SPATIAL DOWNSCALING OF GPM IMERG V06 GRIDDED PRECIPITATION USING MACHINE LEARNING ALGORITHMS

Abstract. According to recent studies, Remote sensing data plays a significant role in filling gaps the poor gauge station, particularly at high elevations and with complex underlying surface features. In order provide high-resolution precipitation estimates over terrain areas, downscaling low-resolution satellite using various environmental variables. this paper, we tried downscale GPM IMERG V06 resolution of (0.1° × 0.1°) nearly 10km (1km 1 km) four machine learning algorithms namely, Decision Trees, Multiple Linear Regression, Support Vector Regressor random forest. Vegetation indices Normalized difference vegetation index (NDVI), Topography, Land Surface Temperature (LST), latitude longitude. This framework can 0.1° product km, by determining importance features, automatically optimizing model parameters. Additionally, ground recorded from rain stations have validated downscaled products. Spatial generally increase accuracy gridded results reveal that spatial is an acceptable way investigating Taiwan.