4th European Conference on Severe Storms 10 - 14 September 2007 - Trieste - ITALY Tornadogenesis: Our current understanding, operational considerations, and questions to guide future research Paul Markowski Department of Meteorology, Pennsylvania State University, University Park, Pennsylvania, USA, [email protected] I. INTRODUCTION to the horizontal buoyancy gradient. Whereas the tilt- ing of environmental horizontal vorticity has been shown Although we know much about the dynamics of to be fundamental to the formation of midlevel meso- midlevel updraft rotation—the defining characteristic of cyclones, the tilting of horizontal vorticity originating supercell thunderstorms—the details of how near-ground within this baroclinic zone has been implicated in the for- rotation arises and is amplified to tornado strength re- mation of low-level mesocyclones (Klemp and Rotunno main a challenge. I will review our current understand- 1983; Rotunno and Klemp 1985), where “low-level” typ- ing of the origins of rotation in supercells and the req- ically has referred to approximately a few hundred me- uisites for tornadogenesis. I also will discuss the chal- ters to ∼1 km above ground level. Recent observations lenges for forecasters and what strategies are likely to be reported by Shabbott and Markowski (2006), however, most fruitful given the current state of our understand- have raised some questions about the importance of hori- ing. I will conclude by mentioning what I believe are zontal vorticity generation within the forward-flank baro- some of the most important outstanding questions per- clinic zone. taining to tornadogenesis and the relationship between tornadic storms and their parent environments. III. REQUISITES FOR NEAR-GROUND ROTATION II. UPDRAFT ROTATION AWAY FROM THE By definition, tornadogenesis requires that large ver- GROUND tical vorticity arises at the ground. If preexisting verti- cal vorticity is negligible near the ground, then vorticity It is widely accepted that vertical vorticity arises stretching near the ground is initially negligible and ver- within thunderstorm updrafts (away from the ground) tical vorticity first must arise either from the tilting of as a result of tilting and subsequent stretching of hori- horizontal vorticity or from advection toward the surface zontal vorticity associated with mean vertical wind shear from aloft. Tilting by the horizontal vertical velocity gra- (Barnes 1978; Rotunno 1981; Davies-Jones 1984). When dients associated with an updraft alone is not effective at the environmental horizontal vorticity is purely cross- producing vertical vorticity near the surface because air wise, updrafts acquire no net rotation, but consist of is rising away from the surface as horizontal vorticity is a dipole of equally strong positive and negative verti- tilted into the vertical. But if a downdraft is involved in cal vorticity extrema that straddle the updraft, with the the tilting process, then vertical vorticity can be advected positive (negative) vorticity extremum being located on toward the surface as it is produced via tilting (Davies- the right (left) flank of the updraft when looking downs- Jones and Brooks 1993), where it subsequently can be hear (Davies-Jones 1984). Updrafts acquire net cyclonic stretched to form a tornado. For these reasons, it has (anticyclonic) rotation when the environmental horizon- been argued that a downdraft is needed for tornadogen- tal vorticity has a streamwise (antistreamwise) compo- esis when preexisting rotation is absent near the ground nent, and the correlation between vertical velocity and (Davies-Jones 1982a,b). (This conclusion depends on ed- vertical vorticity increases as the ratio of streamwise to dies being too weak to transport vertical vorticity down- crosswise vorticity increases, all else being equal (storm- ward against the flow. Furthermore, once a tornado is relative wind strength, growth rate of isentropic surface; established, tilting of surface-layer horizontal vorticity Davies-Jones 1984). by the extreme vertical velocity gradient associated with Three-dimensional numerical simulations of supercell the tornado updraft itself probably contributes to the thunderstorms have shown that baroclinic horizontal vor- near-ground vertical vorticity in a significant way. How- ticity, generated by the horizontal buoyancy gradient ever, such abrupt upward turning of streamlines, strong along the forward-flank gust front, can be tilted into the pressure gradients, and large vertical velocities are not vertical and stretched, just as environmental horizontal present next to the ground prior to tornadogenesis; thus, vorticity associated with the mean vertical wind shear is such tilting in the absence of a downdraft cannot be in- tilted and stretched (Klemp and Rotunno 1983; Rotunno voked to explain the amplification of near-ground vertical and Klemp 1985). This horizontal vorticity tends to vorticity that results in tornadogenesis.) be streamwise because storm-relative winds approaching The aforementioned theoretical arguments for the im- the updraft from the forward flank are generally normal portance of downdrafts in tornadogenesis have been ver- 2 ified in numerical simulations (e.g., Rotunno and Klemp vored in environments having a low cloud base (environ- 1985; Walko 1993). Moreover, nearly countless observa- ments with a low cloud base, i.e., large boundary layer tions exist of rear-flank downdrafts (RFDs), hook echoes, relative humidity, can limit the production of exception- and “clear slots” in close proximity to tornadoes. Fur- ally cold outflow). It seems as though tornadic super- thermore, trajectory analyses in a limited number of ob- cells might benefit from large low-level horizontal vortic- served supercells indicate that at least some of the air ity that is not accompanied by large negative buoyancy; entering the tornado passes through the RFD prior to strong cold pools have a tendency to either undercut up- entering the tornado (e.g., Brandes 1978). Numerical drafts (e.g., Brooks et al. 1993) and/or suppress vortic- simulation results also have emphasized the importance ity stretching beneath the updraft (e.g., Markowski et of the RFD and have shown similar trajectories of air al. 2003). When the ambient horizontal vorticity is only parcels entering modeled vortices resembling tornadoes relatively modest, then perhaps tornadogenesis requires (Wicker and Wilhelmson 1995; Xue 2004). significant enhancement of the ambient horizontal vor- When there is preexisting rotation at the surface, a ticity. Such enhancement might be difficult to accom- downdraft such as the RFD is not needed for tornadoge- plish without strong storm-induced baroclinity (which is nesis. In these cases, near-ground convergence alone can suppressed by large ambient low-level relative humidity), amplify vertical vorticity to tornado intensity. It seems but strong baroclinity is difficult to achieve without fairly as though nonsupercell tornadoes like waterspouts and strong cold pools. landspouts (e.g., Wakimoto and Wilson 1989; Roberts and Wilson 1995), and perhaps most other geophysical vortices, commonly arise in this manner. V. FUTURE RESEARCH There are a number of aspects of supercell thunder- IV. CHALLENGES TO FORECASTERS storms and tornadogenesis that remain poorly under- stood. Among these are the four-dimensional forcings Although supercells might be regarded as being rela- of RFDs and the dynamical role of RFDs in tornadogen- tively easy to anticipate, predicting which supercells will esis, the importance of microphysical differences among spawn tornadoes is one of the most arduous tasks facing supercells and how those microphysical differences arise, operational meteorologists and researchers alike. A re- the thermodynamic characteristics of supercells above cent study in the U.S. has confirmed prior anecdotal evi- the ground, the effects of radiative transfer processes on dence of the relative infrequency of tornadoes even within storm dynamics, the dynamics of storm-storm and storm- supercells; Trapp et al. (2005) reported that only about boundary interactions, and the importance, if any, of a quarter of all radar-detected mesocyclones were associ- meso-γ-scale variability (such as that due to dry bound- ated with tornadoes, using fairly stringent mesocyclone ary layer convection) on storms. I am optimistic that detection criteria. Tornadoes occur over a broad range of substantial gains in understanding can be achieved in the midlevel mesocyclone intensities, with some of the most not-too-distant future as a result of new field projects, intense mesocyclones ever documented being observed in continually increasing computing power, and growing in- nontornadic supercells (Wakimoto et al. 2004). terest in severe convective storms worldwide. Except in rare circumstances, radars only detect tor- nado parent circulations (i.e., mesocyclones)—they can- not resolve tornadoes themselves. One of the most fruit- VI. ACKNOWLEDGMENTS ful strategies undertaken in the U.S. for improving tor- nado warnings has been to combine real-time radar data with observations of the near-storm environment. I am indebted to Dr. Yvette Richardson and our stu- Two parameters seem to offer the most promise in dents at Penn State University: Zack Byko, Jeff Frame, discriminating between nontornadic and tornadic super- Mario Majcen, and Jim Marquis. I also thank my re- cells: (1) boundary layer water vapor concentration and cent collaborators, Mr. Don Burgess, Dr. David Dow- (2) low-level vertical wind