Development & Analysis Of A UK Storm Indicator (part 10)

I continue turning the handle on wind speed data as gathered by 26 Irish weather stations since WWII. Today I settle on a modest model for predicting UK named storms

In my last article we saw what appeared to be a promising approach to predicting the incidence of UK named storms using the 95 percentile extreme windy day count derived from the mean maximum wind speed of seven long-series Irish weather stations (the “Magnificent 7”). Today I am going to expand on this work by developing the very best generalised linear model (GLM) I can muster using the limited data available to me.

We’ve already seen that the 95 percentile Irish windy day count correlates strongly with the UK monthly named storm count (r = 0.765, p<0.001, n=72), but what we haven’t yet seen is that the mean maximum gust speed for the seven long-series stations (Birr, Dublin Airport, Dublin Phoenix Park, Gurteen, Malin Head, Roche's Point, Shannon) also correlates well, with r = 0.684 (p<0.001, n=72). I thus decided to whip up a quick GLM using these two independent variables in the prediction of the named storm count, with the assumption of an underlying Poisson distribution for the…