A new IFORM-Rosenblatt framework for calculation of environmental contours

Abstract

Environmental contours are useful for designing and assessing risk to offshore structures. The Inverse First-Order Reliability Method (IFORM) and the Rosenblatt transformation are widely used to calculate environmental contours but have several challenges. In this approach, the conditional distribution is often expressed in a parametric form with parameters defined as functions of the independent variable. The distribution should be accurate for both extreme and frequently occurring values, for both the dependent and independent variables, and this is a challenging requirement for complex datasets. A new IFORM-Rosenblatt framework is proposed here to overcome this challenge. The framework includes a method to approximate a distribution with a complex form, including parametric, empirical, and mixed parametric/empirical forms, using a reduced set of parameters that are designed to prioritize accuracy for the quantiles of the distribution that contribute to the contour and a set of metrics to evaluate the goodness-of-fit of the resulting contour. A hindcast dataset is used as an example to calculate the 50-year contour using the proposed framework and a traditional approach. The proposed framework is demonstrated to generate accurate contours, although it requires more parameters and is more complex than conventional approaches.