ACCESS-ESM1.6 development

[inh599 (Ian Harman)]

Maps of the difference between @pearseb’s test3 and the new spin up candidate (years 40-60 after the restart - so timings aren’t quite aligned).

Most of these features are tied to differences in precip - so possibly a signal of interannual variability rather than a more systematic change in carbon cycle.

Most of the surface cooling (associated with the revised solar constant) is in the northern hemisphere - though note that the 2K near-Antarctic warming signal seen in previous comparisons between December 2024 runs and February 2025 runs is also mitigated).

[inh599 (Ian Harman)]

Current time series of surface ocean temperature, and volume averaged ocean temperature from the current runs (test3 is an extension of @pearseb’s Spin up started in December i.e. older CABLE) Note that the two April Spinup start from the year 61 of the Test3 ocean (land restart is sourced from MarchEpot test run). Key difference between these simulations is that solar constant is reduced (to CMIP7 value) in the April run (and other land-based differences).

Equivalent plots for surface and volume-averaged salinity. Key difference is that the updates to CABLE-UM coupling have resulted in better water conservation. lprec is a tuning factor used to add fresh water to the ocean (ideally should =0, i.e. green line; blue and red use the same value).

[cbull (Chris Bull)]

Meeting minutes from today’s standup, feel free to update. @dougiesquire, @Jhan, @pearseb, @spencerwong, please see assigned action items and amend notes as requested.

[spencerwong]

Hi everyone,

Following this morning’s discussion, @dougiesquire and I have looked into the UM versions used in @pearseb’s spin up simulation. The git history for manifests/exe.yaml file in the experiments repository shows that the same UM executable was used in all the years that have been uploaded to the repository (0-766). We can confirm this is true for the remaining years once @pearseb has pushed the remaining runlog commits to Github. The UM executable used is at:

/g/data/vk83/prerelease/apps/spack/0.22/restricted/ukmo/release/linux-rocky8-x86_64_v4/intel-19.0.3.199/um7-git.dev-access-esm1.6_access-esm1.6-zpehauutrxs7mi2plup4zmj33kh3ofoi/bin/um_hg3.exe.

The executable was last modified on December 5, which suggests it may have been built from commit 457d1da in the UM repository.

We’ve tried separately rebuilding a UM executable by specifying this commit hash, and reran year 770 of the spinup with the new executable. The output with the rebuilt executable is identical to the original spinup, suggesting that this is the correct version.

It would be great to for the more recent experiments to be uploaded to the experiments repository, as it helps a lot with this type of investigation.

[Jhan]

This certainly brings the CMT change into play. The problem then is - this was a correction to an oversight in CM2, so do we want it?

[Jhan]

@RachelLaw @tiloz - BREAKING NEWS - CMT is not active. namelist element deep_cmt_opt =1, a non-existent case. Is this supposed to be set to case = 2 @arnoldsu ?

[RachelLaw]

This would be the pull request when the CMT fix went in: activate CMT by JhanSrbinovsky · Pull Request #43 · ACCESS-NRI/UM7, dated Dec 17th.

[tiloz (Tilo Ziehn)]

The new field table is now being used. This is the April run (AprilSpinUp) and we can see the big change in the ocean net carbon flux (annual mean values [PgC/yr]). @pearseb is this what you expected? Looks like we went too far the other way.

[tiloz (Tilo Ziehn)]

Update on ocean net carbon flux. Looks a lot better now and very close to zero towards the end of the period run so far.

[clairecarouge (Claire Carouge, ACCESS-NRI Land Modelling Team Lead)]

Meeting minutes from today’s standup, feel free to update. @dougiesquire, @Jhan, @pearseb, @anton, please see assigned action items and amend notes as requested.

[clairecarouge (Claire Carouge, ACCESS-NRI Land Modelling Team Lead)]

From @Jhan :

@cbull or anyone else who knows - how do I add a new table to ESM1.6 Spin-up experiments

Answer: in the edit window, click on the gear icon in the toolbar and “insert table”. But I don’t think you need a new table. I think you need to update the last table. For this, don’t open the edit window, but hover on the table you want to edit and a pen icon appears to open a nicer table editor.
In that last table, the “years” field should contain a comment that it is a continuation from the previous run but the simulated years have been reset. And I thought we didn’t start from 0?

[Jhan]

DOH. I edited the existing table earlier, started this one before I got the notification that there was a response. This morning I was told I needed a new table as the field t able represents a departure point? Anyway, I added this.

The run started from 0, but used mixed restarts. From ~150 from a previous atmosphere run and year 902 for the rest. Technically the year starts from output000, but the main reason I changed it was just because it got confusing for us to talk about. For e.g. if 000 is considered to be 903, then we are currently at 1241 with an ocean that hasn’t settled yet, and then at what year did we add the new ocean field, and it just got messy. Now though Ian has made up some plots yesterday that show this a s a continuation from 902. Next week we are going (likely) change the namelist option for the convective transport scheme. Then the Aust PFTs. Honestly, IDK what people want to call the year now.

[cbull (Chris Bull)]

Now added today’s minutes from this morning’s standup.

As usual it’s a wiki, so please adjust or add additional details.

[cbull (Chris Bull)]

Minutes, please correct as needed.

---

@inh599. As discussed, I’ve followed up with @RhaegarZeng and @rbeucher. RhaegarZeng writes

Rhaegar Zeng: It is already been deployed on module analysis3 in xp65. Due to /g/data4 is offline, the module is not available right now.

What we’ve developed is a CMORiser — a tool designed to CMORise the raw output from ESM1_6. I’ve already had a quick catch-up with the land team to give them a brief introduction last week.

Documentation for the tool is available here:
ILAMB Workflow - New Version of ilamb-tree-generator for ESM1_6

This tool allows users to define which variables they want to CMORise, so at this stage, we haven’t produced a pre-CMORised dataset for them. Instead, they can use the tool to generate CMOR-compliant outputs as needed.

[cbull (Chris Bull)]

Minutes from yesterday’s standup, please correct/amend.

Re: Coordinating development threads for next phase of control run. @All, please fill in the table as discussed.

[cbull (Chris Bull)]

Due to the ACCESS-NRI staff retreat, next week’s standup (May 27) is cancelled.

@clairecarouge can you please cancel the invite.

[spencerwong]

Running AMIP configuration without land use from pre-industrial restart

@RachelLaw following the stand up I’ve tried running the AMIP configuration from a pre-industrial restart, with land use change switched off.

Based on some earlier advice from @MartinDix, the following steps were needed to cut out the coupling fields and set the date from the PI restart, using /scratch/p66/jxs599/access-esm/archive/AprilSpinUp-AprilSpinUp-56aae814/restart942/ as an example.

Prepare the restart file:

cp -r /scratch/p66/jxs599/access-esm/archive/AprilSpinUp-AprilSpinUp-56aae814/restart942  <new-restart-location>
rm -r <new-restart-location>/ocean
rm -r <new-restart-location>/ice
rm -r <new-restart-location>/coupler

# Remove extra coupling fields from restart
module use ~access/modules
module load pythonlib/umfile_utils/access_cm2

python ~access/apps/pythonlib/umfile_utils/access_cm2/um_fields_subset.py -x 95,171,172,173,174,176,177,178,179,180,181,184,185,186,187,188,189,192,250,413,414,415,416,33001,33002 -i <new-restart-path>/atmosphere/restart_dump.astart -o <new-restart-path>/atmosphere/newrestart

rm <new-restart-path>/atmosphere/restart_dump.astart 
mv <new-restart-path>/atmosphere/newrestart <new-restart-path>/atmosphere/restart_dump.astart 

# Update the date in the restart file
python ~access/apps/pythonlib/umfile_utils/access_cm2/change_dump_date.py <new-restart-path>/atmosphere/restart_dump.astart  <<< "1978 1 1"
# Update the date in the calendar file
sed -i "s/end_date:.*/end_date: 1978-01-01 00:00:00/" <new-restart-path>/atmosphere/um.res.yaml

Prepare the configuration:

payu clone https://github.com/ACCESS-NRI/access-esm1.6-configs -B dev-amip -b <new-branch-name> <control-directory>

In <control-directory>/config.yaml replace the restart path with the path to <new-restart>. Also switch off the land use change by commenting out the lines:

userscripts:
    # Apply land use changes after each run
    run: ./scripts/update_landuse_driver.sh

Would l_luc also need to be set to false in the cable namelist?

Let me know if you run into any issues!

[RachelLaw]

Thanks for the instructions.
I got an error on the um_fields_subset.py command:

=== ERROR ===
The use of the #!/usr/bin/env python interpreter line in python scripts
has been deprecated.

Please modify this script:
/g/data/access/projects/access/apps/pythonlib/umfile_utils/access_cm2/um_fields_subset.py

To use either #!/usr/bin/env python3 or #!/usr/bin/env python2
depending on which version of python you require
Alternatively, if you are unable to modify this script
You can load the python2-as-python or python3-as-python
modules depending on which version of python you require

[spencerwong]

@RachelLaw let me know if replacing python with python3 in the command still fails.

[cbull (Chris Bull)]

Minutes from today’s standup, please correct/amend.

Reminder of two handy threads discussed today:

• Coordinating development threads for next phase of control experiment
• ESM1.6 Spin-up experiments