time stepping of canopy%cansto.

[JhanSrbinovsky]

WRT to oldcansto% (re-)initialization within define canopy. NOT the more thorough treatment of %cansto in general AS DISCUSSED IN #162.

@har917 it isnt active in the AM3 model, or the CMIP6 version of ESM1.5. It is blocked by the IF condition. Although it is active in the ESM1.5+CABLE3 version. Probably because offline was doing it. I think your 1,2,3 are in place, I'll have to double check and get back yo you. For 1 however I think my intention was to move it back to offline/cbm where the time-stepping in cbm is the same as in define_canopy anyway and have the same shared canopy for everyone?

Originally posted by @JhanSrbinovsky in #500

[JhanSrbinovsky]

in cable_serial, within the timestep loop, canopy%oldcansto is set to %cansto (i.e. oldcansto=cansto) BEFORE cbm is called. So then when it gets to define_canopy it is just swapping the same thing back i.e. cansto=oldcansto

serial

oldcansto=cansto
cbm

define_canopy

cansto=oldcansto
cansto

In CM2:
explicit driver

oldcansto=cansto
cbm

define_canopy

cansto

implicit driver

PB_oldcansto=oldcansto(from beginning of timestep)
cansto=oldcansto
cansto

oldcansto=PBoldcansto
cansto=oldcansto
cansto

cansto is effectively updated once per timestep. we effectively save it to the restart as/thru canopy_tile (UM field)

ESM1.5+cable3 has:

explicit driver

oldcansto=cansto
cbm

define_canopy

cansto=oldcansto(from beginning of timestep)
cansto

implicit driver

cansto=oldcansto
cbm

define_canopy

cansto=oldcansto(from beginning of timestep)
cansto

So think they are all OK, albeit awkward, and the cansto=oldcansto line in define_canopy can be removed @har917 ?

[har917]

@JhanSrbinovsky This looks okay (for now) but the complexity means that it is prone to be confused and errors creeping in during the future.

Perhaps one thing to check: at the moment we're only placing oldcansto in the prognostic bank (not both). I think we're okay - and changing things now is likely to

---

However if we want to clear things up a bit (in the short-term) I would be looking to :

1. move the pre-cbm setting of oldcansto = cansto into define_canopy,
2. removing the cansto=oldcansto in define_canopy, and
3. adding (in explicit_driver most likely) cansto=oldcansto as needed to keep the logic.

i.e. for serial it goes:

• cbm [ define_canopy [ oldcansto = cansto then update cansto ] ]

in ESM1.6 it goes

• explicit_driver [ cbm [ define_canopy [ oldcansto = cansto then update cansto] ] cansto=oldcansto ]
• implicit_driver [ cbm [ define_canopy [ oldcansto = cansto (which is still cansto from the start of time step) then update cansto] ] ]

and in CM2/AM3 it goes

• explicit_driver [ cbm [ define_canopy [ oldcansto = cansto then update cansto] ] cansto=oldcansto ]
• 1: implicit_driver [ PBcansto=cansto & PBocansto = oldcansto [ cbm [ define_canopy [ oldcansto = cansto then update cansto] ] ], cansto=PBcansto & oldcansto = PBocansto ]
• 2: implicit_driver [ PBcansto=cansto & PBocansto = oldcansto [ cbm [ define_canopy [ oldcansto = cansto then update cansto] ] ] ]

In the longer-term we should be looking to rewrite the code in define_canopy that evaluates the water fluxes (throughfall, evaporation from a wet canopy etc.) and updates cansto so that the fluxes come out of define_Canopy but cansto itself is time stepped separately (in the same/similar way as the soil moisture).

[JhanSrbinovsky]

In the CM3/AM3 I think we only need to PB cansto?

[har917]

In the CM3/AM3 I think we only need to PB cansto?

At the moment we are PB'ing oldcansto not cansto - I think this works as you've laid it out!!

It is confusing - especially the cansto = oldcansto lines that appear in some of the applications of define_canopy and not others.

More generally I find the mix of where the cansto variables are used/timestepped within cbm and define_canopy to be odd - and something that should be tackeld down the track

Actually I suggest that we work towards removing all instances of if um_explicit, if um_implicit etc. from the science routines BUT that's for later.