Closed penguian closed 5 years ago
@mrd599@nci.org.au _uploaded file CABLE_ticket197.pptx
(2419.5 KiB)_
@mrd599@nci.org.au commented
Attached presentation shows the surface energy budget problem in the ACCESS CM2 cable run (u-ax871) that prompted this analysis and various single column model experiments used to track it down and compare possible fixes.
@jxs599@nci.org.au changed status from new
to closed
@jxs599@nci.org.au changed resolution from ` to
fixed`
@jxs599@nci.org.au changed milestone from ` to
1. Closed`
@ccc561@nci.org.au set keywords to maygit
@ccc561@nci.org.au changed keywords from maygit
to keepgit
keyword_keepgit
resolution_fixed
type_ACCESS-CM2
| by inh599The JULES model evaluates the sensible heat flux as proportional to a difference (surface to atmosphere) in potential temperature; CABLE (broadly) evaluates the sensible heat flux as proportional to a difference in temperature. In an attempt to align CABLE with JULES within ACCESS, CABLE-within-ACCESS (UMv8.5) used two near surface air temperatures to force the model. i) The temperature from the UM (TL_1) is used directly within the radiation component of CABLE as %tvrad and ii) TL_1 is increment by an amount determined by the lowest model level as %tk elsewhere CABLE.
However the use of the different temperatures was incorrectly implemented leading to a ~2W/m2 energy imbalance in the surface as seen by the atmosphere (i.e. not within CABLE itself).
A review of the code base and underpinning literature identified three potential solutions
A) removal of the difference between %tvrad and %tk with no increment applied (option 2 in ppt)
B) removal of the difference between %tvrad and %tk retaining the increment in both(option 4 in ppt)
C) ensuring that use of %tvrad and %tk is correct throughout CABLE (option 3 in ppt)
Each of these options solves the underpinning energy balance issue with little change in short-run simulations.
After discussion, the approach identified for ACCESS-CM2 has been to i) retain the code changes (C) so that different temperatures could be used (in the future) but ii) to remove the temperature increment from both %tk and %tvrad (A).
The rationale for this decision is twofold: firstly, removal of the increment aligns the ACCESS version of CABLE closer to the offline model and is easier to document; secondly, the increment method does not address properly the underlying justification for the change and risks introducing other errors (even with a full and correct implementation). Retaining the code changes provides some security to future work and provides backwards compatibility to earlier versions of ACCESS (and possibly ACCESS-ESM1.5).
The underlying issue of the appropriate formulation of the sensible heat flux will require revisiting and impacts both offline and coupled CABLE.
Issue migrated from trac:197 at 2023-11-27 11:26:21 +1100