A Look at Michigan Tornado Statistics

[huronicane]

Having grown up in Michigan, and my interest in weather driven by an initial fear of severe thunderstorms and tornadoes (including one that passed near my preschool when I was a wee lad), I've always found it unusual that there hasn't been much research on the topic of Michigan tornadoes.

In a regional climatology course a few years ago, I decided to take a look at Michigan's tornado history. While searching the literature, I found one article:

"A Look at Michigan Tornado Statistics" - CR Snider (1977), who concluded that "Present methods do not detect all tornadoes" and "Tornadoes are more likely to touchdown (i) on smooth surfaces rather than rough surfaces, (ii) in small towns rather than large cities, (iii) over small bodies of water rather than adjacent land areas."

During additional research, I came across a presentation that John Laurens and Randy Graham put together at the NWS in Grand Rapids, titled "Significant Tornado Climatology for Lower Michigan" which determined "3 unique pattern types which produce F3 or stronger tornadoes across Lower Michigan."

I figured that a good way to start additional research would be to first go through the tornadoes in NCDC Storm Data between 1950 and 2006 (told you it was a few years ago), and look at the when and where tornadoes occurred in Michigan:



As for the when: Michigan has had tornadoes reported in every month except December. Michigan's tornado season, which I defined as being any month that averaged more than 1.4 tornadoes/year, runs from April through August, with a peak during the months of May, June, and July. June was the peak month, with an average of 3.7 tornadoes per year from 1950-2006.

I wanted to take the research further, so I decided to look at Michigan's tornado days since 1997 (the first full year that Michigan was fully under the umbrella of NEXRAD). Using TornadoHistoryProject.com (which uses SPC's tornado database, instead of NCDC's, which shows a tornado that crosses county lines as multiple tornadoes), 218 tornadoes touched down in Michigan during 99 tornado days. A few of the tornado days (for instance, October 19, 2007) were a continuation of the previous day's tornado event. The goal of this research is to classify Michigan's tornado events by morphology, determine synoptic and mesoscale conditions associated with the morphologies, and provide operational guidance for forecasters during Michigan's tornado events.

Some preliminary observations:

Between 1997 and 2010, tornadoes have occurred during meteorological spring, summer, and fall...but not winter. Similar to the historical database, 1997-2010 have a tornado season during May - August, with a peak in June.

Nearly 90% of the tornadoes that have occurred since 1997 have been rated as weak...far more often than in the entire historical database, when tornadoes were rated as weak approximately 71% of the time.

The strongest tornadoes have been rated F/EF-3. The last violent tornadoes to occur touched down April 2, 1977.

I've taken a quick look at the base reflectivity loops for each of the tornado days, and besides seeing lots of green, yellow, orange and red, I've split up the tornado days into a few storm modes:

1. Linear (including LEWP, QLCS, Bow Echos, Bookend Vortices, and Derechos...example: May 31, 1998)
2. Supercellular (examples: July 2, 1997, May 21, 2001)
3. Cellular/Cell Cluster (kind of a catch-all for non-supercell and poorly organized convection...this will need additional inspection to classify properly...but an example would be June 19, 2009, when an EF-1 and 2 EF-2 tornadoes touched down with no tornado warnings)
4. Combination (any combination of the above three)

Perhaps not surprisingly to most Michiganders, 39 of the tornado events appear to be associated with a linear storm mode, 30 under the catch-all, and 15 days each for supercells and combinations.

For you graph and chart fans:






Should anybody want to help me with the classification of any of my catch-all days, or come up with synoptic/mesoscale composites for the storm modes, please let me know.

Discussion welcome.

[Hoosier]

Good stuff. Your findings for MI are similar to mine for IN - a healthy percentage of events are linearish and less strong and especially less violent tornadoes than in previous years, although I think that is also a national trend as a better understanding of damage patterns/building construction has developed.

[SpartyOn]

Great Write Up!

[huronicane]

An additional thing that I've noticed, and will get more in depth in as I continue to dig into this, is that occluded lows seem to play a role in Michigan's tornado activity. Is that something you've seen in Indiana, Hoosier?

[huronicane]

Good stuff. Your findings for MI are similar to mine for IN - a healthy percentage of events are linearish and less strong and especially less violent tornadoes than in previous years, although I think that is also a national trend as a better understanding of damage patterns/building construction has developed.

That could be...a brief look at violent tornadoes since 1950:

1950s: 12 F5 / 108 F4

1960s: 11 F5 / 99 F4

1970s: 14 F5 / 108 F4

1980s: 3 F5 / 64 F4

1990s: 10 F5 / 82 F4

2000s: 2 EF5 / 45 F/EF4

2010s: 6? EF5 / ?? EF4

[huronicane]

I've been back taking a look at Michigan tornado stats...and will be taking a deeper look at tornado days (atmospheric patterns, etc) in the days and weeks to follow.  Let's start with tornadoes for 1997-2011:

 

 

Just looking at the totals for each year...

 

1997-1999 averaged 16.33 tornadoes/year

2000 had 4

2001-2004 averaged 20.75 tornadoes/year

2005 had 5

2006-2008 averaged 15.66 tornadoes/year

2009 had 3

2010-2011 averaged 21 tornadoes/year

2012 had 6 (through 9/30/12...as far out as NCDC has, but I don't think there were any others after that)

 

Could we be looking at a bigger (back to average) year with 15-20 tornadoes in Michigan this year?  While just looking at patterns in the numbers doesn't necessarily correlate to what we'll actually see, perhaps there's some pattern in the overall weather pattern that has something to do with the numbers.

 

Some monthly statistics for 1997-2011:

 

 

May is the winner, thanks to May 2001, with a mini-supercell outbreak on the 21st, producing 19 tornadoes.

 

As always, discussion is welcome and appreciated.

[Trent]

Great stuff.

 

I was in this tornado back in 1999 (write up here): http://www.crh.noaa.gov/dtx/to053199.phphttp://www.crh.noaa.gov/dtx/to053199.php

 

I was directly under the number 6 on the map when it hit. Interesting that that tornado happened to be the most intense in Michigan that year.

 

What always struck me as odd is that there was no warning out for the cell. Perhaps because Midland is so far from the 3 lower peninsula radars? As you can see from the pics and maps that tornado struck a highly busy commercial district on a holiday weekend. It's amazing there were no fatalities. Had that tornado veered a couple hundred feet in either direction or maintained F2 winds a mile longer, it could have been very deadly.

 

 

 

[huronicane]

Between 1997 and 2011, there were 233 tornadoes on 111 tornado days.

 

Tornadoes touched down in 72 of 83 Michigan counties.  The only counties without tornadoes were: Baraga, Benzie, Chippewa, Emmet, Grand Traverse, Houghton, Oceana, Ontonagon, Ostego, Roscommon, and Schoolcraft.

 

Of the 72 counties in which tornadoes touched down, some counties only experienced 1 tornado day, while some counties experienced 8 tornado days:

 

 

(edit: 1/15/2013, made a couple counting mistakes)

 

In map form:

 

 

 

(edit: 1/15/2013, to fix the counting mistakes from above)

 

The map paints a pretty clear picture of "Wolverine Alley", and comparing it to this plot from the Storm Prediction Center:

 

 

generally along a line from K3TR --> KLAN --> KFNT --> K29C appears to be the main Wolverine Alley corridor.

 

Based on 1997-2011 data, the average first tornado day in Michigan is May 6, but dates have ranged between March 28 and June 21.  In 1999, the first day was May 6, and in 2003, the first day was May 5.

[Hoosier]

Interesting to see the higher amount of days as you move into eastern lower...of course it's only 15 years of data so probably shouldn't read too much into that.

[Jonger]

Between 1997 and 2011, there were 233 tornadoes on 111 tornado days.

 

Tornadoes touched down in 72 of 83 Michigan counties.  The only counties without tornadoes were: Baraga, Benzie, Chippewa, Emmet, Grand Traverse, Houghton, Oceana, Ontonagon, Ostego, Roscommon, and Schoolcraft.

 

Of the 72 counties in which tornadoes touched down, some counties only experienced 1 tornado day, while some counties experienced 8 tornado days:

 

 

In map form:

 

 

 

The map paints a pretty clear picture of "Wolverine Alley", and comparing it to this plot from the Storm Prediction Center:

 

 

 

generally along a line from K3TR --> KLAN --> KFNT --> K29C appears to be the main Wolverine Alley corridor.

 

Based on 1997-2011 data, the average first tornado day in Michigan is May 6, but dates have ranged between March 28 and June 21.  In 1999, the first day was May 6, and in 2003, the first day was May 5.

 

Western Livingston does appear to be a tornado magnet, I have attempted to chase a few. Still have yet to see one.

[huronicane]

Interesting to see the higher amount of days as you move into eastern lower...of course it's only 15 years of data so probably shouldn't read too much into that.

 

Even when I looked at the larger set of data (from 1950 on) the greater number of tornado days (and tornado numbers) has been in E/SE Lower Michigan.  

 

My thoughts are that breezes off Lake Michigan stabilize the low levels of western Lower Michigan, minimizing tornado potential, and also (this is limited now) population effects. 

 

Population effects are minimized now due to increased spotting/chasing, Doppler radar allowing us to pinpoint surveys, and expanse in population.

[Hoosier]

Even when I looked at the larger set of data (from 1950 on) the greater number of tornado days (and tornado numbers) has been in E/SE Lower Michigan.

My thoughts are that breezes off Lake Michigan stabilize the low levels of western Lower Michigan, minimizing tornado potential, and also (this is limited now) population effects.

Population effects are minimized now due to increased spotting/chasing, Doppler radar allowing us to pinpoint surveys, and expanse in population.

I'm not even talking about western MI...there's a noticeable dropoff even in south central lower MI where any lake influence would presumably be little to none especially since most tornado days occur with SE to SW low level winds. Of course I'm not really familiar with the population distribution there so that could explain that.

[huronicane]

Ah, down in Hillsdale/Branch/Jackson counties.

 

A look at 2010 Census data: http://www.statemapsonline.com/images/maps/michigan/Michigan-State-Population-Density-Map.jpg

 

The population density is lower in Branch and Hillsdale, for sure, but it is still relatively high in Jackson County.  That's an interesting observation.  It would be interesting to see the cause of that.

[Hoosier]

Ah, down in Hillsdale/Branch/Jackson counties.

A look at 2010 Census data: http://www.statemapsonline.com/images/maps/michigan/Michigan-State-Population-Density-Map.jpg

The population density is lower in Branch and Hillsdale, for sure, but it is still relatively high in Jackson County. That's an interesting observation. It would be interesting to see the cause of that.

Something else I'd be interested to see if you get a chance...take the tornado frequency for the most populated county (I'm guessing that would be Wayne?) from 1950-present, apply it to the rest of the state and see what you come up with. I did it with Indiana (using the county where Indianapolis is) and the results were startling...the state would average almost 60 tornadoes per year compared to a long term average of 21 if every county behaved like that. I restricted it to years since 1980 and came up with a similar result. Even if there are some flaws, it really gave me a sense for how many tornadoes may still be going unreported.

[huronicane]

I'll have to take a look at that.  It would definitely provide skewed data in the Upper Peninsula (and maybe the Upper Lower Peninsula) where the climate is a bit less conducive to tornado development, but it would be interesting to see how the numbers look for the rest of the state.

[huronicane]

Here's an obscure one, that I've only taken a cursory look at.  Just from a brief look, approximately 20 tornadoes crossed a body of water at some point in their lifetime.

[SpartyOn]

The F5 tornado that hit West Bloomfeild MI on 3/20/76 is always an interesting topic. It absolutly devistated that region of Metro Detroit. If it happen today its effects would be even greater given the expansion of population in Oakland Cnty.

[dmc76]

The F5 tornado that hit West Bloomfeild MI on 3/20/76 is always an interesting topic. It absolutly devistated that region of Metro Detroit. If it happen today its effects would be even greater given the expansion of population in Oakland Cnty.

The tornado outbreak that is real interesting to me is the one that I'm not sure on the year is the F-5 tornado that went through Ortonville to Basically Romeo which killed over 20 people. What year was that? I think it was in the early 1900's

[Stebo]

The F5 tornado that hit West Bloomfeild MI on 3/20/76 is always an interesting topic. It absolutly devistated that region of Metro Detroit. If it happen today its effects would be even greater given the expansion of population in Oakland Cnty.

F4

[Stebo]

The tornado outbreak that is real interesting to me is the one that I'm not sure on the year is the F-5 tornado that went through Ortonville to Basically Romeo which killed over 20 people. What year was that? I think it was in the early 1900's

1896

[andyhb]

The tornado outbreak that is real interesting to me is the one that I'm not sure on the year is the F-5 tornado that went through Ortonville to Basically Romeo which killed over 20 people. What year was that? I think it was in the early 1900's

 

May 25th, 1896. Two days before the St. Louis/East St. Louis tornado and a part of that extremely destructive tornado outbreak sequence in the second half of May, 1896.

[andyhb]

Pretty high quality image here of the Hudsonville/Standale F5 on 4/3/56 near peak intensity. Reminds me a bit of the Oakfield, WI F5 when it was at its strongest, in appearance.

 

[huronicane]

Pretty high quality image here of the Hudsonville/Standale F5 on 4/3/56 near peak intensity. Reminds me a bit of the Oakfield, WI F5 when it was at its strongest, in appearance.

 

Here's a nice article from the Electronic Journal of Severe Storms Meteorology about the 4/3/56 outbreak:

 

http://www.ejssm.org/ojs/index.php/ejssm/article/view/36/35

[Hoosier]

May 25th, 1896. Two days before the St. Louis/East St. Louis tornado and a part of that extremely destructive tornado outbreak sequence in the second half of May, 1896.

May 1896 was ridiculous...obviously it doesn't get talked about much because nobody was alive back then...

[andyhb]

May 1896 was ridiculous...obviously it doesn't get talked about much because nobody was alive back then...

 

Yeah, there were numerous deadly/violent tornadoes across virtually everywhere except the SE. Sherman, TX on May 15th, Seneca/Oneida/Reserve/Sabetha, KS on May 17th, May 24th-25th in the Upper Midwest/Great Lakes and of course coming to a climax with the devastating St. Louis area tornado.

 

From reanalysis, it appears a similar pattern to May 2003 took place with several rather low amplitude shortwaves traversing the country within a potently unstable warm sector each time, although it is less reliable considering this is from more than 115 years ago.

[huronicane]

Tornado counts per county from 1997-2011 (if you add all the tornadoes up, it will add up to over 233, because of tornadoes crossing county lines.

 

 

In map form:

 

[Hoosier]

Who moved Genesee county to Oklahoma?

[huronicane]

I've never looked, but that would be an interesting list to see...the counties in the US with the highest tornado count.  You'd think at or around 1 per year would rank you pretty high on the list.

 

Thinking back to my time in Oklahoma, though...somewhere from 2-4 a year would probably be king of that mountain.

[Hoosier]

I've never looked, but that would be an interesting list to see...the counties in the US with the highest tornado count. You'd think at or around 1 per year would rank you pretty high on the list.

Thinking back to my time in Oklahoma, though...somewhere from 2-4 a year would probably be king of that mountain.

http://www.spegweb.com/tornado/us_county_area.php

[Stebo]

Pretty obvious where the local tornado alley is in the state, not surprising considering warm fronts love to stall between the M-21 and M-46 corridors.