A physically-based ensemble of high-resolution regional climate simulations for Sub-Saharan Africa

There is an increasing demand for sound climate information in Sub-Saharan Africa (SSA) for both regional and local scales. While climate information from Global Climate Models (GCMs) are usually too coarse for climate impact modelers or decision makers from various disciplines (e.g. hydrology and water management, agriculture, energy), Earth System Models (ESMs) provide feasible solutions for downscaling GCM output to required spatiotemporal scales. However, it is well known that the performance of regional simulations depends a lot on the physical parametrization, which may vary from region to region. Besides land-surface processes, the most crucial processes to be parameterized in ESMs include radiation, convection, and cloud microphysics, partly with complex interactions. Precipitation generation, for instance, involves many coupled processes between cumulus convection, cloud microphysics, radiation, land and ocean surface, and the planetary boundary layer. Before conducting long-term ESM simulations, it is therefore indispensable to identify a suitable physics parametrization combination. Based on the European Centre for Medium-Range Weather Forecasts (ECMWF) reanalysis product ERA-Interim, we performed a set of 16 high-resolution physics parameterization experiments for SSA, using different cumulus-, microphysics-, planetary boundary layer-, and radiation schemes in the Weather Research and Forecasting (WRF) model for the period 2006-2010 in a spatial resolution of 9 km. Based on traditional (Taylor diagram, probability densities) and more innovative validation metrics (ensemble structure–amplitude–location (eSAL) analysis, Copula functions) and with the use of various observation data for precipitation and temperature, favorable parameterization combinations for whole SSA are identified and will be discussed also w.r.t. the required computing time. Here, we find that complex radiation schemes do not urgently lead to better simulation results for SSA, but increase the computing time tremendously.