Dear Shawn and Phil
This issue has indeed come up before more than once, but not been resolved.
It would help me/us to understand it if you could describe the required
information in more detail and what it is used for. Is a "datum" such as WGS84
a definition of a geometrical surface (ellipsoid) that is used to approximate
the geoid (a surface of constant geopotential)? If so, then it is relevant
both for making more precise what we mean by quantities which are measured with
respect to the geoid, such as sea surface height, and as input to conversions
between lat-lon and map projection coordinates (the latter being why you
raised it). When it has been discussed before, the point has been made that
this information doesn't exist for most models e.g. global climate models,
which treat the Earth as a sphere, so it would be optional (like many other
features of CF).
> For example, should the value of this proposed new attribute be the
> numeric identifier assigned by the dictionary maintainer (e.g. 6326
> for WGS 84), the well-known text identifier (e.g. "World Geodetic
> System 1984"), or either or both of these?
The self-describing principle of CF would indicate that we would use a text
identifier. It could have a numeric component.
It would be good to make use of a dictionary maintained by someone else, like
EPSG, but this depends on their dictionary having exactly the right sort of
contents for us and available in an appropriate format. We have not managed to
do that in other similar cases. Instead we have adapted other sources of
information and imported them into CF. This introduces issues of maintenance,
but on the other hand it makes CF more robust.
Possibly this discussion is related to the one raised by Bert Jagers in the
autumn (see
http://cf-pcmdi.llnl.gov/trac/wiki/MultipleProjections).
Best wishes
Jonathan
Received on Thu Feb 08 2007 - 07:24:03 GMT
