I usually work with data either in WGS84 or ED50 coordinate systems. When I need to perform a conversion from WGS84 to ED50 I always use the common offshore transformation with the following parameters:
If you're using the Create Custom Geographic Transformation tool or the New Transformation dialog in ArcMap, you only need to define it in one direction. In fact, it's very slightly more accurate to define the transformation as-given, because then the software can invert the transformation's rotation matrix (more necessary when the rotation values are relatively large).
If, for some reason, you do need the parameters for the opposite direction, simply change the signs so:
Thank you for your help. In fact that's exactly what I have been doing but what is happening is when I try to convert data from ED50 to WGS84 everything is fine, when I go the opposite way the resulting data has a huge shift. I'm talking about 140 m.
I don't know if I'm missing something else or the problem might be the original data? Any further comments?
I'm having a similar problem but my shift is around 5m. I thought that this wasn't too bad. So this takes me to another discussion, whenever we transform data between 2 different coordinate systems should both FC overlay perfectly or is it normal to have a slight shift because we're using different spheroids?
A transformation is always an approximation, so unless a target layer was created by using the same transformation, you will likely see differences. There will also be differences if positions were re-surveyed--a transformation is unlikely to match the new values. So it's not really that the source/target ellipsoids are different, that's just a part of the whole thing.
Hmmm, I've done this type of thing many times and if both transformation definitions are set up properly, it works fine. You can see small differences if the rotation parameters are large (over 9" or so) because changing the signs is an approximation.
But 140 m sounds like the transformation isn't occurring at all. Some difference in the GCS definitions between what's defined in the transformation and the data or data frame GCS?
This problem has been solved. I tried to make a few transformations using different feature classes and everything went fine. Well, I got these slight rotation differences but nothing as the 140m I had before.
The problem was the original data as I suspected. I've been given a set of data with the wrong coordinate system so with the transformation was getting these wired shifts.