Who owns the EF-Scale?

An essay about responsibility

by

Chuck Doswell

Posted: 02 November 2008 Updated: 24 October 2009: some minor changes, fixed some typos, added a reference link

This essay has been triggered by a continuing irritation with the NOAA agencies connected to severe weather. I believe they have abdicated some of their important responsibilities in service to the public.

Comments can be sent to me at cdoswell # earthlink.net (either use the email hyperlink or cut and paste into your emailer after replacing ' # ' with '@'. If you're not willing to see your comments posted here (or to give me a reasonable explanation for why not), then don't waste my time and yours.

History

I've already posted elsewhere about some of the problems I have with the EF-scale. Some colleagues and I likely soon will have a formal critique regarding the implementation of the EF-scale in the US (Doswell et al. 2009) accepted as a formal publication (when it's available as such, I'll provide an update to this essay). But this essay isn't about the EF-scale, per se. It's about the absence of a process by which the EF-scale could be modified as new science and engineering information becomes available.

To understand this situation, consider some historical perspectives. The F-scale clearly was the brainchild of the late Prof. T. Theodore Fujita. Its development and acceptance can be considered among his voluminous contributions to the science of tornadoes. Whatever flaws it has, it was obvious that any concern over those flaws had a target: Ted Fujita. He embraced the responsibility for his scheme with exuberance and vigor, defending it when challenged. He eventually came to acknowledge that construction quality was a major issue and so introduced a modification of it: the so-called f-scale, which attempted to accommodate the issue of construction quality with the necessity for interpreting damage in terms of windspeed. To whom do we turn with our concerns now?

Obviously, another major flaw was the lack of calibration at the higher F-scale ratings - structural engineers (notably at Texas Tech. University) became increasingly strident with their objections to the wind speed estimates at the high end of the scale. Their basic concern was that they felt that the windspeeds producing F3+ damage were much less than indicated by Fujita's estimates. As a result of the concerns of the engineers (notably Jim Macdonald and Kishor Mehta), an ad hoc group was invited to participate in a series of meetings to do something about the engineers' unhappiness regarding the wind speed estimates at the high end of the F-scale. I was invited to their first meeting (the so-called Fujita Scale Forum) in Grapevine, TX, but did not participate in the process after that. With the assistance of a number of volunteer "experts", a greatly increased number of damage indicators (the primary damage indicator for the original F-scale was the putative "well-constructed frame home") were established, with estimated windspeeds tied to the degree of damage to each indicator.

The EF-scale was "officially" adopted by the National Weather Service (NWS) in 2007. This fact is not documented in any refereed publication - it's simply a matter of policy. The Storm Prediction Center maintains a webpage about the EF-scale, which is about the only recognition of its official status I know about, although some piece of paper somewhere may actually exist that documents this recognition by the NWS.

EF-scale ownership

So this leaves us in the situation that concerns me. Who "owns" the EF-scale? Why do I ask such a question? Ted Fujita evidently owned the F-scale, so long as he lived. His passing actually was the first event leading us to the present situation. The concept of the EF-scale at its initial inception (at the Fujita Scale Forum) was that it would become a "living" thing, rather than some bureaucratic exercise, hardening into solid concrete after some initial formulation. That is, as the science and engineering behind the scale made progress, it was proposed that it would be possible to alter the EF-scale to account for new understanding. However, the ad hoc nature of the original exercise has resulted in the dissolution of the original team. Hence, despite the original intent, the EF-scale presently is not a living concept - once the EF-scale was adopted, the process of revision ended. The structural engineers got what they wanted: downward revisions to windspeeds tied to the high-end F-scale events. They apparently have chosen not to "own" the EF-scale, once it became recognized officially by the NWS. The NWS hasn't ever acknowledged "ownership" of the EF-scale, either. There's no NWS-based EF-scale working group, there's no EF-scale committee, or even an EF-scale "focal point" anywhere in the system.

Imagine some brand new scientific/engineering results become available that indicate strongly that the EF-scale needs some revisions. Who is responsible for investigating the validity of those new results and recommending that the EF-scale be changed? For that matter, to whom would such a recommendation be addressed? Whose approval would be sought to confirm the implementatio of the revisions? By what process would revisions become "official"?

It seems that the EF-scale is now just sort of floating in some sort of black void, with no one apparently responsible for anything. I suppose another ad hoc "forum" could be called and the previous process repeated more or less. But this strikes me as a pathetic substitute for someone accepting ownership of the EF-scale and maintaining a continuing process by which the scale can be reviewed and revised.

Given that the NWS as a whole has not accepted this responsibility, I suppose I'm free to offer my suggestions. My proposed most logical choice would be for the National Severe Storms Laboratory and the Storm Prediction Center to establish in their respective organizations standing groups who would together form and maintain an EF-scale "Forum" in which individuals and organizations could participate in free and uninhibited discussions. Within this forum, there would be an EF-scale "Committee" - with membership appointed on the basis of established credentials in the field - which would approve any changes to the "official" EF-scale document. The document would include:

1. List of approved damage indicators
2. List of approved degree-of-damage measures for each indicator, with an associated statement about windspeeds assigned to that degree of damage
3. List of approved source metadata for each rating

The latter would need to be established if it eventually can be accepted that each rating has to include information identifying the source(s) for the input to the rating process. If needed, events would include multiple sources, of course. The idea is to have this list encompass most of the typical means by which input to the rating could be derived: chasers (of varying credibility!), storm surveys, newspaper accounts, photographs, etc.

This forum would also be responsible for interacting with potential contributors from outside of the USA, where the F-scale has only recently been adopted. See Doswell et al. (2009) for further discussions.

The EF-scale at present

As "recommended" by the official EF-scale document, the EF-scale looks like this:

I see a number of problems with this version of the scale:

• The "threshold" wind speed for a tornado is too high - a wind speed of 58 mph (50 knots) is the current threshold for a severe non-tornadic wind event. A logical choice the minimum windspeed for a tornado would be the same as that considered damaging enough to be "severe" by NWS standards.
• This table suffers from the same illusion of precison as the F-scale did - a wind value 1 mph faster (about 0.4 m s^-1) can result in a category change. See here for more discussion of this.
• Assuming that windspeeds substantially greater than 200 mph (i.e., 300 mph or more) can occur within the lowest 10 m, this scale cannot accommodate them in a meaningful way. Their occurrence or non-occurrence would be important to document, but this scale would lump them in with everything > 200 mph.

From a purely visual standpoint, eyewitnesses (including storm chasers) view the extension of the condensation funnel to the ground (or at least to connect with a debris cloud to be visible evidence of tornado "touchdown". For cases without a visible condensation funnel to the ground, the presence of a debris cloud beneath a rapidly-rotating cloud base (with or without any funnel cloud) is also considered evidence of a tornado. How much wind is needed to raise a debris cloud? In some instances, dust can be raised by winds of 15 mph or so. Would a debris cloud raised by winds of 15 mph be sufficient to call the event a tornado? As discussed elsewhere, the definition of a tornado raises more questions than might seem evident at first glance. If we accept the EF-scale definition literally and without question, then such a "tornado" wouldn't even meet the EF0 threshold. Is there a need for tornado categories below EF0? Or do we exclude from consideration those vortices (some of which might have condensation funnels reaching the ground!) producing debris whirls produced by winds < 65 mph? How would we know if it was or was not a tornado if we couldn't measure the windspeed and had to try to rely on "damage indicators" to substantiate the existence of a tornado? Should we include dust as a damage indicator? What if the dust has been wetted by prior precipitation? Do we need to have a separate indicator for dry and wet soil? What about soil types?

If it's possible (and I believe it is, at least as of this writing) that windspeeds > 240 mph can occur near the ground (i.e., at a height of 10 m or less), the damage they apparently would do (if the engineers are right) to a well-constructed frame home would be indistinguishable from that done by a tornado with winds barely > 200 mph. Assuming that the engineers are correct (and I believe that issue is not yet entirely a closed book), then indicators capable of resulting in higher windspeed estimates need to be developed and approved. Thus, there might need to be additional EF-scale categories beyond EF-5 if this scale is to useful in recognizing the existence of the extreme winds in real tornadoes. Other sources of information, such as photogrammetric analysis or Doppler radars / lidars need to be incorporated so that this windspeed scale can encompass what is plausible in a meaningful way.

Suppose there are new categories added that reach up to 300 mph and beyond. How does this affect ratings from the past? Since there are no past ratings higher than F5, it raises some questions about all the ratings from the current era back as far as anyone has attempted to establish ratings retrospectively. As an example, did the Xenia tornado of 3 April 1974, contain winds much faster than 200 mph? How might we ever be able to know the answer to this? If we were able to travel backward in time and observe that tornado with a mobile Doppler radar, or could do high-resolution 3-d photogrammetry, we might be able to say we measured winds on the order of 300 mph. But wait! There was some photogrammetry done for that tornado - see Fujita (1975; his Fig. 22). His results indicated windspeeds in excess of 300 mph - but only above a height of 200 m, and that speed includes a substantial vertical component. Purely horizontal, tangential wind speeds were measured (by tracking dust in this case) at less than 200 mph - around 160 mph - at a height of 50 m (the lowest height where measurements were obtained, which is about the height where current mobile Doppler measurements end). There certainly was no evidence of 300+ mph at a height of 10 m or less. The issue of whether or not such windspeeds are even possible has yet to be resolved. But do we want to exclude that possibility in our tornado intensity scale?

A suggestion

In the absence of any formal EF-scale revision process, I'm free to propose my own "D-scale" for tornadoes. Since it's mine, I'll emulate Ted and call it the Doswell or D-Scale:

I make no claim that this has been carefully thought out. Like the original F-scale, it is a windspeed scale. There likely are many ways to improve on it, including tossing it onto the rubbish heap and replacing it with a better one! The foremost characteristic of this scale is that the windspeeds in the categories overlap! My rationale for this can be found here. If a damage indicator and degree of damage result in an estimate of the windspeed, then the category midpoint closest to that estimated windspeed would be chosen - it it's exactly midway between two midpoints, then assign it to the lower of the two categories (in mph). Suppose the estimated windspeed is 150 mph - this puts it closest to the midpoint of the D3 category. If the estimate is 145 mph (exactly halfway between the midpoints of D2 and D3), then it's assigned to the D2 category. The overlapping categories offer explicit recognition to the variability in the damage-windspeed relationship.

Like the EF-scale, these are speeds at 10 m above the ground. The values in m s^-1 have been "nudged" in the conversion process to values that are multiples of 5 - hence, they don't quite correspond to the values in mi hr^-1.

The second notable characteristic is that there are now 8 categories rather than 6. This does not imply higher resolution, but is forced by the logic I'm trying to follow. Let me justify my choices (of the moment):

• Any tornado that produces winds that can be shown to be less than those designated as "severe" by the NWS severe weather criterion becomes the lowest possible category, D0. By my definition, this is the "weak" tornado I believe such a category should be designed to include. Most brief touchdowns likely will fall into this category. The degree of damage would be marginal.
• The color scheme for the text indicates the simplest possible three-category scheme I prefer: green = weak, orange = strong, red = violent. Having more than three categories carries with it considerable uncertainty, an inherent aspect of any scheme.
• The highest category, D7, has an upper limit that probably will accept the highest winds observable near the surface in real tornadoes. Windspeed measurements above 10 m (by any means) would need to be extrapolated to values at 10 m by some method, and the method used should be included in source information as part of the record.
• Having more categories as we move up the intensity scale can accommodate the fact that most damage indicators represent only lower bounds for the windspeeds associated with "total" destruction of the indicator. Without some special windspeed or damage evidence, the violent categories would not be reached in most instances. It would simply be understood that if special evidence is not available, an estimate in the D4-D5 range would produce damage typically associated with tornadoes from the past that have been rated as violent (F4-F5). Standing room remnants within an otherwise completely destroyed "well-built" home would be D4 damage. D5 would correspond to most F5 tornadoes in the record. The historical record could not (in general) support a D6+ rating in the absence of information beyond that of "well-built" homes completely swept away.

I repeat - this is just a strawman, not a thoroughly thought-out scheme. Note that I strongly recommend the adoption of "unknown" as a category (DU) when no information is available to allow the estimation of windspeeds (brief touchdowns in open country, tornadoes traversing no object considered useful as a damage indicator, and so forth). Update (24 Oct 2009): it now seems that an EF-Unknown category indeed has become possible (finally!) in the US tornado database.

References

Doswell, C.A. III, H.E. Brooks, and N. Dotzek, 2009: Doswell, C.A. III, H.E. Brooks, and N. Dotzek, 2009: On the implementation of the enhanced Fujita scale in the USA. Atmos. Res., 93, 554-563.

Fujita, T.T., 1975: New evidence from April 3-4, 1974 tornadoes. Preprints, 9th Conf. Severe Local Storms (Norman, Oklahoma), Amer. Meteor. Soc., 248-255.