Deep learning for solving inversion problem of atmospheric refractivity estimation

Atmospheric ducts are typical occurrences in marine environments. They can trap electromagnetic waves in ducts layer and extend propagation ranges significantly. Thus, affecting systems such as radar communications. Because the direct measurement of the atmospheric ducts refractivity has a certain degree of difficulty, the radar sea clutter power, which is easily measured, is used to solve the inversion problem of atmospheric refractivity estimation. In this study, based on the refractivity profile of the evaporation duct and the surface based duct, combined with the deep learning, we established a network mapping model between the sea clutter and the refractivity profile parameters. The model is applied to the inversion problem of atmospheric refractivity estimation, and the inversion results are analyzed to verify the feasibility of deep learning in the inversion problem. We herein report the high-precision inversion results of the atmospheric refractivity estimation.