COSIMA Meeting minutes 2023

Here are some brief minutes from today’s COSIMA meeting. If I’ve missed anything you feel is important please feel free to add it below or edit this post (it’s a wiki post).

DATE: 09/02/2023
Attendees: 20-25

Small items:

  • ACCESS-NRI Compute Hours: Adele/Paul attended the ACCESS-NRI Science Advisory Committee and mentioned that the NRI has ~30MSU for distribution to the working groups (including the ocean-sea-ice working group - i.e. COSIMA) for science projects. If you’re interested please talk to Adele (or Andrew/Ryan who are the other ocean-sea-ice working group co-chairs). Preference would be for runs that are beneficial to the community, but other projects would be considered as well.

  • SWOT Linkage Project Awarded: Shane Keating et al. have been awarded a linkage project to work on SWOT and use it for improving surface current forecasting around Australia. Some possible sea ice work.

  • CMIP7 Scoping Workhop: Andy Hogg brought up the CMIP7 scoping workshop that will be held on the 28th Feb - 1st of March in Aspendale (and online). This will pull together the community to make decisions about a possible Australian CMIP7 contribution (who should run it, where we get storage, etc.).

  • Meeting minutes: Discussion of maintaining minutes for the COSIMA meetings, as is done for the COSIMA TWG meetings (this post). And speakers uploading their talks.

Talk: Alex Fraser from U Tas on “The Antarctic marginal ice zone: Satellite observations of MIZ width and wave attenuation” - lots of questions/discussion, including the possibility that ACCESS-OM3 (which will have coupled ocean-waves-sea ice) will be able to resolve an MIZ of some description.

1 Like

Brief minutes from the COSIMA meeting. If I’ve missed anything you feel is important please feel free to add it below or edit this post (it’s a wiki post).

DATE: 02/02/2023
Attendees: ~20

Talk: Felipe Da Silva from U Tas on “Meanders in the ACC shape the air-sea heat fluxes and water subduction in ACCESS-OM2”.

COSIMA Hackathon Follow-up: Discussion of the successes and failures of last week’s COSIMA hackathon event. Navid took us through how to submit and review PR’s in cosima-recipes. See COSIMA Hackathon v2.0 -- Tuesday January 24th, 2023 for more details, in particular COSIMA Hackathon v2.0 -- Tuesday January 24th, 2023 - #44 by navidcy

1 Like

Here are some brief minutes from today’s COSIMA meeting. If I’ve missed anything you feel is important please feel free to add it below or edit this post (it’s a wiki post).

DATE: 16/02/2023
Attendees: ~15

Talk: Zhaohui Wang from U Tas on “Intercomparison of Ice surface Temperature from Recent Global Atmospheric Reanalyses over Antarctic Sea Ice”, including a JRA-55, ERA-5 comparison.
Brief notes:

  • All reanalyses have very large warm biases in ice surface temperature over the Antarctic region. This is worse in ERA-5 than JRA-55. E.g. the area-averaged bias of JRA-55 is +1.39 K but the bias for ERA-5 is +6.24 K!
  • JRA55 performs well when cloud cover is low.
  • Biases linked to ERA5’s poor representation of clouds through excessive longwave radiation, as well as the representation of sea ice in the reanalyses (binary vs. fractional and ice thickness).
  • Andrew’s comment: ERA5 surface air temperature is also significantly warmer than JRA55, but JRA55-do corrections to JRA55 are excessive, resulting in SAT even warmer than ERA-5 in many places. See here, with more detail here.

ERA-5 in ACCESS-OM2 discussion:
• Ryan gave a summary of progress on an ERA-5 forced version of ACCESS-OM2. Slides can be found here.
• “naïve” ERA-5 forcing implemented with no adjustments, NCAR bulk formula, other simplifications.
• RYF simulations working and bulk circulation metrics and SST/SSS biases suggest they are looking good.
• IAF simulations running into technical issues that we hope to solve soon.
ERA-5do version is in the works from NCAR. Unclear what the time horizon is for this. COSIMA should engage with this as we have lots to contribute (Zhaohui’s work, ACCESS-OM2 as a test-bed).
• Will Hobbs brought up the point that being able to run the model at near real-time is really important for the Antarctic sea ice community. This will disappear sometime this year for JRA-55. So getting ERA-5 onboard is important, and there may be a big gap.
• Aidan pointed out that ACCESS-OM2 might make a great test-bed for new ERA-5 based products as it is well characterised and understood under JRA-55do forcing.

Some action items:

  • Make sure to engage with the NCAR/GFDL/ECMWF people who will develop ERA-5-do forcing.
  • We need more people looking at sea ice in the ERA-5 runs we have at the moment.

Discussion of ACCESS-NRI support for ACCESS-OM2 control runs pushed back to next week.

1 Like

Here are some brief minutes from today’s COSIMA meeting. If I’ve missed anything you feel is important please feel free to add it below or edit this post (it’s a wiki post).

DATE: 23/02/2023
Attendees: ~35

Small items:
- NCI has a new upgrade in the works (Intel Sapphire Rapids) with many new CPUs available (as large as the current Cascade Lake nodes). 52 cores per node. We will be running a high-resolution 1/40-degree pan-Antarctic test on these runs that Adele and others are developing.

Tyler/Laurie/Pat validation of ACCESS-OM2-01-BGC: See this post for slides/follow-up

  • BGC fields in models can often be outside the error bars of the observations, and these error bars are already very large!
  • Negative bias in net primary productivity. One of lowest models. Also lower than ACCESS-ESM.
  • Larger seasonal cycle in NPP than observations - larger maximum bloom. However, the timing is good.
  • O2 → looks good! But initialized from 1979. Andy question: Do we see any trends? Yes but hard to say if it’s physical/expected or not.
  • DIC → mean bias but good gradients. Alkalinity similar as well.
  • Surface CO2 fluxes → overestimated, also issue with seasonal cycle. Spatial structure is good.
  • Mixed layer depth: Spatial structure is good. Some biases but no larger than differences between observational products.
  • Iron: Lots of drift in the positive direction. Questions: Is this the same in ACCESS-OM2-1? ACCESS-ESM? Would be good to compare across all of these.
  • There’s a need to continue the discussion here - possibly through another BGC-focused series of meetings see here

Levente Bodrossy (CSIRO Hobart) talk: See slides here
Extensive genomics dataset around Australia/Southern Ocean. Information on community composition. Interested in collaborating with BGC/ecosystem modellers who can potentially use this dataset as validation. Maybe not for WOMBAT (only one phytoplankton class). However, Tyler suggested it could be used to check/validate the growth rate parameters used in WOMBAT.

1 Like

Here are some brief minutes from today’s COSIMA meeting. If I’ve missed anything you feel is important please feel free to add it below or edit this post (it’s a wiki post).

DATE: 02/03/2023
Attendees: ~30

Update from CMIP7 scoping workshop from Adele:

  • The community wants to put in a CMIP7 submission, because:
    • Critical for credibility of our model
    • Catalyses model development
    • Otherwise might have to co-op another model
  • Timeline: IPCC data deadline - mid 2025-end of 2026. Simulations start mid-2024-end 2025.
  • Model configuration:
    • Ambitious option: ACCESS-CM3 → MOM6, CICE6, UM updates/latest. + ESM3 carbon cycle version.
    • Some doubt within the carbon-cycle community if there is enough time to spin-up given tight timeline.
    • Possible backup: Model based on ACCESS-CM2.
  • No definitive decision. Small working group put together to make a decision/write something up.
  • What this means for COSIMA:
    • Starting simulations mid next-year means there would be a need for COSIMA input to develop MOM6-CICE6 global. We might need to refocus on the global configuration rather than the regional.
    • Benefits to COSIMA from buying into this development of ACCESS-CM2 (from NRI).
  • Andy: Need to be ambitious to convince government to give money to support this.

Wilma Huneke Talk: Australia’s ACCESS-CM2 climate model with a higher-resolution ocean-sea ice component (1/4°)

  • Overview of ACCESS-CM2 with 1/4-degree ocean - comparison to observations and ACCESS-OM2.
  • Costs: 12.5kSU/model year, 4.3yr/day on 1152 cores. (Comment from Adele: This is still really fast despite 1/4-degree ocean. Possible we could go to 1/10-degree. However - benefit from increasing atmospheric resolution might be larger.)
  • Clear improvement in oceanic variability metrics. However, no clear improvement (and often things get worse) in mean state biases.
  • Much of SST bias thought to come from atmosphere.
  • Large multi-decadal variability in the North Atlantic in CM2-025 that is not present in CM2-1. Overtakes ENSO as the largest mode of variability.
  • Andrew K question: How are the fluxes exchanged? Low-res atmosphere means blocky fluxes. There is another approach called “exchange grids” that could help with this.

Aidan on NRI support for ACCESS-OM2 Control runs: See ACCESS-OM2 Control Runs
What should the NRI support in terms of ACCESS-OM2 control runs?

What does support mean:

  • Take over control of data (frees up space for COSIMA).
  • Clean-up of data and documentation/reproducibility (e.g. restarts, configurations).
  • Classes of support:
    • Legacy: If issues are found then flag them. If the issue is big then retract the data.
    • Full support: Rerun the simulations if issues are found during period of full support (TBD)

From the COSIMA side:

  • Would be great to have “legacy”-style support for the RYF/IAF 1/10th degree simulations (these are the biggest in terms of data). Start with those, then move onto 1/4 and 1-degree RYF/IAF.
  • One potential issue is that in the 1/10th many diagnostics are only available for portions of the run. Harder to support.

@wghuneke Feel free to post your diagnostic plots in this thread, or probably better a new one and link back here, to get some detailed feedback.

Here are some brief minutes from today’s COSIMA meeting. If I’ve missed anything you feel is important please feel free to add it below or edit this post (it’s a wiki post).

DATE: 02/03/2023
Attendees: ~20

Small items:

  • Ryan will step down from running the weekly meetings and from the NRI working group co-chair from the end of March. Paul Spence and Ed Doddridge will step in as additional co-chairs (with Adele and Andrew), and take over the weekly meetings.
  • COSIMA will start to focus on the global OM3 configuration to align with CMIP7 plans. Dougie has put together a working initial version of CESM with the NuOPC coupler on Gadi. See GitHub - COSIMA/access-om3: ACCESS-OM3 global ocean - sea ice - wave coupled model if you’re interested in contributing.

Talk: Sebastian McKenna (UNSW): How Pacific Ocean SST biases impact the Indian Ocean using pacemaker experiments in ACCESS-CM2

  • Planned pacemaker experiments that remove ENSO, or remove the mean SST bias, in the Pacific to see impacts on Indian ocean mean and variability.
  • Initial experiments where SST in the Pacific is replaced by model climatology (removing ENSO) show minimal impacts on mean SST elsewhere, except in the North Atlantic which is a sensative area (could be just model internal variability).
  • Discuss of whether this could be linked to weak ENSO diversity/asymmetric in ACCESS-CM2.
1 Like