News Section
Stories from Climate Central's Science Journalists and Content Partners

Hurricanes Likely to Get Stronger & More Frequent: Study

Hurricanes are Mother Nature’s largest and most destructive storms. Fed by warm ocean waters and moist atmospheric conditions, about 90 such storms — also known as tropical cyclones — form worldwide each year. With the population of coastal areas growing daily and sea level on the rise, how these monster storms may change as the climate continues to warm is an increasingly urgent question facing climate scientists, insurance companies, and public officials.

The North Atlantic Ocean Basin has been in a more active period of tropical cyclones since 1995.
Credit: Climate Central.

A new study by Kerry Emanuel, a prominent hurricane researcher at MIT, found that contrary to previous findings, tropical cyclones are likely to become both stronger and more frequent in the years to come, especially in the western North Pacific, where storms can devastate the heavily populated coastlines of Asian nations. Emanuel's research showed the same holds true for the North Atlantic, where about 12 percent of the world's tropical cyclones spin each year.

Emanuel's study casts doubt on what had been the consensus view of most climate scientists — that in most ocean basins, tropical cyclones are likely to become less frequent as the world warms, but that the storms that do occur are likely to contain stronger winds and heavier rains. That view was expressed most recently in a 2012 report from the U.N. Intergovernmental Panel on Climate Change.

Emanuel's study, published Monday in the Proceedings of the National Academy of Sciences, uses the latest generation of global climate models to power a series of high-resolution, regional simulations of tropical cyclones around the world.

The study compared the frequency and intensity of tropical cyclones during the period from 1950 to 2005 against projections for the 21st century, from 2006 to 2100, using a scenario in which global emissions of greenhouse gases, such as carbon dioxide, rise rapidly without being significantly curtailed.

Emanuel’s study notes that the relationship between global climate and hurricane activity is “only beginning to be understood.” For example, scientists still don’t understand why 90 tropical cyclones tend to form worldwide each year, and not, say, 125, or 50. Researchers know that tropical storms and hurricanes derive their energy from warm sea-surface temperatures as well as temperature differences between the sea and the overlying, moisture-rich tropical atmosphere, but figuring out how changing background conditions may affect storms in the future is no easy task. In part, this is because although hurricanes can span hundreds of miles in diameter, they still exist at the very limits of computing power for global climate models to accurately simulate. That is particularly the case for such storms' inner cores, which can be just 20 to 40 miles in diameter and contain the strongest winds.

Emanuel’s study attempts to overcome this problem by embedding high-resolution regional and local models within the broader global modeling framework, which is a frequently-used approach known as “downscaling.” This method — which is akin to constructing a building on top of a preexisting structure, depending on the larger structure for support but being limited by its design — is far from foolproof, and can be prone to mismatches between the underlying calculations used for the global model and the calculations used for the other models. Nevertheless, downscaling is a valuable research tool for hurricane specialists because it can better capture the inner workings of a storm’s core.

In this study, the modeling results showed a significant increase in one key measure of hurricane intensity, known as the “Power Dissipation Index,” which captures both a storm's windspeed and total lifetime. Power dissipation increased by 45 percent during the 21st century simulations the study found, and there was a 40 percent global increase in major hurricanes, of Category 3 intensity or greater.

Projected changes in tropical cyclone track density during the 2006-2100 period compared to the 1950-2005 period.
Credit: Emanuel, PNAS.

Emanuel said the differences between the results from the newer generation of computer models and the older generation may have to do with projected changes in manmade emissions of sulfate aerosols, which recent research suggests may have hindered tropical cyclone activity in the Atlantic Ocean during the 1970s and 1980s.

“ . . . It does go against the consensus which projects (a) declining frequency of global tropical cyclone activity,” Emanuel said in an email to Climate Central. “ This consensus, though, is based mostly on the previous generation of climate models . . . and my technique, applied to those earlier models, shows little change in global tropical cyclone frequency.

“We are not yet sure why the same technique, applied to the current generation of global climate models (CMIP5) shows an increase in activity, though I suspect it has more to do with projected decreases in manmade aerosols than with increasing carbon dioxide (emissions),” Emanuel said.

There are, however, two key points of agreement between Emanuel’s new study and the rest of the scientific literature on hurricanes and climate change. The first is that as ocean temperatures continue to increase, the frequency of high intensity hurricanes is projected to increase as well.

“In practice, these events cause most of the damage,” Emanuel said. For example, Hurricanes Katrina, Rita, and Wilma — all of which were major hurricanes according to the Saffir-Simpson Scale — caused nearly $200 billion in damage alone in 2005, which was the most recent year a major hurricane made landfall in the U.S. With global sea levels on the rise due to climate change, major hurricanes will be capable of doing even more harm in the decades to come, battering coastlines with higher storm surges than they have in the past.

Second, virtually all the studies show that tropical cyclones are likely to dump significantly more rainfall in coming years compared to the historical record, and inland flooding is one of the leading causes of hurricane-related deaths and damage.

James Elsner, an atmospheric scientist at Florida State University who was not involved in this study, downplayed the study’s conclusions given the considerable uncertainties involved with using computer models to simulate complex storms such as hurricanes.

“The results from the new Emanuel are provocative, but in my opinion there is little reason to put much weight on them when considering what might happen to tropical cyclone activity during the next 50 to 80 years,” he said in an email to Climate Central. “This kind of predictive risk analysis can be useful, but it would benefit greatly by being grounded in risk metrics computed from observed data.”

Related Content
Hurricane Tracks: 10 Recent Major Storms to Hit the U.S.
In Wake of Sandy, NOAA Alters Warning Policy
isks of Hurricane Sandy-Like Surge Events Rising
2013 Hurricane Season Could be Extremely Active: NOAA
Top 5 Most Vulnerable U.S. Cities to Hurricanes
Surging Seas: Sea Level Rise Maps from Climate Central


By Manfred P. Gebhard (Rüsselsheim/GERMANY/D-65428)
on July 8th, 2013

That’s essentialy what I heard in lectures at the “Johannes Gutenberg Universität” in Mainz, Germany in the 80’s.
More CO2, more weatherextremes, like more or heavier hurricanes.
Global action since then: inadequate.
However, nice puplication.

Reply to this comment

By Eric Peterson (Front Royal, VA 22630)
on July 8th, 2013

How would you explain the fact that we are in the longest period on record without a major hurricane strike on the US.  My view is it is simply luck: with irregular periodicity of small numbers of events there can be a wide variation in the time between events.  But are there specific weather or climate explanations?

Reply to this comment

By David Schnare (Burke, VA 22015)
on July 9th, 2013

Without grounding models in actual data, the models have little value. Those who suggest model outputs are sufficient to drive policy are no better than the boy who cried wolf.

Reply to this comment

By Larry (Cleveland)
on July 9th, 2013

David -

1. Global Climate Models ARE grounded in reality

2. So what would you have us do?  Stick our heads in the sand and wait until a strom comes along and drowns us?  Prudent risk managment dictates that if the consensus is stronger storms and there is also the potential of more frequent storms, we determine a policy now that can help us mitigate and adapt.

Reply to this comment

By Martin Vermeer
on July 9th, 2013

This David Schnare?

Reply to this comment

By Joe
on July 9th, 2013

NOAA, using 20 predictive models and supercomputers, can’t predict a hurricane track out past 72 hours, yeah, stick with those geniuses.

This is a test: go look at the 20+ hurricane models used to predict that path of hurricanes, the LARGEST event known. Find me ONE, just ONE, hurricane model that is predicting the complete path of a hurricane with 100% accuracy.

NEWS FLASH!  Hurricane strikes are declining in the US: See Eleven Year Running Mean Of Annual Hurricane Strikes

Reply to this comment

By Jay Alt ( IL)
on May 18th, 2014

US / NOAA are short on computing power & forecast programs.

The European mid range forecast model (ECMF) worked well.  It forecast that a Greenland blocking ridge would turn Sandy’s track back into the East Coast.
Their forecast of a NYC/ NJ strike was made 7 days before landfall.

And it took slow NOAA programs 3 more days to converge on the same conclusion.

Reply to this comment

By Charles
on July 10th, 2013

What about the trends of the 30s, and 60s?  Named storms, but not unnamed, subtropical?  Genesis should be considered in all levels of intensity.  Also, the climate central chart of trended averages has a distribution skew.  What about Quikscat?  Landsea found that upon reanalyzing TC counts, the number of storms between 1900 and 1966 increased an average of 3.2 for that time period, and 1.0 between 1966 and 2002 for that time period.  This raises the 1900-2006 mean from 9.2 to 11.5 storms per year during these time periods (Landsea 2007).

Also, Elsner et al. (1999) explain biennial and semi-decadal oscillations in NATCs are influenced by a shift in tropical stratospheric winds at the 30 mb level (quasi-biennial oscillation, or QBO), as well as warm North Atlantic SSTs during La Nina, and a recently discovered near-decadal oscillation, which enhances baroclinically-enhanced TCs based upon North Atlantic SST evaporation rates and solar activity.

Kimberlain and Elsner (1998) also use a more climatological approach to explain 1990s TC activity.  They show that a prolonged ENSO warming period over the Pacific, produced an anomalous Walker circulation and thus Atlantic subsidence, as well as limiting mid-tropospheric humidity, and in turn tropical development.  They also state that tropical upper tropospheric troughs, or TUTTs, were more numerous during the 1990s, increasing shear and thus limiting development.  Combined, it was shown that these patterns accounted for 58% of the variability in NATCs.  Further research showed that Caribbean SSTs were high during 1995 and 1996, and this favored more development during these years as (Kimberlain and Elsner, 1998).  Goldenberg et al. (2001) used similar climatology to explain the increase in activity of NATC in the later 1990s.  They also found that the increase was due to the decreased vertical wind shear, associated with La Nina, which increased Gulf and Caribbean storms during those times. 

Even Shepherd and Knutson conclude that climatology factors largely dictate NATC behavior, and caution that more research is needed on the effects of anthropogenic warming on TC behavior…  And Trenberth (2005) in “Uncertainty in Hurricanes and Global Warming,” argues how statistics may prove that increased greenhouse gases (GHGs) may not necessarily contribute to stronger TCs, but that the sample chosen and data collected may be inadequate.  He shows how increased CO2 may have warmed the atmosphere especially since 1995, which may have increased North Atlantic SSTs and total column water, and thus convective available potential energy (CAPE) in developing North Atlantic thunderstorms.  This is explained to have raised the accumulated cyclone energy (ACE) index to 169% of the median between 1995 and 2004.  It is also explained that this may lead to increased rainfall in NATCs, and more extreme TCs in model runs, as concurrent with the most recent IPCC report (Trenberth 2005).  However, he delicately proposes that although this may be due to increased greenhouse gases (GHGs) and the increased heat capacity in the atmosphere, the anthropogenic factor is hard to prove directly.  Interestingly, there is no mention on an increase or decrease in TC counts.

Come on guys, do some real research.  This came out of my MS thesis, but it’s all citation.  Armchair “climate scientists” don’t understand GCMs…  Spatial statistics are never smoothed or projected properly, and even with x10 CO2 ensemble runs, the biggest appreciable change to North Atlantic tropical cyclones is a 33% increase in the precipitable water column (Knutson and Tuleya 2004).  It was found that a 2.2°C to 2.7°C increase in SSTs due to increased CO2, augmented TC wind speeds by up to 5 to 6% in the North Atlantic, and 3-10% across all basins, by the same study.

The trends with CO2 forcing on TCs for the Atlantic is two-fold: a decrease in FREQUENCY, and an increase in INTENSITY.  All other basins keep a neutral trend in frequency, but also increase in intensity.

Reply to this comment

By PunditPete (Burlington MA 01823)
on July 10th, 2013

Re: Joe’s test—

The study in question is trying to predict the rate of hurricane occurrence and the rate of major damage from hurricanes. This is different from trying to predict the outcome of one particular storm.  It is very hard to decide if an individual will commit suicide, but the rate of suicides can be estimated within reasonable error limits.

Reply to this comment

By FishOutofWater (Pittsboro, NC)
on July 10th, 2013

The subtropical high pressure area in the north Atlantic has been centered north and east of its normal position for the past 3 hurricane seasons allowing hurricanes to track up the eastern seaboard or out to sea. Climate change could be a factor in this tendency of the high to be displaced north and east but it would be incorrect to attribute short term variability in storm tracks to anything but random chance.

Emanuel’s body of work has been closely tied to empirical data. Numerical physical models are the only effective way to make scientific predictions of the effects of climate change on hurricane numbers, tracks and intensity. The criticisms miss the mark.

Reply to this comment

By Eric Peterson (Front Royal, VA 22630)
on July 11th, 2013

PunditPete, Joe’s point is valid. 

From the article: “Emanuel’s study attempts to overcome this problem by embedding high-resolution regional and local models within the broader global modeling framework, which is a frequently-used approach known as “downscaling.””

The downscaled model has to actually predict the path and strength of each hurricane as a weather event.  There can be no statistical results from the broader model about strong storms (or realistically any storms) without the downscaling.  The downscaling is the same as the weather models mentioned by Joe, specifically they use a snapshot from the broader model to initialize the weather model.  Then the weather model runs just like the ones used to (inexactly) predict hurricanes.

One fundamental problem of climate models is lack of fine resolution to model weather features like convective complexes and tropical storms which are really just a set of convective complexes.  What they typically do instead is parameterize.  This is one of the many parameters that control the model “sensitivity” and other predictions.  Tweak up the storm strength and get lower sensitivity, tweak down and get higher sensitivity.  Downscaling is a potential solution to the problem of lack of resolution.  However, as implied by the lack of mention in this article, the results of the downscaled model are not “upscaled” and incorporated into the broader model.  Therefore as the hurricane cools the planet (as each one does ever so slightly), that cooling is not incorporated into the model to offset CO2 warming.

Reply to this comment

Name (required):
Email (required):
Enter the word "climate" in the box below:

[+] View our comment guidelines.

Please note: Comment moderation is enabled. Your comment will not appear until reviewed by Climate Central staff. Thank you for your patience.