Ocean Wave Inversion from HF Radar with BP Neural Network

The Doppler spectra scattered from HF radar contains abundant ocean surface dynamic parameters information, such as ocean surface wind, wave, current and water depth. In this paper, the principle of ocean wave information extracted from HF radar echo is present, and a new method for the significant wave height inversion from HF radar Doppler spectra with back propagation neural network (BPNN) model is proposed to inverse the significant wave height from HF radar sea echo which is designed with three layers (5 input nodes, 10 hide nodes and 1 output node included), and can fit the complicated nonlinear relationship between ocean wave and HF radar echo. Aim to verify the performance of the proposed BPNN inversion model, observation data from HF radar and buoy comparison experiment is done, the coefficient is 0.89 and the root mean square error is 0.25 m. The result shows that the proposed BPNN inversion method is effective and accuracy, which provides a new means for HF radar ocean wave detection engineering application.