2009/10/7 Andy Robinson (blackadder-lists) <[email protected]> > Christopher, > > Useful sanity check, thanks. > > Looking at the the OSTN02 FAQ it states: > > "While it is true that two adjacent OSGB36 triangulation stations are > usually in agreement with each other to within a few centimetres, if we > compare two triangulation stations 10 km apart by measuring a precise GPS > vector, we will typically find an error of several decimetres in their > relative archive coordinates. If the two stations are 100 km apart, the > relative error between them might be four metres. If the stations are at > opposite ends of the country, we will find a twenty metre relative error > between the two, based on their archive OSGB36 coordinates." > > This would suggest that the error in a position of a 10km x 10km 1:25k map > sheet is in the order of centimetres rather than meters. And furthermore > the > OS PAI process found errors only of a less than 5m maximum I think. So I > assume that although these errors are important for precise surveying they > probably don't impact OSM that much? Obviously the error would be > exaggerated over a bigger sheet such as the 7th series but presumably > provided sufficient control points are used the error isn't going to be > significant. > > Thoughts? > > Cheers > > Andy > > Sounds like the PAI can't be producing offsets as much as the 80m reported by others then.
Does sound like something is not right with the transformation if we are getting errors as large as that. If applying the correct 7 parameters (3 shift, 3 rotations and one scale) in a Helmert datum transformation you would expect errors ~5m. Offsets of 80m or more could be created if not using the correct datum shift parameters. Our chums at the OS have a very informative document - http://www.ordnancesurvey.co.uk/oswebsite/gps/docs/A_Guide_to_Coordinate_Systems_in_Great_Britain.pdf Or perhaps I am barking up completely the wrong tree, and should slope off back to my paleotard lair. Did not expect to find myself talking about transformations, projections and datums on talk-gb -- Christopher Osborne www.itoworld.com
_______________________________________________ Talk-GB mailing list [email protected] http://lists.openstreetmap.org/listinfo/talk-gb
