EXTREME WIND DISTRIBUTION TAILS - A PEAKS OVER THRESHOLD APPROACH

We seek to ascertain whether the reverse Weibull distribution is an ap propriate extreme windspeed model by performing statistical analyses b ased on the ''peaks over threshold'' approach. We use the de Haan meth od, which was found in previous studies to perform about as well or be tter than the Pickands and cumulative mean exceedance methods, and has the advantage of providing estimates of confidence bounds. The data a re taken principally from records of the largest daily wind speeds obt ained over periods of 15 to 26 years at 44 U.S. weather stations in ar eas not subjected to mature hurricane winds. From these records we cre ate samples with reduced mutual correlation among the data. In our opi nion, the analyses provide persuasive evidence that extreme wind speed s are described predominantly by reverse Weibull distributions, which unlike the Gumbel distribution have a finite upper tail and lead to re asonable estimates of wind load factors. Instructions are provided for accessing the data and attendant programs.