MPAS crashing with MYNN PBL/SFC scheme

[MattW]

Hi MPAS folks,

I've run into an issue with MPAS where if I try to use the mynn PBL and surface layer schemes with the mesoscale_reference parameterization suite, MPAS will crash as soon as it tries to write out the first hourly restart file. My configuration of MPAS will run successfully with the mesoscale_reference suite of parameterizations, but the 2m temperature has a large cold bias, which is why I'm trying to run it with a different PBL and surface layer scheme. I've copied the physics section of my namelist.atmosphere below

&physics
config_sst_update = false
config_sstdiurn_update = false
config_deepsoiltemp_update = false
config_radtlw_interval = '00:30:00'
config_radtsw_interval = '00:30:00'
config_o3climatology = true
config_bucket_update = 'none'
config_physics_suite = 'mesoscale_reference'
config_gwdo_scheme = 'off'
config_sfclayer_scheme = 'sf_mynn'
config_pbl_scheme = 'bl_mynn'
config_microp_re = true
! config_radt_cld_scheme = 'suite'
config_radt_cld_scheme = 'cld_fraction_thompson'

I've also tried running this with config_gwdo_scheme set to 'by_ysu_gwdo' and got the same error. Any ideas as to what might be causing this? I'm running this on derecho in the /glade/derecho/scratch/mawilson/mpas_OSSE/MPAS-DART/rundir3/param_test directory if that helps.

Thanks,

Matt Wilson

 

[mgduda]

From your log.atmosphere.0000.out file, the error message suggests a problem with the specification of the GWDO scheme:

----- Setting up physics suite 'mesoscale_reference' -----
ERROR:
ERROR: ------------------------------ FATAL CALLED ------------------------------
ERROR: illegal value for gwdo_scheme: bl_mynn_gw
CRITICAL ERROR: MPAS core_physics abort

From your namelist.atmosphere file in /glade/derecho/scratch/mawilson/mpas_OSSE/MPAS-DART/rundir3/param_test, the &physics namelist group looks like

&physics
config_sst_update = false
config_sstdiurn_update = false
config_deepsoiltemp_update = false
config_radtlw_interval = '00:30:00'
config_radtsw_interval = '00:30:00'
config_o3climatology = true
config_bucket_update = 'none'
config_physics_suite = 'mesoscale_reference'
config_gwdo_scheme = 'bl_mynn_gwdo'
config_sfclayer_scheme = 'sf_mynn'
config_pbl_scheme = 'bl_mynn'
config_microp_re = true
! config_radt_cld_scheme = 'suite'
config_radt_cld_scheme = 'cld_fraction_thompson'

/

From Table 6.3 in the MPAS-Atmosphere Model User's Guide, it looks like the only valid specification for config_gwdo_scheme is 'bl_ysu_gwdo'.

 

[mgduda]

Also, from your log file, it looks like you're running MPAS-A v8.0.1:

MPAS-Atmosphere Version 8.0.1

Would it be possible to update to the current release (v8.2.2), at least for a single test run?

 

[MattW]

From your log.atmosphere.0000.out file, the error message suggests a problem with the specification of the GWDO scheme:

From your namelist.atmosphere file in /glade/derecho/scratch/mawilson/mpas_OSSE/MPAS-DART/rundir3/param_test, the &physics namelist group looks like

From Table 6.3 in the MPAS-Atmosphere Model User's Guide, it looks like the only valid specification for config_gwdo_scheme is 'bl_ysu_gwdo'.

Oops I forgot to mention that I ran an additional test setting config_gwdo_scheme to 'bl_mynn_gwdo' and found out it doesn't exist. I'll run it with 'bl_ysu_gwdo' again to get the error file I was encountering earlier so you can look at it. And I can also check out MPASv8.2.2 later.

 

[MattW]

I just tried compiling MPASv8.2.2 for a test run and got a segmentation fault using the same environment I used for compiling v8.0.1. The modules I have loaded are:

Currently Loaded Modules:
1) ncarenv/23.09 (S) 3) cray-mpich/8.1.27 5) craype/2.7.31 7) netcdf-mpi/4.9.2
2) gcc/13.2.0 4) ncarcompilers/1.0.0 6) hdf5-mpi/1.14.3 8) parallel-netcdf/1.12.3

I've also attached the error file I got, and the directory where I'm trying to compile MPASv8.2.2 is /glade/work/mawilson/MPAS_MYNN2/MPAS-Model. Do any of the modules listed above look incorrect?

 

[mgduda]

The Noah-MP land-surface scheme that was added in MPAS v8.2.0 breaks the GNU 13.2.0 compilers. This does appear to be due to a compiler bug and not a result of any problem with the Noah-MP code. Instead, you can use the GNU 12.2.0 compilers on Derecho, and if it helps, there's an updated module list for Derecho in the most recent MPAS-A tutorial guide: MPAS Tutorial — Practice Session Guide .

Thanks for your willingness to give a try with the current MPAS release!

 

[MattW]

Thanks! Changing the compiler to v12.2.0 allowed MPAS to compile. In an unrelated question, is there an MPAS mesh template somewhere that starts at 120 km resolution and refines down to 3 km? The closest I could find was 60 km refining down to 3 km.

 

[mgduda]

As you've found, the 60-3 km mesh is probably the closest that we have to what you're looking for.