Land surface model calibration through microwave data assimilation for improving soil moisture simulations

Soil moisture is a key variable in climate system, and its accurate simulation needs effective soil parameter values. Conventional approaches may obtain soil parameter values at point scale, but they are costly and not efficient at grid scale (10-100 km) of current climate models. This study explores the possibility to estimate soil parameter values by assimilating AMSR-E (Advanced Microwave Scanning Radiometer for Earth Observing System) brightness temperature (TB) data. In the assimilation system, the TB is simulated by the coupled system of a land surface model (LSM) and a radiative transfer model (RTM), and the simulation errors highly depend on parameters in both the LSM and the RTM. Thus, sensitive soil parameters may be inversely estimated through minimizing the TB errors. A crucial step for the parameter estimation is made to suppress the contamination of uncertainties in atmospheric forcing data. The effectiveness of the estimated parameter values is evaluated against intensive measurements of soil parameters and soil moisture in three grasslands of the Tibetan Plateau and the Mongolian Plateau. The results indicate that this satellite data-based approach can improve the data quality of soil porosity, a key parameter for soil moisture modeling, and LSM simulations with the estimated parameter values reasonably reproduce the measured soil moisture. This demonstrates it is feasible to calibrate LSMs for soil moisture simulations at grid scale by assimilating microwave satellite data, although more efforts are expected to improve the robustness of the model calibration. (C) 2015 Elsevier B.V. All rights reserved.