SA) 
s of 
ms 
ual 
of a 
ap. 
ga 
dy 
of 
the 
14 
Brandt, Steffen 
  
3.3 Automatic DTM Generation using Laser Scanning as an Example 
Due to the high accuracy requirements of the data used to produce the terrain model, the quality of laser scanning 
(TopoSys 1998) was verified using analytical comparative surveys of an analysis area (100 kn? in total), giving an 
insight into both the potential and the restrictions of the processes. The regional survey department of North Rhine- 
Westphalia provided the DSK with a second laser scanner terrain model (TopScan, 1996/97) of an area for which the 
two data records overlapped. It was then possible to compare the accuracy of both laser scanners with the analytical 
reference survey. 
The result of the “last pulse” registration of the TopoSys scanner was a grid with a mesh size of 1m. Not least as a result 
of the “last pulse procedure” and the high resolution, it is possible to clearly illustrate the route structure in this 
example, in spite of dense coniferous forest. 
For the digital situation model, a representative value in each of the resulting 1m cells was determined from 4-5 values 
per m°. Analyses at the Institute for Photogrammetry and Engineering Surveying (IPI) produced an altitude accuracy of 
20 — 30 cm compared with the analytically measured reference grid (Koch 1999). The location accuracy is quoted by 
the manufacturer at 0.9 m. The comparison of the DTM by TopoSys with that of TopScan produced a relative location 
displacement in the order of 4 —5 m in the least favourable case. The inaccuracy of the DTM produced by laser scanning 
can be explained by the occurrence of systematic errors. A large proportion of these errors are in the positions and 
direction determination by GPS or INS. Different error influences overlap and are thus difficult to separate from one 
another. As a result, only a hypothetical explanation can be offered concerning elements affecting accuracy (Koch 
1999). DSK implements a third party product for the automatic generation of breaklines (CB-DHM-Laser, by C+B- 
Technik) in order to increase the accuracy and reliability of the digital terrain model. In this way, DSK aims to 
accelerate the manual processing 
The result shows that a large number of breaklines can be identified which must be selected. On the other hand, good 
results can be obtained, depending on the morphology — however the location accuracy of the breakline identification is 
naturally impaired, as described above, by the general location inaccuracy of the laser DTM. For certain tasks, these 
could considerably automate the process in combination with manual surveying. Naturally, the automatic process is not 
capable of, for example, differentiating between artificial breaklines (highways) and natural breaklines (ditches). This 
task will continue to be left to the additional contribution of the operator in the foreseeable future. 
  
Figure 5. Lippeaue: Measuring Elements of Breakline Identification (blue), Manual Recording (other colors) 
The resulting photogrammetric data are entered into the GIS ArcInfo from ESRI to calculate the detailed terrain model. 
  
International Archives of Photogrammetry and Remote Sensing. Vol. XXXIII, Part B7. Amsterdam 2000. 217