多源降水产品驱动的喀斯特中小流域日径流模拟OA

Data scarcity is a major challenge for the development,utilization,and protection of water resources in mountainous watersheds.Precipitation products generated via multi-source data fusion offer advantages such as long temporal coverage and high spatial resolution.Coupling these products with hydrological models provides a feasible approach to alleviating this problem in the short term.Taking the typical small and medium-sized watershed of the Dumu River Basin(in the Wujiang River system,karst region)as the study area,a distributed hydrological model SWAT as the modeling tool,and observed data(OBS)together with three precipitation products(MSWEP,ERA5,CHM_PRE)as model drivers,a systematic analysis was conducted from three perspectives:multi-scale evaluation of runoff simulation accuracy,consistency between precipitation accuracy and runoff simula-tion accuracy,and parameter sensitivity characteristics and rationality.The results showed that:① The daily runoff simulations driven by the three precipitation products generally achieved good accuracy,with significantly better performance during the rainy season from May to October than during the non-rainy season.CHM_PRE showed greater stability across different months and a-chieved the best overall performance,whereas MSWEP performed poorly during the non-rainy season.② The runoff simulation accuracy driven by the three precipitation products was generally consistent with the accuracy of the corresponding precipitation series.However,when compared across different precipitation intensity classes,runoff simulation accuracy was higher than precipi-tation series accuracy.This is because runoff is influenced by the integrated effects of antecedent hydrological processes,whereas precipitation events are relatively independent.③ When the hydrological model was driven by different precipitation products,more than 50%of the parameters reaching the significance level remained stable,indicating that all products could reasonably re-present key hydrological processes.Among the inverse parameters from model optimization,those derived from OBS and CHM_PRE forcing were more reasonable.The findings provide a basis for the selection and fusion of precipitation products and offer a reference for addressing the shortage of hydrological data in mountainous regions.