Optimised Wave-by-Wave Prediction of Spread Waves: Comparison With Field Data

Real-time forecasting of ocean surface waves can be beneficial for many offshore operations such as ship navigation, load mitigation on floating offshore wind turbines, and control of wave energy converters. This study validates a linear, algebraic prediction model for sea states with large directional spreading. The model decomposes observed time histories using a Fourier transform to obtain an approximate representation of the wave field using a small number of directional components. Pre-processing involves the partial removal of nonlinear harmonics by a bandpass filter followed by the attachment of a NewWave-type signal at each end of each record. The model is tested using field data measured using a small array of wave buoys deployed in the Southern Ocean off Albany, Western Australia. It shows good agreement between prediction and target time series. Aggregating and weight-averaging multiple predictions obtained with different sets of optimal representative directions improves the quality of prediction. Based on the linear propagation of representative directional Fourier components, the model is relatively robust to the presence of (unfilterable) higher harmonics and fast enough for real-time predictions.