We captured a landfill back in June and used GCPs. Final result was excellent.
However, we mapped the same area in September and our customer (large surveying company) is saying the elevations are the same on the outer portion of the map but the elevations become “off” when comparing the June DXF vs the Sept DXF.
Specifically we are working with a landfill (upward sloping from outside in). The elevations at the GCPs are correct on both DXF exports. However, the slope is different in parts where there was no dirt work done between each shoot.
Any suggestions from the community on why this is happening? We are using P4P and capturing the area around 200’AGL. We will say this, back in June we shot it at 175’. The reason we changed is because we ran into stitching issues back in June. We eventually got them to work but knew at the time we needed to fly higher the next time we went out.
Hope you are doing well! I just wanted to let you know that I have responded your question internally. As for the accuracy discrepancy between your two maps, we recommend ensuring that you have tied every GCP point to at least 3 images to successfully triangulate the data. Also, remember that, when comparing data, it is important to use the same flight parameters and extended area in order to get similar results.
This is an odd problem if you have thoroughly verified that the GCPs are actually in the previous location (undisturbed if left there) and in the same condition. Remember that with triangulation you have to take into account the other two GCPs it is tying to. Very slight debris or vegetation can cause this to happen as well. The photogrammetry will always capture what it perceives is the most visible object. If you are creating cut/fill scenarios you might think about creating your own DTM using a piece of software that works directly with the surveying or engineering software. We use Carlson Precision 3D.
How are you locating them? GNSS survey gear I am assuming.
You say that the elevations at the GCPs match, but how close are they really?
I actually have a very similar issue! Flew our site in June, it was perfect, lined up exactly with another “control” method we trust (but its a pain and we are trying to replace it with the drone). I flew it again at the end of September and we are having a about a 15% variance in stockpile volumes. Did the backend calculations change at all?
for Chascoadmins Q’s
1: an RTK GPS (Topcon)
2: By dropping the “location” tool on the Dronedepoly portal onto the GCP’s they are about .2 to .3 feet off (higher than) the actual GCP elevation
3: about 1000 ft
Big update here. Our problem was due to our GCP markers. I can’t tell everyone else how important the “Checkerboard” pattern is. Before, the algorithm wasn’t “seeing” our markers at all. We did our best to manually select them, but in the end we ended up having to re-spray them into a checkerboard pattern.
Once we re-flew the area higher with 75 / 75 overlap, the data seems to have been corrected. Now, when we process maps via the GCP engine, we see MUCH more matches than before.
So in summary, be sure your markers are very clear so the engine can find them.
Interesting, I did two sites, both had the issue. One admittedly had a GCP that wasn’t the clearest, but the other site had them all freshly painted and brightened, no obstructions of any kind. I used an X shape though… not a checkerboard one. I wonder if that has anything to do with it… I wouldn’t think so because the flights in June were bang on, with the same GCP’s in the same condition.
We use Topson as well. Just curious as a sidenote, but do you export your GCP point Northing/Easting or Lat/Lon?
I find that the zoom factor available makes it hard to hit the same point or even the desired point, but agree that you should be getting better than 2-3 tenths We should be able to hold the standard tenth tolerance.
I have been learning that 1000’ is about the max distance that I want to have my GCP’s apart, but I have to balance between how long it takes to set them and the rise in chances that one or more are going to be bad. Plus I have to run around and check my “permanent” network before every flight. I have had the most success with 750-1000’ around the perimeter as the permanent network and setting 3 or 4 interior points each flight in active areas that i know won’t survive. Once objects start showing up onsite that should remain I use them on the interior. Watch out for manholes though! I have made the mistake of catching them before they were adjusted to final elevation, lol.