To further expand upon the single-particle calculation discussed in the previous sections, we will increase the particle number release rate until the simulation looks more realistic. If you have not completed the previous sections, please go back and review them before proceeding. Otherwise you can just retrieve
- We ended the last section viewing the hourly concentration contribution of one particle as it was transported and dispersed along its path. Lets start by reducing the number of output frames to every three hours rather than hourly. Open the Concentration / Setup Run / Grids menu and change the last line from 01 01 00 to 01 03 00. Save to close all the setup menus. Then open the Advanced / Configuration Setup / Concentration / Release Number limits (Menu #4) and change the Particles Released per Cycle from 1 to 100. Save to close and exit all menus.
- Now Run Model, open the contour display menu and when complete, view each of the four frames (at 20Z, 23Z, 02Z, and 05Z). Note at the initial time, all the particles are close together and the pattern is continuous, but as the particles spread out due to turbulence and vertical wind shear, the results quickly degrade and by the last frame it almost appears that each of the 100 particles are individually visible. Also take note of the maximum concentration.
- Run this exercise again for 1000 particles and 10000 particles: only the 10,000 particle simulation shows a continuous concentration pattern. Remember in the original example calculation we only released 2500 particles over one hour, a considerably smaller particle release rate, but the concentration pattern was very smooth. This is because we computed a 12 hour average concentration, rather than instantaneous concentrations. The model needs to sample a sufficient number of particles over the averaging interval and concentration grid size to insure robust results. The appropriate number can be determined by trial-and-error as the rate when the air concentration results are no longer sensitive to the particle release rate. Again note how the maximum concentration changes with increased particle number. Why does it go down?
- Previously we found that the simulation results for a given particle release rate were sensitive to averaging time. A similar sensitivity exists with regard to the concentration grid size. To run a simulation with a finer resolution grid, open the Concentration / Setup Run / Grids menu to change the spacing of the grid from 0.05 to 0.005 in both latitude and longitude. One degree spacing is approximately 100 km, so 0.005 degree spacing is about half a kilometer. Note that a grid center specification of zero means that the concentration is centered over the release point. Save to close the menu and then run the model. Note this simulation will take a lot more time than the previous one with the 0.05 deg grid.
- When the run completes, open the display contour menu. With all those nearby particles, the visual look of the plot can be improved by removing the closed contour about each fill color by changing the contour drawing options to none. The final graphic shows much more detail than the previous simulation, but again individual particles are visible.
The simulations with various particle number release rates suggested that more particles are required for more detailed simulations, but at a cost of increased computational time. This may become especially significant for continuous releases. There are other simulation options to reduce the particle number requirements which will be explored in the next section. For future reference, the control (cpart_control.txt) and namelist (cpart_setup.txt) files should be saved.