remove AMIP and PCMDI codes

[graybeal]

In 2008 (At 8:37 AM +0100 4/10/08), Jonathan Gregory wrote:

As I mentioned, for these reasons I think we should remove the AMIP and PCMDI equivalences from the [standard names] table. In any case, they are not being maintained and are out of date, and probably wrong for some GRIB codes.

These haven't been removed, and clutter up the table considerably, so this trac ticket proposes their removal.

[jonathan]

Thanks for opening this ticket, John. I haven't changed my view in the last six years! Hence I support the change.

Jonathan

[graybeal]

Karl Taylor says:

An alternative would be to replace them with the names used for CMIP5 (see ​http://cmip-pcmdi.llnl.gov/cmip5/docs/standard_output.xls ). We'd have to find a volunteer to do that.

Would an external mapping, in this day and technology age, more appropriate? Or does it benefit CF to have the Standard Names tightly coupled to the CMIP5 community?

[davidhassell]

I'm not quite sure what it means here to have an external mapping (the table that Karl points us to already maps CMIP5 names to standard names), nor if there's sufficient interest in putting the CMIP5 names in the standard name table (although I quite like the idea). Slightly more helpfully, I'm fairly sure that helping to do something is within my job description and I'd be glad to put in any effort that I am able to, if required.

All the best,

David

[jonathan]

That's a helpful offer, David!

I think that in this day and age, in Karl's phrase, any table of equivalences should be maintained separately from the standard name table. It does not particularly benefit CF to be related to a particular subset of other standards, I feel. However, it might be useful to have the equivalence between PCMDI CMIP5 names and standard_names available somewhere (the CMOR tables contain this information), and the relationship between GRIB codes and standard names might also be useful to some users of CF, provided they can be kept up to date, and if there is sufficient demand for them.

I still support the proposal as it stands.

Best wishes

Jonathan

[martin.juckes]

I support this proposal. The GRIB codes refer to GRIB 1, which is being replaced by GRIB2, a new standard with different codes. While there is a clearly a need within the community for such mappings, keeping them in the standard name table is a distraction from the core work of the standard. Keeping the mappings up to date is, as Jonathan notes, very important -- and having them embedded in the CF Standard Name table makes this harder. For the AMIP/CMIP names it is clear that the mappings are determined in the CMOR tables, and should be updated every time a new set of tables is produced -- so it makes more sense to direct users to these tables for the mappings.

regards, Martin

[durack1]

The mappings would be fairly trivial to undertake, considering that the CMIP6 (at least the placeholder working versions) are now easily web-accessible through Martins XML format and displayed at ​http://clipc-services.ceda.ac.uk/dreq/index/var.html along with the CMIP6_CVs at ​https://github.com/WCRP-CMIP/CMIP6_CVs/blob/master/CMIP6_variable.json (although I'm sure we're a couple of commits behind the latest). It would also be a great step to ensure that CMIPx and CF are consistent.

If David Hassell is still willing to undertake this step that would be great!

A point that I note, is that for many of the new (and even some of the old) CMIPx variables, a CF standard_name either currently doesn't exist and it's not obvious to me that the data requesters are making the effort to (or in many cases are even aware of the standard_name list) get these new variables "blessed" by CF.

Just my two bobs..

P

[jonathan]

Dear Paul

I believe that Alison is working on the large number of new standard names needed by CMIP6.

Best wishes

Jonathan