[postgis-users] Spatial reference systems

[Frank Warmerdam]

Luis Mota wrote:
> I noticed, however, that there were differences between the definition 
> found at OpenGIS and the definition in the PostGIStable ... Please look 
> at it:
> 
>  From the spatial_ref_system table:
> 20790 | EPSG      |     20790 | 
> PROJCS["Portuguese_National_Grid",GEOGCS["GCS_Lisbon_Lisbon",DATUM["D_Lisbon",SPHEROID["International_1924",6378388,297]],
 >PRIMEM["Lisbon",-9.131906111111112],UNIT["Degree",0.017453292519943295]],PROJECTION["Transverse_Mercator"],PARAMETER["False_Easting",200000],PARAMETER["False_Northing",300000],PARAMETER["Central_Meridian",1],PARAMETER["Scale_Factor",1],PARAMETER["Latitude_Of_Origin",39.66666666666666],UNIT["Meter",1]]
> 
> 
>  From http://www.opengis.org/techno/interop/EPSG2WKT.TXT:
> 
> EPSG=20790
> PROJCS["Lisbon (Lisbon)/Portuguese National Grid",GEOGCS["Lisbon 
> (Lisbon)",DATUM["Lisbon",SPHEROID["International 
> 1924",6378388,297]],PRIMEM["Lisbon",-9.13190611111111],UNIT["degree",0.0174532925199433]],
> PROJECTION["Transverse_Mercator"],PARAMETER["latitude_of_origin",39.6666666666667],PARAMETER["central_meridian",1],PARAMETER["scale_factor",1],PARAMETER["false_easting",200000],PARAMETER["false_northing",300000],UNIT["metre",1]] 
> 
> 
> As you can see, the names of  spheroids, data and projections are 
> different...
> 
> This doesn't seem to be too bad since the numerical data is identical, 
> but to me this is a problem: I need some kind of sure nomenclature in 
> order to refer to reference systems...
> 
> What should I use for this? Is there any standard? Should I use this 
> EPSG number? What about the other ref-systems whose auth_name is ESRI? 
> They are, at least some of them, different from the ones from EPSG...

Luis,

The 20790 from PostGIS seems to be based on the ESRI naming convensions for
GCS, Datum and so forth.  This can be quite hard to relate to anything else
since ESRI "morphs" the EPSG names slightly but does not use the new
authority fields to give a more rigerous identity to the coordinate system
components.

On the other hand, the work done for EPSG2WKT.TXT on the open gis web site
wasn't intended to be absolute, but rather an example.  However, I would
strong like to encourage the PostGIS guys to build off the EPSG2WKT instead
of the ESRI definitions.

In the end, depending on the textual names as a unique id is just not safe.
You will only really have an interchangable coordinate system if you can
refer it back to the EPSG definition (especially important for components
like Datum that cannot be easily defined mathematically).

PS. The ref-systems with ESRI as the authname are coordinate system
provided by ESRI but not (apparently) in EPSG.  They can be useful but are
generally "less exchangable" than EPSG ones.

PPS. At least there seems to be agreement on the PROJECTION[] names.
Unfortunately, the specifications don't peg down exactly what PARAMETER[]
values go with what projections.  An attempt at this is made at:

   http://www.remotesensing.org/geotiff/proj_list

Best regards,
-- 
---------------------------------------+--------------------------------------
I set the clouds in motion - turn up   | Frank Warmerdam, warmerdam at pobox.com
light and sound - activate the windows | http://pobox.com/~warmerdam
and watch the world go round - Rush    | Geospatial Programmer for Rent