Evaluating the L-MEB forward radiative transfer model for the assimilation of SMOS observations

Abstract

Satellite observations play an important role in providing the initial conditions for the Numerical Weather Prediction (NWP) models. Satellite data are assimilated into the first estimate provided by NWP models using a radiative transfer model. The impact of satellite observations significantly depends on the accuracy of the simulation performed by the radiative transfer (RT) models. In recent years, there have been significant advances in RT modeling for microwave and infrared observations, which are not sensitive to the surface. However, in the case of sensitive surface observations such as Soil Moisture and Ocean Salinity (SMOS) satellite observations, the assimilation has been limited by inaccuracy in the forward calculations. This study investigates the accuracy of the L-band Microwave Emission of the Biosphere (L-MEB) RT model for SMOS frequencies using high-quality in-situ observations as input. The L-MEB model is the forward RT model used in the SMOS L2 algorithm, specifically developed to simulate brightness temperature (TB) over the land surfaces at different incidence angles (between 0° and 60°). The L-MEB model simulated the SMOS TB data with the horizontal (H) and vertical (V) polarization at the lowest SMOS incidence angles at the meteorological stations over Iran.The land cover at these stations is either bare soil or low vegetation. The comparison between simulated TB and the SMOS TB products showed a suitable RMSE and a relatively low bias for horizontally and vertically polarized channels. The relatively low bias can justify the assimilation of SMOS observations into the data assimilation systems. However, cross-comparison of the RT models used at the NWP centers and the RT models such as L-MEB, which were mainly developed to work with the SMOS data, is required to ensure that the operational RT models used at the NWP centers meet the same accuracy.