Modelling and Model Building - Environmental Modelling: Finding Simplicity in Complexity - page 10

physical models are very limited in the degree of control
available to the scientist because of the enormous
expense associated with them. They are also very lim-
ited in the scale at which they can be applied, again
because of expense and engineering limitations. So,
the fact remains that, at the scale at which environ-
mental scientists work, their systems remain effectively
noncontrollable with only small components capable of
undergoing controlled experiments. However, some do
argue that the environment itself is one large laboratory,
which is sustaining global-scale experiments through,
for example, greenhouse-gas emissions (Govindasamy
et al ., 2003). These are not the kind of experiments
that enable us to predict (as they are real time) nor
which help us, in the short term at least, to better inter-
act with or manage the environment (notwithstanding
the moral implications of this attitude!). Models pro-
vide an inexpensive laboratory in which mathematical
descriptions of systems and processes can be forced in
a controlled way.
Multiscale, multidisciplinary. Environmental systems
are multiscale with environmental scientists needing
to understand or experiment at scales from the
atom through the molecule to the cell, organism
or object, population or objects, community or
landscape through to the ecosystem and beyond.
This presence of multiple scales means that envi-
ronmental scientists are rarely just environmental
scientists; they may be physicists, chemists, physical
chemists, engineers, biologists, botanists, zoologists,
anthropologists, population geographers, physical
geographers, ecologists, social geographers, political
scientists, lawyers, environmental economists or
indeed environmental scientists in their training but
who later apply themselves to environmental science.
Environmental science is thus an interdisciplinary
science that cuts across the traditional boundaries of
academic research. Tackling contemporary environ-
mental problems often involves large multidisciplinary
(and often multinational) teams working together on
different aspects of the system. Modelling provides
an integrative framework in which these disparate
disciplines can work on individual aspects of the
research problem and supply a module for integration
within the modelling framework. Disciplinary and
national boundaries, research 'cultures' and research
'languages' are thus less of a barrier.
Multivariate, nonlinear and complex. It goes without say-
ing that complex and integrated systems such as those
handled by environmental scientists are multivariate

and, as a result, the relationships between individual
variables are often nonlinear and complex. Models
provide a means of deconstructing the complexity
of environmental systems and, through experimenta-
tion, of understanding the univariate contribution to
multivariate complexity.

In addition to these properties of environmental sys-
tems the rationale behind much research in environmen-
tal systems is often a practical or applied one such that
research in environmental science also has to incorporate
the following needs.

The need to look into the future. Environmental research
often involves extrapolation into the future in order
to understand the impacts of some current state or
process. Prediction is difficult, not least because pre-
dictions of the future can only be tested in the future
(at which point they are no longer predictions). Mod-
els are very often used as a tool for integration of
understanding over time and thus are well suited for
prediction and retrodiction. As with any means of pre-
dicting the future, the prediction in only as good as the
information and understanding upon which it is based.
This limitation may be sufficient where one is working
within process domains that have already been expe-
rienced during the period in which the understanding
was developed, but when future conditions cross a
process domain, the reality may be quite different to
the expectation. Thus we often talk about projecting
into the future rather than predicting into the future, in
recognition of the fact that we are fundamentally lim-
ited to projecting our present understanding into the
future as one possible outcome rather than providing a
reliable forecast of future processes and their outcomes.
The need to understand the impact of events that have
not happened (yet). Environmental research very often
concerns developing scenarios for change and under-
standing the impacts of these scenarios on systems
upon which humans depend. These changes may be
developmental, such as the building of houses, indus-
trial units, bridges, ports or golf courses and thus
requiring environmental impact assessments (EIAs).
Alternatively they may be more abstract events such as
climate change or land-use and cover change (LUCC).
In either case, where models have been developed on
the basis of process understanding or a knowledge of
the response of similar systems to similar or analogous
change, they are often used as a means of understanding
the impact of expected events.