The speaker is an expert on the study of hurricanes in the Atlantic Basin, Dr. Ron Iman, past president of the American Statistical Association, and currently President and CEO of Southwest Technology Consultants in Albuquerque. Dr. Iman will examine models the insurance companies use to project loss cost associated with hurricanes. More information may be obtained from Dr. Conover at jay.conover@ttu.edu.
Abstract of Dr. Iman's presentation:
Quantitative records for the Atlantic basin date back to the mid-nineteenth century. Mann and Emanuel (2006) used these data to find a positive correlation between sea surface temperature and the frequency of tropical cyclones from 1871-2005. Holland and Webster (2007) analyzed frequencies of Atlantic basin tropical cyclones dating back to 1855 and found a doubling of tropical cyclones over the past 100 years. Both papers attributed these results to anthropogenic greenhouse warming. Neither of these papers gave an indication of the uncertainty in their results and both presumed that tropical cyclone frequencies are complete, or nearly so, for at least the past century.
Landsea (2007) has shown that the presumption of complete or nearly complete counts is not reasonable and attributes the “observed increase” in frequency to improved monitoring (geostationary satellite imagery, etc.).
There are several other factors that actually contribute to the “increase” in observed frequencies: (1) the rush to “name” storms such as occurred in 2007 with Andrea and Barry, neither of which was a tropical cyclone and as such, would have been the topic of local forecasts but otherwise hardly mentioned had they occurred in the winter; (2) the National Hurricane Center (NHC) tends to round up wind speed observations to the nearest 5kts; (3) in the past 10 years the NHC counts a cyclone as a hurricane if there is a single gust of at least 74mph, whereas in the past multiple measurements were required; and (4) advanced microwave sounding and the cyclone phase space analyses recognized that four storms from 2003 to 2006 were tropical cyclones and thereby included in the Atlantic database – these storms would have been considered extratropical cyclones and therefore not counted as tropical cyclones prior to the start of the 21st century.
This presentation examines Landsea’s results and offers further modifications which demonstrate that a high percentage of tropical cyclones that remained over the ocean were not counted prior to geostationary satellite imaging. An overview of commercial modeling of hurricanes used for insurance rate filing will also be presented.
