Kernel density estimation (KDE) with adaptive bandwidth selection for environmental contours of extreme sea states

Abstract

The estimation of environmental contours of extreme sea states characterized by significant wave height and energy period for the purposes of reliability-based offshore design is a problem that has been tackled in many different ways. Many of the methods used to generate such contours rely on parametric approaches that require an a priori assumption of the relationship between the variables of interest. These relationships, often given in the form of assumed probability distributions or joint probability structures, may not be flexible across a wide variety of global observation sites. We propose the use of bivariate kernel density estimation (KDE) with adaptive bandwidth selection for generating the joint probability distribution of significant wave height and energy period. This method is nonparametric, straightforward to apply, and lends itself to a characterization of contour uncertainty that is an important aspect when contours are used to generate inputs for numerical or physical simulations of offshore structures. The joint probability distribution of significant wave height and energy period can be queried using a return period of interest to determine environmental contours of extreme sea states. This paper demonstrates that this method provides a robust and flexible characterization when compared to other approaches.