Geoscience Reference
In-Depth Information
morals and æsthetics. Second, there is the realist perspective that truly understand-
ing some entity requires deeper understanding of its structure and the properties
that change it and enable it to change. Realists such as Sayer (1992) have argued
that the language of mathematics is unable to do this (recall the discussion above
of empirical-statistical models as acausal and astructural). Finally, there is the post-
modern 'attack' that science and modelling do not hold privileged positions as
guarantors of objectivity or truth compared with other approaches, and quantita-
tive modelling is just one of many means of geographic description (Cosgrove,
1989). Science, and indeed knowledge, it is argued, are socially constructed
and, as such, are products of the social milieu in which they are created and
embedded.
Another, and related, criticism levelled at geographic modelling is that it fails to
address the important questions of geography. For example, Harvey (1989) argues
that geography is a historical discipline and that the language of mathematics and
the positivist approach are ill-suited to the development of theory in this domain
(see the realist perspective above). He argues that modelling is limited to repetitive
events (cf. Oreskes, 2003, view that prediction is only possible for repetitive systems).
Harvey questions what modelling can teach and has taught us about the important
historical-geographical shifts that he believes should be the focus of human geogra-
phy; he states (p. 212) 'those who have stuck with modelling . . . have largely been
able to do, I suspect, by restricting the nature of the questions they ask' and bemoans
(p. 213) the 'sad degeneration and routinisation of modelling into mere data crunch-
ing, numerical analysis and statistical inference
instead of careful theory building
'
(my italics). Here lies the crux of the debate: to what extent can models and model-
ling contribute to effective theory building in geography?
Conclusions
Modelling occupies a central place in geography and related disciplines, and it
continues to receive considerable attention in the geographic literature. Although
important questions remain about the ontology and epistemology of models and
modelling, models are increasingly used in environmental geography to make pre-
dictions, to improve understanding, to synthesise and integrate data and to aid in
communication. Recent developments in modelling are inextricably intertwined
with developments in technology. As new analytical approaches have been devel-
oped, new sources of data become available, and computer power has increased
and become more readily available, it has become possible to implement ever more
detailed ('realistic'?) models. However, detailed and more realistic 'mimics' are not
a panacea for the long-standing challenges of identifying appropriate representation
and scale. Detailed representation is beguiling, but 'models of this sort may provide
an unjustifi ed sense of verisimilitude' (Levin et al., 1997, p. 335). While the prag-
matic realist might see ever more detailed models as ever-truer representations, the
fact remains that the 'truer' a model, the harder it is to establish its 'truth' (Oreskes,
2003). Likewise, while detailed models may be more empirically adequate, they may
be premature and mask a lack of understanding of the entity being modelled (Frigg
and Hartmann, 2006). Alongside the development of effective tools for model evalu-
ation, fi nding the appropriate level of representational detail remains the key chal-
lenge for modellers and modelling.
