How Well CMIP5 Earth System Models And CORDEX Regional Climate Models Simulate Temperature and Precipitation of Present Climate During JJAS Season Over West Africa

Abstract:

This study investigates whether Regional Climate Models (RCMs) are always adding value at local scale to the performance of Global Climate Models (GCMs). Therefore, for June-September season, we compare Multi Model Ensemble (MMEs) of temperature and precipitation from 16 members from CORDEX ((Coordinated Regional Climate Downscaling Experiment) RCMs, 9 members from CMIP5(Coupled Model Intercomparison Project Phase 5) which are the driving GCMs for RCMs and 29 members from CMIP5 ESMs to the observation gridded data sets. We found that both ESMs and RCMs MMEs are able to capture spatial patterns of precipitation and temperature over West Africa with RCMs MMEs having more local details. For temperature, CORDEX RCMs MMEs shows a cold bias over West Africa compare to ESMs MMEs but for precipitation there is clear bias reduction. Over the Sahel and Gulf of Guinea there is more added value for precipitation downscaling compare to temperature one and uncertainty (spread) is reduced. Rotated EOFs was used to appreciate the ability of the models to capture the inter annual variability. We found that the CORDEX MMEs was able to capture the observed spatial patterns of the first Rotated EOFs but fail to reproduce the inter annual variability compare to CMIP5 and CMIP5_SUBSET. For the second and third Rotated EOFs for precipitation, CORDEX MMES poorly capture the variability with a respective correlation of 0.39 and 0.20. Concerning temperature, only the third mode was well capture with a correlation of 0.78. The global warming trend was also well captured by the models temperature three modes.