Using neural networks to estimate wave and wind parameters from an HF radar

Abstract

High-frequency (HF) radar systems are used to provide measurements of waves, winds and currents from the sea. By the Doppler Effect the radars interpret the backscattered signal from the sea surface. This nformation is condensed in the Doppler Spectrum. From this spectrum, mathematical algorithms extract information such as wave height, wave direction or wave period. The neural networks are an experimental and theoretical method which was born in 1943 and it was always used for academic purposes (investigation, teaching). Its applications cover a wide range such as optimisation and scheduling, pattern recognition or even virtual reality. In this project, neural networks are used to estimate wave measurements from the Doppler Spectrum, in order to compare their accuracy with the existing mathematical algorithms.