Evaluation of RegCM4 driven by CAM4 over Southern Africa: mean climatology, interannual variability and daily extremes of wet season temperature and precipitation

We present an analysis of a present-day climate simulation (1990-2009) for the Southern Africa region with the RegCM4 regional climate model (RCM, dx = 25 km similar to 0.22A degrees) driven by the Community Atmospheric Model version 4 (CAM4) global climate model (GCM, dx = 1A degrees). We assess the capability of the models to simulate the observed climate of the region with emphasis on precipitation, 2-m mean, minimum, and maximum temperature and large-scale circulation. The analysis focuses on seasonal climatologies, annual cycles, interannual variability, and extreme events. In addition to evaluating the performance of the models, we also attempt to assess the added value of the regional model compared to the driving one. With a few exceptions, we find that the models reproduce reasonably well the mean spatial patterns of 2-m temperature and precipitation, along with the associated seasonal cycle and interannual variability over selected sub-regions. Extreme indices of temperature and precipitation are also reasonably well reproduced. However, the RegCM4 substantially improves the simulation of daily precipitation frequency and duration of dry and wet events compared to CAM4 due to its higher resolution. The global and regional models exhibit quite different patterns of bias, an indication of the importance of internal variability and process representation for the simulation of surface climate. Given the good performance shown by the nested CAM4-RegCM4 system, we plan to use these models to generate an ensemble of projections for use in impact assessment studies for the region.