Downscaling and Calibration Analysis of Precipitation Data in the Songhua River Basin Using the GWRK Model and Rain Gauges

Obtaining high-quality precipitation data with both high spatial and temporal resolution is imperative for hydrological and meteorological research. However, the coarse resolution and uncertain data quality of most satellite data, coupled with sparse rain gauge station (RGS), limit their direct applicability in scientific research. Downscaling satellite data, particularly in conjunction with RGS, proves to be an effective approach to overcome this challenge. In this study, we utilize the geographically weighted regression kriging model to downscale global precipitation measurement IMERG monthly precipitation data from 2001 to 2020. Leveraging spatially heterogeneous relationships with digital elevation model, slope, land surface temperature, and soil moisture in the Songhua River Basin in Northeast China, we enhance the spatial resolution from 0.1 degrees to 1 km, initially achieving a 1.4% increase in data accuracy, with a CC value of 0.966. Subsequently, employing the daily fraction method, the downscaled precipitation data are disaggregated to the daily scale and calibrated by merging RGS using the geographical difference analysis method. The outcome is high-quality daily precipitation data with both high spatial resolution and accuracy (CC = 0.818, RMSE = 3.188, and ME = 0.086). An analysis of the annual variation of precipitation in the Songhua River Basin over the past two decades reveals an increasing trend. Spatially, the average annual precipitation variation rate in the basin increases from the middle to both ends, with the increasing trend gradually decreasing from south to north. The proposed approach provides a practical solution for enhancing the spatiotemporal scale of satellite data, improving data quality, and addressing the sparse distribution of RGS.