A question has been popping up quite often recently: We have now had a nine-year stretch without a hurricane landfall in Florida. Is this (ab)normal?
An easy first step in such an investigation is to look back into the history of hurricane landfalls in Florida since 1900, when reporting of hurricane landfalls became reasonably reliable.
A review of Florida hurricanes in the historical data (HURDAT/HURDAT2) published by the National Oceanic and Atmospheric Administration's Hurricane Research Division (NOAA/HRD) reveals that we are indeed in the midst of the longest hurricane landfall "drought"-nine years since Hurricane Wilma (2005). Other top-ranked stretches of hurricane inactivity in Florida are:
- Six years-occurred once between 1979 (David) and 1985 (Kate)
- Five years-occurred three times: 1910-1915; 1987 (Floyd)-1992(Andrew); and 1999 (Irene)-2004 (Charley)
- Four years-occurred six times: 1917-1921; 1929-1933; 1956(Flossy)-1960 (Donna); 1960 (Donna)-1964 (Cleo); 1968 (Inez)-1972 (Agnes); and 1975 (Eloise)-1979 (David)
Hurricane Wilma was the last Florida (and U.S.) landfalling major hurricane (a "major" hurricane is defined as Category 3 or higher). Therefore, the current period is also one with no major hurricane landfall for nine years. This nine-year stretch between major landfalling hurricanes in Florida is common: the longest major hurricane drought is 17 years-between 1975 (Eloise, a Category 3 hurricane) and 1992 (Andrew, a Category 4 hurricane). Other top-ranked stretches without a major hurricane landfall in Florida are:
- 10 years-occurred twice: 1950 (King)-1960 (Donna); and 1965 (Betsy)-1975 (Eloise)
- Nine years-occurred twice: 1995 (Opal)-2004 (Charley); 2005 (Wilma) to 2014
- Eight years-occurred once: 1936-1944
Given this kind of variability, we wanted to look at whether such stretches exist in AIR's U.S. hurricane model's catalog.
The short answer is yes.
The table below gives a comparison of the number of no-Florida-landfall years in the AIR U.S. hurricane catalog and the historical data. It demonstrates that the AIR catalog is fairly consistent with the historical record.
Number of no-Florida-landfall stretches for all hurricanes and for major hurricanes, respectively, renormalized to 100 years, AIR 50K Standard Stochastic Catalog
vs. historical data
| Stretch (Years) | All Hurricanes | Major Hurricanes | ||
|---|---|---|---|---|
| AIR Catalog (50K) | Historical Data (1900-) | AIR Catalog (50K) | Historical Data (1900-) | |
| ≥4 | 7.8 | 9.7 | 10.4 | 10.6 |
| ≥5 | 4.4 | 4.4 | 7.9 | 8.8 |
| ≥6 | 2.5 | 1.8 | 6.1 | 7.1 |
| ≥7 | 1.4 | 0.9 | 4.7 | 6.2 |
| ≥8 | 0.8 | 0.9 | 3.6 | 5.3 |
| ≥9 | 0.5 | 0.9 | 2.8 | 4.4 |
The ability to generate a realistic and yet balanced U.S. hurricane catalog-including landfall frequency as well as lapse period-largely resides in the fact that the annual frequency of hurricane landfalls is modeled using a negative binomial distribution. An important property of a negative binomial distribution is that the variance is greater than the mean. As a result, both the number of years with many landfalls and the number of years with no landfalls is higher relative to a Poisson distribution.
While the underlying drivers behind this clustering are related to the climatological/meteorological changes and are yet to be fully understood, the end results appear to be well captured using a statistical-based stochastic hurricane model.
A final note is that we update the model's catalog each time there is a major update of the historical data (by NOAA/HRD). Therefore, these climatological/meteorological signals, including variability, are well-reflected in the AIR U.S. hurricane catalog.
