Wave Height Mapping From HF Radar-Measured Surface Current Data Based on a Cascaded LSTM Network

Abstract

High-frequency (HF) surface wave radar has been widely used for measuring sea surface current, but its accuracy and spatial coverage of sea wave measurement are limited. A wave height mapping method based on HF radar-measured surface currents is proposed in this article. The tidal currents are firstly estimated and subtracted from the radar-measured surface currents, and then the geostrophic currents are removed. The remaining currents, along with the wind directions, are used as input data to a cascaded long short-term memory (LSTM) network. This network contains two stages. The first stage divides the input data into two groups of high and low sea states, and the second stage performs wave height inversion for each group separately. By combining the inversion results of the two groups, wave height maps are obtained. The algorithm is verified using the experimental data from the southwestern Taiwan Strait. Using the numerical wave height fields as the ground truth, the proposed algorithm has a root-mean-square error (RMSE) of 0.35 m and a correlation coefficient (CC) of 0.90.