Opened 3 years ago
Last modified 3 years ago
#2722 new Task
Branch to import 3rd party GEOMETRIC development
Reported by: | acc | Owned by: | acc |
---|---|---|---|
Priority: | low | Milestone: | Unscheduled |
Component: | LDF | Version: | v4.0.* |
Severity: | minor | Keywords: | |
Cc: |
Description
Plan: Create a branch of 4.0-HEAD (initially equivalent to r4.0.6) to incorporate Julian Mak's GEOMETRIC code. GEOMETRIC is an approach to representing the unresolved turbulent eddies in ocean climate models. and is well introduced here:
https://nemo-related.readthedocs.io/en/latest/GEOMETRIC/geometric.html
Once fully tested, this code will be ported to in a 2021 or 2022 development branch for a future trunk release.
Commit History (3)
Changeset | Author | Time | ChangeLog |
---|---|---|---|
15262 | acc | 2021-09-15T16:41:33+02:00 | Fix restartability issue for ORCA2_ICE_PISCES with GEOMETRIC parameterisation and ln_adv_wave=.true.. Also remove redundant restart variable (zc1). SETTE restartability is now ok for an ORCA2_ICE_PISCES-based test. Still an issue with reproducibility. #2722 |
15260 | acc | 2021-09-14T18:31:24+02:00 | Untested, initial port of GEOMETRIC changes. Includes a bug fix to eken diagnostic in diawri.F90 which, if confirmed, is also relevant to current 4.0-HEAD. #2722 |
15257 | acc | 2021-09-14T15:57:41+02:00 | Create branch for 3rd part GEOMETRIC development. #2722 |
Change History (4)
comment:1 Changed 3 years ago by acc
comment:2 Changed 3 years ago by acc
In 15260:
comment:3 Changed 3 years ago by acc
In 15262:
comment:4 Changed 3 years ago by acc
Ok, it is reproducible but not if you foolishly try to run ORCA2 with ln_beta_plane = .true. . In general though, this is never going to be reproducible (ldfeke.F90):
! if using beta_plane, compute beta and f0 locally IF ( ln_beta_plane ) THEN zf0 = ff_f(1,1) zbeta = (ff_f(1,2) - zf0) / e2t(1,1) IF(lwp) WRITE(numout,*) ' beta plane option is ln_beta_plane =', ln_beta_plane IF(lwp) WRITE(numout,*) ' f0 = ', zf0, ' s-1' IF(lwp) WRITE(numout,*) ' beta = ', zbeta, 'm-1 s-1' END IF
since zf0 and zbeta could be different for every domain (or can we rely on the result being numerically the same if ff_f has been defined correctly on a beta plane?). Otherwise, may need to set values here via the namelist (ref_lat, ref_spacing?)
In 15257: