Summary Report
FUJITA-SCALE FORUM
Prepared by
James R. McDonald, Ph.D., P.E.
Kishor C. Mehta, Ph.D., P.E.

Submitted
December 10, 2002
WIND SCIENCE AND ENGINEERING
Texas Tech University
Lubbock, Texas 79409

Back to Top

• Need to preserve and continue the historical database.

• Need to establish a standard frame of reference for wind speed (3 sec gust at 10m in open terrain).

• Need to provide additional information in the database, including average and maximum damage path width.

• Need for more resources to conduct damage surveys.

• Need for additional damage descriptions and photographs.

• Need for more training of NWS personnel who assign the F-Scale ratings.

• Development of an expert system for consistent assignment of F-Scale ratings.

• Need for more funding for damage surveys.

• Need for involvement of engineers in the damage surveys of intense tornadoes.

• Redefinition of F-Scale wind speed ranges.

• Need for a sliding scale to account for structural variability.

• F-Scale assignment based on maximum or representative damage.

• Interpretation of Doppler measurements of wind speed.

Back to Top

Back to Top

References

Back to Top

• Fujita Scale does not recognize quality of construction, weak links in the load paths, or variation of design criteria with time or geographical locale.

• The damage descriptions do not have specific meanings that can be interpreted by nonengineers, who most often make the storm classifications.

• Calibrate damage with the wind speed categories of the F-Scale.

• Account for different types of construction, i.e. distinguish between the extent of damage to a manufactured home and an engineered building, or between a timber and reinforced concrete building.

• Consider the regional variation of wind design criteria, e.g. structures designed for 90 mph wind will have more damage than one designed for 120 mph wind.

• Factor the tornado translations speed into the classification. A slow moving storm (Jerrell, Texas) will cause more damage than the one that is moving very rapidly.

Back to Top

Steering Committee and Forum Participants

My last comments pertain to tornado surveys. The NWS is responsible for rating tornado intensity Although individual offices and individual personnel give great effort in the surveys, not enough resources are currently available to adequately survey all damage that needs surveyed to determine if a tornado has occurred, and, if a tornado occurred, how intense was it. Response is ad hoc, many times not rapid, and often done without surveying all the damage from the air and from the ground. Of course, not all weak tornadoes can be surveyed in a complete manner, but the indicated problems exist with many strong and violent tornadoes, and most tornado outbreaks. When Ted Fujita was involved with surveys, at least for major events/outbreaks, many of these problems did not exist. Since Ted’s passing, problems have increased. Organized rapid response teams are needed for major events/ outbreaks. More resources are needed at local NWS offices to deal with the more-isolated strong and violent tornadoes. In addition, training is needed for all those participating in the survey activities. Training will also be important for the changes being discussed/recommended by the forum group. It may be possible for the NWS to organize itself to handle all of the surveying, but I think a better solution would be to include others (research meteorologists, wind engineers, university professors, and other experts) in the survey teams. Proposals to do just that have previously been introduced. I think they should be reconsidered. Having recently worked in the NWS, in the training area, I have several specific suggestions for the training activities, but I will hold those suggestions until closer to the time to work with the NWS to develop the training.

Back to Top

1. We should modify the existing database to save information about individual tornado events that is already being gathered (and paid for) during the existing survey process. If WCMs gatherevidence, it should be reflected in the routine database being created at the SPC by modificationsto the format to include more information (as discussed during the Forum), and preserved at the TTU "clearinghouse" for such data. I repeat my long-term concern for the development of digital databases (images and text), both to help with storage and to encourage the use of the accumulating data.

2. If we are going to relate damage to windspeeds by some "enhancement" to the F-scale (I support Roger's "EF Scale" nomenclature) along the lines that were discussed at the Forum, this will be a big step forward. I want the EF-Scale to reflect our uncertainties in an accurate way. The sliding scale and the explicit representation of overlapping distributions of windspeeds associated with a given level of damage is the correct way to go, in my opinion.

3. We must collaborate to develop new materials (Webpages, Tech. Memos, formal publications, …whatever) that can guide non-meteorologists to understand the uses and potential for abuse contained in all the existing databases … Storm Data, the SPC data, the DAPPLE data, Tom Grazulis' data, etc. I have seen and heard some egregious examples and for each one about which I know, there probably are hundreds of abuses about which I know nothing. Our users often make many unwarranted assumptions about the existing data in their analysis.

4. We must develop some consistency in our measurements, such that we can begin to separate the two components of variability: real variability among the events versus what amounts to measurement errors of various origins. I believe we can accomplish this, but it will require more commitment to training the WCMs (who, after all, do most of the work assembling the reports). If we ask the WCMs to do more than what they are doing now, we need to accept the fact that someone must provide them with the resources to do more. It is futile to propose all sorts of enhancements if we are not willing to work toward the actual accomplishment of those enhancements.

5. I have no problem with changes to our methodology and database, so long as we do not lose information we now have. We can always degrade enhanced information to be compatible with older, less comprehensive data, but information loss is irretrievable. There can be no substitute for lost data.

1. I like the suggestions we made regarding the standardized meaning of "windspeed" that was proposed at the Forum (i.e., a 3-s wind at 10 m). However, I have acute heartburn about how we might "reduce" our measurements (via radar, mobile mesonets, or whatever) to that standard. Any assumptions must be made very clear, and no information lost. That is, we should always strive to preserve the "raw" data instead of providing only the "reduced" values, and the details of any reduction process must always be made available. Frankly, I'm very skeptical of any reduction process!
    

2. We need to get behind the effort to reestablish a "fast response" team (to replace what Ted Fujita did for us for so many years) that combines meteorologists and engineers on a routine basis and has access to both aerial and ground survey information. We should be able to get a large list of volunteers to participate, provided we can get funding to support the effort. Our endorsement could be an important part of getting that funding!
    

3. Moreover, any proposal that contributes to the acquisition of enhanced information (beyond that of the typical WCM survey) for more than a handful of tornadoes every year deserves a long, careful look. In most cases, we continue to deal with "small sample" statistics, especially for strong and violent tornadoes. Anything we can do to get better information about what is happening in as many significant tornadoes as possible should be encouraged.
    

4.  Our understanding of the relationship between damage and windspeed is always going to be in a state of flux. We need to consider having some sort of routine process for encouraging meteorologists and engineers to interact on this and related topics. It shouldn't take a special meeting, based on the vision (and funding) of the conveners, to have the opportunity to exchange information. Every time I'm around engineers, I learn new things. I hope they feel the same way about interacting with us meteorologists. From what I have seen, this always has been fruitful when it's aggressively pursued. Why not institutionalize the interaction in some way? Texas Tech. seems like a logical place for such a development.
    

5. The original F-scale was defined as a windspeed scale. That is, its categories were defined in a very precise way using windspeeds directly. The merits and demerits of the F-scale as a windspeed scale never really came up in our discussions. Rather, the connection between the Fscale and damage was the source of the numerous conflicts that we've been having. The F-scale's application has been as a damage scale - the opposite of the way it was defined - and the inference of windspeeds from damage justifiably has been the source of much angst in the engineering community. I have very mixed feelings about the details of what I heard during the Forum regarding proposals for the EF-Scale. In spite of my lack of confidence in an egalitarian approach to a solution, perhaps it is only through the involvement of many more people that a workable solution to the situation can be found. Thus, I am not wildly confident that our Forum discussion provided much more than a starting point for a "solution" to the issues we confront. Perhaps any "solution" we arrive at can only be viewed as another step along a long road. I think that is the way it should be presented.

Back to Top

A further complication is that structural engineers point out that there are a number of factors that can lead to variations in the response of different structures to equal winds. Homes inadequately secured to the foundation can be lifted or slid away in some instances when winds were not sufficient to remove the roof. In other instances the roof may be poorly attached to the top plate and/or the top plate to the walls, leading to removal of the roof and progressive failure of the building. Large garage doors facing the strongest tornado winds can fail, taking part of the home away with the garage. The duration and gustiness of the winds, the amount of flying debris, as well as other factors, can contribute to the variation in damage between structures.

The workshop attendees also expressed a desire for a survey team of engineers and meteorologists to every tornado damage path preliminarily categorized as having F4 or F5 damage. The sentiment was that waterspouts prompting warnings to land or offshore waters should be classified as tornadoes.

Back to Top

JOE GOLDEN

1. I, too, endorse the sliding-scale, "Enhanced F-scale development, along the lines suggested by Jeff Kimball, with some overlap at the higher EF numbers. We need to reflect current uncertainties and be able to adapt the EF-Scale as new tornado windspeed ESTIMATES are made, preferably by two or more sensor types (portable Doppler and simultaneous digital sterophotogrammetry would be a big step forward here). All data should be normalized to 10m heights and 3-sec gusts. We should all be very careful what we release to the media early on, but I know the intense pressures. I am willing to concede that windspeeds have been estimated by portable Dopplers and photogrammetry at 225+ mph in a few tornadoes at tens or hundreds of meters above the ground (at probably different time-averages), but statements to the news media need to reflect also the much lower failure mode windspeeds found by Tim Marshall and the NIST wind engineers in tornado surface structure damage. (We never like to admit that even after 30 years, there is still APPARENT wide disagreement on probably max windspeeds in the level occupied by structures on the ground!)
    

2. Goal for SPC and NCDC (need to get NOAA to commit funds and staff-time!) should be a digital database with metadata indicating local staff and time spent doing surveys and the other items we listed (if done at all, by sfc and aerial mapping) and digital damage photos (many WFO’s now have digital cameras, originally bought with Mark Powell’s (HRD) pushing NWSH to document ASOS site exposures).
    

3. Education and training for ALL WCM’s (and other collaborators) doing ground and aerial damage surveys will be a must, and the work already done in Tech Memo form by Brian Smith and Bill Bunting along with inputs from Roger and all of us is important, along with many more good, well-documented damage photos that can be used to illustrate the EF-Scale (including those with engineering calculations).
    

4. All this effort so far will be for naught if we cannot convince NWSH (John Ogren and his boss, Greg Mandt) and the Regions that it is needed and can be implemented WITHOUT significant addition of new staff. The damage surveys must be organized with cross-discipline teams expeditiously (24 hr or less), and aerial surveys should be carried out for all outbreak and other questionable, significant tornadoes (criteria need to be developed here). I endorse Erik’s shameless plug and the development of a PC and PDA-based "expert system," but we need to convince NWS and OAR that tornado research needs to be higher in the funding priorities (NWS does, after all, pledge to cut its tornado warning FAR by half over the new few years in its Strategic Plan!) I still feel that because of political tinkering by NWSH officials in past "service assessments" that this process should be funded and organized by NOAA (TTU and NIST could approach new NOAA administrator, whoever he/she turns out to be).
    

5. Next steps must be to reach out to the broader meteorological community (good start to approach Severe Storms Committee, as well as AMS Broadcasters and Education Committee), but also to the Wind Engineering community (AAWE and ASCE) which is plugged into the current Congressional Wind Caucus activities that AMS and NOAA do little to encourage. Two upcoming opportunities are the America’s Conference on Wind Engineering at Clemson in early June and the AMS Broadcaster’s Conference also in June, but these may be too early for Jim’s target date for a final report?

Back to Top

TOM GRAZULIS

1. Overlap the wind speed numbers. That would be the single best way to reflect the true complexity of the problem. Even if all homes were engineered, the terrain and the angle of attack would change the degree of damage for a particular wind speed. "Confusion among the public" should not be a concern. Instead of blind acceptance of a number, questions would be raised. This might provide an opportunity for explanation and learning.
    

2. Keep the top F5 wind speed at 318 mph. I believe that there are tornadoes that produce winds of that magnitude. The forum lacked an authoritative voice on vortex structure and modeling to speak to that issue. Until we know what the winds are, lowering the number moves us from inaccurate guess to another. We should keep the top value at Josh Wurman's 318 mph "measurement" in 1999, which confirmed Fujita original guess. Lowering the top F5 wind to 250-280 sets up an unfortunate scenario for the future. There will be other high values based on Doppler radar measurement. When there are others above the F5 limit, the press will have free reign to label the tornado as "F6". The top of the F6 range will hit the headlines. The press won't stop at 318. Years of work needed to get the public perception down to a 318 mph top will have gone for naught.
    

3. The written standards should contain a long list of damage items that are not F5, not F4... and so on. That list should be longer than the list of what damage actually does qualify a tornado to be rated at a particular F-Scale level. The list should be updated annually, by reviewing the Storm Data descriptions and finding the mistakes that have been make. For instance, the carrying of the bell tower for a quarter mile (from a small South Carolina rural church) is not valid F4 damage (yes, that's there). The initial list can be constructed by reviewing the many inappropriate ratings during the past 10 years. An annual published list of the 10 worst ratings of the year would be revealing and instructive.

Back to Top

QUAZI A. HOSSAIN

Hazard Mitigation Center, Lawrence Livermore National Laboratory

In order to prepare the context for my viewpoint on what should be done towards improvement of Fujita Scale, let me list the major shortcomings of Fujita Scale:

• Does not recognize the actual strength difference between two structures having the same or similar physical descriptions, but built to two different design and construction criteria. This error can result in either underestimation or overestimation of wind speeds.

• Application yields only lower-bound intensity classifications, because damage descriptor includes only failed structures. This error results in underestimation of
  wind speeds.

• Itensity can be underestimated because of lack of structures or objects in the tornado path.

• Does not recognize the effects of various parameters that may contribute to failure other than "open terrain" wind pressure: topography, the presence of other structures, rate and duration of loading, distribution of pressure on the structure, etc.

 Several different forms of modifications have been used on an ad-hoc basis to account for these shortcomings of the original Fujita Scale, some by Fujita himself. But none of these are based on a systematic approach of actually calculating the strength of the failed structures in terms of wind pressure. Such an approach is not considered practical when thousands of past tornado records are involved. Instead, the uncertainties and errors in establishing a design wind speed based on historically  observed tornado damage data can be suitably accounted for by developing an Expert System that would yield a wind speed velocity distribution when the user inputs into it the quantitative and qualitative damage data and opinions. In such an Expert System, the characteristics of the wind speed velocity distribution of a given tornado would be compared to a set of standard distribution characteristics (predetermined based on the consensus of the experts) for nominally categorizing the tornado into five wind intensity categories. For simplifying the development of tornado wind hazard curve, the standard wind speed velocity distribution of the categorized tornado will then be used instead of tornado-specific distribution.

Back to Top

JEFF KIMBALL

JEFF KIMBALL

Back to Top

I agree wholehearted to preserve the database as much as possible and provide an increase in training not only for Warning Coordination Meteorologists, but storm spotters and law enforcement in order to provide some standard of verification for the light damage provided by weaker tornadoes. Finally, I think we should provide an increase in engineering expertise in storm surveys to provide a more consistent evaluation of tornado intensity. It is quite amazing that one violent tornado is estimated to 318 mph by Doppler Radar corresponding to the top end of the Fujita estimation. We must be careful using the proposed scale in our second day in that we do not confuse the message with the media when it comes to explain the adjusted scale. I recommend that those who do surveys start with Fujita's modified scale from his Mystery of Severe Storms book.

Thank you for letting me be a part of the forum.

Back to Top

My primary concern, for a number of years now, has been that the current method of reporting tornado damage and occurrence is inadequate for climatological research. My research needs are broad and general (i.e., did a tornado occur, and, if so, was it significant?), and if I find the data inadequate, it is certain that they are not adequate for tornado risk assessment.

At the forum, I presented what I consider to be a classic example for much of the rural United States (the tornado is described in some detail at: http://mrd3.nssl.ucar.edu/~eras/www/SSR/Exp/Expframeset.htm under the "STEPS" experiment). This was a tornado that under normal circumstances would be described as "small and weak". It did not damage any structures because there were none present. Thus, typically, it would receive a "default" damage rating of F0 and default path lengths and widths. A careful damage survey of this tornado revealed that corn was scoured down to bare earth, a damage indicator considered to be F4-F5 by Dr. Fujita. Photogrammetric analysis of a set of very high quality, fixed focal-length photographs revealed that the tornado debris cloud was typically about 350 m in diameter, consistent with the damage envelope, which is a medium-large size tornado. For example, it is about the same size as the Edmond, OK tornado of 3 May 1999. It is my hypothesis that tornado hazard is very poorly estimated in the rural United States, especially the High Plains, because of the mis-assessment of many tornadoes similar to this one.

Although we do not have means available at present to assess the potential damage intensity of rural tornadoes, our forum considered one very significant recommendation for improving the tornado database. It was suggested that a number of pieces of ancillary information be included with the tornado reports. For example, the means through which a damage rating is assigned should be codified.

In my opinion, "guess" or "default" are perfectly acceptable techniques. If we knew that, historically, most rural tornadoes were assessed F0 or F1 ratings through guesses or defaults, we might have an entirely different level of confidence (much smaller) in our estimates of tornado risk in some parts of the United States. And researchers who need to know where and when "weak" tornadoes occur could have some confidence that a tornado truly was weak if a damage survey had been conducted, the tornado did affect structures, and all of the damage was minor. Thus, I reiterate my viewpoint that these "metadata" about the means through which various items in the tornado report were ascertained are absolutely vital for inclusion in our future tornado database. The metadata that were recommended at the forum (green items next to red list) are the essential, minimum set.

Personally, I am happy to discard the idea of including a surveyor’s name and hours invested as part of the metadata if these are a stumbling block to acceptance of the changes to the reports. I think we can achieve a reasonable reduction of uncertainty just by having the means of assessment codified in the report. I will be happy to volunteer to help develop the metadata standards for tornado reporting.

Calibration issues:

I favor the modification of the wind calibration of the F- cale. I like the idea of a sliding technique, with overlap in the wind ranges. I do suggest that we be very explicit about the 10 m, 3 s standard, and that a nominal "well-built frame house" is our primary standard for F-scale assessment. What this means in terms of construction, and why the F-scale slides for non-nominal structures, should be well explained in any publications we produce regarding the new scale. Also, I urge the wind engineering community to establish these windspeeds based on best estimates of minimum failure windspeeds similar to the work Tim Marshall has submitted to the meteorological literature. Put another way, I urge against including gut feeling and guesses of meteorologists in the new scale; that is the technique that was used a few decades ago, and now we find ourselves in a position to make fairly drastic revisions. Finally, I would recommend that we let all users know that the windspeed scale will likely require further revision as our understanding of tornado flow improves through in-situ and close-range remote sensing over the next decade or so.

I also favor leaving the F5 wind range very open at the top unless we wish to retain F6 as a category. This latter option is the most scientifically acceptable route, and it does not in any way imply that we believe that F6 will be observed. But in a statistical sense it is a possibility. So we are faced with the choice of having a probability curve at F5 that is radically different in shape from the other categories (much more skewed toward large windspeeds), or including F6.

I believe that F0 windspeeds should go down to zero; more specifically, they should be identified as being < X. This opinion is based on the practical consideration that we do not know at what windspeed dust starts to swirl below a funnel cloud and the vortex is de facto identified as a tornado. Put another way… do we wish to have NWS personnel querying spotters in realtime asking them if they are confident that the dust is moving faster than 45 m.p.h.? I do not think so. And I believe it is useful to know that the dust did indeed swirl.

Name of revised system:

 I am supportive of the name "Enhanced Fujita Scale" for a revised scale. I believe it will be enhanced through more thorough documentation, training, means for objective application, and better knowledge utilized in the wind estimates.

Shameless plug:

 I want to be very clear that we are now, or shortly will be, in a position to apply many new technologies to gain tornado knowledge. These include mobile Doppler radar at several wavelengths, Doppler LIDAR, in-situ surface state sensors, mobile mesonets, RPV penetration aircraft, and multicamera digital image photogrammetry. I am already undertaking work to develop several of these technologies for application to the tornado problem. However, I am seeking partners. In particular, I want the wind engineering and hazard mitigation communities to be aware of the potential of this new knowledge, the feasibility of obtaining it, and then need for multi- dsciplinary collaboration in data collection, analysis, and funding of these activities. Further, I need everyone to be aware that funds are very scarce (do they exist?) in the meteorological community for this work. We have the ambition and ability, but the only way this work will get done is through collaboration.

Back to Top

TIM REINHOLD

While it is true that well-built homes or structures constructed using conventional construction methods would likely be destroyed by a tornado with winds above 200 miles per hour, tornadoes with these magnitude winds are quite rare. Furthermore, the fraction of the area struck by these storms that actually experiences the extreme winds is also quite small. We have a great deal of information that can be used to educate people that stronger well built buildings can provide significant protection from the most severe hurricanes and from strong tornadoes. However, the high wind speed estimates that are assigned to various types of damage, without recognition of construction weaknesses that pervade much of our conventional construction, frequently blunt this message.

We need to provide accurate assessments of wind speeds that recognize the incredible variabilities in tornado events and the fact that tornadoes can and do produce extremely high wind speeds. At the same time, we need to provide estimates that convey the reality that conventional construction can be absolutely destroyed by relatively low wind speed events. We can certainly build buildings that will have a much better chance of surviving most tornadoes, if they are well built and contain clear load paths and good connections.

Back to Top

THOMAS W. SCHMIDLIN

I was pleased to be invited to the Fujita Scale Forum and applaud Jim McDonald, Kishor Mehta, and others for bringing this group together for the discussion. A consensus developed to:

• keep the name ‘Fujita’ in the revised version,

•  maintain the F0-F5 categories in roughly the same damage classes as Fujita  proposed and as exists in the historical data base,

•  enhance the data and improve consistency in the collection of data in the tornado data base (as time and money are available to WCMs),

•  improve the damage descriptions associated with each F-scale, and

• revise the windspeeds associated with each F-scale.

Regarding the last item, I liked the scheme proposed by Kishor and Tim Reinhold of mid-range wind speeds for each F- scale (with speeds adjusted downward from current winds at F3-F5) and a standard deviation to indicate our uncertainly in the process. If we use 2 times the standard deviation, it effectively puts bounds on the range. The fact that the F-scale is used for hurricanes and downburst wind damage should be considered.

My remaining comments will address the fourth item above, improving the damage descriptions. Motor vehicles are everywhere that people are located in the United States and can be reasonable indicators of wind speed. Private passenger vehicles (called ‘cars’ from here on) have changed a lot in design and safety features since the late 1960s and early 1970s. In addition, we know much more about cars and wind than we did 30 years ago. A selection of our own field and lab work on this topic is cited below.

Our examination of about 300 cars parked outside houses struck by tornadoes showed that at houses with F1 or F2 damage (n=165), only 30% of cars were even moved by the wind and just 4% of cars were tipped over. At houses with F3 damage (n=105), 50% of the cars were moved and 18% were tipped over. Of course, any of the small number of cars that are tipped over make impressive images and are remembered as ‘typical’ of the scene, yet careful inquiry of owners of the damaged homes reveals a different image, as represented in the numbers above.

In addition, our tests of a sedan and minivan in the idealized, but informative, world of the wind tunnel showed that wind speeds of 115 mph to 150 mph at car height (~1 m) were required to lift a wheel and perhaps initiate lifting or a roll-over of the vehicle. These winds correspond roughly to an F3 wind at 10 m height and are consistent with the field work described in the preceding paragraph.

The current Fujita Scale refers to cars in the F1 and F2 categories as "moving autos pushed off the road" and "cars blown off highway," respectively. These should be discarded as vague and difficult to evaluate. Cars pushed or blown off the road are often gone before a survey is done, it is impossible to know exactly why the car ended up sitting on the shoulder or in the median (wind or driver?), and there are many driver-related factors affecting these descriptors. In addition to our data, there is plenty of anecdotal evidence that winds of 100 mph do not cause general upset or destruction of cars.

The current F3 description states "cars lifted off the ground." While some cars are lifted off the ground in F3 damage (although it is difficult to determine whether a car was ‘lifted’ if it was not tipped or moved some distance), our field work and wind tunnel data indicate that most cars are not tipped over in F3 residential damage and (current) F3 winds at 10 m are unlikely to flip over most cars. These are my suggestions for F-scale damage descriptions and 10 m winds related to vehicles. For F0 with wind speeds of about 75 mph, I suggest there be no reference to vehicles in damage descriptions. For F1 with wind speeds of about 95 mph, "Semi-trucks and high-profile vehicles may be tipped over; cars, vans and pick-ups are not tipped over." For F2 with wind speeds of about 125 mph, "Cars, vans, and pick-ups may be moved but fewer than 10% are tipped over." For F3 with wind speeds of about 155 mph, "Cars, vans, and pick-ups are moved and 10% to 50% are tipped over. Vehicles that are tipped over may be rolled tens of meters." For F4 with wind speeds of about 200 mph, "More than 50% of cars, vans, and pick-ups are tipped over. Vehicles that are tipped over may be rolled or lifted and thrown hundreds of meters." The damage surveyor should avoid assessing only those vehicles remaining on-site, which is biased toward the more heavily damaged vehicles, and ask about all vehicles that were outside when the tornado struck. This will include those vehicles that may have had little damage and are in a glass shop or are still being driven and are parked elsewhere.

The description of damage to manufactured homes is difficult due to dependence on wind angle and strength of the tie- own system. I suggest the F1 (~95 mph) description read, "Manufactured homes without tie-downs or with weak tie-downs are rolled over and destroyed, those with strong tie- owns sustain major damage to roof, windows, awnings, etc." The F2 (~125 mph) description could read, "Manufactured homes are destroyed, even with tie-downs."

To evaluate damage to site-built houses, I agree with the general comments at the Forum that the Fscale descriptions should include sections called "weaknesses to be aware of" (poor grammar but useful and brief). These would highlight simple things that anyone with minimal training could inquire about and detect in damaged houses to evaluate whether there were exceptional weaknesses in the load path that would cause premature failure and result in a downward slide in the F-scale rating.

Descriptions of damage to vegetation are not very useful because trees are highly variable in how they fail and the wind speed required for failure. Stripped bark is a poor indicator of wind speed. However, since much wind damage is simply uprooted or broken trees and some assessment of wind speed may be requested, perhaps we can suggest a lower limit for "widespread toppling of trees" in an F-scale description (at F1, about 100 mph I expect).

Related references on vehicles in wind:
Schmidlin, T.W., P.S. King. 1995. Risk factors for death in the 27 March 1994 Georgia and Alabama tornadoes. Disasters 19:170-177.
 Schmidlin, T.W., P.S. King. 1996. Cars and tornadoes: Where is the research? Bull. Amer. Meteorol. Soc. 77:963-964.
Schmidlin, T.W., P.S. King, B.O. Hammer, Y. Ono. 1998. Tornado effects on motor vehicles. Preprints, 19th Conference on Severe Local Storms, p. 428-430, Amer. Meteorol. Soc., Boston.
Schmidlin, T.W., P.S. King, B.O. Hammer, Y. Ono. 1998. Behavior of vehicles during tornado winds. J. Safety Res. 29:181-186.
Schmidlin, T.W, B.O. Hammer, Y. Ono, L.S. Miller, G. Thumann, P.S. King. 2000. Wind tunnel tests of the stability of vehicles in severe winds. Preprints, 20th Conference on Severe Local Storms, p. 427-430, Amer. Meteorol. Soc., Boston.
Hammer, B.O., T.W. Schmidlin. 2001. Deaths in motor vehicles caused by tornadoes in the United States, 1900-1998. Environmental Hazards (in press).
And… Quick Response Reports 98, 106, and 119 from the Natural Hazards Center, University of Colorado, Boulder.

Back to Top

LARRY VANNOZZI

Three main issues were developed during this meeting (modified F-Scale, consistent assessment of damage, and data collection/archiving). I accept the modified F-Scale and consistent assessment issues in their entirety and so will not rehash them in this article. However, I have several comments for altering the third (data collection/archiving) issue. In particular, I will address database preservation, Storm Data, and time spent on surveys. Following those remarks, I will make additional comments regarding training.

It seemed that much of the discussion regarding data collection/archiving at the forum revolved around maintaining the tornado database and the need for additional data to be collected and added to Storm Data. I believe that, assuming the Fujita scale is revised, this group must also address how the revision applies to the current tornado database, if at all. This is a major item of concern and should be added to the data collection/archiving issue. Should revisions apply to the current database? If so, how will that be accomplished? By whom? It is my recommendation that the revision should not apply to the present database, or at least that the present database should be maintained in its current state while an additional new database could be developed by applying the revised scale to the old database. Either way, this is a significant concern that should be addressed as part of the data collection/archiving issue.

Much time was also spent at the meeting discussing the need for additional information to be collected during surveys and for that information to become part of the database by adding it to Storm Data. Based on comments that were made, it is clear that the Storm Data database and its users are disconnected. Further, it is evident that this ‘disconnect’ has existed for a long time. One has to wonder about the viability of a publication that is failing to meet the needs of its users. Although this is a significant issue that should be addressed, I suggest that it be done outside of the future work attempted by the participants in the Fujita Scale forum.

I am concerned with a need that was expressed at this forum. Namely, it was suggested that the number of man-hours that were spent developing the tornado survey be included as part of the data. This recommendation presumably developed from a need to know how much work or effort was put into the survey. I reject this recommendation. I believe that the extent of the survey information will speak for itself; a thorough, detailed report will be the result of a significant effort while a brief and/or skimpy report will be the result of a weaker effort. Further, the number of man-hours will be a false indicator of the effort expended in the survey since that number will depend on (and vary greatly by) the talent of the people that are conducting the survey. Tracking the number of man-hours is an administrative workload that I believe is unnecessary. I strongly believe that the work will speak for itself and that the amount of time spent on the surveys should not be tracked.

In my final comment, I merely want to support the notion that extensive training needs to be developed and implemented to all storm surveyors. There is a significant need for this within the National Weather Service (NWS). The NWS has been weak in this area but is presently trying to improve its damage survey efforts. Some innovative tools and training were discussed at the forum. I hope that these ideas can be refined and implemented as we try to improve the practice of damage surveys.

Much work awaits  those of us interested in improving the Fujita Scale. Even after we can decide upon a new scale, it must be presented and agreed upon by a wide audience. If we can improve damage survey methods and usefulness of the information, then the work will be worthwhile.

Back to Top

ROGER WAKIMOTO

Back to Top