Cut \ fill plan (volume) for a quarry

[Vavania]

Ok, I got it, thank you very much! I'll try to do that next time. May I clarify some so far?

1) I tried to put GCP on diffrent levels. You suggest just to create uniform grid - without worring about levels.
2) As I understand, at fisrt  I should use only "triangles" in Agisoft and check calculated altitude in "circles". If they are correct I am happy, if not - I put altitude in "circles" into Agisoft and recalculate. Is it correct?

BTW, In Agisoft tutorials distances between GCP is quite large - if I'm not mistaken - more than 600 m. However it seems that point density is not enought but problem is not visible.

[SAV]

Hi Vavania,

As mentioned by GPC, the layout of your GCPs could be the cause of your problems. But based on one of your previous statements, I think that the accuracy of your GCPs might be an issue as well.

I put around and inside quarry some boards as GCP. We don't have RTK GPS so I took coordinates for the GCP from the first flight and used them in other for markers. I took GCP elevation from our surface survey.

You might want to invest in a RTK GPS if you are planning to do such work more often. Check out the EMLID REACH RS, which is a low cost option and seems to work very well.
https://emlid.com/reachrs/

Regards,
SAV

[Vavania]

Ok SAV, thank you for info!

My task now is to convince heads that the technology reliable enought to run it in our dozens of quarries. The "problem" is that it is very "easy" to get a result. We get volume and don't know is it correct. The cut\fill method is good control for this. I have seen several cut\fill plans (nice pictures!) presented by UAV staffs, but  have never seen miners happy with those (maybe they exist). I am "mining" proffesional and don't happy with accuracy so far.

I will try to improve the results using the available capabilities and offer solutions to improve accuracy.

[SAV]

Hi Vavania,

Do you mind (privately) sharing the pictures and GCPs with me so I can process them?

So far I have had a very good experience with UAV photogrammetry. I agree, it is almost too easy to go get a result that 'looks good'. But in order to achieve accurate results one needs to follow a best practice workflow and use a sufficient number of accurate and properly distributed control points, besides many other things.

One more thing you could do is to acquire not only nadir imagery, but also add a few oblique images in order to achieve a convergent imaging geometry. As shown by James et al 2014, this can help to reduce the systematic DEM error.

I think that a volume measurement from a traditional stockpile survey has a larger error compared to a that from a UAV photogrammetry workflow, simply due to the 'sampling density'.

I hope we can make your surveys work 

Regards,
SAV

[Vavania]

Thank SAV, I have sent them.

It is just my main task to work out the best practice in this particule issue.
I agree that traditional methods are not ideal, but I have to convince that innovative ones are not only younger, but also better.

[SAV]

Hi Vavania,

Thanks for sharing the data. As I already mentioned in a PM to you, the problem is related to the accuracy of your ground control points which a) have not been surveyed at cm-accuracy and b) have different coordinates for the 2 surveys you tried to compare even though they are the same GCPs.

To solve the problem, (re-)survey these GCPs using a cm accuracy RTK or PPK GNSS and then process your datasets again. Your problem should then disappear 

All the best.

Regards,
SAV

[Vavania]

Ok, thanks SAV!

I understand that we need better acuracy for our GCP, however I don't understand why Photoscan moves GCP in exported files but shows the same position within the application. I will descibe the conditions for our header and contact with producers of the device your recommended.

[SAV]

No worries, Vavania.

You'll be able to achieve reliable results when your GCPs get surveyed at cm-accuracy (e.g., using an affordable EMLID REACH RS). Everything else was already fine (GCP layout, camera quality, image overlap).

Regarding the position of GCPs in PhotoScan. In my case they have always been shown slightly shifted (see attached screenshot of the sparse point cloud + 3 GCPs) in both PhotoScan and QGIS.

All the best.

Regards,
SAV

[Vavania]

Ok SAV, thanks, I got it!