NWC REU 2019
May 21 - July 30



Photo of author

Classification of Rarity of Tornadic Outbreaks

Michael Smith, Makenzie Krocak, and Harold Brooks


What is already known:

  • Calculations of total annual tornado risk show peaks in Oklahoma, Mississippi, and Alabama.
  • Within those peaks the total number of annual tornadoes were similar, but the distribution throughout the year was different.
  • For example, the southeastern United States shows no peak in the annual cycle or the diurnal cycle, whereas Oklahoma has a distinct peak in May.
  • The rarity of a tornado outbreak could affect ones’ perception of risk and thus merits study.

What this study adds:

  • Using EF1 and greater tornado reports from 1954–2015, a rarity factor was calculated by dividing the individual day by the climatology, thus is affected by both a large, single tornado outbreak day or by a small climatology for that location.
  • Considering days with at least 10 grid points reported, 422 days were identified.
  • Three ways to classify rarity were calculated: Maximum Rarity factor (single day/climatology), Area (Rarity>=100), and Intensity (Tornados>= (E)F3).
  • Observed days that had at least 10 grid points reported. Resulting in an overall outlook of 422 days within the data set roughly occurring 7 days a year.
  • These Calculations show that:
    • Intensity occurs earlier in the year than Area, but both classifications occur earlier than Maximum Rarity.
    • Among the top 10 rankings of each factor, 4/10 dates are shared between Intensity and Area; None of the top 10 dates for Maximum Rarity appear on the other lists.
  • Interestingly, while the April 2011 outbreak placed within the top 50 in Area, that month also had 7 consecutive days meeting rarity criteria.
  • Other strings of consecutive days occurred in 1973, which had no single significant days, and May 2019, with multiple tornadic days, none of which were notable. The April 2011 outbreak was quite a significant day, but also occurred within a string of days over a two week period.
  • Intensity shares 4/10 dates of the top 10 rare rankings, presenting some correlation within the two categories, while maximum rarity shares none with either category.
  • There seems to be a trend with consecutive days such as the 1973 consecutive outbreak which didn’t have a single significant day but lead into the month of June. The April 2011 outbreak which did have significant days but occurred consecutively for over two weeks. Also May of 2019 in the state of Oklahoma, where there were multiple tornadic days developing but such as in 1973 not one day stood out.


Tornadic outbreaks cause widespread devastation mentally, physically, and financially. This project focuses on rare tornadic outbreaks by implementing climatological estimates of daily tornadic probability in the United States. The data consists of (E)F1 and greater tornado reports from 1954-2015 from the Storm Prediction Center (SPC). A Gaussian smoother was applied to each day for both space and time. Then a climatology was created by taking the mean of each day over the 62-year span. To obtain the rarity factor, each individual day was then divided by the climatology. Rarity was classified in three states; maximum rarity factor, how large an area was affected, and how intense the event was. These results show that intensity and area have a strong correlation. This topic is critical because rare events cause a state of uproar in the public and with a lack of preparedness cause lead to multiple complications.

Full Paper [PDF]