Abstract: Soil moisture is a critical variable that links climate dynamics with water and food security. The most feasible way to continuously obtain soil moisture spatially explicit information across large areas is using satellite-based microwave remote sensing. There are however limitations of satellite-based soil moisture: 1) information is relevant for the first centimeters of top-soil surface; 2) they have coarse granularity; and 3) there are spatial information gaps. This talk focuses on how hydrologically meaningful terrain parameters and machine learning algorithms could be combined to improve the statistical accuracy and granularity of satellite-derived soil moisture. These analyses will be situated at the local, continental and global scales