Tornado warnings: Officials refine risk levels to convey urgency to public

When residents of Hattiesburg, Miss., woke up on Sunday, Feb. 10, 2013, forecasters at the National Weather Service's Storm Prediction Center in Norman, Okla., had long since posted a heads-up to local forecasters that the weather that day might get nasty.

From Saturday evening, its severe-weather map showed most of Mississippi covered in yellow, indicating a slight chance of severe and possibly tornadic thunderstorms. At 5:03 P.M., a tornado touched down outside of Hattiesburg and grew to a multi-vortex twister that tore through the western side of the city, inflicting more than $38 million in damage and injuring 82 people. Indeed, the storm system that moved through that day would generate 18 tornadoes from southern Mississippi through southeastern Alabama.

The recognition that a forecaster's "slight chance" still could lead to a very bad day for people in the forecast area has prompted the Storm Prediction Center (SPC) to refine the low-end of a scale used to assign severe-thunderstorm and tornado risks.

Under the new scheme, Hattiesburg would have faced an "enhanced" risk of severe thunderstorms, rather than "slight." Areas falling outside of what remained of the "slight" region, but still at some risk for weather related damage, would have fallen into a new category: marginal.

At first glance, the changes may seem like an exercise in semantics. And to some extent they are. For several years, the National Weather Service has been working with social scientists to find better ways of communicating the contents of its forecasts in ways meaningful to emergency managers, radio and TV weathercasters, and the general public, not just experienced forecasters, notes John Ferree, the SPC's severe-storms services leader.

But the "slight" category for severe storms and for tornadoes actually spans a threefold increase in risk, from a 5 percent chance of severe weather to a 15 percent chance, he notes. At 15 percent, the risk for severe weather is considered moderate.

Such numbers may sound low, but even a twofold increase "is a big difference," especially if they relate to tornadoes, he says; it means those in an area covered by the larger number are twice as likely to see tornadoes as people covered by the lower probability.

Indeed, once the risk level rises to 15 percent, forecasters are anticipating "a pretty darn big event," Mr. Ferree says.

Currently, for some parts of the country, a slight risk for severe storms or tornadoes is all they see. Yet one large blob of color on a map for such a wide range of risk does little to provide the best estimate of risk for a given location.

The SPC would like to unveil a web page with the new maps next week and is eyeing April as the month where the refined risk maps go live. The center wants to give potential users time to evaluate the new rating system, provide comments, and to make the changes needed to the forecast and presentation tools they use and that the SPC's maps feed.

Although the new risk scale is slated to go live in April, it will be offered initially on an experimental basis to allow the center and users to continue to tweak the system as needed.

In addition, this year will see another tornado-hazard communications project expand beyond the central US, where forecasters have been testing it during the past two years. Known as impact-based warnings, these incorporate statements about the potential damage a tornado poses.

Such forecasts carry one-word adjectives, such as "considerable" or "catastrophic," to describe the damage threat, as well as more fulsome estimates of the threat to property and lives based on reports that feed into local forecast offices. The goal is to present these with a level of urgency and confidence that the situation demands without raising fear to such a level that people either take the wrong action or no action at all.

After two years of testing, the NWS has found that forecasters have done a pretty good job of identifying and describing the risks from what Ferree calls "high-end events." Emergency managers and first responders have tended to take advantage of the information because is it issued with more confidence.

But the NWS also found that if the warnings are too strong – "you have to get underground or you're going to die" – people won't do the right thing, he says. Instead of heading for adequate shelter in their home or nearby, they'll hop into a car and try to drive out of the area.

"Sometimes that's the worst thing you could do," he says. "You're moving from a very safe place, or at least a place that's survivable, into your car, which may not be."

As with the new system for rating the probabilities of severe weather, the NWS will continue to tweak its approach on impact-based warnings. The goal is to strike the right balance between communicating an appropriate level of urgency regarding the threat without inducing people to take to wrong actions or no action at all, he says.

"We spend a lot of money as taxpayers on equipment to allow us to locate tornadoes before they touch down," he says. "That data and those warnings aren't much good if we don't communicate them well and people don't take action."