Assimilation of ASCAT Soil Moisture and SMAP Brightness Temperature Observations into the NASA GEOS Land Data Assimilation System

Abstract Soil moisture (SM) observations from the Advanced Scatterometer (ASCAT) satellite radar (5.3 GHz) and brightness temperature (Tb) observations from the Soil Moisture Active Passive (SMAP) satellite radiometer (1.4 GHz) are assimilated in the NASA Goddard Earth Observing System (GEOS) land data assimilation system, both separately and together (jointly) from April 2015 to March 2021. The resulting SM estimates are validated using in-situ measurements, independent satellite observations, and data assimilation diagnostics. Assimilating only ASCAT SM (ASC_DA) universally improves the SM analysis estimates relative to a model-only control simulation (CNTL). For example, the anomaly time series correlation coefficient (anomR) vs. in-situ surface SM increases from 0.55 for CNTL to 0.58 for ASC_DA, and the misfit between SMAP Tb observations (not assimilated) and corresponding (3-hour) background forecasts decreases by 4.9%. Assimilating only SMAP Tb (SMP_DA) yields greater improvements in SM analysis estimates than does ASC_DA. For example, anomR vs. in-situ surface SM increases to 0.68, although the misfit between ASCAT SM observations (not assimilated) and corresponding (3-hour) background forecasts decreases by only 1.5%. Jointly assimilating multi-sensor (ASCAT and SMAP) observations (MLT_DA) yields overall SM estimation skill similar to that of SMP_DA. Data assimilation diagnostics suggest that MLT_DA background forecasts are generally improved vs. those of CNTL, but by less than seen in SMP_DA, implying that information from the ASCAT and SMAP observations does not always agree. However, since ASC_DA clearly improves SM estimates, multi-sensor assimilation is nevertheless beneficial by increasing system robustness and extending the period when SM observations are available for assimilation.