[CF-metadata] axis attribute

From: Russ Rew <russ>
Date: Fri, 26 Jan 2007 15:30:46 -0700

Jonathan,

Having finally carved out time to read all 33 messages in this thread,
I agree with your proposal satisfying the requirement to identify the
horizontal coordinate variables.

Ultimately, I think it would be good to satisfy a more general
requirement, to specify the intended meaning of each coordinate
variable, as needed for use in analysis and display software. For
example, we store radar observations that use spatial coordinate axes
corresponding to radial azimuth, radial elevation, and radial distance
from a radar. For a low elevation angle, radial azimuth and radial
distance could be interpreted as horizontal, but I'm not sure how
useful that would be for analysis and display software.

On Sat, Dec 23, 2006 at 10:17:32PM +0000, Jonathan Gregory wrote:
> Dear all
>
> This debate went quiet about a month ago. I'd like to propose the following,
> in the light of what was said then.
>
> Requirement: A way to identify the horizontal coordinate variables.
>
> (NB coordinate variables means 1D, in the Unidata sense, as usual in CF.)
>
> Proposal: It can be done using COARDS rules for units in the case of latitude
> and longitude. In other cases (such as projection coordinates, rotated latitude
> and longitude) the standard name can be used, but that requires recognising
> specific names, and standard names are optional anyway. The axis attribute is
> also optional but if present it can be used to identify horizontal coordinates
> as those having axis="X" or axis="Y". If both X- and Y-axis are identified,
> X-Y-up should define a right-handed coordinate system i.e. rotation from the
> positive X direction to the positive Y direction is anticlockwise if viewed
> from above. The axis attribute is allowed for longitude (X) and latitude (Y)
> coordinate variables. The axis attribute is not allowed for auxiliary
> coordinate variables. Auxiliary coordinate variables can be identified as
> lying in the horizontal surface (e.g. 2D latitude and longitude) by the fact
> of their having dimensions which all belong to 1D horizontal coordinate
> variables.
>
> Change to current standard: Remove the prohibition on axis attribute for
> rotated lat and lon. Introduce the requirement for X-Y-up to be right-handed.
> Add a remark about how aux coord vars can be inferred to be horizontal. Make
> it clear that axis is not allowed on aux coord vars (but that is probably
> already the case, so not a change to the standard).
>
> Cheers
>
> Jonathan
Received on Fri Jan 26 2007 - 15:30:46 GMT