Untitled Document

Trends and Innovations
Storm of progress for tornado forecasts

On Aug. 13, two people died when a tornado careened through the coal-mining town of Wright, Wyo., injuring at least a dozen people and destroying about 50 mobile homes. Residents had only five minutes to prepare.

Doppler radar captured this large tornado as it hit the General Motors plant in Oklahoma City, Okla., May 8, 2003. Meteorological software was used to provide a warning 30 minutes in advance — long enough to safely move 1,200 employees to a shelter. Image courtesy of NOAA.

In 2004, 1,819 tornadoes were recorded in the United States, according to the National Oceanic and Atmospheric Administration (NOAA) Storm Prediction Center, and from this total, 20 were classified as “killer tornadoes” and responsible for 36 deaths. Most occurred during tornado season, which is April to June, and in a region called Tornado Alley — an area spanning from the Dakotas to the Gulf Coast, bordered on the west by the Rocky Mountains and on the east by the Appalachians. But tornadoes can form at any time or location if conditions are right; Hurricane Katrina was responsible for 36 confirmed twisters. The challenge in forecasting tornados is that scientists have yet to fully understand their physics and the conditions that spawn them.

“In terms of using computer models to predict tornadoes, like we currently use computer models to predict weather seven to 10 days in advance, we’re not there yet, and probably won’t be there for a long, long time,” says Kelvin Droegemeier, a professor of meteorology at the University of Oklahoma in Norman, and director of the Center for Analysis and Prediction of Storms (CAPS).

Droegemeier and colleagues, however, in addition to researchers at NOAA, have made progress in recent years. Advancements in radar technology have increased the quality of data, which in turn are used in increasingly sophisticated models. The models serve as the basis of the NOAA Storm Prediction Center’s tornado outlooks, watches and warnings, and the center has been working in conjunction with the media and the National Weather Service to increase the number of people who receive the messages with the maximum time to prepare.

The first step for researchers is to learn more about the components of tornado formation. “Part of the thing about predicting the weather is understanding how it works, understanding the dynamics,” Droegemeier says. A simple explanation for conditions in Tornado Alley is that dry air from the Rocky Mountains meets warm and moist air from the Gulf and cold air from the north. The interaction of hot, cold and dry create ideal conditions for the formation of supercells — rotating thunderstorms necessary for tornado generation. The problem is that tornado formation is complex, and conditions favorable for supercells do not guarantee a tornado.

One way that researchers are trying to learn more about tornado dynamics is through close observations of the twisters themselves, especially from Doppler radar. NOAA operates the Next Generation Weather Radar, or NEXRAD, which since 1988 has employed Doppler technology in a network used to identify and warn about dangerous weather and its location. Now 158 radar stations dot the United States, leaving few holes in the coverage. “It definitely assists us in issuing watches,” says Dan McCarthy, a warning coordination meteorologist at the NOAA Storm Predictions Center.

At large distances, the curvature of Earth prevents NEXRAD from scanning close to the ground. At about 400 kilometers away, the gap can be as large as 5 vertical kilometers — a gap in which there is “a lot of the stuff that is a precursor to tornadoes,” Droegemeier says, and NEXRAD “is missing it.” Thus, 3 out of every 4 tornado warnings are a false alarm, he says.

To “fill in the holes,” Droegemeier is working with researchers at the University of Massachusetts in Amherst, Colorado State University in Boulder, the University of Puerto Rico in Mayaguez, and corporate and government partners on a project called the Center for Collaborative Adaptive Sensing of the Atmosphere (CASA). The notion is to deploy small, inexpensive and low-power Doppler radar units on cell-phone towers, starting with a small Oklahoma test installation in spring 2006. Unlike NEXRAD, they would not scan at long-range distances, but instead would collect data close to the ground. Droegemeier says he hopes that the system will observe precursor structures in storms, before a tornado actually forms. Such data could prove useful to NOAA in issuing warnings prior to tornado development, which is something, McCarthy says, is “the biggest challenge we have.”

Improved radar technology has created an increase in the volume of tornado data, a development that Harold Brooks, a research meteorologist at the NOAA National Severe Storms Laboratory, says is one of the greatest recent advances in tornado forecasting. He says that researchers previously studied conditions surrounding about 20 cases in which tornadoes occurred. Now researchers have datasets for hundreds of storms, including conditions that did not spawn tornadoes. Brooks says that the information will help meteorologists working to distinguish the parameters that make a storm turn into a tornado or not.

The large amount of data has in turn been useful to researchers such as Ming Xue, a meteorology professor at the University of Oklahoma, who has been using technology at the Pittsburgh Supercomputing Center to create what the center calls “the most realistic tornado simulation ever done.” The project reproduced the vortex structure and wind speeds of a tornado spawned by a severe storm in 1977.

Computing power limited previous simulations to linear areas on the scale of about 100 meters, too small to reliably capture forming tornadoes and maintain high resolution. Now, Xue has increased the computable area to capture 50 kilometers on each side of the storm.

To simulate a one-hour storm, the program calculates temperature, pressure and air speed every second for 24 hours and produces 20 terabytes of data. Xue says that he hopes that his research will provide better understanding of tornado dynamics, in addition to developing algorithms to be used in tornado-detecting radar programs such as CASA’s.

Droegemeier thinks that in general, research is leading toward the ability to predict the occurrence of severe storms up to a day in advance. But along with increased prediction capabilities are the issues of how to prepare people in the path of a potential storm. “You can’t look at the science problem just as a science problem,” Droegemeier says. “You actually have an important social dimension.”

McCarthy estimates that about 70 percent of people get weather information from TV, posing a problem when warning for the “tornadoes that work on Sunday nights, when the media goes home.” Longer forecast times may someday alleviate that problem, but for now, NOAA is working to increase the accuracy of watches and warnings. This fall, the agency plans to change the warning zones from generic boxes to more exact county-by-county warnings. “The accuracy is getting good enough now that we can start to get more precise and refine our thinking a little bit, so that we don’t over warn,” Droegemeier says, and do not increase public apathy.

Kathryn Hansen

Center for Analysis and Prediction of Storms
NOAA's National Weather Service Storm Prediction Center
Center for Collaborative Adaptive Sensing of the Atmosphere

Back to top

Untitled Document

Geotimes Home | AGI Home | Information Services | Geoscience Education | Public Policy | Programs | Publications | Careers

© 2018 American Geological Institute. All rights reserved. Any copying, redistribution or retransmission of any of the contents of this service without the express written consent of the American Geological Institute is expressly prohibited. For all electronic copyright requests, visit: