[CF-metadata] Same parameter, different meaning (pressure)

From: Jonathan Gregory <j.m.gregory>
Date: Fri, 29 Aug 2008 08:04:53 +0100

Dear Steve

> If the quantity in question is sea water pressure then
> it should be called "sea_water_pressure". If it is measured at 200m
> then it should have a typical CF-encoded Z coordinate axis to indicate
> that it is at 200m depth.

I agree with that. In the example, it was 200 m height above the sea floor,
which can also be recorded as a vertical coordinate.

However, I think that if it's on the sea-floor we should put that in the
standard name. This is the approach we have use for a lot of existing standard
names. A conceptual reason for this is that quantities on particular physically
important surfaces often do have somewhat different physical uses or
interpretations from quantities on surfaces defined by coordinate variables
in the body of a fluid. A practical reason is that it is rather arbitrary how
to encode the surface using a coordinate variable. The sea-floor could be
recorded as 0 m below sea floor, or 0 m above sea floor, or sigma=0, or H m
below the sea surface, etc. This wide choice would make it inconvenient for
software or humans to identify the quantity as being on the sea-floor. It
seems better to indicate this explicitly, since that is what we want to say.

An argument could be made that the named surface should instead be encoded
in a separate attribute, taking the place of a vertical coordinate variable.
I agree that would work, but I don't think it is preferable because it would
be more complicated for humans or software, which would have to look in two
attributes rather than one. I don't see the advantage in it. As I say, we
have followed the approach of encoding named surfaces in standard names right
since the beginning of CF, and we have a lot of them now in the stdname table.

> Regarding the question of tethered sensors, my initial reaction was
> that this should not be encoded in the standard name.

I agree. If the tethered sensors are actually moving, so that x y z are
functions of t, I agree also that this information could be provided like
a trajectory, as auxiliary coordinate variables. That does not require a
new standard name.

Best wishes

Jonathan
Received on Fri Aug 29 2008 - 01:04:53 BST