• Adlerman, E. J., and K. K. Drogemeier, 2005: The dependence of numerically simulated cyclic mesocyclogenesis upon environmental vertical wind shear. Mon. Wea. Rev., 133 , 35953623.

    • Search Google Scholar
    • Export Citation
  • Adlerman, E. J., K. K. Drogemeier, and R. Davies-Jones, 1999: A numerical simulation of cyclic mesocyclogenesis. J. Atmos. Sci., 56 , 20452069.

    • Search Google Scholar
    • Export Citation
  • Beck, J. R., J. L. Schroeder, and J. M. Wurman, 2006: High-resolution dual-Doppler analyses of the 29 May 2001 Kress, Texas, cyclic supercell. Mon. Wea. Rev., 134 , 31253148.

    • Search Google Scholar
    • Export Citation
  • 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 , 12631274.

    • Search Google Scholar
    • Export Citation
  • Bluestein, H. B., C. C. Weiss, and A. L. Pazmany, 2004: Doppler radar observations of dust devils in Texas. Mon. Wea. Rev., 132 , 209224.

    • Search Google Scholar
    • Export Citation
  • Bluestein, H. B., M. M. French, R. L. Tanamachi, S. Frasier, K. Hardwick, F. Junyent, and A. L. Pazmany, 2007: Close-range observations of tornadoes in supercells made with a dual-polarization, X-band, mobile, Doppler radar. Mon. Wea. Rev., 135 , 15221543.

    • Search Google Scholar
    • Export Citation
  • Burgess, D. W., V. T. Wood, and R. A. Brown, 1982: Mesocyclone evolution statistics. Preprints, 12th Conf. on Severe Local Storms, San Antonio, TX, Amer. Meteor. Soc., 422–424.

  • Darkow, G. L., and J. C. Roos, 1970: Multiple tornado producing thunderstorms and their apparent cyclic variations in intensity. Preprints, 14th Conf. on Radar Meteorology, Tucson, AZ, Amer. Meteor. Soc., 305–308.

  • Dong, K., and C. J. Neuman, 1983: On the relative motion of binary tropical cyclones. Mon. Wea. Rev., 111 , 945953.

  • Doswell III, C. A., and D. W. Burgess, 1993: Tornadoes and tornadic storms: A review of conceptual models. The Tornado: Its Structure, Dynamics, Prediction, and Hazards, Geophys. Monogr., Vol. 79, Amer. Geophys. Union, 161–172.

    • Search Google Scholar
    • Export Citation
  • Dowell, D. C., and H. B. Bluestein, 2002a: The 8 June 1995 McLean, Texas, storm. Part I: Observations of cyclic tornadogenesis. Mon. Wea. Rev., 130 , 26262648.

    • Search Google Scholar
    • Export Citation
  • Dowell, D. C., and H. B. Bluestein, 2002b: The 8 June 1995 McLean, Texas, storm. Part II: Cyclic tornado formation, maintenance, and dissipation. Mon. Wea. Rev., 130 , 26492670.

    • Search Google Scholar
    • Export Citation
  • Fujita, T. T., 1975: New evidence from April 3–4, 1974 tornadoes. Preprints, Ninth Conf. on Severe Local Storms, Norman, OK, Amer. Meteor. Soc., 248–255.

  • Fujita, T. T., and A. D. Pearson, 1973: Results of FPP classification of 1971 and 1972 tornadoes. Preprints, Eighth Conf. on Severe Local Storms, Denver, CO, Amer. Meteor. Soc., 142–145.

  • Fujita, T. T., D. L. Bradbury, and C. F. Van Thullenar, 1970: Palm Sunday tornadoes of April 11, 1965. Mon. Wea. Rev., 98 , 2969.

  • Fujiwhara, S., 1931: Short note on the behavior of two vortices. Proc. Phys. Math. Soc. Japan, 13 , 106110.

  • Glickman, T. S., 2000: Glossary of Meteorology. 2nd ed. Amer. Meteor. Soc., 855 pp.

  • Johnson, K. W., P. S. Ray, B. C. Johnson, and R. P. Davies-Jones, 1987: Observations related to the rotational dynamics of the 20 May 1977 tornadic storms. Mon. Wea. Rev., 115 , 24632478.

    • Search Google Scholar
    • Export Citation
  • Kramar, M. R., H. B. Bluestein, A. L. Pazmany, and J. D. Tuttle, 2005: The “owl horn” radar signature in developing southern plains supercells. Mon. Wea. Rev., 133 , 26082634.

    • Search Google Scholar
    • Export Citation
  • Lemon, L. R., and C. A. Doswell III, 1979: Severe thunderstorm evolution and mesocyclone structure as related to tornadogenesis. Mon. Wea. Rev., 107 , 11841197.

    • Search Google Scholar
    • Export Citation
  • Markowski, P., and C. Hannon, 2006: Multiple-Doppler radar observations of the evolution of vorticity extrema in a convective boundary layer. Mon. Wea. Rev., 134 , 355374.

    • Search Google Scholar
    • Export Citation
  • Marquis, J. N., Y. P. Richardson, and J. M. Wurman, 2007: Kinematic observations of misocyclones along boundaries during IHOP. Mon. Wea. Rev., 135 , 17491768.

    • Search Google Scholar
    • Export Citation
  • Pazmany, A. L., F. J. Lopez, H. B. Bluestein, and M. R. Kramar, 2003: Quantitative rain measurements with a mobile, X-band, polarimetric Doppler radar. Preprints, 31st Int. Conf. on Radar Meteorology, Seattle, WA, Amer. Meteor. Soc., 858–859.

  • Rasmussen, E. N., and J. M. Straka, 2007: Evolution of low-level angular momentum in the 2 June 1995 Dimmitt, Texas, tornado cyclone. J. Atmos. Sci., 64 , 13651378.

    • Search Google Scholar
    • Export Citation
  • Rasmussen, E. N., J. M. Straka, R. Davies-Jones, C. A. Doswell, F. H. Carr, M. D. Eilts, and D. R. MacGorman, 1994: Verification of the Origins of Rotation in Tornadoes Experiment: VORTEX. Bull. Amer. Meteor. Soc., 75 , 9951006.

    • Search Google Scholar
    • Export Citation
  • Seliga, T. A., and V. N. Bringi, 1976: Potential use of radar differential reflectivity measurements at orthogonal polarizations for measuring precipitation. J. Appl. Meteor., 15 , 6975.

    • Search Google Scholar
    • Export Citation
  • Weisman, M. L., and J. B. Klemp, 1982: The dependence of numerically simulated convective storms on vertical wind shear and buoyancy. Mon. Wea. Rev., 110 , 504520.

    • Search Google Scholar
    • Export Citation
  • View in gallery

    Surface map at 0000 UTC 16 May 2003 showing the conditions in the Texas Panhandle as the Shamrock supercell was initiating. For individual surface plots, the upper left is the temperature (°C), the lower left is the dewpoint temperature (°C), the upper right is the surface pressure minus leading 9 or 10 (hPa), the middle indicates any cloud cover present, and the wind barbs indicate wind direction with each full barb representing 10 kt (5 m s−1) and each half barb 5 kt (2.5 m s−1). Approximate dryline (warm front) location shown as white (black) scalloped line. The circles show the location where the two cells that would become the Shamrock supercell initiated on radar. The triangle indicates where tornadogenesis is believed to have occurred near Shamrock, TX, later on at 0300 UTC.

  • View in gallery

    Radar reflectivity at 0.5° elevation angle from the NWS WSR-88D in Amarillo, TX, at 0243 UTC 16 May 2003. The white circle encloses the Shamrock convective storm and the convective line extending from north to south about 50 km west of the radar marks the approximate location of the dryline; range rings are shown every 50 km.

  • View in gallery

    Skew T–logp diagram at Amarillo at 0000 UTC 16 May 2003. Line on the right (left) indicates temperature (dewpoint temperature) (°C). The inset in the bottom left of the figure is the hodograph corresponding to the sounding shown with wind speed values (m s−1) and heights (km AGL). (From the Plymouth State archive)

  • View in gallery

    Diagram showing the NWS estimated tornado paths (black lines) based on a damage survey and the approximate location of SR-1 and the UMass X-Pol radar (black squares). The major highways (blue lines) and towns (gray circles) in the area also are depicted. County lines are shown in red.

  • View in gallery

    SR-1 (left) reflectivity (dBZ) and (right) radial velocity (m s−1) at 0.5° elevation angle collected at (a) 0220, (b) 0229, (c) 0233, (d) 0238, (e) 0245, (f) 0305, (g) 0311, and (h) 0315 UTC 16 May 2003. Black arrows indicate hook echoes in the reflectivity field and the blue arrow indicates a reverse-curving hook. White circles enclose, and numbers indicate, the specific circulation referenced in the text. The data were located at heights ranging from approximately 200 to 500 m AGL depending on the range of the echo from the radar. The radar view angle differs for each scan but ranges from (a) 255°–305° to (h) 290°–5°. Range rings are every 5 km.

  • View in gallery

    (Continued)

  • View in gallery

    SR-1 (left) reflectivity and (right) radial velocity at 3.5° elevation angle collected at (a) 0220, (b) 0229, (c) 0234, (d) 0240, (e) 0243, (f) 0306, (g) 0310, and (h) 0316 UTC 16 May 2003. The data were located at heights ranging from approximately 1.5 to 3.5 km AGL depending on the range of the echo from the radar. The radar view angle differs for each scan but ranges from (a) 255°–305° to (h) 290°–20°. Everything else is as in Fig. 5.

  • View in gallery

    (Continued)

  • View in gallery

    Plots of the approximate ground-relative locations of (a) low-level and (b) midlevel circulations, which correspond to the 0.5° and 3.5° elevation-angle scans, respectively, from SR-1 for the 15 May 2003 Shamrock supercell. The black lines indicate the path each circulation took as it dissipated. The list in the upper left provides the formation and dissipation times for each circulation in UTC time. The time interval between data points (not shown) is ∼75 s. Increasing positive values of the ordinate are toward due north. The origin of the graph is the location of SR-1.

  • View in gallery

    Plots of the approximate normalized storm-relative locations of (a) low-level and (b) midlevel circulations, which correspond to the 0.5° and 3.5° elevation-angle scans, respectively, from SR-1 for the 15 May 2003 Shamrock supercell. All of the circulations start at the origin and the white points indicate the normalized locations in which each circulation dissipated. The time interval between successive marks varies from approximately 5 to 8 min. Increasing positive values of the ordinate are toward due north. The origin of the graph is the normalized approximate location where each of the individual circulations developed.

  • View in gallery

    The approximate maximum radial wind shear as a function of time for (a) low-level and (b), (c) midlevel circulations, which correspond to the 0.5° and 3.5° elevation-angle scans, respectively, from SR-1 for the 15 May 2003 Shamrock supercell.

  • View in gallery

    The approximate diameter as a function of time for (a) low-level and (b), (c) midlevel circulations, which correspond to the 0.5° and 3.5° elevation-angle scans, respectively, from SR-1 for the 15 May 2003 Shamrock supercell. The diameter for circulation L1 is omitted because the maximum inbound and outbound radial velocities were located in adjacent gates, so the actual diameter may have been smaller than the width of the range gates. The diameter of a circulation was not calculated at times when significant data were missing near the approximate center of the circulation.

  • View in gallery

    SR-1 radial velocity (m s−1) at 3.5° elevation angle collected at (a) 0244, (b) 0245, (c) 0247, and (d) 0251 UTC 16 May 2003. White circles enclose, and numbers indicate, the specific circulation referenced in the text. The radar view angles differ slightly for each scan but are approximately 280°–305°. Range rings are every 5 km.

  • View in gallery

    UMass X-Pol radial velocities at 0.5° elevation angle at 0304 UTC 16 May 2003. The scale on the bottom expresses radial wind velocities in meters per second. The white circle encloses, and the number indicates, the specific circulation referenced in the text. Range rings are every 5 km.

  • View in gallery

    SR-1 radial velocity (m s−1) at 0.5° elevation angle collected at (a) 0251, (b) 0254, (c) 0258, and (d) 0300 UTC 16 May 2003 showing an increase in approaching radial velocities (greens and blues) in the rear-flank of the Shamrock supercell. The velocity scale is the same for all four images, so it is labeled only in (a). White circles enclose, and numbers indicate, the specific circulation referenced in the text. The radar view angles differ slightly for each scan but are approximately 270°–320°. Range rings are every 2 km.

All Time Past Year Past 30 Days
Abstract Views 0 0 0
Full Text Views 97 94 13
PDF Downloads 73 71 12

High-Resolution, Mobile Doppler Radar Observations of Cyclic Mesocyclogenesis in a Supercell

View More View Less
  • 1 School of Meteorology, University of Oklahoma, Norman, Oklahoma
  • | 2 Cooperative Institute for Mesoscale Meteorological Studies, Norman, Oklahoma
  • | 3 National Severe Storms Laboratory, Norman, Oklahoma
  • | 4 NOAA/NWS, Amarillo, Texas
  • | 5 ProSensing, Inc., Amherst, Massachusetts
© Get Permissions
Full access

Abstract

On 15 May 2003, two ground-based, mobile, Doppler radars scanned a supercell that moved through the Texas Panhandle and cyclically produced mesocyclones. The two radars collected data from the storm during a rapid cyclic mesocyclogenesis stage and a more slowly evolving tornadic period. A 3-cm-wavelength radar scanned the supercell continuously for a short time after it was cyclic but close to the time of tornadogenesis. A 5-cm-wavelength radar scanned the supercell the entire time it exhibited cyclic behavior and for an additional 30 min after that. The volumetric data obtained with the 5-cm-wavelength radar allowed for the individual circulations to be analyzed at multiple levels in the supercell. Most of the circulations that eventually dissipated moved rearward with respect to storm motion and were located at distances progressively farther away from the region of rear-flank outflow. The circulations associated with a tornado did not move nearly as far rearward relative to the storm. The mean circulation diameters were approximately 1–4 km and had lifetimes of 10–30 min. Circulation dissipation often, but not always, occurred following decreases in circulation diameter, while changes in maximum radial wind shear were not reliable indicators of circulation dissipation. In one instance, a pair of circulations rotated cyclonically around, and moved toward, each other; the two circulations then combined to form one circulation. Single-Doppler radial velocities from both radars were used to assess the differences between the pretornadic circulations and the tornadic circulations. Storm outflow in the rear flank of the storm increased notably during the time cyclic mesocyclogenesis slowed and tornado formation commenced. Large storm-relative inflow likely advected the pretornadic circulations rearward in the absence of organized outflow. The development of strong outflow in the rear flank probably balanced the strong inflow, allowing the tornadic circulations to stay in areas rich in vertical vorticity generation.

* Current affiliation: National Center for Atmospheric Research (NCAR), Boulder, Colorado. NCAR is sponsored by the National Science Foundation

Corresponding author address: Michael M. French, School of Meteorology, University of Oklahoma, 120 David L. Boren Blvd., Suite 5900, Norman, OK 73072. Email: mfrench@ou.edu

Abstract

On 15 May 2003, two ground-based, mobile, Doppler radars scanned a supercell that moved through the Texas Panhandle and cyclically produced mesocyclones. The two radars collected data from the storm during a rapid cyclic mesocyclogenesis stage and a more slowly evolving tornadic period. A 3-cm-wavelength radar scanned the supercell continuously for a short time after it was cyclic but close to the time of tornadogenesis. A 5-cm-wavelength radar scanned the supercell the entire time it exhibited cyclic behavior and for an additional 30 min after that. The volumetric data obtained with the 5-cm-wavelength radar allowed for the individual circulations to be analyzed at multiple levels in the supercell. Most of the circulations that eventually dissipated moved rearward with respect to storm motion and were located at distances progressively farther away from the region of rear-flank outflow. The circulations associated with a tornado did not move nearly as far rearward relative to the storm. The mean circulation diameters were approximately 1–4 km and had lifetimes of 10–30 min. Circulation dissipation often, but not always, occurred following decreases in circulation diameter, while changes in maximum radial wind shear were not reliable indicators of circulation dissipation. In one instance, a pair of circulations rotated cyclonically around, and moved toward, each other; the two circulations then combined to form one circulation. Single-Doppler radial velocities from both radars were used to assess the differences between the pretornadic circulations and the tornadic circulations. Storm outflow in the rear flank of the storm increased notably during the time cyclic mesocyclogenesis slowed and tornado formation commenced. Large storm-relative inflow likely advected the pretornadic circulations rearward in the absence of organized outflow. The development of strong outflow in the rear flank probably balanced the strong inflow, allowing the tornadic circulations to stay in areas rich in vertical vorticity generation.

* Current affiliation: National Center for Atmospheric Research (NCAR), Boulder, Colorado. NCAR is sponsored by the National Science Foundation

Corresponding author address: Michael M. French, School of Meteorology, University of Oklahoma, 120 David L. Boren Blvd., Suite 5900, Norman, OK 73072. Email: mfrench@ou.edu

1. Introduction

The presence of a mesocyclone is used often as a sufficient condition for a convective storm to be labeled as a supercell (Doswell and Burgess 1993). Some supercell thunderstorms are known to periodically produce a series of mesocyclones that undergo similar life cycles (cyclic mesocyclogenesis); such storms are referred to as cyclic supercells (Darkow and Roos 1970). Similarly, some cyclic supercells are known to produce a number of tornadoes in a process known as “cyclic tornadogenesis” (Darkow and Roos 1970; Burgess et al. 1982; Dowell and Bluestein 2002a). Cyclic mesocyclogenesis does not necessarily imply that there is also cyclic tornadogenesis. For the purposes of this study, cyclic mesocyclogenesis refers to the development of a series of mid- and/or low-level mesocyclones. In addition, following the definition used by Dowell and Bluestein (2002a), a mesocyclone is defined as any distinct vortex with a diameter of 2–10 km.

Previous studies of mesocyclones have made use of visual observations, radar observations, and numerical simulations to develop explanations and conceptual models of the cyclic mesocyclogenesis process. Lemon and Doswell (1979) proposed a model whereby cyclic tornadogenesis resulted through the development of multiple mesocyclones and updrafts after occlusion of the initial mesocyclone and updraft. Burgess et al. (1982) used a National Severe Storms Laboratory (NSSL) Doppler radar archive of roughly 100 cases of mesocyclones to develop a conceptual model of cyclic mesocyclogenesis. It was found that mesocyclones had lifetimes spanning approximately 45–90 min. Also, it was found that mean mesocyclone diameter (approximately 4–5 km) and radial (Doppler velocity) shear (9–10 × 10−3 s−1) were roughly constant at all levels and decreased prior to mesocyclone dissipation. The model also consisted of multiple updrafts and mesocyclones as in Lemon and Doswell (1979). In the model, the first mesocyclone is cut off from the inflow region as the rear-flank gust front accelerates forward by the low-level rotation of the first mesocyclone. A new mesocyclone forms at the occlusion point where there is strong convergence, and the cycle described above repeats itself with the next mesocyclone.

Adlerman et al. (1999), after noting a lack of observational studies, used a numerical model to simulate cyclic mesocyclogenesis and found generally good agreement with that of earlier conceptual models, particularly at low levels. Outflow from the rear-flank downdraft (RFD) acted to push the gust front outward, and the formation of an occlusion downdraft (from a downward pressure gradient force) soon followed. The RFD and occlusion downdrafts merged, and their outflow pushed the gust front farther to a point where the initial mesocyclone detached from the gust front was cut off from inflow air and eventually decayed. A new mesocyclone formed downshear of the old one near the bulge in the gust front, and the cycle repeated. The cycling process in the study took over 60 min, a longer period of time than the 40-min cycling period observed by Burgess et al. (1982). In a subsequent study, it was found that the mode of cyclic mesocyclogenesis (occluding or “nonoccluding”) changed depending on the hodograph shape, and the magnitude and distribution of vertical shear (Adlerman and Drogemeier 2005). Nonoccluding cyclic mesocyclogenesis consisted of multiple mesocyclones that propagated southward (in a storm-relative sense) along the gust front instead of moving rearward into the storm as with the typical occluding mesocyclogenesis discussed in most studies.

Dowell and Bluestein (2002a, b), in a case study from the Verification of the Origins of Rotation in Tornadoes Experiment (VORTEX; Rasmussen et al. 1994), formulated a conceptual model of cyclic mesocyclogenesis that was based on that of Burgess et al. (1982), but included more specifics regarding the evolution of cyclic tornadogenesis. In both the numerical simulation studies and the observational studies, it was found that new vortices formed along the rear-flank gust front, while mature vortices moved rearward with respect to the parent storm when they dissipated. More recently, Beck et al. (2006) used mobile, Doppler radar data to obtain high-resolution dual-Doppler analyses of a nontornadic cyclic supercell. The authors found good agreement between the behavior of the mesocyclones in their case and the modified Dowell and Bluestein (2002a) conceptual model. However, Beck et al. (2006) found differences, including a much faster cycling frequency for mesocyclones than observed in other studies. Beck et al. (2006) also developed a conceptual model for hook echo regeneration in cyclic supercells.

Another area of interest related to cyclic mesocyclogenesis is the process whereby the cycling terminates. Previous observations of cyclic tornadogenesis have indicated that a family of relatively short-lived tornadoes tends to be followed by a longer-lived, perhaps, stronger tornado (Dowell and Bluestein 2002a). These latter tornadoes have been associated with extremely long paths of devastating damage. The strength of convective storm outflow was hypothesized to account for the change in behavior from cyclic to long-lived tornadoes (Dowell and Bluestein 2002b). In such studies, it was suggested that more observations were necessary in order to confirm these findings.

The above summary highlights the lack of observational studies of cyclic mesocyclogenesis. Many of the earlier observational studies relied mainly on visual accounts (e.g., Fujita et al. 1970; Fujita 1975). Later observational studies of cyclic mesocyclogenesis made use of data from one or more Doppler radars (e.g., Burgess et al. 1982; Johnson et al. 1987). However, the Doppler radars involved in the studies had relatively low spatial and temporal resolutions. Dowell and Bluestein (2002a, b) used higher spatial resolution data from an airborne, Doppler radar. However, ground clutter contaminated wind observations in the pseudo-dual-Doppler analysis at the lowest levels (i.e., below 500 m), and the temporal resolution (∼5 min between scans) precluded a more detailed analysis of the evolution of specific features. Ideally, radar observations of cyclic mesocyclogenesis would include high spatial (∼200 m) and temporal (∼1 min) resolution data of storm cycling in order to track the individual circulations over their entire lifetime. Only recently with Beck et al. (2006) have higher-resolution radar data been used specifically in studies of the cyclic mesocyclogenesis process.

As in Beck et al. (2006), this study uses high spatial and temporal resolution radar data to investigate the process of cyclic mesocyclogenesis. However, this study focuses on specific features of the circulations at multiple levels, including a detailed analysis of how the circulations evolved individually, an area of study that has not been emphasized in other works. In addition, this study will examine the differences between those circulations associated with a tornado and those that were not.

On 15 May 2003, the Texas Panhandle experienced a daily record number of tornadoes (∼25) for that region. One convective storm formed near Amarillo, Texas, around 0000 UTC 16 May 2003, and developed into a supercell that eventually moved into Wheeler County, around 0300 UTC, where it produced large amounts of damage as it moved through the Lela and Shamrock, Texas, areas. The dataset from the 15 May 2003 Shamrock supercell allowed for an examination of the cyclic nature of the storm early in its lifetime. Data from two Doppler radars provided insight into storm processes that may have influenced the transition from a cyclic supercell to a tornadic supercell that caused large amounts of damage. In section 2 of this study, this unique dataset and the instruments used to collect the data are described further. In section 3, specific observations regarding the process of cyclic mesocyclogenesis are detailed and compared with those from other relevant studies. Section 4 contains a brief discussion of probable mechanisms responsible for the observations described in section 3. Overall conclusions are given in section 5.

2. The 15 May 2003 dataset

On 15 May 2003, atmospheric conditions were favorable for the formation of tornadic supercells. Throughout the day, an area of low pressure developed in New Mexico. At the same time, an upper-level trough propagated eastward into the area and the upper-level wind speeds increased. A strong southerly low-level jet helped to provide ample surface moisture, as dewpoints in the eastern Texas Panhandle approached ∼70°F (∼21°C). A sharp dryline extended southward from the surface low pressure area and a relatively weak warm front stretched eastward from the low (Fig. 1). These two boundaries were the focus of convective initiation throughout the day in the Texas and Oklahoma Panhandles.

The storm that eventually was scanned by both mobile, Doppler radars formed near the dryline, about 40 km south of Amarillo around 0015 UTC. At that time, there were two separate cells that later merged to create an area of multicells (not shown), from which a supercell later emerged (Fig. 2). A sounding from Amarillo at 0000 UTC 16 May 2003 likely gave an accurate indication of the atmospheric conditions near the time and location that the Shamrock supercell developed (Fig. 3). Surface-based convective available potential energy (CAPE) of approximately 2500 J kg−1 and 0–6 km vertical wind shear (speed and directional) ∼25 m s−1 (Fig. 3) made the atmosphere conducive for supercells (Weisman and Klemp 1982).

On the night of 15 May 2003, teams from the University of Oklahoma (OU) and NSSL were able to scan, in a coordinated manner, the supercell that moved through Wheeler County, Texas. 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 just beyond the southern border of Shamrock, collected reflectivity, radial-velocity, and polarimetric data in the supercell from approximately 0240 to 0317 UTC 16 May 2003, though not continuously. 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. 4), collected reflectivity and radial-velocity data in the same supercell continuously from 0202 to 0335 UTC 16 May 2003. During this period, the supercell moved through Wheeler County in the eastern Texas Panhandle and later moved into western Oklahoma.

While data were being collected, it was possible that a tornado formed approximately 10 km (25 km) away from the UMass X-Pol (SR-1). Since the sun had set by the time of apparent tornadogenesis (after 2200 local time, or ∼0310 UTC), there was no visual confirmation of a tornado. An estimated $150,000 in damage was done in and around Lela including damaged homes, businesses, and several overturned vehicles on Interstate 40. A National Weather Service (NWS) damage assessment indicated that the damage associated with the supercell was consistent with at least one tornado of rating F-1 on the Fujita–Pearson (Fujita and Pearson 1973) scale over a fairly long swath. The NWS estimated that the second tornado1 formed near Lela and then moved toward the northeast where it dissipated 13 km northeast of Lela on 16 May 2003 (Fig. 4).

The primary data sources for this study were SR-1 and the UMass X-Pol radar. SR-1 had a magnetron transmitter, a linear horizontal polarization, and a fully operational leveling system. It also had a peak power output of 250 kW, a half-power beamwidth of 1.5°, a range resolution of 75 m, and a maximum unambiguous Doppler velocity of ±20 m s−1 (Biggerstaff et al. 2005). The other radar used, the UMass X-Pol, was built by the UMass Microwave Remote Sensing Laboratory. The radar system had an antenna with a half-power beamwidth of 1.25°, and a range resolution of 150 m (Pazmany et al. 2003). The radar used a marine radar transceiver modified so it could transmit pulses with both horizontal and vertical polarizations (Seliga and Bringi 1976). The radar did not have any levelers. Radial-velocity data collected for this study by the UMass X-Pol had a maximum unambiguous Doppler velocity of ±16 m s−1.

3. Observations of cyclic mesocyclogenesis

Previous discussions of radar-indicated storm-scale circulations have employed several standards to identify mesocyclones, tornado cyclones, and tornadic circulations (e.g., Glickman 2000, 484–485; Rasmussen and Straka 2007). The different standards usually classify circulations based on the diameter of, and the radial velocities in, the circulation, while taking into account the elevation angle of the radar data. In the case of this particular dataset, there are several circulations, all of which have somewhat different diameters and radial wind speeds. In addition, the radar observations were made at night when there could be no visual observations of the circulations. As a result, for the purposes of this study, a threshold is established for defining a circulation: any temporally continuous (in particular, more than two consecutive radar scans at a level) vortex signature of diameter 1–10 km with a minimum difference between inbound and outbound velocities of 20 m s−1. Once a circulation met the above criteria, it was analyzed until it completely dissipated, even if it occasionally fell below the velocity difference threshold. The term “circulation” is used to account for the ambiguity in defining a mesocyclone through use of single-Doppler radar data, as there are shear features in the Doppler wind field that do not meet the above criteria that may have been mesocyclones, and vice versa.

a. Low-level circulation observations

Volume scans were completed by SR-1 at ∼1 min intervals; however, for the purposes of brevity, reflectivity and radial-velocity scans are shown here approximately every 5–10 min (Fig. 5). The first 0.5° elevation-angle scans of the entire supercell were made by SR-1 at 0208 UTC 16 May 2003. At this elevation angle, the data were sampled at heights ranging from approximately 200–500 m above ground level (AGL), depending on the range of the echo from the radar. Rather than a single reflectivity appendage or “hook echo” (hereinafter “hook”) typically associated with a supercell, the storm at 0208 UTC actually had two reflectivity appendages on its southwest side, and two cyclonic shear signatures coincident with the reflectivity appendages (not shown). By 0220 UTC, the easternmost hook had elongated, grown in diameter, and had at least three individual “notches” or smaller hooks embedded within it (Fig. 5a). One of these embedded notches was associated with the first of at least five strong low-level circulations the storm would exhibit during its lifetime (circulation L1) located to the rear of the strongest inbound and outbound radial velocities (Fig. 5a). The high receding velocities likely resulted from outflow from the forward-flank downdraft (FFD), while the high approaching velocities likely indicated outflow from the RFD.

In reflectivity scans at 0229 UTC (Fig. 5b) it is seen that the elongated hook had evolved into two distinctive hooks. Circulation L1, mentioned above, was located at the very tip of the rear hook, along with a much weaker circulation located east of the original. Another broader circulation (circulation L2) formed northwest of a region of rear-flank outflow, and persisted for over 15 min. It is seen that just a few minutes later, at 0233 UTC (Fig. 5c), circulation L1 had diminished almost entirely, and the secondary hook that circulation L1 was associated with had dissipated. At the same time, circulation L2 broadened and strengthened. Also of note was the very tip of the primary hook echo; there was a reverse-curving hook echo attached to the tip of the original hook echo. The corresponding velocity field was indicative of a weak anticyclonic shear region (Fig. 5c; receding yellow and approaching green velocities south of L2).

At 0238 UTC (Fig. 5d), circulation L2 had moved off rapidly to the northwest, while a new circulation, circulation L3, formed. The development of circulation L3 was coincident with another cycle of hook evolution. The main hook echo approached the forward flank of the storm while the tip of the hook extended off to the southeast and developed into the next hook. Both circulations were located at the tips of their respective hooks, about 8 km away from each other. In addition, the reverse-curving hook and corresponding anticyclonic shear regions seen in prior scans had disappeared. By 0245 UTC (Fig. 5e), circulation L2 had weakened considerably and circulation L3 strengthened as it moved toward the north-northwest. In fact, circulations L2 and L3 rotated around each other in a counterclockwise direction and combined to form one circulation, still referred to here as L3, because L3 was much larger and contained stronger radial shear than L2.

Over the next 15 min (not shown), circulation L3 was the main low-level circulation. Circulation L3 maintained its intensity until close to 0300 UTC, after which L3 (and the hook with which L3 coincided) weakened quickly. At roughly the same time, a new circulation, L4, developed (Fig. 5f). Circulation L4 was an isolated circulation, and was associated with one larger hook, unlike the earlier circulations. Circulation L4 marked the end of the rapid formation and dissipation of new circulations during the time mobile radar data were collected. Perhaps not coincidentally, ∼0310 UTC was also the time when the most severe damage from the storm, possibly from a tornado, was being inflicted near Lela. The origin of L5 can be seen as the large area of convergence located north of L4. The area of convergence strengthened as an area of approaching velocities (also located north of L4) increased in magnitude (Fig. 5f). Circulation L4 grew considerably in diameter after merging with a short-lived, cyclonic shear feature (not shown), but diminished by 0309 UTC (not shown). Circulation L5 formed soon after the dissipation of L4 in the area of convergence mentioned previously (Fig. 5g). By 0315 UTC (Fig. 5h), L5 had grown to a relatively large size and, as with L4, was located within one well-defined hook. By 0331 UTC, contamination problems within the 0.5° elevation-angle radar data had developed, and it was no longer possible to track circulation L5 accurately. While L5 still existed at 0329 UTC, it is difficult to conclude definitively whether cyclic mesocyclogenesis ceased with the formation of L5 because the storm had not dissipated prior to the cessation of data collection.

b. Midlevel circulation observations

Scans taken at a 3.5° elevation angle correspond to heights ranging from approximately 1.5 to 3.5 km AGL, depending on the range from the radar. For the purposes of this study, circulations identified in the 3.5° elevation-angle scans were referred to as midlevel circulations (the evolution of the main circulation features were nearly identical in the 3.5° and 5.0° elevation-angle scans). In the first 3.5° elevation-angle scans, taken at 0209 UTC by SR-1, there was a broad, cyclonic circulation and, as in the lowest-level scans, there were strong negative radial velocities near the rear of the storm (not shown). There were no recognizable smaller-scale circulations until about 0216 UTC (not shown). At 0220 UTC (Fig. 6a), there was one relatively broad circulation, circulation M1, with a wider, longer hook than that seen in earlier scans. Also, at the edge of the hook in reflectivity was a small reverse-curving notch (Fig. 6a). The notch corresponded to a weak anticyclonic shear region in the velocity as seen in the low-level scans. However, the anticyclonic shear region in the low-level scans did not appear for another 10 min; the signature at midlevels disappeared in less than 5 min.

At 0226 UTC (not shown), circulation M1 expanded into a larger region of cyclonic shear from which two circulations were evident. It is difficult to determine which circulation was a continuation of M1, and which was a new circulation. The western circulation was renamed M1 and the eastern circulation was treated as a new circulation (M2) because the western circulation continued to exhibit vertical continuity with L1. By 0229 UTC, the supercell rapidly evolved to have two distinct hooks and three circulations (Fig. 6b), including M3, which formed east of M1. At 0234 UTC, circulation M1 had begun to weaken (Fig. 6c). The reflectivity notch, which coincided with circulation M2, had elongated and developed into another rear hook while circulations M2 and M3 maintained their intensity. By 0240 UTC, another new circulation, circulation M4, had formed southeast of the remnants of M3, while circulations M1 and M2 had dissipated almost completely (Fig. 6d). Also during this time period, circulations M2 and M3 rotated around each other in a counterclockwise direction prior to the dissipation of M2. In addition, circulation M4 had developed and strengthened, while circulation M5 formed adjacent (to the east) to circulation M4.

At 0243 UTC (Fig. 6e), another circulation, M6, had formed. It is difficult to determine whether the circulation was actually a regeneration of circulation M2 and M3, or a new, unique circulation. Circulations M2 and M3 had weakened and were now coincident with a large area of convergence that had preceded circulation M6. At this time, circulations M4 and M6 also rotated around each other in a counterclockwise direction for a brief time until circulation M4 and M6 had combined to form one circulation. The combined circulation is still referred to here as M6 because M6 was larger and had much stronger radial shear than M4 prior to the circulation interaction. The apparent rotation was evident during the same time period at which the low-level circulations also rotated around each other; these rotating circulations will be discussed further in subsequent sections. At 0255 UTC, circulation M6 began to retreat rearward relative to the storm and eventually dissipated (not shown). At 0306 UTC, circulation M5 was the only detectable midlevel circulation (Fig. 6f). By 0310 UTC, M5 had dissipated, while, to the north of M5, M7 developed quickly into a large and strong circulation (Fig. 6g). At 0316 UTC, it can be seen that M7 was the only midlevel circulation (Fig. 6h), and was the final unique circulation identified at midlevels during data collection by SR-1.

c. Analysis

1) Circulation propagation

Previous observations of cyclic supercells have focused on the movement of the individual mesocyclones or tornadoes in relation to the parent storm. This dataset has the advantage of having high spatial resolution, single-Doppler radar data of a cyclic supercell at low elevation angles. Such high resolution provides increased detail regarding individual circulations, as shown in sections 3a and 3b. Of interest, then, is how the individual circulations moved in relation to the parent storm and each other, and how the movement of each circulation correlated with its strength and duration. Also of interest is how the final two circulations (in terms of the period of mobile radar data collection) both at low levels (L4 and L5) and midlevels (M5 and M7) differed from the earlier ones. It is likely that one of the last two circulations (most likely L5/M7) was associated with a tornado, while none of the earlier circulations were associated with tornadoes. The last two circulations also were long-lived and isolated, while many of the earlier ones were not, an indication that cyclic mesocyclogenesis had ended, or was slowing. In addition, the behavior of certain circulations, including the rotation of two circulations around each other, warrants deeper analysis.

The three early main low-level circulations (i.e., L1, L2, and L3) all moved in a similar ground-relative direction (Fig. 7a). Early in their respective lifetimes, all three circulations moved generally toward the north-northeast. Other similarities between the circulations included a characteristic turn toward the left halfway through their lifetimes. Circulations L1 and L2 continued to move off to the north as they weakened and dissipated, while circulation L3 regained a northeastward path as it decayed. The two late (i.e., L4 and L5) low-level circulations both moved more toward the northeast, though L5 did turn toward the north at the end of data collection.

All of the midlevel circulations moved with westerly (from the west) and southerly components throughout most of their respective lifetimes (Fig. 7b). The circulations can be categorized as having two different propagation patterns. Circulations M1, M3, and M4 all moved toward the northeast early in their lifetimes, and then turned toward the north. Circulations M2, M5, M6, and M7 also moved toward the northeast, but, with the exception of M7, they never turned toward the north as the other circulations did. Only circulations M3 and M4 moved westward; at both times this movement occurred just prior to the circulation decay. Circulation M7 took a sharp turn toward the north just as the circulation with which it was collocated, L5, did.

To gain a better understanding of how the circulations moved with respect to the motion of the storm, estimated storm-relative, normalized locations were plotted for the five main low-level circulations (Fig. 8a). To do this, an estimated storm motion of ∼10 m s−1 from 245° was determined independently in two ways: 1) tracking the sequence of locations of the core of the storm from SR-1 scans and 2) following a long-lived bounded weak echo region visible in 2.3° elevation-angle scans from the NWS Weather Surveillance Radar-1988 Doppler (WSR-88D) in Amarillo. Both methods provided almost identical results for storm motion. Data averaging over 5–8-min periods places the focus on overall trends in the storm-relative locations. The most obvious finding from the storm-relative locations is the rearward movement of the three early low-level circulations. Circulations L1, L2, and L3 all ended up approximately 6–10 km west of where they began relative to the storm and all three circulations dissipated west of where they developed. Circulations L4 and L5 also moved rearward relative to the storm, but the 2–3-km rearward movement was much less than that of the early low-level circulations. Circulations L4 and L5 were also the only two circulations to move southward relative to the storm. It is thus inferred that the updrafts associated with these circulations were propagating well to the right of the mean wind and ingesting, based on the vertical shear present in the environment (Fig. 3), considerably more streamwise vorticity.

The plot of storm-relative normalized locations for the seven midlevel circulations (Fig. 8b) exhibits the two different propagation patterns mentioned previously. Circulations M1, M3, and M4 all dissipated several kilometers northwest (in a storm-relative sense) of where they formed while M2, M5, M6, and M7 moved at a speed and direction more similar to that of the parent storm. The observations of L1, L2, L3, M1, M3, and M4 agree with previous studies in which it had been found that the vortices that moved rearward in a storm-relative sense eventually dissipated (Burgess et al.1982; Dowell and Bluestein 2002a; Beck et al. 2006). The final, long-lived vortices (M5 and M7) moved, at most, 2 km rearward into the storm. However, M2 and M6, which were shorter lived and not associated with tornadoes, moved in a manner similar to that of the late circulations, contrary to previous observations. In addition, while there were two different propagation patterns for the midlevel circulations, neither of the patterns included a significant storm-relative southward component of motion as in the nonoccluding cyclic mesocyclogenesis described in Adlerman and Drogemeier (2005).

2) Circulation statistics

Aspects of the individual circulations deserving of more detailed study include mean values of circulation diameter, radial wind shear, and propagation speed and direction (Table 1). Mean circulation diameters at both levels were approximately 1–4 km, though the circulation diameters at individual times were often much broader. The mean circulation diameters here are generally smaller than the approximately 4–5 km mean-diameter mesocyclones observed by Burgess et al. (1982). The magnitudes of the radial wind shear (approximately 10–40 × 10−3 s−1) are much higher than the mean shear values in Burgess et al. (1982), which were approximately 9–10 × 10−3 s−1. The higher shear values found in this study are likely related to the smaller observed diameters. In addition, the different results from Burgess et al. (1982) may be a result of the much smaller sample size and/or the higher-resolution radar data used in this study. The highest radial-velocity wind speeds, approximately 55 and 48 m s−1, were located within circulations L5 and M7, respectively.

To better illustrate the strength of the individual circulations, time series of the maximum radial wind shear in the circulations were plotted (Fig. 9). The maximum radial wind shear was calculated by taking the difference between the maximum positive and negative radial-velocity values in the circulation and dividing by the distance between the two locations. Most of the shears for both low- (Fig. 9a) and midlevel (Figs. 9b,c) circulations were between 0.01 and 0.06 s−1. It is difficult to discern any relationship between the shear values of the circulations and the life cycles of the circulations. Some circulations (e.g., L1 and M5) dissipated following a downward trend in maximum radial wind shear, while others (e.g., M3 and M6) dissipated following an upward trend in maximum radial wind shear. The strength of the maximum radial wind shear was a poor indicator of impending circulation dissipation.

Time series of circulation diameters also were plotted (Fig. 10). Diameters of the circulations were calculated as the distance between the locations of the maximum outbound and inbound radial velocities. The above method of determining circulation diameter was subjective and based on the availability only of single-Doppler radar data. Possible biases may have been introduced into the calculations of circulation diameters in cases where clear radial-velocity maxima were not present. The circulation diameters at both levels ranged from approximately 0.5–6 km and the diameters of some circulations changed dramatically during their lifetimes. For example, the diameter of L2 (Fig. 10a) ranged from approximately 1–6 km during its lifetime. At midlevels, the late circulations (Fig. 10c) had wider diameters than the early circulations (Figs. 10b,c). Most of the circulations dissipated following an overall downward trend in diameter (e.g., L3, L4, L5, M2, and M6); however, a downward trend in diameter did not necessarily portend circulation dissipation (e.g., M1 and M5). Those circulations that did not dissipate during a downward trend in diameter (e.g., L2, M4, and M5) underwent a steep drop off in diameter within 10 min of dissipation. The decrease in circulation diameter prior to dissipation at both low and midlevels is consistent with similar statistics calculated by Burgess et al. (1982).

Many of the low-level circulations exhibited vertical continuity with midlevel circulations (Table 2). In general, circulations formed first at midlevels, though the formation at midlevels preceded formation at low levels by just several minutes. Likewise, circulations tended to dissipate first at low levels, again only preceding dissipation of midlevel circulations by short periods of time.

Circulation lifetimes (i.e., the time period when the circulations met the aforementioned criteria) were approximately 10–30 min (Table 2), intervals shorter than the mesocyclone lifetimes observed by Burgess et al. (1982). Previous studies have used a methodology incorporating vertical velocity analyses in order to determine cycling frequency, so an exact comparison of cycling frequencies cannot be made here. However, the relatively short lifetimes of the circulations observed in this study imply a cycling frequency more consistent with that of Beck et al. (2006) than that found by Burgess et al. (1982) and Adlerman et al. (1999). Circulations L5 and M7 had not yet dissipated when SR-1 data collection ceased, so a complete comparison of the lifetimes of the early and late circulations also cannot be made.

3) Additional observations

There were some additional, unique observations of the circulations and their corresponding hooks. As mentioned previously, circulations M4 and M6 rotated around each other in a counterclockwise direction between 0243 and 0250 UTC; the rotation ended as M4 dissipated and M6 strengthened (Fig. 11). Likewise, during the same time period and in the same location, circulations L2 and L3 evolved in the same way, while L2 dissipated (not shown). In fact, as the circulations rotated around each other at both low and midlevels, they also moved closer to each other (Figs. 11a–c), eventually coming so close to each other that L2 and M4 actually combined with L3 and M6, respectively, to form one circulation in each case (Fig. 11d). Circulations M2 and M3 may also have rotated around each other for a brief period of time (not shown), though the rotation was less obvious and occurred over a shorter period of time. The cyclonic rotation among, and simultaneous decrease in distance between, the circulations is an indication of the “Fujiwhara effect” or “binary interaction,” often used to describe the motion of a pair of tropical systems located within a certain distance of each other (Fujiwhara 1931; Dong and Neuman 1983). The Fujiwhara effect has been identified previously as the cause for vortex interaction in dust devils (Bluestein et al. 2004), in misocyclones within various boundaries (Marquis et al. 2007), and among vorticity extrema within a convective boundary layer (Markowski and Hannon 2006). However, it is believed that this is the first time that vortex interaction associated with the Fujiwhara effect has been documented between circulations on the mesocyclone scale within a supercell.

The evolution of the hook echoes, particularly during the formation of circulation L3, agree qualitatively with the conceptual model proposed by Beck et al. (2006). In that model, the existing hook echo shifts into the forward flank of the storm. Then, an area of convergence/confluence associated with a newly formed mesocyclone creates a deformation zone that results in advection of hydrometeors around the southern portion of the new mesocyclone. Between 0229 and 0238 UTC (Figs. 5b–d), the main hook merged with the forward flank while a new hook formed concurrently with circulation L3. In this case, it is not possible to confirm any definitive causal relationship established between the hook echo regeneration and the convergence/deformation fields because dual-Doppler analyses were not available at the time of the hook echo regeneration.

4. Discussion

The Shamrock supercell consisted of at least five low-level and seven midlevel circulations of varying diameters, life spans, and radial wind speeds. A key aspect of the Shamrock supercell that has not been addressed is the Lela tornado. A complicating factor in addressing specifics regarding the tornado is that it occurred at night, so the only information available, outside of radar data, is the NWS damage survey of the reported tornado. While a complete damage analysis comparison with radar data is beyond the scope of this paper, it is likely that the tornado was associated with the late circulations observed in this dataset. More specifically, circulations L5/M7, which exhibited vertical continuity (Table 2), were the only discernible low-level/midlevel circulations present during much of the Lela tornado’s estimated lifetime. In addition the estimated path of the Lela tornado is matched best by the paths of L5/M7. Not coincidentally, L5/M7 contained the strongest radial velocities (Table 1) out of all of the circulations studied. It is also possible that the tornado was associated first with L4/M5 and persisted through the development of L5/M7. Nonetheless, the key point is that a tornado was present only after the rapid cycling had slowed, particularly at midlevels. As a result, a key question to address is “Why were the late circulations tornadic and the early circulations nontornadic?”

A number of differences between the early and late circulations already have been made. To summarize, the late circulations, at both levels, did not move significantly rearward relative to the storm, while the early circulations generally did. The late circulations, specifically L5 and M7, also had the largest diameter and the highest radial wind speeds. As mentioned previously, Dowell and Bluestein (2002b) hypothesized that the strength of the storm outflow at low levels was an important factor in determining whether a tornado would move rearward and dissipate. In their case, storm outflow was initially weak, and strong updraft-relative flow at low levels advected each of the first two tornadoes rearward. The rearward advection of the tornadoes brought them farther away from the main updraft region where there were significant amounts of vertical vorticity due to tilting of horizontal vorticity and stretching of preexisting vertical vorticity. As the outflow strengthened, the updraft-relative flow at low levels in the tornadic region weakened, making it more likely that the tornado would remain near the main updraft and rear-flank gust front, areas containing large amounts of vertical vorticity production.

During the time of SR-1 and UMass X-Pol data collection, the behavior of the Shamrock supercell was qualitatively consistent with the hypothesis of cyclic vortex formation as proposed by Dowell and Bluestein (2002b). In this case, the hypothesis applies to vortices on both the mesocyclone scale and the tornado scale. During the entirety of data collection by both Doppler radars, flow from the forward flank of the storm, which enters the rear flank of the storm (hereafter “forward-flank inflow”) was extremely strong. For example, during the time that the storm likely was undergoing rapid cyclic mesocyclogenesis (approximately 0210–0250 UTC), a consistent area of strong receding velocities can be identified in the SR-1 data northeast of the area of rear-flank outflow (e.g., Figs. 5a–e). At times, the maximum radial wind speeds were over 40 m s−1. After rapid cyclic mesocyclogenesis appeared to slow (after 0250 UTC), the forward-flank inflow did not weaken. In fact, radial wind speeds of almost 50 m s−1 were obtained by the UMass X-Pol at 0304 UTC (Fig. 12; dark reds northeast of the vortex signature). Unfortunately, the character of the inflow winds east and southeast of the circulations could not be determined owing to a lack of hydrometeors in that region.

Though the observed forward-flank inflow in the storm consistently was strong, the strength of the rear-flank outflow changed considerably as the storm evolved. Between 0210 and 0250 UTC, low-level rear-flank outflow in the storm, while occasionally including some strong approaching radial velocities, was generally confined to a small area (Figs. 5a–e). However, between 0250 and 0300 UTC, the character of the winds in the rear flank of the storm changed. This change can be seen in SR-1 scans at that time (Fig. 13). At 0251 UTC (Fig. 13a), the rear flank, outside of L3 was dominated by generally weak radial velocities. By 0254 UTC (Fig. 13b), a curved line segment of stronger approaching velocities developed east and northeast of L3. At 0258 UTC (Fig. 13c), the line segment had grown into a more circular area of strong approaching velocities located in between the dissipating L3 and the developing L4. By 0300 UTC (Fig. 13d), there was a large area of approaching radial velocities ∼20 m s−1 and greater located to the west and northwest of the strengthening L4. In addition, a crude dual-Doppler analysis performed at ∼0305 UTC (not shown) exhibited the qualitative features of a strong low-level mesocyclone (L4) and strong forward-flank inflow. In addition, a comparison between analyses prior to 0250 UTC with those from after 0300 UTC showed a large increase in rear-flank outflow winds, consistent with both sets of single-Doppler radar data.

It is likely that the increase in the outflow winds played a part in the slowing of cyclic vortex formation after ∼0250 UTC and the increase in the size and strength of L4/M5 and L5/M7. The Shamrock storm had such strong forward-flank inflow, that, in the absence of the strong rear-flank outflow, vortices were quickly advected rearward as seen between 0210 and 0250 UTC (Figs. 5a–e). These early circulations were not located near the rear-flank gust front (and likely, the region containing the main updraft) for long periods of time, and, therefore, as they moved rearward, they were removed from areas likely rich in vertical vorticity production, and dissipated. It is speculated that after 0250 UTC, when the rear-flank outflow increased, a balance between rear-flank outflow and forward-flank inflow developed, which weakened the updraft-relative mean flow the circulations were embedded in. Therefore, vortices that formed in the above environment were able to stay close to the rear-flank gust front, in areas that usually contain strong vertical vorticity production. While the specific processes that led to the genesis of the Lela tornado are not addressed here, it is likely that the maintenance of the late low-level circulations at a location near the rear-flank gust front in areas rich in vorticity production allowed the low-level circulations to become tornadic. In addition, Dowell and Bluestein (2002b) speculated that the increased storm outflow in their case came from other convective cells along the supercell’s rear flank, whereas in the current case, the increase in storm outflow seems to be internal to the supercell.

A lingering discrepancy in the scenario outlined above is the propagation of circulations M2 and M6, the early circulations that did not move rearward. There are two possible explanations for the different propagation characteristics of M2 and M6. First, M2 (Fig. 6b) and M6 (Fig. 6e) were the only two circulations that had their origin a significant distance away from the rear-flank gust front. The mechanisms suggested as being responsible for the rearward movement, relatively weak outflow and strong forward-flank inflow, may have had less of an effect on the movement of the circulations if they already were located several kilometers away from the rear-flank gust front. The updraft-relative mean flow in these regions may have been much weaker. Another possibility is that vortex interaction played a part in the movement of M2 and M6. As mentioned previously, M6 interacted with M4 (Fig. 11), and M2 briefly rotated around M3 (not shown). It is possible that the Fujiwhara effect played a part in the relatively abnormal storm-relative forward movement of the circulations. Regardless, the general hypotheses regarding rearward movement of mesocyclones within cyclic supercells should be altered to specify that rearward movement may not occur in storm-scale circulations that do not develop in the “traditional” formation region near the rear-flank gust front.

5. Conclusions

At least five low-level, and seven midlevel, circulations were sampled in a southern plains supercell by SR-1 while data were collected for ∼90 min on 15 May 2003. The increased temporal resolution of SR-1 allowed for in-depth study of the evolution of the circulations over short time scales (∼1 min) that would not have been possible through use of WSR-88D or airborne radar data [e.g., the Electra Doppler Radar (ELDORA)]. The increased spatial resolution of SR-1 (owing to relatively close range and short gate length) allowed for identification of circulations perhaps too small for WSR-88D identification at typical ranges of ∼100 km. The combination of increased spatial and temporal resolution revealed smaller and more rapidly evolving circulations than documented in previous studies.

The 0.5° and 3.5° elevation-angle scans were used as proxies for the low and midlevels of the storm. The early circulations, those that formed before ∼0250 UTC, tended to have a large rearward component of motion relative to the movement of the storm if they formed near the rear-flank gust front. The late circulations, which formed after 0250 UTC, did not have such dramatic rearward storm-relative movement. In addition, the late circulations likely were associated with a tornado that swept through Lela, Texas, and caused heavy damage. Circulations were approximately 1–6 km in diameter and had lifetimes of approximately 10–30 min. Most circulations dissipated following decreases in circulation diameter, while the maximum radial wind shear was a poor predictor of circulation dissipation. Though it is difficult to pinpoint the exact location of the main rear-flank gust front, it appeared that the low-level circulations that dissipated became completely separated from the rear-flank outflow region.

The evolution of hook echo formation qualitatively followed the conceptual model proposed by Beck et al. (2006). The circulations at midlevels tended to be of the same mean diameter and mean maximum radial wind shear as the low-level circulations. However, the later circulations generally had larger diameters and higher maximum radial winds than the earliest circulations. In most cases, mid- and low-level circulations were connected and tended to develop and dissipate at roughly the same time. There was at least one instance in which two circulations cyclonically rotated around each other and simultaneously moved toward each other until the circulations combined.

To address why the late circulations were tornadic, and the earlier circulations were not, single-Doppler radial velocities from SR-1 and the UMass X-Pol radar were examined. Weak rear-flank outflow coupled with very strong forward-flank inflow prior to 0250 UTC likely resulted in strong rearward advection of developing vortices. Increasing approaching single-Doppler radial velocities after 0250 UTC support the idea that increasing outflow aided in keeping circulations L4/M5 and L5/M7 in areas rich in vorticity production. The strong outflow balanced the forward-flank inflow, allowing the late circulations to follow the path of the storm more closely. This study looked mainly at single-Doppler radar data, so causal mechanisms for outflow development could not be adequately addressed.

Acknowledgments

The authors thank Michael Biggerstaff and Alan Shapiro, both of whom reviewed early drafts of this work within the first author’s M.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 (D. Dowell).

REFERENCES

  • Adlerman, E. J., and K. K. Drogemeier, 2005: The dependence of numerically simulated cyclic mesocyclogenesis upon environmental vertical wind shear. Mon. Wea. Rev., 133 , 35953623.

    • Search Google Scholar
    • Export Citation
  • Adlerman, E. J., K. K. Drogemeier, and R. Davies-Jones, 1999: A numerical simulation of cyclic mesocyclogenesis. J. Atmos. Sci., 56 , 20452069.

    • Search Google Scholar
    • Export Citation
  • Beck, J. R., J. L. Schroeder, and J. M. Wurman, 2006: High-resolution dual-Doppler analyses of the 29 May 2001 Kress, Texas, cyclic supercell. Mon. Wea. Rev., 134 , 31253148.

    • Search Google Scholar
    • Export Citation
  • 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 , 12631274.

    • Search Google Scholar
    • Export Citation
  • Bluestein, H. B., C. C. Weiss, and A. L. Pazmany, 2004: Doppler radar observations of dust devils in Texas. Mon. Wea. Rev., 132 , 209224.

    • Search Google Scholar
    • Export Citation
  • Bluestein, H. B., M. M. French, R. L. Tanamachi, S. Frasier, K. Hardwick, F. Junyent, and A. L. Pazmany, 2007: Close-range observations of tornadoes in supercells made with a dual-polarization, X-band, mobile, Doppler radar. Mon. Wea. Rev., 135 , 15221543.

    • Search Google Scholar
    • Export Citation
  • Burgess, D. W., V. T. Wood, and R. A. Brown, 1982: Mesocyclone evolution statistics. Preprints, 12th Conf. on Severe Local Storms, San Antonio, TX, Amer. Meteor. Soc., 422–424.

  • Darkow, G. L., and J. C. Roos, 1970: Multiple tornado producing thunderstorms and their apparent cyclic variations in intensity. Preprints, 14th Conf. on Radar Meteorology, Tucson, AZ, Amer. Meteor. Soc., 305–308.

  • Dong, K., and C. J. Neuman, 1983: On the relative motion of binary tropical cyclones. Mon. Wea. Rev., 111 , 945953.

  • Doswell III, C. A., and D. W. Burgess, 1993: Tornadoes and tornadic storms: A review of conceptual models. The Tornado: Its Structure, Dynamics, Prediction, and Hazards, Geophys. Monogr., Vol. 79, Amer. Geophys. Union, 161–172.

    • Search Google Scholar
    • Export Citation
  • Dowell, D. C., and H. B. Bluestein, 2002a: The 8 June 1995 McLean, Texas, storm. Part I: Observations of cyclic tornadogenesis. Mon. Wea. Rev., 130 , 26262648.

    • Search Google Scholar
    • Export Citation
  • Dowell, D. C., and H. B. Bluestein, 2002b: The 8 June 1995 McLean, Texas, storm. Part II: Cyclic tornado formation, maintenance, and dissipation. Mon. Wea. Rev., 130 , 26492670.

    • Search Google Scholar
    • Export Citation
  • Fujita, T. T., 1975: New evidence from April 3–4, 1974 tornadoes. Preprints, Ninth Conf. on Severe Local Storms, Norman, OK, Amer. Meteor. Soc., 248–255.

  • Fujita, T. T., and A. D. Pearson, 1973: Results of FPP classification of 1971 and 1972 tornadoes. Preprints, Eighth Conf. on Severe Local Storms, Denver, CO, Amer. Meteor. Soc., 142–145.

  • Fujita, T. T., D. L. Bradbury, and C. F. Van Thullenar, 1970: Palm Sunday tornadoes of April 11, 1965. Mon. Wea. Rev., 98 , 2969.

  • Fujiwhara, S., 1931: Short note on the behavior of two vortices. Proc. Phys. Math. Soc. Japan, 13 , 106110.

  • Glickman, T. S., 2000: Glossary of Meteorology. 2nd ed. Amer. Meteor. Soc., 855 pp.

  • Johnson, K. W., P. S. Ray, B. C. Johnson, and R. P. Davies-Jones, 1987: Observations related to the rotational dynamics of the 20 May 1977 tornadic storms. Mon. Wea. Rev., 115 , 24632478.

    • Search Google Scholar
    • Export Citation
  • Kramar, M. R., H. B. Bluestein, A. L. Pazmany, and J. D. Tuttle, 2005: The “owl horn” radar signature in developing southern plains supercells. Mon. Wea. Rev., 133 , 26082634.

    • Search Google Scholar
    • Export Citation
  • Lemon, L. R., and C. A. Doswell III, 1979: Severe thunderstorm evolution and mesocyclone structure as related to tornadogenesis. Mon. Wea. Rev., 107 , 11841197.

    • Search Google Scholar
    • Export Citation
  • Markowski, P., and C. Hannon, 2006: Multiple-Doppler radar observations of the evolution of vorticity extrema in a convective boundary layer. Mon. Wea. Rev., 134 , 355374.

    • Search Google Scholar
    • Export Citation
  • Marquis, J. N., Y. P. Richardson, and J. M. Wurman, 2007: Kinematic observations of misocyclones along boundaries during IHOP. Mon. Wea. Rev., 135 , 17491768.

    • Search Google Scholar
    • Export Citation
  • Pazmany, A. L., F. J. Lopez, H. B. Bluestein, and M. R. Kramar, 2003: Quantitative rain measurements with a mobile, X-band, polarimetric Doppler radar. Preprints, 31st Int. Conf. on Radar Meteorology, Seattle, WA, Amer. Meteor. Soc., 858–859.

  • Rasmussen, E. N., and J. M. Straka, 2007: Evolution of low-level angular momentum in the 2 June 1995 Dimmitt, Texas, tornado cyclone. J. Atmos. Sci., 64 , 13651378.

    • Search Google Scholar
    • Export Citation
  • Rasmussen, E. N., J. M. Straka, R. Davies-Jones, C. A. Doswell, F. H. Carr, M. D. Eilts, and D. R. MacGorman, 1994: Verification of the Origins of Rotation in Tornadoes Experiment: VORTEX. Bull. Amer. Meteor. Soc., 75 , 9951006.

    • Search Google Scholar
    • Export Citation
  • Seliga, T. A., and V. N. Bringi, 1976: Potential use of radar differential reflectivity measurements at orthogonal polarizations for measuring precipitation. J. Appl. Meteor., 15 , 6975.

    • Search Google Scholar
    • Export Citation
  • Weisman, M. L., and J. B. Klemp, 1982: The dependence of numerically simulated convective storms on vertical wind shear and buoyancy. Mon. Wea. Rev., 110 , 504520.

    • Search Google Scholar
    • Export Citation

Fig. 1.
Fig. 1.

Surface map at 0000 UTC 16 May 2003 showing the conditions in the Texas Panhandle as the Shamrock supercell was initiating. For individual surface plots, the upper left is the temperature (°C), the lower left is the dewpoint temperature (°C), the upper right is the surface pressure minus leading 9 or 10 (hPa), the middle indicates any cloud cover present, and the wind barbs indicate wind direction with each full barb representing 10 kt (5 m s−1) and each half barb 5 kt (2.5 m s−1). Approximate dryline (warm front) location shown as white (black) scalloped line. The circles show the location where the two cells that would become the Shamrock supercell initiated on radar. The triangle indicates where tornadogenesis is believed to have occurred near Shamrock, TX, later on at 0300 UTC.

Citation: Monthly Weather Review 136, 12; 10.1175/2008MWR2407.1

Fig. 2.
Fig. 2.

Radar reflectivity at 0.5° elevation angle from the NWS WSR-88D in Amarillo, TX, at 0243 UTC 16 May 2003. The white circle encloses the Shamrock convective storm and the convective line extending from north to south about 50 km west of the radar marks the approximate location of the dryline; range rings are shown every 50 km.

Citation: Monthly Weather Review 136, 12; 10.1175/2008MWR2407.1

Fig. 3.
Fig. 3.

Skew T–logp diagram at Amarillo at 0000 UTC 16 May 2003. Line on the right (left) indicates temperature (dewpoint temperature) (°C). The inset in the bottom left of the figure is the hodograph corresponding to the sounding shown with wind speed values (m s−1) and heights (km AGL). (From the Plymouth State archive)

Citation: Monthly Weather Review 136, 12; 10.1175/2008MWR2407.1

Fig. 4.
Fig. 4.

Diagram showing the NWS estimated tornado paths (black lines) based on a damage survey and the approximate location of SR-1 and the UMass X-Pol radar (black squares). The major highways (blue lines) and towns (gray circles) in the area also are depicted. County lines are shown in red.

Citation: Monthly Weather Review 136, 12; 10.1175/2008MWR2407.1

Fig. 5.
Fig. 5.

SR-1 (left) reflectivity (dBZ) and (right) radial velocity (m s−1) at 0.5° elevation angle collected at (a) 0220, (b) 0229, (c) 0233, (d) 0238, (e) 0245, (f) 0305, (g) 0311, and (h) 0315 UTC 16 May 2003. Black arrows indicate hook echoes in the reflectivity field and the blue arrow indicates a reverse-curving hook. White circles enclose, and numbers indicate, the specific circulation referenced in the text. The data were located at heights ranging from approximately 200 to 500 m AGL depending on the range of the echo from the radar. The radar view angle differs for each scan but ranges from (a) 255°–305° to (h) 290°–5°. Range rings are every 5 km.

Citation: Monthly Weather Review 136, 12; 10.1175/2008MWR2407.1

Fig. 5.
Fig. 5.

(Continued)

Citation: Monthly Weather Review 136, 12; 10.1175/2008MWR2407.1

Fig. 6.
Fig. 6.

SR-1 (left) reflectivity and (right) radial velocity at 3.5° elevation angle collected at (a) 0220, (b) 0229, (c) 0234, (d) 0240, (e) 0243, (f) 0306, (g) 0310, and (h) 0316 UTC 16 May 2003. The data were located at heights ranging from approximately 1.5 to 3.5 km AGL depending on the range of the echo from the radar. The radar view angle differs for each scan but ranges from (a) 255°–305° to (h) 290°–20°. Everything else is as in Fig. 5.

Citation: Monthly Weather Review 136, 12; 10.1175/2008MWR2407.1

Fig. 6.
Fig. 6.

(Continued)

Citation: Monthly Weather Review 136, 12; 10.1175/2008MWR2407.1

Fig. 7.
Fig. 7.

Plots of the approximate ground-relative locations of (a) low-level and (b) midlevel circulations, which correspond to the 0.5° and 3.5° elevation-angle scans, respectively, from SR-1 for the 15 May 2003 Shamrock supercell. The black lines indicate the path each circulation took as it dissipated. The list in the upper left provides the formation and dissipation times for each circulation in UTC time. The time interval between data points (not shown) is ∼75 s. Increasing positive values of the ordinate are toward due north. The origin of the graph is the location of SR-1.

Citation: Monthly Weather Review 136, 12; 10.1175/2008MWR2407.1

Fig. 8.
Fig. 8.

Plots of the approximate normalized storm-relative locations of (a) low-level and (b) midlevel circulations, which correspond to the 0.5° and 3.5° elevation-angle scans, respectively, from SR-1 for the 15 May 2003 Shamrock supercell. All of the circulations start at the origin and the white points indicate the normalized locations in which each circulation dissipated. The time interval between successive marks varies from approximately 5 to 8 min. Increasing positive values of the ordinate are toward due north. The origin of the graph is the normalized approximate location where each of the individual circulations developed.

Citation: Monthly Weather Review 136, 12; 10.1175/2008MWR2407.1

Fig. 9.
Fig. 9.

The approximate maximum radial wind shear as a function of time for (a) low-level and (b), (c) midlevel circulations, which correspond to the 0.5° and 3.5° elevation-angle scans, respectively, from SR-1 for the 15 May 2003 Shamrock supercell.

Citation: Monthly Weather Review 136, 12; 10.1175/2008MWR2407.1

Fig. 10.
Fig. 10.

The approximate diameter as a function of time for (a) low-level and (b), (c) midlevel circulations, which correspond to the 0.5° and 3.5° elevation-angle scans, respectively, from SR-1 for the 15 May 2003 Shamrock supercell. The diameter for circulation L1 is omitted because the maximum inbound and outbound radial velocities were located in adjacent gates, so the actual diameter may have been smaller than the width of the range gates. The diameter of a circulation was not calculated at times when significant data were missing near the approximate center of the circulation.

Citation: Monthly Weather Review 136, 12; 10.1175/2008MWR2407.1

Fig. 11.
Fig. 11.

SR-1 radial velocity (m s−1) at 3.5° elevation angle collected at (a) 0244, (b) 0245, (c) 0247, and (d) 0251 UTC 16 May 2003. White circles enclose, and numbers indicate, the specific circulation referenced in the text. The radar view angles differ slightly for each scan but are approximately 280°–305°. Range rings are every 5 km.

Citation: Monthly Weather Review 136, 12; 10.1175/2008MWR2407.1

Fig. 12.
Fig. 12.

UMass X-Pol radial velocities at 0.5° elevation angle at 0304 UTC 16 May 2003. The scale on the bottom expresses radial wind velocities in meters per second. The white circle encloses, and the number indicates, the specific circulation referenced in the text. Range rings are every 5 km.

Citation: Monthly Weather Review 136, 12; 10.1175/2008MWR2407.1

Fig. 13.
Fig. 13.

SR-1 radial velocity (m s−1) at 0.5° elevation angle collected at (a) 0251, (b) 0254, (c) 0258, and (d) 0300 UTC 16 May 2003 showing an increase in approaching radial velocities (greens and blues) in the rear-flank of the Shamrock supercell. The velocity scale is the same for all four images, so it is labeled only in (a). White circles enclose, and numbers indicate, the specific circulation referenced in the text. The radar view angles differ slightly for each scan but are approximately 270°–320°. Range rings are every 2 km.

Citation: Monthly Weather Review 136, 12; 10.1175/2008MWR2407.1

Table 1.

Several characteristics of the low- and midlevel circulations in the Shamrock supercell. The maximum radial winds in L1 were consistently confined to two adjacent gates so the actual diameter is unknown.

Table 1.
Table 2.

The relationship between low-level (column 1) and midlevel (column 3) circulations with their individual lifetimes (columns 2 and 4, respectively), and the time period when the circulations displayed in each row were collocated (column 5).

Table 2.

1

The initial damage survey on record concluded that there were two tornadoes that caused the damage near Lela, TX, including a tornado that formed northwest of Lela and moved southeast (A. Pietrycha 2006, personal communication). However, it is likely that the first tornado will be eliminated from the database as a result of several factors, including this study.

Save