Calculation Method for Evaporation Duct Profiles Based on Artificial Neural Network

The profile of modified refractive index is crucial for the investigation of the evaporation duct phenomenon. Previous studies have indicated that several similarity functions in Monin-Obukhov similarity theory may be unsuitable for modeling fluxes under stable conditions. Therefore, a flexible scheme for the calculation of the M profile is necessary. This study proposes a numerical profiling method that adopts the artificial neural network and training data from the NCEP CFSR meteorological dataset and the NPS evaporation duct model. Profiling and path loss results are compared when training with air-sea temperature difference (ASTD) < 0 and ASTD > 0, respectively. The proposed method can be applied based on data characteristics instead of Monin-Obukhov similarity theory. Hence, it may be a computationally efficient and promising method for future applications.