WRF-Chem stops after exactly 1 day - Why?

[Leon Kuhn]

Dear WRF-Chem users,

I am currently facing an issue where I try to run 7 consecutive days of WRF-Chem simulation. The run is able to finish day 1, but stops at the beginning of day 2. This was reproducible. I have attached the namelist.input file and rsl.out.0000 with a debug level of 50. Seeing that the file ends with multiple lines of the kind "d01 2018-05-02_00:00:00 in wrf_quilt_write_field", I think it may have something to do with quilting, but I am not sure. On the other hand, it could have something to do specifically with day 2, but the logs give no clear indication what that might be.

Does anybody have an idea of how to proceed on the basis of these two files? Perhaps make some changes to the namelist.input?

Many thanks and best regards!

 

[jordanschnell]

Hi Leon,

From the looks of it, I don't think the quilting/output is the issue, but you can try two different things to be sure.

1) Change the frames_per_outfile = 1
2) Increase the number of nio_groups (try 4?)

You could also try restarting with the restart file, which I assume has been written out correctly (you could change the restart interval to >>1440 in case that is the issue and so it doesn't write a restart file)

Otherwise, it may be an issue with your input data - set debug level >= 300 and make sure none of the other processors are throwing an error (tail rsl.*)

Jordan

 

[Leon Kuhn]

Dear Jordan,

thanks for your reply!

I have attached two more files:

- tail.txt which is the output of tail -20 rsl*
- rsl.out.0000.cropped.txt When running with debug_level = 300, the rsl.out files are huge and can not be uploaded here. So I just uploaded the last lines, which come from increasing the debug_level. As you can see, the rsl.out.0000.cropped.txt starts where the rsl.out.0000 file of the first post ends.

Can you get any more information out of these files? For me they are quite uninformative. Do you have any further ideas on how to inspect input files for errors?

Best regards
Leon

 

[jordanschnell]

Hi Leon,

Unfortunately there isn't much help here.

Have you attempted turning off quilting altogether or increasing the number of nio_groups? You can also try REDUCING the number of total processors.

Jordan

 

[Leon Kuhn]

Dear Jordan,

since some colleagues I have been working with have agreed to your initial idea that it might have something to do with input data, I will now test out whether it runs without any auxiliary input (no bioemissions, anthropogenic emissions, wildfire, exo_coldens or wesely). I will report back after I have tested that.

After that I can also test a run without quilting...

Thanks for your input, it is very much appreciated

 

[Leon Kuhn]

Dear users,

I figured out how to resolve this issue. In my namelist.input I changed the tracer_opt to

tracer_opt = 3, 3

Now everything works. Hope this helps people with the same problem.