The Z coordinate (altitude) can be given with respect to an ellipsoid (ellipsoidal height) or with respect to a geoid (orthometric or geoid height). Both geoid and ellipsoid are mathematically defined surfaces; the ellipsoid (left) is relatively simple, whereas the geoid (right) is much more complex:
More information about the difference between both altitudes can be found in this article: Orthometric vs ellipsoidal height
In order to convert from an ellipsoidal height into geoid or orthometric height, there are three different approaches that could be used depending on the required accuracy:
- Constant height shift
Recommended for small areas where geoid and ellipsoid can be considered parallel in a small portion of terrain. In case of a large area or significant variations between the ellipsoid and geoid, this approach will introduce inaccuracies.
In order to define the height shift in Pix4D, the Geoid Height Above the Ellipsoid function can be used:
More information can be found here: When to use the Geoid Height Above the Ellipsoid function?
- Global geoid model
- Local geoid model
There are four different cases.
- Image Geolocation:
- Standard image geolocation. Most of the drones in the market provide image geolocation which has an accuracy of a few meters.
- Precise image geolocation (RTK/PPK drones). They can reach up to a few cms accuracy.
- GCPs (Ground Control Points). They are usually measured with high accuracy (2-5 cm)
- Case 1. Standard image geolocation and no GCPs
- Case 2. Standard image geolocation and GCPs
As the GCPs are usually measured with a GPS device, the measured height will be ellipsoidal. The recommendation is to convert the heights before they are imported in Pix4D by using a precise local geoid model. Some GPS devices already have the geoid models preloaded, and the transformation to the orthometric (geoid) heights is done automatically.
If the local geoid model is not available, the GCPs can be imported into Pix4D, and a constant shift or global geoid should be applied. If this is the case, the result in Z will not be precise as any of these conversions are not precise either, so the expected accuracy will be similar to the previous case.
- Case 3. Precise image geolocation and GCPs
- Case 4. Precise image geolocation and no GCPs
Convert image geolocation by:
a) applying a constant shift or
b) using a global geoid model
Same as image geolocation (a few meters)
Convert GCPs by:
a) applying a local geoid model beforehand or
b) applying a constant shift or
c) using a global model
a) same as GCP accuracy (2-3 cms usually)
b) (*)
c) 2-3 m
Convert GCPs and image geolocation (optionally) by:
a) applying a local geoid model beforehand or
b) applying a constant shift or
c) using a global model
a) same as GCP accuracy (2-3 cms usually)
b) (*)
c) 2-3 m
Convert image geolocation by:
a) applying a local geoid model beforehand or
b) applying a constant shift or
c) using a global model
a) same as image geolocation accuracy (2-10 cms usually)
b) (*)
c) 2-3 m
(*) Depends on how large the project is and where is located
This article refers to vertical coordinate systems. It is also recommended to read the following Community post about 2D coordinate systems transformations: Horizontal grid corrections and transformations.