Author Archives: Supernasen


1. Relative Höhen berechnen
– “Höhenmodell” – “Geländemodell”

2. Einzelne Bäume identifizieren
– Umkehrung/Invertierung des Rasters mit den relativen Höhen
– Identifizierung der relativen Minima

3. Höhenvarianzen ermitteln

4. Klassifizieren in ein- und mehrstöckiger Wald


How has the DEM been derived from the LiDAR dataset?
The DEM (Digital Elevation Model) consists of the last returns of the LiDAR dataset and was created by using a linear interpolation technique (0.20m cell resolution).

How has the tree height/DSM information be retrieved from the LiDAR dataset?
The DSM (Digital Surface Model) consists of the first returns of the LiDAR dataset.
To get the forest height surface the DEM was subtracted from the DSM. Finally, with a quite complicated “tree height finding model” individual trees were identified.

How has the vertical structure be classified into single- and multi-story areas?
The classification in single- or multi-story area depends on the tree height variance.
Cells with tree height variances <1.54 m were classified as single-story and cells with tree height variances >1.54m were classified as multi-story areas.
(The value is based on the median value between the minimum tree height variance observed in the multi-story plots (2.75m) and the maximum tree height variance observed in the single-story plots (1.21m).)

What is the principle difference between the single-/multi-storage data set and the characterization of the forest areas using the coefficient of variation approach (section 4)?
The method with the coefficient of variation approach (CV) is a continuous surface that is independent of sample size (–> samples of unequal sizes can be compared).

Which errors of the approach can be attributed to post spacing?
For example the side of a tree crown could have been detected while the top of the tree was not measured or trees (e.g. forked trees) could have been counted double.