9.2 Base Configuration Optimization




In the previous section the model was configured to display the plume during the first aircraft sampling pass. Now we will experiment with some basic model simulation parameters. Continuing on from the Concentration / Display / Concentration / Contours menu tab used to create the last simulation graphic in the previous section ...

  1. The large red square in the center of the plot shows the grid cell (to scale) with the maximum concentration. The 25 km resolution of the concentration grid may be appropriate for the full 68 hour simulation, but it is too coarse to examine the model prediction just 200 km downwind. Open the Setup Run / Grids menu and change the resolution from 0.25 to 0.05 degrees. Save to exit, then Run Model and Display / Concentration to see the new model prediction of a slightly narrower plume. The previously larger plume was a computational artifact of the relatively coarse grid resolution.

  2. To examine the effect of using the puff parameterization instead of the default 3D particle, open the Advanced / Configuration Setup / Concentration menu #3 to switch from 3D particle to Top-Hat-Horizontal Particle-Vertical mode and also menu #4 to change the particle release number from 50000 to 5000. Fewer puffs than particles are needed for the same simulation. Save to exit, run the model and display to see the top-hat puff prediction. The plume structure is similar, but much smoother, than the previous 3D particle calculation.

  3. Follow the same process to configure the Gaussian-horizontal particle-vertical puff simulation. The Gaussian plume graphic is similar to the previous top-hat plume showing that the plume structure is controlled more by the horizontal puff distribution due to different transport vectors rather than the details of the internal puff distribution. The particle/puff positions are the same in both calculations.

  4. The model concentrations were computed for the 0-100 m layer above ground level (AGL). However, the aircraft measurements were made at 914 m MSL. To configure the model to compute the layer average concentration centered about the flight level set the KMSL flag in the namelist through the Advanced / Configuration Setup / Concentration menu #2 for Meteorological Subgrid and Vertical Coordinates and choose the Relative to mean-sea-level radio button. This causes the model to treat all heights defined in the CONTROL file as MSL rather than AGL. Most GUI labels are fixed and may not reflect this change.

  5. Now in the Setup Run menu define two layers in the CONTROL file, say 800 to 1000 m MSL, to represent 0-800 and 800-1000, shorten the duration of the sample averaging period from 3 hours to 1 hour, and finally reduce the run time from 13 to 11 hours. Rerun the model.

  6. To show the concentration averaged between 800 m and 1000 m, from the Contours menu, select the Show Each Level radio button. Select the 1000 radio buttons for the From Bottom Level and Through Top Level fields. The resulting graphic is quit different to the ground-level one, showing maximum concentrations over the region of minimum ground level concentrations. The peak predicted concentration is quite similar, in location and magnitude, to the measured one. One explanation is that at the time of the aircraft sampling, around 2200 local time, the boundary layer was no longer well mixed.

  7. We will use this Gaussian configuration for all the subsequent mixing tests when comparing model results to the aircraft measurements. Save the control file settings to conc_case_control.txt and the namelist file parameters to conc_case_setup.txt.

We've found that for this particular case, the Gaussian plume simulation 200 km downwind matched the position and width of the aircraft measured plume. In the next few sections, we will experiment with different dispersion parameters and turbulence settings to determine their effect upon the calculation results.

1 m 54 s