Estimation of extreme 10 minute average wind speed

Hi everybody. this is maybe a very simple question, but at the moment I do not know how to proceed.

I need to estimate the site extreme 10-min average wind speed at hub height with a recurrence period of 50 years for a few wind sites, where we collect aneometer data. Data are in the usual form:

time stamp1, Vaverage1, Vmax1, Vmin1,DevStd1,
time stamp2, Vaverage2, Vmax2, Vmin2,DevStd2,

and so on for 1-2 years worth of data

where the average values are on 10 minutes intervals

I have developed a good method to determine the Weibull distribution best fitting the data. But this is ok for the statistic of the mean values of the wind, not for the extreme values.

Using the same type of data, how could I extrapolate the expected 50 year recurrence of the wind speed for a given site? Somehow I think that the keyword here should be “Gumbel distribution”, but I need to understand how to apply the method (if it is the right one!).

Thanks a lot for your advice
Claudio Pedrazzi


The asymptotically correct distribution for wind maxima is the Gumbel (sometimes referred to as the type I GEV), although you should always check the goodness of the fit against the data before extrapolating, another distribution may prove to be better.