NU Online News Service, June 21, 3:09 p.m. EST
Ming Lee, president & chief executive officer of catastrophe risk modeler AIR Worldwide says he doesn’t understand “what all the fuss is about” when it comes to recent debates concerning hurricane model revisions.
“The overall hurricane risk has not changed,” he adds.
Lee wrote to NU Online News Service to clear up any misconceptions about AIR’s model.
When was the AIR model updated and how does your approach differ?
The last update was completed in June 2010.
All models begin with the same historical data. However, different assumptions used in model development can lead to differences in model output. AIR builds its models from the ground up, validating each component independently. Critically, we also validate the model from the “top down” to ensure that the final model results are both realistic and robust. The AIR U.S. hurricane has provided a consistent, realistic and comprehensive view of hurricane risk. A robust catastrophe model should deliver reliable estimates of potential loss at a granular level and, more importantly, provide that information before actual losses occur.
Did Hurricane Ike change your view of inland risk?
No, inland losses from Hurricane Ike are nothing new. History has repeatedly shown that hurricane force winds can impact inland states. For example, Hurricanes Audrey (1957) and Gracie (1959) caused the insurance industry to experience hurricane losses well inland from the coast, as far inland as Kentucky and Ohio.
In fact, according to the AIR historical catalog, the average time it takes for a landfalling hurricane to dissipate is about 34 hours. Considering a range in forward speeds of 10 to 20 mph, this translates to an average overland distance of 340 to 680 miles, a considerable distance from the coast. Clearly, many storms can penetrate inland even further. Models that are limited to coastal states fail to capture the full extent of losses.
Since its inception, the AIR model has covered inland states such as Arkansas, Kentucky, Ohio, Oklahoma, and Tennessee, in addition to the entire Gulf and Eastern seaboard from Texas to Maine. In 2010, the to include Illinois, Indiana, and Missouri, and now cover 29 states.
Is there new science in the model?
Yes. In the 2010 update, virtually every component of the model was examined. Our understanding of the physics of natural catastrophes is continually improving, as is the sophistication of the models we use to represent them. A critical role for the catastrophe modeler is deciding when to incorporate new scientific theory. In addition to conducting our own original research, one of the most important jobs the scientists and engineers at AIR have is to critically evaluate the latest research findings in order to determine whether new scientific approaches are credible and how much weight to assign to them. Only after careful vetting do we incorporate new science into any model.
What changes did you make to the model?
The 2010 enhancements enabled greater realism in risk assessment at a more granular level. For example, increasingly sophisticated instrumentation aboard reconnaissance aircraft and on the ground provided vast amounts of high quality, flight-level and surface-level wind speed data previously unavailable. This data has advanced our understanding of the physical structure of hurricanes and improved our ability to estimate local wind speeds—both on the coast and as storms move inland—with greater precision. AIR’s recent model release also reflects findings from a multi-year, peer-reviewed study of the evolution and enforcement of building codes at both state and local levels, and the impact of those codes on building vulnerability.
Meanwhile, our view of inland risk remained generally consistent. History has repeatedly demonstrated that damaging winds from hurricanes can penetrate hundreds, and sometimes thousands of miles inland, and the AIR hurricane model has estimated inland risk for many years.
Also, our approach to incorporating warm sea-surface temperatures into the view of hurricane risk was robust and has not changed.
(AIR continues to incorporate storm surge and wind duration into the model as well.)
AIR’s view of risk has been consistent and this update did not fundamentally change our overall view of the risk. The advances in science, technology and engineering—combined with new data at high resolution—provided the opportunity to advance the realism of the model and improve its ability to differentiate risks of different characteristics, which is exactly what we did.
How do you ensure that any changes are justified?
The AIR U.S. Hurricane Model was independently reviewed by three world-renowned hurricane researchers and six widely respected wind engineers. When all the components come together, the final model output is expected to be consistent with basic physical expectations of the underlying hazard, and unbiased when tested against both historical and real time information.
How will rating agencies react to companies wishing to increase their reliance on the AIR model?
To address the confusion and misinformation that existed in the market, we recently confirmed with A.M. Best Co. that a company that changes its reliance from one model to another does not face a de facto adverse impact on their rating. However, companies changing their reliance on one model over another should be prepared to demonstrate and explain their due diligence process.