Only allocate MARBL coupled fields if MARBL is in use

[dougiesquire]

This PR prevents allocation and advertisement of MARBL coupled fields when MARBL is not in use.

See mom-ocean#1653 (comment) for motivation.

[dougiesquire]

Should the following block also be if (cesm_coupled .and. use_MARBL) then? The first comment suggests that when using CIME that is the behaviour.

MOM6/config_src/drivers/nuopc_cap/mom_cap.F90

Lines 598 to 619 in 623e49d

	if (cesm_coupled) then
	! Note that flds_i2o_per_cat is set by the env_run.xml variable CPL_I2O_PER_CAT
	! In CESM, this xml variable is set by MOM_interface's buildnml script and by
	! default it is false unless ICE_NCAT>0 and USE_MARBL_TRACERS=True
	call NUOPC_CompAttributeGet(gcomp, name='flds_i2o_per_cat', value=cvalue, rc=rc)
	if (ChkErr(rc,__LINE__,u_FILE_u)) return
	read(cvalue,*) i2o_per_cat
	if (is_root_pe()) then
	write(stdout,*) 'i2o_per_cat = ',i2o_per_cat
	endif
	! Note that ice_ncat is set by the env_run.xml variable ICE_NCAT which is set
	! by the ice component (default is 1)
	if (i2o_per_cat) then
	call NUOPC_CompAttributeGet(gcomp, name='ice_ncat', value=cvalue, rc=rc)
	if (ChkErr(rc,__LINE__,u_FILE_u)) return
	read(cvalue,*) Ice_ocean_boundary%ice_ncat
	endif
	if (is_root_pe()) then
	write(stdout,*) 'ice_ncat = ', Ice_ocean_boundary%ice_ncat
	endif
	end if

[mnlevy1981]

I ran tests with and without MARBL, and everything ran fine. The non-MARBL run reported the baseline comparison differences, because there are 19 fields that are no longer included in coupler history:

 Could not find match for file2 variable ocnImp_Faoo_fco2_ocn in file1
 Could not find match for file2 variable ocnExp_Faxa_bcph1 in file1
 Could not find match for file2 variable ocnExp_Faxa_bcph2 in file1
 Could not find match for file2 variable ocnExp_Faxa_bcph3 in file1
 Could not find match for file2 variable ocnExp_Faxa_dstdry1 in file1
 Could not find match for file2 variable ocnExp_Faxa_dstdry2 in file1
 Could not find match for file2 variable ocnExp_Faxa_dstdry3 in file1
 Could not find match for file2 variable ocnExp_Faxa_dstdry4 in file1
 Could not find match for file2 variable ocnExp_Faxa_dstwet1 in file1
 Could not find match for file2 variable ocnExp_Faxa_dstwet2 in file1
 Could not find match for file2 variable ocnExp_Faxa_dstwet3 in file1
 Could not find match for file2 variable ocnExp_Faxa_dstwet4 in file1
 Could not find match for file2 variable ocnExp_Faxa_ndep1 in file1
 Could not find match for file2 variable ocnExp_Faxa_ndep2 in file1
 Could not find match for file2 variable ocnExp_Fioi_bcphi in file1
 Could not find match for file2 variable ocnExp_Fioi_bcpho in file1
 Could not find match for file2 variable ocnExp_Fioi_flxdst in file1
 Could not find match for file2 variable ocnExp_Sa_co2diag in file1
 Could not find match for file2 variable ocnExp_Sa_co2prog in file1

The variables that were not removed are bit-for-bit, as expected (and the run with MARBL is bit-for-bit identical as well)

[mnlevy1981]

Should the following block also be if (cesm_coupled .and. use_MARBL) then? The first comment suggests that when using CIME that is the behaviour.

MOM6/config_src/drivers/nuopc_cap/mom_cap.F90

Lines 598 to 619 in 623e49d

	if (cesm_coupled) then
	! Note that flds_i2o_per_cat is set by the env_run.xml variable CPL_I2O_PER_CAT
	! In CESM, this xml variable is set by MOM_interface's buildnml script and by
	! default it is false unless ICE_NCAT>0 and USE_MARBL_TRACERS=True
	call NUOPC_CompAttributeGet(gcomp, name='flds_i2o_per_cat', value=cvalue, rc=rc)
	if (ChkErr(rc,__LINE__,u_FILE_u)) return
	read(cvalue,*) i2o_per_cat
	if (is_root_pe()) then
	write(stdout,*) 'i2o_per_cat = ',i2o_per_cat
	endif
	! Note that ice_ncat is set by the env_run.xml variable ICE_NCAT which is set
	! by the ice component (default is 1)
	if (i2o_per_cat) then
	call NUOPC_CompAttributeGet(gcomp, name='ice_ncat', value=cvalue, rc=rc)
	if (ChkErr(rc,__LINE__,u_FILE_u)) return
	read(cvalue,*) Ice_ocean_boundary%ice_ncat
	endif
	if (is_root_pe()) then
	write(stdout,*) 'ice_ncat = ', Ice_ocean_boundary%ice_ncat
	endif
	end if

That's a good question. Let's leave this block as-is for now, but I can create an issue on the NCAR fork to investigate. I don't think CESM does anything with multiple ice categories unless MARBL tracers are turned on, though.

[mnlevy1981]

Test results are in my previous comment. I have two small suggestions for code changes, so I'll mark this as "Request changes" just to make sure they get seen. If it makes more sense to leave the cesm_coupled .and. use_MARBL, then I'll change my review to Approve.

Thanks so much for putting this together!