Meteorological Extremes

Abstract

This article reviews basic principles of the statistical analysis of large extreme values (EVs), the role of simulations in the development of relevant meteorological data sets, the interaction between the EV modeling process and its application to engineering, and specific issues pertaining to wind, snow, ice, storm surge, and ocean waves. Classical approaches and the peak over threshold approach (POT) to the estimation of univariate extreme value distributions are briefly described. The distributions include: the Gumbel (Type I), Fréchet (Type II), reverse Weibull (Type III) extreme value distributions. One version of the POT approach relevant for meteorological extremes describes the joint observation of the time and magnitude of the threshold exceedance as a two‐dimensional nonhomogeneous Poisson process. It is noted that the dependence on direction of the extreme meteorological quantities of interest plays an important role in the estimation of extreme effects of these quantities on engineering structures. Also noted are the relative paucity of data on freezing rain and associated glaze ice accretion, and difficulties encountered in the estimation of tornado wind speeds. The article is concluded with some brief comments on multivariate extremes.