Porting SST and ice physics improvements from uob_fftw_sit23 branch - #306
Draft
sit23 wants to merge 9 commits into
Draft
Porting SST and ice physics improvements from uob_fftw_sit23 branch#306sit23 wants to merge 9 commits into
sit23 wants to merge 9 commits into
Conversation
(cherry picked from commit e4628a5) Ported onto current master with manual reconciliation: dropped the new binary_ice_albedo switch introduced by this commit, since master independently added the equivalent (and more general) ice_albedo_method ('step_function'/'ramp_function') option in the interim - the two are algebraically identical for the ramp case, so carrying both would just be a redundant duplicate mechanism.
…ea-ice model, the north pole is much warmer than it ought to be. I think this has many knock-on effects for the jet streams, stationary wave patterns etc. to try and patch this, I have added an option whereby a different non-amip sst can be specified over areas of sea-ice. I have been testing this using the 2m temperatures taken from a jra-55 climatology. Awaiting results now. (cherry picked from commit 8af6027) Reconciled against master's specify_constant_sst option (added independently since this branch forked): the constant-SST and file-read SST paths now both feed sst_input, which the sea-ice overlay logic reads from, rather than writing sst_new directly.
(cherry picked from commit 8a1653a) Also fixed a duplicate `specify_sst_over_sea_ice` entry in the mixed_layer_nml namelist declaration list that this commit introduced upstream and which persisted to the branch tip - dropped the duplicate rather than porting it.
(cherry picked from commit 3f695e9)
…d socrates_simcloud The physics port from uob_fftw_sit23_new_maths2intel (previous 4 commits) decoupled land_ice_mask's ice-awareness from update_albedo_from_ice into a new update_land_mask_from_ice flag, defaulting to .false. Both of these existing test cases set update_albedo_from_ice=True without knowing about the new flag, so without this change they'd silently switch land_ice_mask from ice-aware to land-only - affecting qflux zeroing and, for socrates_simcloud (do_sc_sst + specify_sst_over_ocean_only), which cells get SST-relaxed. Setting update_land_mask_from_ice=True explicitly restores their pre-port behaviour.
(cherry picked from commit 0752c6b) Ported the calculate_qflux.py / nc_file_io_xarray.py / cell_area.py changes only - these are the generic, reusable pieces (Socrates variable-name support in check_surface_flux_dims, the ignore_ice_for_calculation option, Gaussian-grid cell-area fallback, and reading a pre-joined file by name). Dropped this commit's hunk to postprocessing/plevel_interpolation/scripts/run_plevel.py: it was Stephen's own scratch run/path config for a specific dataset, and master has independently modernised run_plevel.py since this branch forked (see 4459f5c) - reconciling the branch's run_plevel.py history against that belongs with the numerics/tooling port on uob_fftw_sit23_numerics_2026, not here. This is the script that generated soc_ga3_bucket_jra_55_ice_temps.nc, the q-flux input file used by the Maunder Minimum experiments in exp/socrates_fast_land_ice_cmpi5_ozone/maunder_min/ - needed to regenerate or adapt that q-flux file (e.g. for a different ice treatment, spinup length, or the socrates_fast_land_fft_new_ice family of scripts more generally).
Documents that the ported sea-ice/land-SST physics compiles and runs the real Maunder Minimum q-flux experiment (socrates_fast_land_fft_new_ice_sbm_do_simple_false_bucket_qflux_mm.py), and compares one month of output directly against the original uob_fftw_sit23_new_maths2intel branch: fields tied to the ported physics (albedo, ice_conc, flux_oceanq) match exactly, while dynamical fields diverge with a textbook chaotic-error-growth signature rather than a step change, consistent with two non-bit-identical compiles of the same chaotic model rather than a regression.
Stephen asked whether the branch-vs-branch divergence noted in the previous validation pass was general chaos or something specific to the old branch. Ran the same comparison methodology on a plain, physics-untouched socrates_aquaplanet control case across several pairings: master vs physics_2026 (bit-identical), the 2023 merge-base vs current master (bit-identical), and the original branch vs physics_2026 (diverges, same signature as before). Bisected the divergence by reverting/restoring the ~35 modified source files one at a time / in batches, narrowing it to a single file: socrates_interface.F90, the original branch's own Socrates radiation-coupling changes - unrelated to and untouched by this port's mixed_layer.F90 work. Confirmed physics_2026's socrates_interface.F90 is byte-identical to master's, so whatever the exact remaining issue in that file is, it isn't carried into this PR.
Reframed the socrates_interface.F90 finding: it's not a single-line bug, it's ~2400 lines of independent divergence between the old branch's 2019-vintage Socrates interface and master's now much more developed version (master already reimplements the same Mars-orbital diagnostics more completely, including a fix for an uninitialized use_old_r_inv_squared the old branch's version had). With that confound identified, re-ran the actual Maunder Minimum target experiment comparing physics_2026 against a patched copy of the old branch (everything matched to current master except mixed_layer.F90, left at the branch's original un-ported physics). Result: bit-identical for the full month across every field. This is the direct confirmation that the ported physics reproduces exactly what the original branch would have produced running on current master - the physics port is complete.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
For a while I've been working on getting Isca's basic state as good as it can be. This culminated in some work on the uob_fftw_sit23 branch, and the later version with maths_2 configs. I'd like to merge this in, but it was also bound up with FFTW changes and other things that were done by people at the University of Bristol. Here I'm pull requesting just the physics changes on this branch, and I've had Claude verify that the physics changes perform as they did on the old branch. I need to do some manual testing on this, but so far it looks OK.