sign in sign up
<<
Home , Cyclogenesis, Cyclone, Inflow (meteorology), Mesocyclone, Rear flank downdraft, Storm chasing

2.2 IN – WHAT WE KNOW AND WHAT WE DON’T KNOW

Robert Davies-Jones* National Severe Storms Laboratory, NOAA Norman,

1. COMMON OBSERVATIONS Fawbush and Miller (1954) identified environmental conditions The advent of Doppler , 3D that are conducive to large, long-lived numerical simulations and scientific tornadoes. The sounding generally chasing have led to huge has large convective available advances in our knowledge of potential (CAPE) and strong during the last 35 years. shear associated with veering However, complete understanding of and increasing with height. Although tornadogenesis in supercell storms still both shear and CAPE are important eludes researchers. To put my parameters, theoreticians were discussions of current knowledge and bewildered by cases when tornadoes mysteries in context, I follow formed in environments either with Rasmussen and Straka’s (1997) high CAPE and little shear or with low example by first describing features of CAPE and high shear. The former tornadic supercells that are observed events occurred either in hurricanes or with quite remarkable repetition from in mid- during the cool storm to storm and then reviewing , and the latter ones in mid- viable theories that are congruent with latitudes during the warm season. We these observations. Some of the now know that the low-CAPE, high- following is based on their paper. shear tornadoes form in mini-supercells Background on tornadic storms is with low tops and the high-CAPE, low contained in the most recent review shear tornadoes develop in supercells article (Davies-Jones et al. 2001) and that are not very steady. references cited therein. 1.2 ALOFT 1.1 STORM AND ENVIRONMENT The initial mesocyclone is a rotating updraft with maximum vertical From detailed examination of radar -2 -1 data, which had rather coarse exceeding 10 s and a large resolution by today’s standards, correlation between vertical velocity Browning (1964) concluded that most and vertical vorticity. It forms aloft from tornadoes form within large, steady, tilting of low-level storm-relative and virulent . These streamwise vorticity associated with appeared to be unicellular, so he vertical shear in the large-scale or called them supercells. mesoscale environment. The resulting The introduction of pulsed Doppler vertical vorticity is subsequently into severe-storm research in stretched and advected vertically in 1971 quickly led to new results. the updraft. The storm may amplify the Signatures of and low-level streamwise vorticity in its tornadic vortices were evident in the by stretching it horizontally. Doppler velocity fields of a single Helical environments (i.e., ones with radar. The mesocyclone typically large shear vectors that veer with formed aloft first and then near the height in the lowest few kilometers), ground prior to any tornadogenesis. either on a large scale or around ______baroclinic boundaries are especially *Corresponding author address: conducive to tornadoes. Very large Robert Davies-Jones, National Severe helicity in the lowest 1 km is particularly Storms Laboratory, NOAA, 1313 favorable for tornadoes. This is Halley Circle, Norman, Oklahoma probably because the mesocyclone 73069-8493; aloft has a base that is not far off the e-mail: [email protected] ground. cooling or from a downward 1.3 NEAR-GROUND MESOCYCLONE nonhydrostatic vertical pressure- gradient (NHVPGF). With it In 1949 Brooks (1949) discovered propagates around the rotating from passages near updraft, and its boundary is microbarographs that tornadoes the RFD gust front. The mesocyclone formed within larger , which he at low levels now is divided into updraft named tornado cyclones and which and low base on its left side, nowadays are termed low-level and downdraft (the left side of the mesocyclones irrespective of whether RFD) on its right. (Lemon and Doswell or not they produce tornadoes. In 1979). The right side of the RFD is comparison to mid-level meso- anticyclonic and so lies outside the , near the ground mesocyclone. Convergence at the develops later in the supercell’s lifetime RFD gust front causes the updraft to and by a different process. It seems extend along it and thus become to await the formation of downdrafts horseshoe shaped. At low levels the within the mesocyclone. supercell now resembles an extra- Fujita (1959) collected tropical with the outflow photographs and movies of the cloud boundaries from the FFD and the RFD base and sides of a 1957 tornadic in the warm-frontal and cold-frontal storm near Fargo, , and positions, respectively. The RFD is found that the entire updraft was often visible as a clear slot, a narrow rotating cyclonically (Fig. 1). This was deep slot of cloud-free air that wraps the first visual confirmation of a around the region where the tornado mesocylone. develops about 5-10 min prior to its formation. Rapidly sinking and 1.4 REAR FLANK-DOWNDRAFT evaporating cloud fragments are often seen near its edges. The advancing The data from two radars with RFD occludes the mesocyclone with different viewing angles allowed the the tornado typically forming near the construction of 3D fields inside point of occlusion. At this stage a new severe storms. It soon became mesocyclone may be forming along evident that downdrafts played a role the bulge in the RFD to the right of the in tornadogenesis. Dual-Doppler old one (Fig. 2). analyses and observations by storm chasers revealed that tornadoes 1.5 formed not in the early lifetimes of a supercell when it consisted almost With the advent of radar entirely of a rotating updraft but later came the recognition that tornadic on after downdrafts reached the storms close to the radar often had an ground. Low- (LP) echo with a hook-shaped appendage supercells seldom produce tornadoes, on the right rear side (Stout and Huff probably because they never develop 1953, Markowski 2002). The hook is a significant downdraft. often attributed to precipitation being A supercell is generally considered drawn into a cyclonically rotating to have two main downdrafts (Fig. 2) curtain by the mesocyclone (Browning even though they may not be a gap 1964). If this were true, the hook between them. The forward flank would elongate gradually. However, it downdraft (FFD) forms first and is often seems to form all at once, collocated with precipitation on the left suggesting that it may be associated front side of the updraft in the northern with the sudden development of hemisphere. [For southern-hemisphere precipitation at the rear of a new supercells interchange left and right updraft. The hook is often narrow, throughout this paper.] The rear-flank consistent with chasers’ observations downdraft (RFD) develops at the rear of thin curved rain curtains with of the rotating cloud tower, which may hydrometeors that are advected have a quite circular base initially. The horizontally by the mesocyclonic winds RFD may arise from evaporative as they fall. At the ground the curtains typically consist of large drops. In his study of the Fargo storm, Surface measurements reveal Fujita (1959) realized that an abrupt divergent flow towards and away from lowering of the the mesocyclone axis and locally base was a significant feature and higher pressure in the region of the named it the . It marked the hook. The tip of the hook may flare lower portion of a strong rotating out both cyclonically to the right and updraft (Fig. 1). Supercell tornadoes anticyclonically to the left. are generally suspended from wall until they are overtaken by 1.6 TORNADO CYCLONE divergent air late in their . Rotating wall clouds containing strong Tornadoes have circulations that are upward motions often precede an order of magnitude smaller than the tornadoes by tens of minutes (Fujita circulation of a typical mature 1959). Based on numerical mesocyclone. Fortunately is simulations, Rotunno and Klemp only able to contract a part of a strong (1985) attribute their formation to rain- mesocyclone into a tornado. This part cooled, nearly saturated air that is thought to be a with a typical descends, and flows along the ground core radius of 1 km within the into the updraft. The lowering is due mesocyclone, and is named a tornado to this air having a lower lifted cyclone [not to be confused with level than air in the Brook’s (1949) tornado cyclone in storm’s inflow. Section 1]. The tornado cyclone has been observed occasionally in high- 1.9 TORNADO POSITION IN STORM resolution observations. In one case (Burgess et al. 2005) it Neil Ward chased one of the developed though a deep column tornadic supercells studied by within an intensifying mesocyclone. Its Browning (Browning and Donaldson maximum winds were aloft and a 1963). He observed that the tornado built upward from within it. tornadoes formed in the storm’s main The co-existence of a tornado and updraft and near its gust front. tornado cyclone is also evident in Chasers observed that tornadoes observations of usually form in a region of the storm tornadoes with secondary wind that is a few kilometers away from maxima at distances of 500-1000 m precipitation and the nearest cloud-to- from the axis (Wurman and Gill 2000). ground (Davies-Jones and Thus, the entire tornado cyclone may Golden 1975; Fig. 3). Thus, theories not contract into a tornado. that rely on electrical heating from repeated lightning strikes (Vonnegut 1.7 OCCLUSION DOWNDRAFT 1960) or on a precipitation-loaded downdraft along the axis of rotation The occlusion downdraft is a small- (Eskridge and Das 1976) have no scale downdraft that forms after the observational support. development of intense rotation and Doppler-radar analyses and attendant low pressure next to the chasers’ observations revealed that ground (Rotunno and Klemp 1985). tornadoes generally form close to the Thus, it is a response to the near- circulation center of the mesocyclone ground rotation rather than an and near the interface of updraft and instigator of it. It is driven by strong downdraft. Roughly speaking, they downward NHVPGF associated with a are also near the center of curvature deep low at the surface. Visually, it is of the wrapping rain curtain where the practically indistinguishable from the inward outflow from the associated RFD because it is located near the downdraft converges. front edge of the RFD as it wraps Observations from chase teams around the developing tornado. showed that the tornado below was coincident in time and 1.8 WALL CLOUD with the tornadic-vortex signature (TVS). For the first time could observe tornadoes (or more vertical vorticity by a large persistent precisely their signatures) above cloud and strong updraft or whether it base (Brown et al. 1978). Large originated ultimately from tilting of violent tornadoes were associated with horizontal vorticity. TVSs that extended to near the Stretching of planetary vorticity is tropopause. Note that, if the radar is at odds with the observations because not close enough to resolve the parent it would produce rotation near the tornado cyclone, the TVS may be due ground first where horizontal at least partly to the tornado cyclone. convergence is largest. In contrast, In numerical simulations of Browning and Landry (1963) supercells, vertical motions intensify hypothesized that updrafts in and pressure falls at 1-3 km above supercells rotated cyclonically by tilting ground just prior to tornadogenesis updraft-relative streamwise vorticity (Wicker and Wilhelmson 1995). At present in their inflow. This these levels the poorly resolved mechanism would produce a tornado forms in large gradients of mesocyclone aloft, as observed vertical velocity where the tilting term is initially. Lilly (1982) and Davies-Jones large. (1984) developed mathematical theories of this process. 1.10 ANTICYCLONIC VORTICITY The first successful 3D numerical simulations of supercells settled this At low levels, anticyclonic vorticity debate (Klemp and Wilhelmson 1978). is often located near tornadoes. An The model storms resembled anticyclonic flare is occasionally supercells even with the ’s observed at the tip of the hook and is rotation switched off. In simulations visible in the as anticyclonic with unidirectional (straight) shear, rotation in the base of the flanking storms split as often observed into cloud line to the right of the clear slot severe right- (SR) and left-moving (SL) (Fig. 2). Sometimes an anticyclonic supercells. The updraft in the SR (SL) tornado forms at this location, usually storm rotated cyclonically (anti- when there is a stronger cyclonic cyclonically). Inclusion of Coriolis tornado located on the other (left) side simply made the right mover the of the hook (Knupp and Brown 1980). stronger storm, but only slightly In other cases, the region of because the for anticyclonic vorticity advances around supercells has an order of magnitude a strong cyclonic tornado as the RFD of 100. Veering of the shear vector wraps around the tornado. with height enhanced the right-moving In his photogrammetric analysis of updraft and inhibited the left-moving the 1957 tornado, Hoecker one to a much greater degree. (1960) found a region of anticyclonic Since updraft-relative streamwise vorticity just outside the core at a vorticity is equal to the strength times roughly 100 m from the axis. the rate of veering with height of the updraft-relative winds, updraft rotation 2. WHAT WE KNOW depends on updraft motion. Updrafts propagate towards (away from) the Tornadogenesis divides naturally side where NHVPGF is upward into three stages. The first step in the (downward). Rotunno and Klemp vorticity concentration is the formation (1985) found that in nearly straight of a rotating updraft. shear, updraft propagation depends on the nonlinear part of the NHVPGF 2.1 STAGE 1 OF TORNADOGENESIS arising indirectly from the updraft-shear interaction. The updraft forms a Since some tornadoes occur in midlevel vortex pair by pulling up loops low-shear environments, meteor- of environmental vortex tubes. The ologists debated, prior to 1977, low pressure in these vortices is whether the concentrated vorticity in associated with upward NHVPF below supercell tornadoes was simply a result them. Midway between the vortices of stretching of pre-existing planetary there is high pressure due to water loading and deformation with that either tilt initially horizontal vortex downward NHVPGF below this high. tubes as they advect them downward This configuration of NHVPGF causes or simply transport vertical vorticity the initial updraft to split into a downward. This second stage of cyclonically rotating SR and tornadogenesis, the development of anticyclonically rotating SL updrafts. rotation very close to the ground, is When the hodograph is highly the one that requires much more curved, Davies-Jones (2002) found research. that the propagation depends mainly on the linear part of the NHVPGF 2.3 STAGE 3 OF TORNADOGENESIS induced directly by the shear-updraft interaction (as first proposed by The third stage, the formation of Rotunno and Klemp 1982 for nearly the tornado, appears to be simply the straight shear). result of amplification of vertical Updrafts acquire cyclonic vorticity in air parcels that are being circulation by propagating into cyclonic stretched vertically by an updraft and out of anticyclonic regions. For (Walko 1993). This view is supported example, the SR updraft maintains its by Doppler radar observations of rotation by propagating towards the strong low-level convergence just prior original cyclonic vortex, which stays to tornado formation. Frictional ahead of it because the updraft is interaction between the low-level continually tilting environmental mesocyclone and the ground may aid vorticity upward at its leading edge. this process by inducing radial inflow Davies-Jones (2004) found that the along the ground (Rotunno 1986). At growth of circulation around the edge ‘ground zero’, stretching of vertical of an updraft at a given level is equal vorticity is the dominant term in the to the line integral around the edge of vertical-vorticity equation once vertical vertical vorticity times either the local vorticity is present very near the propagation of the edge normal to ground. Ward’s (1972) laboratory itself or, equivalently, the local tornado simulator exemplified this NHVPGF divided by the local gradient stretching process in a wide updraft of vertical velocity. Thus, the first and reproduced several observed stage of tornadogenesis, the features of tornadoes such as development of a mesocyclone aloft, is characteristic surface pressure profiles, well understood. vortex breakdown, and multiple vortices. Given sufficient time without 2.2 STAGE 2 OF TORNADOGENESIS disruptions such as cold pools spreading beneath the updraft, this Tilting by an updraft of horizontal stage should mimic Ward’s laboratory vorticity fails, however, to produce model. At large swirl ratios, centrifugal rotation very close to flat ground forces prevent the convergence of air because the vertical vorticity is to near the axis of rotation, in which generated as the air is rising (Davies- case either there is a single large weak Jones 1982). This rule could be vortex or multiple tornadoes form violated if streamwise vortex lines were around the periphery of the tilted upward abruptly by a gust front mesocyclone. and stretched by an overhead updraft There is no need to find exotic as proposed for by energy sources for tornadoes since we Simpson et al. (1986). This is an now know that the “thermodynamic effect that is absent in simulations with speed limit” can be broken (Fiedler and limited horizontal resolution. However, Rotunno 1986). The frictional parcels approaching a gust front interaction between the tornado and generally start rising before they reach the ground drives strong radial inflow. it and, since the vortex lines tend Parcels can penetrate closer to the somewhat to be frozen into the flow, axis than the radius dictated by they would also begins lifting ahead of cyclostrophic balance above the the gust front. Tornadogenesis must boundary layer. Because their angular await the development of downdrafts momentum is nearly conserved, they rotate very quickly. Their excessive terminate in a strong updraft to kinetic energy is compensated for by prevent it from filling from above. loss of pressure energy. To conserve mass, the boundary layer erupts 3.1 BAROCLINIC MECHANISMS violently upward into an intense vertical jet along the axis. High-resolution numerical cloud When the convergence increases models now produce poorly resolved and the ambient is tornadoes within simulated supercells constant with height or when the in non-rotating . In these convergence is constant and the simulations, rotation near the ground is angular momentum increases with due to tilting of horizontal vorticity that height, the tornadic vortex forms aloft is generated baroclinically in subsiding because high-angular-momentum air air that has spent considerable time first arrives near the axis aloft. The (around 10 min or more) in a strong vortex then builds downward slowly buoyancy gradient (Rotunno and (tens of minutes) to the surface Klemp 1985). Davies-Jones and through a dynamic pipe effect (Smith Brooks (1993) showed that the vorticity and Leslie 1978; Trapp and Davies- in this air changes from anticyclonic to Jones 1997). If there is insufficient cyclonic during its descent in the angular momentum near the ground, baroclinic zone by the mechanism the vortex remains as a described in Davies-Jones et al. and never becomes a tornado. If the (2001). Its cyclonic vorticity is then convergence and angular momentum greatly amplified as it passes into the are constant with height in the lowest updraft. In the simulations, the near- few kilometers, the vortex contracts surface vorticity maximum or tornado- uniformly and a tornado forms rapidly like vortex is typically located in in 5-10 min. gradients of both and Incidentally, a popular suggestion equivalent potential temperature. for tornado modification is to explode a Unexpectedly, in-situ field device in the core. A laboratory observations made during and after experiment by Chang (1976) showed VORTEX (Verification of the Origins of that this method can disrupt a vortex Rotation in Tornadoes EXperiment) temporarily, but it immediately reforms have failed to detect rain-cooled air at as long as the larger-scale flow near the surface near several strong and the ground remains convergent and violent tornadoes even though the gyratory. tornadoes were close to a wind-shift line. The discrepancy between observations and simulations may be 3. WHAT WE DON’T KNOW due to the microphysics scheme producing too strong a cold pool too As stated above, most of the soon owing to excessive rain mysteries concerning tornadogenesis production too low in the cloud and concern the development of rotation the evaporation rate being too fast. very close to the ground through some The observations do not preclude the process that seems to be very different possibility that air, which enters the from midlevel mesocyclogenesis. The tornado at its base, may have passed possibility exists that the process may slowly through a remote baroclinic not be the same one in all cases. zone that may even be above the Several processes have been ground. suggested and none have been completely verified by field 3.2 POSSIBLE ROLE OF observations. Genesis of a strong MICROBURSTS long-lived tornado requires that the near-ground mesocyclone be Do microbursts sometimes trigger underneath the mid-level mesocyclone tornadoes or help maintain them? so that there is a continuous wide Radar observations and damage vortex column from the ground to near surveys sometimes reveal microbursts the tropopause. The tornado has to prior to the touchdown of tornadoes or on the right sides of tornado tracks. In introduced through the top boundary the radar echoes of several tornadic above the updraft, they fall around the storms, ‘blobs’ of higher reflectivity, periphery of the updraft and drag perhaps associated with microbursts, angular momentum down to the have been observed within the hooks. ground. Some of the angular These descend to low elevations prior momentum that is transported to tornado touchdowns on their left downward advects inwards towards forward sides. The blob in the well the axis as in Fujita’s (1975) “recycling observed 2 June 1995 Dimmitt, process”. Fujita inferred this tornadic storm had an anticyclonic mechanism from eyewitness vortex on its left side and a cyclonic photographs that showed the rotating one that appeared to evolve into the rain curtain tilting inwards towards a tornado on its right side. This vorticity tornado (like as in Fig. 2b). In the configuration could arise from simulation, a tornado forms and downward transport of high-momentum ultimately decays owing to the air (represented by the double arrow in absence of a ‘buoyant cork’ aloft. The Fig. 2). Alternatively, it could originate mechanism is unequivocally barotropic from tilting by a downstream updraft of because differential drag forces baroclinic vorticity. A downdraft that is generate azimuthal vorticity, which heavy aloft owing to water loading cannot be tilted in axisynmmetry flow. and/or evaporative cooling would be In this simulation, the rotating rain encircled by descending quasi- curtain instigates the tornado. horizontal vortex rings with the vorticity The axisymmetric model constrains vectors directed clockwise. These rings the flow unrealistically by making part would spread along the surface upon of the outflow from the downdraft to reaching it, Lifting of the leading focus on (i.e., converge towards) an edges in the RFD gust front imposed axis. However, it has the would give rise to advantage over a fully 3D model of vertical vorticity of the observed being much easier to interpret. configuration. The cyclonic vortex is generally the stronger vortex and the 3.4 WALKO’S BAROTRONIC one that turns into a tornado because MECHANISM it can enter the main updraft. A major question is whether enough circulation Davies-Jones (1982) proposed could be generated for a major that a rear-flank downdraft, although tornado. The other vortex could divergent, is vital to the lowering of become an anticyclonic tornado if it is vertical vorticity and rotation to the stretched by an updraft in the flanking surface. In westerly shear, the line. downdraft would draw down loops of vorticity, producing a north-south 3.3 FUJITA’S BAROTRONIC oriented cyclonic-anticyclonic vortex MECHANISM pair near the surface. The main updraft lies ahead of and to the left of In cases without significantly cool the RFD, in the right position for the air near the tornado, is there a remote with cyclonic vorticity to flow along the baroclinic zone aloft (possibly aloft) or ground and into it. This would result in is a barotropic tornadogenesis a near-ground mesocyclone through mechanism operating? To show that a stretching. Note that the outflow from barotropic mechanism is possible, I the downdraft enhances the low-level built a ‘bare-bones’ axisymmetric model convergence in the updraft. of a mesocyclone without A numerical experiment by Walko . The initial condition (1993) validates this process for an is a Beltrami flow that consists of a idealized flow. In a westerly shear cyclonic updraft (or midlevel flow, Walko used a fixed heat sink and mesocyclone) surrounded by a source to produce a RFD southwest of downdraft. If left unperturbed, the flow an updraft. A tornado-like vortex decays slowly without changing formed barotropically beneath the pattern. When hydrometeors are updraft on the left edge of the cold pool. Although baroclinically mesocyclones within a single supercell, generated vorticity was tilted, it (Burgess et al. 1982). Rotunno (1986) contributed negatively to the Brandes (1978) and Wakimoto and Liu circulation of the vortex and hence to (1997) proposed that the cylindrical its genesis. vortex sheet between cells in a two- celled mesocyclone with a central 3.5 PRE-EXISTING VERTICAL occlusion downdraft could become VORTICITY unstable. The mesocyclone would then transform into two or more Low-level rotation could be the tornado cyclones, as in Ward’s result of a supercell updraft simulator at large swirl ratio. The concentrating pre-existing vertical tornado cyclones could produce vorticity on a front or shear line (Walko tornadoes through frictional interaction 1993). Since the horizontal vorticity in with the ground, sheared supercell environments is an order of magnitude larger than pre- 3.8 TORNADOGENESIS FAILURE existing vertical vorticity, tilting of horizontal vorticity would provide a Why do some imminent tornadoes more abundant source of vertical never develop when strong warning vorticity. But this mechanism cannot signs are present? Possible be dismissed in the case of high- explanations are as follows. CAPE, low-shear tornadoes (see Markowski et al.’s (2003) axisymmetric section 3.10). model with idealized thermodynamics and precipitation shows that a tornado 3.6 VORTEX SHEET INSTABIITY will not form if the air at the ground is too negatively buoyant and cannot be Another barotropic mechanism lifted. This happens in the real world involves shearing instability on the when the outflow is cold and deep and pseudo-. When unstable, the storm-relative inflow is not strong the quasi-vertical vortex sheet at the enough to prevent the density current wind discontinuity rolls up into from propagating ahead of the individual vortices, which can be updraft. Presumably a tornado also stretched by overhead will not form (at the axis) if strong (Barcilon and Drazin 1972; Davies- centrifugal forces prevent Jones and Kessler 1974; Wakimoto convergence. and Wilson 1989; Lee and Wilhelmson 1997). In some simulations of tornadic 3.9 NEGATIVE VISCOSITY supercell storms, two or three cyclonic vortices move up the gust front and Lilly (1969) attributed the existence flanking line to the main updraft where of anticyclonic vorticity just outside the they merge into a tornado-like vortex core of the Dallas tornado to inward (Adlerman and Droegemeier 2005). eddy angular momentum flux, a But, in other simulations of tornadic negative eddy viscosity phenomenon storms, the vortices move down the that would contribute to tornado gust front away from the main updraft, maintenance. This process requires and so play no role in the large asymmetries of the flow, possibly tornadogenesis. Thus, this in the form of spiral bands or inward mechanism cannot be the only one. moving eddies or secondary vortices.

3.7 TWO-CELL MESOCYCLONE 3.10 WARM-SEASON TORNADOGENESIS Agee at al. (1976) explained long- lived tornado families by postulating Storms in high-CAPE, low-shear the existence of smaller-scale tornado environments occasionally produce cyclones that revolved around the strong and violent tornadoes. These parent mesocyclone. Tornado enigmatic storms have rarely been cyclones do exist but tornado families observed during field experiments. are due to a succession of Since there is little shear in the environment, they must either avail of the main storm-scale updraft and themselves of pre-existing vorticity downdraft”. along a frontal or or High–precipitation (HP) supercells manufacture horizontal vorticity are not prodigious tornado producers, baroclinically and then tilt it towards perhaps because their mesocyclones the vertical. Since environmental occlude too rapidly for tornadogenesis inflow winds are light, the cold pool to occur. should spread out ahead of the updraft and inhibit tornadogenesis. 4. SUMMARY There are two factors that may mitigate this effect (Davies-Jones et al. 2001). In this paper I argue that the big First, the updraft may be so strong that unknown in supercell tornadogenesis it creates moderate inflow winds is how rotation develops next to the through suction. Second, the ground. There may be a variety of environmental CAPE is so high that modes for this second stage of parcels at a considerable distance tornadogenesis. Several plausible behind the leading edge of the cold mechanisms are reviewed. pool still have moderate CAPE and can be lifted. Acknowledgments. This work was supported in part by NSF grant ATM- 3.11 CYCLIC TORNADOGENESIS 0340693. I am indebted to Drs. Erik Rasmussen and Jerry Straka for Burgess et al.’s (1982) conceptual allowing me to base some of this model of cyclic mesocyclone core paper on parts of their 1997 formation, based on Doppler radar unpublished article. data, essentially fits the mesocyclogenesis in both simulated and observed storms. Adlerman et al. 5. REFERENCES (1999) found in simulations of cyclic mesocyclogenesis that the baroclinic Adlerman, E. J., K. K. Droegemeier, mechanism was common to all cycles. and R. P. Davies-Jones, 1999: A The succeeding cycles occurred more numerical simulation of cyclic rapidly than the first one because cold mesocyclogenesis. J. Atmos. Sci., air from the previous cycle lends to a 56, 2045-2069. buoyancy orientation that is favorable Adlerman, E. J., and K. K. for horizontal baroclinic vorticity Droegemeier, 2005: A numerical generation. Adlerman and simulation of cyclic Droegemeier (2005) simulated cyclic mesocyclogenesis. (unpublished). tornadogenesis. The model storm Agee, E. M., J. T. , and P. R. produced six mesocyclones, one of Clare,1976: Multiple vortex features which produced two tornadoes and in the tornado cyclone and the two others one each. Although some occurrence of tornado families. of the mesocyclones formed very near Mon. Wea. Rev., 104, 552-563. their predecessors and entrained Barcilon, A., and P. Drazin, 1972: vorticity-rich air from them (as devil formation. Geophys. Fluid postulated by Davies-Jones 1982), Dyn., 4, 147-159. these were nontornadic. Brandes, E. A., 1978: Mesocyclone Dowell and Bluestein (2002) evolution and tornadogenesis: analyzed data collected during Some observations. Mon. Wea. VORTEX of a cyclic tornadic supercell Rev., 105, 113-120. and, in contrast to the simulations, Brooks, E. M.,1949: The tornado found no cold air behind the most cyclone. Weatherwise, 2, 32-33. intense and longest-lived tornado. Brown J. M., and K. R. Knupp, 1980: They concluded that ”cyclic tornado The Iowa cyclonic-anticyclonic formation may result if the horizontal tornado pair and its parent motion of the tornadoes repeatedly . Mon. Wea. Rev., does not match the horizontal motion 108, 1626-1646. Brown, R. A., L. R. Lemon, and D.W. Burgess, 1978: Tornado detection 1974; Tornadoes. Weather and by pulsed Doppler radar. Mon. Modification, W. N. Hess, Wea. Rev., 106, 29-38. Ed., Wiley, 552-595. Browning, K. A., 1964: Airflow and Davies-Jones, R., and J. H. Golden, precipitation trajectories within the 1975: On the relation of electrical severe local storms that travel to activity to tornadoes. J. Geophys. the right of the winds. J. Atmos. Res., 80, 1614-1616.. Sci., 21, 634-659. Davies-Jones, R., and H. Brooks, Browning, K. A., and R. J. Donaldson, 1993: Mesocyclogenesis from a 1963: Airflow and structure of a theoretical perspective. The tornadic storm. J. Atmos. Sci., 20, Tornado: Its Structure, Dynamics, 533-545. Prediction and . Browning, K. A., and C. R. Landry Geophysical Monogr., No. 79, 1963: Airflow within a tornadic Amer. Geophys. Union, 105-114. thunderstorm. Preprints, 10th Davies-Jones, R., R. J. Trapp, and H. Conference, B. Bluestein, 2001: Tornadoes and , DC, Amer. Meteor. tornadic storms. Severe Convective Soc., 116-122. Storms. Meteor. Monogr., No. 50, Burgess, D. W., V. T. Wood, and R. A. 167-221 Brown, 1982: Mesocyclone Dowell, D. C., and H. B. Bluestein, evolution statistics. Preprints, 10th 2002: The 8 June 1995 McLean, Conf. On Severe Local Storms, Texas, storm. Part II: Cyclic tornado Omaha, NE, Amer. Meteor. Soc., formation, maintenance, and 422-424. dissipation. Mon. Wea. Rev., 130, Burgess, D. W., D. C. Dowell, L. J. 2649-2670. Wicker, and A. Witt, 2005: Detailed Eskridge, R. E. and P. Das, 1976: comparison of observed and Effects of precipitation-driven modeled tornadogenesis. Preprints, downdraft on a rotating wind field: 32nd Conf. On Radar Meteorology, A possible trigger mechanism for Albuquerque, NM, Amer. Meteor. tornadoes. J. Atmos. Sci., 33, 70- Soc., CD-ROM, 10R.4. 84. Chang, C. C., 1976: Preliminary work Fawbush, J., and R. C. Miller, 1954: toward “killing the tornado” The type of air masses in which laboratory partial simulation North American tornadoes form. experiment. Proc. Symp. On Bull. Amer. Meteor. Soc., 35, 154- Tornadoes: Assessment of 165. Knowledge and Implications for Fiedler, B. H., and R. Rotunno, 1986: Man, Lubbock, TX, Texas Tech A theory for the maximum University, 493-499. windspeeds in tornado-like vortices. Davies-Jones, R., 1982: observational J. Atmos. Sci., 33, 2328-2340. and theoretical aspects of Fujita, T. T., 1959: Detailed analysis of tornadogenesis. Intense the Fargo tornadoes of June 20, Atmospheric Vortices, L. Bengtsson 1957. U.S. Weather Bureau Tech and J. Lighthill, Eds., Springer- Rep. 5, Severe local Storms project, Verlag, 175-189. University of Chicago, 29 pp. + figs. Davies-Jones, R., 1984: Streamwise Fujita, T. T., 1975; New evidence from vorticity: The origin of updraft April 3-4, 1974 tornadoes. rotation in supercell storms. J. Preprints, 9th Conf. On Severe Atmos. Sci., 41, 2991-3006. Local Storms, Norman, OK, Amer. Davies-Jones, R., 2002: Linear and Meteor. Soc., 248-255. nonlinear propagation of supercell Fujita, T. T., G. S. Forbes, and T. A. storms. J. Atmos. Sci., 59, 3178- Umenhofer, 1976:Close-up view of 3205. 20 March 1976 tornadoes: Sinking Davies-Jones, R., 2004: Growth of cloud tops to suction vortices. circulation around supercell Weatherwise, 29, 116-131, 145. updrafts. J. Atmos. Sci., 61, 2863- Golden, J. H., 1974: The life cycle of 2876. keys waterspouts. I. J. Appl. Davies-Jones, R., and E. Kessler, Meteor.,13, 676-692. Hoecker, W. H., Jr., 1960: Ray, Ed., Amer. Meteor. Soc., 414- and air flow patterns in the Dallas 436. tornado and some resulting Rotunno, R., and J. B. Klemp, 1982: implications. Mon. Wea. Rev., 88, The influence of the shear-induced 167-180. pressure gradient on thunderstorm Klemp, J. B., and R. B. Wilhelmson, motion. Mon. Wea. Rev., 110, 136- 1978: The simulation of three- 151. dimensional convective storm Rotunno, R., and J. B. Klemp, 1985: dynamics. J. Atmos. Sci., 35, 1070- On the rotation and propagation of 1096. simulated supercell thunderstorms. Lee, B. D., and R. B. Wilhelmson, J. Atmos. Sci., 42, 271-292. 1997: The numerical simulation of Simpson, J., B. R. Morton, M. C. non-supercell tornadogenesis. Part McCumber, and R. S. Penc, 1986: I: Initiation and evolution of pre- Observations and mechanisms of tornadic misocyclone circulations GATE waterspouts. J. Atmos. Sci., along a dry outflow boundary. J. 43, 753-783. Atmos. Sci., 54, 32-60. Smith, R. K., and L. M. Leslie, 1978: Lemon. L. R., and C. A. Doswell III, Tornadogenesis. Quart. J. Royal 1979; Severe thunderstorm Meteor. Soc., 104, 189-199. evolution and mesocyclone Stout, G. E., and F. A. Huff, 1953: structure as related to Radar records tornadogenesis. Mon. Wea. Rev., tornadogenesis. Bull. Amer. Meteor. 107, 1184-1197. Soc., 34, 281-284. Lilly, D. K.,1969: Tornado dynamics. Trapp, R. J., and R. Davies-Jones, NCAR Manuscript No. 69-117 1997: Tornadogenesis with and (available from the National Center without a dynamic pipe effect. J. for Atmospheric Research, Boulder, Atmos. Sci., 54, 113-133. CO.) Vonnegut, B., 1960: Electrical theory Lilly, D. K.,1982: The development of tornadoes. J. Geophys. Res., 65, and maintenance of rotation in 203-212. convective storms. Intense Wakimoto, R. M., and J. W. Wilson, Atmospheric Vortices, L. Bengtsson 1989: Non-supercell tornadoes. and J. Lighthill, Eds., Springer- Mon. Wea. Rev., 117, 1113-1140. Verlag, 149-160. Wakimoto, R. M., and C.-H. Liu, 1998: Markowski, P. M., 2002: Hook echoes The Garden City, storm and rear-flank downdrafts: A review. during VORTEX-95. Part II: The Mon. Wea. Rev., 130, 852-876. wall cloud and tornado. Mon. Wea. Markowski, P. M., J. M. Straka, and E. Rev., 126, 393-408. N. Rasmussen, 2002: Direct surface Walko, R. L., 1993: Tornado spin-up thermodynamic observations within beneath a convective cell: Required the rear-flank downdrafts of basic structure of the near-field nontornadic and tornadic boundary layer winds. The supercells. Mon. Wea. Rev., 130, Tornado: Its Structure, Dynamics, 1692-1721. Prediction and Hazards. Markowski, P. M., J. M. Straka, and E. Geophysical Monogr., No. 79, N. Rasmussen 2003: Amer. Geophys. Union, 89-95 Tornadogenesis resulting from the Ward, N. B., 1972: The exploration of transport of circulation by a certain features of tornado downdraft: Idealized numerical dynamics using a laboratory model. simulations. J. Atmos. Sci., 60, 795- J. Atmos. Sci., 29, 1194-1204. 823. Wicker, L. J., and R. B. Wilhelmson, Rasmussen, E. N., and J. M. Straka, 1995: Simulation and analysis of 1997: Tornadogenesis: A review tornado development and decay and a new conceptual model. within a three-dimensional supercell Unpublished manuscript. thunderstorm. J. Atmos. Sci., 52, Rotunno, R., 1986: Tornadoes and 2135-2164. tornadogenesis. Mesoscale Wurman, J., and S. Gill, 2000: Meteorology and Forecasting, P. S. Finescale radar observations of the Dimmitt, Texas (2 June 1995) tornado. Mon. Wea. Rev., 128, 1113-1140. Fig. 1. Diagrams showing relationship of nearby rain shaft and cloud features to (a) tornado (based on Fujita 1959 and from Fujita et al. 1976), and (b) (based on Golden 1974). Fig. 2. Schematic plan view of an isolated SR supercell storm near the surface. Thick line encompasses radar echo (note hook on southwest side). The wave-like gust front structure, resembling a synoptic-scale cyclone, is depicted by classic frontal symbols. Low-level positions of the updrafts are finely stippled. Coarse stippling marks location of forward-flank downdraft (FFD) and rear-flank downdraft (RFD). Favored locations for tornadoes are marked by encircled ‘T’s. The major cyclonic tornado is most probable near mesocyclone center. A minor tornado or a new potentially major tornado may occur at the bulge in the RFD gust front. Southern T also marks position where new mesocyclone core may form as old one occludes. ‘A’ is most probable location of any anticyclonic tornado that forms. Thin arrows depict storm-relative streamlines (adapted from Lemon and Doswell 1979). Fig. 3. Composite view of a typical tornado-producing cumulonimbus as seen from a south-easterly direction. Horizontal scale is compressed. All the features shown cannot be seen simultaneously from a single point. Shelf cloud may not be present or may be south of wall cloud instead of northeast of it. With time a spiral precipitation curtain wraps cyclonically around the wall cloud from the northeast (diagram by C. Doswell III).