New calculation for accounting of root distribution on the transpiration

[ccarouge]

The calculation of ssnow%evapfbl is changed significantly by the ground water work for the non-Haverd2013 case. From what I understand, it is trying to spread the demand for water through the soil layers due to the root distribution.

The current calculation is:

CABLE/src/science/canopy/cbl_dryLeaf.F90

Lines 518 to 521 in 630e2e1

	ssnow%evapfbl(i,kk) = MIN( evapfb(i) * veg%froot(i,kk), &
	MAX( 0.0, REAL( ssnow%wb(i,kk) ) - &
	1.1 * soil%swilt(i) ) * &
	soil%zse(kk) * density_liq )

The new calculation is:

CABLE/src/science/canopy/cbl_dryLeaf.F90

Lines 530 to 547 in 42bf0c5

	! MMY to fix the inconsistence in root water extraction between ! FEEDBACK(MMY?)
	! cable_gw_hydro and above
	xx = 0._r_2 ! demand : transpiration
	xxd = 0._r_2 ! difference : demand - supply
	diff(:) = 0._r_2 ! supply : water available
	DO kk = 1,ms
	xx = evapfb(i) * veg%froot(i,kk) + diff(kk-1) ! demand for layer kk
	diff(kk) = max(0._r_2,ssnow%wbliq(i,kk)-soil%swilt_vec(i,kk)) * soil%zse(kk) * 1000.0
	! supply for layer kk
	xxd = xx - diff(kk) ! deficit in layer kk
	if (xxd .gt. 0._r_2) then ! demand > supply
	ssnow%evapfbl(i,kk) = diff(kk) ! transpiration in layer kk is supply
	diff(kk) = xxd ! deficit in layer kk
	else ! demand < supply
	ssnow%evapfbl(i,kk) = xx ! transpiration in layer kk is demand
	diff(kk) = 0._r_2 ! no deficit in layer kk
	end if
	END DO !ms

I have absolutely no idea if it's a good calculation to introduce. If it's a good idea, I don't know if we want to introduce it now or park it away for later.

@har917 I think that's the last one from the ground water work where I need some input.

[har917]

On this one - yes this is probably a good idea to pull across - it avoids over extraction of liquid water from a layer with lots of ice in it.

But I'm a bit confused as to how this intersects with the code in cbl_remove_trans and remove_transGW (for the non-Haverd2013 case) - this seems only to be setting %evapfbl which isn't then used.

A couple of comments:

• the 1000. at line 537 needs to be a density_liq
• I'd be happier if the factor 1.1 was included - comments from Matthias indicates that this safety net is needed.
• this method does conceptually allow for extraction of water from soil layers where there are no roots.
• this method doesn't actually ensure that evapfb can actually be satisified - or that wbice/wb < frozen_limit

[ccarouge]

Coming back to this, @har917 you are right that the code in cbl_dryLeaf.F90 that updates evapfbl for the non-Harverd2013 case is useless. And looking at cbl_remove_trans.F90, the distribution of transpiration through the layers is already updated to almost exactly what is coming from the groundwater work. In short, we could remove the calculation from cbl_dryLeaf.F90. I'm wondering if the Haverd2013 calculation could be moved to cbl_remove_trans.F90 as well, but unsure I will have the time to look at this within this issue. Some differences noted:

• use of wb instead of wbliq. Could update in cbl_remove_trans.F90 considering your comment in frozen water.
• No factor 1.1 which we may want to add to cbl_remove_trans.F90 again following Ian's comment above.
• Need to check that canopy%fevc*dels/CHL is the same as evapfb (most likely it is)

[ccarouge]

I am comparing remove_trans and remove_transGW. They typically have the same calculation except one is removing the transpiration from wb directly while the other is removing it from wbliq then recalculating wb (noting the difference between GW and non-GW in the wb definition). Taking either approach should end up at the same point since the transpiration should come from the liquid portion of soil water but I'm guessing it would need some extensive testing before we can remove the duplication.