How well do CORDEX models simulate extreme rainfall events over the East Coast of South Africa?

This study assesses the capability of regional climate models (RCMs) in simulating the characteristics of widespread extreme rainfall events over the East Coast of South Africa. Simulations of nine RCMs from the Coordinated Regional Downscaling Experiment (CORDEX) were analyzed for the study. All the simulations cover 12 years (1996–2008). Using the 95th percentile of daily rainfall as the threshold of extreme events and the simultaneous occurrence of extreme events over 50 % of the East Coast as widespread extreme events (WERE), we compared the characteristics of simulated WERE with observations (GPCP and TRMM) and with the reanalysis (ERAINT) that forced the simulations. Most RCMs perform well in simulating the seasonal variation of WEREs over the East Coast but perform poorly in simulating the interannual variability. Based on their rainfall synoptic patterns over Southern Africa, the WEREs in the East Coast can be generally classified into four groups. The first group connects the WEREs with tropical rainfall activities over the subcontinent. The second group links WEREs with frontal rainfall south of the subcontinent. The third group links the WEREs with both tropical and temperate rainfall activities while the fourth group represents isolated WEREs. The RCMs show different capabilities in simulating the frequency of WERE in each group, some perform better than ERAINT while some perform worse. Results of this study could provide information on the usability of RCMs in downscaling the impact of climate change on widespread extreme rainfall events over South Africa.