ESRI has recently published a white paper entitled, “LiDAR Analysis in ArcGIS 9.3.1 for Forestry Applications“. Written by Gordon Sumerling of ESRI Australia the paper seeks to “step through processes to convert lidar data into a format ArcGIS can process, explain methods to interpret the lidar data, and show how ArcGIS can disseminate the data to those who are not geospatial analysts”.
This well written document, although specific to ArcGIS is also valuable in its presentation of basic concepts of working with LiDAR data in a forestry application in general. Looks like a valuable introductory resource.
This whitepaper is similar to the blog they published last summer (http://blogs.esri.com/Dev/blogs/geoprocessing/archive/2008/10/29/Lidar-Solutions-in-ArcGIS_5F00_part-1_3A00_-Assessing-Lidar-Coverage-and-Sample-Density.aspx).
I am not very fond of their technique for processing – I wish they would mention that using last/first return for ground and canopy mapping does not produce accurate results of the canopy, or the ground in a lot of cases (e.g. dense forest). Some 3rd party vendors have tried to produce ArcGIS extensions for performing ground/nonground separation, but they are cumbersome to work with. I guess it is unrealistic for ESRI to advertise that part of the LiDAR processing worlflow should really be outside of their software (e.g. In TerraScan or equivalent).
Most users of lidar data get data that has already been classified by the data provider using 3rd party tools such as TerraScan, LS360 or TLiD or use such tools to pre-process the data. This should have been clarified in the white paper. Such tools are important for performing classification of the points which is not part of the workflow. The workflows make use of the point classification to enable the distinction between the ground and canopy values and not on the first/last return.