I am trying to orthorectify an old aerial photo that has control points. When I run an output and then validate it I find some offset or distortion.

What should I do to fix this?

Does anyone have an example of pre and post orthorectification to help me?

I also seek any additional sources of information on this process that anyone can share.


I ran into the same issues as you while georeferencing images. Sadly there were no real helpful books/texts out there which helped to get information on the offset/distortion. I did the georeferencing using ArcGIS 10.2.2 at that time and when I set all my control points I used aerial images/Google Earth/older aerial images from different WMTS servers to actually reference my image. By asking other colleagues and talking to different people from different universities, they told me that there area always errors and that nothing is perfect. When showing your methodology and stating clearly that there is an error (and maybe try to quantify it, if possible) that you did then enough to explicitly show your "proof of exactness" (that was the exact wording they gave me).

  • Thanks Cliff, I am aware of that as well and you have a good point. I use Agisoft Photoscan along with the Google Earth Pro to get the latitude and longitude and it does pretty good job but the altitude isn't good...so you would to deal with it at the end of the steps in PhotoScan.. This is a good if you have more than 2 overlaps . ArcMap or ERDAS is good enough as well only if you need to do just one imagery. – PROBERT May 8 '19 at 0:41
  • Out of my experience with R (I am not a pro in it) I had the feeling that the results I did were not exactly what they should have been. I ran into projection errors and in the end, my work wasn't that pretty as it should be. The orthorectification I would absolutely do in a GIS and then save it. Doing in R is just a real messing around. – Cliff May 8 '19 at 8:06
  • I don't agree that "stating that there is an error" is enough. Actually what they told you does not add any information: any measurement (and an ortho is a means of measurement) contains errors, which are statistically handled as uncertainty. The magnitude of that uncertainty needs to be assessed in order to give a complete, reliable cartographic product. – RafDouglas May 9 '19 at 15:08
  • @RalfDouglas: Do you know any possibility to calculate the statistical (=significant) error of an georeferenced orthoimage? I always searched for something like this, but never found anything which comes even close to that (regarding significance). – Cliff May 10 '19 at 7:56

A good guide to using QGIS for orthorectifying images and, at the same time, assessing the errors can be found here:

Orthorectification of a large dataset of historical aerial images: procedure and precision assessment in an open source environment

Like any operation which involves measurement and processing, orthorectification implies uncertainties, which are important to be quantified in order to give a valid assessment of the output.

Generally speaking, to improve the accuracy of your orthorectification process, you want to have:

  • more GCPs; this ensures tighter control and better statistical distribution, resulting into lower sigmas and lower average residuals.
  • better GCPs coverage (i.e. GCPs which are more homogeneously distributed)
  • better border coverage (i.e. GCPs at or near the borders of the original image). This avoids areas with overly extrapolated parameters.

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