ESM working group: Meeting notes 2024

Earth System Working Group
, ,
24

Date: 23/5/2024

Participants: 14

Chair: @dkhutch

Agenda:

1. Admin

  • Speakers: Christine Chung from the Bureau has volunteered for the next meeting (June 6th). There are no speakers after that. Please do volunteer to present your science. Message co-chairs or @Aidan to discuss.
  • Remember to add any items you want to discuss in these meetings to ideas for topics
  • Save the date: ACCESS Community Workshop 2-5 September. The planning committee is seeking ideas for (a) invited speakers and (b) breakout topics. Need ESM ideas for both of those things.

@dkhutch on the planning committee. Please send @dkhutch or working group co-chairs for ideas or things you think worked well last year.

  • We need to form an organising committee for the ESM Working Group’s own workshop. Last year we had a half-day, with Wilma as lead organiser, but she is on parental leave this year.

Need ideas for topics! Please get in contact.

Plan to do a half-day ESM Workshop as same as last year.

@eunpalim Any theme? @dkhutch talking it through. One idea is CMIP7, but want to make as inclusive as possible, otherwise could exclude Paleo climate.

Will be $100 fee. Mostly about commitment and some cost recovery. Very small fraction of real cost.

@Aidan Could ACCESS-NRI cover costs for students? @dkhutch Good idea. Let them know if there is no funding.

@eunpalim After last workshop there was feedback sent to @wghuneke. Is that feedback report available? @dkhutch I’m sure we can ask Wilma for that.

2. Shared experiments/ACCESS-NRI resources

  • ESM project lg87 has used 1.74MSU out of 1.9MSU allocation. We have asked, and received, two additional allocations of compute quota (total 1.1MSU extra). Unlikely to be more available this quarter
  • 26TB our of 100TB used of quota on /g/data/
  • Any proposals for shared experiments for next quarter? See guidelines for how to do this.

3. Science talk

Dietmar Dommenget: “ENSO and tropical basin interactions in idealized worlds”

  • Authors @Dietmar_Dommenget and @dkhutch
  • Motivation: model development and ENSO dynamics
  • ENSO Experiments: GFDL CM2.1 and ACCESS-ESM1.5
  • Submitted for publication
  1. Deep paleo study (Eocene 55MYA). Rearrange land and oceans
  • Much harder than you think. Not able to do this with ACCESS-N48. Still not working after 2 years.
  • @dkhutch got ACCESS-ESM1.5 working within 1 year
  • GFDL CM2.1 took ~1 month
  • MPI(1996) took 2 weeks to change bathymetry
  1. A fast CGCM
  • ACCESS-N48 coupled GCM
  • Model biases are on par with early CM2 version
  • Want fast model (100ys/day). Want to explore climate dynamics with many 100+ years simulations.
  • Want fast run, analysis cycle
  1. ENSO dynamics
  • Why is ENSO only in the Pacific?
  • Do the other ocean basis interact with ENSO?
  • Still don’t understand all dynamics of ENSO
  • Models have different behaviour
  • Belief that other ocean basis influence the Pacific and vice-versa, but some controversy
  • Theory: larger basin = larger SST variability, larger basin = longer periods, rest of the world is irrelevant
  • Reality looks different to theory
  1. GFDL CM2.1
  • Atmos: 3° x 3.75°, Ocean 1.0° x 1.5°
  • Single tropical basin, two tropical, and three tropical basin experiments.
  • Plan to simulate eocene, but noticed it was interesting for El Niño dynamics
  • GFDL model: 15 runs was ~1MCPU and 5 days
  • ACCESS-ESM1.5: 5MCPU and 20 days
  • Would take a year to reproduce CM2.1 runs with ACCESS-ESM1.5. Would not be feasible with CMIP like models such as ACCESS-ESM1.5
  • As size increases variability increases, peaks at Pacific size, drops and then increases again at full basin width. Unexpected, and not consistent with ENSO theory. Not expected from ocean perspective, but probably related to atmospheric dynamics. Currently being studied.
  • Power spectrum: peak moves to higher frequencies the bigger the basin. This is the exact opposite of theory. Frequency lowest at Pacific size, but increase with basin size with bigger sizes.
  • Twin ocean basin experiments: two twins hugely increases variability, if one is shallow it is damped.
  • Why is there such a strong change in ENSO if there is a second deep ocean basin?
  • Single basis land is positively correlated. With shallow twin there is no correlation. Two deep basins are very strongly anti-correlated.
  • Triple basin experiments: one larger basis suppresses variability in smaller basis. All identical size variability is similar in each.
  • ENSO does not get stronger with basin size
  • ENSO depends strongly on interaction with other basins
  • ENSO theory cannot explain this

Hypothesis tropical atmospheric dynamics control ENSO

  • Larger basins lead to atmospheric rearrangement and interfere
  • Dynamical ocean basins heat fluxes can provide heat sources/sinks that land cannot
  1. ACCESS-ESM1.5
  • @dkhutch made scripts to mimic these experiments, except for one small connection.
  • Did not repeat all experiments.
  • Generally weaker. Wider basin is much weaker than GFDL2.1. Disagreement with overall variability, especially with largest basin.
  • Peak frequency looks similar, fewer data points, not as strong, but similar patterns.
  • More variability in twins with shallow sea
  • Trio experiments look similar, and larger central also damps small neighbours
  • ACCESS generally has less variability than GDL2.1
  1. Summary

Screenshot 2024-05-23 at 1.36.38 pm
Screenshot 2024-05-23 at 1.36.38 pm1928×1210 237 KB

Could use more resources to complete the experiments.

4. What’s going on?

@Aidan ACCESS-NRI release ACCESS-ESM1.5 progressing. ACCESS-ESM1.5 is building and deploying with ACCESS-NRI systems. Now working on configs: reorganising and adding testing and other quality control measures. Current estimate is a release at the end of June. Can then work on releasing other configurations, and will consult with this working group who will have to decide what configurations should be released and supported, and in what order.

@Dietmar_Dommenget runs crash all the time. Impinges 10 years/day rather than 90 years/day. High priority is a fast and simple model. Make it easier to change configurations easily.

All CMIP6 models are all too slow and complicated.

@gpontes are model crashes are ocean or atmosphere related? @Dietmar_Dommenget not reproducible, something with the operating system or opening files?

@gpontes have been having numeric and technical issues.

@aekiss can automatically sweep and re-run if the crashes are known and can be ignored.

This is a wiki post , if you want to update any of the details in these meeting notes, or to add your own recollection of what was discussed then edit me rather than replying.