Post

2 followers Follow
0
Avatar

Ground Elevation data in incorrect

I flew a large sizeable mission (actually 2 missions, because I was trying to avoid flying over a central area that is controlled by a different land owner). It was overcast so the drone flew at only 4.5 mph. It took 2 days and 10 batteries to complete the mission(s). i included all the ground photos the drone took before takeoff in the upload.

I was also receiving a warning that the vision sensors needed calibration, but I doubt that affects elevation data?

Plus I was receiving the dreaded "Over Flight Maximum Radius" error and had to open the DJI Go app in the background to work around this error.

When I reviewed the first map generated, I noticed the elevation data was about 60 feet BELOW the actual elevation at some known locations. (the average elevations is a pretty flat at 248-252 MSL with one hill maybe 30' above that and a river along the edge of the property  that is about 30 feet lower, the tallest building is 388' MSL on the roof.)

I then uploaded GPS Ground Control Points, and used the Manual Geo-referencing tool to calibrate the image. I assumed this would reduce the elevation error. However, that image reads about 528 feet HIGHER than the known elevation locations.

What is the cause of this error? is there anything i can do to correct it?

Howe, Stanley

Please sign in to leave a comment.

3 comments

0
Avatar

It is better to take your data quickly than it is to have it span multiple days. Doing it over this much time will have moving shadows and affect the accuracy of the reconstruction. You can turn off the slowdown in the Settings. 

The vision sensor warning is something you need to handle in the DJI app. It won't affect the data that is collected but can cause the flight to get stopped prematurely. 

The "Over Flight Maximum Radius" warning comes up with you have a value set in the DJI app. We recommend not having a value there or having it set to 10 km or something like that. 

There is some curvature to your model, most likely for the reasons noted above.

The GCP process does not change the shape of the model. It aligns things to the lat and long points and then attempts to minimize the error vertically. 

Taking out curvature is hard to do over such a large area. That is why we don't do it. You need a lot of super accurate GCPs spread fairly evenly throughout the area to do it properly. Most people are not prepared to provide GCP data of the quality needed to do this correction so we don't offer it.

Zane 0 votes
Comment actions Permalink
0
Avatar

In this case you would have been better off separating the area into northern and southern regions and then merging the results. There were be less curvature that way and the results would likely turn out more accurately. 

Zane 0 votes
Comment actions Permalink
0
Avatar

Yes, I have already re-calibrated the sensors (after the flight) just wanted to make sure it wasn't the cause. 

I tried turning the maximum range off in DJI go, but that only bought me one flight, I then set it to the maximum allowed (I think it was 8000M) and that bought me another flight, i then read to just have DJI-go open would solve it, and I tried that, it seamed to work. 

We usually try to fly on bright days, and can fly the site in 2-3 hrs, however due to FAA restrictions on flying over people we are limited to weekend flights and it has rained nearly every weekend this summer. It was overcast with only a couple brief periods of sunshine(thus no shadows). We have now purchased 2 more batteries and a faster charger to reduce the time spent waiting for batteries to charge which eventually caused us to postpone to the 2nd day.

The elevation data was disappointing, and would have been handy if it was reliable, however taking the time to do more GCP's is not justify-able, because the time spent doing that, we could just go shoot far more accurate GPS points on the points of interest.

Thank you for clarifying the cause, next time, we will attempt to fly all it in one day with better conditions and more batteries.

Howe, Stanley 0 votes
Comment actions Permalink