Just for posterity, there's some relevant discussion elsewhere on this
list in a different thread:
http://www.mail-archive.com/cf-metadata at cgd.ucar.edu/msg01539.html.
-----Original Message-----
From: Jonathan Gregory [mailto:jonathan at met.reading.ac.uk] On Behalf Of
Jonathan Gregory
Sent: 04 June 2010 08:43
To: Jon Blower
Cc: cf-metadata at cgd.ucar.edu
Subject: [CF-metadata] bounds/precision for time axis
Dear Jon
> My worry is the subtle distinction between
> "resolution" and "bounds/cell_methods", which I would have to think
> through properly at some more sociable hour.
I think the distinction is:
* resolution of the time-bounds refers to how accurately you can
characterise
the point at which, or the interval within which, the data was
collected. E.g.
was it at 1200, or from midnight to midnight, with a precision of 1 ms,
or of
1 h? Perhaps that could be recorded in CF by using ancillary_data and
standard_name modifiers (CF 3.3 and 3.4) for the coordinates or bounds.
(I'm
not exactly sure how, since they were designed for data variables.)
* cell_methods refers to how the value in the data variable represents
the
variation within the cell - is it a single value at a particular point
(point),
or a single value selected somehow from all the values within the
interval
(maximum, minimum, median), or a combination of values (mean, standard_
deviation)? I suggested "cell" could be used to indicate, for an
intensive
variable, that it is an ill-defined combination of values which is
somehow
typical of the cell. We don't have to use "cell", which I suggested just
because it already exists. We could use something else e.g. "typical" or
"representative".
Best wishes
Jonathan
Received on Thu Oct 21 2010 - 09:57:56 BST
