We Don’t Live in a Model World

With an inevitable return to something more akin to normal hurricane activity along the U.S. Atlantic and Gulf coasts there’s increased grumbling that this year’s experiences either are a consequence of global warming or indicators of what’s to come. Despite renowned hurricane experts using actual observations to rebut this notion, the claim persists (www.co2andclimate.org/wca/2004/wca_24a.html). Why? Because climate modeling indicates our future will be one where tropical cyclones become slightly stronger as the atmosphere’s greenhouse gas buildup heats things up.
     New York Times science writer Andrew Revkin on September 30, 2004 summarizes findings by Thomas Knutson and Robert Tuleya published in the Journal of Climate this way, “Global warming is likely to produce a significant increase in the intensity and rainfall of hurricanes in coming decades, according to the most comprehensive computer analysis done so far.” Never mind that the authors describe their findings differently. “CO2-induced tropical cyclone intensity changes are unlikely to be detectable in historical observations and will probably not be detectable for decades to come,” they write.
     The resolution of global climate models simply is too coarse to incorporate organized tropical systems. Heck, they barely ‘see’ the Rocky Mountains as influencing weather across the heartland. For that reason, Knutson and Tuleya begin their study with model projections of future sea-surface temperatures, vertical temperature profiles, and vertical moisture profiles over regions where tropical cyclones form. In this way they attempt to define a climate in which they can get away with using a finer-resolution hurricane model to spin up tropical cyclones. They go on to compare the characteristics of model-derived storms in a model-derived future climate with model-derived storms in our actual, observed climate.
     They conclude that in a model-derived future climate, model-derived hurricanes will increase 14 percent in their central pressure fall. Maximum surface winds will increase 6 percent and the average rate of precipitation within sixty miles of a storm center will increase 18 percent. This is by way comparing model-derived hurricanes under today’s conditions. “But what about actual hurricanes under today’s conditions?” astute readers ask. It should come as no surprise that the idealized setting of climate models does not come close to replicating the complexities of the world as we know and observe them.
     In Knutson’s and Tuleya’s modeled atmosphere, carbon dioxide levels increase at a rate of one percent per year, leading to a more than doubled concentration eighty years from now. Meanwhile, in the real world, the concentration of carbon dioxide grows at a rate of about 0.45 percent per year, or less than half that presumed by their model. At this rate, the atmospheric concentration of CO2 eighty years from now will have changed only one-third as much as assumed by the climate modelers. Please note, the rate of growth has been stable for three decades. There’s every indication trends toward increased energy efficiency and less-than-explosive population growth will keep things pretty much as they are into the foreseeable future. In other words, it will take the atmosphere nearly 180 years to reach the concentration the modeled climate achieves in eighty.
     Knutson’s and Tuleya’s modeled hurricanes come into being in a climate ideal for spawning theoretical storms. Theirs is a world virtually without vertical wind shear. Why is this important? Vertical wind shear interferes with development of tropical systems. In effect, it blows the tops off the storms and prevents them from becoming well-organized. El Niño is known to be responsible for increasing vertical wind shear in the tropical Atlantic. A number of studies demonstrate how tropical activity in the Atlantic Ocean decreases in years of El Niño, as does the chance that any storms that develop will make landfall in the United States.
     Many climate models suggest increased El Niño-like conditions in our future; many others don’t. Knutson and Tuleya sidestep the issue and assume not only that will there be no changes in wind shear but virtually no wind shear at all! This ‘other worldly’ assumption alone creates an idealized climate for developing strong hurricanes, one in which a storm’s strength largely is governed by the temperature of the underlying ocean’s surface.
     Knutson and Tuleya note the strong correlation between sea surface temperatures (SSTs) and hurricane intensity — the warmer the sea’s surface, the stronger the storm. Figure 1 shows this relationship between SSTs and hurricane intensity as depicted by Knutson and Tuleya. In their model, sea surface temperatures explain between 45 and 72 percent of the change in hurricane intensity. Because all climate models warm up oceans when the atmospheric CO2 concentration is enhanced (and even more so if it is unrealistically enhanced to levels that more than double within eighty years), more CO2 leads to higher SSTs, which lead to stronger tropical cyclones.

Figure 1. Relationship between sea-surface temperatures and hurricane intensity as measured by minimum central pressure (the lower the pressure the stronger the storm) in models used by Knutson and Tuleya.
    But, the real world is not so hospitable to fledgling tropical disturbances. While the temperature of the underlying ocean surface certainly is a critical factor in tropical cyclone development (the SST must be at least 80ºF for storms even to develop), other real-world factors such as wind shear affect a developing storm (see http://hurricane.atmos.colostate.edu/forecasts/ for more information on hurricane forecasting and the factors that determine hurricane strength and frequency).
     In our Figure 2, we plot the relationship between sea surface temperature in the region of the Atlantic used by Knutson and Tuleya versus two measures of hurricane intensity: average peak wind speed in the 5 strongest storms each year and the total number of intense hurricanes annually. Intense hurricanes are defined as Category 3, 4, or 5 storms. Compare our results with Figure 1.

Figure 2. The observed relationship between sea-surface temperatures and two measures of hurricane intensity: the number of major hurricanes (Category 3, 4, and 5) each year (left), and the average peak wind speed in the five strongest storms in each year (right). (Data source for hurricane information: Landsea et al., 1996 and updates).

    It’s an understatement to say Figure 1 exaggerates the situation. Observations indicate the relationship between hurricane intensity and SSTs to be not nearly as well-defined as implied by the model. In the real world, SSTs explain less than one-tenth of the annual variation in hurricane intensity (as compared with something like two-thirds of the variation in the model).
     If one uses real-world numbers instead of modeled ones, Knutson’s and Tuleya’s results must be adjusted downward. Here is our rough attempt to appropriately adjust them. CO2 concentrations in eighty years only will be about a third as great as those used by Knutson and Tuleya and because the relationship between warming and CO2 concentrations generally is linear, the SSTs will rise only about one-third as much. Given the strong association between SSTs and hurricane intensity in their model, this means the increase in intensity should be about a third as great. Their numbers should be adjusted downward to a 5 percent increase in central pressure fall, a 2 percent increase in the maximum surface wind, and 6 percent increase in average precipitation rate. But we’re not yet finished adjusting their findings.
     Instead of explaining about 66 percent of the variations in hurricane intensity (as indicated by Knutson and Tuleya’s model), SSTs explain only about 8 percent of the variation. This means that instead of SSTs playing the dominant role in determining hurricane intensity, SSTs play a minor role, leaving other more important factors to determine the strength of the storms. Putting all of this together means that no one could detect such minimal change in storm characteristics or impacts. But one thing is certain; such a real-world projection of change will never garner a headline in The New York Times.
     There you have it, another example of observational evidence failing to support a climate model’s dire conclusions concerning future climate. Again we’re compelled to ask, “What has happened to the peer-review process in climate science?”

Knutson, T.R., Tuleya, R.E., 2004. Impact of CO2-induced warming on simulated hurricane intensity and precipitations: sensitivity to the choice of climate model and convective parameterization. Journal of Climate, 17, 3477-3495.

Landsea, C.W., 1996. Downward trends in the frequency of intense Atlantic hurricanes during the past five decades. Geophysical Research Letters, 23, 1697-1700 (and updates).

Have comments or questions?
Contact us at


A Service of
The Greening Earth Society