A Conceptual Model of Nontornadic Supercell Thunderstorms
Open Access
- Author:
- Majcen, Mario
- Graduate Program:
- Meteorology
- Degree:
- Doctor of Philosophy
- Document Type:
- Dissertation
- Date of Defense:
- June 19, 2009
- Committee Members:
- Paul Markowski, Dissertation Advisor/Co-Advisor
Paul Markowski, Committee Chair/Co-Chair
Yvette Pamela Richardson, Committee Member
Johannes Verlinde, Committee Member
Joshua Wurman, Committee Member
Andrew Mark Carleton, Committee Member - Keywords:
- thunderstorm
supercell
nontornadic
tornado
tornadogenesis - Abstract:
- This study uses dual-Doppler observations of nontornadic supercells obtained by ground-based mobile Doppler radars and idealized numerical simulations in order to develop a conceptual model of a nontornadic supercell, particularly at low levels and on the submesocyclone scale. There are relatively few dual-Doppler studies of supercells in the history of severe storms research owing to the relative rarity of supercell occurrences within dual-Doppler radar networks, and the majority of the dual-Doppler studies feature tornadic supercells. Moreover, the submesocyclone scale and lowest few hundred meters generally have not been well-observed in prior studies, which usually have analyzed either pseudo-dual-Doppler airborne radar observations or data from fixed radar networks. In addition to the fact that a steady state must be assumed for relatively long time periods when analyzing dual-Doppler observations from airborne radars (typically 5-7 minutes), the resolution is coarser than what is afforded by ground-based mobile radars because aircraft must maintain larger distances from the mesocyclone for safety reasons. In regards to ground-based dual-Doppler networks comprising fixed radars, the baselines are usually long (40-75 km baselines are common); thus, the centers of the dual-Doppler lobes, where the geometry is most favorable for accurate wind retrievals, are at a greater range from the radars, resulting in a relatively coarse resolution and inability to observe the lowest few hundred meters owing to radar horizon limitations. In the first part of this dissertation, five nontornadic supercell thunderstorms are analyzed using high-resolution dual-Doppler radar data obtained by a pair of mobile ground-based radars. Three out of five observed supercells had well-developed low-level rotation. The observed low-level kinematic fields of the nontornadic supercells with low-level rotation are compared to the low-level kinematic fields of tornadic supercells that have been previously documented. In previous studies, tornadic and nontornadic supercells have had strikingly similar kinematic characteristics on the mesocyclone scale. Thus, discrimination between tornadic and nontornadic supercells has been very difficult (probably fewer than 25\% of supercells are tornadic). It is determined that the observed low-level kinematic structure of nontornadic supercells is qualitatively very similar to the low-level kinematic structure of tornadic supercells, notably two out of three observed nontornadic storms had a ``bent-back' rear-flank gust front just like the tornadic supercells, and one of those also had a dual rear-flank gust front, a feature that previously has been observed only in tornadic supercells. The low-level mesocyclone in the nontornadic supercells extends to the lowest analysis level in the three cases having low-level rotation, but the low-level circulation in nontornadic mesocyclones is much weaker than in tornadic mesocyclones. Also, the divergence associated with rear-flank downdrafts is stronger in nontornadic supercells than in tornadic supercells. Vortex line analyses in the observed nontornadic storms show that the vorticity field structure is consistent with baroclinic generation of horizontal vorticity and subsequent tilting into the vertical by an updraft, as has been shown in recent observational and numerical simulation studies. One major limitation of the observational part of this study is the lack of thermodynamic observations. Thermodynamic retrievals from the dual-Doppler wind syntheses were found to be insufficiently accurate for any rigorous quantitative analysis. In the second part of this study, a series of idealized, dry three-dimensional numerical simulations are used to gain some understanding of the relationship between the low-level thermodynamics and kinematics of supercells, assuming that the idealized simulations can replicate the evolution of the low-level kinematic fields observed in actual supercells. The idealized simulations emulate the generation of near-surface rotation beneath supercell-like (i.e., helical) updrafts in a way consistent with our present understanding of the importance of a downdraft in environments in which vertical vorticity is initially absent at the surface. In the simulations herein, the initial conditions are specified to be horizontally homogeneous but vertical wind shear and veering winds with height are present. A heat source is introduced, resulting in a cyclonically rotating updraft having maximum rotation at midlevels and no rotation at the surface. Once an approximately steady state is achieved, a heat sink is introduced at low levels in proximity to the helical updraft. The resulting downdraft and outflow generate rings of baroclinic horizontal vorticity that encircle the heat sink, and this vorticity is subsequently tilted by the vertical velocity gradients associated with the downdraft-updraft couplet such that a couplet of vertical vorticity develops at the surface, straddling the line that joins the downdraft and updraft extrema. The intensity of the vortices that develop at the surface depends on the degree to which the baroclinic vorticity can be tilted and stretched, which depends in large part on the extent to which the negatively buoyant low-level air originating in the heat sink can be lifted by the overlying updraft driven by the heat source. In all cases, the cyclonic member of the vorticity couplets that develop at the surface become dominant, and this bias is presumably attributable to the overlying updraft possessing updraft-scale cyclonic rotation. The most intense cyclonic vortices develop when the outflow and associated baroclinic horizontal vorticity from the heat sink is strong enough to spread beneath the overlying updraft, but is not so cold that the air parcels within the outflow cannot be drawn upward by the overlying updraft. Three modes of what is referred to as "tornadogenesis failure" also are noted. (Even though the simulations do not explicitly resolve tornado-scale motions, we casually use the terminology "tornadogenesis failure" to refer to the inability of intense sub-mesocyclone-scale vortices to develop in the model.) The first failure mode occurs when the cold pool does not develop or just a weak cold pool (potential temperature deficits near the vortex less than 2 K) develops briefly. In the presence of a stronger cold pool, the baroclinically generated vortex lines are tilted upward and interact with the updraft. Tornadogenesis occurs when the vertical vorticity is stretched by the main updraft. The second failure mode occurs when the cold pool intensity is stronger than in the tornadogenesis case (potential temperature deficits near the vortex about 4--8 K). In this mode of tornadogenesis failure, the gust front bends back in a similar manner to the simulations that resulted in tornadogenesis but tornadogenesis never occurs. In the case of the second tornadogenesis failure mode the vortex lines emanating from the near-ground vorticity maximum form arches that rise only up to about 3 km AGL and then descend in the rear flank to the south of the vorticity maximum. The third failure more occurs when the cold pool potential temperature deficit is 8 K or higher. In that case, the gust front rushes ahead of the main updraft. The vertical vorticity is generated along the gust front but the vortices are very shallow. The vortex lines in this case rise only up to 200 m AGL before tilting into the horizontal.