Extracting independent and identically distributed samples from time series significant wave heights in the Yellow Sea

Abstract

The assessment of extreme significant wave heights is crucial for the design of coastal defences and offshore infrastructures, which requires independent and identically distributed samples to be extracted. To extract independent samples at the regional scale, an automated method is proposed by extending an observational method. In addition to the initial threshold, a minimum interval is used to identify two consecutive storms with a long storm interval, and a minimum level is used to distinguish one storm with a small fluctuation around the initial threshold from two consecutive storms with a short storm interval. By using this method, independent samples in the Yellow Sea are extracted from a 40-year (1979–2018) hindcast of significant wave heights. In this area, storms during tropical cyclones are generally strong, and most storms are driven by winter storms. Comprehensively considering the storm intensity and frequency, the independent sample is divided by an automated method into homogenous samples 1 and 2, depending on the tracks and recorded times of tropical cyclones. Homogenous samples 1 and 2 represent the independent samples in the tropical cyclone and non-tropical cyclone, respectively. Within homogenous sample 2, most independent samples (especially large independent samples) are observed in oceanographic winter. In addition, the difference between the return significant wave heights based on homogenous sample 2 and those based on the winter storm sample is very small. Thus, seasonal declustering is not performed on homogenous sample 2. Consequently, by utilizing the independent and identically distributed samples extracted by two automated methods, a regional analysis of the extreme significant wave heights in the Yellow Sea is performed.