Multiple backward trajectories in time output

Post any defects you find in the HYSPLIT software here. The HYSPLIT Developers carefully monitor this list and will work diligently to repair any reported problems. When posting a bug report, please specify both the HYSPLIT version and operating system you are using.
Post Reply
John.Johansson
Posts: 14
Joined: February 24th, 2016, 2:30 pm
Registered HYSPLIT User: Yes

Multiple backward trajectories in time output

Post by John.Johansson »

I am trying to run multiple trajectories in time backwards from the same endpoint location using GDAS 0.5 degrees data. The trajectories run for 96 hours each and are offset by 3 hours from each other. The endpoints in each file span over 7 days for each run, which gives 56 trajectories in each output file.

This seems to work fine except for one strange feature in the output files: Instead of writing out the true endpoint of each trajectory it seems to write out a duplicate of the endpoint of the first trajectory. The coordinates of the endpoint are of course the same for all the trajectories, but the time and met data should of course be different. Is this a known bug/feature?

I know I could probably get around this by scripting the trajectories into different runs, but I would prefer not to since I have already put some work into this solution.

Here are a couple of additional minor things I have noticed:
  • I am writing out all the possible meteorological output and before the last two columns an extra line break is added. What is the purpose of this? It makes the parsing of the output files more complicated.
  • I have noticed that the terrain height output for the endpoint is a bit different for different runs using the same endpoint location. Shouldn't the terrain height for a fixed point always be the same?
barbara.stunder
Posts: 424
Joined: November 9th, 2012, 4:23 pm
Registered HYSPLIT User: Yes

Re: Multiple backward trajectories in time output

Post by barbara.stunder »

Are you saying that for each trajectory, 3 hours offset from each other, they are all the same trajectory? That is not right. The output file should have all the date/times and trajectory endpoints for the 56 trajectories.

Are you setting the NSTR variable in the SETUP.CFG file to 3? I suggest you do the equivalent run on our web page and compare the CONTROL and SETUP files to your run. http://ready.arl.noaa.gov/HYSPLIT_traj.php

The extra line break when you output all the possible meteorology is probably a formatting issue that was overlooked when we added available fields to output. I’ll look at that for our next upgrade.

The terrain height should be the same at the same endpoint. Please send a tdump file showing this, and the corresponding CONTROL and SETUP.CFG.
John.Johansson
Posts: 14
Joined: February 24th, 2016, 2:30 pm
Registered HYSPLIT User: Yes

Re: Multiple backward trajectories in time output

Post by John.Johansson »

Thank you for the quick response!
barbara.stunder wrote:Are you saying that for each trajectory, 3 hours offset from each other, they are all the same trajectory? That is not right. The output file should have all the date/times and trajectory endpoints for the 56 trajectories.
No, the trajectories seem to be mostly correct. It's just that the time of the first point on each trajectory (chronologically the last point since these are backward trajectories) is the starting time of the entire run. All the other points on the trajectory seem to be fine. Since I know what this time should really be for each trajectory I can of course correct the time myself, but the bigger problem is that the met data is taken from the wrong time.
barbara.stunder wrote:Are you setting the NSTR variable in the SETUP.CFG file to 3? I suggest you do the equivalent run on our web page and compare the CONTROL and SETUP files to your run. http://ready.arl.noaa.gov/HYSPLIT_traj.php
I played around on the web page and realized that this effect is not isolated to backward trajectories. Here is the output from a forward simulation of 4 trajectories with 1 hour offset in between them:

Code: Select all

     2     1
    GFSG    16    10    19     0     0
    GFSG    16    10    20     0     0
     4 FORWARD  OMEGA   
    16    10    20     0    0.000    0.000   500.0
    16    10    20     1    0.000    0.000   500.0
    16    10    20     2    0.000    0.000   500.0
    16    10    20     3    0.000    0.000   500.0
     1 PRESSURE
     1     1    16    10    20     0     0     0     0.0    0.000    0.000    500.0    956.7
     2     1    16    10    20     0     0     0     0.0    0.000    0.000    500.0    956.7
     3     1    16    10    20     0     0     0     0.0    0.000    0.000    500.0    956.7
     4     1    16    10    20     0     0     0     0.0    0.000    0.000    500.0    956.7
     1     1    16    10    20     1     0     1     1.0    0.147    0.024    536.1    951.6
     1     1    16    10    20     2     0     2     2.0    0.284    0.051    553.9    948.0
     2     1    16    10    20     2     0     2     2.0    0.147    0.022    530.9    950.3
     1     1    16    10    20     3     0     3     3.0    0.417    0.081    564.7    946.5
     2     1    16    10    20     3     0     3     3.0    0.286    0.047    549.2    947.8
     3     1    16    10    20     3     0     3     3.0    0.148    0.020    526.7    949.9
     1     1    16    10    20     4     0     2     4.0    0.544    0.113    578.1    945.8
     2     1    16    10    20     4     0     2     4.0    0.419    0.074    566.6    946.9
     3     1    16    10    20     4     0     2     4.0    0.287    0.042    548.1    948.7
     4     1    16    10    20     4     0     2     4.0    0.146    0.017    526.4    950.7
     1     1    16    10    20     5     0     1     5.0    0.663    0.144    594.0    944.3
     2     1    16    10    20     5     0     1     5.0    0.542    0.101    589.2    944.8
     3     1    16    10    20     5     0     1     5.0    0.415    0.065    573.6    946.4
     4     1    16    10    20     5     0     1     5.0    0.280    0.035    552.8    948.5
     1     1    16    10    20     6     0     0     6.0    0.773    0.174    607.3    943.5
     2     1    16    10    20     6     0     0     6.0    0.654    0.127    614.2    942.5
     3     1    16    10    20     6     0     0     6.0    0.532    0.087    607.4    943.2
     4     1    16    10    20     6     0     0     6.0    0.403    0.053    586.3    945.4
     2     1    16    10    20     7     0     1     7.0    0.753    0.153    634.3    940.4
     3     1    16    10    20     7     0     1     7.0    0.633    0.109    642.5    939.1
     4     1    16    10    20     7     0     1     7.0    0.510    0.072    628.7    941.0
     3     1    16    10    20     8     0     2     8.0    0.716    0.130    669.4    936.5
     4     1    16    10    20     8     0     2     8.0    0.596    0.089    670.7    936.1
     4     1    16    10    20     9     0     3     9.0    0.661    0.106    706.1    932.5
You can see that the first four points are one point for each trajectory and they all have the same time and the same pressure. Above that you can see what the real trajectory starting times should be so I could just correct this info myself, but to get the pressure (or any other met data parameter) I would have to extract it separately from the met data files. If this was for the last point on each trajectory it wouldn't have been such a big problem, since I could just run it a little longer and throw away the last point, but the first point of the trajectory is really the one for which the met data is perhaps the most important.
barbara.stunder wrote:The extra line break when you output all the possible meteorology is probably a formatting issue that was overlooked when we added available fields to output. I’ll look at that for our next upgrade.
Ok, great!
barbara.stunder wrote:The terrain height should be the same at the same endpoint. Please send a tdump file showing this, and the corresponding CONTROL and SETUP.CFG.
I'll get back with this when I have more time. Where should I send the files?
barbara.stunder
Posts: 424
Joined: November 9th, 2012, 4:23 pm
Registered HYSPLIT User: Yes

Re: Multiple backward trajectories in time output

Post by barbara.stunder »

You are right that for the trajectories that start after the 1st hour (NSTR is non-zero), the meteorology given at their 1st endpoint is not at the trajectory start time. This is something we need to look into. As you say, running individual trajectories is one way around this.

The individual trajectories could differ slightly from the ones in the re-start mode if the automatic time step differs. The time step can be determined in the MESSAGE file. This is the same as for cases with starting multiple trajectories at different heights. Our web page http://ready.arl.noaa.gov/hypub-bin/traj1.pl has this notice:
NOTE: Please be aware of the following issue: when running more than one trajectory (multiple levels) in the same run, the time step of the model may change based on where the other trajectories are located. The time step is set by the fastest trajectory the last hour, when there are more than one trajectory, this could result in a different time step for any one trajectory than if it were run by itself. When a trajectory is sensitive to the time step, this means that the spatial gradients in the velocity field are too strong and there should be less confidence given to the accuracy of any one trajectory. If this is an issue for your work, run only one trajectory at a time.

Re: different terrain heights at the same location, please post the tdump, CONTROL, and SETUP.CFG here as you did the tdump above, or start a new question.
Ntwali
Posts: 7
Joined: October 13th, 2016, 9:01 am
Registered HYSPLIT User: Yes

Re: Multiple backward trajectories in time output

Post by Ntwali »

Hi,

Please I need your help. How to make seasonal trajectories (both backward and forward ) using HYSPLIT?

Hope to hear from you.

Kind regards.

Didier
barbara.stunder
Posts: 424
Joined: November 9th, 2012, 4:23 pm
Registered HYSPLIT User: Yes

Re: Multiple backward trajectories in time output

Post by barbara.stunder »

Search on this forum. Also see the Tutorial at http://ready.arl.noaa.gov/documents/Tut ... index.html
Post Reply