Capability of CORDEX RCMs in simulating extreme rainfall events over South africa

• Main Author: Abba Omar, Sabina
• Other Authors: Abiodun, Babatunde Joseph
• Format: Dissertation
• Language: English
• Published: University of Cape Town 2014
• Online Access: http://hdl.handle.net/11427/9103

Includes bibliographical references. === In South Africa, extreme rainfall events often lead to widespread destruction, damage infrastructure, displace communities, strain water management and even destroy lives. Past studies have shown that reliable predictions of extreme rainfall events from regional climate models (RCMs) could help reduce the impact of these events. The present study evaluates the ability of nine RCMs in simulating extreme rainfall events over South Africa, focusing on the Western Cape (WC) and east coast (EC) areas. This study defines an extreme rainfall over a location as rainfall that is equal to or above the 95th percentile of the rainfall distribution at that location, and defines widespread extreme rainfall events (WEREs) over an area as events during which more than 50 of the grid-points in the area experience extreme rainfall. The 95th percentile threshold values were calculated over 11 years (1998-2008) of South Africa’s daily rainfall data from the nine RCMs (CCLM, REMO, PRECIS, CRCM5, ARPEGE, REGCM3, WRF, RACMO and RCA35), which participated in the Coordinated Regional Climate Downscaling Experiment (CORDEX) and used ERA-Interim (ERAINT) as their boundary forcing. The simulations were compared to two observation datasets (TRMM and GPCP), and to ERAINT rainfall data to understand whether these RCMs improve on the results from ERAINT. A self organizing map (SOM) was used to characterize WEREs identified in all the datasets into archetypal groups, and ERAINT data is used to describe the underlying circulations for each archetypal rainfall pattern. The number of WEREs mapped to each rainfall pattern for each dataset allows us to get an idea of whether certain RCMs are more likely to simulate certain rainfall patterns.