Date: March 12, 2025
Attendees:
@anton
@dougiesquire
@minghangli
@ezhilsabareesh8
@MartinDix
@aekiss
@AndyHoggANU
@helen
@cbull
Meeting Schedule:
Most attendees will be at Bluelink meeting next week.
Apologies - we forgot to send an invite this week.
OM3-025 Project Board Priorities:
Discussion on the priorities for the OM3-025 release.
Addition of a Docs issue to the project board.
We are close to doing an alpha release and there was some discussion about what (if any) changes are needed before this point. Anton is advocating github should reflect what we have been testing with - initial conditions / grids / bathymetry / run-off fix. We will also allow extra truncations with this.
We talked about the changing MOM parameters - eg dt therm , KPP etc but decided only change dt_therm (MOM tracer timestep) in dev-025deg_jra55do_ryf config and leave KPP as is. Previous similar configs have been run for 30 yeas with current KPP and dt_therm.
- Users - Andy would advocate a group of testers for the release. Chris suggested maybe there will be an OM3 group at how to run a model training sessions next week. Minghang has a list of sensitivity testing that maybe users could engage with.
- Need to stress to uses what alpha release means re support and not-for-science use
- @Cbull will make docs for access-hive for OM3
OneAPI Compiler:
Anton is close to having a build ready with the OneAPI compiler.
Discussion on the need for performance checks and the potential impact on stability. But ultimately we were happy to proceed to minimum sanity testing only as we don’t have a current baseline set of answers / climate to compare to and at this point in development changing answers is ok.
Mom Symmetric Issue:
Turning on Mom Symmetric (in 2025.01.0) breaks restart reproducibility in the main configs and breaks determinism in BGC configs. We have reverted this change for now, but it is needed for regional work. We will provided this as a pre-release in the shorter term, with a plan to investigate and fix the issue, with Dougie Squire leading the investigation.
Upstream Model Component Updates:
We reviewed changes to upstream model components since 2025.01.0 release, and these appear fairly small. We have recently built the model without trouble with latest CMEPS/CDEPS/share. We can just update this now, but probably wouldn’t do a release.
Blue Link Meeting Preparation:
Andrew Kiss and Helen will present at the Blue Link meeting.
Discussion on the content of the presentation, including grid and topography updates, dynamic tracer time step, and potential collaboration with Blue Link.
Action Items:
@dougiesquire: Investigate the Mom Symmetric issue.
@cbull: access-hive style docs for ACCESS-OM3 (025 ryf)