Fixing of CORDEX datasets in lambert conformal conical projection

[Neah-Ko]

Description

Fixing of CORDEX datasets in lambert conformal conical projection.

Some datasets have wrongly defined coordinates, especially on the boundaries.

The fix behaves this way:

1. Checks for differences between the coordinates boundaries and the CORDEX standard domain boundaries

• We check both geographical coordinates and projections coordinates.
  • If the coordinates are matching the standard domain, we fix the metadata and make sure there are bounds
  • This step necessitates to pass from one coordinate system to another to make comparison in the same coordinate system.
    i.e. latitude/longitude are in geographical and projection_x/y_coordinate are in lambert.
  • For now the checks works with some thresholds that could be tuned.

2. Fix faulty coordinate

• if both are wrong: we raise an error
• else
  • if one is wrong and one is right, we use the right one to re-create the wrong one.

Closes #1879

Link to documentation

---

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
• 🧪 and 🛠 Documentation is available
• 🛠 Unit tests have been added
• 🛠 Changes are backward compatible
• 🛠 Any changed dependencies have been added or removed correctly
• 🛠 The list of authors is up to date
• 🛠 All checks below this pull request were successful

---

[codecov]

Codecov Report

Merging #1956 (3dde4cc) into main (72b4169) will increase coverage by 0.04%.
The diff coverage is 100.00%.

@@            Coverage Diff             @@
##             main    #1956      +/-   ##
==========================================
+ Coverage   92.79%   92.84%   +0.04%     
==========================================
  Files         236      253      +17     
  Lines       12438    12544     +106     
==========================================
+ Hits        11542    11646     +104     
- Misses        896      898       +2     

Impacted Files	Coverage Δ
esmvalcore/cmor/check.py	97.56% <ø> (-0.01%)	⬇️
...ixes/cordex/cnrm_cerfacs_cnrm_cm5/cnrm_aladin52.py	100.00% <100.00%> (ø)
...ixes/cordex/cnrm_cerfacs_cnrm_cm5/cnrm_aladin53.py	100.00% <100.00%> (ø)
...ixes/cordex/cnrm_cerfacs_cnrm_cm5/cnrm_aladin63.py	100.00% <100.00%> (ø)
...ixes/cordex/cnrm_cerfacs_cnrm_cm5/ictp_regcm4_6.py	100.00% <100.00%> (ø)
esmvalcore/cmor/_fixes/cordex/cordex_fixes.py	100.00% <100.00%> (ø)
...e/cmor/_fixes/cordex/ecmwf_eraint/cnrm_aladin52.py	100.00% <100.00%> (ø)
...e/cmor/_fixes/cordex/ecmwf_eraint/cnrm_aladin53.py	100.00% <100.00%> (ø)
...e/cmor/_fixes/cordex/ecmwf_eraint/cnrm_aladin63.py	100.00% <100.00%> (ø)
...e/cmor/_fixes/cordex/ecmwf_eraint/ictp_regcm4_6.py	100.00% <100.00%> (ø)
... and 12 more

... and 1 file with indirect coverage changes

Help us with your feedback. Take ten seconds to tell us how you rate us. Have a feature suggestion? Share it here.

[sloosvel]

Thanks for the work @Neah-Ko !! Looks like we are close to a proper fix for these datasets, but first could you please address the following comments?

[sloosvel]

Thanks looking into all of @Neah-Ko !! I tested a bit the fixes with data and there is a bit reorganization to be done. Also with @pepcos we discussed that maybe it would be best to apply the fixes at a dataset level at the end, because for some datasets the results look a bit shifted. I will take what you have already done that is already working nicely and adapt it a bit so that it covers the datasets we have tested.

[sloosvel]

In summary:

• This PR introduces a general check for CORDEX datasets in an LCC system to see if the geographical coordinates are within the domain given by the specifications. The domain description is given by a set of latitudes and longitudes that describe 9 grid points and can be found in the added file npr_rcm_domain_boundaries.json taken from this document.

• The projection coordinates in some datasets are shifted and do not match the domain described by the geographical coordinates (see some plots generated by @Neah-Ko in these notebooks. This PR introduces a fix that computes the projection coordinates from the geographical coordinates doing a crs transformation, guesses the 1D bounds, and then uses the guessed bounds 1D bounds in order to compute the geographical 2D bounds by doing a crs transformation. Coordinate metadata is also fixed. This fix has been tested in datasets ALADIN52, ALADIN53, ALADIN63, ICTP-RegCM4-6 in the EUR and MED domains.

• There is an additonal small metadata fix in dataset DMI-HIRHAM5.