Modification and Application of the Satellite-Based Land Data Assimilation Scheme for Very Dry Soil Regions Using AMSR-E Images: Model Validation for Mongolia-a CEOP data platform

Abstract

At climatic time-scales, soil moisture is one of the most important boundary condition controlling fluxes to the atmosphere. Here, we explore the feasibility of synthesizing distributed fields of soil moisture using AMSR-E observations and a novel application of data assimilation within a hydrological model. We modified our existing Land Data Assimilation Scheme (LDAS) by specifically considering: (1) weak constraint assumptions rather than a strong constraint, thus accounting for the existence of model errors; and (2) the effects of volume scattering within the soil medium in the Radiative Transfer Model (RTM). Adopting the “effects of volume scattering within the dry soil medium” in the RTM is a new step for satellite-based data assimilation techniques for the retrieval of soil moisture.This LDAS can be used to assess model parameters and estimate vertical profiles of soil moisture, especially in very dry regions (volumetric soil moisture is equal to or less than 5-15%) as well as soil-surface and canopy temperatures by comparing passive microwave observations using a unique minimization technique termed Very Fast Simulated Annealing (VFSA). To validate our new LDAS, AMSR-E observations, gathered in Mongolia, were assimilated into the Land Surface Model (LSM) via the modified RTM. Observed micrometeorology boundary conditions for Mongolia were drawn from the CEOP database. In studies that simulate 2-week dry periods, the results of the revised LDAS are in better agreement with observational data than the results of open-loop simulations.