Tripolar north-fold examples (new fold-boundary API)#24
Open
hdrake wants to merge 18 commits into
Open
Conversation
Adds a self-contained section demonstrating how to represent the MOM6 tripolar grid's bipolar Arctic fold with xgcm's face_connections (using the new `fold` connection flag in xgcm). The grid is split into two full-height faces joined periodically in X and by the fold along their northern edge, and the section verifies the fold halo against the geometric neighbour and shows the velocity sign convention via MOM6's seam transport. Also notes that the original GFDL thredds data server is offline and points to the Zenodo archive used by the new section. Requires the bipolar-fold feature from xgcm/xgcm#707. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Replace the dead GFDL thredds download with the full-resolution MOM6 example output archived on Zenodo (record 15420739), and recast the notebook around the variables that archive actually provides: - Load the vertically-integrated mass-budget file directly from Zenodo; drop the now-redundant separate ocean_grid_sym download (geolon_c / geolat_c are already included). - Build the grid for the symmetric-memory layout (xq/yq are the `outer` position). - Recast the vorticity demo as the curl of the vertically-integrated mass transport (umo, vmo), which lands on the cell corners. - Point the bipolar-fold section at the same dataset (no second download). Requires the bipolar-fold feature from xgcm/xgcm#707. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Use boundary={"X": "periodic", "Y": {"fold": "corner"}} on a single grid
tile instead of the experimental two-face face_connections 4-tuple, and
verify the scalar fold halo, the fold-halo latitudes, and the vector
(vmo) sign flip across the seam.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
- 06_tripolar_fold.ipynb: demonstrate the tripolar fold via relative vorticity for MOM6, NEMO eORCA1 (IPSL-CM6A-LR), and Oceananigans; the naive (X-periodic) vs fold-aware vorticity differ only at the fold seam. - 07_oceananigans_tripolar.ipynb: Oceananigans TripolarGrid fold check (tracer zipper + vector sign flip), with the generating Julia script. - scripts/generate_oceananigans_tripolar.jl: reproducible dataset generator. - Readme: document the new examples and their data sources. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
|
Check out this pull request on See visual diffs & provide feedback on Jupyter Notebooks. Powered by ReviewNB |
…rticity - Remove the tripolar fold section from 03_MOM6 (now covered by 06). - Remove 07_oceananigans_tripolar.ipynb (redundant with 06's Oceananigans vorticity demo). - 06_tripolar_fold.ipynb: plot the naive/fold-aware/difference relative vorticity on a North Polar Stereographic projection, so the fold shows as a line across the Arctic between the two grid poles. - Add environment.yml (conda) for running the examples, including cartopy. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
The MOM6 sections of 03_MOM6 and 06_tripolar_fold previously used the vertically-integrated mass transports umo/vmo, which are not the right field for a velocity curl / relative vorticity. Replace them with surface ocean velocities uo/vo from CMIP6 GFDL-CM4 (MOM6 on its native tripolar grid), read anonymously from the Pangeo cloud — mirroring the IPSL-CM6A-LR (NEMO) example. - 03_MOM6: getting-started tutorial now interpolates uo/vo to cell centres for surface speed and computes the velocity curl, on GFDL-CM4 cloud data. - 06_tripolar_fold: MOM6 vorticity-fold section uses GFDL-CM4 uo/vo. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
scripts/run_one_degree_simulation.jl adapts ClimaOcean's one_degree_simulation to run locally on CPU (realistic TripolarGrid + ETOPO bathymetry, ECCO4 init, JRA55 forcing), writing surface u/v/T to JLD2 frequently so an early-stopped run is usable. scripts/jld2_to_netcdf.jl converts the latest snapshot to the NetCDF schema consumed by 06_tripolar_fold (Oceananigans section). Env pinned in scripts/climaocean_env. Output data is gitignored (to be posted to Zenodo). Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
…ample
Replace the minimal synthetic TripolarGrid dataset with a surface snapshot
from a realistic 1-degree global ClimaOcean/Oceananigans simulation
(ECCO-initialized, JRA55-forced, real ETOPO bathymetry), run ~0.75 model-days
past the barotropic adjustment via scripts/run_one_degree_simulation.jl and
converted to NetCDF with scripts/jld2_to_netcdf.jl. The Oceananigans section
of 06_tripolar_fold now shows real Arctic surface vorticity; its U-point
("u") fold pivot is an independent check vs MOM6/NEMO ("corner"). Drop the
superseded synthetic generator. Output data is gitignored (to be posted to
Zenodo).
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Integrate past the initial barotropic adjustment to a spun-up state; save one surface snapshot per model-day so day 10 is the final output. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
…Units) Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Re-run the realistic 1-degree global ClimaOcean/Oceananigans simulation to a spun-up day-10 surface state (t = 864000 s) and regenerate the Oceananigans section of 06_tripolar_fold from it. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
…scaling Add a log-scaled global surface current speed map for the Oceananigans run, interpolating u/v to tracer points (v reaches across the fold) as a clean, metric-free check that the fold gives smooth fields across the Arctic seam. Note that the relative-vorticity panels use index-space finite differences (unscaled), so near-pole magnitudes are dominated by grid convergence; the fold changes only the seam row, isolated by the naive-minus-fold panel. xgcm's reconstructed fold halos were verified to match Oceananigans' own zipper exactly (tracer + both velocity components). Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
…olar Reframe 06 around testing diff and interp across the north fold: - surface current speed (u, v interpolated to tracer points) -- tests interp; - relative vorticity zeta = diff(v,'X')/dx - diff(u,'Y')/dy, now divided by the cell spacings (great-circle distances), so it is the physical vorticity and near-pole magnitudes are no longer inflated by grid convergence -- tests diff. Both are plotted on North Polar Stereographic projections for MOM6 (GFDL-CM4), NEMO (IPSL-CM6A-LR), and Oceananigans, and are smooth across the fold; the naive (no-fold) vorticity differs only on the single seam row. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
…Ro=zeta/f Restructure each diagnostic as a 3x3 grid (columns = models; rows = naive no-fold / fold-aware / difference): - surface speed from interp, shown interpolated to BOTH tracer centres (v->centre crosses the fold) and cell corners (u->corner crosses it), so both velocity components are exercised across the seam; - Rossby number Ro = zeta/f, with zeta = diff(v,'X')/dx - diff(u,'Y')/dy divided by the cell spacings and the Coriolis parameter f = 2 Omega sin(phi). Mask land (missing values, and zeros where they dominate, e.g. the Oceananigans immersed boundary). Mask fold/dateline wrap cells and scale the difference panels by their non-zero magnitude so the (seam-only) difference renders correctly. For all three models both operations are smooth across the fold and differ from the naive calculation only on the single fold row. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
- Relative-vorticity (Rossby) panels use a fixed colour range of +/- 0.05. - All maps zoom to latitudes 60-90. - Land (masked with NaN) is filled grey in every panel. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
…oise The low-resolution models have very small Arctic Rossby numbers, so use a +/-0.01 colour range. Add a note explaining that the band of vorticity at the fold in the Oceananigans panel lives in the model's velocity field (grid-scale north-fold noise), not in xgcm's fold operators -- it is identical in the naive and fold-aware columns. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
The single fold row was hard to read and sat next to a grey "no data" line (the old _wrap_mask dropped the seam, and shading="nearest" + cartopy's PlateCarree wrap smeared/blanked it). Re-plot the fold so its correctness is obvious: - Seamless polar cap: reconstruct the cell-corner lattice *through* the fold (xgcm fold-pad of the centre coords, averaged in 3-D so pole/dateline are clean) and pcolormesh(shading="flat") in self-projected stereographic coords. The Arctic is now one continuous mesh with no masked seam and no antimeridian smear; the difference row, normalised per panel, lights up the exact cells the fold changes -- a thin line across the pole. - Cross-fold transect: continue the surface-speed field across the seam into the halo. The fold fills it with the true partner row (a smooth physical continuation); the naive boundary just flatlines. Shown for all three pivots (MOM6/NEMO "corner", Oceananigans "u"). Builder saved as scripts/build_06.py; the previous grey-seam builder kept as scripts/build_06_maps_v1.py for easy revert. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
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.
Example notebooks for xgcm's new single-tile north-fold boundary
(
boundary={"X": "periodic", "Y": {"fold": <pivot>}}), which replaces theexperimental two-face
face_connections4-tuple. Pairs with xgcm/xgcm#711.06_tripolar_fold.ipynb— demonstrates the fold via relative vorticityfor three models, each with real surface velocities
uo/vo, on aNorth Polar Stereographic projection where the naive-minus-fold difference
isolates the fold as a line across the Arctic:
"corner"pivot."corner"pivot.JRA55-forced),
"u"pivot — an independent third convention.03_MOM6.ipynb— getting-started tutorial rewritten to use GFDL-CM4surface velocities (interp to centre for surface speed, velocity curl).
scripts/—run_one_degree_simulation.jl(ClimaOcean 1° on CPU) +jld2_to_netcdf.jlto produce the Oceananigans dataset (to be posted toZenodo);
environment.yml(conda, incl. cartopy/zarr/gcsfs).Builds on the MOM6 tripolar work in #23.
🤖 Generated with Claude Code