critical dataset issues for joint evaluation

[larsbuntemeyer]

The following issues do stop those datasets from being included in the joint evaluation (ERA5) right now:

• corrupt grid mapping for ICTP.RegCM5-0 #91 (fixable by regridding)
• inconsistent DRS for COSMO-CLM-6-0-clm3 #107 (fixed by sloppy parsing, but is missing masks and ororaphy)
• duplicated timesteps for RMIB-UGent.ALARO1-SFX #115 (has to be fixed by modeller)
• issues with CCLM6-0-1-URB-ESG #66 (has to be fixed by modeller) concerns parts of the dataset, e.g., tas, see issues with CCLM6-0-1-URB-ESG #66 (comment)
• inconsistent grid coordinates for CESAM-UA.WRF451Q datasets across variables #117 Also WRF451Q will require regridding due to different rotated pole grid definition.

[larsbuntemeyer]

@JavierDiezSierra I'll fix #91 today by regridding. If the other issues are resolved by the modellers (they are notified), this data should also be available through the catalog.

[JavierDiezSierra]

@larsbuntemeyer Great!. I will rerun the scripts, once you fix the problem, to add RegCM model. Many thanks!

[larsbuntemeyer]

Ok, I'll also merge #109 today when i added that fix!

[larsbuntemeyer]

I fixed those issues with WRF451Q (by CESAM-UA) and RegCM5-0 by regridding them. RegCM5-0 has still a little messy grid mapping definitions so it's not possible to produce bounds for that yet. In the meantime, ROAM-NBS was added in the meantime which seems to work out of the box.

• CCLM6-0-1-URB-ESG still has problems with non-monotonic time axis,
• COSMO-CLM-6-0-clm3 does not provide masking and orography (also not registered)

Make sure to use the main of evaltools!

I also have to rethink the masking. Probably it's better to skip it completely and only use unmapped_to_nan if regridding to EUR-12. Otherwise we have models masked if regridded and others not. This doesn't make a diffrerence in comparison with EOBS, once you compute a bias with respect to EOBS, it's only important EOBS has nan over water after regridding to EUR-12.

UPDATE: ALARO1-SFX also still missing due to #115

[JavierDiezSierra]

@larsbuntemeyer Many thanks! I’ll update the figures over the next few days. Enjoy your vacation!

[kobebryant432]

#115 Should be fixed - I left the issue open to be double checked but RMIB-Ugent_ALARO.SFX should no longer be blocked.

[vmaurerDWD]

I fixed those issues with WRF451Q (by CESAM-UA) and RegCM5-0 by regridding them. RegCM5-0 has still a little messy grid mapping definitions so it's not possible to produce bounds for that yet. In the meantime, ROAM-NBS was added in the meantime which seems to work out of the box.

* `CCLM6-0-1-URB-ESG` still has problems with non-monotonic time axis,

* `COSMO-CLM-6-0-clm3` does not provide masking and orography (also not registered)

Make sure to use the main of evaltools!

I also have to rethink the masking. Probably it's better to skip it completely and only use unmapped_to_nan if regridding to EUR-12. Otherwise we have models masked if regridded and others not. This doesn't make a diffrerence in comparison with EOBS, once you compute a bias with respect to EOBS, it's only important EOBS has nan over water after regridding to EUR-12.

UPDATE: ALARO1-SFX also still missing due to #115

Hi @larsbuntemeyer , as you are saying, ROAM-NBS was added and seems to work, but it is not displayed as available in any of the availability plots. Is there still something I have to add somewhere?

[jesusff]

Hi Vera. Sorry, the availability plots are not updated automatically (we should include it in the nightly update? @larsbuntemeyer). I just updated the plots and ROAM-NBS is there 🙂