Skip to main content

<i>Science</i>: U.S. Sees Increase in Days with Multiple Tornadoes



On 27 April 2011, 211 tornadoes (warnings shown in red) touched down in the southeastern U.S. These deadly and costly multiple-tornado days are becoming more frequent, a new study suggests. | National Weather Service

Since 1954, there has been a decrease in the number of days per year in the U.S. with any tornadoes, but an increase in the number of days with many tornadoes, according to a new study published in the 17 October issue of journal Science.

"It was surprising to see," said the study's lead author Harold Brooks, senior research scientist at NOAA's National Severe Storms Laboratory. "I would naively expect these trends to go in the same direction."

If the tornado clustering Brooks and his team describe continues to increase, it could lead to an even greater concentration of damaging tornadoes on fewer days. This could impact emergency preparedness crews. "They would have to gather enough resources to handle the big days," Brooks said, "but those resources might sit around a lot."

The increased vulnerability could also impact reinsurance companies, which provide insurance to insurers. "Reinsurance companies have to have reserves to pay off large catastrophe claims quickly," Brooks explained. "If catastrophic payouts become a larger fraction of the total insurance, asset management might have to change."

Brooks and colleagues also observed that despite speculation climate change is causing more tornadoes, the overall number of tornadoes each year since 1954 has remained relatively constant.

The impact of climate change on the activity of U.S. tornadoes has become a hot topic among scientists and the public. To determine if and how numbers of tornadoes have evolved over time, Brooks and his team compiled data on U.S. tornado activity between 1954 and 2013 using the official U.S. tornado database.

Historically, interpreting this database has been difficult for scientists. Doing so requires separating meteorological effects on tornado activity, like favorable atmospheric conditions, from non-meteorological effects, like changes in how storms are reported.

Reports of the weakest tornadoes began to climb in the late 1980s and early 1990s, about the time the National Weather Service began issuing severe thunderstorm warnings, which brought about increased awareness and reporting. The installation of Doppler radar systems also increased coverage and observation of tornadoes, as did the presence of greater numbers of people in once-rural areas.

All of these factors, Brooks explained, resulted in a dramatic increase in the number of weak tornadoes in the database — about 100 per year in the 1950s compared to 800 per year now. Weak tornadoes are events on the Enhanced Fujita scale with winds between 65 and 85 miles per hour, labeled EF0 events.

Not only has the number of very weak tornadoes reported gone up, Brooks explained, the number of strong tornadoes in the database also has been affected, by changes in how path width is measured and damage is assessed.

To overcome these limitations, Brooks and colleagues developed a new method for analyzing the database that focuses on the more middle-of-the-road tornado events. "The big thing we've done is concentrated on what seems to be the best part of the database, the occurrence of EF1 and stronger tornadoes," Brooks explained. "Essentially we're focusing on a limited part of the database that appears to have been reported relatively consistently over the years."

Although the annual number of EF1 or greater tornadoes remained relatively constant over the period evaluated, the researchers did see that tornadoes have been clustering more since the 1970s. A single day in April 2011, for example, featured 211 tornado events.

This variability is not likely to result from changes in the way storms are reported, the researchers say, though it's unclear just what environmental conditions — a changing climate or otherwise — are driving the clustering.

"In the absence of having a physical mechanism we can't be sure it isn't random, but the trends are highly suggestive it's not random," Brooks said.

The results are motivating scientists to reconsider climate model projections and other data to search for possible physical mechanisms. Over the last few decades, for example, researchers have discussed changes in large-scale atmospheric conditions relevant to major destructive tornado events. Such conditions include vertical wind shear, the difference in wind speed and direction over a relatively short distance, and atmospheric instability that can create air turbulence and storms.

To uncover any links between atmospheric variability, climate, and tornado activity, Brooks said he would like to learn more about the last 40 to 50 years' worth of atmospheric conditions, to see how they have changed over time.