Merging AutoCAD drawings into GIS(having cooridinates problems) 1

WH (Visitor) Sep 17, 2001
I am using AutoCAD map 4.0. I am trying to export CAD files into GIS. AutoCAD map will allow you to export CAD files as shapefiles. The problem I am having is the coordinate system. The GIS database was done in TN. State coordinates. The drawings (property boundaries) were done by surveyors. The surveyors didn’t use the TN State coordinate system on the drawings. I know the drawings are not going to overlay perfect with the polygons in the GIS database. How would you get those drawings into the required state coordinates. Somehow I have to assign the property corners to a certain location in the GIS database. I have posted this message on a couple of forums. They keep saying to have the surveyors put it into the required cooridinates. That is not possible for my case. These are old drawings. We plan to scan and vectoize them. Any help would be greatly appreciated.

 

[carlorom]

I am not familiar with your co-ordinate system but if you stste that these are old drawings on a known system a surveyor familiar with the geographical projection used should be able to convert it adequately. So please try your local surveyor or failing that the nearest University teaching surveying. There are also wyas of transforming co-ordinates of drawings of scanned values or from known values. Briefly you must have a well distributed set of common points on both systems. These can then be put through a co-ordinate transformation system with redundancies this will have to be solved by least squares adjustment to minimise errors. A Helmert's Transformation is a very usefull tool in the application you seem to indicate. There is quite alot of literature on this in the average surveying or cartographic textbooks. If you need further information let me know also whic country etc and I can then possibly give you basic formulae or specific texts.
Are you converting plan (ie. Y_X coordinates or is the conversion to be in 3 dimensions involving heights also?)

 

[bundle]

hi...
if you have min 3 or more known points on drawing(coordinates should be known) you can transform that drawing very easily. but you need those points as base points...

those points might be corners of buildngs or better ground control points(GCP) on drawing...

 

[carlorom]

Hi

Your suggestion of three or more known common points is a very simplistic way of doing this and also VERY DANGEROUS as it depends on the accuracy of the common points you are using and other factors. I would only use this if I had no other way or if I needed it for a quick inaccurate solution.

Co-ordinate transformations from systems in GIS are often used indiscriminately thus ending in distortions on maps and then they end up blaming the survey!

The best general rule to apply is to use the BASE \ ORIGINAL data of both sources one wishes to compare or merge. Then analyse common points and obtain an idea errors or accuracy.

Regards

 

[AllenS]

If we are talking about merging survey data into a GIS shape file the survey data will probably be at least an order of magnitude above the accuracy of the GIS, possibly more if the GIS polygons were created form old taxmaps as the often are. So there is probably no reason to go father that a very rough rotation based on a few points along the perimeter.

luck

 

[bundle]

carlorom,

however, it depends on accuracy, as I know these transformations are global and reliable! (affine and projective transformation). I couldnt understand that why you called very dangerous?

and if you use ground control points for transformations ,your result will be accurate. because these points were located with geodetic or photogrammetric ways and must be accurate. as you know maps are produced according to these points.

Regards

 

[carlorom]

Sorry I did not mean to mislead you with the dangerous bit. The transformations as mathematical tools are indeed very reliable and have proven their worth over and over. The danger I refer to is what data is used for the actual transformation. Without lapsing into tedium of error theory I simply wish to warn against using say only three points for starters. This does not give enough redundancies thus helping to establish Errors/accuracies. Most of these transformations use a Least Squres analysis thus minimising errors and yielding a result with errors/uncertainties. See elsewhere in this forum where i have referred to this.

The other aspect is simply to "compare apples with apples" ie. yes ground control points as directly derived or obtained from survey are properly identified on the photo/map, and then used in the transformation.

I just wished to draw your attention to distortions which can occur due to aforementioned methods. Please let me know if you were succesfull in the application you are using this method for? Of course you are welcome to ask anything more you wish to know.

Good luck