Resampling Approach to Extreme Wave‐Height Analysis

A bootstrap resampling approach is developed for extremal analysis of measured and hindcast ocean‐wave data for use in design, construction, operations, and maintenance of coastal structures. The resampling method produces estimates for the extrapolation bias associated with competing extreme‐probability models. A goodness‐of‐fit statistic for the extremal type 1 model is also developed. The bias estimates and goodness‐of‐fit statistic are used in selecting an appropriate probability model for a given record of extreme significant wave heights. The method is shown to provide information about model selection that is not available from present extremal analysis methodology, and may help prevent significant un‐derprediction of extremes that can occur with conventional approaches. Examples using both hindcast and measured wave data are presented to demonstrate application of the methods developed in this paper.