Modelling present and future climates over Southern Africa.

Thesis submitted to the Faculty of Science, Department of Geography and Environmental Studies, University of the Witwatersrand, Johannesburg, in fulfillment of the requirements for the Degree of Doctor of PhilosophyThe representation of contemporary southern African climate by a wide range of general circulation models used in climate studies is evaluated. In addition, projections of regional climate change by the models are interpreted in terms of their present climate performance. Projections of regional climate change by two different types of climate models are considered. First, projections of the equilibrium response to an instantaneous doubling of atmospheric carbon dioxide using atmospheric models linked to simple mixed-layer oceans are assessed. Second, projections of the transient response to gradually-increasing anthropogenic forcing by fully-coupled ocean-atmosphere general circulation models are considered. All of the mixed-layer models considered have been developed since 1990 and are more recent and generally higher-resolution versions of the models considered previously for southern Africa. The improved resolution and model physics result in a general improvement in the representation of several features of circulation around southern Africa. Specifically, these include the meridional pressure gradient, the zonal wind profile, the intensity and seasonal location of the circumpolar trough and the subtropical anticyclones, as well as planetary wave structure at 500 hPa. Atmospheric models forced by observed sea-surface temperatures simulate the large-scale circulation adjustments around southern Africa known to accompany periods of above- and below-average rainfall over the subcontinent. Fully-coupled models simulate the observed features of intra- and intra- annual variability in mean sea-level pressure, although the simulated variability is weaker than observed. Summer rainfall totals throughout southern Africa are overestimated by all of the models, although the pattern of rainfall seasonality over the subcontinent as a whole is well-reproduced. The inclusion of sulphate aerosols in addition to greenhouse gases does not result in a statistically significant improvement in the simulation of contemporary temperature variability over southern Africa. Warming projected by fully-coupled models is smaller than projections by mixed-layer models due to the fact that the transient response of the fully-coupled system and not an equilibrium response of an atmospheric model linked to a mixed-layer ocean is simulated. The inclusion of sulphate aerosols results in a reduction in the magnitude and rate of warming over southern Africa. Projected changes in the diurnal temperature range are seasonally-dependent, with increases in summer and autumn and decreases in winter. Simulated changes in mean sea-level pressure are small but similar in magnitude to observed anomalies associated with extended wet and dry spells over the subcontinent. No change in rainfall seasonality over southern Africa is expected. Nonetheless, little confidence exists in projected changes in total rainfall. While both types of model simulate a 10-15% decrease in summer rainfall on average, projected changes are smaller than the simulation errors and little inter-model consensus in terms of the sign of projected changes exists. No change in the location or intensity of anticyclonic circulation and divergence at 700 hPa in winter is expected. While fully-coupled models provide a more comprehensive treatment of the global climate system and the process of climate change, there is no evidence to conclude that current fully-coupled models should be used to the exclusion of mixed-layer models when developing regional climate change scenarios for southern Africa.Andrew Chakane 201