6 April 2012
There Really Is A Tornado Alley- Three Actually
Posted by Dan Satterfield
The Tuscaloosa AL Tornado on radar. New research shows that the southeast gets more deadly tornadoes than the traditional tornado alley in the Plains.
Any meteorologist gets the same question over and over: Where exactly is tornado alley?? The answer (until now) has been that there is no scientifically defined alley, but that the Great Plains from Kansas to Texas is the best answer.
Enter Michael Frates at the University of Akron. He decided it was high time to delineate the tornado alley’s of America and he did it based on historical data beginning in 1950. Frates finished up his Masters in Geography on this project, and did a great piece of overdue science in the process!
The answer is surprising as well. There are three main alleys and the most long track large tornadoes occur not in the Plains, but instead is the “Dixie Alley” in the Southeast U.S. There are many factors affecting the death toll in tornadoes, and the Southeast has lots of mobile homes (and is a historically poor part of the country), but Frates just looked at EF 3 to EF 5 tracks and image he produced below shows well where the higher risk of deadly tornadoes lies.
From Michael Frates at the Univ. of Akron. Click image for larger version or you can get a pdf of this poster at http://www.uakron.edu/dotAsset/1085452.pdf
A press release from the Univ. of Akron states:
Frates examined the distribution of tornadoes with tracks greater than 20 miles, which occurred between 1950 and 2006 over the geographic area incorporating all four regions. Frates reviewed this entire area as a grid of 3,068 cells, each representing 1,082 square miles. By studying these area units of equivalent dimension, Frates was able to study the area as a whole divided into equal parts — not as individual regions of varying size marked by traditional boundaries such as county and state lines — and ultimately, present these breakthrough findings.
Something Mike Frates did not cover in his research: Virtually every meteorologist I know grew up in one of those “alleys”. I’m from the brown rectangle in Oklahoma 😉
Dan Satterfield has worked as an on air meteorologist for 32 years in Oklahoma, Florida and Alabama. Forecasting weather is Dan's job, but all of Earth Science is his passion. This journal is where Dan writes about things he has too little time for on air. Dan blogs about peer-reviewed Earth science for Junior High level audiences and up.
Excellent post, Dan. Hope you don’t mind if I link to it in my blog.
And, BTW, where I grew up in Wisconsin is right on one of the “borders”!
go for it Kurt!
Fascinating!! I’ve long wondered the same thing, especially since I don’t live in any of the tornado alleys. Funny thing, though….. I know four meteorologists, and they ALL come from the alleys.
Mr. Frates has performed and presented a good piece of research. His method as presented on the poster smacks of geostatistics of spatial variables (krige) which should be used more often in meterological studies.
Living in MS for 40 yr gives me (and TV forecasters) the impression of a distinct seasonality — late Nov early Dec and then again Mar-Apr. Wonder if there is a similar thing in the other ‘Alleys’
Historically, the Plains would see a main season from April to early June with the northern areas seeing storms much later from May into late June. This seems to be changing now with the warmer temps. of the last few decades. We had damaging tornadoes in Michigan in March this year. The extra .8C temp and the resulting increase in atmospheric moisture is without doubt changing the prime season. Am not sure if there is a reliable record for a long enough period to really show it yet.
Dan
One issue is the dataset length. Adding in 2008 would close the gaps considerably between the “alleys.”
Another issue is the spatial smoothing applied, which is related to dataset length. The presence of the small-scale maxima indicates to me that the smoother applied is way too light.
As I pointed out in a paper in 2003, the “colloquial Tornado Alley” is implicitly defined not just by occurrence, but by the repeatability of the season. The Plains have a strong seasonal signal and regions farther east, particularly, in the southeast, do not. I’m not sure what meaning the phrase “Tornado Alley” has, in any event, but the Plains-centric identification is highly a function of the strong seasonal aspect and the repeatability of the season.
I also don’t understand why “Carolina Alley” is identified, but the almost identical magnitude feature in the southern Texas Panhandle is not.
Finally, the NWS never has and never will define a “Tornado Alley.”
For those that do not know- Harold Brooks is one of the world’s main experts on this question.
Oh, and thanks for the contribution Harold.
Hi Dan, According to a UAB study, Alabama has the highest number of deaths since 1980 and most Alabamians don’t even know that. Causes of deaths include high number of mobile homes where most deaths occur, high number of F4 and F5s, rain rapped tornados difficult to see, hilly dense terrain limits visibility. The same study also states that use of protective helmets during tornados would be a game changer by reducing the high number of deaths due to head injury.
Over 30 years ago I wanted to map tornado incidence (creating a tornado-density isoline map) better than what I’d seen (choropleth? Are you kidding?), and the approach I thought of turned out to be one using Voronoi diagrams, with real-world path lengths greater than a single point presenting a problem, since I haven’t the math skills to proceed without highly skilled help. I also want this computerized. Recently I discussed this with someone via LinkedIn, who suggested kriging might work better, omitting Voronoi. Previously I’d asked around trying to find if anyone had tried mapping tornado incidence using Voronoi. None yet. I even asked Harold Brooks, who didn’t reply. Thanks a lot! Very recently I read of yet another mapping approach which uses path lengths and finds a Birmingham-Huntsville high-risk corridor. Someone else made a comment here mentioning the krige method. I still think what I wrote down 30 years ago is valid. But I haven’t the $$ to proceed, and can’t do it on my own. Long-term involuntary unemployment. What’s a person to do in my fix, with no hope out w/o help? Mayday!