feat(esmvaltool): add CMIP7 support to all ESMValTool diagnostics by lewisjared · Pull Request #519 · Climate-REF/climate-ref

lewisjared · 2026-02-09T04:05:51Z

Summary

Add CMIP7 as an alternative data source for all 16 ESMValTool diagnostics in data_requirements
Add CMIP7 to recipe infrastructure for CMIP7 replacing the deprecated drs attribute
Add get_cmip_source_type() helper so update_recipe methods dynamically select CMIP6 or CMIP7 input files

Per-diagnostic pattern

Each CMIP7 DataRequirement mirrors its CMIP6 counterpart with:

frequency filter instead of table_id
variant_label instead of member_id in group_by
SourceDatasetType.CMIP7 for AddSupplementaryDataset

codecov · 2026-02-09T04:17:17Z

Codecov Report

❌ Patch coverage is 71.21212% with 76 lines in your changes missing coverage. Please review.

Files with missing lines	Patch %	Lines
...ef-esmvaltool/src/climate_ref_esmvaltool/recipe.py	50.00%	16 Missing and 6 partials ⚠️
...esmvaltool/diagnostics/climate_drivers_for_fire.py	48.00%	12 Missing and 1 partial ⚠️
..._ref_esmvaltool/diagnostics/sea_ice_sensitivity.py	50.00%	10 Missing and 1 partial ⚠️
...tool/src/climate_ref_esmvaltool/diagnostics/ecs.py	50.00%	8 Missing and 1 partial ⚠️
...e_ref_esmvaltool/diagnostics/cloud_scatterplots.py	74.07%	6 Missing and 1 partial ⚠️
...limate-ref-core/src/climate_ref_core/esgf/cmip7.py	85.71%	3 Missing and 1 partial ⚠️
...ool/src/climate_ref_esmvaltool/diagnostics/base.py	77.77%	2 Missing and 2 partials ⚠️
...climate-ref-core/src/climate_ref_core/providers.py	60.00%	1 Missing and 1 partial ⚠️
...ol/diagnostics/climate_at_global_warming_levels.py	80.00%	1 Missing and 1 partial ⚠️
...ages/climate-ref/src/climate_ref/models/dataset.py	77.77%	2 Missing ⚠️

Files with missing lines	Coverage Δ
...te-ref-core/src/climate_ref_core/cmip6_to_cmip7.py	`92.85% <100.00%> (ø)`
...imate-ref-core/src/climate_ref_core/constraints.py	`96.61% <ø> (ø)`
...-esmvaltool/src/climate_ref_esmvaltool/__init__.py	`100.00% <100.00%> (ø)`
..._esmvaltool/diagnostics/cloud_radiative_effects.py	`100.00% <100.00%> (ø)`
...ool/src/climate_ref_esmvaltool/diagnostics/enso.py	`100.00% <100.00%> (ø)`
.../src/climate_ref_esmvaltool/diagnostics/example.py	`100.00% <100.00%> (ø)`
...valtool/diagnostics/regional_historical_changes.py	`95.53% <100.00%> (ø)`
...e_ref_esmvaltool/diagnostics/sea_ice_area_basic.py	`100.00% <100.00%> (ø)`
...tool/src/climate_ref_esmvaltool/diagnostics/tcr.py	`100.00% <100.00%> (ø)`
...ool/src/climate_ref_esmvaltool/diagnostics/tcre.py	`100.00% <100.00%> (ø)`
... and 13 more

🚀 New features to boost your workflow:

❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

Copilot

Pull request overview

This PR extends the climate-ref-esmvaltool provider to support running all ESMValTool diagnostics against CMIP7 datasets (alongside existing CMIP6/obs4MIPs support), mainly by adding CMIP7 facet mappings, CMIP7-aware recipe generation, and OR-ed data_requirements.

Changes:

Add CMIP7 facet mapping + ESMValTool config entries (drs/rootpath) and CMIP7 path preparation.
Update diagnostics to accept CMIP6 or CMIP7 via OR-ed data_requirements, and select the active CMIP source dynamically in update_recipe.
Add/adjust CMIP7 test cases for regional diagnostics and update the base config test expectation.

Reviewed changes

Copilot reviewed 20 out of 20 changed files in this pull request and generated 3 comments.

Show a summary per file

File	Description
packages/climate-ref-esmvaltool/tests/unit/diagnostics/test_base.py	Update expected `rootpath` entries after adding CMIP7.
packages/climate-ref-esmvaltool/src/climate_ref_esmvaltool/recipe.py	Add CMIP7 facet mapping; tweak `prepare_climate_data` handling.
packages/climate-ref-esmvaltool/src/climate_ref_esmvaltool/diagnostics/base.py	Add `get_cmip_source_type()` helper; add CMIP7 to ESMValTool config and selector extraction.
packages/climate-ref-esmvaltool/src/climate_ref_esmvaltool/diagnostics/zec.py	Add CMIP7 alternative requirement + dynamic recipe update source.
packages/climate-ref-esmvaltool/src/climate_ref_esmvaltool/diagnostics/tcre.py	Add CMIP7 alternative requirement + dynamic recipe update source.
packages/climate-ref-esmvaltool/src/climate_ref_esmvaltool/diagnostics/tcr.py	Add CMIP7 alternative requirement + dynamic recipe update source.
packages/climate-ref-esmvaltool/src/climate_ref_esmvaltool/diagnostics/sea_ice_sensitivity.py	Add CMIP7 alternative requirement + dynamic recipe update source.
packages/climate-ref-esmvaltool/src/climate_ref_esmvaltool/diagnostics/sea_ice_area_basic.py	Add CMIP7 alternative requirement + dynamic recipe update source.
packages/climate-ref-esmvaltool/src/climate_ref_esmvaltool/diagnostics/regional_historical_changes.py	Add CMIP7 data requirement alternatives + CMIP7 TestCases.
packages/climate-ref-esmvaltool/src/climate_ref_esmvaltool/diagnostics/example.py	Add CMIP7 alternative requirement + dynamic recipe update source.
packages/climate-ref-esmvaltool/src/climate_ref_esmvaltool/diagnostics/enso.py	Add CMIP7 alternative requirements + dynamic recipe update source.
packages/climate-ref-esmvaltool/src/climate_ref_esmvaltool/diagnostics/ecs.py	Add CMIP7 alternative requirement + dynamic recipe update source.
packages/climate-ref-esmvaltool/src/climate_ref_esmvaltool/diagnostics/cloud_scatterplots.py	Generalize CMIP requirements (CMIP6/CMIP7) + make plot title project-aware.
packages/climate-ref-esmvaltool/src/climate_ref_esmvaltool/diagnostics/cloud_radiative_effects.py	Add CMIP7 alternative requirement + dynamic recipe update source.
packages/climate-ref-esmvaltool/src/climate_ref_esmvaltool/diagnostics/climate_drivers_for_fire.py	Add CMIP7 alternative requirement + dynamic recipe update source.
packages/climate-ref-esmvaltool/src/climate_ref_esmvaltool/diagnostics/climate_at_global_warming_levels.py	Add CMIP7 alternative requirement + CMIP7-specific matching/grouping facets.

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

...climate-ref-esmvaltool/src/climate_ref_esmvaltool/diagnostics/regional_historical_changes.py

lewisjared · 2026-02-09T12:20:49Z

@bouweandela I was able to run both the cmip6/cmip7 test cases for the historical annual cycle locally.

Let me know if there is any obvious mistakes before I start fleshing out more of the test cases.

bouweandela · 2026-02-09T14:22:53Z

Looks like a great start @lewisjared!

Regarding the CMIP7 data requirements:

the CMIP6 table_id consisted of both the realm and the frequency, so you may want to put entries for realm in the data requirements too
the ESMValTool diagnostics have been developed for global data, so I doubt that anything other than region: glb will work

I noticed some failing tests because facets are different per project (e.g. obs4MIPs doesn't have ensemble members) but that is not reflected in the current implementation, see https://github.com/Climate-REF/climate-ref/actions/runs/21719510099/job/62644893377#logs

bouweandela · 2026-02-09T14:27:51Z

We may also want to consider/check which the diagnostic authors which branding_suffix should be used for the variables in the data requirements. All of them? Only one? Different per variable?

lewisjared · 2026-02-11T04:33:51Z

the CMIP6 table_id consisted of both the realm and the frequency, so you may want to put entries for realm in the data requirements too

According to the global attributes doc "Note that "realm" may be assigned multiple realms, separated by a single space, with the first one listed considered primary.".
I think we will currently compare exactly. I'll make a issue and follow up with this.

Do we have any information about how the external_variables files will be named? I can't find anything on the CMIP7 guidance site.

bouweandela · 2026-02-11T10:40:06Z

According to the global attributes doc "Note that "realm" may be assigned multiple realms, separated by a single space, with the first one listed considered primary.".

I used the CMIP6 table_id instead of just frequency in the data requirements for the ESMValTool diagnostics because I noticed that some of the diagnostics failed to run if the realm was different from expected. ESMValTool recipes will expect the value of mip, which I believe translates to realm in CMIP7 and table_id in CMIP6, to have the value for table_id (minus the "Table " bit) used at the top of the CMOR tables. For CMIP7, the table_id in the CMOR tables appears to be the same as realm: https://github.com/WCRP-CMIP/cmip7-cmor-tables/blob/6737d39d5424ad20550ad117f28512cf69fa2901/tables/CMIP7_aerosol.json#L15, while for CMIP6 it was usually a composite of the first (few) letter(s) of the realm followed by the frequency https://github.com/PCMDI/cmip6-cmor-tables/blob/087fe45d21c082e28723e0f930e4266abe91b853/Tables/CMIP6_Amon.json#L5.

Do we have any information about how the external_variables files will be named? I can't find anything on the CMIP7 guidance site.

They will follow the same naming scheme as other variables. Here is an example of the areacella variable in the CMOR tables: https://github.com/WCRP-CMIP/cmip7-cmor-tables/blob/6737d39d5424ad20550ad117f28512cf69fa2901/tables/CMIP7_atmos.json#L54
and with the CMIP6 as CMIP7 CMORizer example notebook that results in something like MIP-DRS7/CMIP7/CMIP/PCMDI/PCMDI-test-1-0/historical/r1i1p1f3/glb/fx/areacella/ti-u-hxy-u/gn/v20260109/areacella_ti-u-hxy-u_fx_glb_gn_PCMDI-test-1-0_historical_r1i1p1f3.nc

Copilot

Pull request overview

Copilot reviewed 29 out of 29 changed files in this pull request and generated 1 comment.

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

...te-ref-esmvaltool/src/climate_ref_esmvaltool/diagnostics/climate_at_global_warming_levels.py

… handling - Add CMIP7 data requirements to all ESMValTool diagnostics - Add CMIP7 recipe handling with branded variable support - Add DReq mappings and region/branded variable metadata - Add branded_variable_name as derived column in CMIP7 data catalog - Filter CMIP7-only facets from CMIP6 ESGF searches - Add CMIP7 test data specifications to all diagnostics - Use CMIP7-style filenames for converted datasets - Install dev ESMValCore for CMIP7 support - Unify DReq enrichment and clean up CMIP7 data requirements

…so-characteristics The hardcoded TROPFLUX dataset entry was missing the "mip": "Omon" facet that the upstream ESMValTool recipe requires, causing a frequency mapping error during CMIP7 test case execution.

When the CMIP7 conversion cache filenames change (e.g. adding time ranges), stale symlinks from previous test runs remain in the climate_data directory. ESMValCore's glob matches these dangling symlinks and fails with FileNotFoundError. Now removes dangling symlinks from each target directory before creating new ones, preventing stale references from confusing ESMValCore's file discovery.

…atasets rewrite_mip_for_cmip7 only checked top-level and diagnostic-level datasets, missing recipes that place CMIP7 datasets at the variable level (climate-drivers-for-fire, ecs, sea-ice-sensitivity). Refactored detection into _iter_recipe_datasets and mip rewriting into _rewrite_variable_mip for clarity. The rewrite also preserves the original CMIP6 mip on non-CMIP7 additional_datasets (e.g. OBS).

…7 conversion CMIP6 activity_id can contain multiple activities separated by spaces (e.g. "C4MIP CDRMIP") which breaks directory paths and glob patterns. Use only the first activity for CMIP7 DRS paths and metadata. Also add version and sanitized activity_id as global attributes on converted CMIP7 NetCDF files so that parse_cmip7_file can construct correct instance_ids with version directories.

…le CMIP7 per-variable datasets - Add missing realm facet to all CMIP7 FacetFilter definitions (atmos, ocean, seaIce, land as appropriate) - Change test_data_spec source_ids where ESGF data was unavailable: ZEC: CanESM5 -> ACCESS-ESM1-5, TCRE: CanESM5 -> MPI-ESM1-2-LR, cloud-scatterplots-cli-ta: CanESM5 -> CESM2 - Add table_id filter to cloud-scatterplots-cli-ta CMIP7Request to avoid ESGF returning ta from wrong table (AERmonZ instead of Amon) - Fix cloud_scatterplots update_recipe to remove diagnostic-level additional_datasets when using CMIP7 per-variable datasets

…cessor ESMValCore changed annual_statistics to drop the year coordinate by default (keep_group_coordinates=False). Several diagnostic scripts still rely on it, so patch recipes at runtime to set keep_group_coordinates=True on every annual_statistics step.

Use max() instead of next() when globbing the executions directory so the most recent timestamped recipe run is always selected.

lewisjared · 2026-02-16T01:30:46Z

CMIP7 ESMValTool Diagnostic Status

Tested all 18 ESMValTool diagnostics against CMIP7 data. 9 pass, 8 fail, 1 disabled.

Passing (9)

cloud-scatterplots-clwvi-pr
enso-basic-climatology
global-mean-timeseries
regional-historical-annual-cycle
regional-historical-timeseries
regional-historical-trend
transient-climate-response
transient-climate-response-emissions
zero-emission-commitment

Failing — blocked on ESMValCore upstream issues (8)

CMIP7 CMOR tables missing (4 diagnostics):
Variables like lwcre, swcre, and rtnt can't be resolved when project='CMIP7' because the mip value doesn't map to a loadable CMOR table. ESMValGroup/ESMValCore#2980.

climate-drivers-for-fire (tasmax / atmos)
cloud-radiative-effects (lwcre / atmos)
cloud-scatterplots-clivi-lwcre (lwcre / atmos)
cloud-scatterplots-clt-swcre (swcre / atmos)
equilibrium-climate-sensitivity (rtnt / atmos)

The other set of issues which are yet to be resolved is due to the fixes not being applied for these CMIP7 datasets. @bouweandela Is there a gold standard source that we could use that doesn't require fixes?

CoordinateMultiDimError on curvilinear ocean grids (3 diagnostics):
area_statistics fails on multi-dimensional coordinates (e.g., CanESM5 tripolar ocean grid gn) because try_adding_calculated_cell_area runs before using the provided areacello.

enso-characteristics (CanESM5 tos)
sea-ice-area-basic (CanESM5 siconc)
sea-ice-sensitivity (CanESM5 siconc)

CMORCheckError on vertical coordinate metadata (1 diagnostic):
alevel coordinate's standard_name check fails for the cli variable.

cloud-scatterplots-cli-ta (CESM2 cli)

Disabled (1)

climate-at-global-warming-levels — needs scenario data, not yet available in CMIP7

* origin/main: chore: rename changelog to match PR number feat: add file dimensions to all ESMValTool diagnostics feat: add file dimensions to ESMValTool diagnostics

… rename to branded_variable - Add SQLAlchemy hybrid_property on CMIP7Dataset for branded_variable (variable_id + "_" + branding_suffix), eliminating the standalone _add_branded_variable_name() function and load_catalog() override - Rename branded_variable_name to branded_variable across the codebase to match the CMIP7 file attribute name - Add variant_label to diagnostic facets for TCR, TCRE, ZEC, and regional historical trend so CMEC bundle validation passes for both CMIP6 (member_id) and CMIP7 (variant_label) test cases

lewisjared · 2026-02-16T03:19:43Z

...-esmvaltool/tests/unit/diagnostics/recipes/recipe_climate_at_global_warming_levels_cmip6.yml

      operator: mean
    annual_statistics:
      operator: mean
+      keep_group_coordinates: true


@bouweandela Did the default for this change recently?

Yes: ESMValGroup/ESMValCore#2787

At least I'm not going crazy!

bouweandela · 2026-02-16T08:45:29Z

The other set of issues which are yet to be resolved is due to the fixes not being applied for these CMIP7 datasets. @bouweandela Is there a gold standard source that we could use that doesn't require fixes?

You could load the CMIP6 data using esmvalcore.dataset.Dataset, this will apply the fixes and will give you an iris.cube.Cube. You could save that using iris.save or convert it to xarray using ncdata and then you can do the conversion to CMIP7.

lewisjared · 2026-02-16T10:56:42Z

I was trying to avoid the dependency on esmvalcore unless truely needed. The conversions don't just happen in a single standalone script.

bouweandela · 2026-02-16T14:43:07Z

It depends on what needs to be fixed, but most fixes are fairly straightforward, so it may not be much work to copy them from ESMValCore. This one is applied to all datasets: https://github.com/ESMValGroup/ESMValCore/blob/da81d5f67158f3d2603831b56ab6b4fb8a388d86/esmvalcore/cmor/_fixes/fix.py#L360 and then there are project/dataset/variable specific fixes in https://github.com/ESMValGroup/ESMValCore/blob/main/esmvalcore/cmor/_fixes/, organized as project.dataset.Variable.

lewisjared requested a review from Copilot February 9, 2026 05:42

Copilot started reviewing on behalf of lewisjared February 9, 2026 05:42 View session

Copilot AI reviewed Feb 9, 2026

View reviewed changes

lewisjared requested a review from bouweandela February 9, 2026 12:19

Copilot AI review requested due to automatic review settings February 12, 2026 14:41

Copilot started reviewing on behalf of lewisjared February 12, 2026 14:42 View session

Copilot AI reviewed Feb 12, 2026

View reviewed changes

...te-ref-esmvaltool/src/climate_ref_esmvaltool/diagnostics/climate_at_global_warming_levels.py Show resolved Hide resolved

lewisjared force-pushed the esmval-requirements branch from d33be38 to 885bf16 Compare February 13, 2026 07:21

lewisjared added 7 commits February 14, 2026 18:13

chore: update esmvalcore to latest commit

8be7dfd

fix(esmvaltool): add missing mip facet to TROPFLUX OBS6 dataset in en…

6eab71d

…so-characteristics The hardcoded TROPFLUX dataset entry was missing the "mip": "Omon" facet that the upstream ESMValTool recipe requires, causing a frequency mapping error during CMIP7 test case execution.

chore: update the CMIP7 recipe to use parent experiments

00c5a95

chore: add cmip7 parent facets

24c5b1b

lewisjared force-pushed the esmval-requirements branch from 885bf16 to ca866a9 Compare February 15, 2026 05:26

lewisjared force-pushed the main branch from 24c5b1b to 3b5c967 Compare February 15, 2026 05:31

lewisjared added 4 commits February 15, 2026 22:35

fix(esmvaltool): pick latest recipe run when multiple executions exist

d11eb68

Use max() instead of next() when globbing the executions directory so the most recent timestamped recipe run is always selected.

Merge remote-tracking branch 'origin/main' into esmval-requirements

f0d04c2

* origin/main: chore: rename changelog to match PR number feat: add file dimensions to all ESMValTool diagnostics feat: add file dimensions to ESMValTool diagnostics

lewisjared mentioned this pull request Feb 16, 2026

CMIP7 readiness #536

Open

42 tasks

lewisjared commented Feb 16, 2026

View reviewed changes

lewisjared merged commit fb80c2e into main Feb 16, 2026
24 of 25 checks passed

lewisjared deleted the esmval-requirements branch February 16, 2026 03:41

bouweandela mentioned this pull request Feb 16, 2026

Replace out_name by variable_id in CMIP7 paths #539

Open

3 tasks

Conversation

lewisjared commented Feb 9, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Summary

Per-diagnostic pattern

Uh oh!

codecov bot commented Feb 9, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Codecov Report

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Pull request overview

Reviewed changes

Uh oh!

Uh oh!

Uh oh!

Uh oh!

lewisjared commented Feb 9, 2026

Uh oh!

bouweandela commented Feb 9, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

bouweandela commented Feb 9, 2026

Uh oh!

lewisjared commented Feb 11, 2026

Uh oh!

bouweandela commented Feb 11, 2026

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Pull request overview

Uh oh!

Uh oh!

lewisjared commented Feb 16, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

CMIP7 ESMValTool Diagnostic Status

Passing (9)

Failing — blocked on ESMValCore upstream issues (8)

Disabled (1)

Uh oh!

lewisjared Feb 16, 2026

Choose a reason for hiding this comment

Uh oh!

bouweandela Feb 16, 2026

Choose a reason for hiding this comment

Uh oh!

lewisjared Feb 16, 2026

Choose a reason for hiding this comment

Uh oh!

Uh oh!

bouweandela commented Feb 16, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

lewisjared commented Feb 16, 2026

Uh oh!

bouweandela commented Feb 16, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

lewisjared commented Feb 9, 2026 •

edited

Loading

codecov bot commented Feb 9, 2026 •

edited

Loading

bouweandela commented Feb 9, 2026 •

edited

Loading

lewisjared commented Feb 16, 2026 •

edited

Loading

bouweandela commented Feb 16, 2026 •

edited

Loading