mesoscale reference for MPASv7.3 runs stalled citing problem with PBL scheme

[jpen]

My runs with MPASv7.3 for the mesoscale reference suite and mp_wsm6 is stuck until walltime ends. While the log.atmosphere file does not hint at the problem, the slurm log file points to PBL scheme. Is this related to initial conditions? How to resolve it?

 

[Ming Chen]

Your case crashed after about 6 hours of integration with the error message "Floating point exception (core dumped)", suggesting that something is wrong in either physics or dynamics. Your namelist.atmopshere looks fine to me, except that the radiation interval is way too large. Can you set:

   config_radtlw_interval = '00:03:00'
   config_radtsw_interval = '00:03:00'

Then rerun this case. Please let me know whether it works.

Also, what data did you use to create initial condition for this case?

 

[jpen]

Thanks Ming! I will test with your suggested radiation interval and let you know. However, I have run with similar configuration for other dates (in 2024 and 2022) and they ran completely. I am using ERA5 data from Copernicus.

 

[jpen]

The runs are still getting stuck!!!

 

[Ming Chen]

Thanks for testing. I am sorry that the case crashed again.

MPAS-v7.3 is way old. Is it possible that you switch to the latest version (MPAS-v8.3.1)?

I didn't repeat your case mainly because MPAS-v7.3 requires a specific version of PIO for compiling. However, with rapid updates of various compilers and libs, I had troubles to rebuild the old version. Hope the new version of MPAS-v8.3.1 works for you,

If you still see the same issue in MPAS-v8.3.1, then I definitely will repeat your case and find out what is wrong.