Assessment of Data-Inherited Uncertainty in Extreme Wave Analysis

Abstract

Accurate estimation of extreme wave condition is desired for the rational design of offshore structures, but the estimation results are known to have uncertainty from various sources. The quality and quantity of the available extreme wave data differ among ocean regions since the atmospheric causes of extreme waves are not identical. This paper provides insight into how the different extreme wave behaviors influence the uncertainty of extreme wave estimation at each location. A review of extreme waves in four regions, namely the Gulf of Mexico, North West Pacific, Adriatic Sea, and the North Sea, revealed the difference in data uncertainty, shape parameter, and frequency of occurrence. The likelihood-weighted method was introduced to quantitatively assess the impact of each parameter on the uncertainty of extreme wave analysis. Case study based on representative parameters of the Gulf of Mexico and the North Sea revealed the large epistemic uncertainty for a region dominated by tropical cyclones. The assessment conducted in this paper is unique as it evaluates the epistemic uncertainty inherited in the extreme sample data. When the epistemic uncertainty is large, such as the case illustrated for the Gulf of Mexico, the variance from different approaches may not be significant against the epistemic uncertainty inherited in the sample data.