Copula-based joint tropical cyclone-induced wind and wave risk analysis: considering the effect of uncertainty using Bayesian inference

The present study introduces a joint probability analysis method for assessing the risk of tropical cyclone (TC)-induced wind and wave under uncertainty. This method integrates copula functions, log-transformed kernel density estimation (log-KDE), and Bayesian inference based on Markov Chain Monte Carlo (MCMC) simulation into a comprehensive framework, known as the MCMC-KDE-based copula analysis (MKCA) method. The MKCA method is effective in modelling the bivariate characteristics of TC-induced wind and wave while accurately quantifying the inherent uncertainty in the copula parameters. Then, MKCA is employed to evaluate the TC risk for a prospective offshore site within the South China Sea. The results show that log-KDE accurately models the marginal distributions of wind and wave, while the Gaussian copula is suitable for characterizing the dependence structure between them. Additionally, the findings demonstrate a considerable degree of uncertainty when predicting joint return periods and design scenarios with limited data availability. As the predicted return periods increase, the uncertainty ranges also expand. The findings provide valuable decision support for TC risk mitigation and offshore structure design.