Thunderstorm Outbreak Simulation
over the White Mountain National Forest

This is a simulation! Even though the astronomy, geography, and meteorology are instructed to simulate reality , the date and time are not. I picked June 21, 1998 only because it was close to the Summer Solstice. None of this really happened.

The simulation is centered over the small town of Beebe River, NH. The horizontal domain covers a square which is 93 kilometers on a side. The horizontal resolution of grid points for meteorological parameters and surface terrain is 1 kilometer. The vertical resolution for meteorological parameters is 0.5 kilometer. For those of you who don't feel like perusing that "supercells.input" file above (it's big), here are the details on this simulation:

• The simulation begins at 18:00 GMT on June 21, 1998
• It uses a 6 second timestep to integrate the weather through time (non-hydrostatic equations being used)
• Look at that [supercells.input] file above, yep, a 93 x 93 x 35 3D domain contained these thunderstorms
• The horizontal has 1km resolution, the vertical .5 km resolution
• that's 93 kilometers x 93 kilometers x 16 kilometers (depth)
• Radiation is parameterized every 60 seconds
• Surface physics are taken care of every 6 seconds
• Kain-Fritch Cumulus Parameterization is done every 60 seconds
• Simulation Initialized using a single (almost unEarthly) sounding  [jun21.altered.snd]
• I planted a tube of slightly warmer and moister air at low levels along the axis of Franconia Notch to aid thunderstorm formation and cell arrangement
• The simulation was set up to run a couple of hours, it failed due to excessive updrafts at nearly one hour

Nearly one full day (on a wall clock) passed before this one hour simulation was completed. It ran on a Pentium Pro based PC, constructed with an Intel PR440FX motherboard, Ppro 200 Mhz (256 Kb cache), 256 MB DRAM (60 nanosecond ECC), and gobs of QUANTUM ATLAS II SCSI-3 4.5 GB drives. The ARPS run only occupied 125 Megabytes of RAM while it was running, and the VIS5D history data dumps took place every minute of simulated time. These VIS5D images only took up half a GIGAbyte of disk space. The post processing, the editing and image processing needed to make these movies, took much, much more space.