[CF-metadata] ESRI implementation of ellipsoidal-earth

From: Bentley, Philip <philip.bentley>
Date: Wed, 2 Feb 2011 16:02:51 -0000

Hi John,

> Apparently ESRI is willing to add support for CF 1.5
> grid_mapping attributes for ellipsoidal-earth/geodetic-datum
> definitions in the Grid Mappings and Projections specification.

That's good to hear. By coincidence I'm meeting up with a colleague
tomorrow to check out the new netCDF-viewing capabilities within ESRI's
latest ArcGIS 10 release.

> They are looking for sample data, especially using
> ellipsoidal parameters
> (semi_major_axis/semi_minor_axis/inverse_flattening). Does
> anyone have any?

I don't have any netCDF files immediately to hand which contain those
attributes, but what I'll try to do is dig out a suitable file and add
some CRS attributes using NCO's ncatted command. Of course, someone at
ESRI could do this themselves on their own sample files using random-ish
attribute values - presumably, for software testing purposes, they don't
need to be absolutely correct.

>
> Also, they are wondering what to use for the default
> ellipsoid when these are not specified. It seems to me the
> obvious candidates are 1) spherical earth with a standard
> radius, or 2) WGS84 ellipsoid.

Apropos (1), unfortunately there isn't really a standard radius value
for a spherical Earth. A value of 6371229.0 m is used for a number of
'standard' Met Office climate models, but I imagine that a large variety
of values is used in different application domains. To illustrate the
point, there are some 15 different ways of determining the spherical
radius using the 1924 International Spheroid as a basis (see Maling,
Coordinate Systems and Map Projections).

Apropos (2), I'd be inclined to advise against assuming a WGS84 as a
default ellipsoid since that implies a statement about the
georeferential accuracy of a dataset which may well not hold true. In
the observations world WGS84 is probably a reasonably safe bet, though
by no means a certainty (esp. not for pre-GPS data). I'm not aware of
any (global) climate models that use anything other than a spherical
earth, but my knowledge is very limited in this respect so I'd be
interested to hear otherwise.

If a 'datum-less' dataset is used in isolation (i.e. without reference
to other datasets) then the choice of earth figure is not so crucial. In
the case where a datum-less dataset is used with other 'datum-ful'
datasets then one would need to choose one of the latter as a reference
source. And one imagines that this is likely to be context specific.

Regards,
Phil
Received on Wed Feb 02 2011 - 09:02:51 GMT