structure the
erspective of
g to different
ata analysis,
3 reality is
jace or time.
orced to be
ayers. If the
d with the
o adopted as
his approach
eometrically-
nvironmental
ompromises
(Raper and
il to directly
nd analytical
tructure can
modelling.
spects in the
i significant
and system-
echt, 1995),
the data
ould take a
| basis of the
ucture of the
, the direct
objects can
iat first level
ichieved.
the issues of
Jing, object-
attract more
information
> system is
raction in the
nposing the
1 the objects
/ from the
ch, 1993).
1 Livingstone
e traditional
y modified in
ways which
ich towards
hen studying
the object-
presentation
solutions for
ntal models
m design is
orspective to
and from a
proach. But
r reflect the
gner and the
mmunication
between the GIS and the user. Any aspect that is
overemphasized will result in a drawback of further
development of GIS and its applications. If we only
take the approach from the users perspective, by
applying an object-oriented approach, the system to
be developed will only be useful for very specific
cases, as the method and data are encapsulated in
the definition of the object. Although Yuan (1995)
put forward to generalize the users conceptual
schemata and analytical needs and try to implement
them in a GIS, it is not possible to integrate the
users’ conceptual schemata and analytical needs of
all specific application areas into a single
commercial GIS. As the aim of geo-information
theory analyses is to form design criteria and
strategies that can be used by experts to support
and advise users with the development of their
applications (Molenaar, 1995), we believe that both
the system-oriented and the application-oriented
approaches have their points in the design of data
models for coupling GIS and environmental
modelling. In practice, there is a need to take both
and a trade off between them. It can be shown in
the case study In Section 6. In the study of
geomorphology of the coastal zone (in Ameland, the
Netherlands), the coastal zone is decomposed into
landscape units according to the conceptual
schemata of user and environmental models. These
landscape units are represented as objects in the
first level, described by three factors at the second
level, while these factors are referenced by spatio-
temporal data stored in the third level. The data at
the third level can be structured and managed in
different ways from that of the first and the second
level. In this way, the third level of spatio-temporal
database can also be used for other studies such as
hydrological research in the same area, except that
the first and second level definition of objects has to
be adopted.
3. CONCEPTUAL LEVEL - HOW TO STRUCTURE
THE SPATIO-TEMPORAL OBJECTS
The last section discussed the view point for the
selection or definition of entities to be considered in
the spatio-temporal data model. This Section will
investigate the format or structure by which the
entities can be represented, i.e., "what structure in
the environmental problem should drive the
representational basis of the model?". In order to
integrate GIS and environmental models fully, we
should select a structure which can satisfy
requirements of both of them. However, most of the
existing GISs manage static and discrete data while
environmental models deal with dynamic and
continuous phenomena. GIS databases contain
information on location, spatial distribution and
spatial relationships while environmental models
work on a basic currency of mass and energy
transfer. Therefore dynamics and continuity should
be added to spatial data and spatial interaction for
the full integration.
International Archives of Photogrammetry and Remote Sensing. Vol. XXXI, Part B3. Vienna 1996
There are two approaches to represent spatial data
and related processes, field-oriented and object-
oriented. The field-oriented representation takes the
view that the environment is continuous. It is a
locational-dependent representation, ie. the
thematic information of the entities is attached with
their locations or geometric data. Thus, the data is
structured according to their locations, representing
"what exists at where" and "where" (specific
locations) is assumed to be known or fixed first.
This representation is fit for temporal analysis of
thematic information related to fixed locations. The
object-oriented representation considers that the
world is made up of objects which are uniquely
identified by their identifiers. The thematic data and
the geometric data are linked together through the
object identifiers (Ids). It assume that the object
exists first, then the data is structured according to
the object, describing "where is it and what thematic
attributes it has". It is fit for the temporal analysis of
the objects, which location and the thematic
attribute may change in different ways.
As for many natural phenomena of continuous
surfaces, they can be represented by field-oriented
approach, and temporal analysis related to specific
locations can also be easily implemented. E.g., the
height change of terrain surface on several points
(fied horizontal locations). But the temporal
analysis of the whole terrain surface of dune is not
easy to be accomplished if the data is structured by
locations. Firstly, the temporal analysis of the
thematic information is not direct and convenient, as
it can only be revealed after the accessing of the
related locations; Secondly, as the spatial
distribution (locations) are always changing, the
thematic data attached with them have to be
updated according to their changes. Therefore the
field-oriented representation is not suitable to
represent dynamic, natural phenomena, which
position always changes, e.g. the developing dunes
and transportation of polluted materials. In these
cases, the geometric data can not be considered as
a basis or a reference to link the thematic data. But
the object-oriented representation will be able to
deal with such situations.
The difference between field-oriented and object-
oriented is coming from their perspectives of
viewing the world, i.e., what exists first. In the field-
oriented perspective, it assumes the space exists
first, while in the object-oriented perspective, it
assume that the object exists first, in the space with
thematic attributes. For environmental issues, the
object-oriented perspective seems more natural and
acceptable to the user compared with the field-
oriented one. However, for the real representation of
these two perspectives, they don't have so much
difference if we think both of them can be raster or
vector based format. So, at the conceptual level
behind the field-oriented and object-oriented
representation, the representations for locational
(geometric) aspects are same, i.e., raster-based
and vector- based. Both of them can be adopted to
represent the natural phenomena.
