According to Molenaar (1995), GIS data modelling
has four steps or levels which represent the process
of mapping the real world gradually into symbols
readable by computers. These four levels include
spatial modelling, conceptual modelling, logical
modelling and physical modelling. As the physical
modelling deals with the physical storage of data in
computers, users and GIS designers normally pay
their attention only to the first three levels and the
relationships among them. Therefore, we will
discuss how the data modelling at these three levels
can support full integration.
2. SPATIAL MODELLING LEVEL -
HOW TO DECOMPOSE THE REALITY
Users in different application disciplines pay
attention to different phenomena. They will
decompose the complex environmental issues into
simpler ones. In the mean time several types of
entities are selected and defined as the study
objects to represent environmental issues. Later on,
they use environmental models to describe the
interrelationship and behaviours of these objects.
The process of decomposing reality and selecting
typical objects for a specific discipline is called
spatial modelling. Therefore, a spatial model can be
considered as the user understanding of the reality
and the model is for them to describe the reality. It
also implies a method to decompose the reality into
representable entities.
If the user and GIS designer can decompose the
reality in the same way, i.e., GIS designers and
users select the same entities to represent
environmental issues, then the environmental
models and GIS can be easily integrated with each
other. The benefits can be seen from three points.
Firstly, the users of the system can map their
understanding of the environmental issues directly
into GIS without paying attention to the geometric
concepts such as points, lines and polygons.
Secondly, the environmental models created outside
GIS can directly accept the abstract data types
described in GIS and the output of the
environmental models can be directly accepted in
GIS. In this case, the conversion between GIS and
environmental models for the abstract data types
will be reduced. Thirdly, the environmental models
may be developed directly in a GIS as the abstract
data types can be used to define the environmental
models.
There are two ways to decompose the complexity,
which will lead to different criteria and strategies for
the development of the spatio-temporal GIS data
model.
2.1 From the Data-Driven Perspective
Traditionally, the design of a GIS is not directly
related to the requirements of its application. Most
of the designs are started from the stage of
conceptual level. Most available GISs take a
850
International Archives of Photogrammetry and Remote Sensing. Vol. XXXI, Part B3. Vienna 1996
system-oriented approach, i.e. they structure the
data and design operations from the perspective of
a data-driven system, formed according to different
function modules, such as data input, data analysis,
data management and output. The reality is
decomposed into separated layers of space or time.
So the environmental entities are forced to be
segmented and represented in these layers. If the
environmental models are integrated with the
system, their form and nature have to be adopted as
the representational basis of the GIS. This approach
implies an essential adoption of geometrically-
indexed methods for representing environmental
models in a spatial context and forces compromises
on most environmental modelling (Raper and
Livingstone, 1995). They normally fail to directly
map the users’ conceptual schemata and analytical
needs. Consequently, the GIS data structure can
not satisfactorily support environmental modelling.
2.2 From the User Perspective
As the strong emphasis on technical aspects in the
design of most GISs results in a significant
drawback of application-specific data and system-
confined operations (Yuan and Albrecht, 1995),
some researchers suggested that the data
structuring and operation design should take a
users’ perspective. The representational basis of the
data model should be driven by the structure of the
application issues. In such a way, the direct
mapping from users’ concepts to data objects can
be provided (Yuan, 1995). It means that first level
integration of spatial modelling can be achieved.
As structured design does not address the issues of
data abstraction and information hiding, object-
oriented analysis and design approach attract more
and more attention of the experts of information
system design. In this approach the system is
decomposed according to the key abstraction in the
problem domain, rather than decomposing the
problem into algorithm steps, by which the objects
are identified and derived directly from the
vocabulary of the problem domain (Booch, 1993).
For environmental modeling, Raper and Livingstone
(1995), among others, stated that the traditional
science paradigm has been considerably modified in
the last few decades in a number of ways which
demands a more rigorous approach towards
modelling, classifying and discritizing when studying
environmental problems. Adoption of the object-
oriented approach to spatial representation
recognizes these priorities and enables solutions for
some of the problems of environmental models
coupled with GIS.
Therefore, we can see that the system design is
transferred from a system-oriented perspective to
an application-oriented perspective, and from a
data-driven to an object-oriented approach. But
ideally, GIS data models can better reflect the
mental models of both the system designer and the
user in order to best facilitate the communication
