8.5 Test and Optimize Input Files


Previous

HOME


Next

In the previous section the CAPTEX simulation parameters were set without much discussion. In this section, we review a special program HYSPTEST than can be called from the GUI to evaluate the primary model input file CONTROL as well as the optional namelist input file SETUP.CFG. Both trajectory and dispersion input files can be evaluated. However, only limited testing is conducted for a trajectory calculation. The program may suggest changes to these files to improve the simulation.

  1. If continuing from the previous section, delete the WRF meteorology, or retrieve the saved input files captex_control.txt and captex_setup.txt to use only WRF27UW data.

  2. For simplicity, we will evaluate one aspect of the simulation, the particle release rate. In the original configuration, it was increased from the default 2500 to 50000. Open the particle release rate menu and change the release rate back to 2500.

  3. Now the input files are ready for evaluation. Press the Concentration / Special Runs / Test Inputs tab which opens the 3-step Test and Optimize menu.

  4. Pressing the Analyze button of step 1 runs the HYSPTEST evaluation program, the results of which are shown on the simulation log. Abnormal terminations would appear on this log page. In this case the test simulation completed successfully without error but not necessarily without suggested modifications.

  5. Pressing the WARNING button under step 2 shows the contents of the WARNING_mod file which contains a summary of the suggested changes to the inputs. In this instance, two changes were suggested. One to the CONTROL file, noted that the last sampling period ended two hours after the end of the 68 h simulation. This correction would have no effect if there were no measured data during the last sampling period. The second suggestion, was to increase the particle release number from 2500 to 10000. The implication is that the original choice of 50,000 was larger than necessary. Standard analysis messages are written to MESSAGE_mod.

  6. Pressing the CONTROL button will show a side by side comparison of the original and modified files: CONTROL and CONTROL_mod. Pressing the SETUP button will show a similar side by side comparison of the original and modified namelist files: SETUP and SETUP_mod. The modified namelist file will always have more parameters than the original file because the original is created by the GUI, while the modified file is the output of all the namelist variables by the evaluation program.

  7. In step 3, pressing the COPY button, copies the modified input files to the standard names default_conc or default_traj and SETUP.CFG. Now running the simulation will copy the default file to CONTROL and use the new values which can be seen in the MESSAGE file of the simulation.

  8. Note that step 3 only changes the input files. The corresponding variables in the GUI are not changed. However, running the model creates the CONC.CFG file, a copy of the SETUP.CFG namelist file used by the model. Each time you open the Advanced / Configuration Setup menu, it will load the variables in the CONC.CFG file so that the revised particle release rate is automatically changed within the GUI menu. A good practice would be to load the input file changes back into the GUI variables by retrieving CONTROL_mod and SETUP_mod.CFG into their respective menus.

  9. Rerunning the CAPTEX simulation with the 10000 particle release rather than the 50,000 particle release and recomputing the statistics shows that the simulation with the smaller particle number was quite comparable to the original calculation. However, it is always better to release more particles than required.

This section describes a program called HYSPTEST, which is a simplified version of HYSPLIT that will read the various input files such as CONTROL, SETUP.CFG, EMITIMES, etc., and determine if many of the user options are correctly or optimally configured. The program opens all the meteorological data files, releasing, transporting, and dispersing particles in the same manner as a regular simulation, except to reduce the calculation time, only one particle per time step is processed. No valid output files are created except the MESSAGE and WARNING files.

1 m 17 s