Enhancement of RegCM4.7-CLM precipitation and temperature by improved bias correction methods over Central Africa

Precipitation and temperature projections from Regional Climate Models (RCMs) over Central Africa (CA) are of great importance. However, several studies have already shown that the data from RCMs cannot be directly used for climate impact studies on a local scale because of systematic biases that characterize them. Therefore, RCM simulations must be preprocessed in order to make them more representative of climate at a local scale. The present study focuses on improving temperature and precipitation simulations from the RegCM4.7 RCM over CA. For this purpose, two correction methods are used: The adjusted Linear Scaling and Variance (Va) methods. Corrected and uncorrected precipitation and near-surface temperature are compared with Climate Hazards group Infrared Precipitation with Stations (CHIRPS) and the fifth generation of ECMWF reanalysis data (ERA5), respectively. Comparison of the performances of both methods was made during various seasons not only over the whole CA region but also over two sub-domains (Zones 1 and 2) having different characteristics. This was done on the basis of the mean bias and Root Mean Square Error. Results show that the precipitation from RegCM4 is tainted with huge wet biases compared to CHIRPS. Overall, the analysis suggests that the Va method is the most suitable for reducing the biases of RegCM4.7 simulations, particularly for precipitation irrespective of regions or seasons. However, it has been found that both methods fail to improve temperature biases in the Inter-Tropical Convergence Zone.