Cellular automata and agent base models for urban studies: from pixels to cells to hexa-dpi's - Urban Remote Sensing: Monitoring, Synthesis and Modeling in the Urban Environment - page 313

Modelling Environment T1

Agent type A

Agent type B

Attributes:

Agent A - likes to cluster

Agent B - prefers not to cluster

Decision tree:

Agent A - dies if < 2

Agent A - prospers if >4 up to 7

Agent B - clusters if = 4

Agent B - prefers not to if r <3 or >5

Trigger: vital space 0-1 units of distance

Modelling Environment T2

FIGURE 22.4 The agent's environment.

spatial and a-spatial approaches in the same model, but they can
alsobe used as helpful tools to calibratemodels. For example, Sud-
hira et al . (2005) proposed anAB-CA framework for urban sprawl
simulation, where the ABM is used to define the transitional rules
of the CA. Wu and Silva compared 11 recent AB-CA hybrid sys-
tems in urban studies (Wu and Silva, unpublishedwork). In these
11 models, agents usually represented the behaviour of pedes-
trians, vehicles, drivers, and passengers in urban traffic systems,
or householders, stakeholders, and residents in urban systems.
Geographical changes were represented by CA. Nevertheless, the
spatial context was not provided in an explicit way, and no
attempt is made to merge or integrate the data structures.
While there are not many loose coupled models explor-
ing individually CA, ABM and GA. Wu and Silva (2010)
already implemented it in a fully coupled CA-ABM/GA (model:

DG-ABC). the authors present an integrated model that incor-
porates ABM, CA and a GA in order to include both spatial and
a spatial dynamics supplying a new solution for urban studies. In
the model (DG-ABC stands for ''developing genetic-agent based
cells''), the social economic behaviors of heterogeneous agents
(resident, property developer, and government) will be regulated
by GA and the theory of planning behaviour (TpB).
A simplification of integration of CA-ABM-GA models can
be seen in Fig, 22.5. Agents and cells are still confined to the
seed-matrix and the modeling environment, spatial/temporal
synchronization of cells and agents allows adjusting the vari-
ability in scales/time (i.e., agent's actions happening in minutes
and action in cells of 100 ha changing every year). In these
synchronization tables horizontalinteractionpaysapivotalrole,
as it will be the most favourable environment for cells and agents