Novel Extreme Value Estimation Procedures: Application to Extreme Wind Data.
Novel Extreme Value Estimation Procedures: Application
to Extreme Wind Data.
(837 K)
Gross, J. L.; Heckert, N. A.; Lechner, J. A.; Simiu, E.
Extreme Value Theory and Applications, Proceedings.
Volume 1. 1994, Kluwer Academic Publishers, Boston, MA,
Galambos, J., Editor(s), 139-158 pp, 1994.
Sponsor:
National Science Foundation, Washington, DC
Keywords:
wind velocity; wind effects; simulation; monte carlo
method
Abstract:
The past two decades have seen the development of a
large body of extreme value theory based on the
application of the Generalized Pareto Distribution (GPD)
to the excess of the extreme variate over a fixed
threshold. For sufficiently large values of the extreme
variates, the GPD with tail length parameters c > O and
c < O is equivalent, respectively, to the Type II
(Frechet) and Type III (reverse Weibull) distribution of
the largest values. The Type I (Gumbel) distribution is
equivalent to the limit of the GPD as c -> O. Owing to
these equivalences, the GPD can be used to model extreme
data obtained by either the 'peaks over threshold'
approach or the epochal approach. The overall purpose
of our investigation is to assess and use the potential
of GPD/extreme value theory for improving our knowledge
of extreme wind speed behavior. In particular we are
interested in examining the issue of the extreme
distribution tail length.
Building and Fire Research Laboratory
National Institute of Standards and Technology
Gaithersburg, MD 20899