Preconditioning

In Harris, 1996, the author states at the end of section 4 that when fitting a Gumbel distribution to a set of extreme wind speeds whose underlying distribution is a Weibull, the curve fit will be closer to linear (and will lead to more accurate extrapolations) if the probabilities are plotted versus the wind speeds raised to the power of the Weibull k, rather than the raw wind speeds. He then states that in the UK, Weibull k values tend to be close enough to 2 that simply squaring the wind speeds gives virtually the same benefit.

Langreder et al. (2009) apply the term 'preconditioning' to this process of raising the wind speeds to an exponent before fitting the Gumbel distribution. They also confirm that preconditioning can indeed improve the accuracy of 50-year extreme wind speed estimates, and that it makes very little difference whether the preconditioning exponent is k or 2.

The Periodic Maxima Algorithm and the Method of Independent Storms on the Extreme Wind Analysis window both allow you the option of using preconditioning.