Geoscience Reference
In-Depth Information
other developments in IT. If GIS is mainly about digital map information, then GC is about using it
in many different application areas within which the focus is no longer on the original GIS compo-
nents. Nor is GC about evolving new or better data structures for use within GIS or about any of the
GIS research agenda. To put it more bluntly, GIS is merely a database infrastructure which is nice
to have but which is lacking in any science or theory other than the measurement science on which
it is based. GC is not just an add-on to GIS; in fact, it is not really part of it at all.
In essence, GC is concerned with the application of a computational science paradigm to study
all manner of geo-phenomena including both physical and human systems. It probably captures
quite well the broad type of methodological approach that an informed computational physicist
or chemist or aeronautical engineer would adopt if asked to suggest ways of studying subjects as
diverse as river systems to human behaviour. It is not just data mining and it is not necessarily
theory-free; indeed, both extremes of inductive and deductive approaches can be studied via a com-
putational paradigm. GC is all about the use of relatively massive computation to tackle grand chal-
lenge (viz. almost impossible to solve) problems of immense complexity. However, a key feature
is what is termed problem scalability. You start by tackling small and simpler versions of a more
complex problem and then scaling up the science as and when either the HPC systems catch-up or
knowledge of algorithms, models and theory start to show signs of being able to cope. In many areas
of social science and human geography, so great has been the scientific neglect that we can no longer
think about a 1- or 2-year time frame but need a 10-50-year period. Long timescales have not put off
other sciences, for example, mapping the human DNA or fusion reactor physics or laser physics. All
you need is a bold but clear vision of what the end goal is and then a path that connects where you
are at present to where you want to be in some years time. However, not all GCs need only be
big
science. This is fortunate because
big
science is still a wholly alien concept in the social sciences.
Much progress can be made far more readily and with far less risk on small projects. The message
is start small but think big.
However, not all researchers appear to agree with these definitions of GC. The problem appears
to be that most commentators have focused on the content of the various conferences as a means of
defining what it is that
GeoComputationalists
study and hence define the subject of GC. This is not
a particularly good way of developing a definition; for example, the definition of geography based on
the titles and content of the papers presented at the annual RGS/IBG or AAG conferences would at
best be confused and probably somewhat weird! It would be far better to think about the definition
in a more abstract manner.
So far the most detailed study of the subject of GC is by Couclelis (1998a,b). These two essays
contain a delightful mix of fact, useful comment and suggestion blended with hints of confusion and
flashes of future optimism. Couclelis has thought most about the meaning of the term. She starts by
defining
GeoComputation
as 'the eclectic application of computational methods and techniques to
portray spatial properties, to explain geographical phenomena, and to solve geographical problems'
(Couclelis, 1998a, p. 17). She has observed from a study of the content of previous GC conferences
that '… GeoComputation is understood to encompass an array of computer-based models and tech-
niques, many of them derived from the field of artificial intelligence (AI) and the more recently
defined area of computational intelligence (CI)' (Couclelis, 1998a, p. 18). According to her, the key
question now '… is whether GeoComputation is to be understood as a new perspective or paradigm
in geography and related disciplines, or as a grab-bag of useful computer based tools'. Longley also
hints at a similar degree of confusion when he writes: '… GeoComputation has become infinitely
malleable…' (Longley, 1998b, p. 83). At issue here is whether or not there is something special and
new to GC. The short answer is
yes
for reasons that have already been stated. However, if you can
accept that GC is a form of computation science applied to spatial or geo-problems (theory and
data), then much of what Couclelis (1998a,b) is concerned about falls by the wayside.
Couclelis talks about an uneasy relationship with mainstream quantitative geography '… as
evidenced by the relative dearth of GeoComputation-orientated articles and topics in main quantita-
tive geography journals and texts' (Couclelis, 1998a, p. 19). She also adds that 'GeoComputation has
