All the standard disclaimers apply. These are my personal opinions only and do not have any official status. Feel free to e-mail any comments to <email@example.com>
The National Weather Service has recently chosen to implement the so-called Enhanced Fujita Scale for rating tornadoes. As it was originally constituted, the Fujita scale was a windspeed scale. I discussed this in some detail in the F-Scale Assessment Guide I prepared for the National Weather Service, available as a PDF from here, so I won't repeat that in this essay. The advantage to a scale based on windspeeds is that it doesn't have any dependence on the construction practices in one part of the world. It is completely transferable. However, as pointed out in a paper by Don Burgess and myself as long ago as 1988 (also available here), in practice, this windspeed scale is not useful in practice. We get windspeed measurements from tornadoes so infrequently as to make an "intensity" scale based on windspeed completely unworkable. Instead, we have to accept the reality that damage is the best indicator we have on a routine basis, even though the relationship between damage and windspeed is quite far from simple. I discussed this to some extent here, and also in Item #B.11 here. We are forced to live with this, despite the occasional probing of a few tornadoes by mobile Doppler radars. The relationship between the windspeeds sensed by these radars and the winds actually seen down near the surface, where the damage actually occurs, remains to be determined. Some recent studies (e.g., Wurman and Alexander 2005) have begun to explore this topic, but it will be some time before we have windspeed estimates from mobile Doppler radars for even a tiny fraction of the lifetimes of a tiny fraction of all tornadoes. We are going to be using the damage-windspeed relationship for some time to come.
During the late 1960s, Ted Fujita developed the F-scale, and it was implemented nationally with the strong support of Alan Pearson, then head of the National Severe Storms Forecast Center in Kansas City, MO (predecessor to the current Storm Prediction Center) - part of the National Weather Service (NWS). The F-scale became the basis for rating tornadoes in the early 1970s. As part of study to safeguard the nation's nuclear power generating stations, the Nuclear Regulatory Commission (NRC) sponsored an effort to develop F-scale ratings for historical tornadoes back to 1950 through 1976. This was done by paying students to review the newspaper accounts and come up with some sort of estimate of tornado intensity for every tornado in the record. The result of this effort was summarized in a paper by Kelly et al. (1978) - available here - when combined with the fact that F-scale ratings were to be determined thereafter for all tornado reports that made it into the official record, Storm Data (available from NCDC), this made the development of a climatology of tornado "intensities" (based on their F-scale ratings) possible.
As discussed by Brooks et al. (2003) and Doswell et al. (2005) - available here - the historical record of severe storm events is laced with inhomogeneities and secular trends that make it very difficult to interpret. The origins of these inhomogeneities have their roots in what may have been well-intended efforts by individuals but which have had the obvious result that the record is basically a mess when considered in detail. There have been some serious efforts, led by my colleague, Dr. Harold Brooks, to distill as much reliable information as possible from this mess. But the sad fact remains that the historical record is not as consistent as we would like it to be. One of the many factors associated with this is that tornado intensity ratings, as embodied in the F-scales attached to individual events, have many problems. Our understanding of the climatology is necessarily contingent on the accuracy of these ratings, which are notorious for their inaccuracy!
Therefore, our actual perception of the frequency of tornadoes as a function of their intensity (as measured by the F-scale) is plagued with considerable uncertainty, to say the least. There has never been a really serious commitment to obtaining reliable F-scale ratings by the NWS. In fact, it can be argued that even having the NWS involved in the determination of what actually happened during a given severe weather episode has some serious issues relating to the "fox guarding the chickenhouse" syndrome (see here and here for some discussion of this).
After some discussions, including the paper by Speheger et al. (2002) - available here - the NWS created the so-called Quick Response Team (or QRT), a group of people designated as "experts" regarding damage assessments for the purpose of rating tornadoes, after the La Plata, MD tornado was originally rated an F5 tornado by the local NWS survey. Subsequent analysis suggested that this was an overrating of this tornado and the QRT was established to be called in to assist any local NWS survey team in case there was a suspicion that the tornado might be given an F4-F5 rating. I should note that I was designated a QRT member, and have been involved in precisely one survey since being so designated.
For reasons that leave me completely mystified, this has had the clearly unintended effect that the initial, local NWS survey team has consistently avoided giving an event even the chance of a tornado being given an F4 or F5 rating. I don't see any plausible reason why calling in the QRT members to assist in the evaluation of an important event should be avoided, but the result has been that no tornado has been rated an F5 since the 03 May 1999 tornado in the Oklahoma City metro area.
Only a handful of tornadoes have been rated F4 in the past seven years (as of 2006)! And the QRT consistently has not been consulted in numerous tornado outbreaks in this period. What's the cause of this? I don't know and I can only speculate that the NWS is hoping to avoid the added expense of flying in a QRT team "expert" to assist the local NWS survey team. The end result is that the QRT is now moribund - a group of people willing and able to participate, but no one is inviting them to use their knowledge to serve the community, for unknown reasons.
Clarification: Just for the record, I'm not advocating that there's some sort of a conspiracy within the NWS or anywhere else to downgrade tornado ratings or avoid the use of the QRT. I generally have disdain for conspiracy theories - although conspiracies certainly do occur occasionally, so a conspiracy can't be ruled out categorically. Anyway, if an NWS survey team makes a preliminary assessment of F3 or less, and there's any cause to question their call - as in the case where photos of homes swept off their foundations by the storm are available publicly - then it seems to me that they're doing themselves and the system a disservice by not calling in a QRT to confirm their findings. As noted, the presumption seems to be that having to call in outside help is something of an insult to the office. This is simply not the case. The QRT was created to help with the rating process, not simply to second-guess the preliminary team. If the preliminary team missed something or made a mistake, that could be construed as a negative about their preliminary survey, but that seems like an unnecessarily defensive posture for them to take. A "high-end" F3 rating, especially when it's a killer tornado, by the preliminary survey is a perfect example of when a QRT should be called. The use of local personnel should not be considered a substitute for the QRT, whatever their qualifications might be.
It's perhaps also of pertinence that the NWS is not even going through the motions of their virtually useless "Service Assessments" now, even after important tornado outbreaks that result in multiple fatalities, despite the fact that their own criteria would seem to require them to do such an analysis of the service they've provided to the public. It's as if the NWS wishes to avoid any sort of external scrutiny of their performance during major tornado outbreaks, in the fear, apparently, that their service might prove to be less than perfect. Rather than learn what mistakes they might have made, it seems the NWS prefers to avoid learning anything from significant tornado outbreaks. Ignorance is bliss, apparently in the eyes of the NWS bureaucracy.
The failure to use the QRT in the task for which it was created is a sad commentary on the malaise the infects the NWS, top to bottom. The scarcity of F4-F5 tornadoes of late isn't likely to be the result of a real change in the annual frequency of such events, but rather represents just another inhomogeneity in an already ugly historical archive. Removing the effect of such perturbations in the record is pretty much not possible, unfortunately.
Recently, the structural engineers at Texas Tech. University began a discussion with the goal to be to "enhance" the Fujita scale. It has long be felt that the lack of calibration for the Fujita scale, notably at the high end, was leading to an overestimate of the windspeeds associated with F3-F5 damage. It is particularly difficult to simulate just what sort of windspeeds are associated with the "high-end" damage produced by tornadoes. Obviously, we have virtually no in situ wind speed sensors capable of withstanding the winds produced by a significant tornado. We are, therefore, forced to use the damage as some sort of indicator. In most cases, the damage produced by a tornado is to structures that are not even remotely engineered to resist high windspeeds. On rare occasions, engineered structures are found within the damage path of a tornado, and these can, to some extent, serve to "calibrate" the damage-windspeed relationship. For example, if a structure is designed to resist windspeeds of X miles per hour before failing, then when we observe that such a structure fails, the windspeeds must at least equal X at that time and location within the path of the tornado. Unfortunately, such unambiguous indicators are rare.
After the Jarrell, TX tornado of 27 May 1997, some engineers disputed its F5 rating, proposing that its relatively slow movement meant that the duration of the tornadic windspeeds contributed to the complete destruction of homes in a Jarrell subdivision. According to their analysis, much lower windspeeds than those associated with minimal F5 rating (261 mph) could have caused all the observed damage. I'm not a structural engineer, and I'm not prepared to dispute their findings. However, I've pointed out for many years that when destruction is total, the windspeed necessary to produce that damage is only a lower bound to the actual windspeed. We don't and can't know how high the windspeeds were whenever destruction is so complete that nothing is left of a well-constructed frame home. Not all the homes affected by the Jarrell tornado were poorly constructed. The extent to which duration of the high windspeeds plays a role is as yet completely unknown. No one has conducted any sort of experiment to determine the relationship between duration of the wind and the damage produced, especially at the upper end of the Fujita scale.
I've also been disturbed for many years that the very same Texas Tech. engineers pushing a revision to the windspeeds of tornadoes at the upper end of the F-scale have consistently denied that automobiles and other motor vehicles become airborne in some tornadoes. This denial flies in the face of indisputable video evidence and so is completely unjustified, in my opinion. I've wondered why they're so adamant in disputing the clear fact that motor vehicles can become airborne in F3+ tornadoes. The only plausible explanation I can dream up for such steadfast denial of the facts is that they've been promoting a "saferoom" that would evidently not be able to withstand the impact of an airborne motor vehicle. Such a "projectile" would render useless their standard saferoom. Of course, the odds of a saferoom being hit by such would be relatively low, but it's a non-negligible possibility that perhaps they'd prefer did not exist at all. I can offer no other explanation for their consistent and persistent denial of the reality of airborne motor vehicles.
Furthermore, it seems that many in the construction industry have believed that it's basically impossible to build an economical tornado-resistant structure. As I've discussed elsewhere, it's a myth that tornado-resistant construction involves uneconomical construction practice. My view of this as myth is based on input from structural engineers! Perhaps it's impossible to build an affordable home that would completely stand up without damage to an F4-F5 tornado, but even in real F4-F5 tornadoes, only a tiny fraction of the damage path actually experiences those extreme windspeeds, whatever they might be. For the most part, by far the majority of the homes in the 03 May 1999 tornado experienced only F3 or weaker windspeeds, resulting in some structural elements of the homes surviving the tornado. It's not wildly impractical to add some structural elements to existing construction methods in order to improve substantially the odds of having only superficial (not structural) failures for most homes in the path of even a violent (F4-F5) tornado. From my assessment of construction practices in the Oklahoma City metroplex, most of the homes in the area are not even compliant with the existing construction codes, no matter what the cost of the homes might be. Home construction in "Tornado Alley" is pretty shoddy! It seems clear that home builders don't want enhanced building codes, nor do they want improved enforcement of the existing codes.
O.K. - The engineers at Texas Tech. have been very pushy about the revisions to the Fujita scale. Let's now consider the meat of those revisions. These include two significant changes:
a. Damage "indicators"
An important part of this suggested revision of the Fujita Scale is the notion of damage indicators. The participants in the process of "enhancing" the Fujita scale were polled to provide what they subjectively felt were "indicators" of the windspeeds in tornadoes, to add new indicators beyond the "well-constructed" frame home that formed the basis for Fujita's F-scale. The synthesis of that input was an "enhanced" list of damage indicators that would allow the members of a local NWS survey team to make estimates of the windspeeds associated with an observed level of damage. Notably, the windspeeds associated with the high-end indicators, including "well-constructed" frame homes were revised substantially - downward. The details are contained in the documented indicators (see here for the PDF), but an important issue is that according to the current list of EF damage indicators, it's impossible for any damage to a well-constructed frame home to be associated with an EF-5 rating. Contrary to what I was told during the initial meetings concerning the development of the EF-scale, this would mean a notable change in the ratings. Under the old F-scale criteria, if a well-constructed frame home was swept away, leaving nothing but a foundation and no standing walls, then that would be considered F5 damage. Under the new EF-scale system, that would no longer be true - the highest possible rating by the new EF-Scale standard would be EF-4. Only if the structure of a standard frame home was somehow enhanced beyond standard practice throughout the load path, would it be possible for EF-5 damage to be inferred from frame home damage. This amounts to a change to the ratings, and would imply that any F5 rating in the past was, in effect, not valid. I see this as an important violation of I was led to believe was the agreed-upon constraint that any new rating system would not result in a change to the ratings of the past.
b. Windspeed or damage?
Although the original F-scale system was conceived as a windspeed scale, its practical implementation was necessarily that of a damage scale (Doswell and Burgess 1988). So long as the F-scale was ultimately a windspeed scale, however imperfectly damage was related to windspeed, it was a universal scale. That is, it was not related explicitly to specific construction practices. Thus, it could be exported to other parts of the world, where standard construction practice might differ from that of the United States. In the existing implementation of the EF-Scale, however, the ratings are now keyed directly and specifically to construction methods unique to the United States. Thus, the EF-scale is no longer a windspeed scale, but is completely and unambiguously a damage scale. This is a serious problem, in my view, to its adoption elsewhere around the world. On this basis alone, I'd be adamantly opposed to the EF-scale. It now imposes an implicit construction standard unique to the USA, which is a step backward, in my opinion. And no rating system of value should be keyed to specific construction practices in the USA.
The advent of mobile Doppler radars has increased the frequency at which we can observe tornado windspeeds. Unfortunately, a Doppler radar does not observe winds in the same way an anemometer on a 10 m tall mast observes the wind. Although it's not explicitly mentioned by Fujita, he seems to have implied that the windspeeds to which he was referring were at the standard anemometer height of 10 m. There's some debate over which time period is appropriate to consider for the measurements, implicitly connected to the spatial scale of the windfield. Currently, the consensus duration is 3 seconds. A Doppler radar, even when it's only a few km from a tornado, does not see the winds down to a height of 10 m - perhaps at best, they "see" winds down to 50 m or so, so the process of extrapolating the Doppler measurements to a height of 10 m is being considered - no consensus has yet emerged about how this should be done (Note that such an extrapolation of wind observations is being employed today in hurricane windspeed measurements). It's been widely noted that the mobile Dopplers measured a wind of 318 mph during the 03 May 1999 tornado that affected the OKC metroplex. This observation has been ballyhoo'd repeatedly as suggesting the possibility of an F6 tornado (since 318 mph is the nominal top of the F5 class), and has been used by some to suggest that this tornado was the strongest ever measured. Although it might be the strongest ever measured, it is by no means obvious that it was the strongest tornado in our nation's history, as discussed here.
The primary conflict over the F-scale all along has been the damage indicators chosen by Fujita to accompany his windspeed categories. When structural engineers at Texas Tech. University (TTU) became involved in the consideration, the old notions of tornado windspeeds in excess of 500 mph had not yet been completely discredited. Their involvement was very helpful in the process of putting some reasonable bounds on tornado windspeeds. When all things had been considered, it was generally felt that peak windspeeds of 300+ mph were possible in the most extreme tornadoes and, to the best of my knowledge, this remains the consensus view among tornado scientists. Now, however, the structural engineers seem determined to reduce the windspeeds in the F-scale categories because it is their view that windspeeds < 200 mph are sufficient to explain the most extreme damage to "well-built" frame homes - the primary damage indicator that Ted Fujita used for his F-scale assessments. I've already mentioned that when no interior walls of a home are left standing, whatever windspeed is necessary to accomplish that can only represent a lower bound to the real wind.
Also already noted - it's been generally agreed that the windspeed of relevance is a "3-second gust" at a nominal height of 10 meters. For the most part, although we have some scientific evidence that tornadic windspeeds of 300+ mph are possible very close to the surface, it's not yet evident that we can validate the actual occurrence of such windspeeds at a height of 10 m and we certainly have no anemometers that could withstand much more than 200 mph, even if we could somehow put such anemometers into a tornado and be assured they measured the peak gust anywhere within the whole damage path of the tornado. The debate over the relationship between windspeeds and damage is likely to go on indefinitely - until we have high-resolution 3-dimensional windspeed data down to at least 10 m heights throughout the life cycle of a large number of tornadoes. This won't happen in my lifetime.
If we accept the current notion of the windspeeds associated with damage to well-constructed frame homes in the new EF as effectively redefining tornadic windspeeds, only extraordinary damage could ever result in a windspeed of 200+ mph. We have scientific support for the notion that windspeeds in excess of 300 mph can occur in tornadoes, but we now have a windspeed scale where the top category has no upper bound beyond 200 mph - effectively, we now seem to saying that a 200 mph windspeed is about as high as a tornado will ever produce. except perhaps in "megatornadoes" that occur only once every 10 years or so.
For the record, I'm still unsatisfied with the use of "hard" boundaries for the windspeed categories, for reasons discussed here.
I frankly admit I have no way to say confidently that a 300+ mph windspeed (3-second gust at 10 m) ever occurs in real tornadoes. But I don't believe the evidence is yet compelling that we should be "lowering the bar" down from 300+ to 200+ just because most frame homes are not generally well-constructed, according to what amounts to a new norm. We've changed the definition of what's well-constructed - probably with good reason - but is that reason enough to discount the likelihood of estimated windspeeds at 10 m ever reaching much beyond 200 mph? For me, no. Not at this time. I'll remain open to the notion, but I want to see much more careful thought and hard evidence put into this debate before I'm willing to change the windspeeds associated with the F-scale categories.
If the EF Scale represented a great leap forward in terms of the scientific discussion of tornado windspeeds, I'd be all for it. And if the windspeeds in the F-scale categories need to be changed - on the basis of compelling evidence - then by all means we should do so. But sacrificing the historical record on the existing evidence just to add some new and revised damage indicators is not good enough for me. My reservations about implementing this have grown with time since I last participated in the EF Scale discussions, especially since learning that EF5 was no longer possible with a well-constructed frame home.
At least for me, this discussion naturally leads into the question of where any given event fits within the existing record of events in the past. The fact is that we have a very poor notion of what might have happened in the past, so the question of where a modern event fits in the record of the past is pretty much a matter of speculation. What about the infamous Tri-State tornado of 1925? What about the Palm Sunday tornadoes of 1965? What about the 1974 Superoutbreak tornadoes, like that in Xenia, OH? We can't go back and re-survey the contruction of those homes, so what do we do about those ratings? The existing records don't contain enough information about the source for the ratings to allow much of a re-assessment, even if the desire to do such a re-assessment were to be supported. Since we have virtually no reliable measurements for past tornado windspeeds, what possible significance can we assign to a mobile Doppler measurement for a tiny sample of tornadoes? Unless and until mobile Doppler radar measurements are available for most of the tornadoes that occur in a given year, and the data cover the entire life cycle of all those tornadoes (a condition that will not be valid for any time in the foreseeable future), it seems we'll continue to have pretty limited information about what meaning a given observation might offer regarding where it ranks compared to other tornadoes, past or present.
As usual, the media have made much more of the 03 May 1999 OKC tornado than can strictly be justified. Having said that, the 03 May 1999 OKC tornado is still the last tornado recorded in the US to be rated F5 (no longer true - see above). Under the new EF-Scale criteria and with the QRT pretty much dead in the water at the moment, it might be the last to be rated F5 for quite some time. Is this to be interpreted as an indicator of global climate change or is it simply another in the many examples of secular changes in the way records are kept? I'm inclined toward the latter interpretation. I don't like the new EF-scale and I'm inclined to believe that, at best, its implementation is at best premature.
Kelly, D. L., J. T. Schaefer, R. P. McNulty, C. A. Doswell III and R. F. Abbey Jr., 1978: An augmented tornado climatology. Mon. Wea. Rev., 106, 1172–1183.
Doswell, C. A. III, and Donald W. Burgess, 1988: On some issues of United States tornado climatology. Mon. Wea. Rev., 116, 495-501.
Speheger, D. A., C. A. Doswell III, and G. J. Stumpf, 2002: The tornadoes of 3 May 1999: Event verification in central Oklahoma and related issues. Wea. Forecasting, 17, 362–381.
Brooks, H. E, C. A. Doswell III, and M. P. Kay. 2003: Climatological estimates of local daily tornado probability for the United States. Wea. Forecasting, 18, 626–640.
Doswell, C. A. III, H. E. Brooks and M. P. Kay, 2005: Climatological estimates of daily local nontornadic severe thunderstorm probability for the United States. Wea. Forecasting, 20, 577–595.
Wurman, J., and C. R. Alexander, 2005: The 30 May 1998 Spencer, South Dakota, storm. Part II: Comparison of observed damage and radar-derived winds in the tornadoes. Mon. Wea. Rev., 133, 97–119.