ISPRS, Vol.34, Part 2W2, “Dynamic and Multi-Dimensional GIS”, Bangkok, May 23-25, 2001
3.1.3 Spatial Scene Change. Spatial scene is distribution of a
set of spatial objects, but as a whole it has its own properties.
Analogous to population by Pierre Gagnon, Yvan Bedard and
Geoffrey Edwards (1992), for a spatial scene, its primary
changes have location (position) change, direction (orientation)
3.2 Detailed Time Semantics
In time dimension, time semantics are enriched by three time-
varying patterns, i.e., discrete change, stepwise change and
change, shape change and size change (cumulative size change
or population size change) and change of spatial distribution
type (regular, random or grouped distribution), as shown in
Figure 4. Spatial scene change is usually caused by changes of
local objects.
continuous change (Figure 5). Three time-varying patterns
structuralize temporal data, certainly including spatio-temporal
data.
value
value
value
▲ i
k ▲
•
•
—
• •
—
► t
► t
Discrete Change
Stepwise Change
Figure 5. Primary temporal changes
►t
Continuous Change
4 THE COGNITION OF SPATIAL CHANGES
Ecologist James J. Gibson argued that events are perceived, but
time is not (James J. Gibson, 1979). Time scale is generated by
assigning time or date numbers to a sequence of events. For
example, calendar corresponds to the rotations and revolutions
of earth. It is known that geographical space psychologically is a
large-scale space, whose configuration is apprehended by
navigating through it. Large-scale space is larger than human
body, and its configuration can’t be perceived from one vantage
point (Montello, D. R, 1993). In analogy to the line of landmark-
path-configuration cognition, we argue that human beings
apprehend changes from property, to object, to scene. This
conforms to human developmental cognitive rules of local to
overall and part to whole. The changes of essential properties or
most concerned properties result in the change of object
identification. The changes of object properties and object
identifications give arise to a scene change.
Reginald G. Golledge (1995) argued that identity, location,
magnitude and time are first-order spatial primitives, distance,
angle/direction, sequence/order and connection/linkage are
derived spatial concepts, spatial distribution (boundary, density,
dispersion, pattern, shape), correlation, overlay, network,
hierarchy and others are higher-order spatial knowledge. By
contrast, we list location, distance, direction, size and shape as
primary changes at the level of property, but categorize spatial
changes into three levels, property change, object change and
scene change. Some of our considerations are from cognition
and computation. Usually in the large-scale geographical space,
the distribution type seems to be a kind of configuration
knowledge, which formed in the course of human cognition.
More qualitatively, location seems to be a relation between
figure and background. For example, the island is at the center
of a lake. Distance is a relation between two points. For
example, the department store is near to Laval University.
Direction is recognized into three sorts: intrinsic direction,
extrinsic direction and dialectic direction (Holly A. Taylor and
Barbara Tversky, 1996). The front of a car is of intrinsic
direction. The north of America is of extrinsic direction. “The
cinema is to right of a department store” is of dialectic direction
with respect to an observer. The inherent reference frame of an
object, the reference frame of scene and observer are three
basic factors to determine directions. For an object of some size,
its location or distance or direction is often referred to its central
point, but is unrelated to its size. For example, enlarging a forest
stand, but remaining its central point, doesn’t mean the changes
of location, distance, direction of the forest stand. Shape mostly
describes the properties of boundary of an object, such as
concavity/convexity and triangle/rectangle/eclipse. Generally
corner points, curvatures, and compact ratio are used to
measure the shape of an object. In computation, we can
calculate the location of an object with its distance and direction,
or calculate distance and direction with given locations, but we
can’t decide which is prior to the other. In a parallel example, we
are not sure of computing velocity with distance and time, or
computing time with distance and velocity. That is because we
don’t know, for velocity and time, which is first perceived so far.
To some extent, we feel that three levels of change and primary
property and object changes should be somewhat cognitively
plausible.
5 CONCLUSIONS
Motivated by application requirements and current theoretical
work, this paper focus on studies of spatio-temporal semantics
modeling in spatio-temporal databases. Following the time
semantics of event-sequence, we attempt to explore primary
spatial changes possibly involved in spatio-temporal databases.
Specifically, we define spatial changes at three levels of scene
change, object change and property change. At the level of
property change, a set of primary spatial property changes
(location change, distance change, direction change, size
change, shape change), together with geometrical
dimensionality changes (point change, line change, area
change), are proposed. At the level of object change, we present
six kinds of object changes in terms of mapping relationships
between source objects and result objects. At the level of scene
change, a few of integral scene changes and distribution type
changes are posed. Proposed spatial changes, as the content of
spatio-temporal semantic modeling, are expected to bring closer
theories and applications of spatio-temporal databases.
However, our work is still at a very initial stage. Much work is
needed to do in the future, such as to seek more cognitive
evidences to verify our spatial change categorization, to add
spatial relationship changes to our change classification. It is
commonly known that topological relationship is a kind of
