Where have all the violent tornadoes gone?

[Marc Austin]

I too disagree with many of the recent tornado intensity ratings. I won't go into which tornadoes I believe have been under-intensified because that has already been discussed. Something I did want to point out though, was that the estimated winds in the 5/3/99 Moore, OK tornado, which were reported to be around 318mph would fall into the EF-6 category, which doesn't exist in the EF scale. It seems like the scale should include higher ratings. Just because we haven't seen many tornadoes that might be classified as an F5+ doesn't mean they don't exist. I understand I wasn't on the ground and I didn't see any of the damage from these past outbreaks up close, but I have seen footage from previous tornadoes rated F4/F5 and the damage is almost identical. It is good that some effort is being made to enhance the scale to make it more accurate, but I still see flaws in the system.

 

[Greg Stumpf]

Beginning next February, when the new EF-Scale is implemented, it won't matter if a house is swept from its foundation - if there are no changes to the damage indicators between now and then, the highest rating you can get with homes is F4.[/b]

I'm part of the team helping develop NWS training (via WDTB) for the new EF scale. In a recent meeting with the lead scientists who helped develop the EF scale, we discussed the possibility of adding new damage indicators (DI), and higher degrees of damage (DOD) for some of the existing DIs. This includes the possibility of adding a DOD for a well-built well-anchored single/double family residence being swept off a foundation (with the usual caveats - no "sliders", duration of wind, debris from other structures, etc). The discussion is in light of some of the concerns being discussed in this forum.

So, you can all calm down now, and there is no vast government conspiracy hiding the number of violent tornadoes out there!

 

[Jeff Snyder]

So if a house is obliterated and the debris is swept away from the original foundation, that could at the worst be F4? :huh: I understand you're stating fact but that is just astonishing.

Just my 2 cents.
[/b]

Why is that so astonishing? If a tornado with 300mph winds goes through only a forest, the tornado cannot be rated F5. Period. There are many damage indicators (implicit now; explicit when the EF scale goes into effect), and some indicators can only tell you the possible wind range up to a certain point. For example, how do you argue that the complete and utter obliteration of a stop sign means that the tornado had to have had winds >200mph? You can't... You can probably say that the tornado had winds of at least xx mph, but after that, it's uncertain. Engineers have shown that the complete destruction of a house does NOT always require F4-F5 level winds. I wish we could get Tim Marshall on here to give some examples with assessments he's done, but (as you've heard) others who assess tornado damage now agree. As it stands now, it's been said that a 140mph winds can completely destroy some houses and "wipe the foundation clean". The F-scale rates damage, but in the attempt to determine the intensity of a tornado. What good would the scale be if we can label tornadoes with relatively weak winds (~140mph) as F5?

Re: not using a point source damage to say that a tornado was not this or that level (address Justin's comments)... True, just because a mailbox is standing does not necessarily mean that the winds that blew over a house weren't 'violent'. However, you can say just the opposite as well -- just because a house was blown over while leaving a mailbox untouched does not mean that the tornado was violent. I trust that the experts who assess strong and violent tornadoes take ALL damage into account, and make the best assessment based on what they see. I have 0 engineering experience, and I suspect that goes for most of the folks on this board. I can 'guess' what I think the final rating for a particular tornado may be, but when it all comes down to it, there are people who are far more qualified to assess the damage than I (and the majority of this board) are. Do we closely look at the anchoring of the house on the foundation? Do we talk with the home owners? Do we get to look closely at nearby indicators (trees, cars, etc)? No, we don't. We get a handful of pictures, most at relativley low resolutions.

Re: needing more than 6 tornado intensity / damage levels... We haven't had an F5 in 6 years, so I can't imagine that we'd really want an F6 (EF6). If F5s are already extremely rare, how rare would an F6 be? What's the purpose of a scale if we only see the top rating given out for a tornado once every century? Of course, right now, that's moot, since an F6 would not be given, and there will never be an EF6 (there is no upper-bound to EF5).

My hunch is that only a relatively small fraction of past F5s were really F5s (particularly before 1970). So, are we skewing the tornado intensity / damage distribution by making the ratings more "strict"? Is it better continue over-rating tornadoes? Since when is it in the best interest of since to intentionally misrate a tornado? If science is the systematic search for the truth, asking damage assessors to disregard their latest knowledge is contradictory to their purpose. Besides, what would be the motivation to intentioally underrate a tornado? I can see MAYBE if there was very little or no lead time, but that wasn't the case on 4/7. Let's give these guys some credit, especially as we sit in our chairs hundreds of miles away looking at small pictures on the TV or computer screen.

 

[Dan Robinson]

If things like debris impacts and long-duration winds disqualify a swept-clean house from F5, then has there ever really been an F5 in history? Can we expect there to be one in the future? Jarrell was slow moving, and Moore was large and debris-filled - both disqualifiers for F5 under the new criteria.

I see the need to adjust the scale to reflect accuracy - but as it seems now, a tornado today that produces even Jarrell and Moore-like damage will have slim chances getting the F5 rating. If I'm reading the EF document correctly, to get an F5 you'd need things on the order of a large building like a school totally destroyed or high-rise building either deformed or collapsed. Has anything like that ever happened?

 

[Bill Tabor]

Seems like it all really depends on the purpose of the scale ratings. When you consider that all tornadoes before something like 1970 were rated after the fact and by groups of students going through old newspaper articles etc, it introduces a LOT of error in the records. Now combine that with all the tornadoes that never hit anything and are rating F0. Now throw in all the very weak tornadoes that are often reported as 'brief spinups'. Also toss in the fact that US countryside is more populated now than in the past. Also throw in more spotters, chasers, and creation of the old SPC (whatever it used to be called) Basically you get a very poor picture (highly skewed) of what type of tornadoes are affecting the US each year from an objective source. You only get a sample of what was hit. Probably early in last century only F1 or stronger (doing more significant damage) got reported too. Now on top of all that we are switching to EF with all the DI and DOD's. So is the scale for insurance adjustment purpose, scientfic study purposes, or what. Seems from a sci standpoint all we have truly is garbage. Is there a reason for insurance companies to want to rate tornadoes lower? Do lower ratings allow less money paid? If so, perhaps we have found the real basis behind the ratings system because otherwise it seems to have no purpose.

EF may be fine, but the way I see it is you really have to get rid of old data and start fresh to consider it with any accuracy.

EDIT: Just realized scale may not matter for insurance appraisals but it might be a factor for the amount of relief provided by the government as far as definition of a 'disaster area'. I could be off here but it may be easier to get goverment disaster funding if F4's and F5's scoured an area than a group of F1's thru F3's. Maybe not. What do you think?

 

[Bill Tabor]

Another point here agreeing with Alex. EF scale is supposed to insure the consistency of the old ratings data; however if you change how you rate tornadoes quantitatively enough based on new indicators and throw out other old ratings rules such as allowing houses to be rated F5 then...in essence you have broken the continuity of the two ratings systems. BAM! I love it that's it! I just nailed the core reason why EF has a real problem - it isn't really consistent with the old ratings system and it is required to be.

Also, as Alex mentions the old scale didn't care about the wind speeds really. That was just an arbitrary value given as a range by Fujita to rank a tornado scale. As Alex says Fujita layed out some basic suggestions to use for ranking. Simple rules like houses completely removed from foundations. Sure old construction might not be as good as today's construction, but the attempt was not to rate the tornado, but to rank the hazard to life. Since historically (and in those old newspaper articles) we never really knew the quality of construction of the houses you just have to make a determination based on definition what the tornado did. If it blew it away, it blew it away - F5. Knocked it all down but left it on the ground F4...and so on.

So there you have it. I don't think any of you can refute this. Based on historical inaccuracy of tornado ratings and previous assumptions made in ratings if one EF objective is to correlate to old database I don't think it can do it. This is ESPECIALLY true if it changes fundamental rules such as no longer allowing houses to be rated higher than F4!

The only thing you can do is start a new ranking system and perhaps maintain the old side by side. For scientific study only the new scale would be more reliable and accurate. The old scale would provide some level of comparison historically but will always be flawed.

 

[Greg Stumpf]

Let me preface this post with a disclaimer that these are my opinions, and not necessarily the opinions of the EF scale development team, training team, NOAA, NWS, NSSL, or my employer...

If things like debris impacts and long-duration winds disqualify a swept-clean house from F5, then has there ever really been an F5 in history? Can we expect there to be one in the future? Jarrell was slow moving, and Moore was large and debris-filled - both disqualifiers for F5 under the new criteria.[/b]

We've had the same debate about these two events. First, clarifying what I said about debris...I didn't mean to imply that a debris filled tornado should reduce the chance of an F5 designation, but that it could reduce the rating. Consider the 3 May 1999 tornado segment through Moore, west of I-35. There were a few "slabbed" homes in that subdivision. However, these homes were densely packed (1/5 acre plots, I believe), and there is a good likelihood that the surrounding homes that were completely destroyed but not slabbed (rated F4) had debris from other homes coming to rest on their foundations, yet the original home debris was swept away. In other words, to completely "slab" a home, you would have to be sure that debris from other homes would not replace the original debris on that slab.

However, there were a number of places along the OKC path where F5 was unmistakable, especially in Bridge Creek, where the ground was scoured to mud and all vegetation (e.g., cedar trees, bois d'arc) were sheared at ground level. This was in areas where other homes were slabbed. Plus, a group of about 3 well-built residences in the East Lake subdivision were slabbed, and there were no "upstream" homes to replace debris. In the end, the OKC tornado would still remain F5.

The Jarrell tornado rating is one of some debate. The present Fujita scale is used in practice as a damage rating scale, and not a wind scale, because the winds have never been calibrated. However, the Enhanced Fujita scale attempts to bring wind speeds back into the equation, and for good reason (wind speeds are particularly useful for engineering considerations, insurance purposes, etc). This is why each DOD has estaimated winds speeds given in a range, to account for variations in construction of the DI. Therefore, if a DOD of a slabbed well-contructed home was added, then the range of wind speeds might have to take into account things like wind duration, possibly accounting for lower wind speeds for similar damage if the wind duration was longer than normal. Now, I didn't officially see the Jarrell damage first hand until 3 months after the event. One thing that surprised me right off the bat was that the Double Creek Estates subdivision weren't "estate homes" like I had thought, but rather small, ~1000 square-ft four-room lower-income homes, and possibly of questionable construction (the aerial images of empty slabs didn't provide me a good sense of scale). Nonetheless, I can provide no personal estimate of the rating since I wasn't part of the damage assessment team.

That being said, there has been no real attempt to study (to my knowledge) what kind of impacts duration has on degree of damage. Some have pointed toward hurricane winds as a long-duration wind that could be used for comparison. However, it should be noted that straight lines winds are very different from tornadic winds, in that the latter have considerable changes in speed and angular acceleration (especially with embedded suction vortices), and a considerable vertical component in the vortex corner flow regions.

I see the need to adjust the scale to reflect accuracy - but as it seems now, a tornado today that produces even Jarrell and Moore-like damage will have slim chances getting the F5 rating. If I'm reading the EF document correctly, to get an F5 you'd need things on the order of a large building like a school totally destroyed or high-rise building either deformed or collapsed. Has anything like that ever happened?[/b]

Same arguments are being debated by the EF scale team, hence the possibility of adding more high end DODs on existing DIs. New DIs are also being considered, including automobiles, trains (empty and full boxcars), and road asphalt (of varying constrution grade).

 

[J Holder]

I'm really wondering how the insurance agencies are going to use the new enhanced Fujita rating?

JH

 

[Jack Beven]

Josh and all,

The National Hurricane Center over the last few years seems to have gotten more stringent with its rating requirements. For example, Katrina's final rating for its LA landfall was lowered to Cat 3 (despite its extremely low 920-mb landfall pressure) because the wind data just didn't support Cat 4. Similarly, Rita was barely rated a Cat 3 for its TX/LA landfall despite a very low central pressure of 937 mb and the complete destruction of several SW LA communities. It seems the NHC is making an extra effort to really qualify each rating. As can be expected, these downward adjustments in the ratings of these big "headline storms" have proven very controversial with online bloggers.[/b]

It's less a matter of being more stringent with the requirements than a) better data about the surface and low level winds in a hurricane, and B) better understanding of the structural changes that occur during a hurricane's life. On a), data from instuments such as the Stepped Frequency Microwave Radiometer and GPS dropsondes (along with doppler radar and instrumented towers) have given us more data on the surface winds and those below the level that our aircraft fly at. And that has enabled us to do a better job of evaluating the intensity.

On B), since the introduction of the GSP dropsondes we've learned a lot about the structure of the lowest levels of the hurricane and how they evolve. We've discovered the often very strong low-level wind maxima that occur in the eyewalls of intensifying hurricanes. We've seen those maxima decay, or on some occasions seemingly fall to the surface as storms weaken. We've also seen the changes (both horizontally and vertically) that occur as a hurricane undergoes an eyewall replacement cycle. However, we know that we haven't seen everything yet, and that we still have a lot to learn about the hurricane boundary layer.

If you look at the last page of our report on Katrina (http://www.nhc.noaa.gov/pdf/TCR-AL122005_Katrina.pdf), you'll see cross-sections of airbourne doppler radar data taken when Katrina was near peak intensity and when it was decaying. The difference is striking. The wind structure of the classically organized hurricane with the strongest winds in the eyewall on August 28 has to a great extent broken down before landfall on August 29, accompanied by a 20-30 kt decrease in the observed winds aloft. Some of this is likely due to the eyewall replacement, some of it possibly due to the decrease in oceanic heat content north of the Loop Current, and some of it likely due to things we still need to learn about.

I wonder if it gets lost in the debate over Katrina's landfall intensity that as a weakening 920 mb system it did not have the same structure or intensity it did as a strengthening 920 mb system. Ditto for Rita. Both storms underwent noteworthy changes as they started to weaken.

The question that arises: if the NHC is using newer, stricter methodology for rating the intensities of modern landfalling hurricanes, how do we compare these modern hurricanes to big historic ones that may have been overestimated? The NHC is currently doing a reanalysis of all USA hurricane landfalls dating back to 1851. It is widely expected that the estimated intensities of some big, historic storms may be adjusted downward based on the new knowledge and methodologies. The benefit to this will be a consistent standard-- and cleaner comparisons between modern and historic events.[/b]

As a member of the committee who approves all the changes to the Atlantic hurricane best track data base, I first have to say there are a lot of problems with it. There is little or no documentation on how many of the tracks were originally derived, and there are a fair number of storms that have been known to be wrong or questionable for years.

As we've gone through the revisions (we're currently in the 1910's), we've added some storms, we removed some storms, and we've changed the intensity of storms both up and down. We've been given local accounts of landfalls that helped us to refine both track and intensity (and we could use a lot more). We've run across situations where the various source materials violently disagree with one another, and even after extensive analysis it's very difficult to tell which of them (if any) were correct. And we haven't even reached the recon/satellite eras yet, where things are going to get *really* interesting.

Even after we're done, some uncertainty is going to linger on the older storms. In most cases, there's just not enough data to let us even come close to the level of analysis we do today.

Jack Beven
Tropical Prediction Center

 

[Bill Tabor]

Perhaps I didn't make my points clear so let me reiterate and summarize my conclusions:

1) There really has to be an established purpose and intent for using the ratings system and database. Any system should accurately and consistently support that purpose. The purpose should be clearly stated in a 'Mission' statement.

2) As I see it there are currently two purposes primarily at the moment: a) appraisal for insurance and government help in an emergency; b ) study and statistics of historical tornadic storms with climate / future warning implications.

3) There has been a steadily increasing trend with time to more meticulously study tornado reports and type of damage. Greg's notes above illustrate this. Based on this and the way that historical tornadoes were reported or under reported it brings into question a couple of things. a) Is the new and changing system unfair (appraisal wise) to past or future recipients of aid, and are these changes really politically motivated (perhaps by special interest); b ) The consistently changing nature / source of the historical tornado database along with changes in the ratings system (including changing interpretation of how to rate) will / has seriously damaged any ability to accurately compare current tornadoes and tornado outbreaks and systems to those of the past.

 

[Greg Stumpf]

As I see it there are currently two purposes primarily at the moment: a) appraisal for insurance and government help in an emergency; b ) study and statistics of historical tornadic storms with climate / future warning implications.[/b]

More importantly, to assess seasonal and geographic risk and help mitigate impacts. There is considerable interest from the engineering community to help build more storm-resistant structures, etc.

a) Is the new and changing system unfair (appraisal wise) to past or future recipients of aid, and are these changes really politically motivated (perhaps by special interest); b ) The consistently changing nature / source of the historical tornado database along with changes in the ratings system (including changing interpretation of how to rate) will / has seriously damaged any ability to accurately compare current tornadoes and tornado outbreaks and systems to those of the past.[/b]

If there is some special interest group politically changing the way tornadoes are rated, me and my damage survey colleagues are completely unaware of it. I think this assertation is treading on the thin line of government conspiracy theories, which, IMHO, is ridiculous concerning something as simple as a difference in one rating level for certain tornadoes. In the grand scheme of things, this doesn't make much of a difference, and bear in mind that the F-scale and the EF-scale are still loaded with subjectiveness. One need only take the following exercise with a large group of folks to see what kind of distribution of F-scale ratings can be applied among multiple surveyers given the same damage indicators.

As for the changing nature of the historical database, one simply needs to look at the dramatic increase of tornado reports starting in the 1950s. Harold Brooks deals with this all the time in his climatological studies, and there are ways to correct for these reporting changes over the years. However, I share the same concerns that most posters have noted with regard to keeping some consisitency with the past, especially in regard to F5 damage to well-built residences. Otherwise, the new system is a vast improvement over the old, in that there are better ways to determine a rating and estimate the wind speeds associated with said damage. It is also open-ended, allowing for more DIs to be added in the future, recalibartion of wind speeds, etc.

As for the Gallatin F3 versus F4 argument - how many ST subscribers actually saw said damage in person, and have enough damage survey experience to provide an honest assessment? I know that some of the images and descriptions I saw/read look suspiciously like F4, but I wasn't there in person, so my opinion can only remain unofficial. I received second hand reports that Tim Marshall (a National QRT member) may have been involved in that damage survey. Note that the final rating looks to remain F3 (as of today's PNS).

 

[Tony Lyza]

More importantly, to assess seasonal and geographic risk and help mitigate impacts. There is considerable interest from the engineering community to help build more storm-resistant structures, etc.
If there is some special interest group politically changing the way tornadoes are rated, me and my damage survey colleagues are completely unaware of it. I think this assertation is treading on the thin line of government conspiracy theories, which, IMHO, is ridiculous concerning something as simple as a difference in one rating level for certain tornadoes. In the grand scheme of things, this doesn't make much of a difference, and bear in mind that the F-scale and the EF-scale are still loaded with subjectiveness. One need only take the following exercise with a large group of folks to see what kind of distribution of F-scale ratings can be applied among multiple surveyers given the same damage indicators.

As for the changing nature of the historical database, one simply needs to look at the dramatic increase of tornado reports starting in the 1950s. Harold Brooks deals with this all the time in his climatological studies, and there are ways to correct for these reporting changes over the years. However, I share the same concerns that most posters have noted with regard to keeping some consisitency with the past, especially in regard to F5 damage to well-built residences. Otherwise, the new system is a vast improvement over the old, in that there are better ways to determine a rating and estimate the wind speeds associated with said damage. It is also open-ended, allowing for more DIs to be added in the future, recalibartion of wind speeds, etc.

As for the Gallatin F3 versus F4 argument - how many ST subscribers actually saw said damage in person, and have enough damage survey experience to provide an honest assessment? I know that some of the images and descriptions I saw/read look suspiciously like F4, but I wasn't there in person, so my opinion can only remain unofficial. I received second hand reports that Tim Marshall (a National QRT member) may have been involved in that damage survey. Note that the final rating looks to remain F3 (as of today's PNS).
[/b]

I would like to note that the PNS text on the Gallatin tornado is the same text from a few days ago.

 

[Andy Wehrle]

I think this page pretty well hits the nail on the head as to what Alex is trying to get at here:

http://www.crh.noaa.gov/pah/?n=nov1505f4

* EVENT TYPE: F4 TORNADO

* EVENT LOCATION: 1.5 MILES SOUTHWEST OF EARLINGTON TO 7 MILES EAST
OF HANSON.

* PEAK WIND: 210 MPH

... ...

* EVENT TYPE: F3 TORNADO

* EVENT LOCATION: FROM APPROXIMATELY 2.5 MILES SOUTH OF MAYFIELD
IN GRAVES COUNTY...THROUGH MARSHALL COUNTY AND THE LBL
RECREATION AREA...TO SARATOGA IN LYON COUNTY

* PEAK WIND: 200 MPH (PEAK WIND OCCURRED FROM BIG BEAR HIGHWAY TO
MOORS RESORT)

 

[Dan Robinson]

I can see one reason that the leveled large brick homes in Gallatin could get disqualifed from F4 under the new criteria. The subdivision we documented was immediately upstream (South Browns Lane, behind the Lowe's) that would have provided plenty of debris to fill the tornado before it struck the brick homes. The brick homes were only a fifth of a mile or so away from the subdivision we toured. However, from what I could see from actually being there, it appeared that most of the debris from the South Browns Lane area stayed there - as there were piles of rubble where the houses once were and a somewhat minimal 'scatter' of large debris propagating away from each house. We did find roof truss sections immediately behind the Lowe's building, which would indicate that some larger debris could have been airborne at high velocity when the tornado struck the large brick homes.

 

[Jeff Snyder]

The picutres that Sheila shows look like structures just collapsed down. You see tons of debris, which tends to be centrifuged out in violent tornadoes. It's tough to get perspective since the pictures are pretty zoomed in (what does the entire neighborhood look like?) , but the trees in the 2nd pic don't look like they're too bad off, and there isn't much damage to the surroudning houses (in the background). Again, this analysis is pretty trivial, since we don't have the entire overview from which this pictures were taken. If those are two structures destoyed amongst an entire neighboorhood, can we really say much about it? Would a tornado "strengthen" for 40 feet if it's moving at 45mph? I mean, I'm sure it could, but I think chances are better that either the structures were more poorly built or something big smashed into them. Again, however, this analysis is fruitless since we've seen time and again that impressive debris doesn't necessarily equal impressive tornado (intensity-wise). 130mph can "wipe a slate clean", which is why we need some engineering analysis to give us a little objectivity amongst the first impression "wow"s.