Comparison of local indices for regional frequency analysis with an application to extreme skew surges
Abstract
Regional frequency analysis (RFA) is a valuable and well-known method which allows using all the information at the regional scale to improve the actual estimation of the probability of occurrence of extreme events at a given site. In the framework of the index flood method, a local index, representing the local specificities of a given site, is used to normalize at-site observations for the estimation of the regional distribution. It is an essential feature of this model, contrasting with common characteristics shared between the sites of the homogenous region. However, the specification of the local index can be a crucial point. In particular, the performance of the quantile estimator derived from a RFA can depend on the specification of the local index. Four regionalization models are proposed, where the local index is specified by different statistics in each model, and their performances are assessed through Monte Carlo simulations of several regional scenarios. Some guidelines are provided for the selection of the local index which is most adapted to the observed situation (including regional scenarios characterized by some degrees of asymmetry, homogeneity and inter-site correlation). A practical application on extreme skew storm surges is provided to illustrate the results. ©2013. American Geophysical Union. All Rights Reserved.
Keywords
extreme values
local index
regional frequency analysis
skew surges
Essential features
Extreme value
Probability of occurrence
Quantile estimators
Regional distribution
Estimation
Floods
Monte Carlo methods
Specifications
Site selection
asymmetry
comparative study
estimation method
frequency analysis
guideline
homogeneity
index method
numerical model
skewness