Downscaling SMAP Radiometer Soil Moisture over the CONUS using Soil-Climate Information and Ensemble Learning

Soil moisture contributes significantly towards the improvement of weather and climate forecast and understanding terrestrial ecosystem processes. It is known as a key hydrologic variable in the agricultural drought monitoring, flood modeling and irrigation management. While satellite retrievals can provide an unprecedented information on soil moisture at global-scale, the products are generally at coarse spatial resolutions (25-50 km2). This often hampers their use in regional or local studies, which normally require a finer resolution of the data set. This work presents a new framework based on an ensemble learning method while using soil-climate information derived from remote-sensing and ground-based observations to downscale the level 3 daily composite version (L3_SM_P) of SMAP radiometer soil moisture over the Continental U.S. (CONUS) at 1 km spatial resolution. In the proposed method, a suite of remotely sensed and in situ data sets in addition to soil texture information and topography data among others were used. The downscaled product was validated against in situ soil moisture measurements collected from a limited number of core validation sites and several hundred sparse soil moisture networks throughout the CONUS. The obtained results indicated a great potential of the proposed methodology to derive the fine resolution soil moisture information applicable for fine resolution hydrologic modeling, data assimilation and other regional studies.