Geoscience Reference
In-Depth Information
of managing the underlying computer hardware, freeing e-Research applications from being tied
to particular computer clusters. The following section will discuss how these new resources, and
e-Research in general, are changing the way that researchers conduct GC.
10.7 E-RESEARCH AND GEOCOMPUTATION
The early years of GC are largely associated with GIS and the ability to manipulate spatial data,
leading to developments in spatial statistics and multilevel modelling, systems analysis techniques,
mathematical modelling (including SIMs), geodemographics and microsimulation. Developments
in the United Kingdom were certainly stimulated by the Regional Research Laboratory (RRL) pro-
gramme of the 1980s and early 1990s, paralleled by the National Center for Geographic Information
and Analysis (NCGIA) in the United States. The last decade has seen the emergence of a number of
self-supporting laboratories including the Centre for Computational Geography (CCG, Leeds), the
Centre for Advanced Spatial Analysis (CASA, London), the National Centre for GeoComputation
(NCG, Maynooth, Ireland), the Spatial Simulation for Social Sciences (S4, Paris) and the Cities
Centre (Toronto). As noted earlier, these trends are spreading at least as far as China and Australia.
At the same time, a continued progression of computational capacity and e-Research capabilities
has allowed the character of this work to evolve and mature. In this section, we will consider the
major dimensions of this evolution, that is, a progression from the aggregate to the individual and
movement from the inductive to the deductive, from the strategic to the virtual and towards multi-
dimensional visualisation.
10.8 FROM THE AGGREGATE TO THE INDIVIDUAL
Individual-based analytical simulation models have been around for many years. Guy Orcutt's pio-
neering work in economic microsimulation is now well past its 50th birthday (Baroni and Richiardi,
2007), and geospatial applications of this concept can be traced back nearly as far (Hagerstrand,
1957). Discrete choice models of individual behaviour were well understood and theorised by the
late 1970s (Domencich and MacFadden, 1975), and they were pushed out into the geographical
community by Neil Wrigley amongst others (Wrigley, 1982).
The most significant advances in individual-based modelling can almost certainly be related
to the advent of multi-agent simulations (MASs). In the most iconic of the early applications of
MAS, Thomas Schelling developed his model of housing market behaviour to demonstrate and
build an understanding of the phenomenon of segregation in cities (Schelling, 1969). Josh Epstein
and Rob Axtell took these ideas to a new level in the Sugarscape model, combining notions of
migration with epidemiology, spatial economic performance, crime and conflict and even the idea
of social and biological evolution itself (Epstein and Axtell, 1996). In all of this work, there is a
strong focus on notions such as emergence, self-organisation, path dependence and equifinality
which adds a significant new complex systems twist to conventional spatial economic and demo-
graphic models.
Nevertheless, much of the action in this early work takes place against an abstract and stylised
spatial backcloth - in Schelling's case, a simple linear
city
and for Epstein and Axtell, a hardly less
developed grid-based geometry. Typically, this abstraction is justified by a concern for pedagogic
understanding and insight in preference to such unrefined and worldly considerations as predictive
power, accuracy and policy relevance. Although it may be argued that such stylised models depend
on a relatively sophisticated view of software agents, they place few demands on computational
resources and still less on data. Much more interesting, perhaps, are a number of more recent appli-
cations that bridge to more intensive real-world contexts.
For example, in the work of Ferguson et al. (2005), the population of a whole country (Thailand,
with 85 million people) is translated to an epidemic simulation in which cases of Avian bird
flu are capable of growing exponentially and diffusing across an extensive geographical area.
