Absolute Trajectory Error

Up to this point the discussion of trajectory error has been a little abstract and perhaps unsatisfying. Considering the spatial extent of the ground-level concentrations, almost any combination of starting heights or meteorological data would result in a trajectory that passes over some non-zero concentrations. However in addition to the ground-level sampling, several aircraft were equipped with samplers and collected relatively short time-duration samples (6 or 10 min) making passes through the horizontal extent of the tracer plume at several different altitudes and downwind distances. This sampling information can be used to identify the plume center-line and if we compute a backward trajectory from that point, it should pass over or near the tracer release location. That nearest approach distance is one measure of the absolute trajectory error.

1. The sampling data for each of the five different flight levels are contained in files named according to the flight altitude in meters above mean-sea-level (MSL). For instance, the lowest sampling altitude was 914 m MSL and can be found the file named flight0914.txt where the highest concentration of 29,936 pg/m3 was measured over a 6 minute sample starting at 1983 09 26 03 06 UTC with the sample centered over location 41.09N 82.52W.


2. The information about that sample should be transferred to the trajectory setup menu. A little mental calculation is required to determine the run duration. We know that the tracer release started at 17 UTC on the 25th while the sample collection was at 03 UTC on the 26th (rounding to the nearest hour), therefore the run duration should be -10 hours (negative because it is a backward trajectory). Also rename the trajectory endpoints output file from tdump to tdump_0914. Change the meteorology file back to captex2_narr.bin. Upon completing the trajectory setup, it should be saved as traj_0914_control.txt. Note rounding to the nearest hour is suggested for these examples rather than using the minutes field. Otherwise you would need to insure the integration time step is consistent with the start time.


3. One last change is required. The starting height is specified as MSL rather an AGL (the default). MSL designations are typical of aircraft data. You should open Advanced / Configuration / Trajectory menu #2 vertical coordinates and select the MSL radio-button. Save this change and then from the Create Optional Trajectory Namelist File: SETUP.CFG configuration menu save as traj_abse_setup.txt prior to saving and closing the advanced menu.


4. After all the changes have been made go ahead and run the trajectory and then open the trajectory display menu. You can go ahead and create the graphic as we did in previous examples. Clearly the trajectory goes back to near the source location, but we can be more exact by superimposing the previously calculated 750 m AGL forward trajectory on this plot, which would then show the tracer release location. However, rather than using the + symbol between each output file, we will use a different method this time by specifying only one file that contains the file names of all the files we want to superimpose.


5. In this special case the special file name is preceded by the + symbol, which would be the first character in the Input Endpoints field which should contain the file name of traj_files.txt. You then need to use Notepad to create traj_files.txt in the \hysplit4\working directory. At this point the file should only contain two entry lines: tdump_fwrd and tdump_0914. One file per line. The first line was our original 750 m forward trajectory calculation. Now when you run the trajectory display you will get the two trajectories superimposed and this particular aircraft sample based trajectory goes right back to the release location. Select the AGL radio-button in the display menu to show the trajectory in height units rather than pressure units.


6. This first calculation was reviewed in detail. To continue the exercise, go to each of the four remaining aircraft sampling files:
   • flight1219.txt
   • flight1524.txt
   • flight1829.txt
   • flight2134.txt
   and run a backward trajectory from the peak concentration in each file, saving the output files with unique names and then adding each of those to traj_files.txt and the create the final graphic with all the backward trajectories superimposed upon the initial forward trajectory.

In some respects it makes sense that one mid-boundary layer trajectory can be representative of so many different aircraft samples (all collected in the mid-boundary layer). However, the northern most sample had a completely different trajectory and naturally the complex and quite diffuse ground-level concentration pattern cannot be explained with a one-trajectory model. These issues will be explored further in subsequent sections.