At 0025 UTC 25 June 1996 a weak tornado occurred approximately 8 km south of Cheyenne Wells, Colorado. This location is 75 km southwest of a National Weather Service (NWS) Weather Surveillance Radar (WSR-88D) located in Goodland, Kansas (GLD). The tornado was observed for 10 minutes by storm spotters, had a path length of 1.5 km, and was nearly 300 m wide. Since the tornado developed over open country, no injuries occurred and damage was minimal.
Weak tornadoes are relatively common on the eastern plains of Colorado (Brady and Szoke, 1988). However, since they are usually short lived and the associated circulations are weak, radar detection of these events can be difficult. As Lemon and Quoetone (1995) noted, the WSR-88D Storm Relative Mean Radial Velocity Map (SRM) velocities usually will not indicate weak, small scale circulations due to its limited 1 km sampling resolution along the radial. The radar operator must remember to use a higher resolution sample of base velocity which has a limited detection range of approximately 60 km.
This case study will show how another WSR-88D base parameter, the base spectrum width data, was used to obtain radar confirmation of tornadic development within the thunderstorm which produced the Cheyenne Wells tornado. A comparison between the spectrum width data and more "traditional" radar data during this event will also be analyzed.
SYNOPTIC AND MESOSCALE ENVIRONMENTS
At 0000 UTC 25 June 1996 the air mass over eastern Colorado was potentially unstable. Surface temperatures from hourly METAR observations ranged from 28 to 30 degree C while dew point temperatures ranged from 17 to 20 degree C. Figure 1 shows an initial analysis of Convective Available Potential Energy (CAPE) calculated using Nested Grid Model (NGM) model lifted indices with values around 1500 Jkg-1.
Figure 1. Convective available potential energy (j kg-1 x 100).
Figure 2 shows the 700 mb flow associated with the event using the NGM initial analysis. A weak vorticity lobe was rotating across eastern Colorado which supported upward vertical motion. Note that the steering flow winds were from the southwest. As storms developed, cell motions deviated from the steering flow.
Figure 2. 700mb geopotential heights (every 30 dm) and vorticity (10-5 s-1).
A surface analysis at 0000 UTC 25 June 1996 is depicted in Figure 3. A moisture discontinuity was analyzed from Denver, Colorado, to a mean sea-level pressure low in southeast New Mexico. Dew point temperatures across the boundary differed by as much as 10 degree C. A surface trough was analyzed from central Colorado to east-central Kansas. The analysis also showed southeast surface winds at approximately 10 ms-1. The wind profile presented weak low level veering as shown in the Goodland WSR-88D VAD Wind Profile in Figure 4.
Figure 3. Surface Analysis (mb) 0000 UTC 25 June 1996.
Figure 4. Goodland WSR-88D VAD Wind Profile (kt) Centered at 0014 UTC 25 June 1996.
Widely scattered thunderstorms developed on the moisture discontinuity at 2300 UTC 24 June 1996, and continued for approximately three hours. Due to the favorable wind profile and moderate instability, some storms produced severe weather. By 0025 UTC 25 June 1996 a storm spotter reported a tornado just south of Cheyenne Wells, Colorado.
RADAR INTERPRETATION AND THE USE OF BASE SPECTRUM WIDTH DURING THE EVENT
The synoptic and mesoscale environments leading to this event prompted operational forecasters to determine that weak tornadoes were possible. As Brady and Szoke noted (1988), these weak tornadoes originate at low-levels as small vortices produced by shearing instabilities along a convergence boundary. If these vortices become associated with rapidly developing congestus updraft, they are intensified into tornadoes as a result of vertical stretching. With a moisture discontinuity and surface trough in the area, the above determination was made.
The idea of trying the WSR-88D Base Spectrum Width (SW) product to detect small scale tornadoes came from two papers presented at the "27th Conference on Radar Meteorology" at Vail, Colorado. These papers (Golden and Goodall-Gosnell, 1995 and Collins, 1995) were case studies on waterspout events in the southeastern United States. All the previously mentioned factors led the Goodland radar operator to create a time lapse of spectrum width at the .54 nautical mile resolution when convection was initiated. The WSR-88D was in Volume Coverage Pattern 11 while the event unfolded.
The thunderstorm that produced the Cheyenne Wells tornado moved from 160 degrees. This was slightly to the left of the steering flow as indicated by the GLD WSR-88D VAD Wind Profile in Figure 4. The storm remained over the moisture boundary during its lifetime. The location of the moisture boundary and storm at 0009 UTC is shown using the Base Reflectivity and Base Spectrum Width in Figures 5 and 6.
Figure 5. Goodland WSR-88D Base Reflectivity (dBZ).
Figure 6. Goodland WSR-88D Base Spectrum Width (kt) 0009 UTC 25 June 1996.
Based on the Mid- and High-Level Composite Reflectivity (LRM) products, it was determined that the thunderstorm intensified rapidly from the 0009 UTC scan to the 0020 UTC scan. At 0020 UTC, the Echo Tops product (not shown) indicated storm tops around 15 km.
At 0025 UTC when the tornado formed, the SW product indicated a value of 10 ms-1 (19 kt)approximately 8 km south of Cheyenne Wells. This corresponded well to the position of the tornado based on ground-truth reports (Figure 7). From Table 1 note that a SW value of 8 ms-1 (16 kt) or greater indicates severe to extreme turbulence (Operational Support Facility 1993).
Figure 7. Goodland WSR-88D Base Spectrum Width (kt) 0025 UTC 25 June 1996.
|Moderate to Severe Turbulence||
8 to 15 kt
(4 to 7 ms-1)
|Moderately high SW values|
|Severe to Extreme Turbulence||
>= 16 kt
(>= 8 ms-1)
|High SW values|
(> 6 ms-1)
|High shear values|
The SW product was compared to the corresponding Storm Relative Mean Radial Velocity Map (SRM) product at the time of the event. Figure 8 shows the SRM as the tornado developed. In the location where the high SW value was shown, no mesoscale (or smaller) circulation was identified. However, note on this scan and the 0030 UTC scan, that the mean movement in the SRM was nearly 45 degrees off from the actual movement of the cell in question with the speed being very close.
Figure 8. Goodland WSR-88D Storm Relative Velocity Map (kt) 0025 UTC 25 June 1996.
This could have been a possible reason for no rotation to show up. However, using the WSR-88D Algorithm Testing and Display System (NSSL, 1997), a 160 degree movement input into the 0030 UTC SRM still did not yield an indication of a small scale circulation. As Bluestein noted (1985), it is common to observe small scale tornadoes with no mesoscale signatures.
Unfortunately, a Storm Relative Velocity Region (SRR) or a Severe Weather Analysis (SWA) velocity product was not produced during this time. These higher resolution velocity products might have identified the small area of strong horizontal wind shear indicated by the SW product as noted by Lemon and Quoetone (1995).
The 0025 UTC 0.5 degree Base Reflectivity image is shown on Figure 9. In the northwest quadrant of the storm that was in close proximity to the high SW value, there was no definite feature that would indicate a tornado. This was indicative of the Base Reflectivity through the entire event. The Base Reflectivity did not show a pendant, hook, or Bounded Weak Echo Region during the time of the tornado.
Figure 9. Goodland WSR-88D Base Reflectivity (dBZ) 0025 UTC 25 June 1996.
The 0030 UTC SW product (Figure 10) continued to show a value of 10 ms-1 (19 kt), but this had decreased slightly in areal coverage. Again, the SRM suggested no mesoscale or even smaller scale circulation (Figure 11). In the 0030 UTC 0.5 degree Base Reflectivity product (not shown), no features were present that indicated a tornado was occurring.
Figure 10. Goodland WSR-88D Spectrum Width (kt) 0030 UTC 25 June 1996.
Figure 11. Goodland WSR-88D Storm Relative Velocity Map (kt) 0030 UTC 25 June 1996.
The 0035 UTC SW product (not shown) had a very low spectrum width value associated with this storm. Other products at that time also indicated a general decrease in storm intensity. Storm spotters reported that the tornado dissipated approximately at 0035 UTC.
SUMMARY AND CONCLUSION
During the early evening of June 24, 1996, the GLD WSR-88D Base Spectrum Width showed a high value that corresponded to a weak tornado near Cheyenne Wells, Colorado. The same high value was present for two consecutive volume scans which led to a high confidence in the data. This scan to scan continuity is very important since there are often a number of small areas or single pixels with high values of SW. Without using this continuity, it would be nearly impossible to identify tornadoes using SW.
During this time, the Storm Relative Velocity Map product showed little circulation present. However, the tornado was more easily identified by using the Base Spectrum Width, and confirmed the spotter reports. The use of the spectrum width product may help radar operators detect small scale tornadoes when other products fail. Finally, forecasters need to be aware of the environmental conditions that favor this type of tornado before cells actually develop. Further documentation and case studies in this area are definitely needed.
The authors wish to acknowledge Llyle Barker and George Phillips for their insightful reviews of this paper. We also wish to acknowledge John Kwiatkowski for his expertise in using the WATADS software.
Bluestein, H.B., 1985: The formation of a "landspout" in a "broken-line" squall line in Oklahoma. Preprints, 14th Conference on Severe Local Storms, Indianapolis, AMS (Boston), 267-270.
Brady, R.H., and E. Szoke, 1988: The Landspout - A Common Type of Northeast Colorado Tornado. Preprints, 15th Conference on Severe Local Storms, Baltimore, AMS (Boston), 312-315.
Collins, W G., 1995: Pinellas County Florida Waterspout Case of July 8, 1994. Preprints, 27th Conference on Radar Meteorology, Vail, AMS (Boston), 160-162.
Golden, J.H., and C.G. Goodall-Gosnell, 1995: Tornadic Waterspout Detection by the WSR-88D. Preprints, 27th Conference on Radar Meteorology, Vail, AMS (Boston), 7-9.
Lemon, L.R., and E. Quoetone, 1995: WSR-88D and Visual Observations of a Non-Supercellular Storm and Tornado. 27th Conference on Radar Meteorology, Vail, AMS (Boston), 156-159.
National Severe Storms Laboratory, 1997: WATADS: WSR-88D Algorithm Testing and Display System. DOC, NOAA, National Severe Storms Laboratory, 15-17.
Operational Support Facility, 1993: Student Guide for WSR-88D Training Course. Topic 7 Lesson 3 Specific Applications and Limitations of Base Products. DOC, NOAA, NWS Operations Training Branch, Operational Support Facility, 26.