ESM1.5 Reconciliation - Githubissues

penguian commented 2 years ago

keyword_maygit owner:jxs599@nci.org.au resolution_fixed type_model improvement | by srb001@csiro.au

Issue migrated from trac:332 at 2023-11-27 11:41:35 +1100

penguian commented 2 years ago

@jxs599@nci.org.au commented

Updateing ESM1.5 to use CABLE@HEAD

This ticket subsumes #314 which will be closed.

Following initial "merger", differences were noted - however this was always to be expected. We are re-doing this merging incrementally and highlighting differences as we go.

penguian commented 2 years ago

@jxs599@nci.org.au commented

Noted differences are being "plotted" here

https://accessdev.nci.org.au/p66/jxs599/ESM1.5/CABLE3/.

Not exhaustively at all. I am plotting the difference in the monthly mean surface temp. for January (i.e. the first month of the run), where the 2 models being compared are ESM1.5 vs ESM1.5-CABLE3, where I am incrementally replacing existing ESM1.5 code with CABLE@HEAD, and checking the "plot". Where I can achieve ZERO difference, I move on.

My intention is to replace "infrastructure" code first so that we can better focus on the differences due to the differing scientific processes

A known source of differences are updated parameter values for refl and taul used in ACCESS-CM2. Also CM2 uses smaller thresholds for minimum LAI for a cell to be considered as vegetated, and minimum SW radiation for a cell to be considered as sunlit.

I have created a branch on our repo, that contains the imported CMIP6 version of ESM1.5 (boundary_layer/ ONLY).

The first revision includes all modifications made to backport CABLE3 up until the revision where I posted that the 2nd difference was noted: https://accessdev.nci.org.au/p66/jxs599/ESM1.5/CABLE3/Roughness2/

In summary, the 1st difference was not real - caused by the crazy order of operations issue. The 2nd was due to including an extra restriction on the roughness length over ICE regions.*

Note that there are some science routines here already broken up into separate files. They are BUT the executable section is from standard ESM1.5 code. I do the same later with the soilsnow code.

https://trac.nci.org.au/trac/cable/changeset/8794

Further note that this version deliberately restores ESM1.5 PFT parameters and thresholds.

The next revision just seperates the soilsnow subroutines: https://trac.nci.org.au/trac/cable/changeset/8795

As noted above the first difference: https://accessdev.nci.org.au/p66/jxs599/ESM1.5/CABLE3/Roughness/

The second difference: https://accessdev.nci.org.au/p66/jxs599/ESM1.5/CABLE3/Roughness2/