n can 
s with 
) have 
)es of 
bjects 
nts of 
2spect 
el as a 
of all 
ons to 
paper, 
assess 
y help 
nulate 
part of 
utline 
active. 
ance, 
to the 
vell as 
n and 
ystem 
posely 
rm of 
'eality. 
a GIS 
rs and 
query 
ledge 
> data 
ns the 
nents 
  
into information which should be made understandable to 
human by the presentation mechanism. Another important 
role of GIS is maintenance and management of the 
database. The GIS should provide the possibility to update 
the database. The management role is to handle various 
request from the user and activate appropriate operation on 
the database in response to the user's request. 
In the sense of integrated geoinformation two issues must be 
considered: the first one is the spatial model itself; it must 
reflect all those aspects of reality that are relevant for the 
intended use of the GIS. The second one is, whether the 
system offers all means needed to utilise the spatial model, 
which is a matter of system functionality. This means various 
technologies must be adopted to serve different functions; 
e.g., photogrammetry for acquiring some components of 
spatial model, computer aided design (CAD) for graphically 
constructing, editing and visualising the spatial model, 
database technology to manage the spatial model, virtual 
reality to naturally explore the spatial model, etc. To 
effectively exploit the technological development, the spatial 
model should: 
e include three-dimensional (3D) aspects of reality, 
© allow both direct and indirect representation of the real 
world at the desirable level of abstraction, 
® be capable to accommodate data from various sources 
and seamlessly integrate them into one spatial model. 
The first requirement results from the fact that only limited 
spatial analysis of our 3D real world was possible unless we 
have a 3D model. The direct representation referred to in the 
second requirement is suitable for real world objects that 
have well-defined spatial extent. The original observations 
must be maintained by the model so that the knowledge 
about the reality will not be degraded. This implies that 
vector structure may be the more suitable basis of the model. 
In case a real world object cannot be directly represented, an 
indirect representation must be provided by the model, e.g. 
by defining the spatial extent of an object from the 
neighbours by interpolation of property values. In this case, 
the relationships among the neighbours given by the direct 
representation must be used as constraints for the derivation 
of spatial extent to make the indirect representation accurate. 
The third requirement points out the necessity of multiple 
representation in terms of semantics (i.e. an object may have 
different meanings to different observers) and dimensionality 
(i.e. ranging from OD to 3D). The spatial database of a GIS 
should have the capability to store and maintain multiple 
representation. Integrating data from various sources implies 
that redundancy must be minimized whereby human 
intervention is likely required to decide how uncertainty 
should be resolved. Moreover, to achieve a spatial model 
that well-represents relationships among real world objects, 
the underlying data structure must be unified (see Pilouk and 
Tempfli 1994, Pilouk et al 1994). 
The system should be capable in performing both query- 
based and computation-based complex spatial analysis 
across different themes and dimensions and provide the 
295 
possibility to present the information from arbitrary view 
points with realistic visualization. This requirement is beyond 
what 2D GIS can offer. The query-based spatial analysis will 
allow to obtain information based on direct representation 
while the computation-based analysis will help to derive 
information to predict a situation in reality. This information 
may be directly fed into the model again to avoid repeating 
time consuming derivation processes. 
3. FUNCTIONAL COMPONENTS AND RELATED 
DISCIPLINES 
To support the requirements stated in the preceding section, 
the five major functional components of a GIS are outlined. 
This outline may be derived from many existing 2D GISs and 
other systems that are related to spatial modelling, such as 
CAD, DTM, etc., as follows. (see also Maguire and 
Dangermond 1991). 
(1) Data acquisition 
(2) Data structuring 
(3) Data storage and management 
(4) Data processing 
(5) Data output and information presentation 
Each of the above components of a system has been 
separately developed in the past and many of those 
components have become disciplines of their own, e.g., 
database management system, photogrammetry, remote 
sensing, digital image processing, CAD, virtual reality, etc. 
These disciplines have to be brought together as functional 
components of future GIS. The next section discusses 
approaches to system composition by identifying four 
evolution stages. 
4. EVOLUTION STAGES OF SYSTEM ARCHITECTURE 
The composition of a system for integrated geoinformation 
will always depend on the state of the art, policy and 
economical constraints. Since technology develops, we can 
consider various system architectures as stages of evolution 
in handling geoinformation. We use the following fourteen 
criteria to analyse for different evolution stages. 
1) Compactness of the system 
2) Common operating system (OS) or hardware platform 
3) Functional access 
4) Data access 
5) Relationships among components of spatial model 
6) Commonness of user-interface 
7) Investment cost 
8) Maintenance 
9) Data redundancy 
10) Handling of uncertainty 
11) Productivity of geoinformation 
12) Potential toward automation 
13) Supporting personnel 
14) Size of user organisation 
International Archives of Photogrammetry and Remote Sensing. Vol. XXXI, Part B2. Vienna 1996