ynergy of 
sing and GIS 
Int. Arch. 
e Sensing, 
Heipke, C., 
point 
MOMS-02/D2 
Symposium, 
Rigorous 
g of SPOT 
S Journal of 
Sensing, 44: 
the MOMS-02 
mission D2. 
Space (ISPRS 
-50. 
1996 
THREE-DIMENSIONAL GEOGRAPHIC INFORMATION SYSTEMS - 
STATUS AND PROSPECTS 
Dieter Fritsch 
Institute of Photogrammetry, Stuttgart University 
Keplerstrafe 11, 70174 Stuttgart 
Phone: 0711-121-3385 
e-mail: Dieter.Fritsch@ifp.uni-stuttgart.de 
Commission III, Working Group III/IV — Invited Paper 
KEY WORDS: Digital Terrain Models, DTM Integration, 3D Modelling, 3D Data Retrieval, SQL, SQL3 
ABSTRACT: 
From the beginning after the introduction of geographic information systems (GIS) the R&D activities concentrated 
mainly on the computerized link of two-dimensional geometric and various thematic data. Resulting software packages 
— also called GIS products — are used for digital mapping, spatial data analyses, or simply for data retrieval. Currently, 
efficient GIS products are offered by vendors which besides the two-dimensional z, y-geometry integrate a digital terrain 
model (DTM) z = z(z, y) for the representation of the 3rd dimension. The introductory part of the paper reviews con- 
cepts and realizations for the extension of the geometric GIS dimensions. After some metric definitions several approaches 
are presented in more detail to overcome the lack of single valued 3D geometry. It is shown, that most commercial GIS 
products use the layer-oriented D'TM integration. The second part of the paper deals with 3D modelling of other georelated 
disciplines such as geology, geophysics, hydrology, etc. It can be seen, that these disciplines apply modelling strategies of 
solid geometry. The link of the data models applied here with the D'TM integration of the surveying disciplines comes to 
strategies which integrate both: the topographic boundary and solid 3D geometry. The 3D data retrieval differentiated 
in measurement functions, spatial predicates, and object generating functions is discussed. The resulting query language 
should take into acount these classes of data retrieval. The last part of the paper gives an outlook on using object-oriented 
techniques for solving the 3D modelling problem. Using the example of automated extraction of buildings an object class 
builder is shown which takes into account the DTM (regular grid, TIN) and the 3D representation of buildings. Also 
existing object-oriented database management systems will be discussed which are used for 3D data models and its data 
retrieval. 
1 INTRODUCTION 
The fast development of the fascinating field of geographic 
information systems (GIS) was dominated for a long time 
simply by converting analog map data into large digital 
databases. For more than two decades the 'map geometry’ 
(i.e. x,y-coordinates, 2D topology) was digitized, linked 
with attributes of various themes, and then stored in non- 
standard database systems (H.J. Schek, 1996). Most of the 
progress in GIS was initiated during the last 10 years - the 
use of relational database management systems (RDBMS), 
data standardization procedures, object-oriented model- 
ling techniques, and last but not least height data integra- 
tion, to name only few. 
'The problem of integrating some height information is a 
matter of great concern not only for photogrammetry, but 
also for other georelated disciplines which make fully use 
of 3D data. Therefore it is out of question, that GIS R&D 
can not accept any longer only 2D geometry, particularly if 
topographic mapping is concerned, and if the actual shape 
of the Earth’ surface is investigated for exploration and 
environmental purposes. 
First proposals, initiated by the domain of utility inform- 
ation systems, assigned z—values as attributes for the 
(x,y) location. This non-geometric height data integra- 
tion was capable to overcome an urgent need being three- 
dimensional. But research in the recent past has shown, 
that regional and local GIS as well as its software (GIS 
products) that is specifically designed for topographic map- 
ping, should be at least two-and-a-half-dimensional (2.5D). 
For geoscientific data, where objects at one (z, y) location 
only have a single z value, a 2D GIS product might be 
adequate (D. Fritsch/D. Schmidt, 1995, D. Schmidt/D. 
Fritsch, 1996), because the 2D data model can be extented 
to 2.5D. Most 2D GIS products have facilities for perspect- 
ive display of surfaces that are single valued (not more than 
one z value per location z, y). 
Besides that, GIS and belonging products designed spe- 
cifically for geological work, particularly for mining and 
oil exploration, need to be fully three-dimensional, so that 
each data object is characterized by its location in space 
with three spatial coordinates z,y,z (T. Bode et al., 1994, 
M. Breunig, 1996, H. Kasper et al, 1995). The same 
holds true for urban information systems, that can not re- 
nounce on three-dimensional modelling and graphical out- 
put. Therefore, in the following the state-of-the-art of 
height data integration is reviewed. Some interests are 
also directed to the potential of 2.5D and 3D that should 
be opened for GIS applications in near future. 
215 
International Archives of Photogrammetry and Remote Sensing. Vol. XXXI, Part B3. Vienna 1996