In this study, we present the inversion of land surface emissivity (LSE) in the China land area based on the mid-infrared and thermal infrared channel day/night data received by the FengYun-2G (FY-2G) geostationary meteorological satellite. The atmospheric correction of the FY-2G data is based on MODTRAN, which is a radiation transfer model, using the atmospheric profile data, including data with respect to the temperature, humidity, and ozone, provided by the European Centre for Median-range Weather Forecast (ECMWF). The modified Minnaert’s bidirectional reflection distribution function model is used to calculate the hemispherical reflectivity of the mid-infrared channel in the surface direction. Further, the LSE distribution characteristics can be obtained with respect to the different surface vegetation types in China based on the temperature-independent thermal infrared spectral index. Finally, the LSE inversion results obtained from the images of China’s land area are verified by selecting the surface emissivity products from Moderate Resolution Imaging Spectroradiometer (MODIS). The daytime and nighttime results are observed to be in good agreement with the existing LSE products in MODIS. The absolute LSE inversion errors of the hot infrared channels 1 and 2 during daytime are -0.0057 and -0.0068, respectively, and the root mean square errors (RMSEs) are 0.0095 and 0.0103, respectively. Furthermore, the absolute LSE inversion errors of the hot infrared channels 1 and 2 during nighttime are -0.0010 and -0.0035, respectively, and the RMSEs are 0.0094 and 0.0096, respectively. Simultaneously, the LSE inversion results at different time slots on the same day are analyzed. These results indicate that the LSE at nighttime is lower than that at daytime.