Comparison of Bivariate Models of the Distribution of Significant Wave Height and Peak Wave Period

This paper presents the results of a comparison of the fit of three bivariate models to a set of 14 years of significant wave height and peak wave period data from the North Sea. One of the methods defines the joint distribution from a marginal distribution of significant wave height and a set of distributions of peak period conditional on significant wave height. Other method applies the Plackett model to the data and the third one applies the Box-Cox transformation to the data in order to make it approximately normal and then fits a bivariate normal distribution to the transformed data set. It is shown that all methods provide a good fit but each one have its own strengths and weaknesses, being the choice dependent on the data available and applications in mind.