Method For The Remote Sensing Retrieval Of Surface Longwave Radiation

This paper extended the method developed by Tang and Li (2008) with only regard to MODIS/Terra data and with surface altitude only lower than 2.5 km to estimate the atmospheric downward longwave radiation (RL,D). Two separate look-up tables’(LUT) coefficients corresponding to different altitudes and viewing zenith angles (VZAs) were developed for MODIS/Terra and MODIS/Aqua respectively, based on MODerate Resolution Imaging Spectroradiometer (MODIS) data and the atmospheric radiative transfer model MODTRAN4. The upward surface longwave radiation (RL,u) was estimated using the Stefan-Boltzmann law with MODIS land surface temperature/emissivity products (MOD11_L2/MYD11_L2). Combining the RL,D) and the RL,U, the paper proposed the method to estimate the net surface longwave radiation (RL,N) for any elevations.The RL,D and RL,N were validated using some in-situ measurements from two sites of Asian Automatic Weather Station Network Project (AAN) (Dunhuang and Amdo) and three sites of Environmental and Ecological Science Data Center for West China (Huazhaizi, Yingke and A’rou) in Heihe River Basin. In addition, the schemes developed respectively by Bisht et al. (2005) and Brutsaert (1975) were used to estimate the RL,D and RL,N for comparison with the proposed method. The RL,D results showed that the mean root mean square error (RMSE) of the extended method is 25.9 W/m2 for MODIS/Terra data and 25.7 W/m2 for MODIS/Aqua data. The RL,N results showed that the mean RMSE of the extended method is 30.2 W/m2 for MODIS/Terra data and 27.2 W/m2 for MODIS/Aqua data. These results indicated that the RMSEs of the proposed method are the lowest one among the three methods under clear days, and the proposed method can estimate the RL,D and the RLN accurately and feasibly for the high altitude area as the Tibetan Plateau region. Besides, the paper addressed the estimation of the RL,D and the RL,N for the Tibetan plateau region using the MODIS measurements for clear-sky conditions and analyzed the value distribution.