Evaluating a Tornado Index

I’ve been working on a tornado index over the past few years and it’s really basic. It evaluates instability (MLCAPE), wind shear and the potential for upward cyclonic rotation (helicity) and gives an index between 0 and 5*. Based on tornado cases around the United States, equations** were created that rate those three parameters from 0 to 5, with 1 being assigned to the 10th percentile, 2.5 to the 50th percentile and 5 to the 90th percentile. Basically, the scale says that a tornado is unlikely/impossible at a 0, but becomes increasingly likely as the index approaches 5. The average environment for a tornado in the United States gives a mean index of 2.5.

Based on all of this, it’s relatively easy to determine if a predicted setup is likely to produce tornadoes. Yesterday morning, it was clear that the index was showing a possibility of a tornado in the Northeast during the afternoon/evening. After evaluating the mesoanalysis from the time of a tornado in Worcester, Mass., it was determined that the value was 1.76. In other words, a weak/brief tornado was possible based on the index.

While considering the parameters and the index as a whole, it’s been noted that the intensity of a tornado is closely related to how “severe” the parameters are. While a 1 or a 2 on the index means that a tornado is possible, the severity of a tornado in that case should be relatively weak. On the other hand, significant tornadoes are associated with values of 3 to 4, or higher. While rare, cases with the index approaching 5 tend to feature multiple significant or even violent tornadoes.

A quick look back at June 1st, 2011, when a long-track EF-3 tornado ripped through south-central Massachusetts, shows index values into the high 3s. The approximate value leading up to the tornado touchdown was 3.82, which indicates that tornadoes were probable and that a significant tornado was also possible.

I’ve created grids with a 50-mile radius from various stations around the Northeast. This is helpful, because although the environment in a particular area may be favorable for a tornado, tornadoes are small-scale weather events. It’s extremely difficult to predict exactly where a tornado will touchdown, but it’s easier to assess a broader area that has the best probability of seeing a tornado. The image from August 31st, 2014 shows that much of the Northeast had the possibility of seeing a tornado. With decreasing instability, a tornado was unlikely over Cape Cod and the Islands. On the case of June 1st, 2011, the highest values were across western Massachusetts. This is where the EF-3 tornado first touched down.

When applied to notable tornadoes over past years, there is a fairly good correlation between tornado intensity and a higher tornado index. Also, events with very high index values (4+) generally saw major tornado outbreaks.
6/16/2014: 4.50 (Pilger, NE EF-4 tornado)
4/27/2014: 4.26 (Mayflower, AR EF-4 tornado)
7/10/1989: 4.06 (Hamden, CT EF-4 tornado)
5/20/2013: 3.45 (Moore, OK EF-5 tornado)
The Moore tornado case shows that a violent tornado can occur with a value in the 3s, so it should be noted that the index number does not directly relate to an assigned EF-number. Likewise, a 0 does not imply an EF-0. In fact, a tornado index value below 1 is not likely to produce any tornado.

Earlier versions of this index also factored in the lifted index. This proved to be redundant with respect to instability, plus the research papers that were considered to formulate this index did not use the lifted index anyway. This tornado index (sometimes referred to as the Q-Tornado index) is now streamlined, to where it only requires three variables: MLCAPE, wind shear and helicity.

I will continue working on cleaner graphics to display the tornado index values/probabilities. Please free feel to comment below or contact me with feedback. I used to post Excel screenshots, but it became tedious to run the index for different locations that people were requesting. Now I run the index and plug the values into the grid graphic, so the entire region can be viewed at-a-glance to assess the tornado potential.

I have also toyed with converting the raw index value to a tornado percentage. This is non-scientific and still being reviewed, but the current procedure multiples the index by 20. This means that if a location has an index of 1, there is a 20% probability*** of a tornado within a 50-mile radius. If the index were to reach 5, the probability*** of a tornado within a 50-mile radius would theoretically be 100%. ***Again, this percentage probability method is not scientific and has not been tested enough to consider operational or even realistic.

*Tornado index values
0-0.99: Tornado unlikely
1-1.99: Tornado possible (weak/brief)
2-2.99: Tornadoes possible (EF-2 or weaker)
3.3-99: Tornadoes probable (EF-2 or stronger tornado possible)
4-4.99: Tornadoes likely (EF-2 or stronger tornadoes likely)
5-5.99: Major to historic tornado outbreak likely with EF-3 or stronger tornadoes likely

**In order to avoid skewing the index, the equations for each parameter are capped at 6. This means that it is not possible for the index to be higher than 6. As mentioned, I have not even seen a value as high as 5, so anything that high is theoretical anyway.