- 
VW 
TE 0 
SM 
WA NW com NE 
Av 
Nh. 
Schardt, Mathias 
  
Figure 2: 1x1 m forest floor DEM (left) and 1x1 m crown DEM (right) - Alpine test site 
  
Figure 3: 1 x 1 m final tree height maps - cut-out from Alpine test site (left) and hilly test site (right) 
Forest floor DEM 
Because of difficult terrain conditions (steep slopes and small structured relief) high attention was turned to the question 
of generation accurate forest floor DEMs. Raw data (last pulse if available) was processed in a regular grid of 1 x 1m by 
means of a multi-resolution method in combination with an improvement by gradient, figure 2. 
The method searches for the minimum within the respective resolution and increases the minimum by an appointed 
amount as a function of the gradient within the neighborhood. Within every resolution step smoothing and thresholding 
algorithms are performed. Thus, pixel values are resulting within every resolution step. The method starts with a rough 
resolution (e.g. 10 m) and compares the results with better resolutions (e.g. 7, 5, 3,1 m) step by step. Hence, the method 
allows to get forest floor information even in dense stands, as it is assumed that within 10 x 10 m at least one laser signal 
comes from the ground. The multi-resolution method - decisions whether to take the rough or better resolution are made 
by thresholding - assures to take ground results from better resolutions whenever there is a ground signal, in other cases 
(dense stands) the value from the rough resolution is taken. The obtained accuracy varies between 18 and 100 cm (Ruppert 
et al., 2000). 
Crown DEM 
The processing of crown DEMs followed through a simple algorithm, which searches for maximum values within 1 x 1 
m for the regular grid. Smoothing and thresholding algorithms were performed, figure 2. 
Tree height map 
The final tree height maps were calculated as the difference between the crown DEMs and the final forest floor DEMs as 
shown in figure 3 and were appropriated for the further investigations. 
4 STAND-WISE APPROACH USING LASER SCANNER 
Assessment of tree heights 
For the stand-wise assessment of tree heights a maximum filter (window size 3 x 3) was applied to the tree height maps 
to smooth the tree heights with regard to the estimation of top heights from forest inventories. The processed tree height 
maps were analyzed using stand-wise forest inventory data. The stand boundaries were obtained from the end users, 
digitized and imported to GIS system. The boundary information was visually checked by means of aerial photos and 
laser data. 
  
International Archives of Photogrammetry and Remote Sensing. Vol. XXXIII, Part B7. Amsterdam 2000. 1319