Update column weighting upon disturbance

[glemieux]

The column weight for the various upstream types will need to the have the weights updated when a given soil column is disturbed.

See compute_higher_order_weights for reference:

E3SM/components/elm/src/main/subgridWeightsMod.F90

Lines 236 to 253 in 67abd00

	do l = bounds%begl, bounds%endl
	t = lun_pp%topounit(l)
	lun_pp%wtgcell(l) = lun_pp%wttopounit(l) * top_pp%wtgcell(t)
	end do
	do c = bounds%begc, bounds%endc
	l = col_pp%landunit(c)
	col_pp%wttopounit(c) = col_pp%wtlunit(c) * lun_pp%wttopounit(l)
	col_pp%wtgcell(c) = col_pp%wtlunit(c) * lun_pp%wtgcell(l)
	end do
	do p = bounds%begp, bounds%endp
	c = veg_pp%column(p)
	veg_pp%wtlunit(p) = veg_pp%wtcol(p) * col_pp%wtlunit(c)
	veg_pp%wttopounit(p) = veg_pp%wtcol(p) * col_pp%wttopounit(c)
	veg_pp%wtgcell(p) = veg_pp%wtcol(p) * col_pp%wtgcell(c)
	end do
	end subroutine compute_higher_order_weights

Note: lun_pp type is landunit_physical_properties

Tasks

Preview Give feedback

1. topounit weight (col_pp%wttopounit)
2. vegetated landunit weight (col_pp%wtlunit)
3. gridcell weight (col_pp%wtgcell)

[glemieux]

Note that the HLMs seem to have moved towards only using the no_special version of update_column_state:

E3SM/components/elm/src/dyn_subgrid/dynColumnStateUpdaterMod.F90

Lines 84 to 89 in 67abd00

	! NOTE(wjs, 2017-02-24) The implementation involves a few levels of calls to get to the
	! real work routine (update_column_state). This design made sense (in terms of removing
	! code duplication) when there were a number of "front-end" routines, with different
	! special handling. But now it seems we're moving towards having no special handling.
	! If we ditch all of the special handling, we could also ditch these extra levels. This
	! would make the code more straightforward, and would also improve performance.

A search through the dynColumnStateUpdatedMod module appears to confirm this. Additionally, the calls are never made by passing fractional_area. For example:

E3SM/components/elm/src/dyn_subgrid/dynSubgridAdjustmentsMod.F90

Lines 425 to 429 in 67abd00

	call update_column_state_no_special_handling( column_state_updater , &
	bounds = bounds, &
	clump_index = clump_index, &
	var = decomp_cpools_vr(begc:endc, j, l), &
	adjustment = adjustment_one_level(begc:endc))

As such, the gridcell column weight is used to determine the state update using the column_state_updater_type variable area_gained_col. Here is where gain is calculated:

E3SM/components/elm/src/dyn_subgrid/dynColumnStateUpdaterMod.F90

Lines 268 to 272 in 67abd00

	this%cwtgcell_new(c) = col_pp%wtgcell(c)
	this%area_gained_col(c) = this%cwtgcell_new(c) - this%cwtgcell_old(c)
	if (this%area_gained_col(c) /= 0._r8) then
	this%any_changes(clump_index) = .true.
	end if

Note that the grid cell weight is also directly used in the state update:

E3SM/components/elm/src/dyn_subgrid/dynColumnStateUpdaterMod.F90

Lines 911 to 914 in 67abd00

	if (fractional_area_new(c) /= 0._r8) then
	var(c) = (this%cwtgcell_old(c) * var(c) * fractional_area_old(c) + mass_gained) / &
	(this%cwtgcell_new(c) * fractional_area_new(c))
	end if

Note that fractional_area_new is always one since it's never been used as an input (old as well):

E3SM/components/elm/src/dyn_subgrid/dynColumnStateUpdaterMod.F90

Lines 751 to 754 in 67abd00

	if (present(fractional_area_new)) then
	my_fractional_area_new(bounds%begc:bounds%endc) = fractional_area_new(bounds%begc:bounds%endc)
	else
	my_fractional_area_new(bounds%begc:bounds%endc) = 1._r8