Evaluating Validation Methods for Satellite Precipitation Products Using Point Gauge Observations over Complex Topography

The accuracy of surface rainfall observations in complex topographic regions remains challenging, particularly for sub-daily and daily temporal resolutions. This study assesses the accuracy of the Integrated Multi-Satellite Retrievals for Global Precipitation Measurement (GPM) IMERG in the mountainous region of Sumatra using optical rain gauge (ORG) data from Kototabang, West Sumatra, Indonesia [100.32° E, 0.20° S, 865 m above sea level (ASL)] during the 15-year observation period (2002–2016). The validation methods tested are based on the direct [point-to-pixel (PtP)] and the indirect methods of IMERG to the point gauge. The indirect methods used include mean, median (med), linear interpolation (LI), nearest-neighbor interpolation (NI), Inverse Distance Weighting (IDW) interpolation, and kriging (Krg) interpolation. The IMERG data validation generally indicates better correlation coefficient (CC) and critical success index (CSI) values for the indirect methods than the PtP method. For hourly data, the CC (CSI) values across all indirect methods range around 0.28 (0.28), while using the PtP method, CC is 0.25 (0.27). For daily data, nearly all indirect methods show a CC value of 0.50, except for the NI method which exhibits a CC of 0.47, the same as the CC value from the PtP method. Besides CC and CSI, the root mean square error (RMSE) and relative bias (RB) values of the indirect methods also show slightly better values than the PtP method. Despite the differing accuracy and probability detection between the direct and indirect IMERG data validation methods, the discrepancies in CC and CSI values between the methods are not impressive. Thus, the choice of validation method for complex topographic regions can be tailored based on the data processing capabilities and specific needs.