I have spatial data generated in an Archaeological excavation which is in an artificial coordinate system and I want to convert it to a "real" coordinate system. To do this I took real coordinates from already known points (which have artificial coordinates) but I don't know how to proceed now. I am using the latest version of QGIS.


You need to define you own projection system:

  1. Find you x0 and y0 (the origin of you system) based on the points for which you know the real coordinates
  2. In the Coordinate Reference system selector (right click on a layer and Set Layer CRS), use user-defined coordinate-system (use the definition of the CRS you mentioned as "real coordinate" and change your x0 and y0)
| improve this answer | |
  • This would address locating an origin, but wouldn't do much address any transformation/rotation/etc. when converting from the new custom CRS to the desired CRS. Depending on the scale of the excavation, the distances may be too small to matter. – Chris W Feb 25 '15 at 2:43

You do not mention what software you are using. While this solution is ArcGIS based, the principles of the method are the same for any software though the tools to do it and their names may differ or not be available depending on the software.

You have 'artificial' coordinates for all points, and for some of those points you have 'real' coordinates. You first need to determine or know what CRS those real coordinates are in. Once you've established that, you can create a point data set with points at the 'real' coordinate locations. These are known as your control points, and it would be helpful if they have names to match between data sets. There are several ways to do this, including Making XY Event Layer from a table/spreadsheet of coordinates.

If the coordinates you have are in the desired CRS already, great. If not, now would be a good time to project them to the desired CRS. For example, if you captured the coordinates with a GPS they'd likely be in WGS84, and say you want to work and keep things in the UTM zone appropriate to the excavation. You'd use the Project tool to reproject your newly created points to the desired UTM zone, or simply export a new layer of them and choose that UTM zone.

Now you can load in your full point data. It likely won't come in in the right place or scale. You'll use the Spatial Adjustment tool to create what are called displacement links between the points you've created in the correct CRS and their corresponding points in your full point set (basically, 'this point should be here'). When the adjustment is run, it will scale/transform/warp everything to as closely match the two sets of control points as it can.

Note this method may introduce distortions to your original point measurements depending on the accuracy of both your original survey (probably very high) and your control point coordinate collection (possibly rather low with a consumer grade GPS). It also might not be able to form an exact fit. Tsoc's method won't distort the points since it just establishes a link at a single control point (the origin), but it also won't really map them to the surface of the earth or consistent coordinates in your selected CRS. Again, the size of the excavation and your goal for the 'real' coordinates may be significant factors to consider.

| improve this answer | |
  • Dear all, I'm using the last version of Qgis. All the coments are very useful. I can see that there is not a solution that allows good resolution. So better I keep in my artificial system. – Joseba Feb 25 '15 at 12:09
  • I still think that the rotated mercator projection I suggested in the linked question should solve your needs. I did it with QGIS too. – AndreJ Feb 26 '15 at 6:07
  • @Joseba - AndreJ's answer at that other question addresses one shortcoming of Tsoc's answer (mentioned in my comment) by adding an accounting for rotation. Assuming your excavation is laid out on a grid, as long as you have two pairs of real coordinates that have the same artificial x or y coordinates, you can figure out the rotation of that grid. It would then be a simple place and rotate of the points, with no alteration of their relative placement like a spatial adjustment might introduce. – Chris W Feb 26 '15 at 20:22

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.