Combining Thermal and Microwave Satellite Sensor Observations for a Moderate Resolution Soil Moisture Data Product

Many societal applications needs soil moisture data products with moderate to high spatial resolution and reliable numerical accuracy. The thermal infrared observations from geostationary satellite sensors (e.g., GOES Imager and GOES-R ABI) could be used to generate 2-12km resolution soil moisture estimates. Soil moisture retrievals from passive microwave satellite radiometer (e.g. AMSR-E, SMOS and SMAP) soil moisture observations could be accurate but their spatial resolutions are coarse (~40km). To take advantages of these two types of operationally available satellite observations for a moderate resolution soil moisture data product, this study investigates the relative accuracy of the GOES-based soil moisture proxy and the AMSR-E/SMOS soil moisture retrievals using in situ soil moisture observations. The heterogeneity of the GOES-based soil moisture proxy within each AMSR-E pixels of the CONUS domain is evaluated for temporarily interpolating the proxy data. A simple algorithm for down-scaling the relatively more accurate AMSR-E/SMOS soil moisture retrievals (at 25km scale) using GOES soil moisture estimates (at 1/3 of 25km scale) is then examined against in situ soil moisture measurements. The potential of using the algorithm for operationally generating moderate to high resolution soil moisture data products is discussed.