Geoscience Reference
In-Depth Information
of behaviour, is available to every landowner and developer, and they estimate demand for lands,
establish land price, buy and decide on the new land uses, by pure economic reasons. In this case, a
modeller can translate the rules of agents' behaviour into CA state transition rules.
However, the rationality of landowners and developers is never perfect. They may ignore part
of the available information, make incorrect assumptions about the importance of the factors that
influence land prices or make decisions for personal reasons, for example,
I do not like the guy who
wants to purchase my land
. These forms of behaviour cannot be directly translated into CA transi-
tion rules. The more we deviate from the decision maker as a perfectly rational economic creature,
the less we are able to translate agent behaviour into state transition rules.
The motto of AB modelling is simple: let us explicitly reflect human decision making in the
model. Concerning land-use modelling, instead of racking our brains regarding how to translate
the behaviour of landowners and developers into CA state transition rules, let us explicitly represent
them, together with the rules of their behaviour, in the model.
Historically, several steps can be marked on the way from CA to AB modelling in geography; I refer
the interested reader to Chapter 5 of the topic by Benenson and Torrens (2004) entitled
Geosimulation:
Automata Based Modeling of Urban Phenomena
for further details. In this chapter, I focus on the con-
temporary geosimulation view of AB modelling and illustrate the conceptual views with two examples,
one positioned at the abstract end and one at the real-world end, of the AB model spectrum.
9.1.7 t
wo
e
xaMPleS
: e
thnic
r
eSidential
d
ynaMicS
and
P
arking
S
earch
I illustrate a methodology for AB modelling using two models. The first is the famous Schelling-
Sakoda model of residential segregation (Sakoda, 1971; Schelling, 1971), which is usually called
just the
Schelling model
. Readers with some knowledge of AB will be familiar with this model,
which I exploit here for the purpose of presenting the deeper details of AB modelling capabilities.
The reader can reproduce most of the model experiments that I present in the following with the
NetLogo application developed by Dr. Erez Hatna at the Center for Advanced Modeling (CAM) in
the Social, Behavioral and Health Sciences, Johns Hopkins University, which is available at http://
erezhatna.com/SchellingModel/Model.html.* The second example is the PARKAGENT model of
searching for parking spaces and then parking in the city (Benenson et al., 2008; Levy et al., 2012).
The Schelling model investigates an abstract idea that is insufficient, in itself, for expanding
towards real-world residential distribution. The PARKAGENT model also starts with an abstract
idea but is developed as a spatially explicit planning and management tool and fully exploits the
current access to high-resolution GIS data and results from field surveys.
Both models were constructed according to a common methodology and are presented following
the standard
overview, design concepts and details
(ODD) protocol (Grimm and Railsback, 2012).
Let us start with defining agents and objects and assigning their properties and then define the rela-
tions between the model agents and the objects and formulate the behaviour of the agents. After the
models are defined, let us present selected results of their dynamics.
9.2 ASSIGNING PROPERTIES, RELATIONSHIPS AND BEHAVIOURS TO AGENTS
The decision about the kinds of agents in the model, and the properties, relationships and behav-
iours of the agents, are all made with respect to the aim of the model. There are numerous ways
to formalise the same geographic system, and different formalisations will result in different AB
models. Any formalisation, however, starts with the assignment of agents and agent properties and
establishes the relationships between the model agents and the objects. Technically, this process
builds on the rules of relational database theory.
*
Depending on the Java version on your computer, it may be necessary to include this applet in a
list of exclusion. Open Configure Java and go to Security
→
Exception Site List to do this.
