On the Changes in the Number and Intensity of North Atlantic Tropical Cyclones

Abstract

Bayesian statistical models were developed for the number of tropical cyclones, the rate at which these cyclones became hurricanes, and the rate at which hurricanes became category 4+ storms in the North Atlantic using data from 1966 to 2006 and from 1975 to 2006. It is found that, controlling for the cold tongue index (CTI), North Atlantic Oscillation index (NAOI), and the Atlantic Multidecadal Oscillation (AMO), it is improbable that the number of tropical cyclones has linearly increased since 1966, but that the number has increased since 1975. The differences between these two results have to do with the numbers of storms at the start of these two periods: it was easier to say a linear increase was present starting from circa 1975 since the storms in that period were at a low point. The rate at which storms become hurricanes appears to have decreased, and the rate at which category 4+ storms evolved from hurricanes appears to have increased. Both of these results are also dependent on the starting year. Storm intensity was also investigated by measuring the distribution of individual storm lifetimes in days, storm track length, and Emanuel?s power dissipation index. Little evidence was found that mean individual storm intensity has changed through time, but it is noted that the variability of intensity has certainly increased. Any increase in cumulative yearly storm intensity and potential destructiveness is therefore due to the increasing number of storms and not due to any increase in the intensity of individual storms. CTI was not always significant, but lower CTIs were associated with more storms, higher rates of conversion, and higher intensities. NAOI was only weakly associated: the effect was negative for the number of storms, the rate of hurricanes evolving from storms, and intensity, but it was positive for the rate of category 4+ storms evolving from hurricanes. AMO was rarely significant except in explaining the number of storms using the 1966?2006 data. Its direction was always positive as expected; however, higher values of the AMO were associated with more storms, higher rates of conversion, and higher intensities.