SEMINAR: Geotopics today!

[Paul Hagan <phagan@rsmas.miami.edu>]

Geotopics today, 2/18

Dr. Hugh Willoughby and student Javiera Hernandez, FIU

"Statistics of Hurricane Impacts" 

3:15pm slab seminar - refreshments at 3pm

Abstract

Historical US damage and loss of life since 1900 are reasonably well modeled with log-normal distribution if one normalizes for economic factors. The most extreme damage, either seasonally aggregated or from individual storms, however, obeys a Pareto (negative power law) distribution. For example, the most destructive ten seasons caused about 2/3 of damage since 1900. Both earthquakes and tsunami have been modeled with Pareto-like distributions.

 

In contrast with many other geophysical hazards, extreme tropical cyclones are not self similar. It’s not true that the big ones are not like the little ones, only bigger. Hurricane’s maximum  intensity is limited thermodynamically by the oceanic heat source, and their size by the requirement that the Rossby number (ratio of storm-relative rotation to planetary rotation) be large.

 

Consequently, the “fat-tail” distribution seems to be inherited from the size distribution of assets at risk. It is generally known among geographers that urban populations, and (by hypothesis) assets obey  Zipf distributions---equivalent to Pareto distributions with exponent negative one. Simple Monte-Carlo simulations in which random simulated hurricanes strike “Zipfistan,” a virtual country with Zipf-distributed populated places scattered at random, show promise for testing this hypothesis. 

Paul Hagan
Graduate Student
UMiami RSMAS MGG
4600 Rickenbacker Causeway
Miami, Florida 33149 USA
phagan@rsmas.miami.edu
231-835-0100