Enhanced Moisture Retrieval Near Boundary Layer From Satellite Sounder Data Through Atmospheric-Surface Radiance Separation

Abstract

Accurate satellite-based retrieval of boundary-layer water vapor over land is crucial for understanding the Earth-atmosphere system but remains challenging due to the interaction of surface parameters on low-level atmosphere-sensitive channels. This study proposes a novel approach to explicitly extract the upwelling atmospheric radiance ($R_a$) from the total radiance ($R_t$) observed by the Infrared Atmospheric Sounding Interferometer (IASI), using a Residual Multi-Layer Perceptron model. A modified one-dimensional variational algorithm for surface-free radiances is also developed. The radiance extraction model, trained on simulated IASI radiances, is applied to IASI observations over mainland Australia in January and February of 2022. Validated against ERA5 and radiosonde observations, compared with the traditional $R_t$-based method, the $R_a$-based atmospheric profile retrievals show distinct improvements in boundary-layer humidity retrieval with at least 20% error reduction. This approach provides a new thought to enhance humidity retrievals from hyperspectral sounders and benefits other quantitative applications such as data assimilation.