Hi all,
Normally CF doesn't prohibit extra attributes being defined, so I'm not
sure whether the checker should be objecting. That being said, I don't
think the convention should sanction attaching a cell_methods to a
coordinate variable *instead* of attaching it to a regular variable
because then software attempting to determine what the cell_methods are
would have to look both places. Currently, software can always find
*all* the cell_methods in a single place-- attached to the variable.
best regards,
Karl
On 6/1/15 2:35 AM, David Hassell wrote:
Hello Mahalo, Charlie,
My gut feeling was "it shouldn't", but I then thought "why not?" after
I couldn't think of any counter examples hwich would cause problems.
Section 7.3 doesn't disallow it, I think, and I don't think there will
be any conflict or ambiguity arising from its use. Appendix A would
need to be updated and perhaps some text in 7.3 highlighting that it
is OK, clarifying that the variable (the time coordinate in your
example) "defines its own cells"?
This is similar to the recent (ongoing) thread on allowing a
coordinate variable to have an ancillary_variables attribute.
Does that still make sense?
All the best,
David
---- Original message from Charlie Zender (05PM 31 May 15)
Date: Sun, 31 May 2015 17:17:35 -0700
From: Charlie Zender <[email protected]>
To: CF Metadata Mail List <[email protected]>
Subject: [CF-metadata] Are cell_methods attributes OK for coordinate
variables?
User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:31.0) Gecko/20100101
Thunderbird/31.4.0
Aloha CFers,
Is it correct to add a cell_methods = "time: mean" attribute to the
time coordinate when the coordinate is averaged over time?
NCO's ncra does this (though the NERC CF checker doesn't like it).
It's clear from the CF docs that ncra time-averaging a variable like
wind(time) from an array to a degenerate (size 1) array should
add a cell_methods = "time: mean" attribute to wind.
Yet should it add cell_methods to the time coordinate itself?
My take is that it should. The time_bounds variable, if any,
will show the original extent of the temporal range, and the time
coordinate value contains the mean of the original range.
Yet an NCO user is making a good case that cell_methods are only
for non-coordinate variables. His point is that many variables with
different cell_methods can all contain the same coordinate, so that
the coordinate should not have cell_methods. My response is that
cell_methods refers to the bounds variable of the coordinate, not
the coordinate values themselves.
Is the question clear? If not, I can supply CDL...
Mahalo,
Charlie
p.s. the CF list archives on the CF homepage poinst to
http://mailman.cgd.ucar.edu/pipermail/cf-metadata
and that URL currently (and for the past ~week) yields this error:
The requested URL /pipermail/cf-metadata/ was not found on this server.
--
Charlie Zender, Earth System Sci. & Computer Sci.
University of California, Irvine 949-891-2429 )'(
_______________________________________________
CF-metadata mailing list
[email protected]
http://mailman.cgd.ucar.edu/mailman/listinfo/cf-metadata
--
David Hassell
National Centre for Atmospheric Science (NCAS)
Department of Meteorology, University of Reading,
Earley Gate, PO Box 243,
Reading RG6 6BB, U.K.
Tel : +44 118 3785613
E-mail: [email protected]
_______________________________________________
CF-metadata mailing list
[email protected]
http://mailman.cgd.ucar.edu/mailman/listinfo/cf-metadata
_______________________________________________
CF-metadata mailing list
[email protected]
http://mailman.cgd.ucar.edu/mailman/listinfo/cf-metadata
