Downscaling of Global Soil Moisture using Auxiliary Data

Soil moisture is important to land surface modeling and climate modeling, which usually use soil moisture as a critical parameter. Derivation of soil moisture by radar remote sensing has theoretically and practically proven to be possible. However, radar-derived soil moisture is at coarse resolution, nominally about 25 kilometers, which does not satisfy the requirements of models using higher resolution grids. It is desirable to downscale the soil moisture to resolutions finer than 25 kilometers, especially 1 to 5 kilometers. In this study, derived parameters from MODIS have been used to derive the correlation between soil moisture and these parameters and downscale soil moisture. Six downscaling algorithms were proposed and compared. Geographically weighted regression (GWR) was used as the base model. Results showed that GWR performed well in downscaling. Further studies would look into more parameters for base models and higher-order polynomial regression for improving the accuracy of soil moisture downscaling.