Animated map of tornado days per county normalized for area. The data is raw data from the tornado database from Storm Prediction Center. Coding errors in the database such as documented in my "Corrections to the Historic Tornado Database" paper were not applied, but were not expected to make a significant impact on the mapped statistics.

I use "tornado days" as opposed to actual number of tornadoes as I think the number of tornadoes is an overrated statistic. A series of 10 small, brief, weak tornadoes would count 10 times as much as a single long-tracked violent tornado, even though the latter likely would have much greater impact than all of the brief ten tornadoes combined. I also wanted to investigate the likelihood of a tornado developing in a particular county or area based on time of year, while that probability could be skewed by multiple tornadoes on a given days.

For an animation of tornado data (1980-1999) smoothed in both time and space, the National Severe Storms laboratory has this animation available.