[FEATURE REQUEST] Retire met-field LWI in GC-Classic

[lizziel]

This feature request is to (1) remove use of land-water-ice index (LWI) meteorology input in GC-Classic, and (2) remove the GEOS-Chem State_Met%LWI.

The motivation for this is the land-water-ice index in the GEOS-Chem input meteorology will not be available in the GEOS-IT dataset. Updates have already removed the dependency in GCHP by using other input met-fields to determine if each grid cell is predominantly land, water, or ice.

geos-chem/Interfaces/GCHP/Includes_Before_Run.H

Lines 65 to 74 in 412ba59

	! Set LWI based on land type fractions in imports: 0=water, 1=land, 2=ice
	DO J=1,State_Grid%NY
	DO I=1,State_Grid%NX
	IMAXLOC = MAXLOC( (/ FRLAND(I,J) + FRLANDIC(I,J) + FRLAKE(I,J), &
	FRSEAICE(I,J), &
	FROCEAN(I,J) - FRSEAICE(I,J) /) )
	IF ( IMAXLOC(1) == 3 ) IMAXLOC(1) = 0 ! reset ocean to 0
	State_Met%LWI(I,J) = FLOAT( IMAXLOC(1) )
	ENDDO
	ENDDO

There is only one location in GEOS-Chem that uses LWI for computation, in dry deposition.

geos-chem/GeosCore/drydep_mod.F90

Lines 2271 to 2273 in 412ba59

	IF ( State_Met%LWI(I,J) == 0 ) Hplus = HPLUS_ocn
	IF ( State_Met%LWI(I,J) == 1 ) Hplus = HPLUS_lnd
	IF ( State_Met%LWI(I,J) == 2 ) Hplus = HPLUS_ice

GEOS-Chem also passes LWI to HEMCO which uses it in several extensions. HEMCO will thus need to be updated to use the available land type met-fields to do this instead, and I'll make a separate feature request in HEMCO for this.

I will work on getting this update into 14.1. It will impact the benchmark but the impact should be minor.

[yantosca]

For reference, here is the definition of LWI in GEOS-5.

LWI: A Land-Water-Ice mask provided for backward compatibility with the GEOS-4 product. It is 1 over continental areas, 0 over open ocean, and 2 over sea-ice covered ocean. Since in GEOS-5 a grid box can be a combination of these, continental areas are arbitrarily defined as those where FRLAND+FRLANDICE>=0.5. The remaining grid boxes are designated as sea-ice if the ice cover exceeds 50%; otherwise they are open (ice-free) ocean.

Note, this field was introduced in GEOS-5 for backwards compatibilty with the GEOS-4 SURFTYPE variable. LWI is going to be retired from the newer GEOS-IT dataset as it can be created from a combination of other fields:

[lizziel]

Thanks @yantosca. That definition of LWI interestingly does not say anything about FRLAKE (fraction lake). I've been qualifying grid cells as over continental land if FRLAND + FRLANDICE + FRLAKE is greater than both FROCEAN - FRSEAICE (open ocean) and FRSEAICE (ocean ice). If not continental land then I assign open ocean or ocean ice based on which fraction is larger. These are mostly used in HEMCO extensions to determine if over ocean, land, or ice. HEMCO includes an additional qualifier that FRLANDICE >= 0.5 be classified as ice. This is different from LWI that classifies ice as only over ocean.

How tied do you think we should be to the old GEOS-4 definition when switching from LWI to fraction land types?

[yantosca]

@lizziel, we probably shouldn't be too tied to GEOS-4.

[lizziel]

I have a couple draft PRs almost ready for review. Integration tests are in progress.
#1435
geoschem/HEMCO#170

[msulprizio]

These PRs are now included in the dev/14.1.0 branch. I will close out this feature request.