Meeting minutes (17/06/2025)
Chair: @clairecarouge
Attendees: @anton @clairecarouge @paulleopardi @RachelLaw @inh599 @Jhan @dhb599 @dougiesquire @manodeep @micael @rbeucher @spencerwong @tiloz @alexnorton
Minutes: @cbull
Perspectives from UKMO meeting (@paulleopardi @inh599)
- GC5-central looks like it’s mature now. ECS is approximately 3.6K.
- Met Office will look to use UKESM1.3 for CMIP7 FASTTRACK. This is using GC3.1, which has an ECS of 5.6K and is going at ~4 years/day. (They use a relatively up to date UM, ~70 vertical levels with a lot more atmospheric chemistry.)
- Met Office/UK universities will look to use UKESM2 (GC5-central) for CMIP7 Community MIPs.
- Our capability of doing 20 years/day is something we can celebrate/consider an advantage. (Context CM2 is ~10 years/day)
- [Another point @inh599 remembered after the meeting that is worth noting] The UK community realised (then identified) further issues with UKESM1.x when running historical and historical-amip tests (albeit with CMIP6 forcing). We maybe wise to set up/run such configuration/tests soon if only to provide a benchmark for later tests.
Spin-up runs update
@RachelLaw and @Jhan: we have a new June spin up run going, currently at 92 years. New format for phrenology file. Have decided to use CMT=1. The table has been updated. The relevant config is here and build here.
The question then, what is next? CICE5 or Australian PFTs perhaps. [@RachelLaw I think I missed a few other options here? ]
Coordinating development threads for next phase of control experiment
@cbull requested that folks keep the coordination table up to date. @RachelLaw this has been happening, see text Added to branch. @spencerwong noted a few things have been “merged” but that isn’t reflected, will up date.
Australian PFTs
GPP collapse has been fixed, due to using the wrong pft_params input file.
Updates simulation with PFTs 12 and 13 is still not reproducing the old simulation with same PFTs in 2 and 4. Traced back to the albedo in first time step: this is done with MOSES parameters and not CABLE. The differences then propagate.
@Jhan to change the parameters for MOSES and test.
@RachelLaw it raises the question whether there might be other MOSES parameters we are using and we are forgetting about.
@inh599 will look at the long term climate of the longer run to assess if it’s worth digging further.
CICE5 update
@anton: not a lot of activity recently. Spencer has run a CICE5 extension of the run that @anton was doing. Path to output will be shared/posted on GH issue such that @dhb599 and @sofarrell can have a look.
@RachelLaw query about odd salinity. @anton: odd salinity values were due to a Fortran bug, now fixed.
OneAPI status
@manodeep. PIconfig does now work with oneapi executables, the divide by zero workaround has slowed things down. Been using 2025.04.4 [correct number @manodeep?] but the new version doesn’t fix it either. Longer term, looking into lodging a bug report. Looking to coordinate with Anton on a longer CICE test.
Mediterranean salinity issue
@RachelLaw, any updates? @ezhilsabareesh8 is still looking into this, preliminary results are promising but @ezhilsabareesh8 is working on some more east vs west metrics.
Data move
@inh599: @pearseb has moved some data off scratch ( /scratch/p66/pjb581/access-esm/archive/pi_concentrations-expt-c55f7217/).
@manodeep: suggest creating a symlink to not break people’s workflows
Concern that early parts of the spin up are getting close to their expiration date. @clairecarouge would anyone like to follow this up in more detail? @tiloz offered to follow up.