Forecasting tornadic thunderstorms
by Dave Stensrud and Harold Brooks
Left box: Observed frontal location, modeled rainfall during the past hour (green), and observed severe reports (tornado as an inverted triangle; severe wind as a +; hail as a dot) Right box: Model parameters used to determine likelihood of tornadic thunderstorms overlaid on a single figure. Region where tornadic thunderstorms are most likely indicated by the colorfilled sections in the right box that overlap where rainfall is simulated in the left box.
Operational forecasters are very good at identifying days when severe weather occurs across large regions of the country. It is more of a challenge, however, to determine whether these days will be dominated by tornadic thunderstorms or non-tornadic thunderstorms. This difficulty is related to our general lack of understanding of the processes that form tornadoes.
The National Severe Storms Laboratory (NSSL) is exploring this operational forecast problem using a mesoscale numerical weather prediction model. We have been examining hourly output from a number of model simulations to determine whether or not any signal exists that would help forecasters determine whether or not tornadic thunderstorms are likely.
Model results indicate that we can discriminate between days with tornadic and non-tornadic thunderstorms. The key parameter needed to answer this question is the storm-relative wind. When the winds in the atmosphere are much faster than the movement of the thunderstorm, the raindrops produced in the storm updraft get blown away from the storm, and the cooling effect of the rain has little effect on the storm evolution. In contrast, when the winds in the atmosphere are much slower than the movement of the thunderstorm, the raindrops all fall very close to the center of the storm, cooling the air near the ground and producing very strong outflow. In between these two extremes, the winds in the atmosphere are roughly equal to the movement of the thunderstorm, and just the right amount of rainfall reaches the ground near the center of the storm. In this case, the cooling effect of the rain can be used to help generate rotation in the thunderstorm at levels near the ground surface. This increases the likelihood of tornado formation. This conceptual model was developed by Brooks et al. (1994) by examining numerical simulations of thunderstorms but also appears in our mesoscale model simulations of outbreak days.
Three parameters are used to help determine
whether or not tornadic thunderstorms are possible.
One is a measure of the positive buoyant
energy available to a storm (Convective Available
Potential Energy - CAPE), another is a parameter
that has been used to forecast the rotational
characteristics of thunderstorms (Storm-Relative
Environmental Helicity - SREH), and the third a
measure of the wind shear over the lowest 6 km of
the atmosphere (bulk Richardson number shear -
BRNSHR), which is closely related to the stormrelative
wind. When these parameters all fall into
a specified range, tornadic supercell thunderstorms
are more likely than non-tornadic supercell
thunderstorms (see figure). This conceptual model
is presently being used by forecasters at the Storm
Prediction Center (SPC) to help provide improved
severe weather guidance to the nation.
