Premium
More realistic distributions for extreme wind gusts
Author(s) -
Revfeim K. J. A.,
Hessell J. W. D.
Publication year - 1984
Publication title -
quarterly journal of the royal meteorological society
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.744
H-Index - 143
eISSN - 1477-870X
pISSN - 0035-9009
DOI - 10.1002/qj.49711046412
Subject(s) - gumbel distribution , generalized extreme value distribution , extreme value theory , maxima , mathematics , wind speed , parametric statistics , statistics , statistical physics , meteorology , physics , art , performance art , art history
Maximum wind gusts are commonly analysed by fitting the Type I (Gumbel) extreme value distribution. No physical meaning is given to the usual parametric form of the distribution. The stochastic nature of wind gusts provides an alternative distribution form with physically meaningful parameters. For wind gusts the implication of this alternative form is that the most frequent gust size is infinitesimally small. Other stochastically derived extreme value distributions allow for non‐zero modality in the distribution of wind gust sizes underlying the maximum. The assumptions of the stochastic model are qualified in their application to wind gusts. The maximum‐likelihood method may not find parameter estimates of the Type I distribution for a given sample of maximum gusts. However, for any data set a family of size distributions, with non‐zero modes, does provide parameter estimates having greater mean likelihood than the Type I distribution. Seasonal patterns in the parameters are estimated as harmonic functions and expected long term maxima are given.