CSIRO - ACCESS-NRI standup minutes

CMIP7 CMIP7 development
1

Tuesday 15/10/2024 standup
Chair: @cbull
Attended: @sofarrell @RachelLaw @anton @Aidan @harshula @pearseb.
Minutes: @cbull

Pre-meeting agenda items:

  1. @sofarrell: Status of the github repository that ACCESS-NRI was going to set up based on the GSI8.1 sea ice code. @anton was going to link with @paulleopardi last I heard.
  2. @cbull: what does access-nri need to facilitate esm 1.6 release by Dec’? (See Aidan response below)

@RachelLaw: For ESM1.6 ocean work there are broadly 3 threads:

  1. wombat work (@pearseb)
  2. cice update ( @dhb599 / @sofarrell / @anton / @paulleopardi )
  3. pseudo iceberg work ( @dhb599 )

@pearseb: working towards a wombat lite generic tracers release in a few weeks.

@Aidan ACCESS-NRI need code on github on publicly available branch to facilitate release.

@RachelLaw: side point. Rachel would like to see a similar cable 3.1 discussion occurring. Presently, the code is in the atmosphere repo and it needs to be standalone. @Aidan: if useable as library can do CI compile correctness and other tests independently, and easily switch between CABLE versions.

@dhb599 update on current workflow. Currently using ESM1.5 with upgrades, would like to get them working before putting new components into esm1.6. Presently using run script infrastructure at CSIRO not payu, was imagining putting code on github aiming for the end of next month. @Aidan pointed out payu migration was not something @dhb599 would need to do. Can test with existing configs and tweak where necessary, and ACCESS-NRI could do this.

@anton (I think): in this instance, cice4–>cice5 the cm2 code using the oasis coupler is the best placed version. Need to chase @paulleopardi for the status of that.

@RachelLaw: OASIS updates to be compatible with UM7.3?

Time tendency diagnostics issue in CICE4 might be solved by moving to CICE5

@Aidan / @cbull. That timeline is too tight to facilitate the proposed release date. We’re happy to have still being developed-code as a starting point.

Action items:

  1. @cbull to follow up with Claire Carouge regarding a similar meeting for land work.
  2. @anton / @Aidan / @paulleopardi to get status update with cice code being available to csiro on github.
  3. @cbull to follow up with having an offline meeting with @dhb599 after his travel (next week), with the idea that the meeting results in @dhb599 ocean code being on github.
  4. @sofarrell and @dhb599 to have an in person meeting to discuss github workflows.
  5. @cbull will send out a reoccurring invite for the next meeting .

Forward plan:
No meeting 22/10/2024 but the next one on 29/10/2024.

1 Like
2

The 29/10 meeting coincides with eResearch Australasia 2024.

3

Tuesday 29/10/2024 standup
Chair: @cbull
Attended: @sofarrell @cbull @anton @tiloz
Apologies: @pearseb
Minutes: @cbull

Pre-meeting agenda items:
None.

@sofarrell has been cloning some GitHub Cable code and experimenting with making her own local branches. She recently completed a helpful Git tutorial. We had a discussion around: differences between Git / GitHub / svn, querying how to tell if your local Git repository has connected remotes (git remote -v), workflows related to GH issues / branches, git clone syntax (https vs ssh etc), git submodules in relation to --recursive command.

@sofarrell and @anton picked up an ongoing discussion around CICE output being odd in CM2 output as compared to normal CICE6 output re freeboard. Still need to do the comparison when we get the ESM1.6 output (@sofarrell to look when runs available).

@tiloz did not have any questions but is looking to come along in the future.

Next week we’re back on and the release team are available again. Though its a Public holiday in Melbourne.

4

Meeting notes from longer GitHub meeting (4/11/2024)
Present: Dave Bi, Siobhan O’Farrell, Aidan Heerdegen (left at 4pm), Chris Bull
@sofarrell @dhb599 @Aidan @cbull
Meeting scope: assist Dave and Siobhan get the model code they have been using onto GitHub to facilitate ESM 1.6 release.

Summary:
-Dave Bi (DaveBi) added to ACCESS-NRI GH organisation
-Dave Bi shown how to enable 2 FA authentication via text messaging (works!)
-Dave Bi is currently making changes to ESM1.5 code, decided to focus on cice4 for the moment (will need to do MOM too eventually).
-Showed Dave how to make a local clone of git clone https://github.com/ACCESS-NRI/cice4 (note this does not clone the other branches), origin is a shorthand name for the remote repository that a project was originally cloned from. git branch -a lists all the remote branches, noting
remotes/origin/access-esm1.5 is the branch you likely want. Hence git checkout origin/access-esm1.5 -b my-esm1.5 will create a local copy of this branch the string after ‘-b’ will be the local name for this branch.
-git status is very useful for describing the state of the current working directory in terms of the HEAD (“current view”) repository
-discussed how git works in terms of staging areas, adding then committing, then pushing changes. Here is a workflow:

git add filename
git status
git commit -m “This is my 1st commit try”
git push -u origin my-esm1.5

-git remote -v
-git help COMMAND very useful for getting up the help pages where COMMAND is substituted for a git command (e.g. commit)
-On last point for pushing to the remote, Dave Bi’s credential were not correct.

Next actions:
-Dave Bi will try to reset password (suggest username of either davebi or DaveBi) and then push the changes we discussed. See suggested links below.
-Dave Bi will then copy in changes that he’s made to the repository and repeat the above process to push them onto GitHub, he will notify us or let us know if he has any trouble.
Suggest: Dave Bi sets his username and password (see second link below)

Links:

5

Tuesday 12/11/2024 standup
Chair: @cbull
Attended: @Aidan, @pearseb, @sofarrell, @RachelLaw @anton
Apologies: @dhb599
Minutes: @cbull

Pre-meeting agenda items:

  • sinking scheme in generic tracers (move_vertical function?)
  • @cbull “port latest WOMBAT science developments”

@pearseb and @cbull talked about the interest in having variable sinking rates. Ideally it would be possible to have a solution that works for both MOM5 and MOM6 and could be pushed upstream to mom-ocean. Quite a technical discussion, so best followed up offline.l Next step: @cbull will be in touch with @pearseb to have an offline chat about options. Likely after Hobart so that might coincide with when @dougiesquire is back?

@RachelLaw heard from @dhb599 a few days ago was having a water conservation issue in MOM (coming from CICE).

@sofarrell has been waiting on CICE changes from @dhb599 [missed the version information here]

@Aidan gave an overview of the ESM 1.6 project board and how to add issues and manage them as they progress.

Discussion between @RachelLaw and @cbull about what to do regarding the stand up for next week. Given that @anton, @sofarrell, @cbull are both in Hobart for workshop. @RachelLaw was unsure if @dhb599 would be able to make it but will chat to him later in the week (update: @cbull heard from @dhb599 who said they can make it, so @cbull has organised @ezhilsabareesh8 to host the standup next week).

6

Tuesday 19/11/2024 standup
Chair: @ezhilsabareesh8
Attended: @Aidan, @pearseb, @RachelLaw, @dhb599, @harshula and @ezhilsabareesh8
Apologies: @anton, @cbull and @sofarrell
Minutes: @ezhilsabareesh8

Agenda:

  1. Sinking Scheme in Generic Tracer: @pearseb resolved the issue but plans to discuss it further with Dougie upon his return.
  2. Porting Wombat Science Development: @pearseb is implementing it in OM2, which will later be ported to ESM 1.6.
  3. The water conservation issue in MOM from CICE has been resolved.

@harshula provided an update on the Spack pull request for ACCESS-OM2. The PR for building generic tracer components using Spack has been completed and is available on GitHub (PR #86). The changes would take about two weeks to merge, as some modifications are required in the repositories to support the build. Once the work on OM2 is completed, the same approach will be applied to ESM 1.6.

@dhb599 update : Dave shared updates on the testing of the new land ice scheme. The test results indicate a significant improvement in the stability of ocean heat content compared to previous runs, where notable shifts in ocean heat were observed. The new scheme ensures that ocean heat content is balanced with latent heat loss by incorporating runoff adjustments, resulting in stable ocean heat. Dave plans to implement the new land ice scheme in ESM 1.5 and later in ESM 1.6, ensuring compatibility with CICE5 in both models. Aidan highlighted that incorporating the changes into ESM 1.5 would provide immediate value, especially for the paleo community and other users requiring more stable ocean heat content

Aidan proposed that the code changes for CICE and MOM modifications be integrated into ESM 1.5 via a pull request and updating experiment control parameters to align with these changes. Dave recommended conducting initial test runs with the updated code, particularly focusing on the Wombat implementation.

Aidan offered to assist Dave with resolving technical issues in pushing his changes to GitHub.

Next actions:

  • Follow-up discussion on the sinking scheme once Dougie returns.
  • Create ACCESS-ESM1.5 configuration incorporating Dave Bi’s iceberg scheme
  • Plan for integrating stable ocean heat features into ESM 1.5 and 1.6 based on test outcomes.
1 Like
7

Git minutes (21/11/2024)
Attended: @dhb599 @sofarrell @cbull

Following instructions from here.

Create SSH key (Siobhan didn’t have one but Dave already did):

ssh-keygen -t rsa -b 4096

Add SSH key (~/.ssh/id_rsa.pub) by adding in settings, i.e. here.

Then either clone a new repository with SSH keys, e.g., go here, then copy the SSH path, e.g.

git clone git@github.com:ACCESS-NRI/ACCESS-OM2.git

Then make some edits, follow standard practice of git commit then git push etc (Dave Bi was able to push effectively to a new repo’).

In the other case, where the repository already exists, update the link from the local branch to the remote (following cable instructions):

git remote -v #helpful to tell you what's going on!
git remote rm origin
git remote add origin <URL> #get this off the github repository 'code' link
git remote -v
git fetch 
git branch -u origin/<branch_name>
git push -u origin ESM_1.5
#this last command did not work, likely because of a lack of permissions

We’ll have a future meeting once credentials of Dave and Siobhan have been checked.

Action item:
@cbull ~follow up with @Aidan regarding credentials and workflow.~

1 Like
8

Tuesday 26/11/2024 standup
Chair: @cbull
Attended: @Aidan, @pearseb, @RachelLaw, @dhb599, @harshula, @anton, @sofarrell, @cbull
First draft of minutes: @cbull

@harshula asked @pearseb if he could add the cmake build files to the branch he is currently working on. In other words, @harshula asked @pearseb about which generic tracers branch he was currently working on. @pearseb explained that the sinking work is the main focus at the moment but one can look at the issues page. @pearseb thinks wombatlite will be ready by the end of the week and can be rolled out to other modelling components (e.g. ESM, OM etc). Background: the spack build is deployed from a separate development branch so they are looking to update those changes to be more relevant to what @pearseb is working on.

@anton asked @pearseb, if it would be helpful if @matthew.chamberlain were more involved in GitHub activities. @pearseb suggested Matt is working on testing of generic tracers but will feedback via @pearseb for the time being. @Aidan noted that @matthew.chamberlain uses GH in other contexts.

@sofarrell asked about these commands that @dhb599 has used succesfully to push to github

$ git add TEST-pushing-file?
$ git commit -m "second try on 20241125 to push TEST-pushing-file1/2"
$ git push origin HEAD

Key point is that git push origin HEAD, HEAD refers to the local commit viewpoint.

@sofarrell wanted to know how to rename her local branch then push it to GitHub this was done successfully (!) with:

git branch -v
git branch -m <newname>
git branch -v
git remote -v
git push origin -u HEAD

Note that the -u is important as it asks git to add tracking to the local branch.

See the docs for more information on what tracking means.

@RachelLaw mentioned that she’s found the cable docs really helpful.

2 Likes
9

Tuesday 10/12/2024 standup
Chair: @cbull
Attended: @Aidan, @pearseb, @RachelLaw, @anton, @sofarrell, @cbull, @ben, @inh599, @Jhan, @MartinDix, @spencerwong, @TommyGatti
Not present: @dhb599, @harshula, @tiloz, @Arnold_Sullivan, @tammasloughran, @clairecarouge
First draft of minutes: @cbull

@cbull suggested, given the size of the standup that we split into breakout rooms, and then reconvene at the end. This didn’t occur as the discussion went towards issues concerning everyone.

@Aidan, @harshula has been putting together a spack build with different options to allow flexibility given the different model versions currently being used, details here with a test pull request. Options include:

  • with/without generic tracers)
  • cice4/cice5
  • different UM versions
    @clairecarouge feedback would be helpful but she is currently travelling.

@MartinDix has been putting together some standard runs (1% and 4x CO2 Martin and was thinking about using generic tracers). Question then: is there a configuration using generic tracers that we can put in the config’ repo?

@pearseb, there is but the PR related to it is unavoidably slow because @dougiesquire is away and he normally reviews these things. It does however run well in ACCESS om2. @pearseb emphasises that we should be using generic tracers for all wombat work going forward. @MartinDix another isssue is getting the inputs into the standard locations.

@RachelLaw: what are our priorities to do simulations over the summer? Are there test cases or a configuration that we can put together (with possibly some known issues) that looks like ESM 1.6?

@RachelLaw / @pearseb. Tuning of WOMBAT will be different to ACCESS om2 but currently using ACCESS om2 parameter values as a starting point, expect that they will need some tuning.

@RachelLaw is liaising with @Arnold_Sullivan, @Jhan and @dhb599 but there are still considerable changes to the ocean components. @RachelLaw will see @dhb599 later today to get an update. @Aidan : CMT stuff needs an issue on GitHub please. @RachelLaw: ok, meeting with the relevant people (@Arnold_Sullivan, @Jhan – I think?) so we’ll make an issue for that in the meeting today.

@sofarrell is still waiting on CICE updates from @dhb599.

@RachelLaw suggestion to pull together what we have to run some standard simulations to be run over the summer break (see “create initial dev-preindustrial+concetrations branch” issue), thinking of having a meeting in the next ~10 days to organise. @pearseb is keen to facilitate too. @cbull great idea, like a mini hackathon? @Aidan is going on leave from this Friday

Spencer: Dougie put this together, does that have all the inputs that are needed? Pearse: the restart is what will be needed for the ocean with a few extra variables, we can discuss offline.

@cbull any land updates/comments?

@inh599 gave an update, missed the details [@inh599 looking to implement bug fixes in land-atmosphere coupling code this week. These were identified in CM2 development and need porting into ESM1.6. Unfortunately the UM differs enough between version that the fix code has to be developed anew. Should be done as part of/ahead of the hackathon.]

Next actions: @RachelLaw is going to organise a follow up meeting “hackathon” this month to pull together csiro efforts into one executable for test simulations.

— offline discussion with @pearseb @Aidan @cbull
Aidan: all inputs are tracked here. The files are then added to a protected area by a workflow that Tommy set up. Suggestion from Aidan: Pearse can do the modifications to the restart and then share in /scratch/public/username/ (has a very short deletion time; need to add permissions to everyone)
Update: @pearseb has shared the file and put here – /scratch/public/pjb581/ocean_wombatlite.res.nc-pearse

10

Would Tuesday 17th, 11.30-1.00pm, work for the proposed “hackathon”? This is already blocked out as a regular meeting for many of the potential CSIRO participants and also looks free for @pearseb and @matthew.chamberlain. How does it suit any potential ACCESS-NRI participants?

The aim would be to work out what we can build now i.e. how many of our different component changes can we currently build together, what we are aiming for, and if there are intermediate options. The notes from the meeting suggest some of this is in place already

Blockquote @Aidan, @harshula has been putting together a spack build with different options to allow flexibility given the different model versions currently being used, details here with a test pull request. Options include: …

11

I’ll log in from NZ. I am now compiling the ESM with my generic tracers and wombat lite. Will let everyone know how it goes.

12

I am just reading that now. It would be good to get the harvesting in for the summer runs. @tammasloughran @RachelLaw I believe the UM7 version used now is a “copy” of CABLE3 without the harvesting, but since Tammas has done all the work, it should be quite quick to add it.

13

Date: 16/12/2024
People present: @cbull, @matthew.chamberlain, @dhb599, @harshula (briefly), @anton (briefly)
Minutes: @cbull
Meeting purpose was to merge Dave Bi’s iceberg freshwater changes into ACCESS NRI repos such that they can use the ACCESS-NRI build system.

We had a look at current build of mom5 and that does not need any code changes.

Files changed by Dave Bi for CICE4 are located here: /g/data/p66/dhb599/trunk_ESM1.5/submodels/cice4.1

Any files modified after June 26th are his changes. Dave got code from Peter Dobrohotoff on June 26th, from this folder:
/g/data/p66/pbd562/dave

Dave then suggested two folders have relevant changes. Namely

/g/data/p66/dhb599/trunk_ESM1.5/submodels/cice4.1/drivers/access
-rw-r----- 1 dhb599 p66 11570 Sep 11 15:43 CICE_InitMod.F90 
-rw-r----- 1 dhb599 p66 59494 Nov  8 20:10 CICE_RunMod.F90 
-rw-r----- 1 dhb599 p66 66517 Nov  8 20:59 cpl_interface.F90 
-rw-r----- 1 dhb599 p66  8181 Nov  8 21:36 cpl_arrays_setup.F90 
-rw-r----- 1 dhb599 p66 10811 Dec  5 21:27 cpl_parameters.F90 
-rw-r----- 1 dhb599 p66 62537 Dec  5 21:29 cpl_forcing_handler.F90 

/g/data/p66/dhb599/trunk_ESM1.5/submodels/cice4.1/source
-rw-r----- 1 dhb599 p66  37704 Aug 19 16:05 ice_step_mod.F90
-rw-r----- 1 dhb599 p66 166829 Aug 19 22:54 ice_history.F90
-rw-r----- 1 dhb599 p66  51314 Aug 20 15:58 ice_init.F90
-rw-r----- 1 dhb599 p66  37336 Sep 11 15:50 ice_flux.F90

In practice, we did, on this repo

[2046][cyb561.gadi-login-09: cice4.1]$ pwd
/g/data/p66/dhb599/trunk_ESM1.5/submodels/cice4.1
[2045][cyb561.gadi-login-09: cice4.1]$ cp -r  * /home/561/cyb561/davetemp/cice4/

Which led to the following changes (note one additional file)

[2019][cyb561.gadi-login-05: cice4]$ git s
Refresh index: 100% (167/167), done.
HEAD detached at origin/update_DaveBi
Changes not staged for commit:
  (use "git add <file>..." to update what will be committed)
  (use "git restore <file>..." to discard changes in working directory)
	modified:   compile/comp_access-cm_cice.RJ.nP-mct
	modified:   compile/environs.raijin.nci.org.au
	modified:   drivers/access/CICE_InitMod.F90
	modified:   drivers/access/CICE_RunMod.F90
	modified:   drivers/access/cpl_arrays_setup.F90
	modified:   drivers/access/cpl_forcing_handler.F90
	modified:   drivers/access/cpl_interface.F90
	modified:   drivers/access/cpl_parameters.F90
	modified:   source/ice_flux.F90
	modified:   source/ice_history.F90
	modified:   source/ice_init.F90
	modified:   source/ice_step_mod.F90

Build attempt started at (using Dave’s CICE changes):
https://github.com/ACCESS-NRI/ACCESS-ESM1.6/pull/27
Update: compiled succesfully!

Next actions:

14

Tuesday 17/12/2024 standup
Chair: @cbull
Present: @cbull, @spencerwong, @RachelLaw, @ben, @harshula, @pearseb, @Jhan, @matthew.chamberlain, @tammasloughran
Not present: @dhb599, @Arnold_Sullivan, @clairecarouge
First draft of minutes: @cbull

@Chris/Pearse. ESM1.6 with CICE/wombatlite built yesterday! Yes, we had some group permissions issues that were approved for Matt last night.

@Jhan. Asked for an update on the build for ESM1.6?

@Harshula. From his side, he recently did UM7 to the dev-access-1.6 branch (spelling?) which is the the version that @jhan has been using. The PR that Jhan should be using is ‘update cable3 #10’ here.

@chris. More broadly recently there have been 3 main strands of effort for esm1.6:

  • 1
  • 2
  • 3
    Which address update to cable (no generic tracers), update to wombat (wombatlite) using generic tracers, update to wombat and cice.

@Jhan has been using ESM1.5’s versions of MOM5 with no-generic tracers but will start testing the other versions in the next few days.

@pearseb / @spencerwong. There’s a few files pearse would like to add to wombatlite.

@Chris/@pearseb/@harshula. Friday’s wombat PR was using wombatlite-sinking but the new one PR uses the development branch. Chris asked Pearse if this was ok? @harshula suggested that they’ve been merging Pearse development-branches (e.g. wombatlite-sinking) into development periodically behind the scenes. @pearseb, yes that workflow is okay but if I made a change to wombatlite-sinking yesterday what do I do? @harshula, let us know and we’ll merge it in.

@Rachel. Purpose of the hackathon at 11.30 today is to pull these changes together.

@Rachel still outstanding, has been asking @Arnold_Sullivan for the CMT change in the convection part of atmosphere code UM (it’s a small change). @Jhan said he’d follow up with @Arnold_Sullivan, hopefully we can get change before hackathon.

@RachelLaw / @ben. A little confusion on which branch of the access nri cable fork we should all be working off (@chris claire wrote a protocol doc on branch management [@clairecarouge please insert link]). @ben update from claire on what everyone should be using is here, namely

See Claire’s comment re coupled/esm1.6 directory (which is now coupled/esm16 because we don’t want to type full stops all the time)

In CABLE repository, changes for ESM1.6 to be added to coupled/ESM1.6/ directory (directory being created now).
In UM7 repository, changes for ESM1.6 to be pushed to the main branch.

Another update from @ben:

dev-access-esm1.6 branch here https://github.com/ACCESS-NRI/UM7

@RachelLaw. Any thoughts on how we can look at model output?
@tammasloughran. Did have a tool but because of the um2nc tool it doesn’t work with that output (due to the file name change). @spencerwong there is planned update to um2nc which will fix/help this issue.

15

Tuesday 7th Jan 2025

Attendees: @clairecarouge @anton @RachelLaw @spencerwong @ben @Jhan @inh599
Apologies: @cbull & everyone else still on leave

Claire:

  • Do we want to continue the current meeting format ?
    Happy enough to keep going with the current format, extend to the end of March.

Rachel:

  • @pearseb ran an ice-berg and no-iceberg case for ~100years
  • Jhan had two other runs to test CABLE configurations - ran for 40-70 years. Jhan and Rachel has some other tests of CABLE configurations ongoing.
  • CICE5 work still ongoing.
  • Need to determine the right split between liquid and solid run-off for the iceberg scheme. @dhb599 to advise and we can investigate further is needed.

Probably won’t extend CABLE runs until without checking the results in more detail. Rachel is working towards having the correct output to use ILAMB. Needs advice from Martin on how to calculate energy balance.

All CSIRO results are in :


/scratch/p66/rml599/ESM1.6-output/Jhan-run

Scripts are extract_*

Jhan's two runs are 'case2a' and 'case2b'

Ian:

Water balance on the land looks very wrong, will discuss with Martin when he is back. Possibly a missing term has not been connected (or similar)

Clare:

  • Ben & Jhan Working on how to organise CABLE source files in the best way, & to stop duplication between ESM1.5 & ESM1.6

Jhan:

  • Question about the walltime requested for ocean collation at the end of an esm1.6 run. Spencer thinks this is resolved.

  • Issue with changing projects in payu part way through an experiment is not a priority

Next meeting: 14th Jan

16

Tuesday 14th January

Attendees: @clairecarouge @inh599 @tiloz @ben @MartinDix @pearseb @anton @dave @spencerwong @RachelLaw @cbull

Apologies: everyone else.

Pearse:

  • 200 years with/without iceberg flux. Analyses to be started this week by Pearse and Mat. It seems to be relatively stable. Less stable with iceberg fluxes: 5 perturbations in atmosphere. A bit too much but all may not have been necessary.
  • continue the full ESM run for more years.
  • when is @dougiesquire back? Next week.

Rachel:

  • new vegetation distribution to be more compatible with LUH3: we’ll want a run with that vegetation distribution. Need help creating a new restart file.

Ian:

  • Need to check wood thinning in ESM1.6. File reorganisation in CABLE would be good to have asap.

Rachel:

  • water balance script. Martin will look at getting this to run more smoothly for everyone.

Dave:

  • CICE5: working on an issue reading CICE5 restarts. Hopefully might run with esm1.5 in a few weeks. Need help. @cbull / @MartinDix suggested posting an issue on GitHub, here. @anton mentioned there’s a tool for converting restarts from CICE4 to CICE5 here.

Pearse: some updates coming for WOMBAT-lite. Spencer to add to the configs.

Rachel: forcings for pre-industrial for the new data for CMIP6+. input4mips on NCI in qv56.

17

Date: January 21, 2025

Attendees: @clairecarouge, @anton, @inh599, @RachelLaw, @dhb599, @MartinDix, @ben, @dougiesquire, @harshula, @spencerwong, @tammasloughran, @tiloz, @pearseb

Apologies:

  • Jhan
  • Aidan
  • Arnold
  • Shioban
  • Tommy

Agenda:

  1. Recording and Introduction
  2. Issues and Blockers from Land Team
  3. Configuration and Testing Updates
  4. Ocean Team Updates
  5. Forcing Updates

Minutes:

  1. Recording and Introduction:

    • Anton started the recording.
    • Chris Bull welcomed everyone and explained the purpose of the recording for minute-keeping.

  2. Issues and Blockers from Land Team:

    • Rachel mentioned Jhan’s apology and his request for the latest configuration for testing.
    • Ian discussed the syncing between AM3 science, CABLE 3 main and ESM1.6 code includes some changes over the frozen soil which would improve energy and water conservation.

  3. Configuration and Testing Updates:

    • Spencer Wong updated on a bug fix related to the ocean side and discussed with Ian about the inclusion of the bug fix in the tests.
    • Pearse Buchanan provided an update on the sinking scheme issue and its resolution.
    • Rachel discussed the new vegetation distribution and the need to modify the restart file.
    • Ian and Martin Dix discussed the importance of energy and water conservation assessments and the use of Jupyter notebooks for this purpose.

  4. Ocean Team Updates:

    • Dave B. updated on the progress with the CICE 5 upgrade and the resolution of the restart file issue.
    • Chris Bull inquired if Dave needed any input from others, to which Dave responded that he was fine on his own for now.

  5. Forcing Updates:

    • Martin Dix mentioned no new updates since last week.

Action Items:

  • Spencer Wong to contact Jhan regarding the configuration for testing.
  • Rachel to work on creating a restart file for the new vegetation distribution. Martin has some scripts to update UM restart files.
  • Martin to create a python script from his Jupyter notebook for the energy and water balance. Martin to put on GitHub for sharing.
  • Dave B. to continue working on the CICE 5 upgrade and initialization issues.

Next Meeting:

  • Scheduled for January 28, 2025.
18

Date: January 28, 2025

Attendees: @clairecarouge, @anton, @inh599, @RachelLaw, @MartinDix, @dougiesquire, @harshula, @spencerwong, @pearseb, @Arnold_Sullivan, @Jhan, @cbull

Minutes/chair: @cbull

Apologies:

Agenda:

No pre-organised items.

Discussion summary:

Rachel is having an issue with a restart file in cable (change the vegetation distribution) and Lachlan

Ian: could do a short run to check for conservation (checking some changes that jhan did last week). Rachel thinks doing a check with the code in it’s present form is a good idea.

Spencer has set up a trial AMIP configuration (ESM 1.6 atmosphere only, same code but just the atmosphere) that Martin could do a conservation check on (Martin is doing a code review and could do this kind of check as part of the review). Rachel and Jhan are interested in this configuration once it’s gone through review, @MartinDix please share.

Pearse: no updates. Had a run that got stuck over the weekend but have resubmitted the job.

Dougie: Harshula and Dougie will catch up on the build for generic tracers (Dougie has been on leave). Wombat has a code review that Dougie is doing, does not expect that will lead to answer changes.

@dhb599 / Dave is not present this morning, so no update on CICE work.

The meeting was concluded early due to the absence of further updates. The next meeting is scheduled for the following week.

19

Meeting Minutes

Date: February 4, 2025
Attendees: @cbull, @clairecarouge, @micael, @paulleopardi, @MartinDix, @inh599, @spencerwong, @dougiesquire, @pearseb, @dhb599, @harshula, @Aidan, @tiloz, @tammasloughran, @Jhan, @manodeep
Minutes and chair: @cbull
Apologies: @RachelLaw, everyone else…

ACCESS-ESM1.6 project board review

The focus was on reviewing project components, identifying blockers, and discussing any issues or updates.

  • Chris Bull invited attendees to mention any blockers or issues missing from the project board.
  • Micael noted the need to add issues related to the software transformation team’s work. Claire Carouge encouraged adding more components and organizing them as needed.
  • Paul is documenting ancillary work for esm 1.6 for cmip7 and was interested in this issue. Martin can share a suite to look into it.

Land Team Updates

  • Ian: still working on resolving energy balance issues. The fix looks to have an instability in AM3 that they’d prefer not to inherit into ESM. Discussions around updating vegetation distribution.
  • Claire update: Jhan has been working on a large PR that Claire has been reviewing. Vegetation/soil parameters (all the cable namelists) are being read in by the UM rather than being hardcoded making it easier to change them at runtime. Spencer asked if the new namelists files should be on a payu update or vk83. Ian: cautioned that perhaps some parameters shouldn’t be available to be changeable to users. Claire: countered suggesting that we can make them available and it is up to users to change with care.
  • Ian: thinking about making greater use of soil ancillary information in esm 1.6. Martin: would you want to move to the new data sources? Ian: need to think carefully about the best workflow. Paul/Ian: CMIP7 land use cover forcing is indeed part of this. Tilo: what is the status of the restart file such that we can use the new vegetation maps? Ian: Lachlan has been working on this, see the forum topic on it – sounds like he’d appreciate input. Claire: Rachel has provided an update on this, this morning.

Ocean Team Updates

  • Dougie and Parse: no major changes from last week. Pearse: in year 352 in ESM1.6 has modified the deep alkalinity, dic, nitrate and oxygen back to the initialised restart state. This was done due to the bug that we found ~100 year. Model now seems to be more stable 200-250 only required one restart, whereas in the first few hundred years there were quite a few more restarts required (atmosphere was slightly perturbed).
  • Dave Bi: cice5. Still working on cice5 update for ESM1.5 system. Initialisation has gone well, reading in initial files/model set-up looks to run. Had some issues with model code, is getting a seg fault.

Release Team Updates

  • Harshula discussed improvements in SPACK develop instructions and stricter rules from upstream SPACK developers.
  • Aidan Heerdegen emphasized the importance of merging pull requests and ensuring the process works smoothly.
  • Harshula: looking to improve the spack develop process. Been getting some input from the spack developers and they’ve made some suggestions and found a bug which has led to a change (stricter rules) in the spec, this means updating spack.yaml files (example). Related changes have been propagated. ESM 1.6 is affected in minor ways, in this instance where a branch rebase or similar is required a comment should have appeared in PRs (like other infrastructure changes). The release team will try to minimise the impact of this.
  • Aidan: question to Pearse about the bug: is Pearse happy with the branch-develop workflow he’s using? Yes, those fixes are inside wombatlite-sinking branch. Need to eventually get those into main branch. Aidan: Could be a good time to fix the above Spack bug/update.

Closing Remarks

  • Chris Bull highlighted software transformation team are looking to improve performance of esm1.6 over the next few months.
  • We have a meeting next week Claire will be chairing the next meeting.
1 Like
20

Tuesday 11th February

Attendees: @paulleopardi , @clairecarouge , @inh599 , @MartinDix , @tiloz , @anton , @manodeep , @Aidan , @spencerwong , @ben , @micael , @tammasloughran , @Jhan , @TommyGatti

Apologies: everyone else

Land updates

  • Martin: build for checking uninitialised variables and out of bounds. Some problems found in CABLE that will change results. Still on going.
  • Ian: Pb of energy conservation over land in the coupled model. Did a great job improving the water balance though. Question around using ILAMB for QA: can we use the NRI ILAMB release? Need to setup a reusable workflow for MOPPeR.
  • Ian: open question about what the river inflow is in ocean. Need some help from someone who knows the ocean model. Martin: runoff output in ocean and river outflow output in atmosphere. Anton: runoff field in MOM5 called runoff. See Andrew Kiss notes, Anton to follow up.

Ocean updates

Nothing of note

Release updates

  • :warning: Tommy: prerelease current workflow not working as expected for all. Using branches for spec references isn’t working. Workaround for now is to use a commit hash as reference for spack.yaml. You now need to push to the spack.yaml with the new commit hash as well as pushing your code changes.
  • Aidan: it is a longer workflow with 2 pushes instead of one
  • Jhan: Happy with the workaround. Makes it easier to determine which commits are included in a run.

Optimisation updates

  • Manodeep got the configurations he needs. Trying to compile with OneAPI. Aidan: experience with MOM5, never a good idea to go to -O3, can be slower and incorrect. But that was a long time ago. Martin: need to use fpe-model= precise but never investigated in details. Manodeep: criteria is the same run needs to be deterministic. Micael: systematic testing this time and keep record of results for an informed decision. Aidan: load balance often as importance as any model optimisation. Manodeep: using the “sapphirerapids” queues might be a low hanging fruit to improve performance.
  • Anton: Are updates to dependencies on the card (netcdf, mpi,…)? Manodeep: with OneAPI always trying the latest versions of dependencies first. Aidan: target specific dependencies but not all at once. Paul: dependencies on hh5 or access modules that could be replaced.

Actions

  • Ian, Claire and Romain to discuss MOPPeR.
  • Anton to follow up on river inflow in MOM5 with Ian.

Closing Remarks

  • Meeting next week, Chris will be chairing.
1 Like