In this study estimation of extreme sea level from observational data is attempted. Type III asymptotic distribution is employed to achieve that purpose. Application of type III asymptotic distribution to the data requires the determination of distribution parameters. A nonlinear regression method, a skewness method, a maximum likelihood method, and a multinomial discretization method are used to evaluate the parameters. Ochi proposed modified type III asymptotic distribution based on the nonlinear multiple regression method. In the present study the authors applied that scheme to the above mentioned 4 methods. The results show good agreements with the observed data.
Extreme values from observational data are the information of special importance in several areas of engineering applications. In the field of ocean engineering, wave height is main factor to be considered for various design purposes. This paper discusses the methods of statistical estimation of extreme significant wave height which may be encountered for a certain return period. Gumbel(1958) had classified the asymptotic distributions of the extremes as the type I, II, and III asymptotic forms. Ochi(1986) has shown that the type I distribution may yield an increasingly overestimation of the extreme value with increasing variate values. In this study Type III asymptotic distribution is intensively applied to observational data. Estimation methods of distribution parameters must be follow up subjects in extreme value theory. These are the maximum likelihood method, the skewness method, and the nonlinear regression method. In this study, the multinomial discretization method is newly employed in the field of extreme sea level estimation. The calculations using these methods do not give satisfactory results. Ochi(1986) proposed a newly developed modified type III asymptotic extreme value distribution which yields an excellent fit over the entire variate range of the cumulative distribution.