Weather: Why we should expect the unexpected

This month sees the 10th anniversary of the infamous Hurricane that wasn't really a hurricane at all. Weather statistics, however, suggest - paradoxically - that unusual weather is rather common.

The remarkable thing about the weather in this country is that it always seems to catch us by surprise. Summer sunshine regularly brings droughts and hosepipe bans as the water companies cannot cope; autumn leaves falling on to railway lines always seem to come as a shock to train operators, and by the time they have got over it, the first falls of the wrong type of snow cause more disruption. Then in spring, Ascot, Wimbledon and the Lord's Test are washed out by unseasonally rainy weather. The truth is that our weather is more often unseasonal than not. The records show how variable our weather really is, both from place to place and year to year.

Take, for example, the number of mornings in a year in which the inhabitants of Kew wake up to find snow on the ground. During the Eighties, the figures for each year were 0, 23, 7, 0, 22, 13, 13, 0, 0, 0. Our minds are good at detecting patterns, but our memories always give greatest weight to the most recent happenings. So at the start of the Nineties, after three consecutive snowless years, the 11 days of snow in 1990-91 would have come as a bit of surprise. Any type of snow by then would have been the wrong type.

With all but one of the years either very snowy or not snowy at all, the average figure of 7.8 snow-covered mornings is of no practical use. The trouble is that arithmetic means make sense only when they come from a distribution with a low standard deviation (the average deflection from the average). In matters of weather, the variation is so great that any patterns that establish themselves in our memories are liable to be false generalisations.

The position is made still worse by the effect on weather of topographical features. Since most of our rain comes from the Atlantic, the west of the country is rainier than the east; and high ground is always rainier than low ground. So Swansea, for example, has an average annual rainfall of 1,158 mm, while not so far away Llandudno has only 756 mm. Meanwhile, on the other side of the country, Ipswich records 610mm.

But it's the deviations from those averages that make it all so unpredictable. Just look, for example, at the average number of hours of sunshine at Kew Observatory during October. The average figure for the entire month is 102 hours, but the range of recordings spreads from 51 hours in 1894 to 160 hours in 1959.

All of which brings us back to the so-called Hurricane of October 1987 (which was only, strictly speaking, an enormous gale). If you measure the wind speed at Kew or Heathrow, the chances are that it will be between 5mph and 12mph. And wind speeds are generally less variable than rainfall or sunshine statistics. There is only a 1 in 100 chance of recording a wind speed greater than 25mph in Kew, yet in October 1938 it was hit by a 73-mph gale. Deviations from the mean, when they happen, can be very high indeed.

In Britain, hurricanes may hardly happen, but extremes of weather are both less unusual and more extreme than we may think.