By Martin Redfern
BBC radio science unit
A new method to forecast North American hurricane activity has been developed by a team at University College London.
The UK researchers use winds averaged throughout July to predict the severity of the hurricane season that generally runs from August to October.
With Charley, Frances, Ivan and Jeanne, 2004 was a bad year for North American hurricanes. Hundreds died and billions of dollars of damage was caused.
The new work, by Mark Saunders and Adam Lea, is reported in the journal Nature.
It cannot forecast individual hurricanes but it can at least give a strong indication of how destructive a coming season might be.
The team, who operate out of UCL's Benfield Natural Hazard Research Centre, averaged wind speeds and directions throughout the troposphere, from ground level to 8,000m (26,000ft) altitude.
They looked at each hurricane season since 1950 and, in particular, at the height-averaged winds during July, the month immediately preceding the hurricane season.
In an interview for Science in Action on the BBC World Service, Mark Saunders said: "We find over the North Atlantic, North America and the East Pacific that anomalies in height-averaged winds during July are in several parts of this region linked significantly with upcoming hurricane activity striking North America."
To identify the July winds which were critical to a severe hurricane season, the researchers combined data from the 14 most severe hurricane years since 1950 and compared it with data from the 14 least damaging years.
A hurricane is a spinning vortex of winds swirling round a eye of very low pressure
Warm, moist air is drawn upwards around the eye
Cooler dry air is sucked downwards by the low pressure centre
Banks of thunderstorms surround the edges
The differences are shown by the small arrows on the enlarged map at the bottom of this page, and the most critical areas are highlighted in red and orange.
Hurricanes are not caused by wind but by high sea-surface temperature. However, it is the wind that determines how the storms develop and whether they will head for land or not.
It is not enough just to see which way the wind is blowing, says Dr Saunders; you need to combine measurements for a whole month.
"There's lots of day-to-day variability, but by doing these long-term averages, one can bring out predictive links far better."
It is too early to make predictions for the 2005 hurricane season, but the researchers tested their model by seeing if they could predict in retrospect past years using only the data from July.
"One can predict whether seasonal hurricane activity striking North America will be above average or below average from the first of August in nearly 80% of those 55 years," said Dr Saunders.
The new technique will not tell people exactly when or where individual hurricanes will strike so it may not be of much immediate use to individual householders. But it could help in government planning and save the insurance industry millions if companies know in which years they should spread their risks more widely.
WIND ANOMALY DATA
The wind data may allow predictions of how destructive a hurricane season will be
It is not only North America that is threatened by severe tropical storms. Typhoons and cyclones devastate parts of South and East Asia on a regular basis. Dr Saunders hopes his technique will prove useful there, too.
"I think it would be very interesting for other parts of the world where tropical storms are a severe cause of loss and damage, to see whether height-averaged winds offer predictability in those parts of the world, too."