The Fujita Scale Controversy

[Drew Terril]

I'll preface this by saying I'm not a met or an engineer. With that said, a good chunk of my time out of high school was doing residential construction, so I'm very familiar with the processes involved having done it much of my early adult life. In the areas I worked in, codes haven't changed since before I was in the field. Outside of those areas that have actually strengthened their building codes, it's difficult to find EF-5 DIs. With the vast majority of damage pics I see, I find (usually at a glance) where corners were cut in the construction process. In a lot of the poorer areas in particular, so many builds are funded by HUD, which will not pay for anything beyond what's needed to meet local code. My old boss when I was working for him took a HUD contract once, and he was so disgusted with how much they limited him that he never took another HUD contract again. CME foundations, bare minimum code compliance, inadequate amount of anchor bolts (which are already compromised to begin with when using a CME foundation), no hurricane straps. Basically we weren't allowed to build the HUD house to the same standard that we built everything else. I will also add that, for insurance purposes, a house that takes EF3 level damage is almost always going to be a full teardown and rebuild. Even at that point, it's structurally compromised enough that simply repairing what was damaged is not good enough. Think of it like an aluminum can. Once you bend the aluminum, it's much easier to crush it the second go around.

I personally am not of the opinion that DOW readings should be an automatic override when considering upping a rating if the other contextual clues don't match up. I also don't think "I think it was" should be enough either. I'm fully on board with it being yet another tool in the toolbox, but nothing more than that.

I do think a lot of the outcry that we see on social media on the subject tends to fall more or less in one of two groups. One, chasers who were on that storm and want that feather in their cap. Two, the perpetually online who always find something to be outraged about. Obviously, there are exceptions to that, but most of what I see on the subject when I do see it falls into one of those two.

Instead of constantly debating ratings, I would rather find out what can be done to improve building quality and techniques, even on the well built structures. I realize that at a certain point nothing will hold up, but that's the big thing that has me considering civil engineering over mechanical when I start working on my degree again. A big part of my wants to be a part of that side of the process, and while much of the weather enthusiast/chaser community complains about the involvement of engineers in the process, I'd like to see more of them involved. While I'm sure there are some mets out there who have worked construction in the past, or have an engineering background like Tim Marshall has, the vast majority do not have the trained eye that someone like me or a civil engineer would have. And that's where I would be happy to pull the burden of surveying storms from them and let experts in construction take that on. I have no doubt they work hard at it, but it's asking a lot to demand someone that primarily goes off a checklist to assume that workload.

 

[Randy Zipser]

Please see the attachment below: it is a comparison between the original F-Scale and the 2007 EF-Scale, with emphasis on the windspeed-range interval sizes (mph).

The first thing you'll notice is that the F-Scale and EF-Scale differ in interval sizes by a relatively small amount: 40-60mph difference from F1 to F5 (F0 was excluded since it has no specific lower-bound) and 20-40mph difference between EF0 and EF4 (EF5 was also excluded because it has no upper-bound).

One way to improve the current EF-Scale would be to incorporate actual archived data that likely already exists since 2007, when the EF-Scale was first introduced. In fact, Harold Brooks (SPC) and Tom Grazulis (Tornado Archivist & Publisher) might already have these data. The idea is to do an analysis for the CONUS showing the percentage of each EF-level (0-5) with respect to the total number of tornadoes that occurred in each year between 2007 and 2024, the last full year that such data would be available. Then, determine the 17-year average percentage for each EF "level." The end result would be a breakdown by percentage for each EF "level" over the last 17 years, which would be a reasonable period of time for a statistical data sample. Finally, a new EF scale, or as I discussed in a previous post, an "RF scale," would then be devised reflecting similar percentages per level as derived above, but that would not exceed the maximum windspeed limits that other sources of data (structural, photogrammetric, or Doppler radar) are indicating independently.

It's a "game" of playing with numbers, and actual historical long-term data from multiple sources will be input in this revision. This would be a good project for a master's thesis...gee, if only I were in college again (lol)!

 

[Tony Laubach]

Bowdle 2010 was a similar situation; rated EF-4. I don't recall the specifics as it was a long time ago, but we (our TWISTEX crew) were out with NWS for the survey the following day. High tension transmission towers were scoured off at the base and rolled hundreds of yards from their origin point. I recall the conversation about that being EF-5 level damage, however they rated it a 4 because "they didn't want the media attention" as the tornado ultimately missed town and avoided most structures.

My issue with the scale is damage done by a tornado can differ significantly based upon a variety of factors. Debris loading, particularly for a stronger tornado moving through a heavily developed area, will ultimately do more damage since there are more projectiles flying around as opposed to a tornado that moves over a more rural area.

I would love it if it were possible to take the same tornado, same intensity, size, but move it at different speeds over the same area, load more debris and see how that differs, etc. I feel like there are so many factors beyond the actual tornado that can greatly affect the damage done that ultimately gets the tornado assigned a certain rating.

Personally, in MY own opinion, I do not find it coincidental we've lacked a true EF-5 since Joplin, despite evidence of tornadoes mentioned above (Rolling Fork, Mayfield, etc). I'm not an engineer, and I haven't dug into those events enough to know whether those potential EF-5 indicators lacked certain engineering items that may have allowed them to fail in lesser winds. But I certainly wonder about it seeing the aftermath of some of these tornadoes.

Regarding the El Reno tornado... I get it... the damage done did not support the EF-5 rating. So long as that remains consistent, I will accept that for what it is.

But... my last point... we are continuing to add to the data collection mobile radar information obtained in more and more tornadoes, and I hate that most of that data isn't being better utilized. I don't know what the solution is, if there is a solution needed. But I think it goes without saying that the number of violent tornadoes is much higher than the records show, by how much, who knows. But because so many hit in areas that are less developed, they get the lower rating despite their actual strength. It's always been a pondering of mine, but alas, we work with what we've got. And the only true thing every tornado leaves behind is its aftermath... and until we can measure every single tornado somehow, that's probably how it's going to remain.

 

[Mike Smith]

Just as we have observed (and are measuring) changes in our long-term climate trends nowadays, short-term events such as tornadoes and hurricanes may also be evolving in subtle ways (i.e., getting stronger or more frequent?) over time

Hi Randy,

Dislike getting into global warming but there is zero evidence tornadoes are getting worse. Not only are there fewer high-end ratings, normalized tornado damage (for all intensities) is down.

Per the ACE Index, there is no upward trend in hurricanes.

Mike

 

[Randy Zipser]

Hi Randy,

Dislike getting into global warming but there is zero evidence tornadoes are getting worse. Not only are there fewer high-end ratings, normalized tornado damage (for all intensities) is down.

Per the ACE Index, there is no upward trend in hurricanes.

Mike

Absolutely agree, Mike. The statement I posed (using a question mark) was to point out a frequently-heard opinion attempting to relate stronger storms with climate change (notice I intentionally avoided using the term "global warming"). However, to a scientist, this opinion is purely speculation until enough "hard data" can be accumulated and applied to either prove or disprove it through scientific-method studies. Otherwise, we're both in agreement.

The EF-scale now being used has relatively uniformly-weighted range intervals. By weighting these windspeed-range intervals based upon an actual historical data sample, a different picture may (or may not) emerge over time about how tornadoes are rated going forward. The only way we'll ever know is to tweak the classification method and compare with current and previous methods for the purpose of seeing which method more realistically fits whatever damages structural engineeers are observing in the aftermath of actual events. Only more time (and study) will give us a clearer answer. RZ

 

[Boris Konon]

I've plotted the SPC database for tornadoes. No matter how you slice it concerning strong (EF2-3) and violent (EF4-5) tornadoes, the trends are *down*, whether you start at 1950 (period of record start), 1975 (Fujita scale introduced) or 2007 (EF-scale introduced). Yet the MSM and many individuals continue to push "tornadoes are getting worse." This suggests it is more connected to ideology and agenda rather than hard facts and science, and the blanket assumption "*all* weather gets worse in a warmer globe." That's simply not true, is linear thinking, and it is red flag when ppl talk in absolutes when concerning something as complex as the Earth's system.

Why is it that we are currently by *far* in the longest gap for an F5/EF5 tornado in the U.S.? And over time, you would think that since the EF-scale is based on damage, and there are more structures out there all the time to be hit ("targets"), stronger tornadoes, including EF5s, would be more common, yet the opposite is occurring.

I know one could say, "we have been just lucky we have such a long gap, and the strongest tornadoes has missed structures that would reveal their true intensity." But at what point is this not just "luck" and an actual real-world trend? How much time needs to elapse? 5, 10, 50 years?

Concerning hurricanes, the problem is that many think in the U.S., hurricanes all revolve around the Atlantic. Yes, the Atlantic has been very active since 1995, but the Atlantic basin only accounts for 15% of total global ACE annually. Cherry-picking has been going on, and the rest of the globe ignored. After all, if you are talking global warming, you need to include *all* of the globe's changes/trends. So what about the North Pacific? It has seen a marked decline since 1995, w/ many *inactivity* records set. And the North Pacific accounts for 55% of total global ACE annually. See this disparity/disconnect here? But as the media will say, "never let the facts get in the way of a good story (or narrative)." And they rely on ppl taking stories/reports at face value, and not applying any skepticism/critical thinking.
The quip "there is more to the story" is more relevant than ever these days.

 

[John Farley]

Wow, lots of good stuff in this thread. I have a couple thoughts to add. First, I am skeptical that the lack of EF-5s in recent years has much to do with better building standards. The vast majority of buildings out there were built before Greensburg, so when tornadoes hit buildings, they are no mostly hitting buildings built to the new better standards, unless they hit more-or-less entirely rebuilt communities like Greensburg, Joplin, or Parkersburg. Most of the time, they are not.

My other point is that much of the confusion about the Enhanced Fujita Scale is that, although it is technically a damage scale, it is regularly discussed, even by the NWS, as if it were a wind scale. Damage surveys usually include peak wind speed estimates, often to a level of precision far beyond what can be justified. For example, Can we really say that the St. Louis tornado on May 16 was 152 mph, not 150 or 155? Further contributing to the confusion is when measured wind speeds are included in damage surveys but not used to rate the tornado. For example, one of the tornadoes in the tornadofest west of Lubbock last Thursday was rated EF-U (unknown), but the following was added at the end of the NWS Lubbock damage survey report for that tornado: "No damage was observed or reported with this tornado. However, a peak
wind speed up to around 120 mph (EF-2) is possible with this tornado
based on KLBB WSR-88D peak rotational velocity of 45 kts, and NSSL
RaXPol rotational velocity of 80 kts."

I do not have a big problem with this, as it is useful information, but it does likely make people less informed than most of the people on this list wonder why it was rated EF-U. I do kind of like Hannah's suggestion of a wind speed rating separate from the EF rating, at least when we have good wind speed data close to the ground. I am sure that is only the case with a minority of tornadoes, though - probably a small minority.

 

[Sean Ramsey]

I've always believed they could tweak the scale to be more informative yet keep historical data intact. Basically by adding two characters to the EF scale they could measure both wind speed and damage. Using the EF side of things to measure wind speed using best method (whichever gives the most accurate data - Mobile doppler, etc) and then a letter representing the impact and a number rating the impact, a useful gauge could built.

For example:

A wedge measured with 300+ mph winds but does no damage might be an EF5R0.

A rope tornado with estimated winds speeds of 100 mph and damages a barn might be EF1R1.

A stove pipe with 134 mph winds but plows through a small town might be EF2R4.

To me, something like this would be far more informative and useful without compromising past data.

 

[Randy Zipser]

I've plotted the SPC database for tornadoes. No matter how you slice it concerning strong (EF2-3) and violent (EF4-5) tornadoes, the trends are *down*, whether you start at 1950 (period of record start), 1975 (Fujita scale introduced) or 2007 (EF-scale introduced). Yet the MSM and many individuals continue to push "tornadoes are getting worse." This suggests it is more connected to ideology and agenda rather than hard facts and science, and the blanket assumption "*all* weather gets worse in a warmer globe." That's simply not true, is linear thinking, and it is red flag when ppl talk in absolutes when concerning something as complex as the Earth's system.

Why is it that we are currently by *far* in the longest gap for an F5/EF5 tornado in the U.S.? And over time, you would think that since the EF-scale is based on damage, and there are more structures out there all the time to be hit ("targets"), stronger tornadoes, including EF5s, would be more common, yet the opposite is occurring.

I know one could say, "we have been just lucky we have such a long gap, and the strongest tornadoes has missed structures that would reveal their true intensity." But at what point is this not just "luck" and an actual real-world trend? How much time needs to elapse? 5, 10, 50 years?

Concerning hurricanes, the problem is that many think in the U.S., hurricanes all revolve around the Atlantic. Yes, the Atlantic has been very active since 1995, but the Atlantic basin only accounts for 15% of total global ACE annually. Cherry-picking has been going on, and the rest of the globe ignored. After all, if you are talking global warming, you need to include *all* of the globe's changes/trends. So what about the North Pacific? It has seen a marked decline since 1995, w/ many *inactivity* records set. And the North Pacific accounts for 55% of total global ACE annually. See this disparity/disconnect here? But as the media will say, "never let the facts get in the way of a good story (or narrative)." And they rely on ppl taking stories/reports at face value, and not applying any skepticism/critical thinking.
The quip "there is more to the story" is more relevant than ever these days.

You nailed it, Boris! Excellent points you brought up here.

I was "called out" by my master's thesis committee on a statement I had made that tornado wind speeds will likely be lower in the future. I was able to successfully defend that statement, but I knew my committee was skeptical. That was in 1976. Fortunately, for everyone in the path of a tornado, this may not mean much, but a long-term, consistent trend downward is very meaningful over time to our society in terms of injuries, lives lost, and enormous property-damage costs.

And, as you have correctly indicated, there is no direct correlation statistically between tornado wind speeds and any of these three aftermath outcomes. In fact, the very opposite may be true for reasons that have nothing to do whatsoever with what is going on with Earth's atmosphere (e.g., economics, demographics, urbanization/migration trends, etc.). So, "climate change" may or may not be relevant in that equation...but one thing is fact: anything that cannot be proven over a long period of time through acceptable scientific principles is purely speculation, opinion, and in its worst case, conspiracy theory. What we strive for is a factual understanding of the atmospheric phenomena we observe and measure...but arriving at those facts could be elusive, even more so than in our lifetimes.

 

[Randy Jennings]

the Enhanced Fujita Scale is that, although it is technically a damage scale, it is regularly discussed, even by the NWS, as if it were a wind scale.

I made a similar comment in the presence of Dr. Doswell once. He was quick to correct me and inform me that the EF scale is not a damage scale. He said it was a wind scale and that damage was just a proxy for wind. While I still believe that if it looks like a damage scale and smells like a damage scale, it is a damage scale, I also believe that one is foolish to not take his thoughts very seriously. Wish he was still alive today to have a more in depth discussion on this topic.

 

[Randy Zipser]

He said it was a wind scale and that damage was just a proxy for wind.

Dr. Doswell was exactly right in making this statement. The original F-Scale was based upon some modeling that Dr. Fujita (or his students?) supposedly conducted in the late 1960s or early 1970s (I do not know the exact date or specific history since I was not at the U of Chicago). I believe that Mike Smith touched upon some of these details, however, in an earlier attachment in this thread.

The EF-Scale, however, was developed as a result of repeated experimentation by knowledgeable wind-engineers (or aspiring grad students preparing for careers in that specialized field) using materials and designs to replicate materials and designs encountered in actual structures that were damaged by tornadoes. Much of that early work was conducted at Texas Tech under Dr. Joe Minor, if I remember correctly. Those experimental observations were then taken into the field to compare with damage patterns (and causes for structure failure) that were found on aftermath fields surveys. What these findings showed was that engineer-derived windspeeds were consistently lower than the meteorologist-derived model that was being used at the time, which led ultimately to the F- to EF-scale conversion in 2007.

The EF-Scale is still in use today, 18 years later, because the NWS came to believe that wind damage is a more reliable indicator of "true" windspeed near the ground than computer modelling (including Doppler-derived) which is only as good as the data that is input and programming upon which it is based (the expression, "Garbage In, Garbage Out!," comes to mind). It is much harder to manipulate wind data (for whatever reason)
than to play with a computer model to fabricate whatever yields a desired result.

I believe that the right decision was made to use observed structure damage rather than theoretical computer models to come up with the tornado wind speed classification system we have in place today. The trick is to tweak the EF scale, as necessary, if field observations change over a long enough period of time to be called a statistical trend.

I am in the midst of a thunderstorm as I write this, with lots of CG lightning and crashing thunder, so I'll sign off with this discussion!!

 

[Boris Konon]

Wow, lots of good stuff in this thread. I have a couple thoughts to add. First, I am skeptical that the lack of EF-5s in recent years has much to do with better building standards. The vast majority of buildings out there were built before Greensburg, so when tornadoes hit buildings, they are no mostly hitting buildings built to the new better standards, unless they hit more-or-less entirely rebuilt communities like Greensburg, Joplin, or Parkersburg. Most of the time, they are not.

My other point is that much of the confusion about the Enhanced Fujita Scale is that, although it is technically a damage scale, it is regularly discussed, even by the NWS, as if it were a wind scale. Damage surveys usually include peak wind speed estimates, often to a level of precision far beyond what can be justified. For example, Can we really say that the St. Louis tornado on May 16 was 152 mph, not 150 or 155? Further contributing to the confusion is when measured wind speeds are included in damage surveys but not used to rate the tornado. For example, one of the tornadoes in the tornadofest west of Lubbock last Thursday was rated EF-U (unknown), but the following was added at the end of the NWS Lubbock damage survey report for that tornado: "No damage was observed or reported with this tornado. However, a peak
wind speed up to around 120 mph (EF-2) is possible with this tornado
based on KLBB WSR-88D peak rotational velocity of 45 kts, and NSSL
RaXPol rotational velocity of 80 kts."

I do not have a big problem with this, as it is useful information, but it does likely make people less informed than most of the people on this list wonder why it was rated EF-U. I do kind of like Hannah's suggestion of a wind speed rating separate from the EF rating, at least when we have good wind speed data close to the ground. I am sure that is only the case with a minority of tornadoes, though - probably a small minority.

I should have added this caveat to my original post concerning tornado intensity and trends.

It is w/o question many tornadoes are underrated, simply b/c they do not hit anything at all, do not hit anything built well enough to reveal their true intensity, or they just are not at peak intensity along their track when they hit well-built structures.

So many EF4s are actually EF5s, many EF0s are EF1s, and so on. In fact, it is also w/o question some EF0s/EFUs in the database would be EF4/5 if they hit well-built structures. How many times have we seen a large and obviously violent tornado based on its wild cloud motions alone in the middle of nowhere?

Mobile radar studies have suggested that many tornadoes appear to have EF2 winds at some point in their existence. So this ties back to how strong a tornado is (or not) along its track when it hits something substantial.

And this study proposes that more than 20% of all tornadoes could be EF4/5. That is much higher than the 1-2% that our database currently indicates!

https://www.pnas.org/doi/10.1073/pnas.2021535118

This is case where our technological methods simply cannot yet tell us what actually is occurring in the atmosphere. We are getting better, but still a long way to go. The same applies to TCs, except technology has caught up better. We started using GPS dropsondes in 1998, and the SFMR in 2004. This has given us a much better picture of TC winds, esp. really high wind speeds in Cat 5s. But still, a big gap here as well b/c we do not have recons in every TC, and outside of the Atlantic and eastern Pacific, recons into TCs are very sparse or non-existent. Satellite estimates using the Dvorak Technique can only do so much. It works best for TCs 60-105 kt, and it really has trouble w/ those small/tiny TCs than have very tight inner cores/pinhole eyes. Based on satellite alone, EPAC Patricia in Oct 2015 likely would have been capped at 160 kt, but recon showed it was actually 185 kt! Based on this, it is almost w/o question other TCs globally have reached or exceeded this wind speed in the last 40 years alone (WPAC Haiyan in Nov 2013 comes to mind).

 

[Boris Konon]

You nailed it, Boris! Excellent points you brought up here.

I was "called out" by my master's thesis committee on a statement I had made that tornado wind speeds will likely be lower in the future. I was able to successfully defend that statement, but I knew my committee was skeptical. That was in 1976. Fortunately, for everyone in the path of a tornado, this may not mean much, but a long-term, consistent trend downward is very meaningful over time to our society in terms of injuries, lives lost, and enormous property-damage costs.

And, as you have correctly indicated, there is no direct correlation statistically between tornado wind speeds and any of these three aftermath outcomes. In fact, the very opposite may be true for reasons that have nothing to do whatsoever with what is going on with Earth's atmosphere (e.g., economics, demographics, urbanization/migration trends, etc.). So, "climate change" may or may not be relevant in that equation...but one thing is fact: anything that cannot be proven over a long period of time through acceptable scientific principles is purely speculation, opinion, and in its worst case, conspiracy theory. What we strive for is a factual understanding of the atmospheric phenomena we observe and measure...but arriving at those facts could be elusive, even more so than in our lifetimes.

Along the same lines concerning tornado winds and the future. While CAPE increase is likely b/c warmer overall promotes more convective instability, I get the impression that ppl in the wx field and in the public are far more impressed/wow'ed by convective instability/CAPE, rather than the dynamical part of it, esp. wind speed and shear.

You can have all the convective instability/CAPE in the world, but if the wind fields sfc and aloft are weak or disorganized, the vast majority of the time, you will not get tornadoes. Sure, there are exceptions, like deviant storm motion that greatly enhances storm-relative helicity (e.g. Plainfield IL 1990 and Jarrell TX 1997), but those are very rare. And any thunderstorm, or TCU for that matter, has the potential to produce a non-mesocyclonic tornado, but the vast majority of strong/violent tornadoes require strong winds sfc and aloft, and even more important wind shear, and taking it one step further, the correct *type* of wind shear!

Strong winds/wind shear require baroclinicity. As the globe warms w/ the polar regions warming the most and little overall change in the tropics, the temp gradient between the tropics and polar regions decreases globally, which in turn decreases baroclinicity. This means a weaker jet stream, and less intense non-tropical low pressures.

Most of the significant tornado outbreaks and stronger tornadoes occur in the warm sector of a developing low pressure system. Even if you do not have significant sfc low development, the features aloft such as the trough at 500/300 mb and the jet structure can be more than enough (re: 5/3/1999 in OK). Either way though, all of this is dependent on baroclinic zones in one form or another, which in the mean, decrease as the globe warms. We know for a fact the polar regions are indeed warming faster than the tropics, so there little to debate on the large-scale changes here and how that impacts synoptic-scale features like low pressure systems.

Supercell formation almost always requires a certain amount of wind shear in the atmosphere, typically 35 kt or more 0-6 km. That's just to get a mid-level mesocyclone. That in itself is far from a tornadogenesis though. You need good low-level wind shear (0-1 km) for tornadogenesis, and is an almost a must for the stronger tornadoes. And it goes much further as to how conditional things are. For instance, what are the T/Td spreads at the sfc? Is there enough 0-3 km CAPE? These little details can make all the difference between no tornadoes and significant tornadoes, even if all other factors are great. In addition, what is the storm motion? Storms moving E or NE tend to be the best tornado producers while those moving SE are less b/c they tend to rain into their inflow. All these details need to be considered when trying to forecast future changes in climate and the sensible wx that occurs. Sure, it is easier to predict large-scale changes, but not so easy to predict all the small-scale changes, and when it comes to mesoscale phenomena, the details are often everything!

We know this well empirically. How many days does it look great on paper, and high-end tornado outbreak seems for sure, and big fail (so many HIGH risk days - May 20, 2019 comes to mind). Conversely, how many times does it look rather bland or borderline, and it overperforms (Campo CO on May 31, 2010 and Chapman KS on May 26, 2016 are classic examples). Cases like this clearly show we have a lot to learn and a long way to go to really know w/ any decent confidence how things may or may not change in the future. We can't even simulate the atmosphere well enough in real-time many cases for a good mesoscale forecast, and end up being surprised or scratching our heads as to what happened!