Spatiotemporal nonlinear ocean wave prediction with uncertainty quantification and universal predictable zone determination

Accurate spatiotemporal prediction of ocean waves is critical for the efficient operation of offshore renewable energy devices. However, most existing wave prediction methods rely on linear or weakly nonlinear wave assumptions, limiting their applicability in complex sea states. Additionally, current approaches for determining the predictable zone are based on linear assumptions, often resulting in either overly conservative or unrealistic predictions. To address these limitations, in this work, spatiotemporal, phase-resolved prediction of nonlinear ocean waves is investigated, based on variational Bayesian neural network. More importantly, a universal predictable zone determination approach is proposed that, for the first time, enables a prior identification of the predictable zone across different sea states without relying on linear assumptions. The predictive performance of the proposed method is then evaluated using both simulation data based on the higher order spectral method and experimental wave tank data. The results show that the proposed method can achieve accurate wave predictions by using upstream or local past measurements, and the proposed predictable zone exhibits stability and universality across linear and nonlinear sea states. By overcoming the limitations of existing methods, which all fail in strongly nonlinear conditions, this work opens new opportunities for improving the control and monitoring of wave energy converters and floating wind turbines.