Tornadoes - Severe Convective Storms and Tornadoes: Observations and Dynamics - page 359

less well one knows the position of the particle: The observer affects the phenom-
enon being measured. To avoid observing problems laser velocimeters have been
used to map the wind field unobtrusively, but measurements of thermodynamic
variables still require in situ probes. Ironically, some researchers have conducted
numerical simulations of laboratory experiments. Rich Rotunno at NCAR
exploited this approach in the late 1970 and early 1980s.
In a ''Ward''-type vortex chamber, the tornado is separated from the ''storm''
above and the air coming into the bottom of the chamber: the boundaries of the
tornado (exhaust fan aloft and inflow below) are open. Inflow below and outflow
aloft are specified as boundary conditions and can significantly affect the behavior
of the flow. Brian Fiedler introduced an alternative with a closed domain, in
which an updraft is driven by a bubble of positive buoyancy—not by an exhaust
fan. The domain extends upward to the top of the storm at the tropopause. The
Ward chamber is one in which air enters from somewhere else, goes up, and then
out; the Fiedler model is one in which air is forced radially inward below, upward,
and radially outward aloft by localized buoyancy, and then re-cycled. The buoy-
ancy bubble emulates a buoyant updraft in the parent convective storm. In
addition to the Ward chamber and the Fiedler model, there are also models which
use water as the fluid and one in which a propeller is used to force air upward.
An obvious problem with interpreting laboratory model simulations or
idealized numerical simulations of laboratory vortices is that of the difference in
spatial scales: in a simulator, the characteristic wind scale
U 1ms 1 ð 6 : 2 Þ

and the characteristic horizontal length (L) and depth (H) scales are
L H 0 : ð 6 :

3 Þ 1m

Thus, the Reynolds number (the ratio of the acceleration due to inertial forces
Dv

=

Dt to that due to friction)

= m ð 1ms 1 1m Þ=ð 2 10 5 m 2 s 1 Þ 10 4

Re UL Þð 0 : ð 6 : 4 Þ

where the kinematic coecient for molecular viscosity in the atmosphere
m ¼ 2 10 5 m 2 s 1

ð 6 : 5 Þ

has been used. In the case of a real tornado, however,
U 75 m s 1

ð 6 :

6 Þ

L H 100 m ð 6 : 7 Þ

so that the Reynolds number is
Re ð 75 m s 1 Þð 100 m Þ=ð 2 10 5 m 2 s 1 Þ 10 8

ð 6 : 8 Þ

Thus, the characteristics of turbulent, subgrid-scale diffusion may be quite
different in vortices in the vortex chamber and in nature. Recently, Brian Fiedler
and Gabe Garfield at OU (Oklahoma University) have shown how the behavior of