Single-footprint retrieval of clear-sky surface longwave radiation from hyperspectral AIRS data

Abstract

Surface longwave radiation (SLR) derived from remotely sensed data facilitates understanding of the SLR in global climate change. Hyperspectral infrared sounders aboard space platforms provide information on the surface and vertical structure of Earth’s atmosphere. However, currently, SLR products estimated from these observations are unavailable, which hampers their application potential for Earth’s radiation budget in the context of global warming. To address this issue, we developed simple and effective SLR model under clear-sky conditions using at-sensor spectral radiances from Atmospheric Infrared Sounder (AIRS). The model was found to be insensitive to AIRS instrument noise, and showed good performances based on a simulation dataset. The AIRS footprint geometrical model was proposed to match the AIRS and Moderate Resolution Imaging Spectroradiometer (MODIS) data to estimate the cloud fraction. Validation against ground-based measurements found that the surface upward longwave radiation model has a bias of 3.18 W/m2, root-mean-square error (RMSE) of 30.51 W/m2, and R2 of 0.84; the surface downward longwave radiation model has a bias of 0.77 W/m2, RMSE of 29.09 W/m2, and R2 of 0.78. The large validation biases at two ground sites reflect the limited spatial representativeness for AIRS footprints. Terrain-induced altitude differences and spatial inhomogeneity can redistribute the spatial distributions of SLR. Moreover, the model performances were weakly dependent on seasonal variation. The results indicate that the proposed model provides a foundation for the further development of operational SLR products.