<![CDATA[This study evaluates the effectiveness of six bias correction methods, namely Linear Scaling, Delta, second- and third-order Polynomial, Quantile Mapping, and Hybrid Polynomial–Quantile Mapping, in improving satellite-based precipitation estimates and assessing the performance of three satellite rainfall products through validation and verification processes. In addition, the influence of rainfall classification on validation results is examined. Model performance is evaluated using the correlation coefficient, percent bias (PBIAS), Nash–Sutcliffe efficiency (NSE), and the ratio of RMSE to standard deviation (RSR). The results indicate that PERSIANN-CCS, despite having the smallest grid size and the highest spatial resolution, exhibits greater rainfall variability and lower agreement with rain gauge observations, particularly during extreme and minimum rainfall events. In contrast, GSMaP and Power MERRA-2 demonstrate rainfall patterns that are more consistent with observed data. Rainfall classification shows that the calibration dataset consists of 60% normal years and 40% wet years, with no dry years, while the verification dataset does not include wet-year conditions. Based on the calibration and validation results, Power MERRA-2 corrected using the third-order polynomial method provides the best performance at daily, monthly, and annual timescales. Verification results indicate satisfactory performance at monthly and annual scales, as well as improved daily-scale performance under normal-year verification scenarios, supported by cumulative distribution function (CDF) analysis]]>
Copyrights © 2026
