Tornado formation

B. Geerts and E. Linacre


 The environment of tornadic storms

Most tornadoes, especially the violent ones, are spawned by supercell storms. Supercell storms are defined as those that exhibit mid-level rotation (usually cyclonic), with the highest vorticity more or less coincident with the updraft core. This mid-level rotation is known as the mesocyclone, which usually can be seen by a Doppler radar. Supercell storms form in the following environment:

  1. high convective available potential energy (CAPE), i.e. warm, moist air in the PBL and much cooler air aloft;
  2. large wind shear, especially wind backing (veering) with height in the southern (northern) hemisphere;
  3. some convective inhibition, i.e. a stable layer (or lid) at the top of the PBL, about 2 km above the ground.

An additional factor may be the presence of dry air in the middle troposphere. Mid-level dry air may be entrained into the storm, and cooled by evaporative cooling within the mixed air parcel. This may trigger a downdraft. Stronger downdrafts may imply stronger updrafts and a more severe storm.



There are three mechanisms responsible for the intense rotation of a tornado.

The relative importance of these mechanisms is highly variable from case to case, but the first mechanism (vortex stretching) is usually present, especially in the later stages of tornadogenesis.


Why tornadoes usually spin cyclonically 

The question is - why do a majority of tornado funnels (in the USA, at least) twist cyclonically, i.e. clockwise in the southern hemisphere? In the USA an estimated 80-95% of all tornadoes are cyclonic. One would not expect the Coriolis effect to be influential on the scale of a tornado. Strong tornadoes are virtually all tornadoes, whereas many weak tornadoes and gustnadoes have been observed spinning anticyclonically. (Gustnadoes are weak, shallow, short-lived tornadoes that are induced by wind shear and convergence along a gust front. They break off from the gust front of the cool surface outflow of air on the periphery of the storm.)The chance of finding a cyclonic dust devil near a building and in an open field is only about 50%.

The answer is that most tornadoes result from vortex stretching within a mesocyclone. About 80-90% of all mesoscale circulations in supercell storms are cyclonic (the other 10-20% are meso-anticyclones). The prevailing rotation of the mesocyclone also is unaffected by the Coriolis effect. The reason can be found in the wind profile in the supercell's environment. If the wind profile were straight and the winds aloft were to increase only in strength, not direction, then the odds for mesocyclones and meso-anticyclones would be the same. That is because supercell storms often split in two, one drifting north of the mean wind, and one south of it. The storm on the equatorward side (or left storm in the southern hemisphere) has a mesocyclone, and the right-moving storm contains a meso-anticyclone. With a straight wind profile, the odds of survival are equal for the two storms.

In the southern hemisphere, an environment that supports severe thunderstorms usually involves backing of the wind with increased elevation. This is a counterclockwise turning wind profile. For instance, during storms at Sydney there is commonly a northeast wind at the surface, backing to northwest at 500 hPa and westerlies above. In this situation, the left mover thrives after a storm split, and the right mover rapidly decays. The right mover decays because there is not enough low-level shear: the storm moves away from the fuel source, the warm, moist northeasterly inflow, whereas the left mover moves towards it. So mesocyclones are most common because synoptic conditions suitable for deep convection (e.g. in the warm sector of a frontal system) have winds backing with height (in the southern hemisphere), and numerical modelling studies (1) show that this favors left-moving storms with a mesocyclone. This also implies that most supercells in the southern hemisphere move to the left of the deep-layer mean wind, typically about 30.


  1. Weisman, M.L. and J.B. Klemp 1982. The dependence of numerically simulated convective storms on vertical wind shear and buoyancy. Mon. Wea. Rev., 110, 504-520.