Forecasting water resources in south Africa

The scarce water resources of Africa could be better managed if reliable forecasts of drought or flood risk are available. In this paper the potential for prediction of major dam inflows in South Africa is investigated. Naturalized dam inflow data are statistically analysed and climatic predictors are assembled. Multivariate linear regression models are formulated which account for 59% of the variance two seasons in advance. Relationships between predictors such as oceanic winds, pressure and sea surface temperature, and hydrological targets are modulated by the global El Nino. Predictability is greater for southern Africa than in many other areas of the world.