<![CDATA[Reliable precipitation data are essential for hydrological modeling in data-scarce basins. This study evaluates five statistical bias-correction methods—Correction Factor (mean-ratio scaling), Linear Scaling (mean adjustment), Linear Regression, Local Intensity Scaling (LOCI; wet-day threshold and intensity adjustment), and Power Transformation—to improve satellite rainfall for the Gembong Watershed, Pasuruan, East Java, Indonesia. We used daily TRMM (2004–2013) and GPM IMERG (2014–2023) estimates harmonized to a common grid and time step and compared them with gauges using Pearson’s r, Nash–Sutcliffe Efficiency (NSE), and the RMSE-observation standard deviation ratio (RSR). LOCI delivered the best overall balance (NSE = 0.92; r = 0.84; RSR = 0.55), while Linear Scaling achieved a slightly lower NSE but the smallest RSR (NSE = 0.87; RSR = 0.49). Power Transformation showed limited skill (NSE = 0.57; RSR = 0.90) despite high correlation. Ranking prioritized NSE with r and RSR as supporting metrics. The coastal-lowland setting of Pasuruan—with strong convective rainfall and heterogeneous land use—makes accurate bias correction particularly consequential for flood and water-resources analysis. We conclude that LOCI’s adaptive thresholding is well-suited to such regimes and that the comparative framework aids method selection for similar data-scarce watersheds.]]>
Copyrights © 2025
