1. Introduction
A tornadic vortex signature (TVS) in Doppler weather radar scans is a degraded image of an embryonic or fully developed tornado (Brown et al. 1978). In general, it may be defined as some change, at constant radar range, in the radial velocity across adjacent beams (hence, gate to gate). This assumes that the radar beamwidth is greater than or equal to the vortex core of the tornado. When the tornado is sufficiently close to a radar, it may be sampled by several radar beams, resulting in a “tornado signature” rather than a TVS (Brown 1998).
The National Severe Storms Laboratory (NSSL) tornado detection algorithm (TDA), after which we model our TVS identification herein, requires a minimum gate-to-gate velocity change or differential velocity (ΔV) greater than or equal to 11 m s−1 [and subsequently employs other thresholds; see Mitchell et al. (1998)]. The TDA will replace the build 9 (hereafter, current) TVS algorithm in the Weather Surveillance Radar-1988 Doppler (WSR-88D) system1 in build 10 of the radar products generator.
Not all tornadoes have detectable TVSs. This may be due to the stringency of an automated algorithm or to radar sampling limitations. Indeed, radar recognition of a tornado (and mesocyclone) depends highly on the radar beamwidth to vortex core radius ratio, as well as on the position of the center of the vortex with respect to the center of the beam (e.g., Wood and Brown 1997). Even if tornadic-vortex detection is afforded, these sampling artifacts may result in radar-observed wind speeds that grossly underrepresent the actual tornadic wind speeds.
The percentage of nontornadic TVSs (i.e., gate-to-gate velocity signatures meeting TVS criteria, yet not associated with tornadoes) is unclear and has yet to be presented in the formal literature. It is likely low and is most certainly a function of algorithmic definition and range. Recent discussions of tornadogenesis failure (e.g., Trapp 1999), however, suggest that there are also physical explanations for some instances of nontornadic TVSs.
The archetypal TVS is initially detected aloft, within a mesocyclone, and subsequently “descends” to the ground with the embryonic tornado (Brown et al. 1978). This may foreshadow the near-ground spinup of the tornado, or first visual sighting of a tornado debris cloud, by as much as a few tens of minutes. Such descent can be explained theoretically by a process involving the dynamic pipe effect (Leslie 1971), which can arise when “ambient” (i.e., associated with the parent storm or its environment, at a length scale larger than that of the tornado) convergence and/or rotation increases with height above the ground (Trapp and Davies-Jones 1997). The study by Vasiloff (1993) and also a preliminary analysis of WSR-88D data (see Trapp and Mitchell 1995) suggests a few contradistinctions to this archetype [classified by Trapp and Davies-Jones (1997) as mode I tornadogenesis]. For example, some tornadoes are observed by radar to form somewhat uniformly over a several kilometer vertical depth, or appear only at the lowest altitudes and then “ascend”: formation then is via mode II tornadogenesis, which in theory should occur rapidly, within 5–10 min (Trapp and Davies-Jones 1997). Mode II tornadogenesis occurs when ambient convergence and rotation are constant or decrease with altitude above the ground; this condition can be satisfied with supercells and nonsupercells alike.
Acknowledging the sensitivities to algorithmic definition and range, we speculate that the guidance provided by a TVS should increase the probability of a successful (verifiable and timely) operational warning of mode I tornadogenesis. The challenge is to determine how often, and ultimately under what radar-observable conditions, descending TVSs, hence mode I genesis, most likely occur.
The purpose of this study is to estimate, using a relatively large sample of radar-detected tornadoes, the frequency of descending versus nondescending TVSs. This opportunity to examine a variety of tornadoes via TVSs has only recently been provided by the implementation of the WSR-88D network and archival of level II data (Crum and Alberty 1993; Klazura and Imy 1993). In section 2, we introduce this data sample and also our TVS classification criteria. Results of a stratification of descending–nondescending TVSs according to tornado lead time, tornado damage intensity, maximum gate-to-gate differential velocity, and range are given in section 3. For the purpose of illustration, case studies of storms exhibiting descending and nondescending TVSs are provided in section 4. Classification caveats are discussed and concluding remarks are made in section 5.
2. TVS identification and characterization
Tornado events were selected based on existence of WSR-88D level II (base reflectivity, velocity, and spectrum width) data and of damage verification data (as provided, e.g., through Storm Data). Specifically, we sampled the population of all tornadoes detected by WSR-88Ds with level II data recorders, within a range of 150 km. The resultant, geographically diverse dataset comprised 52 events2 and is described in Table 1. Individual events vary from southern Great Plains supercell tornadoes to landfallen tropical cyclone–spawned tornadoes and were observed at radar ranges between 23 and 137 km. Strictly speaking, this sample was not randomly selected, although the tornadoes were not chosen preferentially other than to achieve geographic diversity as a proxy for different environmental conditions.
We pause here to mention that the verification data were at times problematic, forcing the need for modification. As discussed by Witt et al. (1998), tornado formation times reported in Storm Data may be in error by perhaps 1 h, owing to a variety of nonmeteorological reasons. Damage path locations are less likely to contain gross deficiencies. Accordingly, the damage report of each event in Table 1 was compared with the radar data to confirm that a mesocyclone or TVS was at least in the vicinity of the tornado damage path begin point, at the reported tornado formation time. When such confirmation was not possible, formation time was equated to the beginning time of the radar volume scan within which the storm’s (i.e., mesocyclone or TVS) position and damage path beginning point were approximately collocated.
The TVS associated with each tornado was manually identified, thereby placing practical limitations on our sample size. Though time intensive, this manual approach prevented the otherwise possibility of erroneous automated identification owing to improperly dealiased velocity gates by a dealiasing algorithm, for example. Based on an early version of the NSSL TDA (see Figs. 1 and 2 of Mitchell et al. 1998), the following TVS identification criteria were used at each elevation angle, within each of several volume scans per event examined:(i) three velocity gate shear segments, adjacent in azimuth and constant in range, with differential velocity (ΔV) ≥15 m s−1; and (ii) time and height continuity. When the criteria were met, the maximum differential velocity ΔVmax(z) and its attributes of azimuth, range, time, and altitude (z) above radar level (ARL), were recorded for each elevation angle, within each radar volume scan. Note that our criteria allowed the designation of a TVS in absence of a mesocyclone, and, at a single elevation angle, above and below, which need not contain a rotational velocity signature. The reader also should note that the differential velocity threshold of 15 m s−1 is significantly lower than that encountered by users of the current WSR-88D TVS algorithm, and that ΔV = 15 m s−1 at some altitude ARL does not necessarily imply impending tornado formation. Our analysis benefited from a priori knowledge of tornado location and formation time.
An objective means of characterizing a TVS as either descending (D) or nondescending (ND) was employed. The idealized, empirically determined models on which this classification scheme was based are shown in Fig. 1. Briefly, descending TVSs (and associated mode I tornadogenesis) tend to exhibit distinct maxima in differential velocity aloft (>2–7 km) in at least one volume scan prior to tornadogenesis.3 This is shown by an increase with height of ΔVmax(z) within this volume scan. Over time, ΔV at lower heights increases relative to that aloft in prior volume scans. In some events, the peak ΔV at tornadogenesis is found at the lowest tilt, while in others, it remains aloft even though ΔV increases substantially at lower levels. Nondescending TVSs often begin as shallow (depth < 2–3 km) features with peak ΔV values at heights of 1 km or less (as allowed by range). They may build upward in time, as depicted in Fig. 1b, but the peak ΔV typically remains at heights <2 km or so. Alternatively, nondescending TVSs may begin as shallow features with weak to moderate ΔV that is relatively uniform with height. Subsequent intensification occurs uniformly with height.
3. Results
Upon applying the classification scheme in Eqs. (1)–(3) to our sample, we found that 52% of the n = 52 sampled tornadoes had descending TVSs and 48% had nondescending TVSs; the standard error about the percentages is about 7%. If we state as our null hypothesis H0 that 90% of all TVSs descend, we can easily reject H0 (at the α = 0.01 level of significance) in favor of the alternative hypothesis that less than 90% of all TVSs descend. Our results, therefore, are statistically significant at the α = 0.01 level. We can make the same conclusion for H0 based on a proportion of 75%.
We extracted certain quantities for each event, including the peak, pretornadic ΔV and its azimuth, range, height, and time; number of volume scans between tornadogenesis and the first appearance of a TVS; and tornado damage intensity, estimated by the Fujita (F) scale rating. Mean values, and the standard error about these means (≡std dev/
As anticipated by the theoretical work of Trapp and Davies-Jones (1997), the mean tornadogenesis lead time offered by the D events was greater than that of the ND events by two volume scans (∼10 min). Further, the peak, pretornadic differential velocity was found at midlevels (low levels) in the D (ND) cases. The greater mean ΔVpeak associated with the D TVSs is intriguing, yet has no theoretical explanation. In fact, based on mean values of F scale, both TVS classes tended to be associated with weak tornadoes. Of course, this probably reflects the fact that in the population of tornadoes, most are weak.
Because the radar beam broadens with range, the gate-to-gate velocity difference at distant ranges may be associated with a mesocyclonic vortex rather than a tornadic vortex. To explore the possible range dependencies on the TVS character, the quantities in Table 2 were recomputed as a function of range.
As shown in Fig. 3, the percentages of D versus ND TVSs remained approximately 50%, irrespective of range. The “box and whisker” plots (see, e.g., Wilks 1995) indicate that the results expressed in Table 2 also were relatively insensitive to range (Fig. 3). Of course, the minimum height of ΔVpeak is necessarily a function of range; this TVS attribute, incidentally, was associated with the greatest variability. In light of the work by Wood and Brown (1997) that suggests an undersampling of the peak tornadic wind speeds with increasing range, it is interesting to note that there was no well-defined decrease with range in the median ΔVpeak, especially with the D TVSs.
The parent storm of each tornado was classified as either line, cell, or other (see Table 1). Cell refers to a relatively isolated, typically supercellular storm, while line is a nearly continuous line/curve of radar echo, constituting a quasi-linear convective system [i.e., squall line/bow echo; Weisman and Davis (1998)]. Hurricane rainband is an example of other. A contingency table (Table 3) shows that of the seven tornadoes that formed within quasi-linear convective systems, six were associated with nondescending TVSs. In our sample of events, the conditional probability (see, e.g., Wilks 1995) is quite high (0.86) that, given radar-detectable tornadogenesis from a quasi-linear convective system, the TVS will not descend. Mean values, and standard errors about those means, of the TVS attributes for these six cases are listed in Table 2 [“ND (line)”]. Perhaps the most notable attribute is that of the single volume scan (hence ∼5 min) mean tornadogenesis lead time. Clearly, these results suggest very little utility of tornadic vortex signatures in tornado warning decisions, given squall line or bow echo situations.
4. Case studies
For illustrative purposes, we present now two case studies. The events that occurred on these two days should be viewed as examples of the two general, rather broad classes of possible tornado/tornadogenesis behaviors introduced in section 1. Since, in our opinion, tornadoes with nondescending TVSs have received the least attention in the literature, we devote more discussion to events of this type.
On 1 July 1997, several tornadoes occurred in southern Minnesota in association with a squall line that evolved into a bow echo [see comprehensive reports on bow echoes by Przbylinzki (1995) and Weisman (1993)]. From the perspective of WSR-88D KMPX near Minneapolis, Minnesota, the majority of the user-defined TVSs were classified as nondescending. One of the authors (GAT) was the KMPX radar operator during this case and found tornado nowcasting to be particularly vexing, especially along the leading edge of the squall line/bow echo (see also Howieson and Tipton 1998). Events on this day are considered to be good examples of mode II tornadogenesis. As we have just shown, this mode of tornadogenesis can occur quite frequently and by its nature may be relatively more difficult to recognize during warning operations.
Signatures of the 1 July 1997 tornadoes can be contrasted with the descending TVS found in a tornadic storm that occurred in southern Colorado on 22 June 1995. This storm was “low topped” and produced an F1 tornado near Falcon, Colorado. [The reader also may refer to the numerous discussions of the 24 May 1973 Union City, Oklahoma, tornadic storm (e.g., Brown et al. 1978; Golden and Purcell 1978; Lemon et al. 1978). Doppler radar signatures of this storm provide the classic example of mode I tornadogenesis and the concomitant descending TVS.]
a. 1 July 1997 tornadoes in southern Minnesota
A mesoscale convective system (MCS) that originated in central South Dakota moved into western Minnesota by midafternoon, 1 July 1997. In advance of the MCS (hereafter, squall line), a cluster of multicell thunderstorms that initiated over eastern South Dakota developed by late afternoon into an isolated supercell. The squall line and supercell merged and the resultant system evolved into a bow echo. Tornadoes were associated with the supercell, the cyclonic bookend vortex of the bow echo, and the squall line/bow echo itself.
Our examination of the WSR-88D KMPX data begins with the second tornado of the day, which formed near Willmar, Minnesota, at 2245 UTC, in association with the squall line. As depicted in Fig. 4, this occurred before the squall line exhibited appreciable convex shape in radar reflectivity.
Rear- and front- (relative to the line) inflow notches in reflectivity that correspond to locally intense inbound (negative) and outbound (positive) radial velocity, respectively, were pronounced signatures at, and up to 15 min prior to, the time of tornado formation (Fig. 4). The inflow and outflow formed a broad, cyclonic velocity couplet or mesocyclone. This mesocyclone was relatively shallow and of maximum intensity at the lowest radar tilt, prior to tornado formation; a TVS was identified first in the volume scan beginning 2238 UTC. In proximity to the incipient tornado, echo tops and storm-top velocity divergence signatures were found at altitudes ∼15 km ARL, revealing the strong updraft embedded within the line. Tornadogenesis accompanied a surge in the rear inflow, and a tornado with a path length of ∼0.6 km and a rating of F1 resulted.
The squall line merged with the supercell over Wright County, Minnesota, about 50–70 km northwest of KMPX. Consistent with numerous other studies (e.g., Goodman and Knupp 1993; Sabones et al. 1996; Wolf et al. 1996) such a merger favorably influenced the genesis of three tornadoes. For example, an F3 tornado occurred near the town of Buffalo, Minnesota, 43 km northwest of KMPX, at 2343 UTC. Interrogation of the KMPX data suggests that the Buffalo tornado was a remnant of weaker cyclonic shear associated with the supercell and was associated with a nondescending TVS. Furthermore, the data (e.g., velocity at 0.5° elevation and maximum reflectivity at storm top4) suggest that upon squall line–supercell merger, outflow from both the supercell and squall line resulted in localized, enhanced horizontal convergence that subsequently intensified the remnant vertical vorticity. As with the Willmar tornado, a well-defined front-inflow reflectivity notch to the north of surging outflow preceded tornadogenesis. Unlike it predecessor, relatively deep gate-to-gate shear led tornadogenesis by several volume scans.
At 2343 UTC, a significant downburst was occurring across extreme southwest Wright County and northern McLeod County, Minnesota. Low-altitude radar reflectivity possessed a distinct bulge and large gradient at the leading edge of the highest winds: inbound velocities of 45 m s−1 at an altitude of 0.7 km were seen in extreme southwest Wright County. (Storm Data reports from this area indicated convective wind gusts of 30–40 m s−1.) Also at this time, several front-inflow notches were present, the northernmost of which was related to the Buffalo tornado (Fig. 5). This figure shows the notches in what can be thought of as a line-echo wave pattern (Nolen 1959), which is reminiscent of the release of horizontal shearing instability. We have no way of rigorously evaluating the existence of this mechanism or others that explain the tornado’s parent circulation, and this is beyond the scope of this paper. The interested reader may refer to Carbone (1982, 1983) for a discussion of Helmholtz-type shearing instability in severe/tornadic frontal rainbands. The inflow notch just north of the downburst accompanied an F1 tornado that formed at 0000 UTC, near Waverly, Minnesota. Deep, broad cyclonic shear in the vicinity of (and related downburst/extensive low-altitude outflow south of) the inflow notch preceded the tornado by approximately 25 min. A TVS appeared first in the volume scan beginning 2358 UTC, yet only as a shallow, rather benign signature (Fig. 5). Again, echo top growth and intensification of maximum reflectivity aloft corresponded well with the final contraction of the broad shear into a tornado.
Appreciable outward bowing of the line and also of attendant bookend vortex and rear-inflow notch development became noticeable by ∼0030 UTC. An F1 tornado formed within the cyclonic bookend vortex at 0052 UTC, near St. Francis, Minnesota. The TVS in this case extended vertically through the lowest 3 km after formation and, interestingly, was located outside the center of the cyclonically curved reflectivity ring that usually is associated with the bookend vortex (Fig. 6). A TVS within the bookend vortex could not be found prior to the volume scan time of 0049 UTC. Aside from a general intensification of the bow echo, deepening of the rear-inflow notch, and some contraction of the bookend vortex, we could not find clues that might have suggested a genesis time of 0052 UTC (and genesis location of St. Francis). Shrinkage of the bookend vortex core diameter from 12.9 to 6.4 km was seen by Pfost and Gerard (1997) as the precursor of tornadoes that formed on 8 May 1995 near Natchez, Mississippi.
One of the final tornadoes in Minnesota on this day occurred just north of the apex of the bow echo at 0115 UTC, near Forest Lake, Minnesota. The Doppler radar signatures at this time are similar to those of the previous tornadoes, but also are unique in some ways. As evident in Fig. 7a, for example, a pronounced front-inflow notch was absent near the time and location of tornadogenesis. In several volume scans prior to this time, however, an isolated region of intensifying and expanding low-altitude outflow was found to the south of a front-inflow notch (Fig. 7b). There was no clear reflectivity maximum aloft in proximity to the low-altitude circulation. Additionally, the gate-to-gate shear at (prior to) time of tornado formation was relatively weak (nonexistent), although a mesocyclone/broad cyclonic shear associated with the rear-inflow jet preceded tornadogenesis.
b. 22 June 1995 tornado in southern Colorado
From the perspective of the Pueblo, Colorado, WSR-88D (KPUX), radar echoes of the incipient tornadic cell appeared first at approximately 2015 UTC, ∼100 km to the northwest of KPUX. Within 1 h of its presumed initiation, the cell developed a lower-altitude precipitation core with associated radar reflectivity factor of 60 dBZ. Weak mesocyclonic rotation appeared soon thereafter at midaltitudes, and the core of heavy precipitation continued to expand at low altitudes until approximately 2133 UTC. At this time, the original cell split into right- and left-moving members; the tornado developed out of the right-moving member.
A TVS was identified at 2145 UTC in the 3.3° elevation scan (Figs. 2b and 8). The TVS was spatially well correlated with areas of weak reflectivity at low and midaltitudes. On a broader scale, convergent rotation characterized the radial velocity at 3.3° elevation, suggesting a consistency with the mode I tornadogenesis mechanism. In the 0.5° scan, the radial velocity within and surrounding the storm was weak and appeared relatively benign. In the 6.0° scan, in contrast, the radial velocity was strongly divergent and thus revealed the storm summit [with the caveats expressed by Howard et al. (1997)]. The equivalent storm top altitude of ≥9 km ARL supports the storm’s “low-topped” or minisupercell classification (e.g., Kennedy et al. 1993).
TVS descent below an altitude of 2 km presumably awaited further storm organization. As can be deduced from the time–height plot in Fig. 2b, this apparent organization occurred at ∼2220 UTC. Indeed, Fig. 8c depicts the pronounced hook echo found in the 0.5° scan at 2220 UTC. This figure also shows strong inbound radial velocity within the tip of the hook, owing to rear-flank downdraft outflow and the mesocyclone. These and other features suggest that the storm had by this time evolved to its mature stage, during which tornadoes most likely develop (Brandes 1993).
The Falcon, Colorado, tornado reportedly formed at 2240 UTC, and subsequently produced F1 damage. Albeit associated with somewhat weak differential velocity, the incipient tornadic circulation had a readily apparent TVS in the 0.5° scan at 2237 UTC. The short-lived tornado had a pathlength of ∼3 km, and the parent supercell persisted for more than an hour after tornado demise.
5. Summary and discussion
A set of statistics on tornadic vortex signature (TVS) characteristics and behavior was compiled. This work was based on an analysis of WSR-88D archive level II data for 52 tornadic storms. Recall that a TVS in Doppler weather radar scans is a degraded image of an embryonic or fully developed tornado (Brown et al. 1978). Only 52% of the tornadic vortex signatures in our sample were detected first aloft and subsequently descended, apparently reaching the ground with the tornado. The remaining 48% were detected first either near the ground and grew upward, or, appeared nearly simultaneously over a several kilometer depth.
These Doppler radar manifestations of two primary modes of tornado development have been explained theoretically. According to Trapp and Davies-Jones (1997), mode I tornadogenesis occurs when “ambient” (e.g., mesocyclonic) vertical vorticity and/or radial convergence are initially greater aloft than next to the ground (but also when near-ground, ambient vertical vorticity is nonzero). This leads to intensification of mesocyclonic vertical vorticity into an incipient tornado aloft, and sets the stage for a dynamic pipe effect (DPE; Leslie 1971), a type of bootstrap process (see Figs. 14a,b of Trapp and Davies-Jones 1997). The DPE provides the horizontal convergence needed to amplify vertical vorticity at subsequently lower altitudes and is represented in Doppler radar data as a descending or archetypal TVS.
Mode II genesis occurs when ambient vertical vorticity and horizontal convergence are constant with altitude or are maximized near the ground and decrease with altitude (Fig. 14c of Trapp and Davies-Jones 1997). Vertical vorticity amplification by near-ground, ambient convergence into a tornado precludes a DPE, and tornadogenesis is represented in Doppler radar scans as a nondescending TVS.
Compared to the nondescending TVSs, the descending TVSs in our sample were associated with greater pretornadic differential velocity, found at midaltitudes, and were associated with greater tornadogenesis lead time. This result was determined to be more or less insensitive to radar range.
We examined TVS behavior with respect to the parent storm: all but one of the tornadoes that formed within quasi-linear convective systems had nondescending TVSs that provided a mean lead time of approximately 5 min. Many of the nondescending TVSs in our sample also formed within supercells. Thus, one cannot conclude that “nondescending” implies “nonsupercellular” tornadogenesis. In general, these ND events pose a particular challenge to tornado warning decisions and automated detection and diagnosis of tornadic vortices.
Two case studies were presented for illustrative purposes. On 1 July 1997 in southern Minnesota, nondescending TVSs were revealed in the leading edge of a squall line, at the merger of the squall line with a supercell, and later during that day, within the cyclonic bookend vortex of a bow echo. On 22 June 1995 in southern Colorado, a low-topped supercell storm produced a tornado that was associated with a descending TVS.
Some caution should be exercised when interpreting results of this study: (i) differential velocity of 15 m s−1 at some altitude above radar level does not necessarily imply impending tornado formation; (ii) the five volume scan (∼25 min) mean lead time with descending TVSs (see Table 1) can be attributed in part to use of this relatively low threshold and also a priori knowledge of tornado location and formation time; (iii) the classification criteria (1)–(3) were optimized to fit a subjective classification of our 52 events; it is possible that addition of other events may necessitate adjustment to Eqs. (1)–(3); and (iv) although statistically significant results were generated with a sample size of 52, this number is still small compared to the approximate number (∼1000) of tornadoes that occur in the United States annually.
With these caveats in mind, the operational meteorologist may apply our results by examining the vertical distribution of strong, gate-to-gate differential velocity ΔV (or the like) within each radar volume scan. The capability to display ΔV in a time–height presentation, as provided by the NSSL Warning Decision Support System (Eilts 1997), likely aids this endeavor; the meteorologist must additionally interrogate the radar data when deemed necessary. Although not to be used as a “cookbook,” the idealized models depicted in Fig. 1 can provide the basis for differentiating between descending and nondescending TVSs. For example, if (i) the largest ΔV within the TVS is found in the base (0.5° elevation angle) scan, (ii) the radar range to this TVS is within, say, 60 km, and, (iii) a TVS was not detected in previous volume scans, the meteorologist may conclude with a reasonable degree of confidence that the TVS is of the nondescending variety and mode II tornadogenesis is occurring. According to our statistics, the meteorologist should anticipate nondescending TVSs if tornadoes develop in a squall lines or bow echoes. We are still searching for other means to anticipate the mode of tornado development.
Performance evaluations by Polger et al. (1994) and Bieringer and Ray (1996) indicate that WSR-88D use in National Weather Service operations has led to an improved probability of detection of severe local storms and attendant phenomena. A TVS used as an adjunct piece of information for tornado warning guidance purposes likely will increase the probability of a successful (verifiable and timely) operational warning, under some circumstances: a TVS associated with a tornadic squall line or bow echo, for example, likely will not contribute to a successful warning. Of course, it is inappropriate to base tornado warning decisions only on the existence of a TVS (or on any other single signature, for that matter). It is appropriate, however, to recognize while making Doppler radar interpretations that an ostensibly large percentage of tornadoes form (within supercells and nonsupercells alike) in a manner different than that exemplified by a descending or archetypal TVS. Future studies should investigate possible radar-observable conditions and storm environments that are most likely to foster tornadoes with descending TVSs and hence greater lead time.
Acknowledgments
The authors benefited from discussions and/or assistance from the following: R. Brown, D. Burgess, R. Davies-Jones, C. Doswell, S. Ellis, B. Fehrn, D. Floyd, D. Kennedy, M. Lehmann, C. Marzban, R. Prentice, and G. Stumpf. Comments on a draft of this manuscript were provided by H. Brooks and T. Schuur. Radar analysis software packages used in this study were developed and maintained by K. Hondl and other NSSL personnel, who additionally performed the quality control of some of the tornado verification data. This work was initiated while the first author was a National Research Council–NOAA Postdoctoral Research Associate, and subsequently funded in part through COMET, from a subaward (UCAR S97-86995) under a cooperative agreement between NOAA and UCAR. WSR-88D level II data were provided by the National Climatic Data Center.
REFERENCES
Bieringer, P., and P. S. Ray, 1996: A comparison of tornado warning lead times with and without NEXRAD Doppler radar. Wea. Forecasting,11, 47–52.
Brandes, E. A., 1993: Tornadic thunderstorm characteristics determined with Doppler radar. The Tornado: Its Structure, Dynamics, Prediction, and Hazards, Geophys. Monogr., No. 79, Amer. Geophys. Union, 143–159.
Brown, R. A., 1998: Nomogram for aiding the interpretation of tornadic vortex signatures measured by Doppler radar. Wea. Forecasting,13, 505–512.
——, L. R. Lemon, and D. W. Burgess, 1978: Tornado detection by pulsed Doppler radar. Mon. Wea. Rev.,106, 29–39.
Carbone, R. E., 1982: A severe frontal rainband. Part I: Stormwide hydrodynamic structure. J. Atmos. Sci.,39, 258–279.
——, 1983: A severe frontal rainband. Part II: Tornado parent vortex circulation. J. Atmos. Sci.,40, 2639–2654.
Crum, T. D., and R. L. Alberty, 1993: The WSR-88D and the WSR-88D Operational Support Facility. Bull. Amer. Meteor. Soc.,74, 1669–1687.
Eilts, M. D., 1997: Overview of the warning decision support system. Preprints, 28th Conf. on Radar Meteorology, Austin, TX, Amer. Meteor. Soc., 402–403.
Golden, J. H., and D. Purcell, 1978: Life cycle of the Union City, Oklahoma tornado and comparison to waterspouts. Mon. Wea. Rev.,106, 3–11.
Goodman, S. J., and K. R. Knupp, 1993: Tornadogenesis via squall line and supercell interaction: The November 15, 1989, Huntsville, Alabama, tornado. The Tornado: Its Structure, Dynamics, Prediction, and Hazards, Geophys. Monogr., No. 79, Amer. Geophys. Union, 183–200.
Howard, K. W., J. J. Gourley, and R. A. Maddox, 1997: Uncertainties in WSR-88D measurements and their impacts on monitoring life cycles. Wea. Forecasting,12, 166–174.
Howieson, E. D., and G. A. Tipton, 1998: Tornadogenesis associated with the 1 July 1997 derecho—A radar perspective. Preprints, 19th Conf. on Severe Local Storms, Minneapolis, MN, Amer. Meteor. Soc., 192–195.
Kennedy, P. C., N. E. Wescott, and R. W. Scott, 1993: Single-Doppler radar observations of a mini-supercell tornadic thunderstorm. Mon. Wea. Rev.,121, 1860–1870.
Klazura, G. E., and D. A. Imy, 1993: A description of the initial set of analysis products available from the NEXRAD WSR-88D system. Bull. Amer. Meteor. Soc.,74, 1293–1311.
Lemon, L. R., D. W. Burgess, and R. A. Brown, 1978: Tornadic storm airflow and morphology derived from single-Doppler radar measurements. Mon. Wea. Rev.,106, 48–61.
Leslie, L. M., 1971: The development of concentrated vortices: A numerical study. J. Fluid Mech.,48, 1–21.
Mitchell, E. D., S. V. Vasiloff, G. J. Stumpf, A. Witt, M. D. Eilts, J. T. Johnson, and K. W. Thomas, 1998: The National Severe Storms Laboratory tornado detection algorithm. Wea. Forecasting,13, 352–366.
NOAA, 1995: The WSR-88D operators guide. NOAA, 906 pp. [Available from Documentation Section, System Support Branch, Operational Support Facility, 3200 Marshall Ave., Suite 101, Norman, OK 73072-8028.].
Nolen, R. H., 1959: A radar pattern associated with tornadoes. Bull. Amer. Meteor. Soc.,40, 277–1279.
Pfost, R. L., and A. E. Gerard, 1997: “Bookend vortex” induced tornadoes along the Natchez Trace. Wea. Forecasting,12, 572–580.
Polger, P. D., B. S. Goldsmith, R. C. Przywatry, and J. R. Bocchieri, 1994: National Weather Service warning performance based on the WSR-88D. Bull. Amer. Meteor. Soc.,75, 203–214.
Przbylinski, R. W., 1995: The bow echo: Observations, numerical simulations, and severe weather detection methods. Wea. Forecasting,10, 203–218.
Sabones, M. E., E. M. Agee, and M. Akridge, 1996: The Pulaski County and West Lafayette, Indiana tornadoes, 26–27 April 1994: A case of supercell (mesocyclone) and squall line bow-echo interaction. Preprints, 18th Conf. on Severe Local Storms, San Francisco, CA, Amer. Meteor. Soc., 746–750.
Trapp, R. J., 1999: Observations of nontornadic low-level mesocyclones and attendant tornadogenesis failure during VORTEX. Mon. Wea. Rev.,127, 1693–1705.
——, and E. D. Mitchell, 1995: Characteristics of tornadic vortex signatures detected by WSR-88D radars. Preprints, 27th Conf. on Radar Meteorology, Vail, CO, Amer. Meteor. Soc., 517–518.
——, and R. Davies-Jones, 1997: Tornadogenesis with and without a dynamic pipe effect. J. Atmos. Sci.,54, 113–133.
Vasiloff, S. V., 1993: Single-Doppler radar study of a variety of tornado types. The Tornado: Its Structure, Dynamics, Prediction, and Hazards, Geophys. Monogr., No. 79, Amer. Geophys. Union, 223–231.
Weisman, M. L., 1993: The genesis of severe, long-lived bow echoes. J. Atmos. Sci.,50, 645–670.
——, and C. A. Davis, 1998: Mechanisms for the generation of mesoscale vortices within quasi-linear convective systems. J. Atmos. Sci.,55, 2603–2622.
Wilks, D. S., 1995: Statistical Methods in the Atmospheric Sciences. Academic Press, 467 pp.
Witt, A., M. D. Eilts, G. J. Stumpf, E. D. Mitchell, J. T. Johnson, and K. W. Thomas, 1998: Evaluating the performance of the WSR-88D severe storm detection algorithms. Wea. Forecasting,13, 513–518.
Wolf, R., R. Pryzbylinski, and P. Berg, 1996: Observations of a merging bowing segment and supercell. Preprints, 18th Conf. on Severe Local Storms, San Francisco, CA, Amer. Meteor. Soc., 740–745.
Wood, V. T., and R. A. Brown, 1997: Effects of radar sampling on single-Doppler velocity signatures of mesocyclones and tornadoes. Wea. Forecasting,12, 928–938.
Vertical profiles of the maximum, gate-to-gate differential velocity, ΔV (m s−1), at a few times (corresponding to radar volume scan times) during tornado development, for the idealized, empirically determined models on which the (a) descending and (b) nondescending classification is based. The altitude zpeak of the peak differential velocity ΔVpeak within a volume scan, and altitude zlow of the differential velocity ΔVlow at the lowest elevation angle, within the same volume scan, are indicated in (a).
Citation: Weather and Forecasting 14, 5; 10.1175/1520-0434(1999)014<0625:DANTVS>2.0.CO;2
Vertical profiles of the maximum, gate-to-gate differential velocity, ΔV (m s−1), at a few times (corresponding to radar volume scan times) during tornado development, for the idealized, empirically determined models on which the (a) descending and (b) nondescending classification is based. The altitude zpeak of the peak differential velocity ΔVpeak within a volume scan, and altitude zlow of the differential velocity ΔVlow at the lowest elevation angle, within the same volume scan, are indicated in (a).
Citation: Weather and Forecasting 14, 5; 10.1175/1520-0434(1999)014<0625:DANTVS>2.0.CO;2
Vertical profiles of the maximum, gate-to-gate differential velocity, ΔV (m s−1), at a few times (corresponding to radar volume scan times) during tornado development, for the idealized, empirically determined models on which the (a) descending and (b) nondescending classification is based. The altitude zpeak of the peak differential velocity ΔVpeak within a volume scan, and altitude zlow of the differential velocity ΔVlow at the lowest elevation angle, within the same volume scan, are indicated in (a).
Citation: Weather and Forecasting 14, 5; 10.1175/1520-0434(1999)014<0625:DANTVS>2.0.CO;2
Time–height diagrams of maximum, gate-to-gate differential velocity, ΔV (m s−1), from the (a) 11 Nov 1995 tornadic storm near Jackson, MS (a nondescending TVS), and (b) 22 Jun 1995 tornadic storm near Falcon, CO (descending TVS). Bold T denotes tornado times.
Citation: Weather and Forecasting 14, 5; 10.1175/1520-0434(1999)014<0625:DANTVS>2.0.CO;2
Time–height diagrams of maximum, gate-to-gate differential velocity, ΔV (m s−1), from the (a) 11 Nov 1995 tornadic storm near Jackson, MS (a nondescending TVS), and (b) 22 Jun 1995 tornadic storm near Falcon, CO (descending TVS). Bold T denotes tornado times.
Citation: Weather and Forecasting 14, 5; 10.1175/1520-0434(1999)014<0625:DANTVS>2.0.CO;2
Time–height diagrams of maximum, gate-to-gate differential velocity, ΔV (m s−1), from the (a) 11 Nov 1995 tornadic storm near Jackson, MS (a nondescending TVS), and (b) 22 Jun 1995 tornadic storm near Falcon, CO (descending TVS). Bold T denotes tornado times.
Citation: Weather and Forecasting 14, 5; 10.1175/1520-0434(1999)014<0625:DANTVS>2.0.CO;2
(a) Number of descending and nondescending TVSs, given as a fraction of the total occurring in a given range bin. Box plots of descending and nondescending TVS attributes, as a function of radar range: (b) peak, pretornadic gate-to-gate differential velocity (m s−1); (c) height of the peak, pretornadic gate-to-gate differential velocity (km); and (d) lead time (in numbers of radar volume scans). Open circles indicate outliers, which are > the upper quartile + 1.5 × interquartile range or < the lower quartile − 1.5 × interquartile range.
Citation: Weather and Forecasting 14, 5; 10.1175/1520-0434(1999)014<0625:DANTVS>2.0.CO;2
(a) Number of descending and nondescending TVSs, given as a fraction of the total occurring in a given range bin. Box plots of descending and nondescending TVS attributes, as a function of radar range: (b) peak, pretornadic gate-to-gate differential velocity (m s−1); (c) height of the peak, pretornadic gate-to-gate differential velocity (km); and (d) lead time (in numbers of radar volume scans). Open circles indicate outliers, which are > the upper quartile + 1.5 × interquartile range or < the lower quartile − 1.5 × interquartile range.
Citation: Weather and Forecasting 14, 5; 10.1175/1520-0434(1999)014<0625:DANTVS>2.0.CO;2
(a) Number of descending and nondescending TVSs, given as a fraction of the total occurring in a given range bin. Box plots of descending and nondescending TVS attributes, as a function of radar range: (b) peak, pretornadic gate-to-gate differential velocity (m s−1); (c) height of the peak, pretornadic gate-to-gate differential velocity (km); and (d) lead time (in numbers of radar volume scans). Open circles indicate outliers, which are > the upper quartile + 1.5 × interquartile range or < the lower quartile − 1.5 × interquartile range.
Citation: Weather and Forecasting 14, 5; 10.1175/1520-0434(1999)014<0625:DANTVS>2.0.CO;2
WSR-88D KMPX scans, at 0.5° and 2° elevation, of radar reflectivity factor (dBZ) and radial velocity (m s−1) of the 1 Jul 1997 southern Minnesota tornadic squall line near Willmar, MN, at 2238 and 2248 UTC. Range rings are indicated at 20-km intervals. For reference, reflectivity and velocity color scales are provided in Fig. 6.
Citation: Weather and Forecasting 14, 5; 10.1175/1520-0434(1999)014<0625:DANTVS>2.0.CO;2
WSR-88D KMPX scans, at 0.5° and 2° elevation, of radar reflectivity factor (dBZ) and radial velocity (m s−1) of the 1 Jul 1997 southern Minnesota tornadic squall line near Willmar, MN, at 2238 and 2248 UTC. Range rings are indicated at 20-km intervals. For reference, reflectivity and velocity color scales are provided in Fig. 6.
Citation: Weather and Forecasting 14, 5; 10.1175/1520-0434(1999)014<0625:DANTVS>2.0.CO;2
WSR-88D KMPX scans, at 0.5° and 2° elevation, of radar reflectivity factor (dBZ) and radial velocity (m s−1) of the 1 Jul 1997 southern Minnesota tornadic squall line near Willmar, MN, at 2238 and 2248 UTC. Range rings are indicated at 20-km intervals. For reference, reflectivity and velocity color scales are provided in Fig. 6.
Citation: Weather and Forecasting 14, 5; 10.1175/1520-0434(1999)014<0625:DANTVS>2.0.CO;2
As in Fig. 4 except of the tornadic squall line near Waverly, MN, at 2353 and 2358 UTC. Image magnification is twice that of Fig. 4.
Citation: Weather and Forecasting 14, 5; 10.1175/1520-0434(1999)014<0625:DANTVS>2.0.CO;2
As in Fig. 4 except of the tornadic squall line near Waverly, MN, at 2353 and 2358 UTC. Image magnification is twice that of Fig. 4.
Citation: Weather and Forecasting 14, 5; 10.1175/1520-0434(1999)014<0625:DANTVS>2.0.CO;2
As in Fig. 4 except of the tornadic squall line near Waverly, MN, at 2353 and 2358 UTC. Image magnification is twice that of Fig. 4.
Citation: Weather and Forecasting 14, 5; 10.1175/1520-0434(1999)014<0625:DANTVS>2.0.CO;2
As in Fig. 4 except of the tornadic bookend vortex near St. Francis, MN, at 0054 UTC. The radar scan is at 0.5° elevation, and image magnification is twice that of Fig. 4.
Citation: Weather and Forecasting 14, 5; 10.1175/1520-0434(1999)014<0625:DANTVS>2.0.CO;2
As in Fig. 4 except of the tornadic bookend vortex near St. Francis, MN, at 0054 UTC. The radar scan is at 0.5° elevation, and image magnification is twice that of Fig. 4.
Citation: Weather and Forecasting 14, 5; 10.1175/1520-0434(1999)014<0625:DANTVS>2.0.CO;2
As in Fig. 4 except of the tornadic bookend vortex near St. Francis, MN, at 0054 UTC. The radar scan is at 0.5° elevation, and image magnification is twice that of Fig. 4.
Citation: Weather and Forecasting 14, 5; 10.1175/1520-0434(1999)014<0625:DANTVS>2.0.CO;2
As in Fig. 4 except of the tornadic bow echo near Forest Lake, MN. Radar scans are at 0.5° elevation: (a) 0104, 0109, and 0014 UTC; (b) 0044, 0049, and 0054 UTC.
Citation: Weather and Forecasting 14, 5; 10.1175/1520-0434(1999)014<0625:DANTVS>2.0.CO;2
As in Fig. 4 except of the tornadic bow echo near Forest Lake, MN. Radar scans are at 0.5° elevation: (a) 0104, 0109, and 0014 UTC; (b) 0044, 0049, and 0054 UTC.
Citation: Weather and Forecasting 14, 5; 10.1175/1520-0434(1999)014<0625:DANTVS>2.0.CO;2
As in Fig. 4 except of the tornadic bow echo near Forest Lake, MN. Radar scans are at 0.5° elevation: (a) 0104, 0109, and 0014 UTC; (b) 0044, 0049, and 0054 UTC.
Citation: Weather and Forecasting 14, 5; 10.1175/1520-0434(1999)014<0625:DANTVS>2.0.CO;2
WSR-88D KPUX scans, of radar reflectivity factor (dBZ) and radial velocity (m s−1), of the 22 Jun 1995 tornadic supercell near Falcon, CO. Range rings are indicated at 100-km intervals. Radar scans at (a) 0.5° and (b) 3.3° elevation, at 2145 UTC. (c) Radar scan at 0.5° elevation, at 2220 UTC. (d) Radar scan at 0.5° elevation, at 2237 UTC.
Citation: Weather and Forecasting 14, 5; 10.1175/1520-0434(1999)014<0625:DANTVS>2.0.CO;2
WSR-88D KPUX scans, of radar reflectivity factor (dBZ) and radial velocity (m s−1), of the 22 Jun 1995 tornadic supercell near Falcon, CO. Range rings are indicated at 100-km intervals. Radar scans at (a) 0.5° and (b) 3.3° elevation, at 2145 UTC. (c) Radar scan at 0.5° elevation, at 2220 UTC. (d) Radar scan at 0.5° elevation, at 2237 UTC.
Citation: Weather and Forecasting 14, 5; 10.1175/1520-0434(1999)014<0625:DANTVS>2.0.CO;2
WSR-88D KPUX scans, of radar reflectivity factor (dBZ) and radial velocity (m s−1), of the 22 Jun 1995 tornadic supercell near Falcon, CO. Range rings are indicated at 100-km intervals. Radar scans at (a) 0.5° and (b) 3.3° elevation, at 2145 UTC. (c) Radar scan at 0.5° elevation, at 2220 UTC. (d) Radar scan at 0.5° elevation, at 2237 UTC.
Citation: Weather and Forecasting 14, 5; 10.1175/1520-0434(1999)014<0625:DANTVS>2.0.CO;2
Description of dataset. Time, azimuth, and range, all refer to tornado formation, with respect to the radar listed in the third column. F scale is the estimate of maximum tornado intensity based on the Fujita scale. Designator D (ND) refers to descending (nondescending) TVS, and C, L, and O, refer to parent storm type of cell, convective line, or other, respectively.
Mean values, and standard error about those means, of attributes of 25 descending (D) and 27 nondescending (ND) TVSs and associated tornadoes, and also of 6 nondescending TVSs/tornadoes that formed within convective lines (ND line).
Contingency table of TVS character (descending, D, vs nondescending, ND) with respect to parent storm type (convective line, cell, or other). Conditional probabilities of TVS character given radar-detectable tornadogenesis within a particular parent storm type are shown in parentheses.
The current WSR-88D TVS algorithm requires a shear value of 72 h−1 (which now may be reduced to 18 h−1), computed from the maximum inbound and outbound radial velocity within a mesocyclone. According to the WSR-88D Operator’s Guide (NOAA 1995), a “manual” or “user-defined” TVS is associated with a gate-to-gate velocity difference that exceeds 36 m s−1 (70 kt) at ranges between 56 and 93 km (30–50 n mi), through at least the lowest two elevation angles [or exceeds 46 m s−1 (90 kt) at ranges <56 km (30 n mi)].
We show in section 3 that we can achieve statistically significant results, regarding the proportion of descending TVSs, using a sample size of 52.
We treat the beginning time of the volume scan in which the tornado was reported to form as the time of tornadogenesis.
Interpreting storm evolution from changes in the height of echo top must be done cautiously, because of the artifacts of radar sampling (see Howard et al. 1997).
