[CF-metadata] CF-metadata Digest, Vol 99, Issue 8

From: Comiskey, Glenn <g.comiskey>
Date: Wed, 20 Jul 2011 10:43:19 +0100

Thanks for the reply John.

Just to confirm, the vertical coordinate is described by a number of
other metadata attributes such as axis, standard_name,
_CoordinateAxisType etc. and therefore wouldn't expect any problems with
applications discovering the vertical coordinate.

Regards,

Glenn

------------------------------

Message: 4
Date: Tue, 19 Jul 2011 19:13:45 -0600
From: John Caron <caron at unidata.ucar.edu>
Subject: Re: [CF-metadata] COARDS - "positive" attribute
To: cf-metadata at cgd.ucar.edu
Message-ID: <4E262BC9.50805 at unidata.ucar.edu>
Content-Type: text/plain; charset="iso-8859-1"; Format="flowed"

On 7/18/2011 3:42 AM, Comiskey, Glenn wrote:
> Hi,
> The COARDS covention mandates the use of the "positive" attribute for
> a vertical axis that is not a unit of pressure, i.e. if using metres
> to denote altitude. It further qualifies use of the "positive"
> attribute as a means of discovering vertical coordinate variables,
> i.e. the convention states "A vertical coordinate variable will be
> identifiable by: units of pressure; or the presence of the positive
> attribute..." Does this, therefore, restrict use of the "positive"
> attribute to vertical coordinate variables only? NOTE: CF convention
> (v1.5, Appendix A: Attributes, page 50) reports use of the "positive"
> attribute as relating to coordinate data only.
> I ask as I have a number of datasets in which the vertical coordinate
> variable is present, named "altitude" albeit it represents sea level
> with a value of 0 (zero), unit value of metres and has a declared
> "positive" attribute of "up", but also contains tidal data reporting
> the height of the water column above a pressure gauge and an attribute

> is wanted to be added to denote that increasing values represent a
> greater depth of water, i.e. "positive" is down. Would use of the
> "positive" attribute for data variables interfere with the discovery
> of vertical coordinates as defined by COARDS?

Hi Glenn:

The CDM library wont have trouble with that attribute on data variables,
but im not sure about other libraries.

Really theres no good reason not to unambiguously label coordinates,
rather than relying on units or attributes like "positive". Why risk
having a library misunderstand? At this point, the best existing
candidate for CF would be the "axis" attribute. However, people are
usually worried about legacy applications, and theres probably not a
definitive solution for those.

John
Received on Wed Jul 20 2011 - 03:43:19 BST