Experimental model predicted tornado's path hours, not minutes, before it formed

This tornado, predicted by a prototype weather model, struck the town of Elk City, Oklahoma, on May 16, 2017. Its path was predicted hours in advance, rather than minutes, by an experimental NOAA model.

This tornado, predicted by a prototype weather model, struck the town of Elk City, Oklahoma, on May 16, 2017. Its path was predicted hours in advance, rather than minutes, by an experimental NOAA model. (Image credit: Gabriel Garfield/NOAA/Univ. of Okla. CIMMS)

As severe weather brewed in the Texas panhandle late in the afternoon of May 16,  NOAA National Weather Service forecasters alerted residents in parts of western Oklahoma about the potential for large hail and damaging tornadoes that evening, particularly in the area around Elk City.

Ninety minutes later, a dangerous, rain-wrapped EF-2 tornado struck the small town: It killed one, injured eight, and destroyed about 200 homes and more than 30 businesses.

Normally, meteorologists issue warnings based on radar depictions or spotter reports. By then, a tornado could be minutes from touching down. This time, the NWS issued an additional advisory for parts of four counties in southwest Oklahoma stating “... a high probability that tornado warnings will be issued.”

A new, experimental forecast model made it possible.

Forecasters that day were working with researchers from NOAA’s National Severe Storms Laboratory who were testing a prototype Warn-on-Forecast system (watch this video), a new research tool that has the potential to dramatically improve predictions of extreme weather at specific locations up to three hours in advance.

It was the first time the Warn on Forecast (WoF) model was used by the NWS in this way.

“We had a picture of the storms and their evolution before they became life-threatening,” said Todd Lindley, science operations officer with the NOAA NWS Norman Forecast Office in Oklahoma. “We used this model guidance to forecast with greater lead time and greater confidence.”

“Based on the information from the NWS, we knew storms would intensify when they reached our area and were able to activate the outdoor warning sirens about 30 minutes ahead of the tornado,” said Lonnie Risenhoover with Beckham County Emergency Management.

As the storm evolved, a faster timeline of information-sharing

On May 12, the NOAA Storm Prediction Center had already identified the possibility of severe weather in the region, and early in the morning on May 16 they updated their forecast calling for significant tornadoes. At 1:50 p.m., NWS issued a particularly dangerous situation (PDS) tornado watch for 33 counties in western Oklahoma and the Texas panhandle.

Soon after, WoF prototype forecasts began to identify a specific area in the eastern Texas panhandle as the likely starting point for potential life-threatening weather.

“That level of detail and lead time in a forecast is new,” said NSSL Director Steve Koch. “To have information conveying a sense of certainty in so small of an area that far in advance is a success.”

The WoF combines the best weather prediction technologies from NSSL in Norman and NOAA’s Global Systems Division at the Earth Systems Research Lab in Boulder, Colorado.

WoF isn’t operational yet — more works needs to be done — but it represents a significant step on NOAA’s path to providing more precise hazardous weather information to the public sooner.

It’s just one example of how NOAA weather researchers work hand-in-hand with forecasters to develop and test scientific advances to protect lives, property and commerce.

Media contact

Keli Pirtle, 405-325-6933