[FWTools] incorrect data for EPSG:3120 in FWTools2.4.2 and 2.4.6

[Frank Warmerdam]

2010/1/27 Damian Krysian <dkrysian at tmce.pl>:
> Hello, I'm sorry for my English....
> Problem occurs while converting data between EPSG:3120 and EPSG:4326 -
> output geometry is moved ~100m from correct position.
> I must use correct datafiles from FWTools2.2.8.
>
> And another problem:
> when converting data between EPSG:3120 and EPSG:2178 ogr2ogr always return
> incorrect position, but when converting EPSG:3120 --> EPSG:4326 and then
> EPSG:4326 --> EPSG:2178 position is correct! anybody knows why?

Damian,

I suspect you are seeing a difference in behavior because
PROJ 4.6.0 no longer attempts to do a datum shift for
coordinate systems without datum shift parameters.
In the past if there was no +towgs84 parameter and
no +nadgrids parameter PROJ would still apply a change
of ellipsoid of the ellipsoids difference.  In PROJ 4.6.0
this behavior changed.

EPSG:3120 seems to expand to:

+proj=sterea +lat_0=50.625 +lon_0=21.08333333333333 +k=0.9998
+x_0=4637000 +y_0=5467000 +ellps=krass +units=m +no_defs

Note that it uses a different ellipsoid than EPSG:4326 (which
is WGS84), for 3120 the ellipse is Krassovsky, 1942.

It would be best if you could establish an appropriate
towgs84 value for EPSG:3120 and substitute it in the
appropriate files.  This might include the PROJ.4 epsg
init file, and the GDAL gcs.csv file.

http://trac.osgeo.org/proj/wiki/FAQ#WhydoIgetdifferentresultswith4.5.0and4.6.0

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