Downscaling of satellite-based precipitation considering the spatially heterogeneous relationship between precipitation and environmental variables

The satellite-based precipitation products are one of the sources of rainfall estimation. Nonetheless, for usage in the local regions and, or for parameterizing of meteorological and hydrological models at basin scales, their spatial resolution is often coarse. Therefore, in this study, a downscaling– calibration method was developed for global precipitation measurement (GPM) satellite estimates (at 0.1° spatial resolution), for one year from 01/04/2014 to 31/03/2015, by considering the spatial heterogeneity of the relationship between precipitation and the environmental variables using the mixed geographically weighted regression (MGWR) model for Golestan province. In obtaining improved precipitation data with 1 km spatial resolution at an annual scale, the results showed that (1) the proposed method not only improved the spatial resolution of precipitation but also increased accuracy; (2) the downscaled and calibrated precipitation data (CC = 0.74, bias = 0.23) performed better than the original data (CC = 0.58, bias = 0.35) against ground observations.