CRITICAL: Netcdf output not CF-compliant

[MuayScience]

The netcdf output for the model is not CF-compliant. This is true of the dyn, phys, MOM6 and CICE5 files. The MOM6 and CICE5 files cannot be post-processed correctly using ESMF tools as a result.

[JessicaMeixner-NOAA]

Is this just an issue of using a different diag_table or ice_in? If not this will need to be addressed by the model components individually.

[MuayScience]

It needs to be addressed wherever the attributes for the netcdf output are being set. It's mostly things like non-compliant units attributes, but it turns out ESMF at least is extremely sensitive to any departures from CF-compliance.

[DeniseWorthen]

Would it be possible for you to list the non-compliant units attributes? Denise

…

On Tue, Dec 3, 2019 at 9:52 AM MuayScience ***@***.***> wrote: It needs to be addressed wherever the attributes for the netcdf output are being set. It's mostly things like non-compliant units attributes, but it turns out ESMF at least is extremely sensitive to any departures from CF-compliance. — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub <#6?email_source=notifications&email_token=AJU7JZHCAIC2BUXUDY2RKNLQWZXDHA5CNFSM4JRNJLJKYY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOEFZUKCQ#issuecomment-561202442>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AJU7JZCSGKU5BLAIY2RIQ6LQWZXDHANCNFSM4JRNJLJA> .

-- Denise Worthen IMSG at NOAA/NWS/NCEP/EMC Denise.Worthen@noaa.gov

[MuayScience]

Hi Denise - I've attached a file here that gives a full CF-compliance check for the 6-hourly cice output. It includes warnings and errors. Is this a useful format?
CF-Convention compliance check for cice.h2.06h.pdf

[DeniseWorthen]

Yes, that helps. These are component-defined fields however, so making them CF-compliant would be an issue for the CICE5 developers---or more importantly for the CICE6 developers since CICE5 is frozen. Are there MOM6 non-CF compliant fields?

…

On Tue, Dec 3, 2019 at 10:29 AM Benjamin A Cash ***@***.***> wrote: Hi Denise - I've attached a file here that gives a full CF-compliance check for the 6-hourly cice output. It includes warnings and errors. Is this a useful format? CF-Convention compliance check for cice.h2.06h.pdf <https://github.com/ufs-community/ufs-s2s-model/files/3917356/CF-Convention.compliance.check.for.cice.h2.06h.pdf> — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#6?email_source=notifications&email_token=AJU7JZEBAOXOHGBGQZFK3Y3QWZ3MHA5CNFSM4JRNJLJKYY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOEFZYLQI#issuecomment-561219009>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AJU7JZGYEOUHEUUVMEY6H73QWZ3MHANCNFSM4JRNJLJA> .

-- Denise Worthen IMSG at NOAA/NWS/NCEP/EMC Denise.Worthen@noaa.gov

[MuayScience]

Yes, although I don't have a CF report at the moment in the same format. It's the same kind of issues - for example, fields that don't have units will have a 'units=none' attribute instead of just not having a units attribute. I'll try to upload an example later today.

Are there no plans then for making changes to model components within UFS? If we identify an issue like this in CICE5 as implemented in UFS is the community just stuck with it until CICE6?

[DeniseWorthen]

I think CICE5 is a little bit of a special case because no-further development of CICE5 is on-going. We could certainly accept PRs on the NOAA-EMC CICE5 repo for this issue but the priority on our end is moving to CICE6. For MOM6, if the non-compliance arises from how MOM6 itself is writing the fields, then I do believe that is an issue to be raised w/ GFDL. Denise.

…

On Tue, Dec 3, 2019 at 10:40 AM Benjamin A Cash ***@***.***> wrote: Yes, although I don't have a CF report at the moment in the same format. It's the same kind of issues - for example, fields that don't have units will have a 'units=none' attribute instead of just not having a units attribute. I'll try to upload an example later today. Are there no plans then for making changes to model components within UFS? If we identify an issue like this in CICE5 as implemented in UFS is the community just stuck with it until CICE6? — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#6?email_source=notifications&email_token=AJU7JZA7PFP527JHG5OXLD3QWZ4YXA5CNFSM4JRNJLJKYY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOEFZZXAA#issuecomment-561224576>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AJU7JZEHRBPEJIKHGSA7MBDQWZ4YXANCNFSM4JRNJLJA> .

-- Denise Worthen IMSG at NOAA/NWS/NCEP/EMC Denise.Worthen@noaa.gov

[MuayScience]

Here is a sample CF check for the mom6 static file. It has a good example of a typical kind of error in the mom6 files, where a non-dimensional field has 'units=none' instead of just not having a units attribute.
CF-Convention compliance check for mom6.static.pdf

[arunchawla-NOAA]

@DeniseWorthen was this taken up with GFDL for MOM6?

[DeniseWorthen]

As far as I know, there has been no action on this issue.

[DeniseWorthen]

I'm unsure of who exactly created this issue but I plan on closing it. I believe that EMC's conclusion was that this was an issue best raised w/ the maintainers of the authoritative repos in question.