Browse by author
Lookup NU author(s): Dr Lee Fawcett,
Dr David WalshawORCiD
This work is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License (CC BY-NC 4.0).
Accurate and precise estimation of return levelsis often a key goal of any extreme value analysis. For example,in the UK the British Standards Institution (BSI) incorporateestimates of ‘once-in-50-year wind gust speeds’ – or50-year return levels – into their design codes for new structures;similarly, the Dutch Delta Commission use estimatesof the 10,000-year return level for sea-surge to aid the constructionof flood defence systems built to protect against seaflooding. In this paper, we briefly highlight the shortcomingsof standard methods for estimating return levels fromthe univariate extreme value toolbox, before presenting anestimation framework which we show can greatly increasethe precision of these estimates. Some of our work exploitsrecent developments in the estimation of the extremal index,a key parameter which characterises the strength of serialdependence present in our environmental extremes. Turningfrom frequentist ideas, we consider the Bayesian paradigmas a natural approach for building complex hierarchical orspatial models for extremes. We recommend the Bayesianposterior predictive value as the most satisfactory representationof a return level for use in practice.
Author(s): Fawcett L, Walshaw D
Publication type: Article
Publication status: Published
Journal: Stochastic Environmental Research and Risk Assessment
Print publication date: 01/02/2016
Online publication date: 29/07/2015
Acceptance date: 01/01/1900
Date deposited: 27/09/2016
ISSN (print): 1436-3240
ISSN (electronic): 1436-3259
Altmetrics provided by Altmetric