Re: [Geotools-devel] Coordinate transformation "boundaries"

[Martin D. <mar...@ge...>]

Andrea Aime a =E9crit :
> I'm wondering if there's any way to get a validity area for a generic=20
> transform,
> that is, a bbox in source coordinates telling me where the transform=20
> works, and
> were it starts failing.

If all the following conditions are meet:

* epsg-hsql, epsg-access or epsg-postgresql is available in the classpath=
;

* The source and target CRS contain their EPSG code in their metadata (it
   should always be the case if the CRS were created from an EPSG factory=
,
   even if this factory was epsg-wkt);

* The "sourceCRS to targetCRS" transform is one of those that are explici=
tly
   defined in the EPSG database (in the "Coordinate_Operation" table);

Then the "area of validity" should be defined as part of the CoordinateOp=
eration metadata. This area=20
of validity can be defined in various way, including string, geographic b=
ounding box, envelope or=20
arbitrary polygon (all this flexibility is provided by the ISO 19115 "Ext=
ent" object). It sound like=20
that I didn't wrote a convenience method for extracting an envelope from =
a CoordinateOperation - we=20
should probably open a JIRA task as a remainder that we should add such c=
onvenience method.

In the main time, there is a convenience method for extracting an envelop=
e or a geographic bounding=20
box from a CRS. So the easier way to check the "area of validity" of a tr=
ansform is probably to invoke:

     Envelope areaOfValidity =3D CRS.getEnvelope(operation.getTargetCRS()=
);

Javadoc:

http://javadoc.geotools.fr/snapshot/org/geotools/referencing/CRS.html#get=
Envelope(org.opengis.referencing.crs.CoordinateReferenceSystem)

Reminder: just in case you wonder what is the difference between an Envel=
ope (from ISO 19107) and a=20
GeographicBoundingBox (from ISO 19115): the Envelope can be in arbitrary =
CRS. The=20
GeographicBoundingBox is garantee to be (longitude,latitude) in something=
 close to WGS84.


All the above work only for CRS or CoordinateOperation explicitly defined=
 in the EPSG database. If=20
we want a general mechanism for computing the area of validity of an unkn=
own coordinate operation,=20
this require some though. There is no such general mechanism right now in=
 Geotools.

	Martin.