"I don't think we're in Boulder anymore!" may have been the refrain of many during a tornado alert on 6 June 1997. It may not have been in the class of the recent F5 Texas tornado, but the tornado that hit eastern Boulder (Figure 1) the first Friday in June did have some resemblence to its snake-like counterpart from "The Wizard of Oz."
A Denver Cyclone Scenario
Early June is the peak time for tornado activity in this area, particularly when there is a Denver Cyclone, also known as the Denver-Convergence Vorticity Zone (DCVZ; Szoke 1991). Climatology for the 1980s indicated that when a well-formed DCVZ is present in June, there is better than a 70% chance of a tornado forming somewhere in or near the zone. Such a feature was present on 6 June, forming when the low-level flow on the eastern plains was from the southeast. In addition, this southeasterly flow was very moist, with dewpoints in the middle to upper 50s, along with low clouds and fog over the area in the morning.
Figure 1. 6 June Boulder tornado (top) forming a waterspout in the Baseline Reservoir, three miles from the FSL building (bottom), where researchers view the storm from the rooftop. (Top photo by Dongsoo Kim; bottom photo by Will von Dauster, FSL.)
The missing ingredient for a well-organized severe storm with midlevel rotation, or a mesocyclone, was a lack of significant wind shear above the surface. In fact, winds were somewhat remarkable in that they were southeasterly through the entire troposphere. Coupled with the high moisture content in the atmosphere, this suggested that the main threat on 6 June was the potential for heavy rains that could produce flooding, given the high level of runoff currently in the area rivers and streams, with snowmelt from an above-average winter snowfall season. Indeed, the first storms of the day formed quite early in the morning well west of Boulder, near Nederland (Figure 2), coating the ground with small hail by midmorning, and ominously reforming over the same area during the late morning. These initial storms produced under an inch of rain in the foothills, and did not create a serious flood threat.
Figure 2. Map showing approximate locations where the Boulder tornado touched down, as well as surrounding areas where funnels have touched down, i.e., near Nederland and Denver. Map prepared by Julie Smith, FSL, using DeLorme mapping software.
Visible satellite images, wind profiler data, and surface mesonets have played an important role in learning more about complex mountain-valley circulations. Figure 3 (a-d) shows a sequence of four visible satellite images on the Weather Forecast Office (WFO)-scale from FSL's WFO-Advanced workstation. The first image (top, taken at 1700 UTC, 11:00 a.m. local time) shows that the low cloudiness near Boulder had burned off, with an area of sunshine prevailing to the east of Denver. However, there was still quite a bit of low cloudiness farther east through much of eastern Colorado. Interestingly, this low cloudiness in southeastern Colorado revealed a beautiful circulation (arrow), verified by the northwest wind at LaJunta and the southeast wind at Lamar, which is that area's equivalent of northeastern Colorado's Denver Cyclone. Studies by Benjamin et al. (1986) and others showed that such a circulation should exist to the north of the Raton Ridge (near the Colorado-New Mexico border). The sparsity of surface data in the area, though, usually makes it difficult to detect the phenomenon with certainty, unless it can be revealed by fog or low cloudiness, as in this case.
Figure 3. WFO-Advanced satellite images showing a) top, at 11:00 a.m. local time, low cloudiness burned off near Boulder but persisting east of Denver, with cumulus clouds stretching north-south 40 miles, along either side of DIA; b) at noon, line now consisting of cumulus and towering cumulus; c) at 1:00 p.m., strongest storms near northern and southern ends of the line; d) at 2:00 p.m., near the time of the tornado.
It did not take much sunshine to produce the first developing cumulus along the DCVZ, which can be seen in Figure 3a (top) stretching north-south about 40 miles along either side of the Denver International Airport (DIA). The next image (Figure 3b) at noon is zoomed in a bit, and clearly shows the line that consisted of cumulus and towering cumulus. Storms were still evident in the foothills, and at the time, forecasters thought that they would be slow to push off onto the plains because of the deep southeasterly flow. The forecast, though, was quickly changed as it became apparent that the line of clouds along the DCVZ was developing into thunderstorms. The visible satellite image at 1 p.m. (Figure 3c) indicates the strongest storms were near the southern and northern ends of the line (arrows).
In addition to being a spawning ground for thunderstorms because of its converging wind field, the DCVZ is also a favorable area for generating nonsupercell or "landspout" tornadoes, so named (originally by Howard Bluestein of the University of Oklahoma) because the formation mechanism for such tornadoes is similar to that of a waterspout. In a landspout, the circulation that develops into a tornado begins as an ambient circulation at lower levels along a boundary, possibly originating by a type of shearing instability as the winds come together along the boundary. Often Doppler radar, like the National Weather Service's WSR-88D radars, can observe a number of incipient circulations along a boundary like the DCVZ even before any clouds develop, with the radar's ability to detect clear air return well below 0 dBZ. A tornado can then form if a towering cumulus cloud should grow rapidly (into an eventual thunderstorm) over one of these circulations, with the updraft of the developing storm stretching and tightening the circulation into a tornado. The magnitude of such tornadoes is limited by the stretching that can occur and the strength of the initial circulation, and typically ranges from F0 to F2.
The detection of the DCVZ was made possible after PROFS (the Program for Regional Observing and Forecasting Services, a precursor organization to FSL) mesonet was installed in 1980. A study by several scientists of the 1981 Denver tornadoes (Szoke et al. 1984) identified the zone as a major factor in the tornadic development. A later study of a 1985 tornado near Erie, using Doppler radar, enabled documentation of the stretching mechanism mentioned above. Over the years, forecasters at the Denver WFO have become well aware of the DCVZ and its ramifications through interaction with FSL and severe weather workshops.
One difference between the events on 6 June and previous "typical" tornado development along the DCVZ is that the initial storm development along the zone failed to produce any tornadoes between 12:30 and 2:00 p.m. It could be that the converging wind field into the DCVZ simply did not generate any significant low-level circulations, but the usual scenario is for tornado development to occur as the storms are rapidly growing along the zone, with the tornado developing before much, if any, downdraft has been generated.
Genesis of the Boulder Touchdown
On 6 June, the initial storms developed along the DCVZ, produced heavy rains and hail through the Denver area and near DIA, and then sent outflow boundaries westward toward Boulder and the Front Range. This westward direction was a bit unusual, but occurred because of the deep southeasterly flow both at low levels and aloft.It was from a thunderstorm that generated along one of these outflow boundaries that the tornado developed east of Boulder around 2:10 p.m., actually touching down a mile west of my house. Two of the early observers of the initial stages of the tornado from the FSL storm observatory (the 5th floor roof), John Brown and Thomas Schlatter, noted that between 2:10 and 2:15 p.m. there was mainly a funnel extending up to a third of the way to the ground, then a gap, then an occasional debris cloud at the surface. This is typically how Colorado landspouts appear, unlike the lengthy funnel in Figure 4, since the circulation is usually not strong enough, in our relatively dry air, to generate a funnel cloud to the surface. The observation also indicated that early in its lifetime, the tornado was not always on the ground.
Bicycle Storm Survey
In talking to some folks near where the tornado touched down during my "bicycle storm survey" on Saturday, they concurred with the above observation, noting that there was some light debris picked up from an initial touchdown near a rather large house north of Baseline Road and west of 75th Street (see map). The most significant damage from the tornado occurred as it continued to move to the southwest (also a rather unusual direction of movement) and struck a house located on the north side of the curve of Baseline Road just at the northeast end of Baseline Lake. A free-standing shed next door was pushed into a wall and heavily damaged. Two large trees were uprooted at the home, and a roof (which I was told was attached with hurricane clips to the walls) was taken off of a nearby garage, lifted up and onto a corner of the house, heavily damaging a portion of its roof. At the other end of the house, the upper portion of a wall and roof corner were taken off, at the point where an air-conditioning unit protruded from the wall. Rather impressive was a 2x4 board that punctured the wall and was left protruding out from just below this damage area. Some tools from the garage were lifted up and deposited about 100 feet away into the yard. The residents of the home took shelter in the basement and were not injured.
Figure 4. Another view of the tornado on 6 June 1997. (Photo by Dongsoo Kim, FSL.)
The tornado crossed Baseline Road, spinning a car around in its path, and then entered the field across the street. The somewhat flattened tall grass was clear evidence that the tornado had visited this field. It was quite remarkable to me how narrow the damage was - no more than 30 to 50 feet in width, and this is consistent with the damage at the house. The tornado apparently went through some trees (though I could see no damage there) then entered the lake, becoming a full-fledged waterspout. It was this transition to a waterspout that made the tornado more visible to FSL employees and others gathered on the roof, between 2:15 and 2:20 p.m. In fact, calls in to the Denver WFO came from quite far away, which led to some confusion as to the exact location and number of tornadoes. When we called in the tornado at 2:15 p.m., we were fairly certain of its location, though our speculation became more certain when it began picking up water.
Playback of a video made by another observer showed the tornado emerged on the southwest side of Baseline Lake. From the roof we could see evidence that some light debris (perhaps plant material and tree branches) were being picked up, before the tornado lifted around 2:20 p.m., for a total path length from the first touchdown of about two miles.
The tornado will likely be ranked as F1 strength, but with a very narrow width, fairly typical of tornadoes of this type. While it is not common for tornadoes to touch down in Boulder proper, a fairly high frequency of tornadoes, usually of the landspout type, does occur just east of Boulder, closer to Interstate-25, which is near where the DCVZ typically sets up. Although the last tornado to actually hit Boulder was in October 1980, last year a tornado touched down west of us in Nederland, and another was easily visible just south of Boulder, so the local area is certainly not immune to tornadoes.
The 6 June tornado was relatively minor, but it could have been much worse because it came within a quarter of a mile from Platt Middle School, just as the last day of the school year was ending. There were many people outside on the school grounds when the tornado approached, and they were whisked inside to the gym. In retrospect, of course, a safer place would have been in the hallways, because it was possible that even a tor-nado of this minimal magnitude could have caused some of the gym roof to break off and fall down on the students and staff.
On a personal note, my daughter and her classmates had quite a scare as they were walking back toward the school along a bike path, from a last-day school trip. About a third of the way back they saw the funnel, but the decision was made to press on toward the school. It did not take long for most of them to realize that they were walking, or running, toward the tornado, which was also moving toward them! As the hail began to fall, near panic ensued as they finally made it back to the school. Luckily the tornado touchdown was east of the school grounds.
Improved Warnings
While we know a lot about how tornadoes such as this form, the ability to issue warnings remains a considerable challenge. Because of their small size, even the advanced WSR-88D Doppler radar has difficulty detecting the circulation beyond about 20 to 30 miles from the radar. In this case, some of us tried to find evidence of a circulation using WFO-Advanced after the event, but could not do so. This points out the importance of reliable spotter information for such events, in tandem with radar and other data on WFO-Advanced, to have any hope of getting out reliable and timely warnings for this type of tornado. It should be noted that the type of tornado discussed here is quite different from a tornado associated with a storm that has a mesocyclone - the mesocyclone at midlevels that precedes a possible tornado is clearly visible on Doppler radar.One interesting possibility for helping to detect tornadoes is a low-frequency sound detection system that is being developed by Dr. Alfred Bedard, a scientist in NOAA's Environmental Technology Laboratory, also in Boulder. Still in its experimental stage, the system can apparently detect both weak and strong tornadoes, and if it is successful in providing a reliable signal, it may be a most useful complement to the WSR-88D radars.
References
Szoke, E.J., M.L. Weisman, J.M. Brown, F. Caracena, and T.W. Schlatter, 1984: A subsynoptic Analysis of the Denver Tornadoes of 3 June 1981, Monthly Weather Review, 112, 4, 790-808.Benjamin, S.G., R. Brummer, E.-Y. Hsie, E.J. Szoke, and J.M. Brown, 1986: Comparisons of a Nested Grid Model Simulation to Observations of a Local, Topographically Induced Circulation. Preprints, 11th Conference on Weather Forecasting and Analysis, Kansas City, Missouri. American Meteorological Society, 5 pp.
Szoke, E.J., 1991: Eye of the Denver Cyclone. Monthly Weather Review, 119, 5, 1283-1292.
(Edward Szoke is a scientist in the Local Analysis and Prediction Branch, headed by Dr. John McGinley. He can be reached by e-mail at szoke@fsl.noaa.gov. More photographs of the Boulder tornado are available on the FSL Website )
