the same holds for DTM nodes and DTM areas (Fritsch, 
1991). A more detailed description of the attributes for 
these primitives is given in Höhle (1991). 
  
Figure 11: Attributes for DTM edges 
Using the data model of figure 1 a DTM can be genera- 
ted from objects which are characteristic for terrain only 
and which are characteristic for terrain and situation. 
The total integration of DTM allows 3D presentation 
and analyses of the terrain in combination with the situa- 
tion data what extends considerably the query space of 
a GIS based on 2D topology. 
4.2. DTM data structures in GIS 
  
In the following, problems of integrated data manage- 
ment, according to the different DTM data structures 
will be discussed in more detail. It is differentiated in 
DTM generation, data management and data analyses 
under the aspect of using a GIS. 
4.2.1. GRID and GIS 
The generation of a GRID as DTM data structure is 
solved by approximation and interpolation using the 
original terrain data. The result is a new data set with a 
regular structure which forms the surface of the terrain. 
Furthermore, if geomorphological line information is 
integrated into the GRID this information can be hand- 
led objectwise. Thus the object "height model" is to 
subdivide in raster cells and additional line information. 
Within data management two different data sets for the 
terrain have to be stored and managed what causes some 
problems. A DTM update can only be done when the 
original data set is accessable and on the other hand 
every update in the original terrain data set requires a 
   
502 
new DTM generation to be consistent in the ID GRID 
presentation. But there remains still an other consisten- 
cy problem using situation data which serve also as 
terrain information because there is normally a differen- 
ce between the height resulting from the DTM and the 
original height. In this sense consistency problems exist 
a priori. A way out of this dilemma whould be a rigorous 
separation of situation (2D) and terrain data (1D). But 
this reduces the possibilities for data analyses provided 
by the data model of figure 1 and makes redundant data 
storage unavoidable. 
terrain 
original data 
Figure 12: GIS data model for GRID integration 
        
  
terrain 
derived data 
    
4.2.2. TIN and GIS 
Within a TIN the original data represent the DTM. 
Updating the terrain surface within a TIN is much easier 
than in a GRID because changes in the original data 
have only local influence on topology and position; there 
is no recomputation of large parts of the DTM. The TIN 
contains implicitely all accessable terrain information 
without any approximation therefore the complete data 
model of figure 1 can be used for data analyses in con- 
trary to GRID. Consistency problems will not occur, 
when it is guaranteed that all 3D information takes part 
of the object "height model". 
4.2.3. HYBRID and GIS 
In the HYBRID data model the driving force is a GRID. 
The advantage of HYBRID is the better fidelity of non- 
regular phenomena such as break lines and ridges, spot, 
peak and pit points within the derived DTM what makes 
it better suited for data analyses which includes height 
Bi 
Bi 
Eb 
Eb