| Method of Independent Storms | |
The Method of Independent Storms (MIS) is one of the algorithms that Windographer uses to calculate the 50-year extreme wind speed. This is a published algorithm (Harris, 1999) that Windographer implements almost exactly as published, so the Harris article is the best place to look for details and explanations. The only two adaptations in Windographer's implementation of the algorithm are:
Compared to the Periodic Maxima method, the principal distinguishing feature of MIS is the way it builds the set of points to which it fits a Gumbel distribution. Rather than finding the maximum wind speed in each period of fixed length, MIS searches through the wind speed time series looking for storms, meaning events in which the wind speed exceeds a threshold value for some period of time before dropping back below the threshold. The peak wind speed value from each storm event is then used in the Gumbel fit procedure.
The MIS uses only the single peak value from each storm event, regardless of the length of the event. The graph below, for example, shows four storm events of varying length that occur in a single day, using a threshold wind speed of 20 m/s. Windographer colors the storm events alternately blue and yellow, and identifies with a red flag the peak wind speed occurring within each storm event:
By default, Windographer enforces an independence criterion of 48 hours, meaning it merges storm events that are separated by less than 48 hours of time. Doing so in this example reduces the four storm events shown above to a single event:
If you experiment with different values of the threshold wind speed and the independence criterion, you should find that they do not drastically affect the resulting 50-year extreme wind speed. This is one of the strong points of this algorithm. Both Cook, 1982 and Harris, 1999 report that they found the algorithm results to be insensitive to the value of the threshold wind speed.
Tip: If you click the Set Recommended button, Windographer will set the threshold wind speed to a value that results in approximately 20 storms per year. Although Harris recommended a target of 100 storms/yr, the default independence criterion of 48 hours makes that target impractically high. We found that a target of 20 storms/yr and a 48-hr independence criterion resulted in approximately the same threshold wind speed as a target of 100 storms/yr with no independence criterion.
In his 1999 paper on the MIS, Harris used the Harris 1999 algorithm to fit a Gumbel distribution to the highest few storm maxima values. The Harris 1999 algorithm deliberately chooses only the top few storm maxima to use in the curve fit. (The algorithm includes logic to choose the number of storm peaks to use in the curve fit.)
Windographer uses the Harris 1999 algorithm for the MIS by default, but if you wish, you can instead choose to employ the Simple algorithm for Gumbel fitting or the Harris 1996 algorithm, both of which use all available storm maxima. Windographer will display the best-fit Gumbel parameters, and if you choose the simple Gumbel fit algorithm, it will also display the R2 value.
Tip: Until mid-2022, the option to choose any algorithm other than the Harris 1999 algorithm is available only to the customer that helped fund this improvement.
As with the Periodic Maxima method, you can choose to precondition the wind speed values by applying an exponent to them before performing the Gumbel curve fit. Preconditioning can often improve the convergence of the curve fit and therefore improve the estimate of the 50-year extreme wind speed.
See also