21st Conf. on Severe Local Storms

8.4

A large eddy simulation of a tornadic supercell: Comparison with observations

Louis J. Wicker, NOAA/NSSL, Norman, OK; and D. C. Dowell, Y. P. Richardson, and R. Wilhelmson

Numerical simulations of supercells containing low-level mesocyclones and even smaller scale vortices were first generated during the late 1980's and early 1990's. With the existing numerical methods and computational resources of the day, one could reproduce in some detail the observed features within tornadic storms. However, our lack of knowledge regarding the turbulence near the surface, representation of appropriate surface effects, and the inability to further refine the numerical grid limited our confidence as to whether the models were capturing the correct physics. Therefore conclusions regarding the physics of tornadogenesis were limited. Another significant limitation was the lack of observational data on a scale similar to the model grid. Validation of modeling results with observations was essentially impossible.

With the advent of mobile Doppler radars (Doppler On Wheels, UMASS 3mm Doppler radar, SMART radars) in the mid- to late 1990's, there now exist a significant number of Doppler data sets that have captured a wide variety of mesocyclones and their tornadoes. Combining these data with detailed thermodynamic and surface-based measurements around tornadic supercells (e.g., VORTEX-95) now enables modelers to quantitatively compare many aspects of their numerical simulations to these new observations.

We are now engaged in producing a large eddy simulation of a tornadic supercell using a variety of new numerical modeling techniques. Horizontal grid spacing over much of the thunderstorm is approximately 50 meters with vertical grid spacing near the surface approximately 20 meters. This increase in grid spacing as well as improved model numerics likely represents an increase in the effective model resolution by a factor of 3-4 over previous simulations.

While the main purpose of the simulation is to study tornadogenesis, our first task will be to compare the simulated velocity and reflectivity fields to those observed by the mobile radars. Comparison of the spectra of simulated and observed data will be performed in order to access how well the simulated storm represents eddy structures observed in the real data. Time permitting, additional analyses will be presented discussing the potential role of surface friction and precipitation processes in the production of tornadoes.

extended abstract  Extended Abstract (588K)

Session 8, Numerical Modeling of Severe Storms
Wednesday, 14 August 2002, 8:00 AM-10:00 AM

Previous paper  Next paper

Browse or search entire meeting

AMS Home Page