Synthesis of US and Caribbean Hurricane Impacts on a Warmer Globe

The future of humanity may well depend upon worldwide reduction of global greenhouse gas emissions. Because Tropical Cyclone (TC) landfalls are newsworthy and often tragic, advocates cite increased human and economic impacts to support low-emission policies. Skeptics argue that an unambiguous TC signal due to Anthropogenic Global Warming (AGW) has not yet emerged above large random variations. In this context it is essential to distinguish between political skepticism (or enthusiasm) that muddles the issue, and the scientific skepticism essential to nuanced understanding.

Thermodynamic arguments for increased Maximum Potential Intensity (MPI) as the sea warms are theoretically compelling and consistent with observations. By contrast, data issues may lead scientifically skeptical analysts to question pessimistic interpretations of some observational studies. Here, we seek clarity through in-situ statistics of hurricane occurrence and impacts.

Controversial “normalization” of continental-US hurricane damage for local population, individual wealth, and inflation yields sensibly zero 1900-2008 trend. A salient argument against this result is that stronger building codes and better forecasts reduce present-day damage. It is well documented that better forecasts have reduced US hurricane mortality by 90% compared with what would result with mid-20^th Century practice. The consensus among wind engineers is that Florida’s present High Velocity Hurricane Zone building code would mitigate 50% of wind damage from a recurrence of 1992’s Hurricane Andrew. Nonetheless, much of 21^st Century damage has resulted from storm surge and inland flooding rather than from wind, and “grandfathering” means that implementation of stronger building standards takes decades to make its way into the legacy housing stock.

Fitting an exponential trend line to 1900-2016 nominal (i.e., “then-year” no inflation correction) US hurricane damage yields an annual growth rate essentially the same as that of US nominal Gross Domestic Product. The statistical distributions of both normalized and detrended US damage are approximately log-normal, but somewhat leptokurtic and negatively skewed. The logarithmic standard deviations represent factors of approximately ten times the geometric mean seasonal damage. These values mean that damage would have to double on century timescales for the increase to become statistically significant.

From 1900-2016, there was a small, not statistically significant, increase of Accumulated Cyclone Energy (ACE) in a “US Waters” domain that encompasses the Gulf of Mexico and extends 500 km offshore along the East Coast to the Canadian border. This result contrasts with the reported substantial basin-wide increase of ACE, which can plausibly be attributed to improved observations of the remote North Atlantic. Nonetheless, a Caribbean domain exhibits a statistically significant increase of major hurricanes and a not-yet-significant increase of ACE, which may become significant when 2017 data are included. Historically, this domain has been heavily traveled by shipping, received priority for aircraft reconnaissance, and the analyzed time interval was chosen to avoid aliasing of the Atlantic Multidecadal Oscillation into the trend. Moreover it is in the Caribbean domain where one would expect the MPI response due to warming seas to be most pronounced. Hurricane Maria’s devastating impact in Puerto Rico is consistent with this interpretation.

Category 5 hurricanes in the Atlantic Basin fall into two classes. Those that made US landfall during the 20^th Century as category 5 hurricanes (the 1935 Labor Day Storm, Camille and Andrew) all intensified rapidly in the hours before US landfall. A different pattern emerges from examination of the 19 Atlantic category 5 hurricanes since 1977. Except for Andrew, 18 of them intensified rapidly over the Caribbean or the Gulf of Mexico loop current, often undergoing multiple eyewall replacements that increased their overall size. Of these, 12 then weakened by an average of two Saffir-Simpson categories before US landfall as major, but not extreme hurricanes. Only 6 missed the continental US entirely. Hurricane Irma of 2017 fit this pattern, as did Matthew of 2016, post-tropical storm Sandy of 2012, and Katrina, Rita and Wilma of 2005. This pattern suggests a nuanced interpretation of the evolving threat on a warming Earth. While there is no reason to expect the danger of US category 5 hurricane landfalls to diminish, it is reasonable to imagine that increases in Caribbean major hurricanes will result in both more windstorm disasters in the islands and more flooding disasters caused by formerly extreme, but currently weakening, Caribbean TCs that bring heavy rainfall and raise devastating storm surges on US shores. If this speculation stands the test of time, the Caribbean will experience more Marias and the continental US will experience more Harveys.