I think DGLC needs area correction factors for fluxes sent back to the mediator

[billsacks]

As in ESCOMP/CISM-wrapper#69, I think DGLC needs to apply a correction factor between the model's internal areas and the grid cell areas on the mesh file that is used to create the mappings between components. As in that CISM issue, I think DGLC should have area correction factors for the runoff fluxes sent to the mediator, though I'm not 100% sure of this, and we should think this through further.

@mvertens @Katetc

[billsacks]

Now that DGLC does global sums, we also need to verify that we're using the right areas (mesh areas vs. internal areas) in doing those global sums.

[mvertens]

@billsacks - where does this issue stand? DGLC only knows about mesh areas - since it does not have any internal model areas. So I am confused as to how this area correction could be done.

[billsacks]

I had a long discussion about this and the related CISM issue with @Katetc @whlipscomb and Gunter a couple of weeks ago. Kate and Bill are going to lead the effort to fix this in CISM and DGLC, and I'll support as needed. We decided that, given the uncertainties about what is right here, we want to do some careful validation of conservation, probably by setting up artificial fluxes for which it will be easier to tell if this is being done right; this validation will probably encompass a lot of the needed effort.

DGLC does know about the internal areas:

CDEPS/dglc/dglc_datamode_noevolve_mod.F90

Lines 285 to 292 in 2684ab7

	! Determine "glc_area" ;
	! Sg_areas is in radians
	! SHR_CONST_REARTH is the radius of earth in m
	! model_internal_gridsize is the internal model gridsize in m
	do ng = 1,lsize
	Sg_area(ns)%ptr(ng) = (model_internal_gridsize(ns)/SHR_CONST_REARTH)**2
	end do