(RPI) and its sponsors used the occasion of RPI's 10th anniversary in 2004 to assess accomplishments and set new goals. A major change for RPI's 10th year is a realignment of the
RPI Research Groups to better reflect RPI sponsors' interests in climate change and a growing variety of hazards.Ten years ago, Bermuda's innovative catastrophe reinsurers and BBSR scientists
formed RPI as an experiment to interest scientists in problems important to insurers. The catalyst for the formation of RPI and many Bermudian reinsurers occurred two years earlier
when Hurricane Andrew pummeled Florida in 1992 and caused approximately $20 billion (in 2004 dollars) in insured losses.
These unexpectedly large losses initiated a shift in business practices by
surviving catastrophe reinsurers and the formation of new reinsurance companies that use complex risk models as a basis for a more technical
and scientific approach to writing catastrophe reinsurance. The companies' interest in RPI was driven partly by an interest in building the best science
into new risk models and business decisions. The large number of tropical cyclone landfalls so far in 2004 and a recent RPI workshop on wildfires
provide two examples of the importance of RPI's efforts to link science and business.
Risk models are based on the assumption that the historical record
provides a reasonable approximation of future variability in hurricane activity. Thus a risk model's estimate of the probability of a storm such as
Hurricane Andrew forming and striking land is ultimately based on the historical record of storm strikes.
are certainly on the outer bounds of the
historical record. By press time, four major hurricanes have struck Florida this year, the largest number of Florida landfalls since historical records
began in 1851. In addition, the last time a state was hit by four storms in one year was in 1886 when Texas earned the dubious distinction of being the
favorite storm target. Florida is not suffering alone. So far this year Japan has been struck by 10 tropical cyclones with winds of tropical storm strength
or greater. The old record was seven tropical cyclones striking Japan in one year. The unusual number of tropical cyclone landfalls in 2004 naturally begs the
question "why?" Are the record numbers of landfalls within the range of natural variability associated with a relatively stable climate, or do they
herald the onset of a significant change in climate? If climate is changing then there is a host of related questions, such as: What is causing the
change? How much will climate change? How fast will climate change? How will a change in climate alter the probability of a natural hazard?
Existing risk models are not designed to account for changes in climate. However, RPI sponsors are interested in knowing as much as possible
about how climate changes will influence the probability and characteristics of natural hazards. A major area of interest for the RPI Forecasting
Research Group is learning how climate change affects the probability of tropical cyclone landfall so that future risk models will be able to account for climate change.
Wildfires are another natural hazard closely linked to climate variability. When wildfires cross the wildland-urban interface, they can cause significant
losses. For example, in fall 2003 there was a series of large wildfires in southern California that caused nearly $2 billion in insured losses. Last
year's fires capped a series of increasingly frequent and relatively large losses from wildfires and prompted RPI to host a May 2004 workshop
entitled "Fires at the wildland urban interface: What does the future hold?"
The aim of the workshop was to identify scientific questions that need to be
solved to develop realistic wildfire risk models. This was a timely workshop because a number of reinsurers and modeling companies are currently
developing wildfire risk models. Workshop speakers presented results of their work on a number of fire-related topics, including the influence of
climate on fire frequency and extent, how fire frequency might change in response to climate change, and the development of interannual forecasts
for wildfires that account for climate variability driven by phenomenon such as the El Niņo-Southern Oscillation.
A significant finding of the workshop was that the large wildfires of 2003 are
only possible under specific conditions: sufficient fuel, dry air and sustained strong winds, such as the Santa Ana winds associated with a high pressure cell over the western United States.
Developing risk models that account for a changing climate's effects on wildfires, hurricane landfall and other natural hazards will be a significant
challenge. Over the past 10 years, RPI has worked to meet the challenge of making science relevant to its sponsors. In the future, RPI will continue this
work and focus on supporting science that can be used to develop new risk models and test the existing ones.
