AbstractsMathematics

Effects of Seasonality on the Estimation of Environmental Extremes: A Study of Non-Homogeneous Methods for Estimating Environmental Extremes:

by A.C. Trahan




Institution: Delft University of Technology
Department:
Year: 2013
Keywords: Extreme Value Analysis; Non-Homogeneous Poisson Process; Hydraulic Boundary Conditions
Record ID: 1264994
Full text PDF: http://resolver.tudelft.nl/uuid:6694eaf1-489b-4ad4-87ea-61c1eea63fd0


Abstract

Extreme value theory is commonly applied in ocean engineering to estimate extreme environmental conditions (e.g. wave height and wind speed). Extreme value models generally assume data are identically distributed and homogeneous in time. Because there are often different dominant generating mechanisms in different seasons, the sample distributions vary through the year, potentially violating the identical distribution assumption. Two methods for overcoming this are (1) dividing the data into seasons whose distributions are homogeneous (‘time-stratified’) and (2) employing a time-varying extreme value model whose parameters change through the year, such as a non-homogeneous Poisson process (‘time-varying’). The difference between the results of these two methods and the classical method indicates the influence of violating the identical distribution assumption. These methods also provide seasonal extreme estimates, which are useful for design, construction, and maintenance of ocean structures. Based on analysis of model and observation data from thirteen sites in the North Sea (4 wind speed, 9 wave height), the differences between all-year extreme estimates from the classical method, the time-stratified method, and the time-varying method prove not to be statistically significant. This implies the violation of the identical distribution assumption is also not significant. The seasonal estimates of the time-stratified method and the timevarying method are also compatible, validating the use of the less man-hour-intensive and more versatile time-varying method for calculating seasonal extremes.