Microtopographical Analysis

[noahmkorne]

Hi all, brand new to this forum and software. I'm hoping to get some guidance on how to accomplish some of my project objectives described below. If others have asked similar questions I apologize.

I am working on a project analyzing microtopography at arctic breeding shorebird nests. Nest sites were surveyed with a drone, and the files converted to .las point clouds using Pix4D. Microtopography refers to small (typically less than 30 cm) bumps/hummocks that cover the landscape. I've attached a screenshot of a point cloud for reference. The scale of this project is quite small, with the point cloud measuring roughly 10x15m^2.

I am hoping to find ways to
a) quantify the surface with regard to: number of hummocks; mean and sd of hummock height; individual hummock height; etc.
b) assess some sort of terrain ruggedness index
***for a,b) it would be important to be able to conduct these analyses without taking into account overall slope of the cloud
c) conduct above analyses within a certain radius of the nest, is there a way to segment by a pre-measured circle?
d) compare the results of a,b) between the nest site and a random point to determine if there are significant differences between the two.

Thank you so much for your time, I'm looking forward to learning more about this software.
Noah

[daniel]

Hum, hard to give you precise directions as each problem needs specific processings and steps (no 'out of the box' tool here ;).

Look for curvature and roughness computation tools (see http://www.cloudcompare.org/doc/wiki/in ... le=Methods).

Maybe it's possible to automatically detect nests with the Canupo plugin (for classification): http://www.cloudcompare.org/doc/wiki/in ... O_(plugin)

This way you would be able to extract the nests and at least it will help the next steps.

[ReneeDegutis]

Hum, hard to give you precise directions as each problem needs specific processings and steps (no 'out of the box' tool here ;).

Look for curvature and roughness computation tools (see http://www.cloudcompare.org/doc/wiki/in ... le=Methodstools).

Maybe it's possible to automatically detect nests with the Canupo plugin (for classification): http://www.cloudcompare.org/doc/wiki/in ... O_(plugin)

This way you would be able to extract the nests and at least it will help the next steps.

Hi Daniel,

Does the Canupo plugin allow to compare the results of the classification? And is it only for 3D items? I wonder if it's possible to complete my geomorphology research analysis based on CorelDraw processed images.

Regards,
Renee

[daniel]

Hi Canupo can be used to classify the points (basically to flag the points with the information about whether they are corresponding to a class or not)
But there's no comparison option. And it only works on 3D point clouds indeed.

If you want to go 2D, then there are probably plenty of tools I guess?

[noahmkorne]

Hi, thanks for the replies.

I am curious if there is a way to extract raw elevation data, with regards to the number and heights of peaks in the point cloud for further analysis?
I have begun to explore the roughness and scalar functions, and looked at the .csv export from their related histograms, but that doesn't give me the flexibility needed.


Also, I have attached a photo of a code for computing a Terrain Roughness Index from a paper by Riley et al. 1999. Full admission, I do not fully understand the computation, however I am wondering if there is a way to implement a similar kind of calculation in CloudCompare? It calculates the sum change in elevation of a cell from its 8 immediate neighbours. In the paper this is done at a km^2 scale, however it would be much smaller in this case. So a further question would be if it is possible to specify the 'cell' size used.

Thanks,
Noah

[noahmkorne]

Also, to repeat a question from the original post,

I am wondering if there is a method to use the segmentation tool (or other) to select a circular section of the point cloud for analysis.

Noah

[daniel]

- you can extract the elevations as a scalar field (Edit > Scalar fields > Export coordinate(s) to SF). And then you can segment the points based on their elevation (with 'Edit > Scalar fields > Filter by value'). But it's the absolute elevation and not the local one relative to the local peaks.

- the code you provide is for a raster (not a point cloud). On a cloud you could do the same by using the nearest neighbors inside a sphere (but you would end up with a sum based on different number of neighbors for each point) or with a fixed number of neighbors (but the extent of the neighborhood would change for each point). In this case it's better to normalize the result (e.g. by the number of neighbors in the first case). And in this case it's equivalent to computing the elevation scalar field gradient (Edit > Scalar fields > Gradient).

- there's no official method to select a circular section of a point cloud. You could of course hack the code of an existing tool (if you have some programming skills). Or more 'manually', you can create a sphere entity of the right radius (with the 'File > Primitive Factory' tool). Then move the sphere where you want to extract the points (you can enable the 'stippling' mode on the sphere so as to make it partly transparent). And then compute the distances between the cloud and the sphere (you can put a maximum distance slightly higher than the sphere radius to avoid losing to much time for the computation). And then keep only the points inside the sphere (i.e. with a negative distance) with the 'Edit > Filter by value' point.