GCP necessary for Roof survey?

[PVSimon]

Hey there!

I am working in small engineering company and we are planning middlesized to largescale pv systems.
For the planning of middel sized rooftop pv systems we´re considering UAV mapping with our phantom 4 pro.
I downloaded the latest trial version of Photoscan and already got the first test results.
The first results with a 100 Image project were really impressive!

I used an high overlap of 95 % for side- and frontlap (dronedeploy app). I flew NADIR at around 74 m. The area was around 4.000 m2.
I did some control measurements afterwards on the ground with a handheld laserdevice and the deviation was between 2-7 cm!
No GCPs were used. Just the onboard p4p GPS.

Since I just tried this with two small flights I am sceptic if I was just lucky.
I read a some posts about accuracy and that it is highly recommended to use GCP for a precice model.
But since I got that good results I am wondering if I even need those.

Can I generally expect such good results without using GCP or onboard RTK for Roofs?
Note that we´re not after 2cm accuracy. 10-15 cm relative accuracy would still be ok. We don´t need absolute accuracy for roof mapping.
Important is the horizontal accuracy. The result would probably be an Orthofoto for Autocad import.

The areas could be bigger than the one i tested on though. Sometimes we have complexes of several buildings.
One example in the near future would be an 16.000 m2 area if surveyed all in one (alternatively we could split it into several projects).

I would be very happy to hear from your experiences with roof surveying! What accuracy do you expect without GCPs for those cases?

Best from Berlin,
Simon

 

[Meta4]

I would be very happy to hear from your experiences with roof surveying! What accuracy do you expect without GCPs for those cases?

GCP's are necessary to tie your survey to the local datum and give accurate heights.
If you are only interested in inspecting roof condition, there should be no need for GCP's and relative heights should be fine.

 

[PVSimon]

GCP's are necessary to tie your survey to the local datum and give accurate heights.
If you are only interested in inspecting roof condition, there should be no need for GCP's and relative heights should be fine.

Thanks for the Reply! I just read that you can expect around 2 x GSM horizontally reletively and around 3 x GSM for the vertical accuracy. Can you confirm that? Could there be cases where I get dramatic shifts of accuracy in one 3D-Model? Again 10cm accuracy would be fine for me, since I just have to know where I can fit a certain amount of solarpanels on th e roof...

 

[Meta4]

Thanks for the Reply! I just read that you can expect around 2 x GSM horizontally reletively and around 3 x GSM for the vertical accuracy. Can you confirm that? Could there be cases where I get dramatic shifts of accuracy in one 3D-Model? Again 10cm accuracy would be fine for me, since I just have to know where I can fit a certain amount of solarpanels on th e roof...

What accuracy do you need?
Accuracy as mentioned above is how close the surveyed location matches the local datum - not how accurately the orthophoto shows the roof you photograph.
Without GCP's you can get a very accurate orthophoto but where it fits in the landscape could be a little out.
With the software I use, any altitudes are relative to my launch spot unless I use GCP's.

I can't imagine that GPS's and that kind of accuracy is at all important if you are just wanting a good representation of what's up there on the roof.

 

[PVSimon]

What accuracy do you need?
Accuracy as mentioned above is how close the surveyed location matches the local datum - not how accurately the orthophoto shows the roof you photograph.
Without GCP's you can get a very accurate orthophoto but where it fits in the landscape could be a little out.
With the software I use, any altitudes are relative to my launch spot unless I use GCP's.

I can't imagine that GPS's and that kind of accuracy is at all important if you are just wanting a good representation of what's up there on the roof.[/QUOTE

Thats really good to hear! Like I said I just need the relative accuracy within the Model, not the absolute. I was just skeptical because the results were unexpectedly good and I would like to know if I can expect that kind of accuracy for my next projects