ISPRS, Vol.34, Part 2W2, “Dynamic and Multi-Dimensional GIS", Bangkok, May 23-25, 2001 
Managing 
MTEROP’ 
TOPOLOGIC DATA STRUCTURE FOR A 3D GIS 
(SVG)”, 
Mattias Pfund 
îss et. al, 
Institute of Geodesy and Photogrammetry 
Swiss Federal Institute of Technology (ETH), Zurich 
ETH HOnggerberg, CH 8093 Zurich, Switzerland 
idelsnand 
Tel: ++41-1-633 3045, Fax: ++41-1-633 11 01 
ri, Vol. 8, 
E-mail: pfund@geod.baug.ethz.ch 
ABSTRACT 
The extension of Geographic Information Systems into the third dimension is gaining more and more importance every day. A huge 
number of GIS-users need as completion of the usual planimetric representation of the landscape (cartographic maps and 2D-GIS) also 
some altimetric data and the volumetric extension of interesting objects. 
Today most applications and data structures for 3D GIS are optimised for visualisation. They usually omit topological information in 
order to get a better performance. In consequence the functionality provided by this systems is often limited to visualisation and 
supports few analysis functions. If however the whole spectrum of GIS functionality we use in 2D-GIS (data acquisition, management, 
analysis and visualisation) has to be available for a 3D-GIS, then it needs an adequate geometrical data structure. 
This paper first describes the requirements of a suitable topological data structure for a 3D GIS. The goal is to model any object of the 
3-dimensional space exactly or at least with a good approximation. Secondly, the implementation of a prototype system supporting a 3D 
data structure is described. 
1 INTRODUCTION 
2.1 Data Aqulsition and Management 
Geographic Information Systems enable a numeric and 
abstract description of spatial objects and phenomena of the 
the real world. Although our world and the objects situated in it 
are three-dimensional, commercial GIS usually reduce spatial 
data by its third dimension and project it onto two dimensions. 
One of the main reasons for this is the complexity of an 
implementation of a complete three-dimensional GIS. First of 
all there ist still the requirement of an adequate geometric data 
model with an appropriate object building and an extended 
topology. Furthermore new methods for data acquisition and 
administration must be developped as well as algorithms for 
processing and analysis of 3D spatial data (Pfund 1999). 
The photogrammetric approach uses the profound knowledge 
of the community about the acquisition of 3D data. The main 
goal is usually to produce and visualize city models of a finite 
area. Starting with a detailed aquisition of the rooftops and a 
elevation model, they produce surface models of buildings and 
other objects. This techniques are quite fast and accurate. The 
disatvantage is, that these methods don’t provide much 
thematic information about the objects other than .house’, .tree’ 
and so on. Furthermore there is no clear concept recognizable 
about data management. The prevalent philosophy is: Acquire 
once and never touch it again! 
2 STATE OF THE ART 
For a cuple of years several efforts were made in 
order to realize a 3D GIS. Depending upon the 
initial position different strategies and system 
architectures were used. Basically one can 
distinguish between three different approaches, 
determined by the data aquisition method: A 
photogrammetric, a CAD and a GIS approach. In 
order to rate these methods one should keep in 
mind the four main tasks of a GIS: 
• aquisition (new, import, update) 
• management (modeling, modifying, storage) 
• analysis (queries, manipulations) 
• output (visualization) 
From a Geoinformation System point of view, a 
software must be able to perform all tasks listed 
above. If only one capability is missing partially, 
the usability of the system as a GIS is reduced 
significantly! 
Fig 1: Different evolution steps on the way to a real 3D-GIS: 
a) analogue/digital cartographic maps; b) 2D-GIS; 
c) 3D-visualisation of 2D-data; d) 3D-GIS with query functionality 
The realisation of a "real” 3D GIS is thus extremely difficult due 
to technical and conceptional problems, for reasons of the 
complexity as well as the temporal and economic 
developpment expenditure for the 
implementation. 
CAD systems provide a large amount of editing capabilities for 
3D data. Data can be aquired with high precisition and a lot of 
details. Because the lack of suitable capabilities for data 
Cartographic Map 
2D-GIS 
3D - Visualisation 
3D-GIS