1. Introduction
Violent tornadoes (those rated F-4 or F-5 on the Fujita scale or EF-4 or EF-5 on the enhanced Fujita scale) often occur as part of a series of tornadoes produced by a single parent storm. Cyclic tornadogenesis, the process whereby a single storm generates multiple tornadoes (Darkow and Roos 1970; Burgess et al. 1982; Adlerman and Droegemeier 2000), has been documented numerous times using airborne or ground-based mobile Doppler radars (e.g., Burgess et al. 2002; Dowell and Bluestein 2002a,b; Alexander and Wurman 2005; Wurman et al. 2007; French et al. 2008; MacGorman et al. 2008). Tornadoes and tornadic supercells have also been sampled by polarimetric radars (e.g., Ryzhkov et al. 2005; Kumjian and Ryzhkov 2008; Romine et al. 2008; Van Den Broeke et al. 2008; Palmer et al. 2011), including mobile polarimetric Doppler radars (Bluestein et al. 2007a; Kumjian et al. 2008; Snyder et al. 2010; Pazmany and Bluestein 2011).
If a cyclic, tornadic supercell (CTS) produces a violent tornado, it may exhibit multiple tornado production modes (French et al. 2008) with the following general characteristics:
the storm produces relatively weak (e.g., ≤F-3 or EF-3) short-track tornadoes at nearly regular intervals, or
the storm produces long-track, violent (e.g., ≥F-4 or EF-4) tornadoes (possibly accompanied by satellite tornadoes).1
There are few mobile Doppler radar datasets that document the transition of a CTS from tornado production mode 1 to mode 2 (Dowell and Bluestein 2002a,b; Alexander and Wurman 2005). The number of such datasets is necessarily limited by the rare combination of circumstances required for data collection: a mobile Doppler radar must be collecting data in a CTS, at least one violent tornado must be documented, and at least one weaker tornado preceding the violent tornado(es) must be documented as well. Successful and safe deployment of a mobile Doppler radar under these circumstances requires a great deal of forecasting skill, patience, situational awareness, and, sometimes, luck (Bluestein and Wakimoto 2003).
The purpose of this study is to document the University of Massachusetts X-band, mobile, polarimetric Doppler radar (UMass X-Pol; Bluestein et al. 2007a) data collected during the early life cycle of the 4 May 2007 Greensburg, Kansas, CTS (hereafter “the Greensburg storm”).2 The Greensburg storm was significant in a number of ways:
It produced the first EF-5 rated tornado since the introduction of the enhanced Fujita scale (LaDue and Mahoney 2006) by the National Weather Service (NWS) in early 2007 and the strongest tornado recorded up to that date since the 3 May 1999 Moore–Bridge Creek, Oklahoma, F-5 tornado (McCarthy et al. 2007; Lemon and Umscheid 2008).
It caused 12 fatalities, 11 of them in the town of Greensburg, which was 95% destroyed by the aforementioned EF-5 tornado (hereafter “the Greensburg tornado”; shown in Fig. 1) (Lemon and Umscheid 2008; Marshall et al. 2008).
It had an unusually long period of tornado production, generating at least 22 tornadoes over a 6-h period (Lemon and Umscheid 2008).
It had a complex origin, being the 14th in a sequence of splitting storms, complicating prediction efforts (Bluestein 2009).
The Greensburg tornado, illuminated by lightning, as seen from 29 km southwest of Greensburg at 0234 UTC, when its damage path was at least 1.5 km wide. View is toward the NNE. (Photograph courtesy of R. Fritchie.)
Citation: Monthly Weather Review 140, 7; 10.1175/MWR-D-11-00142.1
The UMass X-Pol collected volumetric data in the Greensburg storm for over 1 h (0124–0234 UTC), sampling 10 tornadoes and capturing the mode-1 to mode-2 transition of the Greensburg storm, which occurred around 0200 UTC. This mobile Doppler radar dataset, while far from being the first collected in a CTS, is one of only a handful to document both an EF-5 tornado and the mode-1 to mode-2 transition with polarimetric observations. We use these data, combined with observations from the nearest operational NWS Weather Surveillance Radar-1988 Doppler (WSR-88D) at Dodge City, Kansas (KDDC), to characterize the Greensburg storm, as well as features of the Greensburg storm related to tornado production.
A brief meteorological overview and early timeline of the Greensburg storm are given in section 2. In section 3, we describe the deployment of the UMass X-Pol near the Greensburg storm, the data collected, and limitations of those data. Tornadoes produced by the Greensburg storm are tracked and characterized in section 4. In section 5, we examine the polarimetric observations of the Greensburg storm, as well as dual-Doppler analyses generated from UMass X-Pol and KDDC observations. We provide a summary and concluding remarks in section 6.
2. Event overview
The meteorological background of the Greensburg storm has already been extensively discussed by both Lemon and Umscheid (2008) and Bluestein (2009), so only a brief overview and description of the early tornadoes will be given here. The reader is referred to Bluestein (2009) for an extended discussion of the prediction aspects of the Greensburg storm, including the synoptic setup, the preceding sequence of storms, and a storm “genealogy.” The Greensburg storm was the 14th [“Storm N” in Bluestein’s (2009) genealogy] in a series of splitting and merging storms that started around 2200 UTC 4 May 2007 near Pampa, Texas. At 0000 UTC 5 May, a 500-mb trough was located west of Dodge City and, combined with the southeasterly low-level flow, furnished 34 m s−1, 0–6-km AGL bulk shear, a parameter value deemed sufficient for supercells. Mixed layer CAPE east of the dryline was estimated to have been as high as 5100 J kg−1 (Lemon and Umscheid 2008).
The first radar echo from the Greensburg storm appeared around 0030 UTC 5 May southwest of a prior storm [“Storm J” in Bluestein’s (2009) genealogy], along its surface outflow boundary. For the duration of its life cycle, the Greensburg storm remained the southernmost storm in its cluster. Over a period of 45–60 min, the Greensburg storm organized into a supercell as it moved northeastward (Fig. 2), first exhibiting a mesocyclone and distinct hook echo (in KDDC observations) around 0100 UTC and rapidly intensifying thereafter. Between 0132 and 0156 UTC, the Greensburg storm produced at least four relatively small EF-0 or EF-1 tornadoes (tracks 1–4 in Fig. 2; Table 1) between Sitka, Kansas, and Greensburg, one of which came within 8 km of Protection, Kansas, and was visible to the UMass X-Pol crew (Fig. 3b).
Objectively analyzed regions of KDDC reflectivity ≥35 dBZ at 1.5 km AGL between 0029 and 0258 UTC 5 May 2007. For clarity, only every sixth volume is contoured, shaded in alternating light and dark gray. Surveyed tornado damage tracks are plotted in heavy gray contours and numbered chronologically (following Lemon and Umscheid 2008). Thin black lines denote county boundaries, and the heavy black line denotes the border between Kansas and Oklahoma. (Tornado damage tracks are courtesy of J. Hutton of the NWS forecast office in Dodge City, Kansas.)
Citation: Monthly Weather Review 140, 7; 10.1175/MWR-D-11-00142.1
Chronology and damage survey information concerning the first 10 tornadoes produced by the Greensburg storm. Tornadoes were cyclonic unless otherwise noted. (Information courtesy of J. Hutton and M. Umscheid of the NWS forecast office in Dodge City, Kansas.) Table entries were used to identify tornadoes in the UMass X-Pol data, and have not been retroactively modified.
(a) The UMass X-Pol truck and its attendant crew members (C. Baldi, H. Bluestein, and J. Snyder) in 2008 (photograph © R. Tanamachi). (b) Tornado 2 (22 km WNW) as seen by the UMass X-Pol crew at 0138 UTC. The town of Protection is located by the grain elevator on the horizon. Note a second lowering (15 km W) in the foreground near the left edge of the frame; this lowering developed into tornado 3 at 0148 UTC (photograph © R. Tanamachi; contrast enhanced). (c) Wide view of the storm to the WNW about 5–10 min after that in (b); composite of video frames (© H. Bluestein).
Citation: Monthly Weather Review 140, 7; 10.1175/MWR-D-11-00142.1
A transition from tornado production mode 1 to mode 2 occurred between 0155 and 0200 UTC. The Greensburg tornado (track 5 in Fig. 2) formed at 0200 UTC, moved generally northeast for most of its mature phase, attained a maximum path width of 2.7 km, and inflicted severe damage in the city of Greensburg between 0245 and 0250 UTC. The Greensburg tornado dissipated at 0305 UTC, following a narrowing, looping path such that the end of the damage track was located northwest of Greensburg. During its mature stage, the Greensburg tornado was accompanied by at least five smaller satellite tornadoes (tracks 6–10 in Fig. 2; Table 1). These first 10 tornadoes are summarized in Table 1 (M. Umscheid and J. Hutton 2007, personal communication).
Lemon and Umscheid (2008) describe the structure and evolution of the Greensburg storm as inferred from KDDC data and the NWS warning process during this episode (which is widely credited with minimizing the number of fatalities). They also detail the subsequent evolution of the Greensburg storm; 12 additional tornadoes after the Greensburg tornado (a few of which can be seen in Fig. 2) were documented, some of which were also long tracked. In the context of this study, a transition from tornado production mode 2 back to mode 1 occurred after the dissipation of tornado 15. The Greensburg storm maintained supercell characteristics until about 0800 UTC and a distinct updraft until about 0900 UTC. Since these events occurred after data collection by the UMass X-Pol ended, they will not be discussed.
3. UMass X-Pol data
a. Instrument description
The UMass X-Pol (Juyent 2003; Bluestein et al. 2007a), built at the Microwave Remote Sensing Laboratory (MIRSL) of the University of Massachusetts, is an X-band (3-cm wavelength), dual-polarized, pulse-Doppler radar system installed on a modified Ford F350 truck (Fig. 3a; Table 2). Data from the UMass X-Pol have been incorporated in studies of severe storms and tornadoes over the past decade (Bluestein and Wakimoto 2003; Kramar et al. 2005; French et al. 2006; Bluestein et al. 2007a; French et al. 2008, 2009; Snyder et al. 2010). As part of these research efforts, the UMass X-Pol was deployed in severe storms by participants from the University of Oklahoma throughout the spring of 2007 (Bluestein et al. 2007b).
Characteristics of the WSR-88D and UMass X-Pol systems.
Beginning in 2007, onboard power was provided by a 3.5kW pure-sine inverter powered by deep-cycle marine batteries which, in turn, were recharged via the truck’s alternator while in motion. Initial difficulties with the recharging system limited the length of the data collections, as will be discussed later.
The UMass X-Pol data have an inherent sample width of 150 m as dictated by the transmitted pulse width. Received echoes were, however, oversampled in range with 60-m gate spacing. Similarly, the azimuthal resolution is dictated by the 1.2° antenna beamwidth (and slightly broadened owing to the scan rate), but the data were oversampled in azimuth at intervals of 0.8°.
b. Deployment in the Greensburg storm
On 5 May 2007, the UMass X-Pol was deployed 4 km east of Protection (Fig. 2) to collect data in the approaching Greensburg storm. A wall cloud, associated with an early low-level circulation center in the Greensburg storm, was located 14 km west-southwest of the UMass X-Pol at 0113 UTC and served as the initial radar target. Once deployed, the crew chose to remain stationary to maintain data continuity, apart from a 6-min period (0131–0136 UTC) when the driver moved the truck from the south to the north side of the road to minimize beam blockage from telephone poles to the west. (Parts of the life cycles of two tornadoes fell into this gap.)
As the Greensburg storm moved toward the north-northeast, past the latitude of the UMass X-Pol, the radar operator shifted the scanned sector clockwise (toward the north), keeping the hook region near the middle of the sector. Although the terrain around the deployment site was relatively flat, several clusters of trees within 2 km of the truck (some of which can be seen in the foreground of Fig. 3b) prevented data collection at elevation angles lower than 3.0°. Between 0113 and 0123 UTC, 106 single-elevation sector scans were collected at elevation angles of ~3.0°–4.0°. The radar operator then switched the radar into volume collection mode, wherein sequential sweeps were collected at increasing elevation angles, starting at 3.0°. Initially, the maximum elevation angle was 10°, but as the storm moved away from the radar, this angle increased to 15° at 0214 UTC, and finally, to 20° at 0224 UTC. (In effect, the depth of the volume increased with time.) In addition, the maximum range of the radar was increased from 30 to 60 km at 0220 UTC. Between 0124 and 0234 UTC, 81 volumes were collected over various azimuthal sectors toward the west and north, for a total of 539 sweeps (Table 3).
UMass X-Pol data collection in the Greensburg storm. Elevation angles are relative to the truck bed, while azimuthal angles are clockwise from north.
The UMass X-Pol crew visually observed tornadoes 2, 3, 4, and 5 during the deployment (Figs. 3b,c). Intervening precipitation, tree blockage, the onset of darkness, and distance prevented visual observation of the other tornadoes. During tornado 5, a large lowering in the cloud base (not shown), illuminated by lightning, was visible to the UMass X-Pol crew.
UMass X-Pol data collection ceased at 0234 UTC when the onboard battery charge was depleted. (Recall that the batteries were recharged while the truck was moving, and the truck had been stationary for over an hour by that time.) Therefore, no UMass X-Pol data were collected during the exact time period when tornado 5 severely damaged Greensburg (0245–0250 UTC). (The crew was unaware that Greensburg was about to be hit.) However, the 5 May 2007 UMass X-Pol dataset contains data collected from the 10 tornadoes reported during the deployment, including the complete life cycles of at least 7 (3, 4, and 6–10) and the genesis through mature stages of the Greensburg tornado (tornado 5).
c. Data limitations
The exact heading of the truck was not recorded, but the UMass X-Pol driver attempted to align the truck exactly north–south using the gridded roads for guidance. For the purposes of this study, the azimuthal information was corrected on a volume-by-volume basis by matching stationary ground clutter targets to buildings (e.g., the Protection grain elevator; see Fig. 3b) in satellite imagery. These azimuthal corrections never exceeded ±3.1°.
During postanalysis, it was found that the UMass X-Pol computer clock was approximately 2.5 min (±15 s) ahead of the KDDC clock. This time offset was determined by calculating the time offset between KDDC and UMass X-Pol sweeps that appeared to show tornado vortex signatures (VSs; small areas of relatively strong inbound and outbound velocities in close proximity) in the same location. UMass X-Pol data were corrected by subtracting 2.5 min from the time stamp on all sweeps in the dataset; the corrected times appear in all figures, tables, and text in this manuscript.
Large areas of attenuated X-band reflectivity, presumably resulting from heavy rain and/or large hail (>3-cm diameter; National Climatic Data Center 2009), were evident on the far side (with respect to the UMass X-Pol) of the Greensburg storm core (Fig. 4a). Such large hail falls into the Mie scattering regime at X-band (3 cm) wavelengths, with a reduction in the backscattering efficiency compared to Rayleigh scattering (Kumjian et al. 2008). Therefore, the reflectivity field in this dataset is more appropriately referred to as the equivalent reflectivity field, with units of dBZe (Doviak and Zrnić 1993).
UMass X-Pol (a) uncalibrated equivalent reflectivity (in dBZe), (b) aliased velocity (in m s−1), (c) dealiased and edited velocity (in m s−1), (d) differential reflectivity (in dB), and (e) copolar cross-correlation coefficient (unitless) at 0229 UTC. All data are from an elevation angle of 3.1° except for (e), where 4.1° data are shown because 3.1° data were contaminated by beam blockage. The scans intersect the tornado at an altitude of 1.9 km AGL [2.5 km AGL in (e)]. At this time, the damage track of the mature Greensburg tornado was 2.5 km wide, and (c) shows the same Doppler velocity data after manual dealiasing and some editing. Range rings are 15 km apart; spokes are 30° apart.
Citation: Monthly Weather Review 140, 7; 10.1175/MWR-D-11-00142.1
Attenuation in the hail core accounts for some of the differences seen between the observed reflectivity patterns from KDDC (whose S-band signal is not as susceptible to attenuation) and those from UMass X-Pol. However, for this particular deployment, UMass X-Pol reflectivity data were not well calibrated, as it was the first in-storm deployment of the 2007 season. In most instances, the reflectivity values measured by the UMass X-Pol in areas of the Greensburg storm not heavily affected by attenuation were ~30 dBZe lower than corresponding KDDC observations. Therefore, UMass X-Pol reflectivity fields were not directly compared to observed reflectivity fields from KDDC or any other radars,3 and were only used to infer qualitative details of the structures of the Greensburg storm and tornadoes.
UMass X-Pol Doppler velocity data were aliased around a maximum unambiguous velocity of ±19.2 m s−1, resulting in multiple “folds” near the centers of tornadoes (Fig. 4b). For ease of study, UMass X-Pol Doppler velocity data associated with second-trip echo and clutter were manually masked in Solo II (Oye et al. 1995), and the remaining Doppler velocities manually dealiased (Fig. 4c).
During the UMass X-Pol deployment on the Greensburg storm, owing to a software coding error, the complex-valued numerator of the polarization correlation coefficient at zero lag (ρhv; see Doviak and Zrnić 1993) was inadvertently stored to a real variable. As such, only the real part of ρhv was retained. We denote the resulting alternative field
However, values and spatial patterns of
4. Greensburg storm structure and evolution
a. Comparison of UMass X-Pol observations and surveyed surface damage
Lemon and Umscheid (2008) reported at least 22 separate tornadoes spawned from the Greensburg storm based on surveyed tornado damage tracks. Approximate start and end times for the first 10 tornadoes (based on KDDC data and other evidence furnished by eyewitnesses, including reports, photos, and videos; Table 1) were used to associate UMass X-Pol VSs with tornado damage tracks 1–10. VSs for tornadoes 1–5, 9, and 10 are shown in Fig. 5; those for tornadoes 6–8 are omitted because they were smaller, shorter-lived, and less distinct.
Evolution of the hook echo of the Greensburg storm in UMass X-Pol (left) uncalibrated equivalent reflectivity (Z; in dBZe) and (right) dealiased Doppler velocity (υr; in m s−1): (a),(b) 0136 UTC, 8.8°; (c),(d) 0142 UTC, 7.7°; (e),(f) 0156 UTC, 3.7°; (g),(h) 0205 UTC, 4.0°; and (i),(j) 0222 UTC, 9.3°. (a)–(d) The early tornado production stage with numerous small, short-lived vortices, (c),(d) the transition to a single, long-track tornado production phase, (e),(f) the developing phase of the Greensburg tornado, and (g)–(j) the mature Greensburg tornado with two satellite tornadoes, one of which is anticyclonic. Range rings are plotted every 5 km, azimuthal spokes every 10°. VSs associated with tornadoes are numbered according to Table 1.
Citation: Monthly Weather Review 140, 7; 10.1175/MWR-D-11-00142.1
The structure of the Greensburg supercell’s hook changed considerably throughout the UMass X-Pol deployment. Between 0124 and 0200 UTC, a number of VSs formed and dissipated along a roughly northwest–southeast axis (Figs. 5a,b). At 0139 UTC, a maximum of four simultaneous VSs were detected by the UMass X-Pol (Figs. 5c,d), representing tornadoes 1–3 in various stages of their respective life cycles, as well as a VS that was not associated with a surface damage track. Later, after 0200 UTC, a large reflectivity spiral pattern and strong VS (with velocities sometimes exceeding 80 m s−1 on the outbound side) are evident (Figs. 5g–j), roughly coinciding with the early portion of the Greensburg tornado (tornado 5) damage track. Smaller VSs, occasionally accompanied by reflectivity hooks, appear near the Greensburg tornado at various times between 0206 and 0234 UTC (e.g., Fig. 5j). In general, those to the east of the Greensburg tornado are associated with damage tracks of satellite tornadoes (6–10; e.g., Figs. 5i,j), but at least two weaker VSs to the west of the Greensburg tornado (not shown) are not associated with any tornado damage track.
Most of the tornadic mesocyclones first appeared in the Doppler velocity data as broad, midlevel (2.0–4.0 km AGL) VSs that gradually contracted and intensified over periods on the order of a few minutes (not shown). In some cases (tornadoes 2–5), these midlevel VSs preceded the start times and persisted after the end times of associated tornadoes. In the case of tornado 4, the midlevel VS was distinct as early as 0139 UTC and persisted until 0204 UTC, well outside of the NWS-reported start and end times of the tornado (0150–0156 UTC; Table 1). Tornado 4 also had a visible funnel (not shown) as late as 0200 UTC.
VSs were subjectively located in all UMass X-Pol radial velocity sweeps collected at and after 0132 UTC. Most of the VSs above the surface associated with tornadoes 1–5 were located west of the corresponding surface damage (Fig. 6). The notable exception is tornado 4, whose VSs aloft were east of its surveyed damage track. Tornado 4 tilted toward the northeast later in its life cycle, so the eastward offset of the UMass X-Pol VSs from the surface damage is consistent with the radar data. The UMass X-Pol did not collect data in any of the tornadoes at altitudes below 800 m (or higher as the hook region moved farther away from the UMass X-Pol) because the storm was so far away that low elevations were below the beam and/or the beam at the lowest elevation angles was blocked by ground targets. It is therefore conceivable that the VSs aloft and surface damage do not correspond perfectly in time and space.
UMass X-Pol–detected VSs (dots; shaded by altitude in m AGL) associated with tornado damage tracks (Table 1). Times (in UTC) of the first and last VSs for each track are indicated; the times for tornado 5 are in bold. A gap in the data from 0131 to 0136 UTC results in breaks in the VS tracks for tornadoes 1 and 2. Data collection ceased at 0234 UTC; Greensburg was struck at 0245 UTC. Recall that “shallow” UMass X-Pol volumes were collected until 0214 UTC and “deep” volumes after, so there is more information about the VS at altitudes above 5 km after that time.
Citation: Monthly Weather Review 140, 7; 10.1175/MWR-D-11-00142.1
The tracks (Fig. 6) of the UMass X-Pol VSs associated with tornado 2 (3) and tornado 4 (5) are nearly contiguous. Tornadoes 2 and 4 appear to have originated from the same midlevel circulation, although this circulation became very broad (>4 km) during the interval between the two tornadoes (0139–0150 UTC; Table 1). Similar observations of multiple low-level circulations originating from the same midlevel circulation in the 15 May 2003 Shamrock, Texas, tornadic supercell were reported by French et al. (2008). The low- to midlevel VS associated with tornado 3 moved out of the UMass X-Pol sector between 0151 and 0153 UTC, and when the sector was shifted clockwise to include it again at 0153 UTC, a dissipating circulation (presumably the remnant of tornado 3) was being absorbed into the east side of the broad precursor circulation of tornado 5 (Figs. 5e,f).
b. Reflectivity features
Most of the VSs in UMass X-Pol velocity data correspond to a reflectivity hook (although some are embedded within the overall reflectivity structure of the Greensburg storm and are difficult to discern). In addition, several of the tornadoes (1, 3, 4, 5, 10; e.g., Fig. 5i) exhibited a weak-echo hole (WEH; or local reflectivity minimum). During the mature phase of the Greensburg tornado, the WEH was a continuous, weak-echo column (WEC) extending above 10 km AGL (Fig. 7). (Hereafter, this reflectivity minimum will be called a WEH when it is discussed in the context of a single sweep, and a WEC when discussed over a depth of more than one sweep.) This WEC is similar to those observed in previous radar datasets collected in tornadoes (e.g., Fujita 1963; Wakimoto et al. 1996; Wurman and Gill 2000; Dowell and Bluestein 2002a). We address polarimetric characteristics of the WEC in the next section.
A pseudo-RHI of UMass X-Pol (a) uncalibrated equivalent reflectivity (in dBZe), (b) Doppler velocity (in m s−1), (c) Zdr (in dB), and (d)
Citation: Monthly Weather Review 140, 7; 10.1175/MWR-D-11-00142.1
The vertical continuity of the mature Greensburg tornado’s WEC is striking. WECs in tornadoes are commonly attributed to the centrifuging of hydrometeors and debris from the center of the intense vortex (Dowell et al. 2005). We can perform a scale analysis of the centrifugal acceleration and radial acceleration from frictional drag, referring to Eq. (12) in Dowell et al. (2005):
In this formulation, u is the radial velocity and υ the tangential velocity. Subscripts p and a refer to particles (large raindrops) and air, respectively. The terminal velocity wt for large raindrops is 10 m s−1. At 0225 UTC, at a radius r = 200 m from the center of the vortex, we estimate that υp = 80 m s−1 (25 m s−1) at 1.5 km AGL (8.0 km AGL), making the centrifugal acceleration ~32 m s−2 (3 m s−2). Assuming that |υa − υp| ~ 30 m s−1 and (ua − up) ~ −1 m s−1 at all levels, the radial acceleration from frictional drag is ~ −3 m s−1. Centrifugal acceleration is an order of magnitude larger at 1.5 km AGL than it is at 8.0 km AGL. We therefore conclude that centrifuging caused the bottom part of the WEC, but is less likely to be the cause aloft. What, then, caused the upper part of the WEC?
Azimuthal shear associated with the WEC (as measured by the UMass X-Pol) decreased by an order of magnitude between 2 and 14 km AGL (Fig. 8). The decrease in azimuthal shear (and hence, vorticity) with height implies a downward-directed perturbation pressure gradient force in the vortex, leading to speculation that the WEC aloft resulted from an axial downdraft combined with sublimation and/or evaporation of hydrometeors. Many observational, physical, and numerical modeling studies indicate the presence of such an axial downdraft in the centers of tornadoes and tornadolike vortices (Church et al. 1977; Lewellen 1993; Lee and Wurman 2005). However, this process does not adequately explain the low scatterer concentrations near the top of the WEC (inferred from low
Maximum azimuthal shear (in s−1) within a three-gate radius of the VSs associated with the mature Greensburg tornado and its parent mesocyclone as a function of height.
Citation: Monthly Weather Review 140, 7; 10.1175/MWR-D-11-00142.1
A bounded weak-echo region (BWER) with an upper-level echo overhang, indicative of an updraft region or “vault” (Browning and Donaldson 1963) in the Greensburg supercell, is apparent on the near side of the tornado (Fig. 7a), and from the viewer’s perspective would have come out of the page and wrapped around the front of the tornado. A partial reflectivity curl (Fig. 7a) associated with a horizontal vortex (Fig. 7b) appears in the echo overhang. Wakimoto et al. (1996) found similar features in the echo overhangs of supercells sampled by airborne radars during the Verification of the Origins of Rotation in Tornadoes Experiment (VORTEX), which they refer to as “echo curls” (see their Fig. 6) and attribute to the recycling of hail embryos into the supercell updraft.
c. Comparison of tornado and vault motion
Dowell and Bluestein (2002b) observed that cyclic tornadogenesis (i.e., mode 1) in the 8 June 1995 McLean, Texas, storm resulted from “a mismatch between the horizontal motion of successive tornadoes and the horizontal velocity of the main storm-scale updraft and downdraft.” As a corollary, long-track tornadoes (i.e., mode 2) were observed when the horizontal motion of a tornado closely matched that of its associated updraft and downdraft. To see if the Greensburg storm exhibited similar behavior, we used reflectivity and velocity data from KDDC to locate subjectively the supercell vault (as a proxy for the updraft, in the absence of vertical velocity measurements) and the low-level (1.5–2.0 km) VSs from the UMass X-Pol data (as proxies for the tornadoes; Fig. 9), and compared the motions of each.
Tracks of low-level (1.5–2.0 km AGL) UMass X-Pol VSs (gray dots) and the vault of the Greensburg storm (black triangles) overlaid on top of tornado tracks (as in Fig. 1). Arrows and circled numbers mark the locations of eastward jumps in the vault’s track.
Citation: Monthly Weather Review 140, 7; 10.1175/MWR-D-11-00142.1
Because the vault of a supercell spans a considerable depth (about 1.0–8.0 km AGL), and KDDC data had more consistent coverage at altitudes at and above 5 km than those of UMass X-Pol, particularly during the early tornadoes (1–4), KDDC data were used to infer the vault’s location. KDDC collected volumetric reflectivity and Doppler velocity data every 4.1 min in the Greensburg storm. According to the conceptual model of Browning and Donaldson (1963)—later refined by Lemon and Doswell (1979)—and the results of numerical simulations (e.g., Weisman et al. 1983), the updraft of a mature Northern Hemisphere supercell is typically located near the highest reflectivity gradients on the left (with respect to storm motion) side of the hook echo at low levels, adjacent to the notch region, and on the near-hook side of the BWER at upper levels. We assumed that the updraft remained in the same place with respect to these reflectivity features. The latitude and longitude of the vault were subjectively located (to within ±2 km) by comparing reflectivity fields at 1.5 and 7.5 km AGL (taken to be the altitudes of the low-level hook and upper-level echo overhang, respectively).
The vault exhibited fairly steady motion toward the northeast at most of the analysis times, except for three eastward “jumps” in its track that occurred between 0136 and 0140 UTC, 0144 and 0148 UTC, and 0152 and 0156 UTC (Figs. 9, 10) during the Greensburg storm’s mode-1 tornado production phase (0132–0200 UTC). We speculate that such jumps to the east could have been caused by new updraft pulses at the apex of a rear-flank gust front (RFGF) as it wrapped around and occluded the hook region of the supercell. Occlusion might also have served to advect the lower portions of any ongoing tornadoes left and/or rearward with respect to the updraft motion, thereby causing a mismatch in the horizontal motions of these two features (Dowell and Bluestein 2002b). The tilt of tornado 4 toward the updraft with height near the end if its life cycle (Fig. 6) may be evidence of this updraft-relative rearward advection at low altitudes.
The (a) u and (b) υ components of motion (in m s−1) of the Greensburg storm vault (solid line with no markers) and VSs associated with tornadoes (marked lines) as depicted in Fig. 9. Tornadoes 7 and 9 are not depicted because neither lasted longer than the duration of a full KDDC volume scan (4.1 min), so their velocities could not be computed. Circled numbers indicate the times of eastward jumps in the vault’s track, as shown in Fig. 9.
Citation: Monthly Weather Review 140, 7; 10.1175/MWR-D-11-00142.1
In general, the earlier tornado damage tracks (1–4) had a component of motion to the left of the vault’s motion, while the Greensburg tornado tracked more or less parallel to the vault during its mature phase (Fig. 9). However, the surface damage tracks did not contain information about the tornadoes’ horizontal speed, so UMass X-Pol data were used to estimate the speeds of the tornadoes.
UMass X-Pol VSs between 1.5 and 2.0 km AGL (the lowest altitude range over which most of the tornadoes and mesocyclones were sampled) were sorted, by tornado, into 4.1-min intervals corresponding to the KDDC volumes. The average motion of these VSs was computed and compared to the vault motion over the same 4.1-min intervals. The mean u and υ components of motion of the vault and VSs associated with each tornado were randomly resampled 1000 times, and the bootstrap mean components for each VS were compared to the 90% confidence interval for the vault’s components of motion. If a VS’s bootstrap mean u or υ component lies outside the vault’s 90% confidence interval, then the VS’s mean component is significantly different from that of the vault. It was expected that the mode-1 VSs would have a smaller u component of motion, while the mode-2 Greensburg tornado would move with the vault. It was found that the mean u components for all mode-1 VSs fell below the 90% confidence interval for the vault’s u component, meaning we can say with 90% confidence that the mode-1 VSs moved eastward more slowly than the vault. There was no statistical difference, however, between the υ components of motion for the mode-1 VSs and the vault. Additionally, both components of the mode-2 VSs’ motion fell within the 90% confidence intervals for the vault’s motion, so the motion of the VSs was not significantly different from that of the vault. We conclude that the motion of the Greensburg tornado matched the motion of the updraft more closely than the mode-1 tornadoes, particularly when the Greensburg tornado was approaching its mature phase (0206–0234 UTC; Fig. 10). This behavior persisted for at least 20 min, and may have continued after UMass X-Pol data collection ended.
The reader is cautioned against interpreting Fig. 10 in a strictly quantitative manner, since the vault and VSs were located subjectively. At times, it was difficult to determine the precise location of a VS, particularly when it was broad, weak, or other VSs were close by. The precursor circulation of tornado 5 (which was initially very broad; see Fig. 5f) is believed to have ingested the remnant circulation of tornado 3. Therefore, the location of the precursor circulation of tornado 5 was particularly uncertain, and may explain, at least in part, the apparent substantial westward motion (u < 0) of tornado 5 prior to NWS-reported tornadogenesis (0200 UTC; Table 1; Fig. 10a).
5. Polarimetric observations
As previously mentioned, variables relating to the transmitted UMass X-Pol signal phase (in particular, ΦDP and KDP) were unavailable for this study. In this section, we focus on those variables related to transmitted signal power: differential reflectivity (ZDR) and copolar cross-correlation coefficient (
a. Hook echo features
Patterns of ZDR and
UMass X-Pol (a),(e),(i) uncalibrated equivalent reflectivity (in dBZe); (b),(f),(j) dealiased Doppler velocity (in m s−1); (c),(g),(k) ZDR (in dB); and (d),(h),(l)
Citation: Monthly Weather Review 140, 7; 10.1175/MWR-D-11-00142.1
In tornadoes 3, 4, 5, and 10 and their parent mesocyclones, the distribution of ZDR around the ring was uneven (e.g., Figs. 4d, 11g,k), while
Low-ZDR (≈0 dB), low-
At 0226 UTC, it can be seen in polarimetric data collected in a “deep” volume that the low-ZDR, low-
A conceptual diagram of the ZDR and
Interpretations of the structures of the Z hook and ZDR and
Citation: Monthly Weather Review 140, 7; 10.1175/MWR-D-11-00142.1
b. ZDR in the forward flank
When the forward-flank precipitation shield was included in the UMass X-Pol sector, relatively high values of ZDR (≥3.0 dB) were observed at and below 4 km AGL (a “ZDR shield”) (Romine et al. 2008), with particular enhancement (≥5.0 dB) along the inflow edge of the forward-flank precipitation shield (a “ZDR arc”; Kumjian and Ryzhkov 2008) (e.g., Fig. 4d). These features have been reported in other Great Plains supercells sampled by S-, C-, and X-band polarimetric radars (Schuur et al. 2001; Kumjian and Ryzhkov 2008; Romine et al. 2008; Van Den Broeke et al. 2008; Kumjian and Ryzhkov 2009; Kumjian 2011; Palmer et al. 2011). Synthesizing these studies, it is generally accepted that the ZDR arc signifies the presence of larger, more oblate drops, and that the source of these large drops is melting hail.
At least two different explanations have been promoted to explain the observed enhancements of ZDR in the forward flank. Romine et al. (2008) argue that collision, coalescence, and breakup processes (CCB) are responsible for large drop–dominant distribution in the ZDR shield, based on a synthesis of studies of tropical cloud microphysical processes. Kumjian and Ryzhkov (2008, 2009) hypothesize that the ZDR arc (specifically) results from drop size sorting as a mesolow in the storm core deepens and low-level convergence increases. In their model, small drops in the forward flank are preferentially advected toward the interior of the storm’s core by powerful near-surface inflow, so that the hydrometeor distribution near the edge of the precipitation shield consists predominantly of larger, more oblate drops. We speculate, based on examples of forward-flank ZDR enhancement provided by Romine et al. (2008) and Kumjian and Ryzhkov (2008, 2009), that CCB processes could be responsible for, say, 2–3 dB of enhancement in the ZDR shield, while size sorting along the inflow edge results in an additional 2–3 dB in the ZDR arc. However, owing to inconsistent UMass X-Pol sampling of the forward flank (the sector was selected to focus on the hook echo region of the Greensburg storm rather than the forward flank) and the lack of empirical (e.g., disdrometer) observations of drop sizes in the forward flank, we cannot make a definitive statement about any association between the ZDR fields and changes in, for example, the low-level wind shear or drop size distributions. We await the results of disdrometer-based supercell drop size distribution studies, including those conducted as part of VORTEX2 (Dawson and Romine 2010; Wurman et al. 2010).
c. Comparison with dual-Doppler analyzed features
To ascertain more objectively the relative locations of updrafts and downdrafts of the Greensburg supercell, dual-Doppler analyses (Armijo 1969) of the u, υ, and w fields were generated from KDDC and UMass X-Pol data at seven selected analysis times. In general, we consider the later dual-Doppler analyses (0207, 0220, and 0231 UTC, during the mode-2 phase) more reliable than the earlier ones (0127, 0138, 0147, and 0157 UTC, during the mode-1 phase), because the UMass X-Pol collected deeper volumes (Table 3) and because the Greensburg storm’s hook region was closer to the center of the northeastern dual-Doppler lobe at the later times (Fig. 13). In the following discussion, we focus primarily on the relative locations of updrafts and downdrafts, rather than their magnitudes.
Dual-Doppler lobes (stippled) between KDDC and UMass X-Pol (shown with a maximum range of 60 km and minimum between-beam angle of 20°). Tornado tracks are plotted on top of the dual-Doppler lobes; see Fig. 2 for an enlarged view.
Citation: Monthly Weather Review 140, 7; 10.1175/MWR-D-11-00142.1
The dual-Doppler analyses at 0127 (not shown), 0138, and 0157 UTC (Fig. 14) contain vorticity maxima at 2.0 km AGL whose locations are consistent with observations of tornadoes 1, 2, 3, and 5 in UMass X-Pol data. At 0138 UTC, vorticity maxima corresponding to tornadoes 1 and 2 are analyzed along updraft–downdraft interfaces at 2.0 km AGL (Fig. 14d). The ZDR values in the VSs of tornadoes 1 and 2 are smaller (1–2 dB) than they are elsewhere in the hook echo (4–5 dB; Fig. 11c), a possible indication of smaller, more spherical drops ingested from the downdrafts (Kumjian 2011). However, corresponding
Dual-Doppler analyzed storm-relative velocity vectors (arrows) at 2.0 km AGL at (a)–(c) 0138 UTC and (d)–(f) 0157 UTC, overlaid on (a),(d) UMass X-Pol uncalibrated equivalent reflectivity (filled color contours in dBZe) and UMass X-Pol Doppler (radial) velocity (black contours in m s−1); (b),(e) dual-Doppler analyzed vertical velocity (filled color contours in m s−1) and vertical vorticity (black contours in 10−5 s−1); and (c),(e) UMass X-Pol ZDR (filled color contours in dB) and dual-Doppler analyzed vertical velocity (black contours in m s−1). Vorticity maxima associated with tornadoes are numbered according to Table 1. Positive (negative) values are plotted as solid (dashed) lines.
Citation: Monthly Weather Review 140, 7; 10.1175/MWR-D-11-00142.1
At 0146 UTC, however, similarly small ZDR values were found in the precursor circulation of tornado 3 (1–2 dB; Fig. 11c) relative to the rest of the hook (3–5 dB), including the precursor circulation of tornado 4. In this case, the hook echo was much closer to the UMass X-Pol and beam blockage effects were minimal, so the corresponding
Tornado 5, which formed at 0200 UTC (Table 1), appears as a relative minimum in both ZDR and
Dual-Doppler analyzed storm-relative velocity vectors (arrows) and (a),(c),(e) vertical velocity (black contours in m s−1) at 2.5 km AGL overlaid on UMass X-Pol ZDR (filled color contours in dB), and (b),(d),(f) vertical vorticity (black contours in 10−5 s−1) overlaid on UMass X-Pol
Citation: Monthly Weather Review 140, 7; 10.1175/MWR-D-11-00142.1
As in Fig. 15, but at 5.0 km AGL.
Citation: Monthly Weather Review 140, 7; 10.1175/MWR-D-11-00142.1
At 0220 and 0231 UTC, the eastern portion of the Greensburg mesocyclone’s ZDR ring is thinner than the western portion (Figs. 4d, 15g,k). In both cases, this decrease in ZDR occurs downstream of the downdraft analyzed in the immediate vicinity of the tornado. Inward transport by the tornado of smaller, more spherical drops in this downdraft might have eroded the east side of the ZDR ring (Kumjian 2011). There is another possible interpretation of this apparent “erosion,” however. The main source region for the large drops in the ZDR ring appears to be the heavy precipitation core north of the tornado. Large drops revolving cyclonically around the (unusually wide) tornado vortex would fall out faster than small ones (Gunn and Kinzer 1949), so one would expect the ZDR ring to become thinner farther from the large-drop source (i.e., on the east side of the tornado). The smaller inner diameter of the
6. Summary and conclusions
Two distinct cyclic tornado production modes occurred in the Greensburg storm: in mode 1 (0124–0200 UTC), the Greensburg storm generated four short-track, weak tornadoes. A transition to mode 2 occurred around 0200 UTC, and the long-track, violent Greensburg tornado resulted. Additional brief satellite tornadoes formed to the south and east of the Greensburg tornado. Mode-1 tornadoes moved to the left with respect to the parent updraft. The mode-2 Greensburg tornado (and those accompanying satellite tornadoes for which the motion could be determined) moved at the same speed and in the same direction as the parent updraft (Fig. 10). Dowell and Bluestein (2002a) also found that mode-1 tornadoes moved left with respect to the supercell updraft, while the mode-2 (McLean) tornado exhibited similar horizontal motion to the updraft.
A WEC extending as high as 10 km was found inside the Greensburg tornado, and likely formed from centrifuging of hydrometeors and debris, followed by rapid upward transport of relatively scatterer-free air (Fig. 7). The WEC was observed in both mode-1 tornadoes (3, 4) and mode-2 tornadoes (5, 10), and was coincident in each of these cases with a low-ZDR, low-
Based on ZDR and
In this study, we sought to illuminate the mode-1 to mode-2 transition and polarimetric characteristics of the Greensburg storm purely through analysis of the radar data. To evaluate thoroughly the dynamical mechanisms for tornado formation, maintenance, and dissipation, additional high-spatiotemporal resolution numerical analyses will be required to retrieve the full three-dimensional wind fields using the UMass X-Pol data. Experiments to simulate this storm in a numerical computing framework, using the ensemble Kalman filter to assimilate UMass X-Pol and WSR-88D data, are ongoing at the time of this writing.
Acknowledgments
This research was supported by National Science Foundation Grants ATM-0637148, ATM-0934307, and ATM-0802888, and by NOAA Grant NA08OAR4320904. Portions of this manuscript were completed while the first author was a graduate research assistant at the School of Meteorology and Cooperative Institute for Mesoscale Meteorological Studies in Norman, Oklahoma, and subsequently, a postdoctoral research associate at the Center for Analysis and Prediction of Storms (CAPS) in Norman.
Two residents of Protection, Kansas, assisted the UMass X-Pol crew in locating truck services and a deployment site. Jeff Hutton and Mike Umscheid of the NWS–Dodge City, Kansas, forecast office provided the damage-track shape files, tornado chronology (Table 1), and a great deal of other useful information about the Greensburg storm and its tornadoes. Dr. Louis Wicker provided computing support. Discussions with Les Lemon, Dr. Chuck Doswell, Dr. David Dowell, Dr. Morris Weisman, Vijay Venkatesh, Jeffrey Snyder, Dr. Alan Shapiro, and Matthew Kumjian were illuminating and helpful. We used the software package ViSky, developed by Gordon Carrie, to generate the image of the dual-Doppler lobes. Robert Fritchie supplied the photograph of the Greensburg tornado. The comments of two anonymous reviewers resulted in substantial improvements to this manuscript.
APPENDIX
Description of the Dual-Doppler Synthesis
The UMass X-Pol observed the hook region of the Greensburg storm from ranges of 10–60 km at elevation angles from 3° to 20°; KDDC also observed the hook region from ranges of 65–75 km at elevation angles from 0.5° to 19.5° (Fig. 13). Analysis times were selected at roughly 10-min intervals (0127, 0138, 0147, 0157, 0207, 0220, and 0231 UTC), and the closest (in time) KDDC and UMass X-Pol volumes were objectively analyzed and synthesized on a 30 km × 30 km × 10 km analysis grid with a horizontal and vertical grid spacing of 250 m, centered on the hook region of the Greensburg supercell.
To correct for the translation of features within the radar volumes, storm motion was estimated by subjectively tracking reflectivity features not immediately associated with the mesocyclone, hook echo, or vortices. The UMass X-Pol and KDDC sweeps were translated either forward or backward to a common analysis time, then objectively analyzed to the analysis grid using a distance-weighted exponential analysis scheme (Oye and Case 1995). We used a smoothing parameter ranging from −4.9 to −10.0, calculated in a manner consistent with the methodology of Trapp and Doswell (2000).
The dual-Doppler analyses of UMass X-Pol and KDDC observations were synthesized using Custom Editing and Display of Reduced Information in Cartesian Space (CEDRIC) software (Miller and Fredrick 1993). Lacking other near-surface wind data, UMass X-Pol data were extrapolated from the lowest elevation angle to the surface, following which vertical velocities w were retrieved by assuming a lower-boundary condition of w = 0 and integrating the mass continuity equation upward. Retrieved wind fields within 20° of the baseline were discarded, and those within 30° of the baseline were considered more tenuous than those closer to the center of the dual-Doppler lobe.
REFERENCES
Adlerman, E. J., and K. Droegemeier, 2000: A numerical simulation of cyclic tornadogenesis. Preprints, 20th Conf. on Severe Local Storms, Orlando, FL, Amer. Meteor. Soc., 17.2.
Agee, E. M., J. T. Snow, and P. R. Clare, 1976: Multiple vortex features in the tornado cyclone and the occurrence of tornado families. Mon. Wea. Rev., 104, 552–563.
Alberty, R. L., D. B. Burgess, and T. Fujita, 1980: Severe weather events of 10 April 1979. Bull. Amer. Meteor. Soc., 61, 1033–1034.
Alexander, C. R., and J. Wurman, 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.
Armijo, L., 1969: A theory for the determination of wind and precipitation velocities with Doppler radars. J. Atmos. Sci., 26, 570–573.
Bluestein, H. B., 2009: The formation and early evolution of the Greensburg, Kansas, tornadic supercell on 4 May 2007. Wea. Forecasting, 24, 899–920.
Bluestein, H. B., and R. M. Wakimoto, 2003: Mobile radar observations of severe convective storms. Radar and Atmospheric Science:A Collection of Essays in Honor of David Atlas, Meteor. Monogr., No. 52, Amer. Meteor. Soc., 105–136.
Bluestein, H. B., M. M. French, R. L. Tanamachi, S. Frasier, K. Hardwick, F. Junyent, and A. L. Pazmany, 2007a: Close-range observations of tornadoes in supercells made with a dual-polarization, X-band, mobile Doppler radar. Mon. Wea. Rev., 135, 1522–1543.
Bluestein, H. B., and Coauthors, 2007b: Preliminary results from the fielding of a disparate triad of mobile Doppler radars to study severe convective storms. Preprints, 33rd Conf. on Radar Meteorology, Cairns, Australia, Amer. Meteor. Soc., 13A.2. [Available online at http://ams.confex.com/ams/pdfpapers/122770.pdf.]
Brooks, H. E., 2004: On the relationship of tornado path length and width to intensity. Wea. Forecasting, 19, 310–319.
Browning, K. A., and R. J. Donaldson, 1963: Airflow and structure of a tornadic storm. J. Atmos. Sci., 20, 533–545.
Burgess, D. B., V. T. Wood, and R. A. Brown, 1982: Mesocyclone evolution statistics. Proc. 12th Conf. on Severe Local Storms, San Antonio, TX, Amer. Meteor. Soc., 422–424.
Burgess, D. W., M. A. Magsig, J. Wurman, D. C. Dowell, and Y. Richardson, 2002: Radar observations of the 3 May 1999 Oklahoma City tornado. Wea. Forecasting, 17, 456–471.
Church, C. R., J. T. Snow, and E. M. Agee, 1977: Tornado vortex simulation at Purdue University. Bull. Amer. Meteor. Soc., 58, 900–908.
Darkow, G. L., and J. C. Roos, 1970: Multiple tornado producing thunderstorms and their apparent cyclic variations in intensity. Proc. 14th Conf. on Radar Meteorology, Tuscon, AZ, Amer. Meteor. Soc., 305–309.
Dawson, D. T., II, and G. S. Romine, 2010: A preliminary survey of DSD measurements collected during VORTEX2. Preprints, 25th Conf. on Severe Local Storms, Denver, CO, Amer. Meteor. Soc., 8A.4. [Available online at http://ams.confex.com/ams/pdfpapers/176115.pdf.]
DOC/NOAA, 1998: Service assessment: Spencer, South Dakota, tornado May 30, 1998, NOAA/NWS, 18 pp. [Available online at http://www.nws.noaa.gov/os/assessments/spenccov.htm.]
Doviak, R. J., and D. S. Zrnić, 1993: Doppler Weather Radar and Observations. 2nd ed. Academic Press, 562 pp.
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, 2626–2648.
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, 2649–2670.
Dowell, D. C., C. R. Alexander, J. M. Wurman, and L. J. Wicker, 2005: Centrifuging of hydrometeors and debris in tornadoes: Radar-reflectivity patterns and wind-measurement errors. Mon. Wea. Rev., 133, 1501–1524.
French, M. M., H. B. Bluestein, D. C. Dowell, L. J. Wicker, M. R. Kramar, and A. L. Pazmany, 2006: The 15 May 2003 Shamrock, Texas, supercell: A dual-Doppler analysis and EnKF data-assimilation experiment. Preprints, 23rd Conf. on Severe Local Storms, St. Louis, MO, Amer. Meteor. Soc., 14.6A.
French, M. M., H. B. Bluestein, D. C. Dowell, L. J. Wicker, M. R. Kramar, and A. L. Pazmany, 2008: High-resolution, mobile Doppler radar observations of cyclic mesocyclogenesis in a supercell. Mon. Wea. Rev., 136, 4997–5016.
French, M. M., H. B. Bluestein, L. J. Wicker, D. C. Dowell, and M. R. Kramar, 2009: An example of the use of mobile, Doppler radar data for tornado verification. Wea. Forecasting, 24, 884–891.
Fujita, T., 1960: A detailed analysis of the Fargo tornadoes of June 20, 1957. U.S. Government Printing Office, Research Paper 42, U.S. Weather Bureau, Washington, DC, 67 pp.
Fujita, T., 1963: Analytical mesometeorology: A review. Severe Local Storms, Meteor. Monogr., No. 27, Amer. Meteor. Soc., 77–125.
Gunn, R., and G. D. Kinzer, 1949: The terminal velocity of fall for water droplets in stagnant air. J. Meteor., 6, 243–248.
Juyent, F., 2003: Design, development and initial field deployment of an X band polarimetric, Doppler radar. M.S.E.E. thesis, Dept. of Electrical and Computer Engineering, University of Massachusetts—Amherst, 121 pp.
Junyent, F., S. Frasier, D. J. McLaughlin, V. Chandrasekar, H. Bluestein, and M. French, 2005: High resolution dual-polarization radar observation of tornados: Implications for radar development and tornado detection. Proc. Geoscience and Remote Sensing Symp., Vol. 3, Seoul, South Korea, IEEE Int., 2034–2037.
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, 2608–2634.
Kumjian, M. R., 2011: Precipitation properties of supercell hook echoes. Electron. J. Severe Storms Meteor., 6 (5). [Available online at http://www.ejssm.org/ojs/index.php/ejssm/article/view/93/65.]
Kumjian, M. R., and A. V. Ryzhkov, 2008: Polarimetric signatures in supercell thunderstorms. J. Appl. Meteor. Climatol., 47, 1940–1961.
Kumjian, M. R., and A. V. Ryzhkov, 2009: Storm-relative helicity revealed from polarimetric radar measurements. J. Atmos. Sci., 66, 667–685.
Kumjian, M. R., J. Snyder, A. Ryzhkov, D. Zrnić, S. Frasier, and H. Bluestein, 2008: Comparison of polarimetric radar observations of tornadic supercells at S, C, and X bands. Preprints, 24th Conf. on Severe Local Storms, Savannah, GA, Amer. Meteor. Soc., 5.5. [Available online at http://ams.confex.com/ams/pdfpapers/142020.pdf.]
LaDue, J. G., and E. Mahoney, 2006: Implementing the new enhanced Fujita Scale within the NWS. Preprints, 23rd Conf. on Severe Local Storms, St. Louis, MO, Amer. Meteor. Soc., 5.5. [Available online at http://ams.confex.com/ams/pdfpapers/115420.pdf.]
Lee, W.-C., and J. Wurman, 2005: Diagnosed three-dimensional axisymmetric structure of the Mulhall tornado on 3 May 1999. J. Atmos. Sci., 62, 2373–2393.
Lemon, L. R., and C. A. Doswell, 1979: Severe thunderstorm evolution and mesocyclone structure as related to tornadogenesis. Mon. Wea. Rev., 107, 1184–1197.
Lemon, L. R., and M. Umscheid, 2008: The Greensburg, Kansas tornadic storm: A storm of extremes. Preprints, 24th Conf. on Severe Local Storms, Savannah, GA, Amer. Meteor. Soc., 2.4. [Available online at http://ams.confex.com/ams/pdfpapers/141811.pdf.]
Lewellen, W. S., 1993: Tornado vortex theory. The Tornado: Its Structure, Dynamics, Prediction, and Hazards, Geophys. Monogr., Vol. 79, Amer. Geophys. Union, 19–39
MacGorman, D. R., and D. W. Burgess, 1994: Positive cloud-to-ground lightning in tornadic storms and hailstorms. Mon. Wea. Rev., 122, 1671–1697.
MacGorman, D. R., and Coauthors, 2008: TELEX: The Thunderstorm Electrification and Lightning Experiment. Bull. Amer. Meteor. Soc., 89, 997–1013.
Markowski, P. M., J. M. Straka, and E. N. Rasmussen, 2002: Direct surface thermodynamic observations within the rear-flank downdrafts of nontornadic and tornadic supercells. Mon. Wea. Rev., 130, 1692–1721.
Marquis, J., Y. Richardson, P. Markowski, D. Dowell, and J. Wurman, 2012: Tornado maintenance investigated with high-resolution dual-Doppler and EnKF analysis. Mon. Wea. Rev., 140, 3–27.
Marshall, T. P., D. McCarthy, J. LaDue, J. Wurman, C. Alexander, P. Robinson, and K. Kosiba, 2008: Damage survey of the Greensburg, KS tornado. Preprints, 24th Conf. on Severe Local Storms, Savannah, GA, Amer. Meteor. Soc., 8B.3. [Available online at http://ams.confex.com/ams/pdfpapers/141534.pdf.]
McCarthy, D., L. Ruthi, and J. Hutton, 2007: The Greensburg, KS tornado. Preprints, 22nd Conf. on Weather Analysis and Forecasting, Park City, UT, Amer. Meteor. Soc., J2.4. [Available online at http://ams.confex.com/ams/pdfpapers/126927.pdf.]
Miller, L. J., and S. M. Fredrick, 1993: CEDRIC: Custom Editing and Display of Reduced Information in Cartesian space. Reference Manual, Mesoscale and Microscale Meteorology Division, Boulder, CO, 130 pp. [Available online at http://www.eol.ucar.edu/instrumentation/airborne-instruments/eldora/eldora-help-center/manual/cedric-reference-manual/cedric_doc.pdf.]
National Climatic Data Center, cited 2009: Storm Events Database. [Available online at http://www.ncdc.noaa.gov/stormevents/.]
Oye, D., and M. Case, 1995: REORDER: A program for gridding radar data. Installation and use manual for the UNIX version. Atmospheric Technology Division, Boulder, CO, 20 pp. [Available online at http://www.eol.ucar.edu/instrumentation/airborne-instruments/eldora/eldora-help-center/manual/reorder-reference-manual/unixreorder.pdf.]
Oye, D., C. Mueller, and S. Smith, 1995: Software for radar translation, visualization, editing, and interpolation. Preprints, 27th Conf. on Radar Meteorology, Vail, CO, Amer. Meteor. Soc., 359–361.
Palmer, R. D., and Coauthors, 2011: Observations of the 10 May 2010 tornado outbreak using OU-PRIME: Potential for new science with high-resolution polarimetric radar. Bull. Amer. Meteor. Soc., 92, 871–891.
Pazmany, A. L., and H. B. Bluestein, 2011: Mobile rapid-scanning X-band polarimetric (RaXPol) Doppler radar system. Preprints, 35th Conf. on Radar Meteorology, Pittsburgh, PA, Amer. Meteor. Soc., 16B.2. [Available online at http://ams.confex.com/ams/35Radar/webprogram/Paper191294.html.]
Romine, G. S., D. W. Burgess, and R. B. Wilhelmson, 2008: A dual-polarization-radar-based assessment of the 8 May 2003 Oklahoma City area tornadic supercell. Mon. Wea. Rev., 136, 2849–2870.
Ryzhkov, A. V., T. J. Schuur, D. W. Burgess, and D. S. Zrnic, 2005: Polarimetric tornado detection. J. Appl. Meteor., 44, 557–570.
Schuur, T. J., A. V. Ryzhkov, D. S. Zrnić, and M. Schönhuber, 2001: Drop size distributions measured by a 2D video disdrometer: Comparison with dual-polarization radar data. J. Appl. Meteor., 40, 1019–1034.
Snyder, J. C., H. B. Bluestein, G. Zhang, and S. J. Frasier, 2010: Attenuation correction and hydrometeor classification of high-resolution, X-band, dual-polarized mobile radar measurements in severe convective storms. J. Atmos. Oceanic Technol., 27, 1979–2001.
Speheger, D. A., C. A. Doswell, and G. J. Stumpf, 2002: The tornadoes of 3 May 1999: Event verification in central Oklahoma and related issues. Wea. Forecasting, 17, 362–381.
Trapp, R. J., and C. A. Doswell, 2000: Radar data objective analysis. J. Atmos. Oceanic Technol., 17, 105–120.
Van Den Broeke, M. S., J. M. Straka, and E. N. Rasmussen, 2008: Polarimetric radar observations at low levels during tornado life cycles in a small sample of classic Southern Plains supercells. J. Appl. Meteor. Climatol., 47, 1232–1247.
Wakimoto, R. M., W.-C. Lee, H. B. Bluestein, C.-H. Liu, and P. H. Hildebrand, 1996: ELDORA observations during VORTEX 95. Bull. Amer. Meteor. Soc., 77, 1465–1481.
Weisman, M. L., J. B. Klemp, and J. W. Wilson, 1983: Dynamic interpretation of notches, WERs, and mesocyclones simulated in a numerical cloud model. Proc.21st Conf. on Radar Meteorology, Edmonton, AB, Canada, Amer. Meteor. Soc., 39–43.
Wurman, J., and S. Gill, 2000: Finescale radar observations of the Dimmitt, Texas (2 June 1995), tornado. Mon. Wea. Rev., 128, 2135–2164.
Wurman, J., Y. Richardson, C. Alexander, S. Weygandt, and P. F. Zhang, 2007: Dual-Doppler and single-Doppler analysis of a tornadic storm undergoing mergers and repeated tornadogenesis. Mon. Wea. Rev., 135, 736–758.
Wurman, J., L. Wicker, Y. Richardson, P. Markowski, D. Dowell, D. Burgess, and H. Bluestein, 2010: VORTEX2: The verification of the origins of rotation in tornadoes experiment. Preprints, 25th Conf. on Severe Local Storms, Denver, CO, Amer. Meteor. Soc., 5.1. [Available online at http://ams.confex.com/ams/pdfpapers/176068.pdf.]
One might characterize tornado production modes as a sliding spectrum, rather than discrete categories. It is primarily because of the vast differences in potential human impact (as denoted by their EF-scale ratings) that two modes were delineated in this study.
Historically speaking, the date of the Greensburg storm is 4 May 2007, because the tornado struck Greensburg at about 9:45 p.m. local time (2145 CDT) on that day. For the purposes of this study, however, we will follow meteorological convention and use UTC time hereafter; for example, 2145 CDT 4 May 2007 corresponds to 0245 UTC 5 May.
The Greensburg storm was detected by a number of other WSR-88Ds, including those at Vance Air Force Base near Enid, Oklahoma (KVNX; ~130 km away), and the NWS forecast office in Amarillo, Texas (KAMA; ~330 km away). KDDC was by far the closest in range (~65–75 km away).
However, the low values of