[gdal-dev] SRS-Transformation(DHDN/GK=>ETRS89/UTM) with towgs84 - Accuracy

[Gregor Fikoczek]

Hi list, 

i am trying to transform some point (in a gml-file) from DHDN/GK3 - srs  to 
ETRS89/UTM - srs in some diferent ways:


some points(gml-File) in start-srs
-----------------------------------------------------------------------------------------------------------
Pointnr.	     |		X[m]	      |		Y[m]
-----------------------------------------------------------------------------------------------------------
5216100002 |	2552401,862    |	5616707,524
5414100047 |	2554778,697    |	5614549,001
5508100055 |	2555906,314    |	5608696,986

some points(gml-File) in target-srs
-----------------------------------------------------------------------------------------------------------
Punktnr.	      | 	East[m]	| 	North[m]
-----------------------------------------------------------------------------------------------------------
5216100002	32340408,033		5617072,487
5414100047	32342695,230		5614819,749
5508100055	32343584,901		5608927,638


1.) using the srs defined in the proj-epsg-File:

ogr2ogr -s_srs EPSG:31466 -t_srs EPSG:25832 -f GML points_etrs_gdal.gml 
points.gml

-----------------------------------------------------------------------------------------------------------
Results 					|		difference
East[m]	     |	North[m]		|	E	    |	   N	    |  distance
-----------------------------------------------------------------------------------------------------------
340411,269  |	5617068,431	|	-3,236  |	4,056   |	5,188
342698,445  |	5614815,708	|	-3,215  |	4,041   |	5,164
343588,142  |	5608923,646	|	-3,241  |	3,992   |	5,142


2.) using a nadgrid (BETA 2007)

ogr2ogr -s_srs "+proj=tmerc +lat_0=0 +lon_0=6 +k=1.000000 +x_0=2500000 +y_0=0 
+ellps=bessel +units=m +nadgrids=./BETA2007.gsb +wktext" -t_srs "+proj=utm 
+ellps=GRS80 +datum=WGS84 +zone=32" -f GML points_BETA2007_gdal.gml 
points.gml

-----------------------------------------------------------------------------------------------------------
Results 					|		difference
East[m]	     |	North[m]		|	E	    |	   N	    |  distance
-----------------------------------------------------------------------------------------------------------
340408,064  |	5617072,481	|	-0,031  |	0,006   |	0,031
342695,246  |	5614819,723	|	-0,016  |	0,026   |	0,031
343584,918  |	5608927,610	|	-0,017  |	0,028   |	0,033

3.) using towgs-option (Transform-Parameter)

ogr2ogr -s_srs "+proj=tmerc +lat_0=0 +lon_0=6 +x_0=2500000 +y_0=0 
+ellps=bessel +units=m 
+towgs84=590.1181,99.6226,340.1890,1.185095,-1.334952,-2.904521,1.00001651559" -t_srs "+proj=utm 
+ellps=GRS80 +datum=WGS84 +zone=32" -f GML points_7P_gdal.gml points.gml

-----------------------------------------------------------------------------------------------------------
Results 					|		difference
East[m]	     |	North[m]		|	E	    |	   N	    |  distance
-----------------------------------------------------------------------------------------------------------
340408,056  |	5617072,821	|	-0,023  |	-0,334  |	0,334
342695,237  |	5614820,058	|	-0,007  |	-0,309  |	0,309
343584,897  |	5608927,933	|	0,004   |	-0,295  |	0,295


Parameter:
dX	+590.1181m
dY	+99.6226 m 	
dZ	+340.1890 m
rX	+1.185095"
rY	-1.334952"
rZ	-2.904521"
dm = 1.00001651559


=========
Accuracy:
=========
1.)  ~ 5m
2.)  ~ 0.033 m
3.)  ~ 0.3	m	<=== Why so bad ???

The accurancy by using the towgs-option is worse then expected, i was 
expecting ~ 0.02 m. Aspacielly the difference in North 
(-0,334, -0,309, -0,295) is remarkable. 

So maybe an algebraic sign is incorrect or what ever, anybody a suggestion?

Best regards, 
Greg



---------------------------------------
WhereGroup GmbH & Co. KG
Siemensstraße 8
53121 Bonn
Germany

Gregor Fikoczek
Email: gregor.fikoczek at wheregroup.com
Fon: +49 (0)228 / 90 90 38 - 25
Fax: +49 (0)228 / 90 90 38 - 11

info at wheregroup.com
www.wheregroup.com
Amtsgericht Bonn, HRA 6788
-------------------------------
Komplementärin:
WhereGroup Verwaltungs GmbH
vertreten durch:
Olaf Knopp, Peter Stamm
---------------------------------------