Auto-Encoder Based Inversion for Inhomogeneous Evaporation Ducts from Sea Clutter

Abstract

Inverting the modified refractivity profile of the evaporation duct from radar sea clutter has prominently pragmatic utility and widespread industry prospects. Notwithstanding, the modified refractive index (M) profile structure of the evaporation duct will exhibit the inhomogeneous characteristics in the range direction as the multifarious meteorological conditions over the sea, which will seriously affect the tangible detection performance of radar. Numerous attempts have been made to ameliorate the accuracy of the inverted parameters of evaporation duct profiles by utilizing various intelligent models. Nevertheless, modeling of the range direction inhomogeneous evaporation duct, conventional modeling approach of extracting the dimensional features of the high dimension M profile is not befitting the feature extraction where the range direction is persisting nonlinear relation. To tackle this issue, we propose a novel model, namely one-dimension residual convolutional autoencoder, to extract features from inhomogeneous evaporation ducts and to overcome the dimension issue inherent in modeling the range direction parameters of evaporation duct profiles. Experimental results demonstrated that the proposed approach is more befitting and efficacious for evaluating the range direction dependent inhomogeneous evaporation duct M profile.