crashes when running with noah lsm

[xtian15]

I was making forecasts with atmosphere_model that was initialized with ERA5. It kept running into crashes that ended up returning

Segmentation fault - invalid memory reference

and preceded by

Flerchinger USEd in NEW version. Iterations= 10

in the stdout with the default

config_physics_suite = 'mesoscale_reference'

. If I add the

config_lsm_scheme = 'off'

into the namelist, the simulation can go thru the point it would've crashed. The namelist, log.atmosphere.0000.out, and stdout have been attached to this post. Would you mind taking a look at what might have caused the crash as I still want to leave the LSM option on?
Thanks in advance,

 

[mgduda]

In my experience, the "Flerchinger" errors most often turn up when there are unreasonable soil moisture or soil temperature values in the initial conditions. As a first check, it might be worth plotting or otherwise scanning through the soil moistures and soil temperatures to see that they all look reasonable.

 

[xtian15]

Hi, Michael,
After running in the login node, it turns out that the crash is caused by the RRTMG_SW with the error message of "MCICA_SUBCOL: KISSVEC SEED GENERATOR REQUIRES ... <cut>..", which will completely terminate the simulation process. I also saw an open issue of this at https://github.com/MPAS-Dev/MPAS-Model/issues/139. I wonder how it should be addressed.
Much thanks!

 

[mgduda]

As Brian mentioned in MPAS-Model Issue #139, it's not clear why the check

if (pmid(i,1).lt.pmid(i,2)) then

is actually needed. As suggested in the issue, I think a reasonable work-around for now would be to just comment-out the check in src/core_atmosphere/physics/physics_wrf/module_ra_rrtmg_sw.F around line 1701:

         !   if (pmid(i,1).lt.pmid(i,2)) then
         !      stop 'MCICA_SUBCOL: KISSVEC SEED GENERATOR REQUIRES PMID FROM BOTTOM FOUR LAYERS.'
         !   endif

 

[xtian15]

Thanks for your response!