The longwave spectral emissivities of desert and vegetation are considerably different from blackbody emissivity. A dominant majority of current atmospheric GCMs still treat the surface as blackbody and ignore spectral variations of surface emissivity. Charney (1975) proposed a positive feedback in arid areas via interactions among solar radiation, surface albedo, and atmospheric motion. This leads us to postulate that similar feedback as in Charney (1975) could operate in the longwave and such longwave feedback might not be properly represented by current GCMs. We incorporate realistic surface spectral emissivity over the Sahara and Sahel regions, where the emissivity is as low as 0-6-0.7 over the IR window region, into the NCAR CESM v1.1.1, while keeping treatments for the rest of the globe unchanged. Both the standard and the modified CESM are then used to carry out a 10-year simulation with prescribed climatological SST. Compared to the standard CESM simulation, the mean surface radiative temperature in the modified CESM simulation increases by 1.6 K over the region. However, the net upward longwave flux at the top of the atmosphere is decreased by 2.33 Wm-2 because the low emissivity of desert leads to less longwave emission over the IR window region. Energy budget analysis shows that the atmospheric radiative cooling over the region is decreased by 1.33 Wm-2 in the modified CESM simulation. The changes in 500-hPa vertical velocities indicate in average enhanced descending motion over the region, result in suppression of convection, which in return enhances arid situation in the region. Our findings demonstrate that change in surface LW spectral emissivity can influence simulated climate in the Sahara and Sahel regions in a way, to some extent, similar to the mechanism proposed by Charney (1975).