Points projected from EPSG:4326 to NAD83 UTM is introducing error due to changes in TOWGS84 parameters. This was observed in the 2.7 EPSG WKT, but I believe it has been there in older and current releases of EPSG hsql as well.
In 2.6.5 (TOWGS84) projecting a point -114.0, 51.0 to EPSG:26911 produced a
point: 710488.8570345972, 5654109.178707749
In 2.7-RC2 (TOWGS84) produced a
point: 710488.3667198857, 5654108.761620307
Independent projections of the same point in ArcGIS produces: 710488.857, 5654109.179
and gdaltransform produces: 710488.85703878 5654109.17860521, which are comparable with what older geotools versions generate.
Not sure how many CRSs this affects.
CodeHaus Comment From: aaime - Time: Wed, 2 Mar 2011 01:56:29 -0600
The official EPSG database provides a long list of methods to do the datum conversion between NAD83 and WGS84. However GeoTools is only able to use one, it has to choose.
Previously the code was picking EPSG::1188 (which is valid for continental north america), however now for some reason we're ending up with EPSG::1252, which is valid for Hawaii instead.
In an ideal world we would be able to load all of them and use them according to the area being projected, but that would require very significant changes to the existing code.... I guess I'll have to find a way to better sort the existing choices in order to get the most popular on top
CodeHaus Comment From: aaime - Time: Sat, 9 Apr 2011 04:55:17 -0500
This one has been fixed one week ago by having the code prefer the TOWGS84 alternative that has the larger area of influence given the same accuracy.
In particular this means continental USA values will be preferred to Hawaii ones
CodeHaus Comment From: aaime - Time: Sat, 28 May 2011 04:19:55 -0500
Mass closing all issues that have been in "resolved" state for more than one month without further comments