An ongoing project I'm working on is to periodically fly and map a road construction project and compare sequential ground surface clouds to track changes/etc.?ÿ This is being done with photogrammetry only, no LIDAR, and nadir-only photos with about 80% overlap.?ÿ I flew last month and established a surface that checked well with ground control points.?ÿ I flew for the second time last week, but when comparing the surfaces there is one area that isn't matching well, and I know that no work has been done there yet.?ÿ On the attached photo it is the area in orange (orange is about 2 feet of cloud-to-mesh distance on the current scale).?ÿ The land adjacent to the road in that area is open dirt/wetlands.?ÿ I have drawn the approximate locations of my current ground control with red boxes.
Would adding ground control adjacent to the orange area help with the photo alignment and surface generation for future flights??ÿ If not, what other methods might help with that?
?ÿ
Are you flying RTK or standard photo? If standard photo, a few more GCPs would help. Is this all one flight or multiple? There is a big gap through the middle that you would need some temporary GCPs to tighten up your data.
Are you flying RTK or standard photo? If standard photo, a few more GCPs would help. Is this all one flight or multiple? There is a big gap through the middle that you would need some temporary GCPs to tighten up your data.
I'm flying RTK. The entire area pictured is one flight, but this is only about 1/4 of the full job. The remaining job area is covered by three more flights. Would temporary GCPs in the gap help with correcting the alignment, or would the alignment remain the same but my RMSE for my GCPs would get worse?
Bump, to the top, for further beneficial discussion. Good topic.
Are you flying RTK with a base or are you dialed in to a network? When I lose network connection during a flight, the first thing to go is the vertical. But that's always been LiDAR for me. I fix it by downloading a block of base data from the network provider and reprocessing.
Also, short of setting more GCP's, have you done any ground truthing in that area to see which flight is actually wrong? Two feet is quite a bit and it wouldn't take long to take a few shots to find it.
Thanks for the bump the and responses. I'm using network RTK. It's unlikely that I lost a connection at any point in that flight as there is excellent cell coverage there. We did do ground truthing, and the first flight, being used as the reference surface for the photo I posted, matched it very well. The biggest change to happen between flights in that area is that most of the trees were removed, which I suspect is contributing to a lack of definable tie-points for the most recent dataset. My current plan is to add two GCPs in that area before the next flight and see what difference that makes.
Problem resolved. Thanks to information I learned from BC.surveyor's recent post and his video tutorials I discovered that I wasn't not utilizing the full accuracy of the RTK data. With changes made to my settings while processing I was able to get consistent results across the entire project without adding any additional GCPs.
This site R O C K S !
