lasheight normalization causes artifacts in tree shape along steep gorges

[Martin Isenburg]

Hello.

A user of LAStools came recently up with an interesting question. He noticed that the height-normalization process may encounter difficulties when lasheight is applied to forest landscapes with steep topographical variations. He wrote:

"I am currently analyzing tropical forest data. When the topography is relatively level, lasheight (used with "-replace_z" and "compute height above ground points") returns nice tree crown shapes for the raw data.

However, when the topography contains very steep gorges, lasheight cannot handle the trees along the gorge and causes severe shape artifacts (attached is an illustrated example). Maybe this is caused by trees growing close to the gorge edge with lots of light, hence bending towards the gorge opening...

Is there any simple adjustment which could eliminate this problem? I am looking forward to hear from you."

And he has illustrated the issue in the attached two images. I think it is the same issue that we encountered in the Canary Islands that I pointed out in my SilviLaser 2012 keynote talk with the attached slide.

Any "simple adjustments" folks can suggest to address this problem?

Regards,

Martin

[Terje Mathisen]

I think this might be related to another issue I have noticed in steep
hillsides:

The ground points determined by lasground/lasground_new or most
commercial products seems to be very likely to cut off sharp edges,
smoothing them into a sort of average slope angle.

Here in Scandinavia this results in two things:

Smaller ledges with a cliff in front of them ends up as approx 45-degree
slopes, and the ground returns from those ledges instead turn into low
or medium height vegetation, and any actual trees growing there become
artificially high.

In order to find more of those cliffs and ledges I've had to write my
own custom post-processing of the lasground(_new) generated "wilderness"
ground points:

I look for any patches of some minimum size where there are no ground
points and where the height vary significantly (at least 25-30 degrees
average slope) from one side to another, and then I look for the lowest
vegetation points in each square meter cell inside the patch: If those
points seems to correspond to a ledge I reclassify them as ground,
before I go on the the rest of the standard pipeline including slope
angle images, contour generation etc, etc.

Having trees that actually lean out would just make this issue even
worse, I think I would try to determine the ground height on all sides
of the area covered by the tree crown, and either pick the highest of
all those ground points or some kind of median value which would filter
away outliers.

Terje

Martin Isenburg wrote:
> Hello.
>
> A user of LAStools came recently up with an interesting question. He
> noticed that the height-normalization process may encounter
> difficulties when lasheight is applied to forest landscapes with steep
> topographical variations. He wrote:
>

> "I am currently analyzing tropical forest data. When the topography

> is relatively level, lasheight (used with "-replace_z" and "compute
> height above ground points") returns nice tree crown shapes for the
> raw data.
>
> However, when the topography contains very steep gorges, lasheight
> cannot handle the trees along the gorge and causes severe shape
> artifacts (attached is an illustrated example). Maybe this is caused
> by trees growing close to the gorge edge with lots of light, hence
> bending towards the gorge opening...
>

> Is there any simple adjustment which could eliminate this problem? I

> am looking forward to hear from you."
>
> And he has illustrated the issue in the attached two images. I think
> it is the same issue that we encountered in the Canary Islands that I
> pointed out in my SilviLaser 2012 keynote talk with the attached slide.
>
> Any "simple adjustments" folks can suggest to address this problem?
>
> Regards,
>
> Martin

> --
> 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


--
- <Terje.M...@tmsw.no>
"almost all programming can be viewed as an exercise in caching"

[Anahita Khosravipour]

Hi everybody,

Last year, we published a paper and only focused on this issue.

On steep slopes, the raw elevation values located, for example, on either the downhill or the uphill part of a tree crown are height-normalized with parts of the DTM that may be much lower or higher than the tree stem base, respectively.

Based on this observation, we theoretically and experimentally quantified the effect of slope on the accuracy of treetop detection and its effect on the tree height estimation before and after height normalization, in the paper.

This is the like to the paper:

http://www.sciencedirect.com/science/article/pii/S0924271615000581

I hope you find the information helpful.

Best regards,

Anahita