Benchmark for the Sources of Uncertainty in Extreme Wave Analysis

Abstract

This paper discusses the uncertainty in extreme wave analysis from different sources. Poor data quality and small sample size will lead to uncertainty in the extreme wave analysis. Extreme value estimation methods are developed based on various assumptions, and each would lead to unique estimation results. In addition, the cause of extreme waves varies among regions, directly affecting the extreme behavior. The aim of this study is to provide insight into how the uncertainty of extreme wave estimation is influenced by the different source of uncertainty, namely data uncertainty, method selection and extreme behavior at each location. Key parameters to describe the extreme wave events are the frequency of occurrence and its tail-behavior. We use these two parameters as a benchmark to assess the extreme wave characteristics. We focus on four regions, namely Gulf of Mexico, North Sea, Adriatic Sea, and North West Pacific. Meteorological cause of extreme events and known extreme wave behavior are reviewed based on previous studies. Model inter-comparison revealed the shortcomings of wave models to reproduce extreme wave events, and the magnitude of data error was unique to each location. Numerical experiments were conducted to evaluate the possible impact from poor data quality and small sample size on epistemic uncertainty. Case study based on representative parameters of Gulf of Mexico and North Sea revealed the difference between two locations. These results provide a benchmark for the source of uncertainty and its impact on extreme wave analysis. Among them, extreme waves dominated by tropical cyclones were most vulnerable to have large epistemic uncertainty. The importance to adequately quantify epistemic and aleatory uncertainty is reconfirmed.