Results from Assimilating AMSR-E Soil Moisture Estimates into a Land Surface Model using an Ensemble Kalman Filter in the Land Information System (LIS)

Soil moisture, because of its very high spatial variability and relative paucity of in situ observations, is well suited to take advantage of assimilation of remotely-sensed data. However, the limitations of microwave data for estimating soil moisture (shallow penetration, vegetation and surface roughness effects) complicate data assimilation efforts within land surface models. The goal of this study is to improve the representation of the soil moisture profile in the SHEELS (Simulator for Hydrology and Energy Exchange at the Land Surface) model by assimilating microwave satellite estimates of soil moisture based on X-band (10.65 GHz) brightness temperatures from the Advanced Microwave Scanning Radiometer-EOS (AMSR-E) sensor flying on the NASA Aqua satellite. These estimates represent soil moisture for only the top 1-2 cm of soil and are valid only over regions of sparse to moderate vegetation cover, but still may be useful for specifying surface boundary conditions in a modeling system. Further research is needed to better characterize the benefits to numerical weather prediction gained by assimilating AMSR-E soil moisture estimates. SHEELS is a land surface model tailored to accurately characterize soil moisture and temperature profiles, and is designed to facilitate assimilation of soil moisture data. The number and depth of soil layers is user-defined, permitting higher vertical resolution near the surface where temperature and moisture gradients are large. We have integrated SHEELS into the Land Information System (LIS), a framework for running land surface models from a common interface using common inputs (atmospheric model outputs or base forcing, and observational data). In LIS, we have coupled SHEELS with the Weather Research and Forecasting (WRF) model. LIS includes an Ensemble Kalman Filter (EnKF) data assimilation algorithm. We have added AMSR-E soil moisture as a LIS observation type, which enables us to assimilate soil moisture observations with the EnKF algorithm. We will present results showing the impact of assimilating AMSR-E surface soil moisture estimates on the modeled soil moisture and temperature profiles, and validate against in situ observations from the Little Washita River Basin Micronet and the Oklahoma Mesonet for a period in summer 2003.