1. Introduction
Recent decades of scientific advancement have resulted in a much greater understanding of tornado formation, although challenges still remain. Along with this increased knowledge has come the realization that tornadoes can actually form in several different ways. This is illustrated in part by the evolution of the definition of a tornado. In 1959, the Glossary of Meteorology (Huschke 1959) defined a tornado as “a violently rotating column of air, pendant from a cumulonimbus cloud.” In 2000, the second edition of the Glossary (Glickman 2000) defines a tornado as “a violently rotating column of air, in contact with the ground, either pendant from a cumuliform cloud or underneath a cumuliform cloud.” This new definition illustrates the acceptance of a broader range of vortex events such as landspouts, waterspouts, and gustnadoes.
Whether or not tornadoes are defined according to their causative mechanisms or according to their physical appearance can/should make a difference in their potential identification and classification in Storm Data. For instance, if any cumuliform cloud occurring over water creates a columnar vortex, this vortex is defined as a waterspout and can only be labeled a tornado if it travels onto land. Consequently, waterspouts can occur with either thunderstorms (even supercells) or with weaker nonglaciated convective clouds. However, the landspout (see Bluestein 1985) is always called a tornado and is more typically produced by nonglaciated convective clouds (although such clouds can grow to the glaciated stage). These landspouts are typically produced by slow-moving cumulus congestus clouds or by seemingly nonsevere thunderstorms, with vorticity spinup in the boundary layer (which is not in association with a mesocyclone). These landspouts receive their name due to their one-cell laminar vortex structure (referred to in fluid mechanics terminology as a soda-straw vortex) and, additionally, because they have a similar physical appearance to the well-known waterspout. Additional confusion in tornado recognition comes with the gustnado phenomenon, an inertial instability that forms along a vortex sheet that is produced locally along the gust front of any thunderstorm (although typically more prevalent in the outflow of stronger thunderstorm events). Forbes and Wakimoto (1983) go far in documenting the occurrence of such vortex events in a severe weather outbreak near Springfield, Illinois, on 6 August 1977.
It is also noted that the strongest tornadoes (F5/EF5) are most likely produced by mesocyclonic supercell storms (see Trapp et al. 2005), such as the events in Moore, Oklahoma, on 3 May 1999; in Greensburg, Kansas, on 4 May 2007; and in Parkersburg, Iowa, on 25 May 2008. These tornadoes (defined as type I) are produced by a strongly rotating updraft interacting with a rear-flank downdraft, although most supercell tornadoes are weaker and some may not require the RFD for tornadogenesis. Other types of tornadoes of significant strength (F2/EF2–F4/EF4) can be produced by a number of mechanisms including the tilting of solenoidal horizontal vorticity by a cold pool, and by rear-inflow jets (RIJs) in quasi-linear convective systems (QLCS) (Taylor et al. 2002; Trapp and Weisman 2003; Weisman and Trapp 2003; Trapp et al. 2005), as well as by bookend vortices (BEVs) in bow echoes (also see Fujita 1978). Even nonsupercell tornadoes of the landspout variety can achieve intensities as high as F2 (see Wakimoto and Wilson 1989).
Most meteorologists would likely say that every vortex event associated in any manner with any type of thunderstorm or convective cloud is a tornado (i.e., Glickman 2000). However, this approach could be viewed as counterproductive in establishing an appropriate system for the classification and archiving of tornado events. Florida, for example, has an elevated tornado count due to the frequency of landfalling waterspouts. Further, in this era of global warming with the anticipated increase in CAPE and convective storms in the twenty-first century (see Solomon et al. 2007; Trapp et al. 2007), it might be useful to separate the more locally produced weaker vortex events from those more directly associated with supercells (type I) and QLCS (type II). The predicted increase in the number of severe thunderstorms in the United States in future decades offers the opportunity to see potential changes in the occurrence and distribution of the different tornado types, if an appropriate database can be established.
The objective of this paper is to introduce definitions for three types of tornado classifications, including 15 tornado species, thus offering a more detailed identification methodology with improved labeling of events in Storm Data archives. Such an improvement would also provide additional meteorological value to Storm Data reports, in addition to the engineering estimates of wind speeds based on damage assessment. This classification taxonomy is somewhat consistent with the views provided by Davies-Jones et al. (2001). Individual states in the United States would then have the total count for all tornado types, and better representation could be made in assessing any future climatological trends of tornadoes, including national and regional differences. It is further proposed that National Oceanic and Atmospheric Administration (NOAA) adopt this taxonomy for their reporting practices (which is clarified in more detail in the following sections).
2. The type I vortex event
The nomenclature introduced for the discrete supercell tornado event is the type I vortex, which represents the fundamental (and most complicated) series of meteorological events that lead to tornado formation. In this case the supercell thunderstorm possesses the key ingredient of a strong local updraft (SLU) extending vertically through a horizontal wind field that is rich in horizontal vorticity, resulting in the development of a well-defined mesocyclone. An example of this precursive condition to type I tornado formation is shown in Fig. 1. Doppler radar does well in the identification of these mesocyclones [see National Weather Service (NWS) definitions] and these can often be detected without the subsequent formation of a tornado since the strongest rotation is usually above the planetary boundary layer. Another key development in the series of events for type I tornado formation is the rear-flank downdraft (RFD) that strategically produces vertical vorticity in the PBL, in close proximity to and interacting with the SLU. The combined interaction of the SLU and the RFD, resulting in the tilting of horizontal vorticity and the stretching of vertical vorticity, creates the perfect environment for tornadogenesis. Subsequently, this tornado cyclone (see Agee et al. 1976; Glickman 2000) can become increasingly stronger [even of tornado vortex signature (TVS) strength] and type I tornado formation is completed. It is noted, however, that type I tornadoes range over the full spectrum of tornado intensity from F0 to F5 (now EF0 to EF5) and RFD involvement is not always required for tornadogenesis. If one examines the Beltrami vorticity equation, it is evident that several processes can affect the local reservoir of horizontal and vertical vorticity. These include the convergence and stretching of existing vertical vorticity; the tilting and stretching of horizontal vorticity; the role played by baroclinicity and preexisting boundaries, which are also favorable sources for eventual concentration of vorticity into the developing tornado (see Houston and Wilhelmson 2007); and the effects of frictionally induced convergence. The intent here is to not explain the variety of intricate details in the evolution of a type I tornado, but to present the infrastructure or boilerplate format for identification (and thus classification). In summary, the type I tornado is produced by a discrete mesocyclonic supercell storm with an SLU that interacts favorably with the environmental wind field. This process moves sequentially through the strong mesocyclone formation (often observed with Doppler radar) and can subsequently result in spinup in the PBL (either through the RFD or other boundary layer processes), thus completing the formation of the tornado vortex column (typically in the right-rear quadrant of the parent thunderstorm). The presence of the parent storm cell in a field of strong helicity (V·∇ × V) adds to the rotational properties of the SLU and the intensification of the mesocyclone and the likelihood of tornadogenesis. It is further noted that the movement of the parent storm cell with respect to the environmental wind field can maximize the intake of streamwise vorticity (i.e., storm-relative helicity) particularly in the lowest 500–1000 m (see Thompson et al. 2003). The combined properties of CAPE and storm-relative helicity (see Johns and Doswell 1992) are critical to the formation of severe rotating thunderstorms that are capable of producing the type I tornado (as well as the type II tornado discussed below). Such events can also occur in low tropopause conditions with high helicity and minimal CAPE, resulting in “low top” strongly rotating minisupercells (see Kennedy et al. 1993). It is further noted that hurricane environments can produce tornadoes from shallow supercells (see McCaul 1987; McCaul and Weisman 1996).
3. Type II vortex events
The type II vortex will represent those tornadoes that are produced by QLCS, which is typically an MCS that has taken the form of a squall line or bow echo. The concept of a meteorological event known as the QLCS was first introduced by Weisman and Davis (1998) and the tornado climatology associated with QLCS was established by Trapp et al. (2005). Interestingly, the definition of the QLCS was not listed in the latest edition of the Glossary of Meteorology (Glickman 2000). The QLCS concept will include all “lines” of convective storms with features such as line echo wave patterns (LEWPs), bows, BEVs, RIJs, or any type of mesovortex event. The terminology used in this paper should not be confused with that presented by Gallus et al. (2008), as all nine categories of their storm morphologies basically equate to lines or cells. One could argue that a “perfectly linear” system would not produce a type II tornado, since each of the above features/events results in some kind of irregularity in the line. The “cold pool process” is typically fundamental in producing the tilting of horizontal vorticity, thus aiding in tornado formation. An example of a bow structure that developed within a line is shown in Fig. 2a. Also, the bow feature can often produce a BEV as illustrated in Fig. 2b. Further, the mesoscale development of RIJs can add to the production of shear-induced mesovortices that form along the QLCS. As stated earlier, the type II tornadoes can be as strong as most of the type I tornadoes (but generally not beyond F4/EF4 intensity). Finally, another example of a QLCS with type II tornadoes is shown in Fig. 2c, which contains a number of mesovortices that developed in association with tornadogenesis [a phenomenon that has been numerically simulated by Trapp and Weisman (2003)].
4. Type III vortex events
As stated earlier, both the type I vortex (supercells) and the type II vortex (QLCS) events will be labeled as tornadoes with a capability of producing significant damage (≥F2/EF2). Definitions and conceptual discussion of type I and II events have been presented (and along with type III events are summarized in a formal taxonomy presented in the next section). Interestingly, in the type III tornado the act of vortex formation is somewhat uniquely controlled by the updraft (e.g., in nonglaciated cumuliform clouds) or by the downdraft (e.g., in association with the gust-front boundary), typically resulting in weaker tornado events. Thus, the waterspouts, landspouts, and gustnadoes would fall into a third category of tornado events (type III). Stretching of shear vortices in the PBL by the updraft of nonsupercells has, in extreme cases, produced F3 tornadoes (see Wakimoto and Atkins 1996). The best example of a locally produced, shear-driven vortex is the gustnado (see Fig. 3). This type III vortex event can produce tornado damage, but typically is rather benign and only observed in dusty conditions along thunderstorm gust front boundaries. The gustnado, however, can experience a metamorphic process, where it is advected into the core of a strong mesocyclone and subsequently evolves into a type I tornado (see Bluestein et al. 2002). The concept of a gustnado was presented some years ago (e.g., see Doswell and Burgess 1993). One of the authors can personally recall a talk by D. Burgess at the 11th Severe Storms Conference with a slide presentation showing gustnadoes along a dusty gust front in an Oklahoma thunderstorm. Crowley (2008) presents an excellent illustration of our need to properly identify tornado events and separate out the gustnadoes. The proposed type I, II, and III nomenclature presented here will address this issue. Further, tornado statistics will be more accurately presented for each state when totals are given for each tornado type. Smith (2008) also shows the dilemma of vortex types and the need for appropriate classification.
The landspout (see Bluestein 1985) is another type III vortex that is typically associated with a weak cumuliform cloud that has often not even achieved the glaciated stage. Radar observations show no indication of any suspect severe weather or tornado development [as seen in Crowley (2008) and in Smith (2008)]. Figure 4 shows a typical landspout, photographed by a local citizen near West Lafayette, Indiana, on 25 June 2006, which was produced by a slow-moving cumuliform cloud. This landspout was observed by one of the authors (as the “parent” cloud drifted from the southeast to the northwest in light wind conditions) and certainly was viewed as similar to the waterspout events in the Florida Keys. Local Doppler radar gave no indication of any significant convection, let alone a tornado circulation. Such an event is not attributable to strong low-level vertical shear in the horizontal wind. Thus, it is different from the gustnado and any inertially driven instability vortices that form in the boundary layer (but all such events would be classified as type III tornadoes and, thus, are distinctly different from the type I and II vortex events).
Finally, some brief comments are offered about hurricane-induced tornadoes (HITs). The occurrences of type I and II HITs are noted, and these are the more common types of hurricane-induced tornadoes. However, if vortices develop in the eyewall region, they would be viewed as a shear-induced inertial instability and thus represent type III tornadoes (see Fujita 1993; Montgomery et al. 2002).
5. Taxonomy
Based on the above introductions and descriptions of types I, II, and III tornadoes, it is now appropriate to present an organized taxonomy for classifying tornadoes. Figure 5 shows the organization of the three proposed types of tornadoes, which encompass 15 identifiable tornado species.
a. Type I
The type I tornado is produced by the discrete supercell mesocyclone, as well as the low-top minisupercell (previously discussed). It is also noted that the supercell is typically severe right moving (SR), but in rare instances it can be severe left moving (SL), depending on the characteristics of the wind hodograph. Supercells that split into SR and SL storms have the potential to develop cyclonic and anticyclonic type I tornadoes, respectively. Third, another type I tornado is produced by the low-top minisupercell associated with landfalling hurricanes. These tornadoes are typically associated with convective cells located in the right-front quadrant of the hurricane (and are somewhat rare). Important to the development of discrete supercells is the presence of low-level directional shear (as well as magnitude shear), a distinguishing feature in cells versus lines.
b. Type II
The type II tornado is produced by a variety of physical processes that are associated with QLCS. These tornadoes occur in association with mesocyclones and other mesovotices that arise in conjunction with cold pool dynamics, embedded convective cells, bow echoes and bookend vortices, rear-inflow jets, and line echo wave patterns (previously discussed). In addition, spiral bands in the right-front quadrant of a landfalling hurricane can produce tornadoes of the QLCS type. There is a total of six tornado species proposed as type II events, as illustrated in Fig. 5.
c. Type III
The type III tornadoes are all of those events not classified as type I (supercells) or type II (QLCS). More specifically, these are localized convective and shear vortices. The type III convective events include landspouts, waterspouts, and cold air funnels. The type III shear events are gustnadoes, anticyclonic secondary vortices, and shear vortices associated with inertial instabilities in the eyewall of hurricanes. In addition, it is noted that type III tornadoes can occur in association with supercells and QLCSs (namely gustnadoes), even when no type I or II tornadoes occur. Further, it is noted that type I and II tornadoes can occur over water and thus, according to the Glossary of Meteorology (Glickman 2000), would be labeled as waterspouts, but in this classification system these events (with landfall) would not be labeled as type III tornadoes.
d. Guidelines for classification
Finally, Table 1 provides a description for each of the 15 tornado species and is intended to offer guidance for the classification of tornadoes. As noted above, there are three species in type I, six species in type II, and six species in type III.
6. Summary and conclusions
The concepts and descriptions of type I, II, and III tornado events have been presented and are consistent with the definition of tornadoes provided in the Glossary of Meteorology (Glickman 2000), as well as the views presented by Davies-Jones et al. (2001). The type I tornado is produced by the discrete supercell thunderstorm, where a mesocyclone has formed due to an SLU interacting with a horizontal wind field with large values of horizontal vorticity. The resulting tilting and stretching processes create favorable conditions for interaction with RFD development or any other processes that help with vortex spinup in the PBL. If supercells merge into a line, and thereafter a tornado is produced, the event would be classified as a type II tornado (since there would be a QLCS structure at the time of tornadogenesis).
The type II tornado events are defined as those associated with QLCS, and such events are often associated with cold pool and shear-induced mesovortex events that form along the line. Some of these type II tornadoes can also be associated with bow echoes and BEVs that often develop in the comma-head. RIJs along the line (e.g., see Taylor et al. 2002) can also add to shear-induced vertical vorticity and tornadogenesis. The type III tornado is produced by locally concentrated updrafts in weaker cumuliform clouds, as well as by local wind shear effects in the planetary boundary layer associated with the gust fronts of stronger thunderstorm events. Therefore, the occurrence of waterspouts, landspouts, cold air funnels, and gustnadoes would typically be classified as type III tornadoes. In general, the class of type I tornadoes (supercells) and type II tornadoes (QLCS) would be stronger than those classified as type III tornadoes. It is also noted that an individual supercell could “parent” both type I and III tornadoes, and similarly for QLCS events.
Further, it is recommended that NOAA adopt this taxonomy when archiving tornado events in Storm Data, drawing on the expertise of NWS personnel in the interpretation of radar data, field observations, and visual evidence. Undoubtedly, some tornado events would be difficult to classify due to a lack of appropriate supporting information. This “typing” could also allow the NWS to issue more informative tornado warnings and severe weather advisories. Further, a more meaningful tornado climatology can potentially be established that better represents the different types of tornadoes both nationally and regionally, as well as for individual states (e.g., Florida versus Oklahoma). This tornado climatology could also be useful in assessing the effects (if any) of the climatic warming trend expected to continue through the twenty-first century and the hypothesized associated increase in severe thunderstorm activity (see Trapp et al. 2007). Decades of tornado typing could allow for the detection of changes in tornado patterns that could potentially be linked to climate change or other cyclical weather patterns.
Acknowledgments
The authors are grateful to the two anonymous reviewers that first reviewed this manuscript when it was originally sent to the Monthly Weather Review. Their comments have been most helpful and have contributed much to the improvement in this revised manuscript. One reviewer recommended acceptance (with major revisions) in Monthly Weather Review, and the other reviewer recommended that the paper be sent to Weather and Forecasting in order to be exposed to a larger and more appropriate audience. The authors are further grateful to the three Weather and Forecasting reviewers who provided constructive criticism and made several suggestions for improving the revised manuscript.
REFERENCES
Agee, E. M., Snow J. T. , and Clare P. R. , 1976: Multiple vortex features in the tornado cyclone and the occurrence of tornado families. Mon. Wea. Rev., 104 , 552–563.
Bluestein, H. B., 1985: The formation of a “landspout” in a “broken-line” squall line in Oklahoma. Preprints, 14th Conf. on Severe Local Storms, Indianapolis, IN, Amer. Meteor. Soc., 267–270.
Bluestein, H. B., Weiss C. C. , Pazmany A. L. , Lee W-C. , and Bell M. , 2002: Tornadogenesis and tornado-vortex structure in a supercell. Preprints, 21st Conf. on Severe Local Storms, San Antonio, TX, Amer. Meteor. Soc., 12.1. [Available online at http://ams.confex.com/ams/SLS_WAF_NWP/techprogram/paper_46228.htm.].
Crowley, D. J., cited. 2008: Is it a tornado? NWS Southern Region Tech. Attachment SR/SSD 96-39. [Available online at http://www.srh.noaa.gov/topics/attach/html/ssd96-39.htm.].
Davies-Jones, R., Trapp R. J. , and Bluestein H. B. , 2001: Tornadoes and tornadic storms. Severe Convective Storms, Meteor. Monogr., No. 50, Amer. Meteor. Soc., 167–221.
Doswell C. A. III, , and Burgess D. W. , 1993: Tornadoes and tornadic storms: A review of conceptual models. The Tornado: Its Structure, Dynamics, Prediction and Hazards, Geophys. Monogr., Vol. 79, Amer. Geophys. Union, 161–172.
Forbes, G. S., and Wakimoto R. M. , 1983: A concentrated outbreak of tornadoes, downbursts and microbursts, and implications regarding vortex classification. Mon. Wea. Rev., 111 , 220–235.
Fujita, T. T., 1978: Manual of downburst identification for Project Nimrod. Satellite and Meso-Meteorology Research Proj Rep. 156, Dept. of Geophysical Science, University of Chicago, 104 pp.
Fujita, T. T., 1993: Wind fields of Andrew, Omar, and Iniki, 1992. Preprints, 20th Conf. on Hurricanes and Tropical Meteorology, San Antonio, TX, Amer. Meteor. Soc., 46–49.
Gallus W. A. Jr., , Snook N. A. , and Johnson E. V. , 2008: Spring and summer severe weather reports over the Midwest as a function of convective mode: A preliminary study. Wea. Forecasting, 23 , 101–113.
Glickman, T. S., Ed. 2000: Glossary of Meteorology. 2nd ed. Amer. Meteor. Soc., 855 pp.
Houston, A. L., and Wilhelmson R. B. , 2007: Observational analysis of the 27 May 1997 central Texas tornadic event. Part I: Prestorm environment and storm maintenance/propagation. Mon. Wea. Rev., 135 , 701–726.
Huschke, R. E., 1959: Glossary of Meteorology. 1st ed. Amer. Meteor. Soc., 638 pp.
Johns, R. H., and Doswell C. A. III, 1992: Severe local storms forecasting. Wea. Forecasting, 7 , 588–612.
Kennedy, P. C., Westcott N. E. , and Scott R. W. , 1993: Single-Doppler radar observations of a mini-supercell tornadic thunderstorm. Mon. Wea. Rev., 121 , 1860–1870.
McCaul E. W. Jr., , 1987: Observations of the Hurricane “Danny” tornado outbreak of 16 August 1985. Mon. Wea. Rev., 115 , 1206–1223.
McCaul, E. W., and Weisman M. L. , 1996: Simulations of shallow supercell storms in landfalling hurricane environments. Mon. Wea. Rev., 124 , 408–429.
Montgomery, M. T., Vladimirov V. A. , and Denissenko P. V. , 2002: An experimental study on hurricane mesovortices. J. Fluid Mech., 471 , 1–32.
NCDC, 2006: Storm Data. Vol. 48, No. 6, 486 pp.
Smith, R., cited. 2008: Non-supercell tornadoes: A review for forecasters. NWS Southern Region Tech. Attachment SSD96-8. [Available online at http://www.srh.noaa.gov/topics/attach/html/ssd96-8.htm.].
Solomon, S., Qin D. , Manning M. , Marquis M. , Avery K. , Tignor M. M. B. , Miller H. L. Jr., and Chen Z. , Eds. 2007: Climate Change 2007: The Physical Science Basis. Cambridge University Press, 996 pp.
Taylor, J., Lashley S. L. , Smith R. E. , Murphy P. B. , and Logsdon J. A. , 2002: The 24 October 2001 tornado outbreak. Preprints, 21st Conf. on Severe Local Storms, San Antonio, TX, Amer. Meteor. Soc., 14.4. [Available online at http://ams.confex.com/ams/pdfpapers/46381.pdf.].
Thompson, R. L., Edwards R. , Hart J. A. , Elmore K. L. , and Markowski P. , 2003: Close proximity soundings within supercell environments obtained from the Rapid Update Cycle. Wea. Forecasting, 18 , 1243–1261.
Trapp, R. J., and Weisman M. L. , 2003: Low-level mesovortices within squall lines and bow echoes: Part II. Their genesis and implications. Mon. Wea. Rev., 131 , 2804–2823.
Trapp, R. J., Tessendorf S. A. , Godfrey E. S. , and Brooks H. E. , 2005: Tornadoes from squall lines and bow echoes. Part I: Climatological distribution. Wea. Forecasting, 20 , 23–34.
Trapp, R. J., Diffenbaugh N. S. , Brooks H. E. , Baldwin M. E. , Robinson E. D. , and Pal J. S. , 2007: Changes in severe thunderstorm environment frequency during the 21st century caused by anthropogenically enhanced global radiative forcing. Proc. Natl. Acad. Sci. USA, 104 , 19719–19723. doi:10.1073/pnas.0705494104.
Wakimoto, R. M., and Wilson J. W. , 1989: Non-supercell tornadoes. Mon. Wea. Rev., 117 , 1113–1140.
Wakimoto, R. M., and Atkins N. T. , 1996: Observations on the origins of rotation: The Newcastle tornado during VORTEX 94. Mon. Wea. Rev., 124 , 384–407.
Weisman, M. L., and Davis C. A. , 1998: Mechanisms for the generation of mesoscale vortices within quasi-linear convective systems. J. Atmos. Sci., 55 , 2603–2622.
Weisman, M. L., and Trapp R. J. , 2003: Low-level mesovortices within squall lines and bow echoes. Part I: Overview and dependence on environmental shear. Mon. Wea. Rev., 131 , 2779–2803.
A Doppler radar image of a supercell thunderstorm near Oklahoma City, OK, on 3 May 1999. (left) A nice example of a hook-echo radar feature that often accompanies the mesocyclone. (right) The Doppler velocity field and the color velocity couplet that locates a TVS.
Citation: Weather and Forecasting 24, 2; 10.1175/2008WAF2222163.1
Examples of different QLCS structures that are viewed as being capable of type II tornadogenesis: (a) an example of a bow echo near San Angelo, TX, on 7 Apr 2002; (b) a bow echo and associated comma head over central IL on 4 Jun 2002, when a weak tornado (F0) was reported by the IL State Police; and (c) QLCS in southwest IL during the Bow Echo and Mesoscale Convective Vortex Experiment at 2300 UTC 6 Jun 2003. Several tornadoes were produced by this event.
Citation: Weather and Forecasting 24, 2; 10.1175/2008WAF2222163.1
Gustnado in Australia on 7 Apr 2007, which formed along a dusty thunderstorm gust front.
Citation: Weather and Forecasting 24, 2; 10.1175/2008WAF2222163.1
Dusty landspout near West Lafayette, IN, on 25 Jun 2006. No precipitation occurred near this vortex, and radar gave no indication of any possibility of severe weather (see also NCDC 2006).
Citation: Weather and Forecasting 24, 2; 10.1175/2008WAF2222163.1
Schematic of tornado types and associated species.
Citation: Weather and Forecasting 24, 2; 10.1175/2008WAF2222163.1
Criteria for applying tornado taxonomy.
