Predicting Joint Return Period Under Ocean Extremes Based on a Maximum Entropy Compound Distribution Model

Abstract

In this paper, we proposed a novel 2-dimensional (2D) distribution model based on the maximum-entropy (ME) principle to predict the joint return period under ocean extremes. In detail, we first derive the joint probability distribution of the extreme wave heights and the extreme water-levels during a typhoon by using the maximum-entropy principle, and then we nest this distribution with the maximum-entropy distribution of discrete variables to form such a maximum-entropy 2-dimensional (ME 2D) compound distribution model. To evaluate the performance of our model, we conduct experiments to predict the N-year joint return-periods of the extreme wave heights and the extreme water levels in two areas of the East China Sea. According to the experimental results, our model performs better in predicting in the highly unpredictable joint probability of extreme wave heights and water levels in typhoon affected sea areas, compared with the widely-used Poisson-Mixed-Gumbel model in ocean engineering design. This ascribes to the fact that unlike other models whose corresponding parameters are arbitrarily assigned, our model utilizes both the new 2D distribution and the discrete distribution which are based on the ME principle.