Due to its high spatial variability and spectra, the microwave land surface emissivity calculation in the JCSDA Community Radiative Transfer Model (CRTM) (Han et al. 2005) is still one of the greatest challenges. The inaccurate surface emissivity results in large errors in CRTM simulated satellite brightness temperatures over land and rejection of satellite data in the NCEP Gridpoint Statistical Interpolation (GSI), especially for surface sensitive channels.

This study is continuing our efforts to improve the microwave land surface emissivity model (Weng et al. 2001) with more accurate land surface parameters, canopy optical parameters and an alternative vegetation dielectric constant calculation (Ulaby and El-Rayers, 1987; Dobson et al., 1985). We established sets of soil and vegetation parameters which are required for the new emissivity calculation, such as soil clay and sand fractions, as well as vegetation water content, specific density, leaf-area-index (LAI), and leaf-thickness. Second, based on the three-layer medium model, the more accurate formula of total upwelling radiance emanating from the surface is derived, considering impact of ground upwelling radiance which is important for low microwave frequency channels, especially for the desert and semi-arid regions. The enhanced model shows better improvement in a reduction of bias and root-mean-square error in simulated brightness temperature, as well as an increase in the number observations assimilated in the GSI, compared to the results when using a previous land surface emissivity scheme.