to deal with
ad On Three
put forward
ntations, i.e.
iain factor to
jects. When
atio-temporal
act and its
led from the
domain, an
… geometry,
utes (Cheng
atio-temporal
to construct
on that time
zation factor,
as a basis to
anize objects
rganize the
work of the
nporal model
ich a model,
the spatial,
aspects. An
proposed to
thought less
was done
proposed a
in models of
t (1995) put
up of object-
presentation
) each other.
ains spatial,
utes (Figure
ed according
cation-based
event-based
ribute-based
ute aspect),
by applying
event-based
ge of these
e structured
alysis in an
os relates to
ector-based)
of individual
be traced,
should be
Jes are the
represented,
0 sufficiently
l., 1995).
Nonspatio-temporal
characteristics
Spatial Temporal
characteristics characteristics
Figure 1: Three components of a spatio-temporal object
4. LOGICAL LEVEL- HOW TO IMPLEMENT THE
SPATIO-TEMPORAL DATA MODEL
Logical modelling is the bridge to link the conceptual
models and physical data models. One of the most
important models at this level is the relational data
model, which is used by relational database
management systems to implement conceptual
models in computerized databases. However,
object-oriented database management systems
attract more attention recently as they have
advantages over relational data models.
This section will concern the implementation of data
models by applying the object-oriented approaches
proposed above. The paper distinguishes a loosely-
coupled method and a tightly-coupled method. The
three characteristics of objects are tightly linked in
the first case. It means these three characteristics of
an object are stored together under the same
identifier. In the second case, the geometric,
attribute and temporal characteristics of objects are
loosely linked. They can be stored separately, e.g.
in different files or in different DBMSs. The objects
can be represented as random combinations of
these three characteristics. The first approach
provides a tool to extract the change of an object as
a whole. It may not be easy to identify the changes
happened to which of the three aspects. The second
method is convenient to organize the objects which
frequently change in all three aspects: i.e., the
spatial, temporal or attribute aspects. It is not easy
to quickly query the situation of a specific object.
Several examples are given in the following section
to illustrate these two approaches.
4.1 A Tightly-Coupled Approach
A time-based approach is selected to implement a
unified and tightly-coupled spatio-temporal model in
(Cheng et al, 1995) The following steps are
proposed for the physical implementation of the
conceptual model:
(1) Set up lists to store the identifiers of objects
existing at time t1 (assuming t1 is the base state),
according to different object classes (i.e. body,
surface, line and point);
(2) Set up a “history-list” for each object, keeping
the temporal topology for each object;
International Archives of Photogrammetry and Remote Sensing. Vol. XXXI, Part B3. Vienna 1996
(3) Set up dynamic linked list to represent the spatial
composition of the object, keeping spatial topology at
the same time.
Body Class
T1 Bi B2 ..., Bn
T2 B1.B2 .... Bn
Tn B1 B2 ... Bn
B1 |B1,1|B1,2] ... | Bt,k|
|
des 5s
B1,1 [attribute] S1,1 ... S1,k1
|_»|B1,2 [attribute] S1,2 ... S1k1
B1,k [attribute] S1,k ... S1,kk
Figure 2: A tightly-coupled approach
The structure of the data model can be illustrated in
Figure 2. Such a data structure will have the following
characteristics:
(1) It is a unified representation of spatial, temporal and
attribute information of 4-D geographic objects;
(2 The geometric and temporal topological
relationships are explicitly recorded. It is convenient for
topological queries about spatial and temporal aspects
within and between objects;
(3) The changes of the objects are recorded explicitly,
which makes it easy to detect changes along time.
Using the tightly-coupled approach, however, it is not
easy to reduce the data redundancy for both geometric
and attribute aspects. The model shown in Figure 2
provides an effective storage for geometric data while it
might have high redundancy in attribute aspects.
4.2 Loosely-Coupled Approach
Yuan (1995) put forward a spatio-temporal data
model to manage wildfire information. It has three
domains of semantics, time and space. The data
model is shown in Figure 3. The semantic domain
consists of wildfire's concrete or abstract concepts
of aspatial and atemporal properties, such as
names of individual fire events, fire intensity, fire
types, or forest stands. The temporal domain
consists of temporal objects of points and lines,
which represent instance time and time intervals
respectively. The temporal domain supports
analysis and reasoning, such as fire frequency or
fire cycles. The spatial domain is composed of
spatial objects of points, lines, polygons, cells, and
volumes. Each of them represents zero-, one-, or
three- dimensional spatial units. It was suggested
that each domain could have its own database
management system (DBMS) for data storage,
