dtm from sfm-pointcloud in hilly forest area

[kristina....@gmail.com]

Dear all,

I am aiming to generate a DTM from a photogrammetric point cloud using lastools. The terrain of my study area is quite hilly and the relief changes a lot. It contains mostly scattered tress which sometimes do create a closed canopy but also buidlings and a lot of bushwork in other areas.

The inital pointcloud has a spacing of 0.03m and covers about 300,000 m2. As a fist step to improve processing I created tiles of 40x40 m and used lasthin with a pixelszie of 0,05 m extracting the lowest point. Afterwards I followed a variation of this workflow https://rapidlasso.com/2017/07/04/removing-excessive-low-noise-from-dense-matching-point-clouds/#jp-carousel-2597 to remove lownoise and generate a DTM with 5 cm resolution.

Doing so, I have issues to calibrate lasground_new properly since the feature that ought to be removed vary greatly. As for now I used the following settings

step 4

bulge 0.1

spike 0.1

down spike 0.1

offset 0.01

-extra_fine sub 7

The result is relatively satisfying in most parts but does classify some tree tops as well as areas of rooftops as ground points. When increasing the step size and/or reducing the sub the issues disappear but then large planar areas at the borders of the tile are falsely classified as nonground points as well (the effect is much worse on other tiles but i display this one because of the roof)

topview

stepsize 4

stepsize 5

Appearently, I am still struggling to fully grasp the calibration of the settings. What I understood is that bulge, spike and offset may be set low, when the relief doesn’t change abruptly since it keeps the surface relatively smooth removing edges and bumbs. Downspike should also keep the lownoise that is included in the tin at a minimum, correct?

For these reasons I was mainly playing with the stepsize and sub..  But why do the large areas at the border appear and how to avoid it?

link to the data of two tiles https://drive.google.com/open?id=1OHCS6hIDciP354sLs9K1AtNfpL1Cm0wt

Any hint is more than welcome!

Greetings,

Kristina

[Martin Isenburg]

Hello,

When people are asking me to "have a look at their data" I choose not to accept download links to point clouds tht are in uncompressed LAS format. These are  difficult (or even expensive) to download when I am in remote areas with slow Internet speeds (and I usually am). The LASzip compressor or the LAZ format were made especially for the purpose of faster sharing of point clouds over the Web or across other bandwidth-limited networks. The development of this open source compression scheme and getting acceptance for it within our industry has been a labor of love and tears for many years (not just for me, but also my collaborators and sponsors) so I've got to be very strict here ... (-;

Now this one here is to be expected: "When increasing the step size and/or reducing the sub the issues disappear but then large planar areas at the borders of the tile are falsely classified as nonground points as well (the effect is much worse on other tiles but i display this one because of the roof)." You need to use buffers that are at least as large as the step size you use with lasground/lasground_new or you will get *serious* edge effects - especially on sloped terrains.

https://rapidlasso.com/2015/08/07/use-buffers-when-processing-lidar-in-tiles/

Regards,

Martin @rapidlasso 

--
Download LAStools at
http://lastools.org
http://rapidlasso.com
Be social with LAStools at
http://facebook.com/LAStools
http://twitter.com/LAStools
http://linkedin.com/groups/LAStools-4408378
Manage your settings at
http://groups.google.com/group/lastools/subscribe