Hurricane Wilma wreaked havoc when striking Florida in 2005, being the last of four major hurricanes to affect the United States in that same year. Few would have imagined what followed: 10 years without a major hurricane landfall along the US coast!
In particular after the 2004/2005 hurricane seasons, there was an almost unanimous consensus that hurricane activity in the North Atlantic was elevated compared to the long-term average. Correspondingly, all hurricane risk assessment models used in the (re-)insurance industry offered a near-term risk view reflecting increased activity levels.
The hypothesis of increased hurricane activity level is generally linked to the Atlantic Multidecadal Oscillation (AMO). The AMO describes a cyclical warming and cooling of sea surface temperatures (SST) in the Atlantic Ocean on a multi-decadal scale. Since the mid-90's, we're considered to be in a "warm" phase with above average SSTs. Looking back over the last century, there is a significant correlation between warm AMO phases and increased hurricane activity (Goldenberg et al., 2001).
Last year's seasonal hurricane forecasts by NOAA contained the statement "The current SST anomaly pattern … suggest that the warm AMO signal is absent this season", and a scientific article by three well-known hurricane researchers published in a leading journal asked: "Active Atlantic hurricane era at its end?" (Klotzbach et al, 2015). While NOAA linked its statement to the specific 2015 El Niño season, the scientific paper raised valid questions but provided no conclusive answers. However, both helped to revive the discussion around hurricane activity levels, a topic of significant relevance for the (re-)insurance industry.
From a personal standpoint, here some key thoughts around the topic:
1. The AMO link to hurricane activity sounds simple and straightforward: higher SST = higher hurricane activity. Obviously, the way hurricanes form (or are inhibited) is more complex, with wind shear, atmospheric stability, air humidity and other factors playing important roles beside SSTs. Still, an AMO index from NOAA - basically the monthly SST averages north of the equator (link) - does not seem to show any obvious weakening trend currently, i.e. this particular AMO index remains clearly in a warm phase.
2. It has been proposed that the thermohaline circulation (THC) in the oceans drives the AMO. However, given scarce historical data on ocean circulation, such a link is not generally accepted. Over the past decade, the THC seems to have weakened (Srokosz & Bryden, 2015). So if (a) this weakening persists and if (b) the link to AMO is for real, then hurricane activity could be on its way back towards long-term average levels.
3. Climate models indicate that the below-average hurricane activity from 1970-1994 may have been significantly influenced by air pollution (Dunstone et al., 2013). If this were a significant factor, then the elevated activity since around 1995 would not be a cyclic feature connected to AMO, but rather the "new normal" of hurricane activity in today's climate environment.
4. Often ignored, the AMO signal is linearly de-trended for global SST warming. In simple terms: the "warm" AMO phase in 1930 was similar (in absolute °C/°F) to the "cool" AMO phase in 1990. Looking at hurricanes from a purely thermo-dynamical point of view the de-trending does not seem intuitive. But research suggests that relative differences in SST between ocean basins are more important for hurricane activity than absolute SST levels (Vecchi & Soden, 2007). Still, climate models generally suggest that maximum wind speeds – and thus the number of intense hurricanes – will likely increase in a world with warmer SSTs (IPCC, 2014). There is low confidence in such projections, but it shows that absolute temperature changes ignored in AMO signals may also influence hurricane activity.
5. The absence of major hurricane landfalls in the U.S. over the last decade is remarkable. However, a recent paper showed that even small changes in the hurricane intensity metric used for counting landfalls lets the U.S. "hurricane drought" in the last decade all but disappear (Hart et al., 2015). The particular "U.S. major hurricane landfall" view obscures an overall active hurricane decade in the full Atlantic basin.
Above is a small, subjective selection of current research. It highlights that a variety of factors are at play in a complex atmosphere/ocean environment. While a significant correlation between AMO and hurricane numbers exists, there seems little scientific consensus as to the robustness and cause of it. This is exciting news with respect to academic curiosity, but certainly a challenge for business decision makers.
Faced with the task to make (re-)insurance hurricane risk assessments, I tend to stick to the simple facts: firstly, SSTs play an important role in hurricane formation; secondly, since 1995 SSTs have been and continue to be elevated in the NOAA AMO index; thirdly, this AMO index and hurricane numbers show significant correlation. So based on these three high-level considerations, it appears premature to expect lower hurricane numbers just yet. The era of elevated hurricane activity levels may well linger for a while longer.
Category: Climate/natural disasters: Disaster risk, Floods/storms