icut.gif (1037 bytes)
intdots.gif (546 bytes)
white1.GIF (834 bytes)

Generations of Hurricanes

A vitalized symbol
Of earth and of storm,
Of Chaos contracted
To intricate form.

Oliver St. John Gogarty,
The Crab Tree

Hurricanes are the most serious of earthly storms. In the United States, damages from the rainfall of intense hurricanes continually top the charts for property losses, both insured and uninsured. But technology advances have improved survivability against these fierce storms—even as luxury coastal development has made property losses seem to skyrocket. In fact, hurricanes of the past were much deadlier. In 1938, for example, more than 600 people died in a hurricane that hit New England; and in 1900 more than 6,000 people were killed in a hurricane that pelted Galveston, Texas.

So far in 1999 there have been five intense hurricanes (wind speeds above 130 miles per hour, or categories 4 and 5 on the Saffir-Simpson scale), higher than the average of 2.2 per year over the last 50 years.  In searching for what caused this uptick in intense hurricanes, can we look to our usual suspect—greenhouse gas global warming?

First, we must review what is known and unknown about hurricanes. Over the last 50 years, keen observers noticed that the number of intense Atlantic hurricanes declined from 3 per year (1944-1964) to about 1.5 (1965–1994) (Figure 1). Since 1995, hurricane activity seems to have returned to its earlier, active phase, with the exception of 1997, when the strong El Niņo suppressed hurricane formation in the Atlantic.

Figure 1

Figure 1.  Number of intense (category 4 and 5) Atlantic hurricanes, 1944–1999 (blue line); updated from C. Landsea NOAA/AOML, Miami.  The change in annual Sahel region rainfall is also shown (dashed red line, courtesy of M. Hulme, University of East Anglia).

One of the keenest observers, William Gray of Colorado State University, has found an interesting link between the decadal pattern of intense Atlantic hurricanes and rainfall in the Western Sahel region of sub-Saharan Africa—sandwiched between the northern desert and the southern wet zones. Sahel rainfall matches the pattern of changes in intense Atlantic hurricanes. The connection might not be chance: Proto-hurricanes in their early, tropical-storm stage are spawned in the eastern Atlantic near Western Africa. Sahel rainfall does not create hurricanes per se; rather, it indicates conditions that favor tropical storm formation.

Tropical cyclones in the North Pacific also show a decadal change, according to P.–S. Chu and J. D. Clark (Figure 2). Between 1966 and 1981 fewer tropical storms and hurricanes formed than recently (1982–1997). As Chu and Clark note, the sudden, steplike jump in the North Pacific storm activity is unlikely to be connected to a more gradual, greenhouse gas warming.

Figure 2

Figure 2.  Number of Northern Pacific tropical storms and hurricanes, 1966–1997 (from Chu, University of Hawaii) (blue line), and change in eastern equatorial Pacific sea surface temperature (dashed red line; from Mitchell, University of Washington).

Is there a simple explanation for the change in Pacific storm activity? Chu and Clark suggest storm activity changes because the local sea surface temperature changes—higher local sea-surface temperature tends to promote tropical storm genesis there.

The sea surface temperature of the Eastern equatorial Pacific (6°N to 6°S, 180°W to 90°W), also plotted in Figure 2, compared with the number of Pacific storms, suggests such a correspondence. The jump in temperature starting circa 1982, beginning with the strong El Niņo of that year, initiated a period of two more storms per year than the average over the previous period.

The relationship between sea-surface temperature and storm frequency, while statistically significant, is not all that strong, as Figure 2 shows—further evidence for the complexity of the global warming/hurricane picture.

Just as the rise in the number of Pacific storms is too sudden to be related to a gradual greenhouse gas warming, so is the jump in Atlantic storms. The increases in storminess and temperature are unrelated between the two oceans because they describe only local conditions. If the independent storm patterns of both oceans follow the decadal averages they are currently showing, then both regions are currently in phase with conditions that tend to promote high storm activity. Coincidence? Absolutely.

These new empirical results on storm activity highlight the enormous difficulties in engaging in storm prediction for a greenhouse-enhanced climate. Modeling storminess requires paying attention to the small spatial scales on which storms are formed, while understanding influences of variability that occur on much broader scales such as the El Niņo–Southern Oscillation index.

From Princeton’s GFDL model, F. Vitart and colleagues point out some deficiencies of storm modeling. For example, the modeled storms are less intense and shorter-lived; and the distribution of storms disagrees with the observed locations. Notably, the model incorrectly predicts two storms per year spawning over land, and one storm per year in the south Atlantic, where none is seen. The most obvious disagreement between the model and reality is in the number of storms predicted for the Eastern North Pacific, three per year, compared with the recorded average of 17 per year.

Studying real-world storminess using a GCM is still a challenge, and models cannot yet match the observations. Model predictions about storms in a greenhouse gas–enhanced world remain premature and unreliable.

This bimonthly contribution is made possible by the George C Marshall Institute, Washington, D.C., where Sallie Baliunas is senior scientist and Willie Soon is a visiting fellow.


Chu. P.-S., and J.D. Clark, 1999, Decadal variations of tropical cyclone activity over the central north Pacific, Bulletin of the American Meteorological Society, 80, 1875–1881.

Vitart, F., et al., 1997, Simulation of interannual variability of tropical storm frequency in an ensemble of GCM integrations, Journal of Climate, 10, 745–760.