Various improvements to transect capability

[James Warner (jwarner8)]

Broad collection of improvements to the transect capability in CSET, addressing the following issues/enhancement requests for pressure level transects;

• Producing an infographic/inset showing location of transect [Plot infographic plots for transect and verification plots #830]
• Callback to fix varname from ML models that prevents colorbar from being fixed.
• Support aggregation options for transect.

Co-pilot was used to generate some of the code ideas used for the transect inset, which was further reviewed and adapted by JamesW.

Contribution checklist

Aim to have all relevant checks ticked off before merging. See the developer's guide for more detail.

• Documentation has been updated to reflect change.
• New code has tests, and affected old tests have been updated.
• All tests and CI checks pass.
• Ensured the pull request title is descriptive.
• Ensure rose-suite.conf.example has been updated if new diagnostic added.
• Conda lock files have been updated if dependencies have changed.
• Attributed any Generative AI, such as GitHub Copilot, used in this PR.
• Marked the PR as ready to review.

[github-actions]

[James Warner (jwarner8)]

Status 26/03
Modifications to transects have now been made

Started plumbing the aggregation side of things, need to debug why all models are being loaded for one model aggregation. It seems to load model 1 and 2, i.e. the parbaked recipe looks like

file_paths:

• ~/cylc-run/cset_workflow/run11/share/cycle/*/data/1
• ~/cylc-run/cset_workflow/run11/share/cycle/*/data/2

this leads to

Constraint doesn't produce single cube: ConstraintCombination(ConstraintCombination(ConstraintCombination(Constraint(), Constraint(cube_func=<function generate_cell_methods_constraint.<locals>.check_no_aggregation at 0x148eceefaae0>), <built-in function and_>), Constraint(name='x_wind'), <built-in function and_>), Constraint(coord_values={'pressure': <function generate_level_constraint.<locals>.<lambda> at 0x148eceefac40>}), <built-in function and_>)
0: x_wind / (m s-1) (time: 16; pressure: 28; latitude: 186; longitude: 142)
1: x_wind / (m s-1) (time: 16; pressure: 28; latitude: 186; longitude: 142)
2: x_wind / (m s-1) (time: 8; pressure: 13; latitude: 281; longitude: 321)
3: x_wind / (m s-1) (time: 8; pressure: 13; latitude: 281; longitude: 321)

A separate problem here is that cubes don't merge in a given model, but primary issue here is two models being loaded. Need to determine if other aggregation recipes do the same.

[James Warner (jwarner8)]

Progress:

• Looking at comparison between transect hour of day agg vs generic spatial agg, both have the same input path. Now examining differences in the loader.
• Changing to loop over model in itertools seems to work, more generic issue of cubes not merging (also seen in pressure level aggregation) now being looked at.
• Change operator from read_cube to read_cubes allows it to now work.

[James Warner (jwarner8)]

Testing for other directions/transects looks OK

[James Warner (jwarner8)]

Now ready for review, note their are no new tests, as its not clear how we would test a transect inset (other than trying it for different use cases and visually inspecting the output).

[James Frost (jfrost-mo)]

Not run myself, but the code change looks good. A couple minor suggestions but it is up to you if you take them.

[James Warner (jwarner8)]

Not run myself, but the code change looks good. A couple minor suggestions but it is up to you if you take them.

Thanks for the review, these suggestions sound sensible and have now implemented, will go ahead and merge.