Evaluation of rainfall simulations over Uganda in CORDEX regional climate models.
Résumé
The study evaluates the ability of ten regional climate models (RCMs) to simulate the present-day rainfall over Uganda within the Coordinated Regional Downscaling Experiment (CORDEX) for the period 1990–2008. The models’ ability to reproduce the space-time variability of annual, seasonal, and interannual rainfall has been diagnosed. A series of metrics have been employed to quantify the RCM-simulated rainfall pattern discrepancies and biases compared to three gridded observational datasets. It is found that most models underestimate the annual rainfall over the country; however, the seasonality of rainfall is properly reproduced by the RCMs with a bimodal component over the major part of the country and a unimodal component over the north. Models reproduce the interannual variability of the dry season (December–February) but fail with the long and short rains seasons even if the ENSO and IOD signal is correctly simulated by most models. In many aspects, the UQAM-CRCM5 RCM is found to perform best over the region. Overall, the ensemble mean of the ten RCMs reproduces the rainfall climatology over Uganda with reasonable skill.