inconsistent grid coordinates for `CESAM-UA.WRF451Q` datasets across variables

[larsbuntemeyer]

WRF451Q datasets have inconsistent grid coordinates across variables of the same simulation. This came up when I was trying to align those datasets. This is an example:

import xarray as xr

wrf_tas = xr.open_dataset("/mnt/CORDEX_CMIP6_tmp/sim_data/CORDEX-CMIP6/DD/EUR-12/CESAM-UA/ERA5/evaluation/r1i1p1f1/WRF451Q/v1-r1/mon/tas/v20250520/tas_EUR-12_ERA5_evaluation_r1i1p1f1_CESAM-UA_WRF451Q_v1-r1_mon_198001-198912.nc")
wrf_pr = xr.open_dataset("/mnt/CORDEX_CMIP6_tmp/sim_data/CORDEX-CMIP6/DD/EUR-12/CESAM-UA/ERA5/evaluation/r1i1p1f1/WRF451Q/v1-r1/mon/pr/v20250520/pr_EUR-12_ERA5_evaluation_r1i1p1f1_CESAM-UA_WRF451Q_v1-r1_mon_198001-198912.nc")  

now compare:

wrf_tas.rlon.data

gives

array([-10.7745409 , -10.67769241, -10.58078861, -10.48377323,
       -10.38669205, -10.28950214, -10.19225407, -10.09489632,
        -9.99748039,  -9.89994907,  -9.80235958,  -9.70466328,
...

while

wrf_pr.rlon.data

gives

array([-10.32560158, -10.23278809, -10.13992214, -10.04694939,
        -9.95391273,  -9.86077309,  -9.76757717,  -9.6742754 ,
        -9.58091831,  -9.48745155,  -9.39392757,  -9.30030251,
...

Unfortunately, we can not rewrite those coordinates because WRF451Q also has another own definition of EUR-12 in rotated pole coordinates, see WCRP-CORDEX/archive-specifications#28

[larsbuntemeyer]

Pinging @dmoreno11 (as registrator of CESAM-UA).

[dmoreno11]

Hi Lars, just detected the error, the cmorization code was picking up a staggered set of coords instead of the unstaggered set. Running the code now to fix this, should upload fresh set of vars during next week. Thanks for detecting this.

[dmoreno11]

Regarding the rotated pole coordinates, I've re-checked the archive specs doc, version 2 (https://doi.org/10.5281/zenodo.10961068)

and my files seem to comply with the specs:

[larsbuntemeyer]

Thanks for the quick fix! Regarding the rotated pole definition, that is fine, in deed! My comment just refers to the rlon and rlat coordinates, which are different from the original definition of EUR-12, e.g, different number of grid boxes, different lower left coordinates. Having a common rotate pole grid in the ensemble always makes comparison easy. However, I have no experience in running wrf so i guess this is the original grid of the simulation or did you do some regridding during cmorization?

[dmoreno11]

Hi again Lars, no regridding done, this is raw output grid from the WRF model. However, I just noticed that the difference between the nº of gridboxes between the original definition of EUR-12 and the ones from my files are the 10 relaxation points all around the domain that should be removed from the wrf output files (thus originating less 20 points in lat and less 20 points in lon). Will apply that fix in the new batch of files I'll upload next week.