[1] In chasing there have been many occasions when we hear reports of tornadoes that have been seen by some citizen or spotter very near to our position. In the past, we may have heaped abuse on the poor quality of spotting only to discover later in some cases that the reports were quite accurate and it was our perspective that was faulty for some reason. This is not always so, but it happens often enough that I am not willing to consider every such report a "bogus" (non-existent) event. I have learned to be somewhat cautious in rejecting the validity of reports that I cannot verify visually. Of course, some fraction of these reports are "bogus," perhaps even most of them in some situations. But I believe we should be willing to entertain some doubts about our own perspective before we judge the reports of others too harshly.

[2] The term "mesocyclone" can be said to be a misnomer if it is regarded to be a contraction of the term "mesoscale cyclone." Meteorologists consider "mesoscale" to encompass a range of sizes from roughly 1000km down to about 100 km (arguably). Thus, the cyclone associated with a convective storm is too small by something up to 2 full orders of magnitude to be considered a mesoscale phenomenon, as the term is widely accepted. "Mesocyclone" in the context of supercells and convective storms is historical, rather than conforming to our ordinary usage of "mesoscale" terminology.

[3] A tornado data base that doesn't include information about the sources for its input numbers is not worth very much. It becomes impossible to know how to interpret the data, either as individual events or in aggregates.

[4] Discussion added on 26 June 1998:

This addition has been prompted by some exchanges on an e-mail discussion group. Other than the Glossary definition, all definitions I've seen, including my own (see disclaimer, above), are not peer-reviewed ... hence, none have any formal "approval" except the Glossary definition, which I see as seriously flawed.

A vortex has to meet some threshold intensity at the surface for it to be considered a tornado. However, when we see a debris whirl at the surface, it's not obvious how to go about measuring the windspeeds to see if they meet some windspeed threshold. Lacking a Doppler lidar, the visible evidence of a tight circulation capable of producing a debris cloud at the surface is all we have to go on in ascertaining whether or not a tornado is present.

It gets worse, though, when we consider how to define the threshold windspeed at the surface for a vortex to be considered a tornado. The Fujita scale windspeeds, as discussed on my "Pet Peeves" Website, aren't calibrated, so although they might be accurate in some sense, it would be purely coincidental ... they're essentially a WAG. I won't repeat my gripes about the F-scale here.








Windspeed (mph)







Thus, according to Ted Fujita, the threshold is 40 mph for the weakest of tornadoes. You can use this "information" however you wish. You might also want to see my discussion about "records" regarding tornadoes.

From my perspective, a tornado is defined (by any reasonable definition), as a vortex. A vortex is a kinematic entity ... a process described completely by the specification of the velocity field as a function of time and space. The dynamics of vortices ... a description of the physical processes that give rise to the vortex ... probably vary considerably from one event to another, and in general are not very thoroughly known. It makes no sense to me to base a vortex classification system on dynamics when those dynamics aren't known definitively and certainly can't be assessed just by looking at the vortex and attendant storm. We're only marginally capable of saying anything about the velocity field, much less the processes that give rise to it. Instead, I'm arguing that we put some essentially arbitrary (but, hopefully, reasonable) thresholds on which vortices we are going to call tornadoes ... as I am doing in this essay ... from a purely kinematic perspective. When and if the dynamics become well-known and can be determined definitively from available observations, then they can and should be the basis for a revision of the classification system.

Chasers are free to use any terminology they wish. Of course, if they submit postings somewhere using terminology that succeeds in passing my irritation threshold (a poorly-specified function of time), I feel free to gripe about it.

[5] Note added 27 June 2000:

Mr. Jim Means of San Diego sent me an e-mail regarding this definition. My responses are included (in boldface):

>I've been reading your proposal for a new definition of a tornado, and
>I'm bothered by one aspect. You say that a tornado is "...intense enough
>at the surface to do damage..." I find this part of the definition
>almost unusable, because it doesn't define what sort of damage the
>tornado must be capable of. Maybe I'm nitpicking, but if a vortex passes
>over the straw house of the little pig it might do extensive damage and
>be considered a tornado, but then when it goes over the ICBM bunker it's
>not a tornado anymore? 
You misinterpreted the definition ... it is NOT a definition predicated on 
the actual OCCURRENCE of damage.  My definition merely requires that winds 
at the surface be strong enough to do damage ... i.e., it is CAPABLE of 
doing damage, irrespective of whether or not damage is done.  In effect, it's 
a windspeed criterion ... but I have NOT specified the windspeed.
>It seems like the ability to do damage in a
>particular environment is too subjective, and in the case of waterspouts
>I think most of the time it would be impossible to judge, since the
>there is usually nothing to damage.
Obviously, I agree.
>I would suggest that you amend your definition to:
>A vortex that is clearly visible at the surface that extends upward at
>least as far as the cloud base (with that cloud base associated with
>deep moist convection) should be considered a tornado.
Sorry, but what does it take to be "visible"?  Condensation?  A tracer 
(like dust or water) to depict the vortex?  This is also rather subjective 
and full of contingencies.
>With this definition one doesn't have to quibble about the ability to
>cause damage, and all waterspouts should fall into the definition of a
Waterspouts have "the ability to cause damage" even when they DON'T "damage" 
anything.  You've missed the point I was trying to make.

A second e-mail exchange went:

>Actually, I don't think I
>misinterpreted your definition, I was just pointing out that even
>potential damage needed to be defined. I didn't see any way of judging
>the ability to do damage for something like a waterspout, it's just too
>That's why I thought requiring a visible vortex would be easier to
>judge.  Obviously if there's a condensation funnel it's visible, or if
>there is a dust swirl at the ground, or if there is damage debris, or if
>there is a spray column.  If there is none of that, then we wouldn't be
>able to evaluate the ability to cause damage anyway--we wouldn't know it
>was there! I think just asking for some evidence of the vortex
>circulation at the ground is much less subjective.
Subjectivity is inescapable, and I would argue that your proposed 
solution involves subjectivity as well, but of a different sort than 
mine.  If I thought it was useful, I would prefer a windspeed criterion 
rather than depending on the presence of something to enable the 
visualization of the flow.  The absence of (a) accurate and extensive 
windspeed data in tornadoes, and (b) clearly-defined criteria for what 
might constitute "damage potential" makes this an unworkable solution.  
Hence, I chose to be intentionally vague.  Your definition requires 
that someone actually SEE the vortex ... a damage swath 10 miles long 
and 100 yards wide could not be classified a tornado if no one 
actually saw it happening.
Philosophically, it's annoying to me that scientists have this 
compulsion, virtually to the point of it being a neurosis, to remove 
subjectivity in any and all forms.  I have nothing against "objectivity"
and to the extent that objectivity can be introduced, it is NOT 
something inherently BAD.  But many scientists with whom I have to 
deal seem to follow the principle:  subjective = unscientific crapola, 
objective = true science.  This ignores many things and, imho, is both 
unrealistic and not justified.  But this topic, in general, deserves 
another whole essay ...

Yet a third exchange went:

>> Your definition requires that someone actually SEE
>> the vortex ... a damage swath 10 miles long and 100 yards wide could not be
>> classified a tornado if no one actually saw it happening.
>Well, if that's true for my definition then, I think that's more or less
>the case with your definition, too, because you require the connection
>to the cloud base, and without observers (perhaps observing with
>portable doppler radar) you can't make that connection either from a
>damage swath.
As noted, I never had any intention of developing an objective 
definition, so this is a moot point.  However, it would be unlikely 
that anything other than a tornado connected to a deep convective 
cloud would create such a damage swath, whether it was observed or 

Obviously, I like discussions like this. These are the means by which we can all come to an enhanced understanding of the phenomena. I hope you readers of this page enjoy the debate ... and I always invite participation.

[6] Note added 15 August 1998:

As already noted, chasers (and spotters and even the general public) have encountered many things that have at least some resemblance to tornadoes. The growth of chasing has provided more examples where events fail to fit into neat categories than most non-chasers can possibly imagine. I would call the whole spectrum of things with which we might have to deal convective vortices ... even dust devils fit within this spectrum, since they are associated with convection ... it's just dry convection! The whole plethora of convective vortices (part of which is discussed in the preceding section) now must include such oddities as dust devils being enhanced through interaction with cumulus clouds, gustnadoes developing into circulations reaching cloudbase, funnels around the edges of dissipating Towering CUmulus clouds, dust devils reaching damaging intensity, etc. The only things not included are such oddities as "mountainadoes" since they are not associated with any convective event.

If we take the Fujita scale seriously (a dangerous thing ... see my Pet Peeves), its lower limit for an F-0 tornado is 40 mph winds. On this basis, it surely is possible to have F-0 dust devils. When observing dust and debris clouds ... the visible manifestation of a circulation at the surface ... it generally is not possible to estimate windspeeds on a visual basis. Thus, we can't simply look at a surface debris cloud and determine if it's met Fujita's criterion (or anyone else's windspeed criterion, of course) to be "tornadic" windspeed.

From a purely scientific viewpoint, having a "bestiary" of convective vortex events represents no serious problem. Given that vorticity is often present in the atmosphere, owing to gradients in velocity, it shouldn't be surprising that from time to time that vorticity is "spun up" in ways that don't seem to fit what most of us would call a "true tornado". Since the atmospheric boundary layer typically has a lot of horizontal vorticity (due to strong vertical wind shears), it ought not to be surprising that some of it might be tilted and stretched by convective processes, including dry convection. All one needs to intensify pre-existing vorticity is to stretch it ... and the presence of strong vertical motions during convective processes seems to promote that vortex stretching (a fluid dynamicist's view of the conservation of angular momentum). Although scientists are not immune from wanting to create names for the events we see, nor are we immune from arguing about the classification system (i.e., the taxonomy), ultimately most of us recognize the potential difficulty with any taxonomy. Thus, science is relatively comfortable accepting the "bestiary" of convective vortices. Of course, we may not have the resources to study all of them equally intently ... "cold air funnels" don't get much research attention because they don't do much damage!

When it comes to the public, it's difficult to generalize. However, it's likely that if we wanted to acquaint most ordinary (non-meteorologist) citizens about the convective vortex spectrum, they would probably not stay interested very long, if at all. Let them be seriously affected by a convective vortex, whatever its physical orgins, though, and it's a different story ... "It struck without warning!" If it was a convective vortex, they probably will feel that "It was definitely a tornado!" What the public probably wants is for us to warn them in advance of every event that affects them, and at the same time, never to warn them about events that don't influence them personally. This, of course, is impossible. I've discussed this elsewhere.

Some really tough questions regarding taxonomy are faced by operational forecasters. They have to make the tough decision to warn or not to warn. If there is a whole spectrum of events out there, they should at least be aware of the spectrum. Operational offices often receive reports of strange things ... as it stands now, the typical response is to check the radar. If a significant thunderstorm isn't there, the tendency is to "blow off" these oddball reports. It's bad enough with non-supercell (or low-topped supercell) tornadoes, that often are associated with storms that don't look very menacing on radar ... imagine a dust devil being enhanced by a cumulus cloud! Such events can attain damaging proportions, although they almost certainly would not be classified as tornadoes by most meteorologists. These oddball things definitely can happen, but they are not commonly intense enough to do significant damage. Note that "gustnadoes" probably are pretty common, whereas dust devils enhanced by cumulus clouds are probably not at all common. If we were to warn for all convective vortices that might become damaging, and expect to catch all of them, we might as well issue a tornado warning everywhere for the entire thunderstorm season! Science offers no way to predict most of these weird non-tornadic convective vortices. We have enough trouble with events that are pretty unambiguously true tornadoes!

Operationally, it seems to me that there is a certain subset of convective vortices that ought to be called tornadoes and those would be the events for which we should realistically try to warn, to the best of our scientific and operational ability. Those should simply be "Tornado Warnings" with no indication that we can predict their size, intensity, or duration ... since we can't!! If the oddball things happen, it's useful for forecasters to have some idea of what they might be ... no one calling in an oddball report should be "blown off" by the forecaster simply because there's no beast of a radar echo in the vicinity of the putative event. Perhaps a special statement describing the situation could be issued if the events are persistent, but do not warrant a tornado warning. This means that forecasters need to be trained to recognize the possible convective vortex events they might encounter on the job that are not clearly tornadic. Training is, sadly, not a high priority within the NWS, but that's another story.

[7] Note added: 17 June 1997: A Mr. William Lewitt has e-mailed me and said:

Earlier in your essay, para 1 I believe, your definition of a tornado included having damaging tornadic winds at the ground. If that premise is accepted, then gaps in the damage pattern would have to be a "skipping" tornado, or perhaps (as you suggest) the termination of one tornado and the start of another.

This is an excellent point. If the damage at the surface ceases because the intensity decreases, then strictly by my definition above, it is no longer a tornado. If a vortex continues aloft, it has become a "funnel cloud" and so, in this sense, the vortex is "skipping." In the case where the vortex continues aloft, it seems silly to call each path segment between gaps (caused by temporary cessation of damaging winds at the surface) a separate tornado. I find myself troubled by this point, though, for at least two reasons

Nevertheless, it is an interesting dilemma that follows from the definition I have given earlier. Perhaps that definition could be altered as follows:

Tornado -- A vortex extending upward from the surface at least as far as cloud base (with that cloud base associated with deep moist convection), that is intense enough at the surface to do damage at one or more points along its path should be considered a tornado.

A vortex that continues over some period producing episodic damage at the surface would then be a single tornado, even though it might have segments where it weakens (at the surface) to the point where damage ceases temporarily. In such a case, it is likely that a vortex flow (albeit a weak one) continues at the surface such that the vortex is, in this sense, continuous in time and even vertically (see my Vorticity Primer discussion). The continuing nature of the vortex in spite of fluctuations in its intensity is really the reason why I included this discussion of skipping in the first place. Phases where there is no surface damage are "funnel cloud" phases of a continuous tornado ... the word "tornado" then applies to the vortex, not to the surface damage path points. If the funnel cloud never produces any surface damage, the vortex is not considered a tornado ... although it might have achieved tornadic windspeeds somewhere aloft. This is an interesting issue for me and I am happy to consider other perspectives.

[8] [Discussion suggested by Dr. Jeffrey Trapp, NSSL] To the extent that we can know the dynamics producing a particular vortex, a dynamical definition seems like an appropriate way to distinguish between damaging mesocyclones and tornadoes. That is, if we can distinguish clearly the processes producing a particular vortex, we might find it easy to classify a particular vortex as a mesocyclone or as a tornado. This is predicated on the plausible assumption that the physical processes producing a mesocyclone are quite different from those producing a tornado. This is a fine concept in principle but (a) we still do not know clearly the tornadogenesis process(es) in general, and (b) it would be difficult to put into practical use, since it we typically do not have direct information about the dynamical processes associated with a particular event as it is happening. I certainly find it easy to endorse dynamical distinctions, as opposed to arbitrary thresholds, for classifying events whenever it is possible. As noted above, however, we are still rather far from being able to understand and identify the dynamical processes responsible for any particular tornado. Hence, as I've said, this idea must remain unrealized for the time being.

[9] 26 June 1998 addition: I note for the record that in the absence of video/film documentation of the event, it can be very difficult to assess the character of the event producing damage. Subtle indications (see above) are often subject to misinterpretation. In one case, a well-respected team of scientists did an aerial survey and concluded that the event was a microburst, only to have to retract that assessment when shown film of a tornado that produced the damage path they surveyed ... even the "experts" can be fooled. Further, politics can affect the decision ... at times, the political pressures to call an event a tornado (or not) will override any other evidence, unfortunately. Sadly, there does not appear to be much of a commitment anywhere to ensuring that damage assessments are done carefully, thoroughly, and by people given sufficient training to be able to rely on their judgements. See here for more discussion of this.

[10] 15 March 2001 addition: I am using the word "circulation" here outside of the formal definition of circulation as an integral of the flow field. That is, the flow resembles a vortex (and may be either cyclonic or anticyclonic). Only scientists familiar with fluid dynamics will even be concerned about the use of this verbiage. I'm trying to be careful, for their benefit. See my Vorticity Primer for more information regarding this topic.