Pointers for the Future - Environmental Modelling: Finding Simplicity in Complexity - page 436

empirical approaches to these disciplines possess very
poor explanatory power. But not all models provide
the same level of explanatory power. While different
models are perhaps more suited to different methodolog-
ical contexts, we should beware of placing too high a
burden on some types of model. This issue poses some-
thing of a problem, though, when the results of our
modelling pass out of the hands of the modeller and
into the policy domain (Engelen, Mulligan - Chapters
20 and 21). Oreskes et al . (1994) have suggested that
this problem means that we should place strict limits on
the ways models are employed. According to Mulligan,
modellers should maintain a strong and continuous link
with the users - and uses - of their models. Models are
an important way of communicating our results (see
below) but we should be careful to consider that science
is a socially situated activity. As we saw in Chapter 18,
there are complex levels of behaviour and interaction
that control decisions relating to environmental ques-
tions. Ultimately, this process becomes a recursive one,
where the model results are employed within a wider
framework that then controls what sort of research is
carried out, and thus the sorts of future models that
are produced. Climate modelling, as discussed by Harvey
(Chapter 9) is a very clear example here as is that of
catchment hydrological modelling (Baird - Chapter 10;
Mulligan and Wainwright - Chapter 2).
But we should remember that the social situation is
not simply an external issue. It occurs within science
itself and the practice of modelling too. We live through
scientific 'fashions', where certain explanations tend to be
preferred over others, despite the fact that there is no clear
rationale for making one choice over another. Future dis-
coveries and methodologies may mean that either choice
is ultimately incorrect, so we should beware of becoming
too dogmatic about our explanations, and continue to
question current orthodoxies. Major advances in science
have tended to develop in this way. Preconceptions, as we
pointed out in the introduction (see also Favis-Mortlock)
are always with us; we accept them at our peril! They
may relate to disciplinary boundaries, which prevent the
rapid advancement of our understanding (Mulligan and
Wainwright - Chapter 11) or progress in communicating
it (Mulligan). There is sufficient commonality across the
modelling methodology that is carried out within tradi-
tional disciplinary boundaries for us to be able to discuss
issues and overcome the limitations posed by such myopia
and, after all, models can help us to communicate across
these very same boundaries, because the language of mod-
elling is common to them all. In environmental modelling

(a)

(b)

Figure 27.1 Comparison of (a) an 'absolute' and (b) an
abstract representation of a landscape.

the explanatory loop by providing the link between cause
and effect. Some would deem this process as tinkering
to produce the correct answer (Oreskes et al ., 1994).
While the naıve calibration of models to produce the
'right' answer against some measured value is certainly
problematic in this respect, the use of this criticism is
itself questionable, as pointed out by Young and Leedal
(Chapter 7). As noted by Haraldsson and Sverdrup, we
can learn more when the result is incorrect than when
it is correct. Models are always an approximation, and
always by limited in terms of their 'truth' to the extent
that their use does not go beyond the assumptions made
in making that approximation. The truth is out there!
In disciplines such as ecology (Fisher, Perry and Bond
and Mazzoleni et al . - Chapters 12-14) and geomorphol-
ogy (Brazier - Chapter 15, Hergarten - Chapter 16, van
der Beek - Chapter 19), this explanatory loop enables us
to tackle difficult questions relating to the link between
process and form (see below). Without this link, most