Summary: | Hurricane track data from the NHC HURDAT best track dataset from 1900-2003 are analyzed with respect to landfalls along the east coast of Florida. Using a Markov Chain Monte Carlo method changepoint analysis technique, a changepoint in the landfalls along the coast of east Florida is identified in 1969. The changepoint represents a significant decline in the number of hurricane landfalls along the coast of east Florida starting in 1969 relative to the period 1900-1968. Overall Atlantic activity and U.S. landfalls are analyzed with respect to this date and it is found that despite an increase in hurricane activity in the Atlantic basin and little variance in the number of landfalls elsewhere in the U.S. between the two periods, landfalls along Florida's east coast have declined significantly. Using spatial intensity analysis, hurricane activity shows a marked decrease around the Florida peninsula, the Bahamas, and the western Greater Antilles since 1969, as compared to the period from 1900-1968. A domain is constructed that covers this area of decreased activity. Using NCEP/NCAR reanalysis, NOAA Extended Reconstructed SST, and NOAA/CPC SOI and NAO data, an analysis of the regional climate with respect to variables that have previously been found to be important to hurricane formation and distribution is performed. Reanalysis data for seven thermodynamic and dynamic parameters (SST, 500-700hPa average RH, surface air temperatures, 200-850hPa zonal vertical shear, 925hPa relative vorticity, 200hPa divergence, and sea-level pressure) as well as observations of the Southern Oscillation Index (SOI), the North Atlantic Oscillation (NAO) index, and the Quasi-Biennial Oscillation (QBO) are used to form yearly seasonal averages, covering the months of August through October over the domain from 1948-2003. These yearly seasonal averages are then used to construct a forward stepwise Poisson regression model that predicts the number of landfalls along the east coast of Florida. The variables that contribute significantly to ...
|