A New Analysis Method for Extreme Wave Statistics by Poisson-GPD Model

Description

This paper describes a new mothod for a frequency analysis of extreme wave heights, by employing the generalized Pareto distribution (GPD) assuming Poisson process of event occurence. A principal advantage of using GPD is the connectivity to the asymptotic extreme value distribution, whose shape parameter’s value is identical to that of GPD. The inference method is based on the maximum likelihood for estimation of return wave heights and a Goda’s spread parameter, and the confidence intervals are due to the delta method. A wave data set hindcasting off Kodiak is applied as an example. The present analysis shows that extreme value of the sample follows the FT-III distribution, which is not employed as a candidate distribution in the previous studies.

Journal

Details 詳細情報について

Report a problem

Back to top