MOM6 static file mask missing values

[hkershaw-brown]

Description:

CESM3 MOM6 TL319_t232 (new workhorse grid) has missing/fill values in the static file for geolat|lon|u|v written out by CESM.

This pull request:

• checks for fillvalue in the grid and uses the fillvalue locations as the mask to pass to quad_utils_mod.
• The grid == fillvalues are also set to a dry land lat,lon. This is to stop set_location blowing up with lon = 1e+20.
  Note the missing value locations are not the same as the ocean_geometry 'wet' mask.
• Forces lon to be [0-360]

Also I have set the MOM6/work/input.nml cutoff to something not giant. Not sure if 0.03 is appropriate for the ocean.
& set the perfect_model_obs namelist init_time_days to -1 -1.

Fixes issue

fixes #685

Types of changes

• Bug fix (non-breaking change which fixes an issue)
• New feature (non-breaking change which adds functionality)
• Breaking change (fix or feature that would cause existing functionality to not work as expected)
• Documentation update

Documentation changes needed?

• My change requires a change to the documentation.
  • I have updated the documentation accordingly.

Tests

Please describe any tests you ran to verify your changes.

Running cesm3_0_alpha03a
/glade/derecho/scratch/hkershaw/DART/Crocodile/DART_Runs/c.TL319_t232.001
perturbed from a single instance.
CESM:

/glade/derecho/scratch/hkershaw/DART/Crocodile/CESM((cesm3_0_alpha03a))
/glade/derecho/scratch/hkershaw/DART/Crocodile/Runs/c.TL319_t232.001
../CESM/cime/scripts/create_newcase --run-unsupported --res TL319_t232 --compset C_JRA --case c.TL319_t232.001 --project P86850054

Checklist for merging

• Updated changelog entry
• Documentation updated
• Update conf.py

Checklist for release

• Merge into main
• Create release from the main branch with appropriate tag
• Delete feature-branch

Testing Datasets

• Dataset needed for testing available upon request
• Dataset download instructions included
• No dataset needed

[hkershaw-brown]

note added this fix to https://github.com/NCAR/DART/tree/mom6-scripting
which has the mom6 assimilate.sh that Enrico is kicking the tires on.

[mjs2369]

DART/models/MOM6/model_mod.f90

Lines 621 to 624 in f04bf42

	! mom6 has missing values in the grid
	mask(:,:) = .false.
	call nc_get_attribute_from_variable(ncid, 'geolon', '_FillValue', fillval)
	where (geolon == fillval) mask = .true.

Why don't we need to repeat this process done in the lines above for geolon_u and geolon_v and have a separate mask array for each? These mask arrays, for example, mask_geolon_u would be passed into their respective calls to set_quad_coords. Example:
call set_quad_coords(interp_u_grid, geolon_u, geolat_u, mask_geolon_u)

Is this because the missing values are always going to be in the same place for gelon, geolon_u, and geolon_v? And therefore the mask array would be simply be duplicated for geolon_u and geolon_v?

[hkershaw-brown]

Is this because the missing values are always going to be in the same place for gelon, geolon_u, and geolon_v? And therefore the mask array would be simply be duplicated for geolon_u and geolon_v?

There are in the same place for the case , but you're correct there's no guarantee that that will always be the case. I'll add a mask for each

[mjs2369]

There are in the same place for the case , but you're correct there's no guarantee that that will always be the case. I'll add a mask for each

Should we have a separate array for the fillval as well or is that not needed?
fillval, fillval_u, fillval_v

[hkershaw-brown]

There are in the same place for the case , but you're correct there's no guarantee that that will always be the case. I'll add a mask for each

Should we have a separate array for the fillval as well or is that not needed? fillval, fillval_u, fillval_v

Do you mean move the where (geolon_u == fillval) geolon_u = 72.51 in case the fill values are different for each variable? Yeah probably wise.

call nc_get_attribute_from_variable(ncid, 'geolon_u', '_FillValue', fillval)
where (geolon_u == fillval) mask_u = .true.  
where (geolon_u == fillval) geolon_u = 72.51`

[mjs2369]

Do you mean move the where (geolon_u == fillval) geolon_u = 72.51 in case the fill values are different for each variable? Yeah probably wise.

call nc_get_attribute_from_variable(ncid, 'geolon_u', '_FillValue', fillval)
where (geolon_u == fillval) mask_u = .true.  
where (geolon_u == fillval) geolon_u = 72.51`

Yep that's exactly it

And do where (geolat_u == fillval) geolat_u = 42.56 as well before the next call to nc_get_attribute_from_variable

So the full thing would be

call nc_get_attribute_from_variable(ncid, 'geolon_u', '_FillValue', fillval)
where (geolon_u == fillval) 
  mask_u = .true.  
  geolon_u = 72.51
where (geolat_u == fillval) geolat_u = 42.56

[mjs2369]

Everything looks good. Nice fix Helen

I ended up testing with the case you had in /glade/derecho/scratch/hkershaw/DART/Crocodile/DART_Runs/c.TL319_t232.001

I was getting the error message shown in the corresponding issue when running filter on the main branch, but I was able to run filter successfully on the mom6-missing-grid branch with all the latest commits.