Comparing Bias Adjustment Methods for CMIP6 Extreme Precipitation Projections in the San-Pédro River Basin (Côte d’Ivoire, West Africa)

West Africa (WA) is highly vulnerable to flooding and needs accurate projections of extreme precipitations to improve flood preparedness. However, selecting appropriate bias adjustment method for such projections remain challenging. This study assesses four bias adjustment methods, namely Cumulative Distribution Function Transfert Singularity Stochastic Removal (CDFt SSR), Empirical Quantile Mapping (Eqm), Delta, and Scaling in refining seventeen Coupled Model Intercomparison Project Phase 6 (CMIP6) models and their ensemble mean (EnsMean) for projecting extreme precipitations under three scenarios (SSP1-2.6, SSP2-4.5, and SSP5-8.5) over the San-Pédro River basin. Analyses were performed at the annual and long rainy season timescales. Using Climate Hazard Group Infrared Precipitation with Station (CHIRPS) observational data, seven flood-related precipitation indices (PRCPTOT, R1mm, SDII, CWD, R99pTOT, Rx5day, and Rx1day) were computed over 1991–2020. Bias adjustment methods were calibrated (1982–2001) and validated (2002–2014) using statistical (R and Pbias) and graphical evaluations. Results revealed significant discrepancies among models and methods. While Delta-adjusted models achieved the best statistical performance (R > 0.8 and Pbias < 30%), CDFt SSR-adjusted models most accurately reproduced observed daily precipitation and indices distributions. However, limitations persisted for CWD and R99pTOT. Future projections indicate increase in extreme precipitation in the near (2031–2060) and far (2061–2090) futures, relative to the baseline period (1985–2014) across all scenarios, heightening flood risks, threatening agriculture, and challenging hydropower operations. CDFt SSR emerges as the most robust method for projecting extreme precipitations, offering a robust foundation for climate impact assessments and adaptation planning in WA.