Benchmarking of L-band soil microwave emission models

A first step before assimilating Soil Moisture and Ocean Salinity (SMOS) L-band brightness temperatures (Tb) over land is to couple land surface models (LSMs) to microwave emission models. In this study, the Interactions between Soil Biosphere and Atmosphere (ISBA) LSM is coupled to the Community Microwave Emission Modelling platform (CMEM). Simulations of Tb are performed over a 3-yr period (2003-2005) for a bare soil field in southwestern France, at the Surface Monitoring Of the Soil Reservoir EXperiment (SMOSREX) experimental site. Both ISBA and CMEM present several options for the representation of the soil moisture and soil temperature profiles. Simplified 2-layer simulations are compared with more detailed multilayer simulations. In the 2-layer simulations, the soil is divided into two layers (a thin surface layer and a bulk reservoir), and Fresnel laws are used in CMEM to model the smooth surface emissivity. In the multilayer simulations, the ISBA soil diffusion scheme is used (with 11 soil layers represented) together with either the Wilheit or Fresnel option of CMEM. The Tb simulations are compared to the Tb ground observations available for the SMOSREX site, at H and V polarizations and at different angles, and the impact of soil roughness is assessed. It is shown that taking surface soil moisture into account in the calculation of soil roughness improves the representation of the seasonal cycle and increases the correlation for all the versions of CMEM. The Tb derived from the most complex multilayer simulations correlate slightly better to the observations than the Tb derived from the 2-layer model (r = 0.84 and r = 0.82, respectively, for the pooled dataset). This is partly due to a better representation of the soil moisture profile. Finally, the multilayer model is used to investigate the L-band sampling depth for contrasting soil texture profiles. For a variety of soil textures, it is found that Tb are mainly driven by the top 15 cm soil layer.