Evaluation of Bias Correction Methods for Coupled Model Intercomparison Project Phase 6 Model and Future Rainfall Projections over Muda River Basin

Accurate rainfall projections are vital for effective water- resource and agricultural planning in Malaysia’s Muda River Basin. This study evaluates three bias correction methods—Linear Scaling (LS), Local Intensity Scaling (LOCI), and Empirical Quantile Mapping (EQM)— applied to rainfall simulations from the CMIP6 MPI-ESM1.2-HR model. Historical observations (1989–2014) were used for calibration and validation, while future projections (2015–2100) were analyzed under SSP1-2.6, SSP2-4.5, and SSP5-8.5 scenarios. The performance of these methods was assessed based on three statistics: wet-day frequency, mean wet-day intensity and the 90th percentile of wet-day rainfall. LOCI achieved the highest overall skill with <10% errors for all statistics, followed by EQM (10–16% errors) and LS (21–36% errors). Projected rainfall, based on LOCI-corrected data, suggest a nonlinear relationship between rainfall response and emission scenarios in the near- (2026–2050) and mid-term (2051–2075) future. However, by the end of the century (2076–2100), higher emissions are associated with increased annual rainfall. While mid- term rainfall projections are relatively stable across scenarios, greater deviations emerge toward the end of century. The findings demonstrate that appropriate bias correction substantially improves accuracy of CMIP6 projected rainfall.