Design wave heights are estimated stochastically by calculating the probability of significant wave heights occurring within a specific return period, utilizing observed or hindcast wave data over a given timeframe. In this study, a preliminary analysis was conducted on the acquired wave data to identify randomness and the presence of outliers. Parameters for four types of probability distributions were estimated using the L-moments method, followed by the selection of the optimal distribution model. Extreme value analysis was performed on the Annual Maximum (AM) and Peaks Over Threshold (POT) series extracted from 30 years of hourly wave observations at Deokjeokdo, with jittering applied to ensure numerical continuity. The results of various goodness-of-fit tests and information criteria demonstrated that the GPA distribution estimated from the POT series is the most suitable model for determining design wave heights in the Deokjeokdo region.
Oh et al. (Sat,) studied this question.