate model to the
located much further
rainfall

east in the South Indian

Ocean

southern

African

domain.

led to a considerably smallerresponse circulationand in

over southern Africa.

The model results therefore suggestthat it is the warm part of the BY SST dipole, and particularly its proximity to land, that is important for the response References

Acknowledgments. Partial funding from the South African Water ResearchCommissionand DACST is gratefully acknowledged. RossMurray assisted with the graphics.

overeastern SouthAfrica / Mozambique. Furthermore, Behera, S. K., and T. Yamagata, Subtropical SST dipole theseresults reinforce previous observational work[e.g. eventsin the southern Indian Ocean, Geophys.Res. Left., Walker,1990;Mason,1995]whichfoundsignificant sta28, 327-330, 2001.
Jury, M.R., H.R. Valentine, and J.R.E. Lutjeharms, Influence of the Agulhas Current on summer rainfall along the southeastcoast of South Africa, J. Appl. Meteor., 32, Further north (i.e. Zimbabwe), comparison the of 1282-1287, 1993. Set 2 and 3 resultssuggests that there may be some Mason, S.J., Sea surface temperature - South African rainsensitivity to the cold part of the SST dipole. Set 3 fall associations,1910-1989, Int. J. Climatol., 15, 119-135,
tistical links between South African summer rainfall and

SST in the AgulhasCurrent region.

(withoutthe coldpole)shows weakincrease evena a or

decrease in model rainfall here whereas Set 2 showed a

1995.

statistically significant increase (Figure5). The important difference in the circulation fields between these 2

sets is that the relative enhancement the ascending of branch of the Walker circulation is noticeably stronger over tropical southernAfrica in Set 2. This resultsbecause dipoleSST forcingin'Set2 enhances zonal the the SST gradient across the tropical South Indian Ocean

and therebyfacilitatesa relativestrengthening (weak-

ening) the maritime of ITCZ over western the (eastern)

part of the basin.

Concluding remarks
Three sets of SST anomaly experiments have been performedwith an atmospheric GCM. These havebeen

Rayner, N.A., E. B. Horton, D.E. Parker, C.K. Folland, and R.B. Hackett, Clim. Res. Tech. Note 74, UK Meteorological Office, Bracknell, 1996. Reason, C.J.C., Evidence for the influence of the Agulhas Current on regional atmosphericcirculation patterns, J. Climate, in press, 2001. Reason, C.J.C., C.R. Godfred-Spenning, R.J. Allan and J.A. Lindesay, Air-sea interaction mechanisms and low frequency climate variability in the South Indian Ocean region, Int. J. Clirnatol., 18, 391-405, 1998. Reynolds,R. W., A real-time globalseasurfacetemperature analysis, J. Climate 1, 75-86, 1988. Rocha, A., and I.H. Simmonds, Interannual variability of southern African summer rainfall. Part II: modelling the impact of sea surface temperatures on rainfall and circulation, Int. J. Climatol., 17, 267-290, 1997. Tyson, P.D., Climate Change and Variability in southern Africa, 220 pp., Oxford University Press, Cape Town,
1986.

motivatedby recentobservations [BY] of SST dipole Walker, N.D., Links between South African summer rainfall and temperature variability of the Agulhas and Benguela events in the South Indian Ocean which are signifiCurrent systems, J. Geophys.Res., 95, 3297-3319, 1990. cantly correlatedwith summerrainfall over large parts Xie, P., and P.A. Arkin, Analysesof global monthly precipof southeastern Africa. The model results reinforce the itation using gauge observations,satellite estimates and [BY] findings suggest the response south- numerical model predictions. J. Climate, 9, 840-858, 1996. but that over
easternAfrica is strongerif the warm pole is located
closer to the subcontinent, and that the responseover
C.J.C. Reason, Oceanography and EGS, University of

thetropical (subtropical) landmass sensitive sen- Cape Town, Private Bag, Rondebosch,7701, South Africa. is (not sitive)to the coldpart of the dipoleoff Western Aus- (e-mail:cjr@egs.uct.ac.za)
tralia. In all cases,the model response includesa nearDecember 2000;accepted 8, March3, 2001.) surfacelow over the warm SST forcing in the South (Received
_