las2dem vs lasgrid

[Ty Kennedy-Bowdoin]

Martin and lastools group,

I know this has been discussed a few times, but there are advantages to both lasgrid and las2dem, but neither tool has it all.  Overall, I prefer las2dem because it is much faster , but there is no switch to choose the highest, lowest, average, etc. point. 

Would it be possible to add this functionality to las2dem and blast2dem? or is it already available by some other combination that I havent thought of?

Thanks, 

Ty

--
Ty Kennedy-Bowdoin
Carnegie Institution for Science
260 Panama St.
Stanford, CA 94305
USA
+1 831 227-3885
Skype: Ty.bowdoin

[Veldman, Edgar]

Hi Ty,

 

It’s because las2dem/blast2dem is based on triangulation and interpolation across long triangles, where lasgrid is based on cell based gridding, and filling across voids.

Triangulation in combination with highest/lowest/average etc. just doesn’t work.

 

Edgar

--
Download LAStools at
http://lastools.org/
Visit the LAStools group at
http://groups.google.com/group/lastools/
Be social with LAStools at
http://www.facebook.com/LAStools
http://www.twitter.com/LAStools
Say goodbye to LAStools at
http://groups.google.com/groups/opt_out
 
 

[Ty Kennedy-Bowdoin]

Yes, I can see that this would be a problem. Although I dont know how to implement it, what I am looking for is  a hybrid approach, something like this:

1. if there are many points in a cell, I want to choose which value to use for the cell (min for ground model; max for canopy surface model).
2. if there are no points in a cell, I want to interpolate a value from the tin of all adjacent values.

In reality most models have some cells of both types in them, making it difficult to have to chose between approaches.

Just throwing out ideas in case anyone can imagine a good way to do this.

Thanks for the response!

Cheers,

Ty

[Martin Isenburg]

Hello,

so you want to triangulate LiDAR that has been min or max (aka '-lowest' or '-highest') gridded with a certain '-step n' size and then create a TIN-interpolated DEM through rasterization. Then your command line is:

lasthin -i lidar.laz -highest -stdout | las2dem -stdin -o dsm.tif

lasthin -i lidar.laz -lowest -stdout | las2dem -stdin -o dtm.tif

I have the feeling you are really after creating a Canopy Height Model (after all, it's CAO Stanford calling). What is it that you are trying to achieve that the standard

lasground -i lidar.laz -replace_z -o normalized_laz

las2dem -i normalized_laz -first_only -o chm.tif

does not do to suit your needs? Could you share some background? -Besides curious me - there are lots of laser foresters on the list.

Regards,

Martin @rapidlasso

[Veldman, Edgar]

No very elegant, but this will do that in a couple of steps:

LAS -> Lasgrid -> ASC -> las2las -> LAS -> blastdem -> GRID

[Martin Isenburg]

Hello,

Hello Edgar,

what you propose below works as well, but I would prefer the approach using lasthin. The reason is that with lasthin you get the highest (lowest) point at its original x/y position whereas with lasgrid you get the highest (lowest) elevation "quantized" to be aligned on a grid ... possibly more information is lost (although the sampling is more regular). Also ... the grid alignment of the points from the ASC raster is actually the "worst case" input for las2dem or blast2dem since points on a grid are "hard" to Delaunay triangulate and require resorting to a lot of "high-precision" arithmetic.

Regards,

Martin @rapidlasso

[Veldman, Edgar]

Hi Martin,

 

Yes, lasthin would probably be the way to go for simple operations like highest and lowest.

 

But if for some reason one would like to be able to use the full range of lasgrid operations (density, user_data etc) they’ll  will have to use this alternative approach, although I see no practical use to for instance interpolate ‘point_density’ … I also know too well that lastools sometimes struggles with gridded inputs :), so let’s go with your approach.

 

Edgar

[Ty Kennedy-Bowdoin]

Thanks for the suggestions, this gives me quite a few ideas to test out. 

I am currently using this type of procedure to calculate DSMs and canopy height models, and have always used a triangulated approach (with both lastools and terrascan) because it fills all cells and works very quickly but as always, I am on the search for a nicer model. This approach does have trouble when you have a lot of first returns on the ground or mid canopy because then you have a pretty dissected looking canopy surface (normalized or not). This has become even more of a problem as we get more sensitive sensors higher resolution and etc. because now we have far more points other than the really bright returns from the surface.

The gridded methods do work really well when you have really over sampled data and/or you can grid at a much lower resolution than you planned for.  Rivers, lakes, and poor spot spacing cause this method to end up really messy.

So, ideally you would have a combination of the two methods to help you get nice clean looking DTM, DSM, and canopy height models without having big holes for rivers lakes, or areas with lower spot spacing. I just got an email from a developer at Applied Imagery who tells me this is precisely what QT modeler uses to make their surface models.

Ok, so one idea is to thin which sounds good on paper, but I guess I will see when I try it out. Another is to do it both ways and combine. To do this, I guess I could just make two sets of rasters (one triangulated, and one gridded), then combining the two using the nodata value to select which model to use. This would fill in the holes in the gridded model with interpolated data. It could of course have some ugly artifacts in transitions, but it would be interesting to try.

Anyway, I'll let you know if I come up with the magic combination. Martin, do you think that this could be implemented into future tools somehow?

cheers,

Ty

[Tony]

Hi, 

Years have passed, but do you have a solution now? I am also looking for a way to combine lasgrid and las2dem. Like you said, the "grid" approach is better to me than "interpolation", but where point density low, a lot of empty holes appear. Just wanna keep the results from lasgrid when producing a DSM, and at the same time fills all the holes (like interpolation does).

Best,

Tony

[Seweryn Cieslik]

Hi

Just bumping this issue.

I have similar issue with my datasets - quite narrow corridors resulting in gaps

Regards

Seweryn

[Martin Isenburg]

Hello,

a combination of lasthin and las2dem or lasgrid and las2dem might be what you are looking for. If you could provide a lot more detail on your issue with lasinfo reports (*) of your input LAS/LAZ files, the command line(s) you have tried and a (not too large) image of the result(s) that those command lines give you, then it would be easier to talk about the problem you are having.

Regards,

Martin @rapidlasso

(*) how to generate lasinfo reports: http://rapidlasso.com/lasinfo

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