Tools for Building Spatial Dependence Structure of Extreme Wave Heights at Regionally Neighboring Ports

Multivariate normal distribution is widely used in the general statistical analyses. Extreme wave heights should be, if possible, examined according the extreme value theory, which is constructed by multivariate normal distribution, because it has great advantages like as the general normal statistics. In this study we clear the essential and useful properties of the model and propose a method of analysis as well as show the items in the toolbox: another simpler model, accordance index to grasp the dependencies, KS test of bivariate symmetry, Kendall distribution to check the bivariate dependency, etc. We remark the constraints also and demonstrate an example for the dependencies of annual maxima of wave height at the ports located along the pacific coast in Japan.