Global 1km Soil Moisture derive from SMAP Using Gap-Filled Satellite Land Surface Temperature Data

Soil moisture (SM) is a key factor for numerical earth science applications, including hydrological, meteorological, agricultural and environmental fields. The remotely sensed SM data products have been provided by a few satellites, especially the Soil Moisture Active Passive (SMAP) satellite, which can be used to retrieve SM from L-band passive microwave radiometer observations at a native spatial resolution of 36 km. A downscaling algorithm based on the vegetation condition modulated thermal inertia relationship between SM and land surface temperature (LST) change was developed using LDAS (Land Data Assimilation System) model output variables and implemented on the 1 km MODIS (Moderate Resolution Imaging Spectroradiometer) LST and NDVI. However, the MODIS LST retrievals derived from optical and thermal bands are often restricted by the cloud cover issue. The Advanced Microwave Scanning Radiometer 2 (AMSR2) onboard GCOM-W satellite provides passive microwave radiometer brightness temperature (TB) observations at 10 km resolution for all-weather conditions. Therefore, the AMSR2 TB at 36.5 GHz, v-polarization were acquired to build linear regression model with MODIS LST on time scale for each month and the model predicted LST were used to gap fill the original MODIS LST data and downscale the SMAP SM at 1 km. The results show that the downscaled SMAP SM product significantly have higher coverage than the original downscaled product, especially in the tropical/sub-tropical watersheds, e.g., Lower Mekong River Basin and Amazon River Basin. The validation results for the downscaled SM derived from the gap-filled LST also show a good correlation with the ISMN (International Soil Moisture Networks) in situ measurements.