This worksheet covers LAS datasets which is a standard in distributing LiDAR point cloud observations. Since computing time and available software is quite a topic, this worksheet is optional in many cases but in the end you has to come up with a way to read it.
LiDAR point cloud datasets are generally distributed as LAS files which include the LiDAR point clouds and meta information. You can think of the point cloud like an XYZ file structure where the height information Z is stored for each XY location which has reflected a laser pulse. Since there is quite often more than one return signal from the same position, multiple entries with the same XY coordinate are possible. In addition to the Z information of each individual return, the number of this return from the given position along with the number of all returns from this position is stored. Actual LiDAR scanners also register the intensity if the return. Standard post-processing included accurate geo-location and at least a classification of the LiDAR returns into presumably ground- or canopy-based (canopy in this sense also includes houses etc.) For more information on the LAS data set and classification specifications, see the descriptions for the appropriate LAS version at [ASPRS-LAS].
The LiDAR datasets used by this course are provided by the Hessische Verwaltung für Bodenmanagement und Geoinformation (HVBG). As stated above, the LAS format can store a variety of attributes for each XY location. The classification usually provided by the data vendor into ground (i.e. a subset of all last returns) and canopy (i.e. a subset of all first returns) points is of course mandatory to derive some kind of elevation models.
The data set from the study region comes in an ETRS89/UTM 32 north projection and the height information is DHHN92. The returns have been classified by the HVBG as follows:
For more information see this presentation by Carsten Dorn.
While R is very handsome for many, many tasks, handling raw LAS data is not something you should do using plane R routines. For example, the gridded LiDAR data used later within this worksheet covers 8 square kilometers and has been compiled based on 95,503,394 LiDAR returns originally stored into 8 different LAS datasets. This is just to much to handle in e.g. spatial point data frames or something similar in an acceptable amount of computing time.
Hence, more efficient program language solutions (e.g. C) must be used for such tasks. One software to mention in this context is ESRI's ArcGIS which - starting from version 10.2 - has a quite handy LiDAR processing environment. A free alternative would be SAGA GIS which also offers LAS import (and subsequent handling like it was a point cloud). GRASS GIS also has a LiDAR workflow included and for those who want a purely command line but effective solution, the (non-free) LAStools is worth a try.
Just try any out if you want to!
The task of this worksheet is simple: read the LAS data provided for the study region of our Marburg University forest which is available through GeoMOC. The dataset to search for is “LiDAR point cloud for the Marburg University forest”.
Please use e.g. ArcGIS to read the provided LAS datasets.
After reading, play a little with the visualization options to become familiar with the data.