1. Introduction
To keep an accurate tornado climatology, employees from local National Weather Service (NWS) weather forecast offices conduct ground surveys after damaging convective events to determine whether a tornado is responsible for the damage. If it is determined that a tornado caused the damage, an approximate intensity is assigned to the tornado based on several damage indicators by using the Fujita–Pearson (Fujita and Pearson 1973) scale or, more recently, the enhanced Fujita scale (Marshall 2004). In many cases, the determination of whether a tornado occurred is a formality, since there are often credible sightings of tornadoes, or instances in which tornadoes are videotaped or photographed. However, there are also times in which there are no credible sightings of tornadoes and the cause of the damage in the convective event is unknown. In particular, convective events that occur at night provide a challenge to those conducting damage surveys because, in most such cases, only the damage is reported, with little information as to the cause of the damage. In addition, there are times, for various reasons, that an NWS team cannot perform a complete survey after a convective event. One possible aid in such situations is to use ground-based, mobile, Doppler radar data of the convective event.
Over the past ∼10 years, mobile, Doppler radars have been used as tools for obtaining higher temporal- and spatial-resolution data of mesoscale phenomena, such as tornadoes, that evolve over short time scales (e.g., Bluestein and Pazmany 2000; Wurman and Gill 2000; Wurman 2002; Burgess et al. 2002; Bluestein et al. 2003, 2004, 2007; Alexander and Wurman 2005; Wurman et al. 2007). The use of NWS Weather Surveillance Radar-1988 Doppler (WSR-88D) data in the study of, or the identification of, tornadoes is more limited because the data are of comparatively lower spatial and temporal resolutions. In addition, the data obtained even at the lowest elevation angles may be located well above the surface. However, the higher spatial resolution (d ∼ 100 m, where d is the distance between adjacent observations) of mobile, Doppler radar data can be used to identify small-scale circulations, even tornadoes, and the higher temporal resolution (volumetric scans every ∼1 min) of the radar data can be used to track small-scale circulations. Then, the radar data can be combined with global positioning system technology to estimate the approximate location of the storm circulations and compare them with the location of the damage associated with the storm. As a result, mobile, Doppler radar data may be able to confirm the presence of a low-level circulation in the general area in which a tornado is alleged to have occurred. The radar data would be useful especially for tornadoes that are believed to have been on the ground for relatively long time periods (>10 min), since the motion of the tornado could be compared with the motion of the storm or the motion of circulations identified in the radar data. Previously, estimated wind speeds in tornadoes derived from ground-based, mobile, Doppler radar radial-velocity data have been used to assess the F-scale ratings that the tornadoes were given following NWS damage surveys (Burgess et al. 2002; Wurman and Alexander 2005).
This note provides an example of how mobile, Doppler radar data can be used as a tool in the determination of whether reported damage is from a tornado. Section 2 is a brief background of the mobile, Doppler radars used to collect storm data. Section 3 provides comparisons between the damage path and the mobile, Doppler radar data of the supercell at several times. Conclusions are presented in section 4.
2. Background
For details about the synoptic environment of the storm, and for more information about the radars used in this study, the reader is referred to French et al. (2008).
On the evening of 15 May 2003, teams from the National Severe Storms Laboratory and the University of Oklahoma scanned the Shamrock, Texas, supercell with ground-based, mobile, Doppler radars. The radars were located south of the storm as the storm moved through Wheeler County. The radar data used in this study were obtained by the 5-cm-wavelength (C band) Shared Mobile Atmospheric Research and Teaching Radar 1(SR-1; Biggerstaff et al. 2005), which was located east of Samnorwood, Texas (Fig. 1). SR-1 collected reflectivity and radial-velocity data in the supercell continuously during the time the supercell was ostensibly tornadic; SR-1 had a peak power output of 250 kW, a half-power beamwidth of 1.5°, a gate spacing of 75 m, a pulse duration of 0.5 μs, and a maximum unambiguous Doppler velocity of ±19.95 m s−1 (Biggerstaff et al. 2005). The other radar used in data collection was the University of Massachusetts 3-cm-wavelength (X band) radar (Pazmany et al. 2003; Kramar et al. 2005; Bluestein et al. 2007; hereinafter UMass X-Pol), which was located south of Shamrock (Fig. 1). UMass X-Pol data are not shown in this study, but are referenced periodically. Integrated Data Viewer (IDV; Murray et al. 2003) software from Unidata was used to import and view the mobile, Doppler radar data and the locations of the damage paths.
3. The relationship between storm features and observed damage
Following the passage of the Shamrock supercell, a team from the NWS Weather Forecast Office in Amarillo, Texas, surveyed the damage from the storm. The initial conclusion was that two tornadoes, both associated with the Shamrock supercell, formed in Wheeler County (Fig. 1). It was estimated that the first tornado formed at ∼0243 UTC (hereinafter all times are given in UTC) northeast of McLean, Texas, moved southeastward, and abruptly turned toward the northeast, before dissipating near Lela, Texas. The assessment of the first tornado was made without a damage survey (except at the very end of the path in the Lela area), but instead from credible reports of a number of downed trees. It was estimated that the second tornado formed at ∼0300 near Lela, moved northeastward, and dissipated near Twitty, Texas. The determination of the second tornado was made after a survey of damage in the area. The NWS damage assessment indicated that the damage associated with the tornadoes was consistent with a rating of F-1 on the Fujita–Pearson scale. According to the National Oceanic and Atmospheric Administration (NOAA) publication Storm Data, an estimated $150,000 in damage occurred in and around Lela, including damaged homes, businesses, and several overturned vehicles on Interstate 40.
A complicating factor in this study was that the results of the final version of the damage survey by the NWS were not entered into the official database. A preliminary survey was entered mistakenly instead of the final survey (A. Pietrycha 2006, personal communication). The preliminary survey included much of the same information as the final survey but concluded that the two damage paths were connected and resulted from a single tornado. In addition, another mistake was made when the (incorrect) damage survey was entered into the database. In Storm Data, the single damage path appears as beginning near Lela and ending near Kelton, Texas, instead of beginning northeast of McLean and ending near Twitty. For the purposes of this study, the results of the preliminary damage survey will be ignored and only the results of the final survey (Fig. 1) will be evaluated.
The first tornado damage path was oriented in an unusual direction relative to what would be expected given the storm motion. The motion of the storm was estimated from both SR-1 and the WSR-88D to be ∼10 m s−1 from 245°. The track of the first damage path was oriented from the northwest to the southeast (Fig. 1), a direction approximately normal to the path of the supercell. It is possible for a tornado to move in a direction different from that of its parent storm; however, it is unlikely that a tornado would move in a direction normal to the storm over the period of time covered by the supposed first tornado (∼15–20 min). SR-1 was scanning the Shamrock supercell continuously during the entire time period that the first tornado supposedly occurred (∼0243–0300). The continuous scanning of the supercell by SR-1 allowed for the locations of circulations that may have been associated with tornadoes to be tracked every ∼70 s. To investigate further the validity of the first tornado track, SR-1 data from ∼0220–0320 were viewed in relation to the location of the damage path attributed to the first tornado. Using IDV, the radial-velocity data from the supercell and the location of the first damage path were plotted and compared for several times (Fig. 2).
The supercell cyclically produced at least five identifiable low-level circulations1 during the time it was scanned by SR-1 (French et al. 2008). At 0236, approximately 5 min before the estimated start of the first tornado, the supercell was located in the vicinity of the first damage path, but both the rear flank of the storm (where Doppler velocities approached 20 m s−1) and the only identifiable circulation were located several kilometers to the southwest of the damage path (Fig. 2a). The damage path was embedded within the forward flank of the supercell with low-level radial velocities of ∼10 m s−1. At 0245, the first tornado damage path was still located within the forward-flank region of the supercell as a new low-level circulation developed. The new circulation also was located to the southwest of the damage path (Fig. 2b). The same circulation then moved north-northeastward over the next 10 min but still was not in the vicinity of the damage path (Fig. 2c). At 0305, after the approximate end time of the first tornado, a new low-level mesocyclone formed and was located to the southwest of the first damage path as well (Fig. 2d). The term “mesocyclone” is used in reference to the circulation in Fig. 2d because UMass X-Pol data from this time period (see Fig. 12 in French et al. 2008) also show a strong circulation southwest of the damage path, and a strong low-level mesocyclone can be identified in a crude dual-Doppler analysis (not shown) synthesized from both sets of radar data. As at the previous times, the damage path was located within an area of relatively weak radial velocities in the forward flank of the supercell. At 0316, another new circulation was located southwest of Lela as it moved northeastward, overlapping the end of the first damage path (Fig. 2e).
At no point during the time the supposed first tornado occurred was there any identifiable mesocyclone, circulation, or shear feature near the beginning or middle of the first damage path, nor was there any feature that moved in the same direction as the damage path (Fig. 3). In addition, rarely did radial velocities near the damage path reach or exceed 20 m s−1. The lack of any storm-scale circulations in the area of the damage path, the orientation of the damage path relative to the movement of the storm, and the lack of an actual survey of most of the damage path are all factors that cast serious doubt on the validity of the first damage path and/or its cause. More-plausible explanations for the damage will be explored later in this section.
The second damage path (Fig. 1) was oriented in a direction similar to the path of the supercell, and the damage was surveyed by the NWS team. To assess further the validity of the second damage path, SR-1 data from 0300 to 0335 were analyzed and compared with the location of the path. In this case, a large low-level circulation, containing stronger radial velocities than those seen in earlier circulations, was located in the vicinity of the second damage path. For example, in the SR-1 data at 0322, the center of the circulation was located close to an early part of the second damage path (Fig. 4). The center of a low-level circulation is a location where tornadoes often are found, although, in this case, it is difficult to determine exactly where the circulation’s center was using only the single-Doppler radar data (UMass X-Pol data from this time period were extremely noisy). Between 0310 and 0335 (the end of SR-1 data collection), radial velocities between 30 and 40 m s−1 often were located in the vicinity of the damage path. In this case, the radar data are consistent with the conclusions reached by the NWS survey team.
The most intense damage took place at the end of the first damage path, in and around the Lela area. The orientation of the end of the first damage path, after the sudden change in direction, was very similar to that of the second damage path (Fig. 1). As mentioned previously, the only time that any of the low-level circulations crossed the first damage path was near the Lela area at the end of the path (Fig. 2e). A possible hypothesis is that the damage at the end of the first damage path in Lela was caused by a tornado. That tornado then either briefly lifted before touching down again where the second damage path began or the tornado never lifted (meaning the gap between the two paths was an error in the survey). This hypothesis would explain the intense damage in Lela and the supposed change in the tornado’s direction at the end of the first damage path. The indicated damage in the Lela area was consistent with a tornado, although the damage observed along Interstate 40 at the end of the first damage path could have been caused by rear-flank outflow (C. Alexander 2006, personal communication).
If the damage along most of the first damage path was not caused by a tornado, then what caused the damage? The first damage path was not surveyed, so the damage may not have been reported accurately. However, assuming there was an extended path of downed trees, as reported, there are other possibilities of what may have caused the damage. The supercell responsible for the damage was located just ahead of an impending squall line that moved through the area shortly after the supercell moved through Wheeler County. It is possible that high winds associated with the squall line caused the downing of trees and that the squall-line damage was mistaken for tornado damage. Both SR-1 and the UMass X-Pol had stopped data collection by the time the squall line moved through the area covered by the first damage path; however, the squall line was sampled by the WSR-88D in Amarillo (Fig. 5a). As an example, the radial-velocity data from the WSR-88D in Amarillo at 0348 are shown in relation to the first damage path (Fig. 5b). There were radial velocities of at least 30 m s−1 at or approaching the first damage path (oranges and browns located at the western end of the damage path and dark browns located west of the damage path in Fig. 5b). In addition, as the squall line moved through western Oklahoma a short time later, the Cheyenne Oklahoma Mesonet site recorded surface wind gusts of over 35 m s−1. It is plausible, though purely speculative, that the sustained winds at the surface, although likely lower in magnitude than the values shown in Fig. 5b—perhaps 25 m s−1, were still gusting strongly enough to down trees in the vicinity of the first damage path. If the reports of downed trees formed a line segment, the resulting damage path may have been misconstrued as a tornado path, especially given that the area was not surveyed by an NWS team. Another possibility is that individual low-level mesovortices within the squall line caused pockets of wind damage along the first damage path, though mesovortices were not identified in the WSR-88D data for this case. It has been suggested that mesovortices may be responsible for much of the “straight line” wind damage caused by squall lines (Trapp and Weisman 2003).
In this particular case, it would have been difficult for the NWS in Amarillo to perform a similar analysis using WSR-88D data. Scans from both the Amarillo (KAMA) and Frederick, Oklahoma, (KFDR) WSR-88Ds at ∼0305 have been provided (Fig. 6) for comparison with SR-1 data at approximately the same time (Fig. 2d). The first damage path was located approximately 125 (160) km from the Amarillo (Frederick) WSR-88D origin, resulting in a center beam height at that location of ∼2 (3) km above ground level (AGL) and an azimuthal resolution of ∼2 (2.75) km. It would be difficult for the local office to identify small-scale features in their data or compare such features with damage at the surface. In comparison, SR-1 was ∼23 km from the first damage path at 0305, resulting in a beam height AGL (azimuthal resolution) of ∼200 m (600 m), providing more confidence that identifiable features were being reflected at or close to the surface. The higher spatial resolution of the SR-1 data can be seen in the ability to identify approximate circulation centers (Fig. 2). Conversely, it is difficult to identify circulation centers in the WSR-88D scans (Fig. 6) with a degree of confidence suitable for comparisons with the locations of damage paths.
The existence of low-level circulations in an area where damage occurred does not mean necessarily that there was a tornado. Similarly, the absence of a low-level circulation in an area where damage occurred does not mean that there was no tornado. As it pertains to this case, the possibility that a tornado that was too small to be detected by radar caused the damage in the first path cannot be ruled out. However, in this case, a lack of evidence that there was a tornado was used to come to the conclusion that most of the first tornado damage path likely was erroneous. Nonetheless, caution always should be used when speculating on what phenomena Doppler radar signatures represent. The lack of evidence again emphasizes the importance of visual observations, regardless of the quality of the other observing tools used.
4. Conclusions
Ground-based, mobile, Doppler radar data were used to examine the conclusions reached from an incomplete damage survey performed after a damaging supercell moved through Wheeler County, Texas, on the night of 15 May 2003. It was concluded from the survey that two tornadoes formed in the immediate area and that the first tornado moved from the northwest to the southeast for most of its lifetime. The ground-based, mobile radar data for the storm were compared with the location of the first damage path. There were no storm-scale circulations identified near the beginning or middle of the first damage path as this area was embedded within the forward flank of the storm. The information presented herein challenges the conclusion that the damage in the first path was caused by a tornado. A more likely scenario is that the tree damage in the area occurred from straight-line winds in a squall line that passed through the area shortly after the passage of the supercell.
The NWS in Amarillo is currently in the process of changing the database to remove the first tornado and to extend the path of the second tornado as a result of several factors, including this study. In cases in which damage surveys do not provide clear answers as to the cause of damage, or when damage surveys cannot be completed, local NWS offices, especially offices located in the southern plains, should be aware of the possibility that mobile, Doppler radar data of the storm responsible for the damage were collected. The radar data may be used in concert with damage survey information or other available information to get a better idea of what type of phenomenon was responsible for damage in a convective event.
Acknowledgments
The authors thank Michael Biggerstaff and Alan Shapiro, both of whom reviewed early drafts of this work within the first author’s master’s thesis at the University of Oklahoma. Thanks also are given to Curtis Alexander and Al Pietrycha, who provided useful damage information. This study was supported by NSF Grants ATM-0241037, ATM-0637148, and ATM-0437898 (DCD).
REFERENCES
Alexander, C. R., and Wurman J. , 2005: The 30 May 1998 Spencer, South Dakota, storm. Part I: The structural evolution and environment of the tornadoes. Mon. Wea. Rev., 133 , 72–97.
Biggerstaff, M. I., and Coauthors, 2005: The Shared Mobile Atmospheric Research and Teaching radar: A collaboration to enhance research and teaching. Bull. Amer. Meteor. Soc., 86 , 1263–1274.
Bluestein, H. B., and Pazmany A. L. , 2000: Observations of tornadoes and other convective phenomena with a mobile, 3-mm wavelength, Doppler radar: The spring 1999 field experiment. Bull. Amer. Meteor. Soc., 81 , 2939–2951.
Bluestein, H. B., Weiss C. C. , and Pazmany A. L. , 2003: Mobile, Doppler radar observations of a tornado in a supercell near Bassett, Nebraska, on 5 June 1999. Mon. Wea. Rev., 131 , 2954–2967.
Bluestein, H. B., Weiss C. C. , and Pazmany A. L. , 2004: The vertical structure of a tornado near Happy, Texas, on 5 May 2002: High-resolution, mobile, W-band, Doppler radar observations. Mon. Wea. Rev., 132 , 2325–2337.
Bluestein, H. B., French M. M. , Tanamachi R. L. , Frasier S. , Hardwick K. , Junyent F. , and Pazmany A. L. , 2007: Close-range observations of tornadoes in supercells made with a dual-polarization, X-band, mobile, Doppler radar. Mon. Wea. Rev., 135 , 1522–1543.
Burgess, D. W., Magsig M. A. , Wurman J. , Dowell D. C. , and Richardson Y. , 2002: Radar observations of the 3 May 1999 Oklahoma City tornado. Wea. Forecasting, 17 , 456–471.
French, M. M., Bluestein H. B. , Dowell D. C. , Wicker L. J. , Kramar M. R. , and Pazmany A. L. , 2008: High-resolution, mobile Doppler radar observations of cyclic mesocyclogenesis in a supercell. Mon. Wea. Rev., 136 , 4997–5016.
Fujita, T. T., and Pearson A. D. , 1973: Results of FPP classification of 1971 and 1972 tornadoes. Preprints, Eighth Conf. on Severe Local Storms, Denver, CO, Amer. Meteor. Soc., 142–145.
Kramar, M. R., Bluestein H. B. , Pazmany A. L. , and Tuttle J. D. , 2005: The “owl horn” radar signature in developing southern plains supercells. Mon. Wea. Rev., 133 , 2608–2634.
Marshall, T. P., 2004: The enhanced Fujita (EF) scale. Preprints, 22nd Conf. on Severe Local Storms, Hyannis, MA, Amer. Meteor. Soc., 3B.2. [Available online at http://ams.confex.com/ams/pdfpapers/81090.pdf].
Murray, D., McWhirter J. , Wier S. , and Emmerson S. , 2003: The Integrated Data Viewer: A Web-enabled application for scientific analysis and visualization. Preprints, 19th Int. Conf. on Interactive Information and Processing Systems (IIPS) for Meteorology, Oceanography, and Hydrology, Long Beach, CA, Amer. Meteor. Soc., 13.2. [Available online at http://ams.confex.com/ams/pdfpapers/57870.pdf].
Pazmany, A. L., Lopez F. J. , Bluestein H. B. , and Kramar M. R. , 2003: Quantitative rain measurements with a mobile, X-band, polarimetric Doppler radar. Preprints, 31st Conf. Radar Meteorology. Seattle, WA, Amer. Meteor. Soc., 858–859.
Trapp, R. J., and Weisman M. L. , 2003: Low-level mesovortices within squall lines and bow echoes. Part II: Their genesis and implications. Mon. Wea. Rev., 131 , 2804–2823.
Wurman, J., 2002: The multiple-vortex structure of a tornado. Wea. Forecasting, 17 , 473–505.
Wurman, J., and Gill S. , 2000: Finescale radar observations of the Dimmitt, Texas (2 June 1995), tornado. Mon. Wea. Rev., 128 , 2135–2164.
Wurman, J., and Alexander C. R. , 2005: The 30 May 1998 Spencer, South Dakota, storm. Part II: Comparison of observed damage and radar-derived winds in the tornadoes. Mon. Wea. Rev., 133 , 97–119.
Wurman, J., Richardson Y. , Alexander C. , Weygandt S. , and Zhang P. F. , 2007: Dual-Doppler and single-Doppler analysis of a tornadic storm undergoing mergers and repeated tornadogenesis. Mon. Wea. Rev., 135 , 736–758.
Final NWS-estimated tornado paths based on a damage survey and the approximate locations of the mobile, Doppler radars (black squares). The damage paths attributed to tornadoes 1 and 2 are depicted as black lines. The major highways (blue lines) and towns (gray circles) in the area also are depicted. County lines are shown in red.
Citation: Weather and Forecasting 24, 3; 10.1175/2008WAF2222147.1
SR-1 radial velocity (m s−1) at 0.5° elevation angle collected at (a) 0236, (b) 0245, (c) 0255, (d) 0305, and (e) 0316 UTC 16 May 2003. The thin black line marks the location of the first damage path as estimated by the NWS in Amarillo. Black circles outline low-level circulations identified in the radar data. Towns in the area (filled circles) and county lines (red lines) also are depicted. The scale for radial velocity is the same for all scans. The approximate heights AGL for the circulations shown in (a)–(e) are 310, 290, 280, 190, and 180 m, respectively.
Citation: Weather and Forecasting 24, 3; 10.1175/2008WAF2222147.1
Estimated paths of five low-level circulations (L1–L5) identified in the Shamrock supercell in relation to the location of the first damage path. The approximate direction of storm motion (from 245°; arrow) is shown. Towns in the area (circles) and county lines (thick lines) also are depicted.
Citation: Weather and Forecasting 24, 3; 10.1175/2008WAF2222147.1
As in Fig. 2 but at 0322 UTC 16 May 2003. The thin black line marks the location of the second tornado damage path as estimated by the NWS in Amarillo. The circulation is located approximately 200 m AGL.
Citation: Weather and Forecasting 24, 3; 10.1175/2008WAF2222147.1
KAMA WSR-88D scan showing (a) radar reflectivity (dBZ) and (b) radial velocity (m s−1) at 0.4° elevation angle collected at 0348 UTC 16 May 2003. The thin black line marks the location of the first damage path as estimated by the NWS in Amarillo. The western end of damage path 1 was located ∼120 km east of the WSR-88D in Amarillo. The height of the center of the radar beam varied from ∼1.9 to 2.3 km AGL along the length of the first damage path. The black circle outlines a circulation associated with the Shamrock supercell as the storm was overtaken by the squall line. Towns in the area (filled circles) and the OK–TX border (blue line) also are depicted.
Citation: Weather and Forecasting 24, 3; 10.1175/2008WAF2222147.1
WSR-88D scans of radial velocity (m s−1) at ∼0.4° elevation angle from (a) KAMA at 0303 UTC 16 May 2003 and (b) KFDR at 0306 UTC 16 May 2003. Both scans are centered at the same location. The location of the first damage path (thin black line), towns in the area (filled circles), and county lines (red lines) also are shown. The scale for radial velocity is the same for both scans.
Citation: Weather and Forecasting 24, 3; 10.1175/2008WAF2222147.1
The term “circulation” is used here instead of “mesocyclone” because of the inherent ambiguity involved in defining a mesocyclone through use of single-Doppler radar data. During most of the time that the supercell was undergoing rapid cyclic mesocyclogenesis, UMass X-Pol data were not available.
