Scientists Reveal Groundbreaking Research into Future Risks

NewYork, NY (CompNewsNetwork) - Scientists from the Willis Research Network (WRN), part of Willis Group Holdings Limited (NYSE:WSH), the global insurance broker, have outlined groundbreaking new research on future hurricane activity and storm clustering today, including unparalleled use of supercomputers to tackle Caribbean and Gulf of Mexico hurricane risk.

In his keynote presentation at the latest Willis Research Network Seminar, “Supercomputers, Climate Change and Catastrophe Modelling,” Dr Greg Holland from the National Center for Atmospheric Research (NCAR) described how NCAR's premier models for weather and climate, along with the computing power of its Bluefire supercomputer, are being used to dramatically enhance research into changes in future location, frequency, wind severity and rainfall of hurricane tracks across key Atlantic regions.

“Combined advances in supercomputing capacity and innovative approaches to climate modelling are enabling us to substantially improve the capacity for projecting future variations and trends in weather systems of high impact to society,” said Dr Holland.

Modelling high impact weather systems, such as local flash flooding, hurricanes and European Windstorms, requires computer models of a complexity and scale hitherto only possible for short-term weather forecasts.

NCAR is working to develop the ability to make long-range forecasts by combining its advanced weather forecasting model with its IPCC climate model.  The combination is expected to enable predictions of changes in the characteristics of high-impact weather on decadal time scales. After three years of testing and development, the program is currently nearing the completion of a 50-year projection of variations in Atlantic hurricanes and western US precipitation.

“The first few months of 2009 will be extremely exciting for us as the first results begin to come in,” said Dr Holland. “These results will be used to develop new approaches to assessing risk and vulnerability to severe weather, by combining the model projections with statistical assessments based on existing weather records.”

Rowan Douglas, Chairman, Willis Research Network, commented, “This research will have far-reaching consequences throughout the scientific and financial services communities. Willis' focus now is to apply this work quickly and practically to the risk modelling work we do for our clients in the re/insurance industry.”

Also presenting at the conference, which was attended by over 250 insurance professionals, scientists and academics from around the world, was Dr Renato Vitolo, Willis Research Fellow at Exeter University. Dr Vitolo premiered the results of his research into European Extreme Windstorm Clustering, work he has undertaken with fellow WRN members from the Walker Institute at Reading University.

Dr Vitolo revealed that he had discovered “significant clustering of intense windstorms for a large area of northern Europe,” and surmised that this clustering was the result of atmospheric variability at low frequencies, as represented by teleconnection indices. The research also pointed to large-scale clustering in several North Atlantic locations, including Florida and the Caribbean, a conclusion supported by work with HiGEM, the Met Office Hadley Centre's Atmospheric Model, which used 130 years of historical tropical cyclone tracks to reveal clustering of yearly counts in the Pacific and Atlantic Oceans.

This seminar is the latest in a programme of seminars held by Willis Re and The Willis Research Network (WRN). The WRN is the world's largest partnership between academia and the insurance industry. Willis has so far teamed up with fifteen leading institutions across a full range of disciplines from atmospheric science and climate statistics, to geography, hydrology and seismology, to the impacts on the environment via engineering, exposure analysis and Geographic Information Systems. Additional information can be found at www.willisresearchnetwork.com

Share This Article: