Extreme wave statistical methods and implications for coastal analyses

Abstract This work reviews methods for extreme value analysis of ocean waves, including: site specific univariate analysis, environmental contour analysis and spatial analysis. A variation of regional frequency analysis is proposed based on the hypothesis that the distribution of extreme wave events at different sites within a homogeneous region are identical, except for a shift due to a location parameter. With this assumption the shape and scale parameters of the distribution can be determined with greater certainty by pooled measurements, while the spatial variation is captured by the location parameter which is equal to a threshold wave height and may be more reliably sourced from a computational model (as compared to the extreme values themselves). This approach attempts to overcome the lack of accuracy of most wave models in extreme wave events by basing the shape and scale parameters on measurements, while using a wave hind-cast in a range where it is acceptably accurate to find the spatially varying location parameter. The methods are applied to a case study of the Canadian Pacific Coast. Un-sheltered measurement sites were found to satisfy statistical homogeneity and were used to define a regional, shifted extreme value distribution. A high resolution numerical model was used to define the spatially varying location parameter equal to the 98th percentile of significant wave heights. In general the results are satisfactory and suggest this approach may be a useful tool in understanding the spatial variability of wave extremes. However, it is acknowledged that this approach relies on the availability of multiple measurement sites with sufficient duration to support extreme value analysis.