Abstract
Statistics on extreme wave heights are needed to evaluate and classify risks at wave energy project sites with respect to opportunities for wave energy extraction. They are also used to evaluate and classify the extreme environmental wave loads, e.g., those based on significant wave heights occurring at 50-year return periods, Hs(50), as required by standards to design wave energy converters. The present study goal is to calculate, map and classify values of Hs(50), and the relative risk ratio, Hs(50)/Hs(mean) , within the economic exclusion zone of the US Coastline. These statistics are derived from Hs time series outputted from validated high-resolution 30-year wave climate hindcasts at spatial resolutions of 200-300 m at millions of wave sites. Two standard methods for calculating from extreme value distributions are compared and evaluated to select the best practical approach for extreme wave height estimation: A peak-over-threshold (POT) method to derive an exponential extreme value distribution model from a partial series based on a selected threshold value and lag period, and an annual maxima method that applies the Gumbel distribution model to a simple annual maxima series. Both methods are accepted by international standards. While the partial series approach is generally considered to be more accurate, particularly for short historical periods of record, several statistical tests are required to justify the threshold value selected and threshold value selection can be subjective and time consuming. The annual series approach, by contrast, requires no selection of parameters to derive the sample population used for the modeled extreme value distribution. Comparison of the two approaches for eighty-five sites along the West Coast indicates reasonable agreement, supporting the use of the annual series approach for resource and conditions classification due to its relative simplicity and ease of computation. Model diagnostics for the POT method establish both independence of peak wave heights through auto-correlation, and goodness of fit tests. Further evaluation of the annual maxima method is needed before its uniform adoption to all US wave climates.