understanding of one of Mother Nature's more disastrous forces.

The Risk Prediction Initiative (RPI) programme is aimed at providing scientists with a better understanding of the past which will aid reinsurers, and countries, in the future.

According to assistant research scientist David Malmquist, the project started after a meeting between Tony Knap, Director of the Bermuda Biological Station for Research and some of the Island's leading reinsurance companies.

There to raise funds for BBSR, Dr. Knap came upon a means through which the reinsurers could benefit through scientific studies conducted by the experts.

It was decided to work towards a means by which BBSR would provide a rapid and up-to-date connection between the science of experimental prediction and the parts of the business community affected by environmental change and catastrophe -- in this case, hurricanes.

"Hurricane Andrew which hit Miami in 1992 was the largest insurance loss ever -- at $15 billion,'' said Mr. Malmquist. "It came as a great surprise to reinsurers.'' Part of the reason for the surprise, was that traditionally, reinsurers calculated the probability of disasters based on the frequency of events over the past 150 years. Until 1992, most information resulted in losses much less than what occurred. The change, was not a reflection of an unusual increase in hurricane strength, but the result of a large increase in Miami's population and an increased material wealth in the average household.

As such, it was decided a new means of risk assessment would have to be created.

RPI is working to show that some cycles in the global climate system are longer than the 150-year observation period. And through an analysis of ponds, trees and caves, the team at the Bio Station, along with experts outside Bermuda, are achieving that goal.

"When a hurricane passes over land, it leaves records,'' he explained. "For example, trees are knocked down and storm surges occur. In many places, and Spittal Pond is one example, when there is a storm surge, there is a layer of sand which lands on top of the mud. When the surge subsides, the mud is placed back on top again. So if someone took a core analysis of Spittal Pond, they would be able to get a record of the incidence of hurricanes over a period of time from which we could asses the probability of a future strike. So instead of a statistical extrapolation of two storms, we have physical evidence that a storm of some size has passed over that area.'' An alternative method of obtaining records for predicting storms are by looking at a cross-section of tree trunks in various parts of a country as trees are typically sensitive to climate change, he said. As such, he would be particularly grateful if persons removing or cutting down trees would drop them off to him for analysis.

A third method has the scientists examining rock formation in Bermuda's caves.

"Rain that comes from a hurricane differs, in terms of isotopic signs, from regular rain,'' he explained. "Atoms come in three different isotopes -- 16, 17 and 18. O18 has two extra neutrons than O16 and is therefore heavier.

"As a raindrop falls to the ground, it evaporates on the way down and gets heavier as the O16 evaporates and the O18 stays put.

"In a hurricane, the air is typically so saturated with moisture, that very little evaporates before the raindrop hits the surface and so those raindrops end up being lighter than regular ones. Caves form by rainwater falling on the surface and dripping into rocks.'' A cut rock sample from a cave, he said, will show various layers which indicate its growth.

"You won't see individual hurricanes but you might see periods of twenty or thirty years where there were more or less hurricanes than at other times.'' According to Mr. Malmquist, the hope is that RPI will create novel communications between climate science and insurance by: Making the science of climate variability and forecasting available, understandable, useable and relevant to insurers.

Making the science community aware of the needs, interests and approaches of the insurance industry.

Identifying the likely contribution of a public, academic, not-for-profit entity in benefitting the activities of the insurance and science communities.

It is also hoped, he said that RPI will improve and create data and forecasting products on topics of mutual interest and benefit to the climate science and insurance/reinsurance communities by: Focusing on understanding the variability of hurricane/typhoon landfall and the predictability of future landfall probabilities as different from the recent historical record.

Focusing initially on El Nino as the predictable climate phenomenon, with additional emphasis on other predictable climate features relevant to hurricanes, and funding academic research towards making predictions on time-scales of years to decades.

Working to expand the available datasets for Atlantic hurricanes and for typhoons in the Pacific that affect Japan and Australia and studying the correlation between landfall events in different basings.

Identifying novel approaches to measuring the return periods of strong hurricanes and typhoons and funding some of the novel, directed academic research in these areas.

Laying the groundwork for, and perhaps funding pilot research projects on other climate perils that affect insurers and for which a body of academic research exists.

"Insurers are concerned, if we're heading into a period of increased activity as many scientists believe, that these hurricanes will be hitting the US coastlines with more people and more buildings. So we could be heading for a major disaster of the type we saw with Hurricane Andrew.'' An analysis of mud taken from Spittal Pond HURRICANES SUPPLEMENT HUR