Geoscience Reference
In-Depth Information
emphasis on computation as a problem-solving device grounded in a scientific approach. It seeks to
exploit the new opportunities for modelling, simulation and analysis of human and physical systems
that major new developments in HPC have created. In trying to achieve this function, it is quite nat-
ural that GC should also seek to make good use of both old and new tools, particularly those emerg-
ing from AI and CI backgrounds that are computationally based. However, it has not really been
suggested anywhere that before you qualify as a
GeoComputationalist
, you need simultaneously
lots of data, a lack of theory, massive amounts of HPC and heavy use of the latest AI or CI tools.
To summarise, GC is
•
Not another name for GIS
•
Not quantitative geography
•
Not extreme inductivism
•
Not devoid of theory
•
Not lacking a philosophy
•
Not a grab-bag set of tools
1.6 GEOCOMPUTATION RESEARCH
In many ways, the current GC research agenda reflects and evolves around that of its constituent
parts. The most important of these are HPC, AI and its more generalised expression as CI and a
global GIS that has stimulated the appearance of many large spatial databases. However, there is no
single dominant factor and others of more traditional importance probably need to be added, such
as statistical techniques, mathematical modelling and computer simulation relevant to a geographi-
cal context.
HPC is a most significant technological development. As computers become sufficiently faster
and offer sufficiently large memories, HPC really does provide new ways of approaching geography
based on a GC paradigm, which encapsulates the flavour of a large-scale computationally intensive
approach. It involves both porting and moving current computationally intensive activities onto
HPC platforms, as well as the application of new computational techniques, algorithms and para-
digms that are dependent upon and can take particular advantage of supercomputing.
However, it is once again important to stress that it is much more than just supercomputing or
HPC for its own sake. The driving factors are threefold: (l) developments in HPC are stimulating
the adoption of a computational paradigm to problem-solving, analysis and modelling; (2) the need
to create new ways of handling and using the increasingly large amounts of information about the
world, much of which is spatially addressed; and (3) the increased availability of AI tools and CI
methods (Bezdek, 1994) that exist and are readily (sometimes instantly) applicable to many areas
of geography suggesting better solutions to old problems and creating the prospect of entirely new
developments. GC also involves a fundamental change of style with the replacement of computa-
tionally minimising technologies by a highly computationally intensive one. It also comes with
some grand ambitions about the potential usefulness that may well result from the fusion of virtu-
ally unlimited computing power with smart AI and CI technologies that have the potential to open
up entirely new perspectives on the ways by which we do geography and, indeed, social science. For
instance, it is now possible to think about creating large-scale computing machine-based experi-
ments in which the objects being modelled are artificial people living out their artificial lives as
autonomous beings in computer-generated artificial worlds (Dibble, 1996). HPC provides a labora-
tory within which many geographical and social systems can be simulated, studied, analysed and
modelled; see also Gilbert and Doran (1994) and Gilbert and Conte (1995). A fusion of microsimu-
lation and distributed autonomous intelligent agents is one way forward. The hardware, software,
data and core algorithms largely exist. Perhaps the greatest obstacle is the difficulty of acquiring
research funding for revolutionary ideas far beyond the conventional and then of gaining access to
sufficiently powerful HPC to make it practicable.
