Problem with trajectories re-entering metgrid with higher resolution

[John.Johansson]

I am running backward HYSPLIT trajectories using metdata on 3 different resolutions. I include metdata for all 3 resolutions in the CONTROL file. The trajectories start in the grid with the highest resolution. There doesn't seem to be any problem when trajectories pass over the edge from a higher resolution grid to a lower resolution grid, but sometimes trajectories will turn around and re-enter the higher resolution grid. In these cases, the trajectories seem to be stopped. I'm running many trajectories with different starting points and heights in the same run, but it seems like only the trajectories reentering from a lower resolution grid to a higher resolution grid are stopped, while the rest continue. Has this been observed before? Is this a known limitation?

[ariel.stein]

Can you please send us the CONTROL file that produces that feature?

[John.Johansson]

I have now chosen a suitable example case from a large batch of trajectories that I have been running and simplified it slightly to rule out some very simple possible explanations. I have rerun the simplified case and checked that the problem still arises. Here is the CONTROL file from my example case:

Code: Select all

2013 08 12 21 00
4
53.330000 6.280000 10.0
53.330000 6.280000 100.0
53.330000 6.280000 300.0
53.330000 6.280000 500.0
-96
0
10000.0
3
/home/john/test_20130812_2100/
fnl_weekly_arl_20130808
/home/john/test_20130812_2100/
wrf_d01_weekly_arl_20130808
/home/john/test_20130812_2100/
wrf_d02_weekly_arl_20130808
/home/john/test_20130812_2100/
test_20130812_2100

This example only has 4 trajectories starting from the same point, but different heights. At the end of my simulation only 1 of the trajectories was still running. If you need any of the other files from my example run (such as SETUP.CFG, the metdata files or the output file) you can download them here:

https://chalmersuniversity.box.com/s/p4 ... o0cbirzojc

[ariel.stein]

Is there any reason you need to select all the meteorological variables along the trajectory? Try selecting only one.

[John.Johansson]

Do you mean setting tm_tpot, tm_tamb, tm_rain, tm_mixd, tm_relh, tm_sphu, tm_mixr, and tm_dswf to 0? I tried setting them all to 0 now, but the trajectories still stopped when re-entering the higher resolution domain. I also tried setting only one of these parameters to 1, but it didn't change the outcome either.

[ariel.stein]

That is strange, because when I run the model I get the trajectories going back. See below.

Try setting the vertical motion method to 5 (divergence).

[John.Johansson]

Yes, setting the vertical motion method to 5 worked. The resulting trajectories were also more similar to the ones you posted above than before, but not exactly the same. Were your trajectories computed with vertical motion method 5, or did you also get it to work with method 0?

Is it possible that the version of HYSPLIT that I'm running is to old? Could this have been fixed in updates during the last year or so?

[ariel.stein]

My trajectories were produced with the vertical motion method set to 5. I don't think we made any updates regarding this issue in the last year or so. What version of the model are you running?

[John.Johansson]

Ok, so it works with method 5 but not method 0. I also tried method 4 (isosigma) and it also worked? Any idea why method 0 specifically wouldn't work?

I don't know what the best way to check the version is, but when running hyts_std, it prints the following message:

Code: Select all

 HYSPLIT4 - Initialization
 Last Changed Rev: 836
 Last Changed Date: 2016-09-29 15:10:42 +0200 (Thu, 29 Sep 2016)

[ariel.stein]

I assume that the vertical velocity field from your meteorological data is what is causing this problem. The vertical method 5 is just using the horizontal divergence to estimate the vertical velocity.