Research on the Application of the Radiative Transfer Model Based on Deep Neural Network in One-dimensional Variational Algorithm

As a typical physical retrieval algorithm for retrieving atmospheric parameters, one-dimensional variational (1DVAR) algorithm is widely used in various climate and meteorological communities and enjoys an important position in the field of microwave remote sensing. Among algorithm parameters affecting the performance of the 1DVAR algorithm, the accuracy of the microwave radiative transfer model for calculating the simulated brightness temperature is the fundamental constraint on the retrieval accuracies of the 1DVAR algorithm for retrieving atmospheric parameters. In this study, a deep neural network (DNN) is used to describe the nonlinear relationship between atmospheric parameters and satellite-based microwave radiometer observations, and a DNN-based radiative transfer model is developed and applied to the 1DVAR algorithm to carry out retrieval experiments of the atmospheric temperature and humidity profiles. The retrieval results of the temperature and humidity profiles from the Microwave Humidity and Temperature Sounder (MWHTS) onboard the Feng-Yun-3 (FY-3) satellite show that the DNN-based radiative transfer model can obtain higher accuracy for simulating MWHTS observations than that of the operational radiative transfer model RTTOV, and also enables the 1DVAR algorithm to obtain higher retrieval accuracies of the temperature and humidity profiles. In this study, the DNN-based radiative transfer model applied to the 1DVAR algorithm can fundamentally improve the retrieval accuracies of atmospheric parameters, which may provide important reference for various applied studies in atmospheric sciences.