Highest risk corridor?

[Shawn Schuman]

I don't know if this data is easily accessible to anyone, but insured losses (adjusted for inflation) per person would be an interesting and perhaps easy way of comparing 2 different areas? I would think it would show a similar trend, with the most insured losses per capita occurring in the areas in question, from say OKC through Alabama.

I'm working on something like that, I'll post it shortly. But yes, it's more or less the same trend no matter what metrics you use. A large area of enhanced risk through the Southeast and into the Southern and Central Plains, with spikes in northern/western Alabama and central Oklahoma (and occasionally in central Mississippi and south-central Kansas). The only major exceptions are Massachusetts and Indiana, which often show up when you're looking at the data per capita and/or per square mile. Of course, Massachusetts' high ranking comes almost entirely from the 1953 Worcester tornado, with some contribution from the 2011 Springfield tornado. Indiana is often ranked fairly highly because of the 1965 Palm Sunday outbreak and, to a lesser extent, the 1974 Super Outbreak.

 

[Mark Blue]

The entire time while I was reading the article I couldn't help but think it's more indicative of long-track tornadic events versus the actual location of tornado alley, but who am I to say? It was nothing more than a gut feeling.

 

[Dan Robinson]

I'm sure everyone has seen these maps, but just for reference, here they are (copied from a past post in this thread). From the University of Nebraska-Lincoln High Plains Regional Climate Center these show all tornado reports (left) and violent tornado tracks (right) from 1950 to 2005:

Also of interest: Jon Davies' blog post listing the top ten worst tornado outbreaks in US history, ALL outside of the Plains (aside from a small part of the May 1927 event clipping the central Plains):

http://davieswx.blogspot.com/2011/04/27-april-2011-tornado-outbreak-stunning.html

It's clear the South matches the Plains on every count - violent tornado frequency, number of tornadoes, tornado damage and deaths. James Spann might even have the edge on Gary England when it comes to the number of events he's covered live that have affected the Tuscaloosa/Birmingham viewing area.

http://www.youtube.com/results?search_query=abc+33+40+tornado+live&aq=f

However, it's become increasingly clear how the Plains' violent tornado history has been underaccounted for by damage-only scales. Like others have said, I'm inclined to believe the Plains and the South are probably very close if not equal in this category.

 

[Shawn Schuman]

Okay, few new maps to add. All of the monetary damage data is in dollars and has been adjusted for inflation, housing and population.

Damage per tornado:

Damage per square mile:

Yearly damage per capita:

All significant (F2+) tornadoes:

All violent (F4-F5) tornadoes:

All strong (F3+) tornadoes, not including 4/27/11:

Same as above, including 4/27/11:

 

[Shawn Schuman]

Well I feel unoriginal now. Rich Thompson apparently had the same idea I had (using area x intensity to make comparisons), way back in 1998. He called his the "Destruction Potential Index," which sounds a lot better than my name. I believe he also used kilometers rather than miles or feet (only skimmed the paper), so I've updated to do the same thing.

http://www.spc.noaa.gov/publications/thompson/dpi/dpi.htm

The results are still more or less the same, but I'll post the new ones tomorrow. Also have some stuff by season and a few other things.

 

[Mike Johnston]

Really appreciate your work on this stuff, Shawn. You're a gifted analyst and a credit to this forum.

 

[mkavulich]

Including path width seems like a bad idea for trying to determine the area impacted by a tornado, for several reasons:

1. Path width is the maximum path width, which is almost always much higher than the average path width. This will undoubtedly lead to large overestimates in the area impacted.
2. The path width of a violent (or really any) tornado will always be larger, usually much larger than the area actually affected by the most destructive winds.
3. The path width is determined based on the widest area of damage, which in forested and densely populated areas will lead to biases in the data.

Perhaps I'm overestimating the impact of these problems, but I feel that including path width data will be a net negative if you're trying to determine the true risk to a given area.

 

[Shawn Schuman]

It's actually worse than that, since path width used to be recorded as the average, but is now the maximum width. But besides that, there are biases everywhere. Literally the only way to avoid running into biases in your analysis is to not do any analysis at all. Path length is extremely inconsistent, as is width. Intensity estimates are generally terrible. Even the tornado count is badly flawed. Fatality numbers are prone to many biases as well. In reality the tornado record tells us more about our (in)ability to document tornadoes than it does about the tornadoes themselves. But, that's all we really have to work with. There are still things to be gleaned from analysis, we just have to bear in mind that the data we're using is objectively terrible for any number of reasons.

 

[Jeff Wright]

More news on the subject
http://www.wsfa.com/story/23767875/new-study-claims-tornado-alley-may-be-in-southeast
http://birmingham.raycomweather.com...n-tornadoes-and-miniature-tornado-alleys.html

 

[Dave Gallaher]

More news on the subject
http://www.wsfa.com/story/23767875/new-study-claims-tornado-alley-may-be-in-southeast
http://birmingham.raycomweather.com...n-tornadoes-and-miniature-tornado-alleys.html

I'm especially fascinated with the notions in that second link. That area around Moulton, AL shown as a touchdown point has always been a good place to be in a line of approaching storms, and I've noted some very interesting small-scale phenomena there as well.

 

[Scott A. Kampas]

Shawn did some very good original work here.

Path width is even more complicated in that the definition of tornado matters to determining it and there's a lot of confusion, as well as difficulty in making determinations even if one is informed on the issue. This is increasingly apparent from ground level tornadocyclones producing damage and from radar observations showing irregular shaped tornadoes, tornadoes that are not visible to the naked eye or whose damaging winds extend outside the condensation/debris cloud, damaging RFDs and inflow jets, and the like. The width and definition issues have been covered for many years by the likes of Doswell and Rasmussen, respectively.

Harold Brooks et al have done the definitive work on tornado frequency and "density".

It boils down to the southern Plains having the most consistent significant tornado occurrences per annum. It's pretty clear that's Oklahoma, north and parts of west Texas, and Kansas. Farther east has big tornado events less frequently (as does farther north or west in the Plains where population density reduces sightings and rated intensities), but gets some very large outbreaks with very long path lengths. Partly there's more shear causing that. There's also favorable more favorable trajectories advecting warm, moist Gulf air in the off season which leads to an extended but more diffuse season of tornado activity in the MS/OH valleys and Southeast. The proximity of the Southeast to the Gulf is a factor meteorologically, as are the population density and lower visibility contributing social factors. For deaths, housing construction and lack of basements also matters. If Dixie Alley leads, other than for getting some really massive outbreaks with intense tornadoes, occurrences of exceptionally dense corridor outbreaks, and very long tracks, it leads on tornado deaths. Historically, there were many very high fatality events so 1973 wouldn't make as stark a difference as a starting point for fatalities. On the other hand, the Southeast was settled before the Plains which confounds historical data from very long ago.