Additional IO needed for reproducibility in ACCESS2.0

[penguian]

keyword_nogit owner:jxs599@nci.org.au resolution_fixed type_ACCESS-CM2 | by m.decker@unsw.edu.au

---

Two new option, or_evap and fwsoil_switch='Haverd2013' introduce variables that need to be saved each time step to ensure reproducible runs independent of restarts.

or_evap=.true. The variable canopy%sublayer_dz needs is kept between timesteps and must be output and read in. Currently canopy%sublayer_dz is set to a constant in cable_um_init_subrs.F90 on the first timestep.

Fix: Create an additional UM variable in the D1 array (Stash code: section 0 836) to write sublayer_dz to the restart dumps. For simplicity, new ancil files have been created, instead it is set to a constant value at the start of the run.

Haverd2013: canopy%fwsoil is used in the dryLeaf subroutine before it is calculated (in contrast to the other options). It needs to be output and read in. Offline simulations have shown the model is not sensitive to not saving fwsoil, but the model cannot reproduce simulations once restart. There are two different ways to fix this issue. Fix #1: set canopy%fwsoil=1 at the start of the define_canopy subroutine. Fix #2: Follow the same methods as was done for sublayer_dz (create D1 variable).

I have chosen Fix#1 for simplicity, and set canopy%fwsoil=1.0.

The code including both changes is at ^branches/dev/markdecker/stable_vn10.6_CABLE_SSGW@10900

I have not yet tested the impact of these fixes.

---

Issue migrated from trac:194 at 2023-11-27 11:26:01 +1100

[penguian]

m.decker@unsw.edu.au commented

---

An updated rose suite that sets canopy%sublayer_dz is u-av593.

The new stashcode is: 0 | 839 |CABLE VISC LAYER DZ (m)

[penguian]

@jxs599@nci.org.au changed status from new to closed

[penguian]

@jxs599@nci.org.au changed resolution from ` tofixed`

[penguian]

@jxs599@nci.org.au changed milestone from ` to1. Closed`

[penguian]

@ccc561@nci.org.au set keywords to nogit