On Sun, 2004-03-28 at 08:15, Matthias Basler wrote:
> Hi GeoTool members,
> I got some other questions concerning seemingly "inconsistencies" between
> GeoTools (cts-coordtrans) and the EPSG database.
> Firstly, very general: GeoTools tries to follow the OpenGIS specifications. Does
> the EPSG database comply to them too? Does/did GeoTools follow the EPSG
> database structure? Or just ... who complies to whom?
It is my understanding that OGC and EPSG are different entities with
different goals. The EPSG database describes names and parameters for
coordinate systems (projected and geographic) and datums around the
world. This effort started in the 1980's to provide a standard set of
names and parameters (and documentation) for surveyors to use. The
format of the database follows a standard relational layout that
somewhat conforms to our concepts about coordinate systems (datum table,
operation methods table, etc).
The OGC created a specification to describe coordinate systems in the
late 1990's. Some of this work agrees with the EPSG concepts, but there
is not a 1 to 1 relationship and the ogc includes many things that the
epsg does not. If you look at the CTS specification (01-009) you will
see that the list of parameterized transforms (section 10, page 30) is
similar to the list given in the epsg guidance note #7 ("Coordinate
conversions and transformations including formulas"). Note that the
classes the OGC use also conform to our notions about coordinate
systems, making them similar to the epsg database.
I do not know if there are any efforts underway at the epsg to conform
to the ogc standard.
> The first thing is that GeoTools2 is obviously switching to the new OpenGIS
> CoordinateSystem Specification, which distinguishes between Coordinate System
> and Coordinate Reference System. How far is this gone?
> Does this mean that f.e. the Object "CoordinateSystem" gets or will be renamed
> to "CoordinateReferenceSystem" in order to be consistent with the new OpenGIS
Martin mentioned in his other email that he is almost done creating the
new interfaces and will start implementing them in May (at the
> But in practical use I encountered two specific questions?
> The EPSG database CRS object has following strongly simplyfied structure:
> - Geographic CS (=CRS) = Coordinate System + Datum
> - Datum = Ellipsoid + Prime Meridian
> - Coordinate System = Unit + Axis information
> In the contrary to this, GeoTools has following structure
> - Geographic CS (=CRS) = Unit + Axis information + Datum + PrimeMeridian
> - Datum = Ellipsoid (+ Conversion parameters)
> So in GeoTools2 the prime Meridian is separate from the geodetic datum, what I
> think, is strange. Additionally it somewhat complicates the work with the EPSG
> factory. Will this change with the implementation of the new specification?
Personally I find the OGC (geotools2) way more logical. A datum
describes how an ellipsoid fits to the earth. This does not need a prime
> The EPSG database has (beside others) the CRS types
> - geographic 2D
> - geographic 3D
> - compound
> where the latter two are (mostly/always?) used in 3D space.
> In GeoTools I only find the corresponding objects
> - GeographicCoordinateSystem (for 2D geogr. CS as the JavaDoc suggests) and
> - CompoundCoordinateSystem (for 3D geogr. CS).
> How should I map "geographic 3D" CSs as for example code 4979 (WGS84)? Looking
> at the EPSGFactory source code I would guess I should use
> GeographicCoordinateSystem as well. Is this correct?
I do not know and cannot find this crs code in the 6.4 version of the