A time-varying copula approach for describing seasonality in multivariate ocean data

Characterizing multivariate ocean variables is quite critical for reliability design and risk assessment of marine structures. A robust, precise, and practical multivariate statistic model is necessary for comprehending ocean characteristics. As the time-varying characteristics exist in the ocean data, it is unreasonable to employ a simple constant statistical model to characterize all the multivariate data at one time. Therefore, in this paper, a time-varying copula approach is developed for modeling time-varying multivariate ocean data. Considering climate variations, a time-varying formula for return period and environment contour is also derived. The developed approach is demonstrated based on a site-specific ocean dataset collected from a buoy on the US coast. The climate effects associated with the multivariate ocean variables are characterized. The developed time-varying copula approach is also compared to the conventional copula and the conditional model in estimating the return period. The results showed that the time-varying model is helpful to explore the most critical environmental conditions for marine structures.