[geotk] Spherical-earth CRS

[Martin Desruisseaux]

Hello Jon

Le 30/03/11 18:15, Jon Blower a écrit :
> I just noticed DefaultGeographicCRS.SPHERE in Geotoolkit-3.17.  I was slightly surprised to find that I could create a MathTransform between this CRS and DefaultGeographicCRS.WGS84 without Geotoolkit complaining about the datum shift (I didn't specify lenient=true).

I just tried a CRS.findMathTransform(DefaultGeographicCRS.SPHERE, 
DefaultGeographicCRS.WGS84) and I get an OperationNotFoundException with 
geotk-SNAPSHOT. Maybe the requirement for Bursa-Wolf parameters were disabled on 
a library-wide basis on your machine? This check can be disabled by a call to 
Hints.setputSystemDefault(Hints.LENIENT_DATUM_SHIFT, Boolean.TRUE).

> Does this mean that I can safely use this transform to take spherical-earth data into WGS84 coordinates?  This might be useful for us since numerical model data often uses a spherical-earth approximation.
With lenient datum shift, I think it can be used as an approximative transform. 
But there is many spheres. The DefaultGeographicCRS.SPHERE one as a radius of 
6371000 metres. The /GRS 1980 Authalic Sphere/ (i.e. a sphere with the same 
surface than the GRS 1980 ellipsoid) has a radius of 6371007 metres. I guess 
there is also a /WGS84 Authalic Sphere/, but I don't know its radius.

Do you know what are the radius of the sphere used in the numerical model?

     Regards,

         Martin