to a data set of minimum size and, onthe other hand, the 
response time of a system should be reasonable. In order 
to develop three-dimensional data structures terrain in- 
formation must be connected with situation information 
(figure 1). In terms of height integration geometry can 
be represented by 2D data (situation), 1D data (terrain) 
and a combination of both which leads to closed 3D data 
models. However, the integration of DTM into GIS is 
only a first stage in 3D modelling - the final objective 
should be a 3D boundary representation of the earth 
surface as reference not only for man-made construc- 
tions above and underneath the terrain but also for 
complex 3D natural terrain phenomena. 
   
  
      
terrain 
and 
situation 
  
    
Figure 1: GIS data model 
2. DTM DATA STRUCTURES 
Digital terrain models are represented by three different 
data structures (Ebner,Fritsch,1986). These data struc- 
tures may evolve during data acquisition and data appro- 
ximation, respectively. For that reason we can differen- 
tiate between: 
- anirregular structure, which is triangulated and the- 
refore forms a network being called triangular irre- 
gular network (TIN). The triangulation should care 
for constraints between nodes, for instance, to inte- 
grate breaklines, and ridge lines (figure 2). Obvious- 
ly this is the most direct representation of the prima- 
ry data. It is suited best for all tasks where strong 
reference to the measured data is required (e.g. 
volume calculation) or if a fast and rigorous revision 
of the DTM surface is necessary. 
498 
Figure 2: TIN data structure 
- aregular data structure, in which the data is organi- 
zed by a GRID of fixed or variable grid size. For the 
generation of the GRID methods for interpolation 
and filtering are used (Kraus,1972; Koch,1973; Eb- 
ner,ReiB,1978; ReiB,1985). With regard to better 
quality terrain representation the grid is often inter- 
sected with geomorphological information (figure 
3. 
  
  
| 
ae iio of 
| 
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
Figure 3: GRID data structure 
- acombination of both data structures resulting from 
the integration of additional geomorphological in- 
formation. This information is considered by local 
TIN's in a GRID. With regard to merging two diffe- 
  
Pra 
dat: 
pro 
The 
files 
posi 
imp 
isa 
late: 
shoi 
posi 
well