[GRASSLIST:3156] Re: processing of LIDAR data

[Hamish]

> Does anybody has any experience with LIDAR data processing within
> GRASS or could point me to any resources?
> 
> I am looking especially for ways to extract a DTM from LIDAR data of
> forested areas, but also for any general advice on LIDAR data
> processing within GRASS.
> 
> My main problem at the moment is still the extraction of a DTM (points
> representing the bare earth) from the LIDAR point cloud. So, I am
> looking for ways to filter effectively point clouds. I will use LIDAR
> data of dense Sitka spruce stands. This data has a quite high overall
> point density (3-4 points per sqm), but the number of ground hits is
> naturally very low ("penetration" rate less than 20%).
> 
> The only thing I have read so far is the GRASS Book (Open Source GIS:
> A GRASS GIS approach) and Brovelli's paper (LIDAR Data Filtering and
> DTM Interpolation Within GRASS). 


I've never used LIDAR data, but from what you describe, s.cellstats
might be useful. How do you determine what is good data in the 'clouds'?
If it is just the lowest elevation in the cloud, you can just use
s.cellstats taking the minimum value within each cell, but I expect
there will be two clusters of elevations (1. ground, 2. canopy) + noise
from signal scattering; and you want to somehow separate out the lower
cluster and take the mean of that?

Perhaps use s.cellstats to calculate the first quartile value for each
cell, then modify the code to only do the mean of values below that
threshold? Or maybe just use the first quartile map as is.


?
Hamish