Scheduled Downtime
On Friday 21 April 2023 @ 5pm MT, this website will be down for maintenance and expected to return online the morning of 24 April 2023 at the latest

segmentation fault when running atmosphere_model

L

laetis

New member
Hi,
I encountered a problem with MPAS atmosphere software. I wanted to change the GFS source by using ifs data. The init_atmosphere code runs well but when I launch atmosphere_model it fails with segmentation fault and no error printed. The only thing that is different from usual in the init_atmosphere_model run are these messages:
"
Setting SST from SKINTEMP
Bad sm_fg 0.00000 4 755
Bad sm_fg 0.00000 4 777
Done with soil consistency check
"
which are not pointed as error. What do they mean and could it be the origin of the problem?
Thanks for any tips or insights.

Best
 
The message below indicates that soil moisture is wrong in your input data.

Code: 
Bad sm_fg 0.00000 4 755
Bad sm_fg 0.00000 4 777

Specifically, sm_fg(4, 755) and sm_fg(4,777) are 0.0. However, the MPAS initialization program will correct such errors by specifying sm_fg = 0.001. This is why your init_atmosphere can run to the end successfully.

I am suspicious that your input data contains some other errors, which lead to the failure of model run. If this is the case, then the model should crash immediately after it is launched. Let me know if I am wrong.

Please clarify
(1) Is your case a global or regional MPAS run?
(2) How did you get IFS data? Is it a global dataset?
(3) Can you upload two single intermediate files from IFS for me to take a look?

Thanks.
 
Last edited:
I looked at your IFS files, which look fine. However, your SST file is from GFS, not from IFS. The slight difference in LANDSEA mask between IFS and GFS leads to issues in soil data interpolation, which explains why you get the errors in sm_fg.

Can you add GFS LANDSEA to your SST data, then try again? Let me know whether this can fix your issues.
 
Hi
Thank you a lot for your answer.
I tried to add the LANDMASK ans it didn't changed anything.
I also tried to take the SST from the ifs data so that everything is from the same source, still no success. It always fails at the begin of the second iteration and I always have the same warning in the init_atmosphere_mode log file.
Best
 
Hi,

This is an issue caused by inconsistent spatial distribution of LANDSEA and soil data, e.g., at certain grids of IFS, LANDSEA shows they are land points while soil data implies they are water points, and vice versa.

Currently MPAS has no capability to correct such kind of errors in the input data. Please avoid using IFS at present. Both GFS and ERA5 should work fine for MPAS.
 
You must log in or register to reply here.
Top