Projected climate change in the Sahel region could strongly affect water resource availability and further compromise food security. Like other Sahelian countries, water resources in Burkina Faso face this threat. This study evaluates five precipitation bias correction methods on climate models outputs in the Mouhoun river basin. Climate Hazards InfraRed Precipitation with Stations (CHIRPS) products merged with Comité permanent Inter-Etats de Lutte contre la Sécheresse dans le Sahel (CILSS) rain gauges and regional climate model RCA4 simulations driven by three global climate models were used. The methodology consisted of identifying models' outputs biases through comparison based on rainy season main characteristics. Wilcoxon test was used to asses bias significance between observed and simulated series. Correction methods performance was assessed using the mean absolute error (MAE) and relative error. As results showed, the rainy seasons simulated by climate models are longer than those observed characterized by an early start and late end. The models overestimate the number of rainy days (+14.89 to 52.93%) and underestimate the 90th percentile of rainfall (-38.18 and -35.40%). The five biases correction methods tested in this study, LOCI, LS, GQM; EQM and GPQM; showed their ability to reduce the biases of models simulated data. Overall, EQM and GPQM are more adapted to reduce the biases of climate models simulations in the Mouhoun basin but, due to its flexibility, EQM is more indicated to correct projected precipitation for impact studies.

Keywords: Climate modeling, rainy season characteristics, bias correction, Mouhoun, Burkina Faso.