Change in CF-1.7: _FillValue and missing_value allowed for all auxiliary coordinates

[zklaus]

In CF-1.7 Table A.1. Attributes in Appendix A was changed to allow _FillValue and missing_value in all auxiliary coordinates.

QA-DKRZ allows them only in discrete sampling geometries in accordance with CF-1.6 (tag: CF_251e).

@h-dh Can this be changed to allow those attributes for all auxiliary coordinates?

[h-dh]

I am going to attend a national meeting tomorrow and will be back on this on Thursday. cheers, hdh

…

On 6/25/19 2:36 PM, Klaus Zimmermann wrote: In CF-1.7 Table A.1. Attributes in Appendix A was changed to allow |_FillValue| and |missing_value| in all auxiliary coordinates. QA-DKRZ allows them only in discrete sampling geometries in accordance with CF-1.6 (tag: CF_251e). @h-dh <https://github.com/h-dh> Can this be changed to allow those attributes for all auxiliary coordinates? — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <#16?email_source=notifications&email_token=ACJZOW5PAAMNJ5HAFHTWLDLP4IGMXA5CNFSM4H3H2XP2YY3PNVWWK3TUL52HS4DFUVEXG43VMWVGG33NNVSW45C7NFSM4G3RD23Q>, or mute the thread <https://github.com/notifications/unsubscribe-auth/ACJZOWZQJPKFQNDP44EKL5TP4IGMXANCNFSM4H3H2XPQ>.

-- Dr. Heinz-Dieter Hollweg Abteilung Datenmanagement Deutsches Klimarechenzentrum GmbH (DKRZ) Bundesstraße 45a • D-20146 Hamburg • Germany Phone: +49 40 460094-212 FAX: +49 40 460094-270 Email: hollweg@dkrz.de URL: www.dkrz.de Geschäftsführer: Prof. Dr. Thomas Ludwig Sitz der Gesellschaft: Hamburg Amtsgericht Hamburg HRB 39784

[h-dh]

I admit that this slipped my attention.
Do you have a reasonable example where missing values in auxiliary coordinates make sense?
cheers,
hdh

[zklaus]

It is always useful when you want to attach another mask to data. One example where I discussed this recently with a colleague is for climate indices where you might be able to compute perfectly reasonable values for all points, but can only ascertain the quality where the reference dataset is available. Similar ideas could apply for quality control of observational datasets, like cloud products where perhaps the quality level can not be determined at all points.

I admit that this is not very specific and thought through, but I can tell you that we are seriously exploring these applications.