The relationship between El Nino and drought over Botswana

Numerous studies have shown that drought occurs in Botswana in association with most Pacific ENSO episodes. Using composite analysis, an attempt is ma de to identify characteristic differences between those episodes during whi ch drought occurs and those in which it does not. The evolution of the Sout hern Oscillation index (SOI) and of SSTs in key sectors of the Atlantic, Pa cific, and Indian Oceans during ENSO episodes between 1946 and 1992 is exam ined. The most pronounced contrast is in SSTs in the Atlantic and Indian Oceans. For the "dry Botswana" composite, SST anomalies are predominantly positive throughout the latter half of the episode, when drought tends to occur. For the "wet Botswana" composite, SST anomalies are overwhelmingly negative in the Atlantic and generally negative or near normal in the Indian Ocean. Di fferences are also apparent in the seasonal evolution of the SOI and Pacifi c SSTs. In the Pacific there is weaker and less consistent warming in the w et Botswana composite than in the dry Botswana composite, especially in the central and western Pacific. Also, the SOI is less negative and the low in dex period is less persistent in the wet Botswana composite. Because the SOI measures atmospheric, rather than oceanic, response, this m ay imply a weaker atmospheric response and weaker atmosphere-ocean coupling during the Pacific ENSO episodes that do not evoke drought in Botswana. Th is, rather than specific patterns of SSTs, appears to be the critical facto r. The Pacific ENSO events that produce drought tend to be the Type 1 event s identified in earlier studies: those with strong central Pacific warming that persists well into the boreal winter. This is the key to the developme nt of midlatitude teleconnections. SSTs alone are of limited value in forecasting drought occurrence in Botswa na. However, the statistical associations between ENSO and rainfall have so me forecast potential. In all six rainfall regions examined, the probabilit y of a wet season is very low at any time during the second half of an ENSO episode, but it is particularly low for the April-June (AMJ) season (i.e., AMJ of the post-ENSO year). During that season the likelihood of rainfall being more than one-half of a standard deviation below average is about 70% or better in all regions.