tinychaseTornado kiosks · the measurement record, 1950 to 2025
Width Creep
The average tornado in the record is three times wider than it was in 1975. Almost none of that is weather. Before 1995 the Storm Prediction Center logged a path's mean width; from 1995 it logged the maximum. In 2007 the EF scale changed how damage is surveyed, and surveys have kept getting more thorough since. Drag the tape to stretch the old records and see how much of the climb survives.
Recorded width, year by year
What the surveyor wrote down
Share of all rated tornadoes recorded at each exact width. For 45 years, "10 yards" was the answer more than a third of the time.
Ground swept, by decade
Path length times width, summed. This is the basis of the Direct Hit kiosk's recurrence odds, so whatever the tape does to old widths it does to those odds.
Data: TerraPulse tor dex (SPC tornado history cross-matched with NCEI Storm Events; US federal public domain), exported 2026-09-27, rated tornadoes 1950 to 2025 with a recorded path length and width. Widths are as logged by SPC: mean path width before 1995, maximum after. The "1995 jump" preset is the ratio of 1995 to 1999 medians to 1990 to 1994 medians, averaged across ratings; it is an empirical patch, not a physical conversion, and the true mean-to-max ratio differs by tornado. Stretching only the pre-1995 rows cannot address the 2007 EF-scale change or the growth in survey effort since, which is the point. Path length is not stretched and has tripled on its own. Swept area treats each path as a rectangle. Rating thresholds mix F and EF scales.