A0353
Title: Bayesian mixture of spectral density functions for ocean waves
Authors: Matt Moores - University of Wollongong (Australia) [presenting]
Sankalpa Fonseka - University of Wollongong (Australia)
Jeff Hansen - University of Western Australia (Australia)
David Gunawan - University of Wollongong (Australia)
Abstract: Ocean waves are a fundamental part of the environment and play an important role in the physical and biological processes of the oceans. The distribution of energy versus wave frequency, known as a wave spectrum, provides a way to quantify the behaviour of ocean waves. Accurate estimation of the physical parameters, such as wave height, peak frequency, and velocity, is crucial for a wide range of applications, including offshore engineering, coastal management, and maritime operations. A mixture of spectral density functions is introduced to model ocean wave spectra with multiple peaks, i.e., wind seas and swell. Informative priors for the parameters are defined using expert elicitation. Posterior samples are obtained using Hamiltonian Monte Carlo. Simulation-based calibration is employed to quantify the accuracy, consistency and coverage of the estimates. The model is fit to ocean buoy measurements from King George Sound in Western Australia.
