[Proj] GGRS87 TO WGS84 PROBLEM(?)

[Oscar van Vlijmen]

From: "Andreas T"
Date: Wed, 6 Dec 2006 23:44:19 +0200
Subject: [Proj] GGRS87 TO WGS84 PROBLEM(?)

> I have trouble trying to reproject a point from greek grid to WGS84.
> ....

> Supposing that these params are official and accepted enough,
> ....
I can't find better ones at the moment, so one might assume they're correct.

> ....
> [CODE]
> // Projections declaration
> $projInObj=ms_newprojectionobj
> ("proj=tmerc,ellps=GRS80,towgs84=-199.87,74.79,246.62,0,0,0,0,units=m,
> lon_0=24,k=0.9996,x_0=500000,no_defs")
> $projOutObj=ms_newprojectionobj(
> "proj=latlong,ellps=WGS84,DATUM=WGS84,no_defs");
> // Shape to be reprojected
> $myshape = ms_newpointobj();
> $myshape->setXY(552000, 4098077);
> ...
> The expected result (according to very reliable software like ArcGIS and
> others...) is:
> lon=24,351076881,lat=37,014812887
> Though! the result that I get from mapscript is:
> lon=24.585568772,lat=37.028433901

>...
> I can't relize any systematic error.
>...

This is a tough problem indeed! I have no clue.
Going from x, y = 552000, 4098077 through Transverse Mercator in the GGRS87
system should produce lat, lon = 37.027448, 24.584634 deg. A geocentric
shift as mentioned should produce lat, lon = 37.030039, 24.586333.
The ArcGIS values seem very unlikely; on the GRS80 globe they are nearly 21
kilometers away from the regular TM values. Datumshifting from GRS80 to
WGS84 cannot result in such large differences; the ellipsoids are nearly the
same and the cartesian shifts are less than 250 meters.