Hydrologically Relevant Bias Correction of GPM-IMERG: Performance Across Quantiles in Upland Tropical Watershed

• Identifies systematic monthly biases in satellite rainfall over upland tropical basins, affecting operations and risk analysis. • Benchmarks linear, nonlinear, and optimization-based corrections using distribution-aware diagnostics across quantiles. • Linear methods best recover central tendency; nonlinear and optimization approaches strengthen upper-tail fidelity and extremes. • Proposes a staged workflow frequency/threshold adjustment followed by shape correction for hydrologically realistic, operational use. • Outlines limitations and future directions, encouraging sub-monthly testing, regime-aware calibration, and cross-validated tuning to enhance transferability. Satellite-based rainfall (GPM-IMERG) exhibits systematic monthly biases in upland tropical watersheds, undermining reservoir operations and flood-risk estimates. This study tests whether five bias-correction methods, Linear Scaling (LS), Linear Regression (LR), Genetic Algorithm-based Correction Factor (GA-CF), Local Intensity Scaling (LOCI), and Power Transformation (PT), improve monthly totals and extreme-sensitive quantiles over the Lesti Sub-Watershed (2001-2020). Performance was evaluated using the correlation coefficient (R), Nash–Sutcliffe Efficiency (NSE), the ratio of the root-mean-square error to the standard deviation of observations (RSR), and quantile diagnostics (Q25-Q75, Q90-Q99). All methods enhance skill; LS provides the best overall monthly agreement (NSE = 0.84, R = 0.92, RSR = 0.39). LR and LOCI yield intermediate improvements by correcting level/variance and wet-day frequency, respectively. GA-CF and PT show greater benefits for the upper tail of the rainfall distribution, reducing underestimation beyond Q90 and better reproducing hallmark wet years. The contribution presents a distribution-aware, monthly-scale comparison and directly evaluates the staged LOCI–PT workflow, while identifying GA-CF as a promising standalone alternative for improving upper-tail rainfall behavior. Limitations include five gauges, month-wise parameterization, and evaluation at the monthly aggregation level; future work should assess sub-monthly behavior, regime-aware calibration (ENSO/IOD), and multi-objective tuning under more rigorous validation frameworks to improve generalizability across upland tropical watersheds.