November-December 1997 STORMTRACK features the Jarrell, Tx Tornado.


  1. COMMENTARY by Tim Marshall

The tornado outbreak in Texas on May 27, 1997, otherwise known as the Jarrell, Texas event, made me once again think about what we know or donít know about forecasting tornadoes. I have looked at this event several times over and I am left scratching my head. Tornado outbreak? No way. Will storm cells discretely develop southwestward along a stationary boundary? No chance. And will the tornadoes that develop move southwest too, last a fairly long time, and be up to wedge status? Donít make me laugh. Once again, mom nature taught me a lesson.

Some may say, "Well at least you chased that day". True, but I wasnít really serious about it. Low level wind fields were weak and any storm that moved east or northeast would have very low storm relative helicities. With all the instability around I figured lots of storms would develop. The slow moving front passed Dallas that morning and I headed southeast around 11am to keep ahead of it. A line of towering cumulus became visible as I headed southeast of downtown. I stopped and set up my satellite dish and watched The Weather Channel. The first tornado watch box was issued for northeast Texas. The box was placed south and parallel to a pseudo-warm front and extended from Dallas to near Shreveport. I was in the western part of this box in what seemed like the perfect place. At 12:30pm, the radar on The Weather Channel only showed two isolated storms developing near Longview, Texas. These "diversion" storms held my interest for awhile and I continued to monitor them with the false assumption that anything which developed further southwest (i.e. near Waco) would head my way! Hearing that a tornadic storm somehow popped up between Waco and Temple, I headed west only to find Interstate 20 closed due to a major traffic accident. All traffic was at a standstill for about one hour. By the time I got back to Dallas and headed south on I-35, the storm was passed Belton heading southwestward for Jarrell. It wasnít long before I heard that I-35 was closed at Salado because a highly visible tornado just off the interstate. Need I say I missed the show.

I have studied tornado outbreak days for more than twenty years. I have seen many "traditional" tornado days with the classic (textbook type) Colorado low and dryline. There also have been many "unique" tornado days (i.e. Plainfield, IL, Jarrell, TX). Over the years, I find myself becoming less confident in my forecasts. This is because more and more "unique" tornado situations are being added to my data base each year. Soon, I will have more "unique" events than "traditional" events. The end result seems to be chaos!! Given this scenario, I see two possibilities for chasing: 1) chase everything on any risk day (which I canít possibly do) or 2) try to pick and choose the best chase days knowing that Iíll miss them most of the time. The moral of this story is clear: Donít give up your day job!


The Iowa National Weather Association Severe Storms and Doppler Radar Conference will be held 3-5 April 1998 at the Holiday Inn Des Moines Airport (where it was last year) in Des Moines, IA. More than 200 are expected to attend second annual gathering! Reserve your room at the Holiday Inn for the special rate of $65 by calling 1-800-248-4013. The subject will be tornadoes, tornadoes, tornadoes. The conference will include a storm chasers video-fest (show and tell session). Additional conference details posted at: Registration before January 1 is $75.00, $50 for students, and $250 for vendors (includes 10 minute presentation). Registration should be mailed to: Central Iowa NWA, Box 7512, Urbandale, IA 50322. John McLaughlin. his email:

The 16th Conference on Weather Forecasting and Analysis will be held January 12-16 at the Phoenix Convention Center in Phoenix, Arizona. Registration for this conference ranges from $120 to $315 depending on the time of registration and whether you are a member of the American Meteorological Society. Sessions on severe weather forecasting will be held on Thursday, January 15th. For more information, contact the AMS at 45 Beacon Street, Boston, MA 02108, phone: 617-227-2426.

David Dowell and Howie "Cb" Bluestein have published a paper entitled: "The Arcadia, Oklahoma Storm of May 17, 1981: Analysis of a Supercell during Tornadogenesis" in the October 1997 issue of Monthly Weather Review.


Marty Feely sent in excerpts of an article by Chris Olson from the Omaha World-Herald dated October 2, 1997: ""Paris has the Eiffel Tower. St. Louis has its Gateway Arch. Seattle has the Space Needle. Now, Robert Hogenmiller Jr., an Omaha commercial artist, proposes a 610-foot-high stainless-steel landmark for Omaha: Vortex, the Tornado Tower. The tower, at an estimated cost of $35 million, would include a restaurant, a tornado museum, a virtual-reality theater depicting a tornado, an education-conference center on tornado safety, an amusement ride to the top and observation decks with 25-mile views. At night, lasers would give the tower the illusion of spinning counter-clockwise like a tornado, said Hogenmiller, who doesn't yet have financial backing for his idea..... ' We certainly wouldn't want people to think Omaha is glorifying a tornado,' said T.J.Knight, president of Downtown Omaha, 'But once you get beyond that, the tower would be a wonderful educational and entertainment facility. í " Wow, should we take up a collection?

Andy Kula submitted a news report from the Des Moines, IA Register: "A flying cow and a Doppler Radar tower are among the features to be unveiled on September 14th at a dedication ceremony for Eldora, Iowaís latest recreational area: TWISTER PARK. The 12,000 square foot area will be home to the usual jungle gyms and swings, along with unique playground equipment including a "Dorothy" and a flying truck. The parkís wooden structures will have the appearance of a ravaged community. The park has been financed solely through community donations. Cost: $70,000."

Antonio Caridade sent in a newspaper article about a tornado in Japan from the Asahi Evening News: "Twisters and storms hit the Kyushu region on Tuesday, October 14th, killing one person in Nagasaki Prefecture and seriously injuring two students in Fukuoka Prefecture. Hailstorms raged in Izuka, Fukuoka Prefecture, and over Saga. Meteorological officials said the unstable weather was caused by a cold front moving south. Officials said the twister began gathering force around 1:40pm on the sea about 500 meters from the town of Gonoura, Nagasaki Prefecture. The twister had grown to 400 meters in height by the time it arrived in the neighboring town of Iki. It turned over five boats and blew off the roofs of ten homes."

Colin McIntyre send in a newspaper article about a tornado in Alberta, Canada: "A twister over Missawawi Lake near Lac La Biche caused $250,000 in damage, but no injuries, when it hit cottages and trailers on June 8th. (Lac La Biche is about 100 miles northeast of Edmonton, about 54.5 degrees north latitude). The tornado was on the ground for 40 seconds, but wasnít detected by radar."


Rich Thompson ( writes: "Jarrell's been discussed quite a bit in the past few months, but not many people have tried to explain why that storm managed to produce multiple, significant tornadoes in that environment. Here's what I've noticed about this event:

1) Vertical shear - 12z soundings from FWD (Fort Worth) and DRT (Del Rio), as well as area vertical wind profiles did not suggest a supercell shear profile (weak Bulk Richardson Number Shear, weak storm relative helicity). The local vertical wind profiles looked pretty sad even near the time of the event...though the profile probably varied more in close to the storms.

2) Temperature/moisture profile - exceptional CAPE (Convective Available Potential Energy) in the 12z soundings (near 5000 J/kg unmodified), with a classic "loaded gun" profile. Boundary layer dewpoints ranged from 74-78 degrees Fahrenheit.

3) Mesoscale boundaries - A north-northeast/south-southwest oriented surface boundary (weak quasi-stationary front?) across central Texas. This boundary was well-defined in surface observations and visible satellite imagery as a narrow line of cumulus. Visible imagery also suggested a gravity wave (from collapsing storms over Arkansas/Mississippi originated in Oklahoma the night before) propagating southwestward and "intersecting" the surface boundary near where the Jarrell storm complex initiated.

4) Event evolution - It's my best guess that the Jarrell storm(s) did not originate as supercells, certainly not in the classic sense. Perhaps the enhanced baroclinic (horizontal) and vertical vorticity (cyclonic horizontal shear) along the surface boundary provided a source for storm scale rotation through tilting/stretching. Once rotation was established (from the bottom upward to the mid-levels), then supercell processes may have become relevant. The observed CELL motion was very slow to the south (not to be confused with mesoscale convective system motion which was south-southwest at 20-25 knots), so storm-relative flow through the troposphere would have been optimized.

Perhaps the real importance of the storm-relative winds was to keep the new updrafts removed from developing rain cores such that the storm was not undercut by outflow prematurely from the north or northeast. I still don't think you can apply a classic conceptual model to this case, because storm motion didn't result from continuous propagation as with most long-lived, violent-hose-producing supercells (i.e., it was not just a single storm). It seems to me that any pressure perturbation effects would have helped the storm move more south-southeast (off the boundary), while the gravity wave phase speed seemed to agree well with storm motion and may have played an important role in keeping the storms on the surface boundary.

From eyewitness accounts, pictures and video, it is not clear that the storm or individual tornadoes displayed supercell characteristics PRIOR to tornadogenesis (e.g., lowered, rotating wall cloud, clear slot, etc.). The tornadoes had some "landspout" characteristics and appeared to originate from flat updraft bases with no apparent RFD (rear flank downdraft) or precipitation nearby. Since I'm not aware of any 7000 CAPE Denver Cyclone cases, who's to say that these tornadoes weren't largely non-supercellular given the outrageous contribution from vertical stretching? I suppose it's possible that there were some mixed supercell/non-supercell processes going on, which might explain the bizarre evolution of the Jarrell tornado from what looked like a stationary, strong (maybe) landspout to a suddenly south moving multi-vortex tornado more characteristic of high-end supercells. I hope this storm doesn't become another "Plainfield" where everyone and his dog tries to force this event neatly into the same part of the spectrum as, say, the Red Rock/Andover supercells (the Plainfield event brought about an endless search for large storm relative helicity values, since storm relative helicity was thought to be the main player by many people at the time)."

Erik Rasmussen ( says: "I had the privilege two weeks ago to give a talk at a conference of National Weather Service forecasters in Dodge City, Kansas, and to COMAP in Boulder. Here are some of the ideas from that talk and other recent discussions regarding what is and is not a supercell and a supercell tornado. First, let me say this: we need to start thinking in terms of the physical processes, and move away from semantics. I think this is true in all areas of severe storms... from spotting and chasing to research. For the sake of discussion, let's define two types of tornadoes: 1. Landspout... the rotation is supplied by concentration (convergence/stretching) of pre-existing vertical vorticity ONLY, and 2. Non-landspout... the rotation is supplied through tilting or downward transport of vorticity, implying nearby-downdrafts are fundamentally important. The landspout tornado requires an updraft to move over a pre-existing meso-circulation in the boundary layer. Not much is required of this updraft except that it be strong enough, and persist long enough, to do the convergence/stretching thing.

The non-landspout tornado requires a bit more. It requires pre-existing rotation aloft, at SOME scale larger than a tornado (as Gene Moore pointed out... this spans a big range of observed circulationís in real storms). It does NOT require the rotation to be pre-existing for a certain period of time; rather, at some moment it exists and is available to be moved to the boundary layer via tilting or downward advection. Further, it requires a downdraft near the initial circulation to do the tilting and/or downward transport. It requires an updraft that is strong enough, and persistent enough, to then concentrate the vorticity that has been brought to the lowest levels into a tornado, and maintain that tornado for a certain period of time. In general, I think it's becoming apparent that this process requires that the updraft not be a single pulse that then gets loaded with precip and collapses; i.e., a certain amount of storm-relative upper flow (a.k.a. deep shear) is required.

Does the non-landspout require a mid-level meso? No. Does it require persistent rotation for 45 minutes? 30 minutes? 20 minutes? No. It requires a source of vorticity to initiate a larger-than-tornado circulation in the lowest few kilometers; a downdraft; and an updraft associated with the circulation that can persist long enough to form and maintain the tornado. All of the foregoing is laced with my opinions, but they are informed opinions (in my opinion !), and are the subject of several investigations and articles. To me, growing understanding of the non-landspout tornado (a.k.a. supercell tornado) may require us to change our notions of what supercells are, or at least their importance to tornado formation. Further, I think that we almost certainly will need to change our operational, chase, and spotter perceptions of what is and isn't important in the tornado formation process. Essentially, any strong low-level rotation in an environment locally supportive of supercells should be viewed as potentially tornadic. On the other hand, in my opinion (!), the presence of a persistent or even long-lived mid-level mesocyclone is not of much significance to the tornado formation problem. My point in the earlier Jarrell post was that, owing to enough storm-relative UT flow in that storm, the precip and updraft were segregated, and the precipitation fell down shear, leading to a forward-flank boundary and an environment locally supportive of a tornadic supercell. This is largely speculation on my part. Many of you have looked at a lot more data. So please prove me wrong.

Rich Thompson responds: "This entire Jarrell discussion focuses on the MECHANISM behind the Jarrell tornadic storm, and how it may have varied from the "classic" violent tornado producing supercell. That's not to say that Jarrell wasn't a supercell, but if the definition gets so large as to be inclusive of almost all storms that rotate (and looking at radar data nationwide MANY do rotate!), then it's of little use. From the perspective of a national forecaster, do I have to assume that a tornadic storm is possible in virtually every combination of shear and instability except for zero? The mechanisms Erik is describing seem to apply most in the run of the mill storm days where something is clearly less than ideal, thus some sort of local augmentation is necessary for tornadoes. However, by nature, these events tend to be VERY localized and affect a relatively small area. Sure, if the Jarrell storm hits a town it's devastating, but what the heck would have happened if the Kellerville/Allison storm had moved over populated areas? It seems to me that a long-lived mid-level meso gives you MORE opportunities for tornadoes, all else being equal. Also, can't the longevity of the mid-level rotation correspond to the longevity of a tornado in some cases? We know there are many environments where long-lived rotating storms fail to produce tornadoes, but many of the most important cases of violent tornado outbreaks DO consist of storms with long-lived mid-level rotation. If you're going to make the claim that mid-level mesocyclones are not terribly relevant, then I would at least like to see some evidence showing that storms with long-lived mid-level mesocyclones are in some way LESS LIKELY to produce tornadoes than other storms encountering similar local environments. Do you really believe that a long-lived supercell (even non-tornadic) that encounters an outflow boundary (just an example) is less likely to produce a significant tornado than a non-supercell storm that encounters the same conditions?

By the way, I've seen a summary version of the article, and the claim is that the extreme damage was a function of the open exposure and time each dwelling experienced the strongest winds. For a time period of 90-150 seconds, the engineers from Tech estimated the damage would result from winds at or below 200 mph. This estimated wind corresponds to F3. I think they may be right, and this just may be a case where Fujita's wind estimates don't account for enough factors to be practically useful."

Erik Rassmussen ( responds to Rich Thompsons post: "Once again, our intense focus and debate on the definition of a supercell may be somewhat unproductive. I don't have to issue watches and warnings for a living (fortunately for the public!), but I think it's still fair to say that if it rotates, it's worth some special attention. I.e., if it's a supercell by the definition of "updraft well-correlated with rotation", then it's probably a storm capable of producing some severe weather. I think the tornado production issue is perhaps a somewhat separate issue. I'm having a heck of a time disproving the hypothesis that local augmentation is almost always necessary. I now look back over a couple of decades of chasing and wonder if most/all of those successful chases were the result of local augmentation, and those zillions of crappie chases (on "supercell days") were the result of lack of augmentation. Am I crazy? Don't answer that!

On the brighter side, my climatology work (in agreement with a lot of good work in the past such as Jon Davies') has also shown that significant tornadoes are generally associated with a threshold combination of shear (e.g. SRH) and CAPE, so NO... as a national forecaster, I would think that not much has changed. It's just that if there's a gob'o'CAPE, and there's enough storm-relative upper flow to remove the precip from the updraft, and there's a boundary hanging around or the storm can generate one itself, then a non-landspout tornado is possible. Here, I would have to reiterate my earlier rant about a need to shift focus to the physical process of tornadogenesis, once we begin to understand it, instead of the ingredients for supercells in general. These two things are obviously not mutually exclusive, but maybe not as intimately tied as we once thought (remember when we were all told that > 50% of mesocyclones produce tornadoes?).

I can argue the following: if there are mid-level mesocyclones, this is evidence that the environment is providing some certain threshold level of CAPE and shear (or whatever the relevant parameters are). This means that it is probably that much easier to get the local augmentation. For instance, a boundary could double the SRH locally. Or, deep shear is strong enough that storm-generated forward-flank boundaries could be effective at getting low-level rotation going. With confidence I can point you to a number of cases of NON-LANDSPOUT tornadoes in which the rotation first intensified near the ground, and then built quickly upward through the storm, followed in short order by a tornado. In these cases, I think the mid-level meso was of little significance, being a consequence of low-level processes, not the cause.

Gene Moore ( says: "I remember one of the most important discoveries that came out of early tornado research was the "mid-level mesocyclone". The fact that this circulation we were watching not only extended 25,000 feet into the storm, but it started there. What a revelation that was! Today, I fear the (radar) detection of mid-level rotation is commonly thought of as the "beginning of a tornado". When a radar operator sees mid-level rotation on the monitor, supercell comes to mind....does it not? In reality that mid-level mesocyclone may not be that uncommon for thunderstorms moving in a good current of (mid to upper level) westerlies. An artifact of covering the country with Doppler radarís. We now see what was always there. Mid-level rotation alone, does not make a supercell. Quite a few (isolated) thunderstorms that come off the front range contain rotation. It's probably why outflow boundaries are so important in the high country. It triggers this garden variety (high based) storm to reach below the boundary layer inversion and become a "real supercell". Without the connection to the low level inflow/circulation that cell is nothing more than another "thundershower" (its a good thing Chuck's not on this thread). Can we have a thunderstorm with a mid-level mesocyclone and not call it a supercell? That might be progress.

Iím still a fanatic on the subject of boundaries, but since the early days I have made extreme changes to my chase methods. With the exception of my old 200/300 mb rules my whole approach is different. I try to determine where the initial base of the storm will be and make my forecast for that level....that becomes my surface. I ignore what's below that. So I negate all shallow boundaries the cloud base can't "see". It also throws out shallow warm fronts. My favorite boundary remains the downstream outflow boundary from anvil precip. This year I had the joy of watching a cumulas-calvas (A large TCU with a small anvil, but no precip) that moved north and made contact with the forward flank downdraft of an existing cell. In fact it may be further downstream than that...I'm not sure how far the accepted FFD extends. Since that boundary is so beautifully vertical through such a tall column it really works well with an approaching updraft (from another source). Another feature that makes this boundary the best (in my estimation) is that it's hard for it to "pulse" like the outflow from the main convective elements of the storm. I'm talking about pulsing in both wind velocity & temperature. Since it's from the anvil the flow is smoother, more consistent. When the updraft hits that vertical wall of deep down motion a strong mesocyclone winds-up fast. Since this downdraft is not tilted with height like the storm that it originated from the mid-levels of the new updraft are forced to tilt alongside the boundary instead of over it. Sometimes this tilt increases the S-R of the new updraft. This particular case netted a couple of tornadoes.

Ed Calianese ( writes: "I'm happy to see the continued discussion of the Jarrell event - fairly extensive discussion has been occurring intermittently here in North Texas concerning this event, as it struck "close to home". As Lon Curtis mentioned, the storm developed in the WSFO FTW county warning area (CWA), produced several strong tornadoes with significant damage in our CWA, and later devastated several areas south of our CWA. The event stirs up interest due to its occurrence in a non-classic violent tornado setup (marginally sheared environment). While composing Storm Data for our CWA, I had the opportunity to go through the Archive IV 88D data from KGRK (Granger) several times, and discuss the event extensively with one of the individuals who saw the storm first-hand, Lon, in an effort to logically put the event together. Alan Moller and Mike Foster were also involved in some of these discussions. It was concluded that indeed most of the tornadoes that occurred were associated with a supercell thunderstorm. One tornado (a likely landspout) did developed in northern Bell County in association with a small non-rotating storm (per radar data) that formed about five miles south of the supercell. This storm was quickly "eaten up" by the larger complex.

Though I agree with much of what has been stated previously regarding this "non-classic" event (theoretically speaking it was non-classic), I must interject the fact that the radar data was more classic than has been suggested. The storm initially formed south of KACT (Waco) and was not indicating rotation from the onset. Our first severe thunderstorm warning went out quickly after the storm developed when it still appeared to be non-severe. In fact, it had obtained a VIL (vertically integrated liquid) of only about 25 by that time (our warning forecaster warned due to his anticipation of rapid severe weather development due to the ungodly amount of instability present). Within 30 minutes, radar indicated storm-scale rotation and the first tornado warning was issued (Lon was viewing a tornado at about this time). The storm proceeded to strongly indicate supercellular characteristics on radar (very tight reflectivity gradient on the southwest flank, deep storm-scale rotation, and what I like to call a screaming eagle reflectivity appearance. I have yet to see a non-rotating storm exhibit this latter characteristic. In addition, both reflectivity and velocity data indicated the presence of a rear flank downdraft. Lastly, many of the radar indicated tornado signatures were embedded in a weaker and broader circulation. This was especially true of the actual Jarrell circulation that was embedded in a several nautical mile diameter mesocyclone and had at least 50 kt gate-to-gate shears (from KGRK) for three volume scans before Jarrell was hit (the tornado associated with this signature touched down several miles north of Jarrell)."

Alan Moller says: "I see no reason to redefine "supercells" at the present time. It is extremely important that operational forecasters identify mid-level mesocyclonic storms (vertical vorticity of at least 10-2 sec-1 through at least 1/3 of the storm's depth, lasting for tens of minutes) as supercells - tornadoes are not the only threat from these beasts! We at Fort Worth began in 1997 to explain to advanced spotters that the "cascade paradigm" is not the only route (not even the most common route) for tornadogenesis, and that "spotting" low-level conditions under storms is of the utmost importance - especially at distances from the radar. Further, when issuing warnings I want to hear from the most knowledgeable spotters (i.e., chaser-spotters) as far as detecting low-level circulationís outside of radar range. Good chasers are the people who will understand what they are seeing. Dadgum, chasers CAN also be spotters! Lon Curtis proved that at Jarrell. Good spotters always help, but some of this stuff is rough to handle unless one has seen many storms...

I support Ed Calienese's view (with some caution pending another look at the data), that Jarrell was a supercell. Mid-level mesocyclones/echo overhang were common, and several times small BWERs (bounded weak echo regions) were identifiable. Tornadoes generally were preceded by very strong low-level convergence along the gust front, occasionally with a low-level meso, and followed by a TVS (tornadic vortex signature) with the tornadoes. "Landspout- like" stretching likely figured into tornadogenesis, but the tubes would not have been so nasty without the parent supercells contributions (whatever THAT is). Granted, the "propagation" was a mixture of continuos/intermittent development, but (I think I recall) so were Browning's 1964 storms. Different supercells in '64 fired from individual flanking line towers associated with the first storm - with each succeeding supercell producing tornadoes. I don't have the Jarrell data at the present time, but I think someone calculated SRH's (storm relative helicities) of 100-200 m2/s2. I must admit that I would have been thinking primarily of wind/hail-filled HP storms, but knowledge of Plainfield and presence of the May 27th boundary should have keyed any forecaster with his/her salt into considering the tornado threat.

How many great dry line tornadic storms have we seen - away from areas of favored terrain? I'm also with Gene Moore here. If severe storms go up on boundaries ahead of the dry line, they often have better tornado potential. More examples - the 1980 Grand Island storm (thermal gradient area northeast of a sub-synoptic surface low, which was well ahead of the dry line.) The 6-13-76 Jordan Iowa event - (warm front ahead of dry line), etc. Chasers who opted for dry line were slashed by both events. When to be on the dry line?... Big dynamics, dry line bulge, pressures falling rapidly, strongly backed winds, etc.

On the subject of gravity waves, yes, I guess that they can initiate storms. Operationally, however, I have seen morning gravity waves, propagating through the inversion, ruin environments favorable for severe storms far more often than otherwise. These "packets" of gravity waves often veer low-level winds and decrease moisture convergence along boundaries. Further, how can we ever be sure that a gravity wave was necessary for storm initiation? (Granted, evidence seems pretty strong in the 5/10/85 case)."

Gene Rhoden ( presents some of his ideas: "After reviewing the incredible sequence of photos taken by Lon, I must say that this event is a little more "classic" than what I had previously thought. Several of his photos show a tornado prior to the Prairie Dell event with what I would consider as a classic RFD (rear flank downdraft) occluding off the updraft (no doubt in my mind), even though the vortex has a cylindrical double/triple end-wall landspout "looking" structure. Lon's latter photos of the Prairie Dell rope show the base of the vortex holding almost stationary while the "top" seems to be pulled into a newly developed updraft core.

The vortex then appears to be stretched until it segments and the new updraft reorganizes the circulation rapidly just north-northeast of the subdivision. This looks like a type of "hand the baton off" possibly not completely unlike what happened with the Wichita/Andover tornado between Haysville and the Hydraulic area of town. Its simply amazing to see how the Prairie Dell "Tight Rope" transitioned into a maxi vortex just as it struck the subdivision (as if that was its only purpose) then rope out just south of the addition. The storm base at the time of the transition showed a rather large bowl lowering or "pregnant appearance" typical of rather large tornadic mesos with latter shots showing evidence of large scale RFD wrapping around the circulation as the tornado reached maximum size. I must also agree with Edís' post regarding the radar showing mesocyclonic signatures just prior to the first reported tornadoes with the storm taking on a classic "flying eagle" appearance. There are no storms that I know of that have had that appearance and were not classified as "supercells".

I have also just reviewed the damage photos that Tim Marshall shot of the event. I must say that I was IMPRESSED! It is quite likely that with the relatively slow translation coupled with the immediate widening of the vortex over the subdivision, that extreme winds could have lasted for a couple of minutes over any one point which is evidenced by some slabs with peripheral anchor-bolts showing much caked mud on them as well as a concrete stair set in front of a former pier-and-beam house looking more like a ramp due to the amount of caked on tornado "adobe" mixture of hay and mud. While the time the tornado had over any given structure may have been longer than "average" which would have contributed to the reason why the houses were so well cleaned away, I still am quite confident that the winds in this tornado were on the order of 250 mph. I do not feel that an F-3 rating for this storm holds much merit given comparison with other storm damage surveys that I have done.

Back to the Jarrell vortex, since nature works in a continuum of storm "types" without solid boundaries between our devised LP/CL/HP categories, why wouldnít there be a continuum of vortex types and for that matter of vortex development processes and combinations thereof i.e. vortex hybrids? Does all "supercell" tornadogenesis have to occur in the same fashion (whatever that may be) to be considered non-landspout or supercellular? Two events, the Ash Valley KS event, the first several of which most invariably were landspouts, with the "big one" being a "I'm not so sure" and the Lazbuddie II event two years ago captured in great detail photographically by NWS personnel from upstate NY show side by side transition between typical landspout processes and classic mesocyclonic processes within a short amount of time and space. These aforementioned cases and possibly others would be good initial candidates for trying speculations about vortex "hybrids" (if they exist).

Lon Curtis responds to Gene Rhodensí post. "What a truly fascinating series of posts you guys are making! I've tried to stay "on the sidelines" in the debate over whether the Jarrell tornado produced F5 or F4 or F3 damage ... I think the ultimate resolution of the wind speed issue lies in photogrammetry, but I don't know that anyone is doing that work. With respect to the supercell or not debate, I confess that have assumed the storm was a variation between Classic and HP. After about 1800Z, I was not in a position to see the storm from a distance, I was under the updraft (except for the 1840-1900Z period when I had to cross the precip core to get back ahead of the storm). The rain-free base was huge by the time the storm was south of Belton. The separation between the RFB and precip was on the order of 3-5 miles for most of the period from 1930Z to 2045Z, except that both the Morgans Point-Lake Belton tornado and the Jarrell tornado became rain-wrapped toward the end of their cycles. One of my photos (from 0.4 miles south of the town of Jarrell, looking back north at 2032Z) shows classic supercell features (large, bowl-shaped wall cloud lowering and tail cloud extending from the precip core back toward the RFB/wall cloud), as mentioned by Gene R. I suspect that the demise of the Jarrell tornado was hastened by a fairly strong cell which developed just north and northeast of Georgetown (~12 miles S of Jarrell) as the Jarrell tornado evolved. I think this cell, which was producing +RN and hail by 2045Z, disrupted the juicy air SE of the Jarrell storm and helped cause its early demise. I think we also need to remember that the Jarrell tornado was at least the seventh tornado produced by this system, and it didn't stop producing tornadoes at Jarrell. The Cedar Park tornado was (fortunately) not as intense as either the Jarrell or the Morgans Point-Lake Belton ones, but the Lake Travis tornado was (last time I checked) credited with F4 damage.

Williamson County Tornadoes on May 27, 1997 by Craig A. Green, N5WEH

The 27th of May was to be my last day to work before my scheduled vacation which will complete the week. I had to supervise one of my stations to be installed in Austin and was halfway expecting some chance of severe weather after the Red River storms of Monday the 26th. I was surprised to find that the SPC (Storm Prediction Center) had issued a moderate risk for severe weather for central Texas and as Pamela and I packed for the Austin trip, we included our chase gear. The surface dewpoints were in the 70s with lifted indices in the -9 to -12 range for central Texas. With backing 30 to 40 knot mid-level winds from the west, sufficient shear was present for supercell development. The forcing mechanism would most likely be the weak cold front which was moving south toward the region. Additionally, the presence of weak upper troughs moving over the area would add upper air support to an already very dangerous situation!

Pamela and I drove to Austin early in the morning and kept the weather on the back burner. After lunch, the SPC issued a watch box for central Texas which included Williamson county at the extreme western end. At about 1400 local time, I logged in for some weather data on the Internet and found that some storms were beginning to fire north and NE of Austin. Having completed my mission, we started north, thinking that we could monitor NOAA radio and amateur frequencies for guidance. A quick stop at my usual place in Austin, and a quick peek at TWC (The Weather Channel) yielded two storms. One was due north of us in Bell county and the other one was in east Texas. There seemed to be no particular concern for the Bell county storm at about 1500 when we watched TWC, so I decided to head north and decide later, which is often the case.

As we drove North on I-35, we began to hear reports of a tornado in Bell county on the 145.15 machine and decided on the north storm. As we passed the Georgetown exits, Many Texas Department of Safety marked units (Black and Whites) passed us at over 90 MPH. We could see a dark lowering to the north over I 35 about five or six miles away and as we reached the 269 mile marker at 1535, traffic stopped on the Interstate due to DPS roadblocks. Pamela began videotaping the Jarrell tornado which was now visible about five miles up the road. Not wanting to get trapped on I-35 with what was apparently a large wedge tornado headed south towards us, I drove through the grass on the right side of the roadway to get on the northbound access road. We had escape routes to the east and south, so we drove up to a DPS road block at a crossover near the 270 mile marker. I was just about to explain our spotting/chasing posture to the officer when, directly across the highway, in a plowed field next to the southbound service road, large dust whirls began lifting and swirling off the ground at 1540. About 200 yards away a tornado was spinning up on the west side of I 35! The DPS officer glanced at he weak tornado, uttered an expletive as he entered his vehicle, and sped away southbound from his post. This of course eliminated the problem of explaining our actions, but presented another more immediate issue requiring resolution.

We continued up the northbound access road taping the dust tube tornado with a second eye on the big wedge bearing down on us. About 100 yards up the access road a county road slanted off to the ENE and we drove down about 1/4 of a mile out of harms way. We jumped out of the chase truck at 1542 and began our documentation of events. The dust tube was still grinding away to our WSW and I nervously glanced north. No sign of the wedge because a rain area was in the way. Back to our NE was the big rain core with what was most likely gorilla hail. We continued taping and clicking away until Pamela said Its Time To Move CRAIG!. We jumped back into the truck and sped to a 90 degree turn to the south-southeast and after about 1/2 a mile we stopped. Above us, back to the west was the parent cloud of the dust tube which had now narrowed to a skinny rope with a condensation funnel curving away 45 degree from the cloud base, extending halfway to the ground. Debris could still be seen moving on the ground below the tip of the funnel. Looking back to the SW, a new mesocyclone was forming with a growing, rotating lowering. Looking toward the NW, an immense RFD notch was visible which now extended almost to the new lowering.

As we continued taping, with the rain and hail bearing down on us, a large cylindrical tornado became visible from behind the rain area at the right edge of the RFD notch. The radial motion of the dying Jerrall wedge was astounding; somewhat like a white barber pole sped up ten times. The cylinder bent into an elephant trunk and as we continued our observations, a speeding utility vehicle with flashing grille lights was approaching from the NW. The emerging public servant advised us that golf ball size hail was approaching. Pamela perceived his arrival as a portent to move, so we headed down the road to a right turn back to the northbound access road of I 35. We tried to hide under a bridge crossing at mile marker 269 but the DPS officer offered to lodge me at the county caboose if I entered the on ramp. We settled for the crowded underpass at the 268 marker and found fortune with a spot on the wrong side of the road. After a ten minute stay of heavy golf balls, we were evicted by a deputy into 1/2 inch soft hail. Not wanting to join the circus in Austin with the Cedar Park tornado which we heard on the 146.94 Travis County ARES net, we departed North at about 1615.



I started the day around 6:30 am with the usual review of model output from 00Z the night before, early morning forecast discussions from Texas offices, and the morning severe weather outlook. The SPC (storms prediction center) had issued a moderate risk for portions of eastern Texas, southern Arkansas and northern Louisiana, but the western edge of the moderate risk extended west into central Texas near Austin and Waco. The forecast soundings for the afternoon looked very unstable and when I ran the 09Z RUC output valid at 21Z through PC-GRIDDS, the results looked promising for deep, intense convection, but forecast winds, both surface and aloft, looked pretty weak. Instability was forecast to be extreme, with CAPES (convective available potential energies) of 4000+ j/kg and LIs (lifted indices) of -10C in central Texas by forecast time.

My 11Z hand-analysis of surface data placed a weak surface low on a dying cold front right at Dallas, with a trough extending southwest to near Del Rio. An outflow boundary from overnight convection in eastern Oklahoma and northeastern Texas had pushed across east Texas and was draped from Fort Polk to Conroe to College Station to the low near Dallas. Ahead of the front, surface winds were light southeasterly except in areas impacted by the outflow boundary. North of the front, winds were northerly with some evidence of a secondary front across central Oklahoma. The most striking feature on the map was an area with dewpoint temperatures 75 degrees F or better which was to the right of a Del Rio-Junction-Waco-Palacios line.

I put on my work clothes (suit, dress shirt, tie, etc.) and went to the office. During the morning, I watched the situation evolve via internet sites. The first significant data to arrive was the 12Z soundings, which I obtained from RAP-UCAR. Del Rio was showing CAPE near 4700 j/kg and LI=-12 C (at 7am CDT!!!) while Fort Worth was CAPE near 3000 j/kg and LI almost -10 C. (I didn't look at Corpus Christi but have been told it was almost 8000 j/kg). Such extreme instability before daytime heating took effect foretold incredible instability for the afternoon hours. The 15Z SPC update of the SWO (severe weather outlook) continued a moderate risk of severe storms for portions of central Texas eastward, but noted that the relatively weak wind fields would mean that the primary threat was hail and damaging winds, with only a chance for an isolated tornado. Experience and many hours of reading research papers from AMS conference preprints told me to watch for evidence of storm deviating in the extreme from the hodograph, which suggested storm motions of ~300 degrees at 12 knots.

My 15Z hand-analysis placed the surface low nearer to Waco than Dallas, suggesting that it had drifted southwestward. The outflow boundary to the east was still visible on satellite imagery, as was the trough-cold front to the west. The cooler air east of the outflow boundary appeared to protrude westward to near Corsicana, northeast of Waco, creating something resembling a triple-point at Waco. I left at 11:30am for lunch so I could drop-off film shot in south central Oklahoma Sunday for processing (two tornadoes near Duncan Sunday afternoon), telling the secretary I'd be back by 1:00pm. Lunch was from Chick-Fil-A in Temple and a large cumulus tower to the north grabbed my attention and sent me home for a quick look at data. I ate the sandwich as I drove.

The ~12:40 SRM products off of KGRK and KFWS (thanks, Freese-Notis!) both indicated strong shear suggestive of a developing mesocylcone in the storm I'd seen visually, now located in southern McLennan County. A quick look at a local television radar showed the storm was developing a reflectivity pendant on the WNW side. I put the SRMs and the pendant each together and went to put on my jeans and then headed for McLennan County. The chase was "on"! Come on along! (All times are CDT.)

About 1:00pm I left my house in far western Temple and traveled north on SH317 to Moody. I called the office to report I'd be a little late getting back. A severe thunderstorm watch was issued for the McLennan Co. storm at 12:50pm ("rapidly developing thunderstorm near Woodway"). Passing Moody, I heard the DPS dispatcher at Waco asking a trooper to check the "back side" of that cell near Woodway. I took FM2113, which took me right into the rear of the storm. The sky darkened. FM2113 passed through farm and ranch land broken by several small streams, and about 5 miles northeast of Moody, I came up out of a valley and around a curve and there, a mile or two in front of me, was tornado #1. After several photographs and a chance to observe the very slow motion of the storm, I crept up toward it. Additional photos are shot from ~1/2 mile away. The tornado was undercut by outflow and died. I went down Mackey Ranch Rd to see if help was needed; a mobile home was blown-away but no one was hurt. This tornado moved from ENE to WSW over a distance I would estimate to be a couple of miles. Path width was probably 75 yards or less. Reports were made to NWSFO/FTW when the tornado was spotted and when it dissipated.

At this point, I made one of only two mistakes in the afternoon chase which is about to evolve. Hearing a tornado warning being issued for Falls County, southeast of my location, I continued on to Hewitt and then south on IH35 toward Eddy and Falls Co. In the process, I almost missed seeing tornado #2. Approaching Eddy, it was obvious that to get back south of the storm, I'd have to punch the core, because the storm was not moving southeast, it was going almost due south. Most motorists had abandoned the Interstate as I barreled south at the speed limit or better with golf ball hail everywhere and smaller hail beginning to cover the highway. (Perhaps that explains why I've never worried about repainting the chase truck where the hood is peeling and rusting!)

The scanner was absolutely full of spotter and law enforcement reports of a new tornado between Moody and Troy. Passing through Troy, I took FM1237 back west toward Pendleton and SH317. Topping a hill near Pendleton, I could see tornado #2 off to the northwest. I watch for several minutes as until it became rain-wrapped and disappeared. I'm low on gasoline and I radioed the Belton Fire Dept. to watch the area near Moffat and Morgans Point for new development (southwest of the last tornado) as I stop for fuel near SH317 and SH36 just west of Temple.

I headed on south on SH317 and while crossing from Temple into Belton, hear that a new tornado (#3) is on the ground at Morgans Point. I turn west of FM439 and from the south side of Belton Dam, shoot several photos of the large tornado that is roughly 3 miles away. I declined to cross the dam and approach the tornado for fear that a traffic accident or sightseers will obstruct my only path of escape. Instead, I repositioned to US190 at FM1670, west of Belton, in time to see that #3 is rain-wrapped. I shoot one photo with aid of a telephoto lens as it emerged from the rain and ropes-out. I radioed to the Fire Department that we will need to watch the Lake Stillhouse area for new development, and I head for Stillhouse Hollow Dam on FM1670. While crossing the dam, I glanced left and see a funnel cloud extending earthward; looking down into the valley east of the dam, I see that there was a condensation funnel forming and debris flying; tornado #4 is developing, but soon dissipated. I continued on south on FM1670, and radioed the Belton Fire Department to notify the Salado Fire Department to be particularly alert; I expected the next threat to be from near Youngsport (west of Salado on FM2484) to Salado itself.


I passed through Salado southbound on the IH35 west side frontage road. On a hill south of Salado at the FM2268 overpass (one of my favorite 'spotting' locations), I saw a funnel cloud to me southwest. I continued south on the frontage road and about MP282 on the Interstate, just south of a rest area, stop to watch tornado #5, a thin rope-like tornado as it wandered around in a pasture about 3/4-mile away. After several minutes of observing little movement other than 'wandering', I moved-up on the tornado, eventually closing to ~0.3 mile, documenting with more photos. I realize maybe I'm getting a little bold, and retreated to the Exit 280 overpass and make several more photos, as the pencil-like tornado suddenly begins to move southward and accelerates. I take the east side service road southbound and turn my back to the tornado for about 2 minutes. Halfway to Jarrell, I stop for another quick series of photos. I am amazed ... my "pencil" is now a multi-vortex tornado and growing rapidly in size. (Later surveys and interviews with residents show that the "pencil' likely dissipated as the multi-vortex tornado formed (tornado #6).

I rush southward to Jarrell (another 2 miles or less) and notice that what looks to be most of the residents are outside watching the tornado grow and move toward their city. Just prior to observing this fact, I radioed the Belton Fire Department to have them notify the Jarrell Fire Department that a large tornado is approaching from the north. I stayed with the east side frontage road through Jarrell and take a position 0.4 mile south of the center of town. (Time 3:30pm) I begin to photograph the tornado again, now a large cone which becomes a wedge, and continue to do so until debris falling from the storm make my position untenable.

Turning to "flee" (the chaser is beginning to feel like the chased), I am confronted with the sight of another tornado (#7) on the ground about 3 miles to my southwest. My last view of the Jarrell wedge was as it becomes rain-wrapped. Subsequent survey led me to believe that it was entering the Double Creek subdivision at that time. I moved south on the frontage road and about 3 miles south of Jarrell, stop to 'shoot' the large tornado as it emerged from the rain and slowly begins to rope-out. (#7 turns out to be thankfully brief.)

I now began to get south of the parent thunderstorm, which had overtaken me near Jarrell, so that I can stay ahead of it. As I turn northwest on SH195 north of Georgetown, Bruce Haynie called from Austin (it is 3:55pm, give or take a minute) and wanted to know where I was and where he should go. I tell him to head northwest, out US183, toward Cedar Park and Liberty Hill. Bruce gets out in time to catch the next tornado at Cedar Park, but I'll let him tell that story himself. I continue to Cedar Park, hoping perhaps to get on the west side of the storm again, but to no avail. The traffic jam caused by the tornado there and the hail and rain put an end to my chase at roughly 4:30pm. I buy two beers (among the best I've ever had), and visit with the convenience store customers and clerks about the events I've just witnessed.