# What is the difference between theoretical error and the error to GCP initial position?

The Theoretical Error S(X,Y,Z)  and Error to GCP Initial Position are computed after Step1 and they give a good estimation of the accuracy of the block.

The Theoretical Error S(X,Y,Z)  informs us about the relative accuracy whereas the Error to GCP Initial Position informs us about the absolute accuracy.

Note: A GCP can have high Error to GCP Initial Position and small Theoretical Error S(X,Y,Z). If that is the case, the point has been marked correctly on a sufficient number of images, so that the relative accuracy is correct, but the GCP could be located far away from what it should be which means that the absolute accuracy is not correct.

A better explanation can be found below.

Theoretical error S(X,Y,Z)

The Theoretical Error S(X,Y,Z)  defines the three error ellipsoid axes of a GCP and it does not take at all into account what is the Initial (input) position of the GCP. It can be seen in the rayCloud when clicking on a point. If the Theoretical Error S(X,Y,Z) is low, it means that the error ellipsoid for that particular GCP is small so that there is a good relative accuracy.

If the Theoretical Error S(X,Y,Z) is high, it means that the error ellipsoid for that particular GCP is big so that the relative accuracy is not so good. That can happen when:

• The GCP was marked in very few images

The image below shows a point that has been marked on only two images with a small baseline (the distance between the center of the images). As a result, there is a large Theoretical Error in the direction of the rays: When the point is marked on a new image, the Theoretical Error gets smaller: • The GCP was wrongly marked in some images

The example below shows a point that has been marked on only 4 images with a small baseline and on one of them (2) it is marked erroneously: When the point is correctly marked on more images, the following result is obtained: Note:
• The error in Z is larger than the error in X and Y for nadir aerial images.
• If the error is larger than three times the GSD, the estimation of the 3D point needs to be improved by manually refining the point on the images using the rayCloud Editor.

Error to GCP Initial Position

The Error to GCP Initial Position gives the difference between the initial and computed GCP position in each direction (X,Y,Z). If the error is low, it means that the project has a good absolute accuracy whereas if it is high, it indicates problems with the georeference. Information: There is a blog post which can help to understand the difference between absolute and relative accuracy: What is accuracy in an aerial mapping project?