Visualize: Insights that power innovation

Preview of the 2017 Atlantic hurricane season

By Eric Uhlhorn, Ph.D., and Michal Clavner, Ph.D. June 22, 2017
From AIR Worldwide
Hurricane Matthew
Lumberton, NC, USA--October 11, 2016--High flood waters remain in neighborhoods impacted by Hurricane Matthew. Jocelyn Augustino/FEMA.

By convention, the Atlantic hurricane season begins June 1 and ends November 30, but these dates describe only when most tropical cyclones form; storms can develop earlier or later in the year. For the third year in a row a tropical cyclone formed in the Atlantic before June 1, but this year’s fizzled harmlessly mid-ocean the next day and proved to be a false start to the season. Given that early storms do not necessarily herald an active season, what can we expect this year?

Tropical cyclone predictors

Several factors influence tropical cyclone development, but one of the most important is the El Niño–Southern Oscillation (ENSO), which is described by the National Oceanic and Atmospheric Administration (NOAA) as “a periodic fluctuation in sea surface temperature and the air pressure of the overlying atmosphere across the equatorial Pacific Ocean.” ENSO can have a significant impact upon the world’s weather systems, and can either foster or hinder hurricane development. A period with elevated sea surface temperatures (SSTs) is known as El Niño, and the corresponding cool period is La Niña. El Niño tends to increase vertical wind shear in the Atlantic, and in doing so helps to temper tropical cyclone formation.

The recent La Niña was declared over in February 2017, but as SSTs increase in the equatorial region of the central and eastern Pacific, however, El Niño conditions are more likely than not to return later this season—unusually soon after the strong El Niño of 2015. Figure 1 shows the plume of ENSO predictions as of mid-May, with a weak El Niño indicated by the consensus (dotted line).

Figure 1

Another major influence is the Atlantic Multi-decadal Oscillation (AMO), a natural variability of SST in the North Atlantic. Alternate periods of warm or cool SST anomalies occur from the equator to Greenland and typically last 20 to 30 years. Warm periods correlate with increased tropical cyclone frequency and cool periods with less activity.

The current warm phase began in the mid-1990s, and unusually quiet hurricane seasons in 2013, 2014, and 2015 have led to speculation that the warm phase is transitioning into a cool phase. However, the relatively active hurricane season in 2016, including two U.S. landfalling hurricanes, has added to the uncertainty and it is still far from clear that a transition has begun. Moreover, the AMO index itself remains positive, with no indicative decreasing trend. Figure 2 shows the month-to-month variability in sea surface temperature anomaly from 1950 to the present.Figure 2

Table 1Uncertainty remains

Forecasts suggest average, or slightly elevated SSTs in the region of the Atlantic where most tropical cyclones form. However, secondary factors such as El Niño weather impacts, the Saharan Air Layer, the Madden-Julian Oscillation, and the North Atlantic Oscillation also influence hurricane activity.

Although scientists understand the links between these phenomena and hurricane formation, the inherent randomness and variability of hurricane formation and tracking toward landfall is a much greater influence on landfall risk than all of these factors put together. Even perfectly accurate SST forecasts would not make landfall risk more predictable, given the current state of knowledge.

Since 2007, AIR has released two catalogs for its U.S. hurricane model: a standard catalog that reflects hurricane risk under average climate conditions, and a supplemental catalog, which “conditions,” by way of climate indices, the standard catalog using data only from those years since 1900 when SSTs have been higher than the long-term average. It should be noted that because these indices are based on a subset of years, the warm sea surface temperature (WSST) catalog is characterized by higher uncertainty. Thus, it is offered as a supplement to, rather than a replacement for, the standard catalog.

Compared to the standard catalog, the WSST view of risk is approximately 10% higher in hurricane counts nationwide, although regional differences can be higher still. The actual storm count has tracked near or between the standard and WSST views, dipping below the standard view some years and rising above the WSST view in others.

What will the 2017 hurricane season bring?

The moderate El Niño previously forecast for the peak of the season is now likely to be a weak one, if it materializes at all, and forecasts for hurricane activity have been revised to slightly above average levels. It is important to remember, however, that above average activity does not necessarily presage elevated insurance losses. There is no strong correlation between the number of tropical storms that form and U.S. landfalls in any given season. An active season can produce many storms but few U.S. landfalls.

And while not all landfalling systems impact exposure and produce significant damage, just one major hurricane can inflict record-breaking losses—as Hurricane Andrew did 25 years ago. Conversely, a series of tropical storms and lesser hurricanes can have as much cumulative impact on the insurance industry as a major hurricane. Furthermore, as Hurricane Matthew so ably demonstrated last year as it moved northward along the coast of Florida just offshore, even a bypassing storm can inflict significant damage.

It is tempting at the outset of another Atlantic hurricane season to try to extrapolate a picture of the season’s risk from recent climatic activity. Yet the complexity underlying landfalling tropical cyclone risk is the very reason that researchers at AIR have long advocated for a robust approach to estimating risk based on as many years of data as possible.

AIR’s standard and WSST catalogs both provide a long-term and stable view of risk from landfalling hurricanes. Each incorporates the latest scientific research and has undergone extensive peer review by leading scientists. By providing two catalogs for the U.S. Hurricane Model, AIR encourages clients to assess variability and uncertainty, which are fundamental to managing risk. One thing is certain: seasonal forecasting is inherently uncertain and, from a loss standpoint, anything can happen.


Eric Uhlhorn, Ph.D., is a manager and principal scientist and Michal Clavner, Ph.D., is a scientist with AIR Worldwide.