Modeling asymmetrically dependent multivariate ocean data using truncated copulas

Characterizing multivariate ocean parameters is quite important for offshore engineering reliability design and risk assessment. To fully understand ocean conditions, a robust and accurate multivariate model is essential for the analysis and estimation of the ocean state. Therefore, advanced simulation of the ocean parameters helps to improve practices in offshore engineering. In this work, the principle of a new type of copula, namely truncated copula, is developed and adopted for modeling the multivariate ocean data. Unlike previous studies on modeling asymmetric ocean data by purely mathematical fitting techniques, this study proposes a truncated method based on physical limits to study asymmetrically dependent ocean data. The truncated copula method is contrasted with the conventional symmetric and existing asymmetric copula from the literature using real environmental observations for the demonstration. Various commonly used traditional copula models are modified by the proposed truncation technique and applied to fit multivariate ocean data collected in buoys off the US coast. Based on the fitting of ocean data, this paper compares the advantages and disadvantages of different copula models. The properties of different copula models for data simulation and extreme value prediction are also discussed.