Information Technology Reference
In-Depth Information
ing's decision making. However, the complexity of urban growth determines that the
classic CA must be modified in order to deal with practical issues (the details are
described in
[4]
). In this paper, we develop a modified CA model for understanding
the spatial and temporal processes of urban growth based on dynamic weighting con-
cept and project-based approach to be described below.
2.1 Temporal Heterogeneity (Dynamic Weighting)
[13]
applied logistic regression method for modelling land development patterns in
two periods (1979-1987 and 1987-1992) based on parcel data extracted from aerial
photos. They found that the major determinants of land development have changed
significantly, e.g. from proximity to inter-city highways to proximity to city streets.
Likewise, if we shrink the long period (1979-1992) to shorter period such as 1993-
2000 and also from the whole city to smaller part. The same principle should be
working as well. As a consequence, the factors influencing local growth should be
assigned with dynamic weight values.
Obviously temporal pattern from time
t
1
to
t
n
,
is influenced by highly complicated
spatial and temporal processes. However, similar patterns can result from numerous
different processes. As a
consequence, the understanding of process is more important
than that of pattern. Pattern is only a phenomena but process is the essence. The inter-
action between pattern and process is a non-linear iteration function like other phe-
nomena: fractal, chaos etc. which are typically represented by non-linear iteration
function (eq.1).
X
t+1
=f (x
t
)
(1)
In the case of urban growth, temporal complexity might be indicated by:
•
Compared with major roads, minor roads especially in new zones, which are also
new development units, may have certain time delay in affecting local growth, i.e.
between
T
0
and
T
n
, not immediately from
T
0
;
•
The spatial impacts of various factors such as road, center, rail are not simultane-
ous temporally in effecting local growth;
•
Neighbourhood effects may suffer from temporal variation, for example, it may
be stronger in
T
0
than in
T
n
,
or vice versa.
Figure 1 is only an example of temporal complexity involved in urban growth.
T
1
,T
2
,T
3
indicate time series. The same spatial pattern results from three (in reality,
more) distinguishing temporal process, which reflect the spatial and temporal interac-
tions between road-influenced and center-based local growth. The arrows indicate the
trend of temporal development, from which we can define them as three different
temporal processes (convergence, sequence and divergence). The basic principle be-
