[CF-metadata] Horizontal Data Coordinate System Definition [long]

[Richard P Signell]

John & Co,

I think that requiring either ellipsoid paramaters OR EPSG code is a good
idea, but I'm not sure we should limit the EPSG codes people can use.
While I agree it would be pretty unreasonable to expect clients to support
*all* EPSG codes, CF is already so complex that clients will probably only
implement selected features. The IDV is an example.

If someone wants to write a client that will access the EPSG database and
implement every projection and transformation in, we shouldn't stop them.
And also it would be sad to tell people that their model using Swiss
Oblique Cylindrical coordinates cannot be described with CF. ;-)

-Rich

From: John Caron <caron at unidata.ucar.edu>
The EPSG databases are pretty big, and it may be unreasonable to expect
clients to implement all of them. I think a happy medium might be to
require either 1) the ellipsoid axis and flattening parameters or 2) a
few very commonly used EPSG codes that we explicitly list, such as
WGS84. Similarly, rather than allow any EPSG code for projections, we
could allow the EPSG codes for just the projections we explicitly define.