Accuracy without GCPs

I have used GCPs before, but have a request to map a 143 acre area and they want me to map it without GCPs. What kind of elevation error are we looking at? Using a Phantom 4 Pro at 150’ agl 77/72 overlap 3000 images. I have read most of the material but am still a bit confused on what to expect.

2 Likes

Most of ours have had a 1m shift in the images from the geotag to the photogrammetry alignment. Ground accuracy on Z is usually within +/- 2-3ft. This means that on one end of the project you could be 2ft higher than actual and the other side could be 2ft lower so a 4ft variance. I have seen it get better but that was in total overcast and no/low winds.

Without GCPs, your results will be relative to your takeoff level.
There will be no way to link the relative altitude to your local datum

I recommend this paper, which discusses the accuracy of Phantom 4 RTK with and without GCPs:

Hope this helps!

1 Like

The elevation of a map created without GCP’s can be adjusted in DroneDeploy using these instructions:

https://www.google.com/url?sa=t&source=web&rct=j&url=https://support.dronedeploy.com/docs/calibrate-elevation-kaitlin%23:~:text%3DCalibrate%20Elevation%20Data%20within%20DroneDeploy%26text%3DDrag%20the%20calibration%20icon%20to,which%20you%20know%20the%20elevation.&ved=2ahUKEwjh4pKmn-rxAhXEAp0JHQxzC6EQFjABegQIAxAF&usg=AOvVaw2nqaxMqGMEP2Yn91X1j3sz

1 Like

Thanks! So we are talking a possible error of 2-3 feet give or take. IF they want data for understanding the elevation for eventual build out but do not want to spend the hi price of having a survey crew do it, would there be a percentage we could apply? I know this is not going to be exact. But say 10% one way or the other cut/fill volume. Just curious.

As an example I processed a map twice. Once with GCP’s and once without. I used the single point elevation calibration to force the non-GCP map to an elevation as established by the GCP. I then went and checked two other GCP’s and they were off 2-3ft. Unfortunately with the lack of accuracy the results won’t be reproducible.

As an alternative if you do not have a way to set GCP’s you can setup what we like to call Rough GCP’s (RGCP) where we basically fly the first map and then create points based on the values. We then use those points as a GCP file moving forward. Error gets better but you are still around 1-2ft vertically. Horizontal should be good unless you are doing a large site, larger than 40-50 acres. The most beneficial thing this does is force the maps to align near perfectly horizontally whereas they can shift up to 3 meters with any kind of control. Problem is that you have to be on at least a Business plan and you would have to pay per map. This is a shortcoming of DD but is where we are at right now.

1 Like

I posted this question in another response thinking I was responding this post, but what do you think the relative accuracy would be? Vertical - Horizontal?
Not in relation to the rest of the world i.e. absolute accuracy. I am not looking for maps to align just merely give me fairly accurate elevation/contour maps.

Just replied to the Sub-M Map thread.