Accuracy of RTK drone with GCP

Could you please share your insights on the accuracy of RTK drones for surveying and mapping applications, specifically in relation to the necessity of Ground Control Points (GCP) for data calibration? In my experience, while conducting flights at an altitude of 100m to monitor morphology changes, I utilized GCP. However, I observed an RMSE of 1.5m in the z-axis, while the x-axis had an RMSE of 0.005m and the y-axis had an RMSE of 0.03m. I would greatly appreciate hearing about your experiences with RTK drones at similar altitudes and your thoughts on whether GCP usage significantly enhances data accuracy in morphology change monitoring.

Hi there

I’ve been flying with Mavic 3E for a few months now. What I’ve observed is that x ja y are pretty much always spot on but in the z-axis there is always some variance. Z-axis is sometimes spot on and sometimes it’s 15cm off. That’s why I’m still using GCP’s but fewer that when I used when I was flying with Phantom 4 Pro.

I use a WingtraOne with PPK. Horizontal is usually 1-2x gsd and vertical is 2-3x gsd. Only put out targets for check points.


Generally, the expected relative accuracy of a correctly reconstructed and precisely georeferenced project is:

  • 1-2 x GSD horizontally (X,Y coordinates).
  • 1-3 x GSD vertically (Z coordinate).

That means that GSD has a great impact on vertical accuracy.

Having 15cm off in the Z axis can be a good or bad achievement, depending if the GSD was 10cm or 1cm.

Also note that the absolute accuracy of the results cannot be higher than the accuracy of image geolocation or GCPs’ accuracy.

In general, when possible, we suggest always using Ground Control Points (GCPs), at least to be used to assess the absolute accuracy.

You can find additional information here: