So far, it appears that none of the elevations are accurate without GCP’s. That even includes when the same set of pictures is re uploaded to DD. DD has not released any supporting data for it’s elevation feature that I can find.
We’ll have to wait for elevation results from data which includes GCP’s to see if there’s any benefit from this feature at all.
Here’s a map I flew yesterday to check GCP accuracies. Again, the launch position returned an elevation greater than 0, therefore skewing the elevations of the other points. Once the vertical offset at the launch position is taken into account, however, the map is within it’s advertised accuracy of 1.3m
Details of the mission:
70% and 70% overlap front and side
80m agl (from launch position-- so, 80m-ish)
flight duration: 5minutes 20 seconds
temp: 4ºC, winds calm, bright sunshine
area: 7.29ha
Did you do any X-Y analysis to see if everything meshed on the stitch? What equipment are you using for flying? How did you identify the GCPs in the field?
I am am trying to do the same thing and was curious of your workflow so I can compare when I do the same exercise… Great Job!
@dragonflyAS: I was the surveyor on this site providing layout for the underground facilites (water and sewer piping systems), the dirtwork and a portion of the curbwork layout. I used one of my permanent control points as a GCP for the purpose of this mission (GCP3). The other three GCPs were simply elevation shots to provide a comparison. So- the RTK system is a Topcon HiperV and the survey itself is tied to local benchmarks (real-world coordinates), the UAV is a Matrice 100 with Zenuse X5 camera.
I checked the coordinates for the survey control point against the coordinates provided by Drone Deploy - check this out:
If I’ve done everything correct - this is insane! (note: it HAS to be a coincidence just where I set the notation pointer after zooming in as close as I could on my aerial target…) still, though!!
WOW! That is dead on. As an engineer I could design with that X-Y accuracy and the Z is good enough, provided that I had my key tie-ins and major features surveyed traditionally.
For our application, at the moment anyway, as long as the z is relative to itself within the map, that’s all we need for volume calculations. The x-y accuracy is good to know, though!
I agree, key tie-in points still require boots on the ground. (I’m just saying that to save the surveying profession!)
WAIT!! - I double checked everything - I forgot a k!!! The difference is 0.00434km - so 4.34m. The radius of the earth in the formula is, of course, in kilometers, therefore the result would be kilometers, NOT meters as shown on the table…sorry for the confusion…!
Still within the accuracy of the map:
Is everything accurate, relative to the site? Or maybe a better way to ask is, if a distance in the real world is 10’ is it measuring 10’ on your map? Also if an elevation is +10’ in real world is it +10’ on map? I know global location may shift, but is it dead on relative to itself?
horizontal distance check #1: actual distance is 16.821m (0.461m - 1.5’ error)
horizontal distance check #2: actual distance is 42.728m (0.102m - 0.3’ error)
sorry the photographs are grainy- that’s MicroSoft’s snipping tool…
hope this helps
ANOTHER UPDATE
Here’s me next to my 2013Tacoma Access Cab (208.1inch overall length = 5.285m)
This is DEAD ON! I don’t know what to think.
Looks like the profile of a truck 
Also, some of those slight differences could be based off of which exact pixel the DD “Dot” measures to. Either way this is all great info!
Definitely some good numbers! With an rmse that low, I would say you have solid data.
How were those Distance Checks done? Did you measure after calibration via GCPs or was that just a raw image captured in a run of the mill mission?
Sorry it took so long to get back to you. There were no calibrations after the fact; this is the image as it was returned to me from DD.
Amen to everything you are saying! I couldn’t agree more and be more supportive of this inquiry. I know I am late to the party on this but how have your experiences been since you posted these? I would GREATLY like to know. How are your accuracies been turning out? Elevation, volume, distance, etc.? Thanks so much in advance!
Thats awesome! Thanks for getting back to me. Thats some really accurate data for not using GCPs. I wish elevations were that good. Do you know the accuracy of volume measurements compared to distance? Are they nearly that precise?
Mostly I fly for stockpile and overburden volume calculations at our aggregate pit, without the use of GCPs. For my purposes, I don’t require real-world elevations or coordinates, as long as the piles are relative to itself (if that makes sense). Before the UAV, I would manually walk these piles with my RTK GPS and calculate the volumes that way. Being in my late 40s now, with bad knees, I would take a shot every 10’ or so, averaging roughly 250-300 shots per pile. Once I got the UAV and flew my first mission, I found the accuracy to be within 10% of my ground survey. With as many calculation points as the DroneDeploy mission returns, I would say it’s the more accurate of the two methods. For certain it would include more of the high and low peaks and valleys of the stockpile than I would manually walking (and huffing/puffing) the whole time! I hope that helps.
