Modelling bivariate distributions of significant wave height and mean wave period

The use and scope of bivariate parametric probability distribution functions in the joint modelling of significant wave height and mean zero-upcrossing period are discussed. It is suggested that, for some applications, the calculation of probabilities can be made with kernel density estimates, instead of adopting parametric models. In view of the statistical variability that seems to exist in data sets from different years, an approach is proposed for long-term models of the sea-state parameters including the concept of random population. A bivariate density function is proposed for fitting significant wave height and mean zero-upcrossing period data jointly. Numerical results are given with data from Figueira da Foz. (C) 2002 Elsevier Science Ltd. All rights reserved.