[GRASS-user] Processing Atlas Lidar Data - Problem(s)

[Stuart Edwards]

I've been trying to get my head around this recently too ........

On Mar 10, 2011, at 1:23 PM, TimNorwey wrote:
> 
> I´m trying to generate a DEM/DSM out of LIDAR data that I have downloaded
> from the website http://atlas.lsu.edu/lidar/ . 
> 
> The data can be downloaded as .csv-file. Within this file no attribute names
> are given. For test purpose I manipulated the .csv, so that "only" 10,000
> points were left and I added a row at the top "x,y,z" as attribute names.

So this is raw three column data with no indication of the return number
> 
> After doing that I imported the .csv into QGIS using the Plugin "Add
> Delimited Text Layer", stored the points as .shp-file and imported them into
> GRASS GIS database using v.in.ogr (as 3D points).

May be easier to use r.in.xyz - especially if you want to run statistics.  Or v.in.ascii if you can get a text file generated by las2txt (http://liblas.org/).  This assumes you have access to a .las file rather than the simple xyz file - many published sources offer it both ways.  The advantage of .las is that it will include the arrival numbers (and other information)
> 
> Then I tried to follow the workflow described on the website
> http://grass.osgeo.org/wiki/LIDAR#LIDAR_Tools  (I started at "DEM/DSM
> separation the more complex way").

Always start with the hard stuff  ..........
> 
> The first step should be the use of v.extract to separate first and last
> pulse laser points.
The example in the wiki does seem to indicate that this can be done simply by pulling the firsts and lasts out of the data set by name.  While not explicit, the example must be using a data set that includes the arrival numbers stored in a column named 'return'.  It didn't work for you because you have no such column.

Also, the suggested approach may have a problem in that the 1s are only 'firsts' if there is a 2 following; otherwise they are 'first' and 'last', and should be processed with the 'lasts'.  
> Because of the reason that I have no information about
> first/last pulse laser points, I skipped this step and tried to follow the
> other steps.
If your scene includes tree or brush cover, this will be a problem.  If not, it may be ok.
> But I think this was for nothing, because v.lidar.edgedetection
> needs the last pulse laser points and v.lidar.growing needs the first pulse
> laser points.
Actually they both run on last points - Again, it depends on how much vegetation you have to create multiple arrivals as to whether it's a problem not to know the difference.
> So I don´t write more about this...

> I really don´t know how to generate a DEM/DSM based on this data and it
> would be nice if anyone of you has an idea. Is there a possibility to do
> generate a DEM/DSM without using the steps that are described within the
> WIKI page above?
The most useful reference that I have found is Open Source GIS - A GRASS Approach (Neteler and Mitasova).  The wiki is actually pretty good too - probably best to start at the beginning rather than the end  :-)  and get familiar with the tools.  I quickly realized that this is not a trivial exercise.
> 
> What do you think about the import of the .csv into GRASS GIS via QGIS? Is
> there an easier way to import it?
see above.
> 
Good luck -

Stu