Evaluation of modeled microwave land surface emissivities with satellite‐based estimates

An accurate estimate of the microwave surface emissivity is necessary for the retrieval of atmospheric quantities from microwave imagers or sounders. The objective of this study is to evaluate the microwave land surface emissivity modeling of the Community Radiative Transfer Model (CRTM), providing quantitative statistic information for further model improvements. First, the model-simulated emissivity is compared to emissivity estimates derived from satellite observations (TELSEM, Tool to Estimate Land Surface Emissivities at Microwaves). The model simulations agree reasonably well with TELSEM over snow-free vegetated areas, especially at vertical polarization up to 40 GHz. For snow-free surfaces, the mean difference between CRTM and TELSEM emissivities at vertical polarization is lower than 0.01 below 40 GHz and increases to 0.02 at 89 GHz. At horizontal polarization, it increases with frequency, from 0.01 at 10.6 GHz to 0.04 at 89 GHz. Over deserts and snow, larger differences are observed, which can be due to the lack of quality inputs to the model in these complex environments. A further evaluation is provided by comparing brightness temperature (Tbs) simulations with AMSR-E observations, where CRTM emissivity and TELSEM emissivity are coupled into a comprehensive radiative transfer model to simulate the brightness temperatures, respectively. The comparison shows smaller RMS errors with the satellite-derived estimates than with the model, despite some significant bias at midday with the satellite-derived emissivities at high frequencies. This study confirms and extends to the global scale previous evaluations of land surface microwave emissivity model. It emphasizes the needs for better physical modeling in arid regions and over snow-covered surfaces.