Downscaling TRMM satellite-based precipitation data using non-stationary relationships between precipitation and land surface characteristics

Satellite-based precipitation dataset has been widely used to estimate precipitation, especially over regions with sparse rain gauge networks. However, the low spatial resolution of these datasets has limited their application in localized regions and watersheds. So, having an accurate estimation of precipitation by satellites along with the adequate spatial scale in hydrologic studies is the main goal of this study. In this research, Geographically weighted regression (GWR) method was investigated to downscale the Tropical Rainfall Measuring Mission (TRMM-3B42 Version 7) over the DEZ river basin in the southwest of IRAN for 2010-2011. Downscaling was performed based on the non-stationary relationships between the TRMM precipitation and the Digital elevation model (DEM) derived products, the Normalized difference vegetation index (NDVI), the Enhanced vegetation index (EVI) and the Land surface temperature (LST). The result shows that the downscale precipitation at 1 km spatial scale had significantly improved spatial resolution, and agreed well with data from the rain gauge stations. For the 16-day precipitation, Mean square root means square error (RMSE) and absolute mean error (MAE) values are 22.7 mm and 7.45 mm, respectively. However, the accuracy of the model varies in a different location and depends on the vegetation condition.