On 12/13/2010 1:50 PM, Karl Taylor wrote:
Dear all,
Does anyone remember, why we didn't allow the "bounds" attribute to be
attached to a scalar coordinate variable? Currently CF requires the
user to include a dimension a size one if he wants to define
coordinate bounds:
"The advantage of using a coordinate variable is that all its
attributes can be used to describe the single-valued quantity,
including boundaries."
Is there any good reason for this restriction on use of a scalar
coordinate variable?
Hi Karl,
I cannot at the moment think of any outright contradictions that arise
from allowing scalar variables to have the status of full-fledged
(degenerate) coordinate variables, as long as they are pointed to by
some dependent variable using a "coordinates" attribute.
Having said that, I think we should always weigh new proposals against
Michi Henning's advice about creating standards: 'the ability to say
"no" is usually far more important than the ability to say "yes."' By
offering two encodings for the same information we have added no power
to CF, but we have made the task of writing interpreters harder. We
have opened the door wider to unanticipated contradictions; we have
closed a door on alternative interpretations of a syntax that we might
be wishing for in the future.
[Note that we also, don't allow an "axis" attribute to be attached to
a scalar coordinate variable, and I also don't remember why we did this.]
There are valid reasons to argue that coordinate variables should be
self-identifying -- i.e. that their interpretation does not rely on the
presence of some other variable bearing a "coordinates" attribute that
points to them. This requirement would allow the coordinates in a file
to be identified in a single pass; it would ensure that the
interpretation of the file is not subtly corrupted by the removal of a
dependent variable. Traditional netCDF coordinate variables (dimension
name = variable name) pass this test. All forms of auxiliary coordinate
variable (including scalar coordinates) pass this test _only if_ they
possess an "axis" attribute. (The units= attribute is not a reliable
indicator of an independent coordinate.) So from my pov scalar
variables should be _required_ to include an "axis" attribute. Ditto
for other auxiliary coordinate variables. (Jonathan Gregory and I
recently concluded that there are ambiguities in the interpretation of
the "axis" attribute as currently documented.)
- Steve
=======================================
Best regards,
Karl
_______________________________________________
CF-metadata mailing list
CF-metadata@cgd.ucar.edu
http://mailman.cgd.ucar.edu/mailman/listinfo/cf-metadata
_______________________________________________
CF-metadata mailing list
CF-metadata@cgd.ucar.edu
http://mailman.cgd.ucar.edu/mailman/listinfo/cf-metadata
