Downscaling algorithms for annual TRMM data based on climatic and orographic variables over the Qinling Mountains, China

Abstract

Obtaining the gridded precipitation data with a high resolution in mountainous area is of importance in hydrology, meteorology, and ecology. However, rain gauge observations and satellite-based precipitation products have its own shortcomings. Precipitation in mountainous area has correlation with variables like elevation, slope, and temperature. In this study, we applied a downscaled algorithm called Geographically Weighted Regression (GWR) to obtain a fine resolution (1 km) gridded precipitation data from the Tropical Rainfall Measuring Mission (TRMM) data at 0.25° resolution based on an assumption that precipitation in mountainous area has correlation with some orographic factors (elevation, slope, and aspect) and climatic factors (temperature, wind velocity, and humidity). The results indicated that (1) GWR improved the accuracy of TRMM data in the Qinling Mountains (r = 0.86, BIAS = − 2.77%, and RMSE = 93.24 mm for annual downscaled precipitation during 2013–2015 periods, and r = 0.71, BIAS = − 3.60%, and RMSE = 99.31 mm for annual TRMM data during 2013–2015 periods). (2) GWR showed a good performance in the southern part of the Qinling Mountains, while it showed a worse performance in the northeast part of the Qinling Mountains. (3) Not only orographic factors but climatic factors were all essential in downscaling precipitation in mountainous areas. The more input factors, the more accurate downscaled result derived from GWR.