A new hazard model for tornadoes

Hazard models for disasters (earthquakes, volcanoes, etc.) are designed to estimate how often a particular place will be affected by a disaster and how intense the event will be. Traditionally, tornado hazard models have relied on how often a tornado has hit close to a location to estimate how often that location will get hit. Since tornadoes are rare events at any location, the estimates are difficult to verify and can be the estimates are difficult to verify and can be very sensitive to the length of the record. Recently, Harold Brooks of NSSL has been working on a new approach to making these estimates, using a combination of statistical models of the frequency of occurrence and path length and width of tornadoes to produce “artificial” tornado data. Most of the models take advantage of data from larger areas of the country than traditional approaches.

The average length of time between successive events at a place is known as the return period. The new hazard model estimates that the shortest return period for a location getting hit by an F2 or stronger tornado is a little less than 4,000 years. For an F4 or stronger tornado, it’s about 18,000 years. The most likely location in the contiguous United States to be struck by strong or violent tornadoes is in south central Oklahoma. The least likely location is in central Nevada, where an F2 or stronger tornado is only predicted to occur once every 2 million years.

The research was done in collaboration with Michael Kay of the Storm Prediction Center and summer students at the Oklahoma Weather Center’s Research Experience for Undergraduates program from 1999 to 2001.