Various fixes for CORDEX model data

[bouweandela]

Description

• Update the list of facets used to create the alias facet and group the datasets in the ensemble_statistics preprocessor function so they work for CORDEX data
• Update the list of facets used to create the ensembles in the ensemble_statistics preprocessor function with rcm_version, driver, and grid. Related to Ensemble creation in current ensemble_statistics preprocessor is not optimal #2044.
• Use standard long_name and standard variable names for coordinates for all variables from model HadREM3-GA7-05 because these are non standard
• Set use_standard_grid: true by default for models COSMO-crCLIM-v1-1, RACMO22E, and RCA4 because these models are missing bounds on the lat/lon coordinates. This could alternatively be solved by computing the lat/lon bounds from the rotated lat/lon bounds.
• Use the standard grid_latitude and grid_longitude for model WRF381P with driver MPI-M-MPI-ESM-LR Dataset problem: rlut_EUR-11_MPI-M-MPI-ESM-LR_historical_r1i1p1_IPSL-WRF381P_v1_day_19510101-20051231.nc has invalid projection coordinates #3145.
• Remove height coordinate from variable snw of model COSMO-crCLIM-v1-1 as it does not make sense for this variable
• Correct the units of sic from model HadREM3-GA7-05, the NetCDF attribute says the units are in %, but from the values it seems more likely that the units are 1.
• Correct the units of clivi and prw for model CCLM4-8-17 with driver MOHC-HadGEM2-ES. Based on comparison with the other drivers, these units are likely off by a factor of 1000, i.e. the units in the NetCDF attribute are kg m-2, but based on the values they are likely Mg m-2. Dataset problem: units of v20150320/clivi_EUR-11_MOHC-HadGEM2-ES_historical_r1i1p1_CLMcom-CCLM4-8-17_v1_day_20010101-20051230.nc are off #3147

Testing

These changes can be tested with recipe_perfmetrics_CORDEX-CMIP5.yml from ESMValGroup/ESMValTool#4199. It should also be possible to test the changes to the ensemble_statistics preprocessor function by using it in that recipe.

---

Before you get started

• ☝ Create an issue to discuss what you are going to do

Checklist

It is the responsibility of the author to make sure the pull request is ready to review. The icons indicate whether the item will be subject to the 🛠 Technical or 🧪 Scientific review.

• 🧪 The new functionality is relevant and scientifically sound
• 🛠 This pull request has a descriptive title and labels
• 🛠 Code is written according to the code quality guidelines
• 🛠 Unit tests have been added
• 🛠 Changes are backward compatible
• 🛠 The list of authors is up to date
• 🛠 All checks below this pull request were successful

---

To help with the number pull requests:

• 🙏 We kindly ask you to review two other open pull requests in this repository

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 96.26%. Comparing base (46124b9) to head (f2b4988).

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #3135      +/-   ##
==========================================
+ Coverage   96.25%   96.26%   +0.01%     
==========================================
  Files         274      280       +6     
  Lines       16189    16258      +69     
==========================================
+ Hits        15582    15651      +69     
  Misses        607      607              

☔ View full report in Codecov by Harness.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.
• 📦 JS Bundle Analysis: Save yourself from yourself by tracking and limiting bundle sizes in JS merges.

[ghossh]

I got some of this issues from the recipe yesterday. I'm going to test this PR.

[bouweandela]

Thanks! I'll stop working on this branch then and start a new branch for further fixes.

[ghossh]

You can keep making changes. I thought you finished when you put me in the reviewer. Just ping me again when you finish all changes.

[bouweandela]

It's finished for the perfmetrics recipe, but then I tried adding another recipe plotting the models without first regridding and ran into #3145.

[ghossh]

I tried the recipe and it never finishes

26-07-12 10:03:50,282 UTC [506934] INFO    diagnostic/sst                [########################################] | 100% Completed | 222.17 s
2026-07-12 10:03:50,371 UTC [506934] INFO    Successfully completed task diagnostic/sst (priority 10)in 0:12:21.248472
2026-07-12 10:03:50,533 UTC [506696] INFO    Progress: 1 tasks running, 1 tasks waiting for ancestors, 9/11 done 

not progressing from there!! There is a problem with lst variable likely. Let me check further!.

[ghossh]

The recipe doesn't finish along with lst in 4hr. But it does work seperately but takes longer time.

[bouweandela]

What settings are you using for Dask? The current recipe_perfmetrics_CORDEX-CMIP5.yml runs in about 25 minutes on Levante with these settings:

$ cat ~/.config/esmvaltool/dask.yml
max_parallel_tasks: 10
dask:
  use: threaded
  profiles:
    threaded:
      num_workers: 8

and produces this figure: ESMValGroup/ESMValTool#4199 (comment)
I did have issues with the distributed scheduler, saw a warning about a large task graph: I suspect that some of the preprocessor functions insert one or more large objects into the task graph.

[ghossh]

I was able to run the recipe successfully.