The spatiotemporal accuracy of precipitation products is critical for their application in watershed hydrology, meteorology and drought monitoring. This study introduces a novel, multi-scale assessment framework to evaluate the accuracy of four widely used precipitation products the Global Precipitation Measurement Integrated Multi-Satellite Retrievals (GPM IMERG), the Multi-Source Weighted Ensemble Precipitation (MSWEP), the China Meteorological Forcing Dataset, and the long-term, gauge-based gridded precipitation dataset for the Chinese mainland (CHMPRE). A key innovation is the integration of precipitation zoning, achieved by applying the Rotated Empirical Orthogonal Function method to delineate the upper and middle reaches of the Hanjiang River Basin (UMHRB) into distinct precipitation zones. We then systematically assessed the performance within these zones in terms of daily statistical accuracy, capacity to capture spatiotemporal evolution, and skill in detecting precipitation events and extremes. This study validates the applicability of dividing the study area into five precipitation zones, as determined from observed data (1975–2018). A key finding is CHMPRE is recommended as the most reliable product across all zones. In terms of precipitation event probability, MSWEP excels at reproducing the probability density distribution across various intensities, whereas CHMPRE tends to overestimate the frequency of non-precipitation events (< 0. 1 mm/d). Critically, a major discrepancy emerges in extreme precipitation detection: although all products show high capability at the watershed scale, their performance deteriorates markedly at the zonal scale. This was most pronounced for MSWEP, whose Kling–Gupta Efficiency for Rx1day dropped to as low as 0. 07 in Zone 5 and 0. 09 in Zone 3, indicating a failure to effectively capture extremes in these regions. This multi-faceted, zoning-level assessment provides a nuanced understanding of product performance, offering valuable insights for the informed selection of appropriate datasets tailored to specific hydrological and meteorological applications.
Bo et al. (Wed,) studied this question.