6.1 Frequency Analysis




In this section we will review different ways to analyze multiple trajectory output files. In contrast to the previous section, where multiple trajectories were written to a single output file, in this analysis approach, each output file contains a single trajectory resulting from a single execution of the trajectory model. This method would be used rather than the multiple trajectories in-one-file approach if additional meteorological information is required with each trajectory or to more easily sort trajectories associated with selected air samples. In this example, we will investigate the probability of conducting a successful CAPTEX tracer release during the month of September 1983.

  1. Although a regional extract of meteorological data for September of 1983 is already provided for the tutorial, the complete global file can also be downloaded from the ARL server through the GUI meteorology menu tab. Enter the year and month into the download menu and the file name will be constructed automatically. A progress bar will be shown until the download completes. Remember all files are directed to the \working directory unless another location is specified. Note that downloads may be blocked by the Windows firewall and may need an administrative user to enable.

  2. Now go to the trajectory setup menu and configure the run for the single mid-boundary layer trajectory as before (traj_fwrd_control.txt), but first change the starting time to the beginning of the month 83 09 01 00, reduce the run time to 48 hours, rename the output file to fdump to avoid later confusion with previous simulations, and lastly change the meteorology file (first clear) to the global file, either in the /captex directory or the working directory, if you just downloaded the file RP198309.gbl. Save the changes as traj_freq_control.txt in case we need to rerun this case and then save to exit the menu.

  3. The run model tab will only create a single trajectory but the Trajectory / Special Runs / Daily tab will run a script based upon the start time previously entered in the SETUP menu and a duration time (days and hours) entered into the run daily menu. Change the duration from 30 to 28 because September only has 30 days. To complete a 48 h trajectory you can't start trajectories on the 29th or 30th without adding October to the meteorology input file list. New trajectories are started at the indicated interval (6 hr in this case). The script execution creates a CONTROL for each trajectory starting time, runs the model, and creates a unique trajectory output file for each run: fdump{yymmddhh}.

  4. After the script completes there should be 112 files starting with fdump in the \working directory. Rather than trying to display them individually, select the Display / Frequency tab. This opens the frequency menu. There are two steps involved in this process. First, enter the base name of the output files: fdump, this will be used as a wildcard to Create file of trajectory filenames when the yellow button is pressed. The file of filenames is named by default as INFILE.

  5. Check the working directory to insure that no fdump files exist without the number suffix. These would have been created by the run model tab. This file should be deleted and a new INFILE created, or you can just edit INFILE to delete the line that shows the unwanted file name.

  6. The next step is to just Execute Display which runs two processes. The first is to count the number of trajectories that fall within each grid cell that covers the area. The grid resolution defaults to 1.0 degree. Other resolutions can be set through the menu. The trajectory frequency (F) is just the sum of the number of trajectories (T) that passed through each (i,j) grid cell divided by the total number (N) of trajectories analyzed:

    • Fi,j = 100 Σ Ti,j / N
    • Σ Ti,j ≤ 1.0

  7. Note that all trajectories are counted once in the source location grid cell and each trajectory is counted once per intersecting grid cell if the residence time radio-button is set to No. When the residence time option is set to Yes, multiple intersections can occur, up to the number of endpoints (P) along any on trajectory:

    • Fi,j = 100 Σ Ti,j / N P
    • Σ Ti,j ≤ P

    Normally all trajectories are counted that pass over a grid cell. However, a height range can also be selected to exclude trajectories. Options available through the command line, but not the GUI, include computing the frequency by dividing by PxN or the maximum value of T. The trajectory summation results are written to a file and a normal completion will result in a simple termination message.

  8. The second part of Execute Display automatically opens the display of the trajectory frequency output which shows flow in all directions during the month with one preferential direction to the northeast skirting along the edge of the CAPTEX sampling network. If the computation is repeated, but with the residence flag set to Yes, then the results show more transport to the east.

The results shown here present a simple approach to analyzing multiple trajectories. There are other methods which will be explored in more detail in subsequent sections. For instance, counting a trajectory more than once in a grid cell because multiple points along the trajectory may stay within the same grid cell is a residence time weighted frequency. This type of particle trajectory analysis is the basis of the concentration calculation.

3 s