Soil Moisture Downscaling from Remote Sensing Footprints

The study focuses on downscaling of soil moisture from coarse remote sensing footprints to finer scales. Two approaches are proposed for soil moisture downscaling. The first approach provides the probability distribution functions at the finer scales with no information about the spatial organization of soil moisture fields. The second approach implements a multiscale ensemble Kalman filter (EnKF) that assimilates remotely sensed coarse scale soil moisture footprint, attributes of fine scale geophysical parameters/variables (i.e., soil texture: %sand, vegetation: NDVI, topography: slope, and precipitation) and coarse/fine scale simulation into a spatial characterization of soil moisture evolution at the finer scales. To downscale the remotely sensed coarse scale soil moisture to another spatial scale, the multiscale EnKF uses a bridging model. The bridging model infers the pixel-specific scaling coefficient from the compatible geophysical parameters/variables that influence the soil moisture evolution process at that particular spatial scale. Data from diverse hydroclimatic regions from the semiarid Arizona, the agricultural landscape of Iowa, and the grassland/rangeland of Oklahoma are used in the study to implement the multiscale downscaling algorithm. The results demonstrate that the bridging model of multiscale EnKF helps to characterize the evolution of soil moisture within the remotely sensed footprint. Validation conducted at the finest scale also shows reasonable agreement between the measured field data and the simulated downscaled soil moisture evolution.