<![CDATA[Satellite-based rainfall products such as CHIRPS are essential in data-scarce tropical regions, but they require bias correction to improve reliability. This study compares five correction techniques—Linear Regression, Linear Scaling, a static Correction Factor, a Genetic Algorithm (GA)-optimized Correction Factor, and a Python-based Temporal Analysis—against gauge observations in the Petung Watershed, East Java, Indonesia. The GA method optimized nonlinear correction coefficient by minimizing RMSE through iterative selection and mutation processes. The Temporal Analysis applied monthly dynamic scaling using Python scripts to account for seasonal rainfall variability. Model performance was assessed using the Nash–Sutcliffe Efficiency (NSE), Pearson correlation (R), and the RMSE–Standard Deviation Ratio (RSR). Linear Scaling achieved the best results (R = 0.857, NSE = 0.724, RSR = 0.547), followed by Linear Regression. The GA-based approach showed marginal improvement over the static factor (NSE = 0.658 versus 0.639). Temporal Analysis improved correlation (R = 0.813) but showed poor performance overall (RSR = 1.425), indicating residual errors exceeding natural data variability. While statistical methods performed best in this case, the poor results of the complex methods reflect implementation limitations—rather than inherent inferiority. This study also highlights the importance of including RSR alongside conventional metrics to expose residual structures often masked by high correlation.]]>
