Simulations Of Present-Day And Future Climate Over Southern Africa Using HadAM3H

A high resolution (~150 km) GCM, HadAM3H, is used to obtain enhanced regional information from a fully coupled atmosphere-ocean GCM, HadCM3, in time-slice experiments of the periods 1961-90 and 2071-2100. For the future scenarios, the IPCC SRES A2 and B2 emissions scenarios have been used, and the model results of present-day and future climates are analysed over southern Africa. The model is generally able to capture the circulation dynamics of the present-day climate, reproducing the primary features of observed circulation and the general pattern of seasonal change. In summer, however, convergence into the Intertropical Convergence Zone and Zaire Air Boundary is too strong, resulting in a positive rainfall bias over much of southern Africa. In addition, there is a surface temperature cold bias over much of the land, which is related to an excess of thick cloud and less incoming solar radiation reaching the earth’s surface, as well as the possibility of greater evaporative cooling in the model. In winter, surface temperatures south of 30°S are colder than observed, whereas between 10°S and 25°S there is a warm bias. These biases are related to cloud and circulation anomalies. In the climate change experiments, the weaker forcing in the B2 scenario results in intermediate values of surface temperature compared to the controls and the A2 scenario, with the results being less clear for precipitation. The A2 future scenario over southern Africa suggests an average warming over the region of 3.9°C in summer and 4.1°C in winter. There tends to be a drying over western and central tropical and subtropical land areas in summer, whereas equatorial regions tend to become wetter, with more intense and extreme rainfall. The same is true in general for equatorial regions in winter, but over the South Western Cape winter rainfall region of South Africa, there is a reduction in mean seasonal rain and a tendency for extreme rainfall events to become less likely.