Calculate rockfall volume from 3D point clouds

[cullenni]

Hi Daniel,

I have 2 TLS point clouds and have used the M3C2 plugin to identify areas of significant change (i.e. rockfalls, see attached figure - significant change in grey). I would like to isolate individual rockfalls and calculate the volume for each. The image attached is just a small section of the cliff face.

I have tried the 2.5D volume calculation but I am not sure if I am doing it correctly. Its very difficult to draw an exact polygon around each rockfall so I extract the the area of change identified by the M3C2 plugin for each point cloud by selecting both clouds (normals),drawing a box around the area of change and extracting.I then select both extracted areas and open the 2.5D volume tool. The settings used are shown in attachment and result below.

Volume: -0.063
Surface: 1.259
----------------------
Added volume: (+)0.022
Removed volume: (-)0.085
----------------------
Matching cells: 93.1%
Non-matching cells:
ground = 1.5%
ceil = 5.5%
Average neighbors per cell: 7.9 / 8.0

Re the surface area - is this the entire planar or entire 2.5D surface area I extracted? or is it the planar or 2.5D surface area of change only? Should I only select the exact area of change?

Re the volume calculations - is there any way to determine the error on this value? It's a vertical cliff face so in theory there should not be any added volume only loss, hence I'm slightly concerned about getting added volume in the results. Perhaps this can be used to estimate an error term?

Finally, I have hundreds of rockfalls to calculate volume on is there any way to automatically select them? I have seen on other posts about connected components but I am still not clear on how it works.

Thank you in advance for any help on this,

Niamh

[daniel]

Once you have applied M3C2, switch the the 'significant change' scalar field, and use the 'Edit > Scalar fields > Filter by value' tool to extract the points with significant change (it will help a lot to isolate them).

Then for the 2.5D Volume calculation tool, make sure the points are orientation so that the change is mostly occurring along X, Y or Z. You should use the rotate/translate tool on the clouds if it's not the case.

This tool also needs a good density of points, so if your area of change is too small/sparse it maybe not be very accurate.

The surface is the sum of elementary surfaces (cell area) for the cells that have points in both grids/clouds. Therefore it's the approximate 'surface' over which the volume difference has been computed.

There is always an 'error' since you add elementary volume units (the cell area multiplied by the difference of the min/average/max height in equivalent cells. Therefore, the bigger the cells are, the higher the estimation error is. And of course it is also related to the cloud noise, density, etc. It's hard to estimate the exact error. And once again, this tool needs a good density to perform correctly. You shouldn't use it on very small areas.