[CF-metadata] New standard name: datetime_iso8601

From: Jonathan Gregory <j.m.gregory>
Date: Tue, 19 Mar 2013 19:34:37 +0000

Dear Aleksandar

Thanks for clarifying your proposal. I am one of those who misinterpreted you,
then. Sorry about that. Is the proposal for the use of date-time strings in
auxiliary coordinate variables only, not in (Unidata) coordinate variables,
to provide a human-readable equivalent to the encoded time coordinate variable?

In that case, could we require that there is also a time coordinate variable,
encoded numerically as usual? This is, of course, redundant, and that means
it will give rise to inconsistency, due to mistakes being made. Inconsistency
could be detected automatically e.g. by the CF checker, and I suppose the
numerical coordinate variable should take precedence. Is the facility to store
human-readable strings in the netCDF file sufficiently valuable to justify this
increased complication?

Is the proposal only for the real-world calendar, or do you also propose the
date-time strings to be valid for model calendars? I guess the ISO standard
only applies to the real world.

I agree that this would not be a fundamental change to the convention. It
would be a material change, however, if there were new requirements such as
above; those would be additional rules in the convention, so they'd need to be
proposed in a trac ticket.

This email list is where standard name proposals are discussed. I agree that
a standard name for strings of this kind would be needed.

Best wishes and thanks

Jonathan
Received on Tue Mar 19 2013 - 13:34:37 GMT