Modelling and Model Building - Environmental Modelling: Finding Simplicity in Complexity - page 8

in this topic and abstract from them the key elements that
might be required to build a useful simulation model.
Abstraction is a difficult skill to acquire in adults (we tend
to overcomplicate) though young children have the skill
well honed as they operate their own mental models of
how the world works before parents and teachers provide
them with alternative models. A good exercise in judging
your own abstraction skills may be carried out with a
simple piece of paper. Think of all the faces that you
know: the short round ones, the long thin ones, the Euro-
pean, African, Asian and South American ones; the ones
with beards and those without. How might we abstract
from this sea of faces a simple model for the human
face? Try that on your piece of paper. Give yourself two
minutes.
Our guess is that you made it too complex. The bare
minimum we need is a circle, dots for eyes and a upwards
facing curve for a mouth. The yellow smiley face is a
good example and is one of the most common images in
modern life. If you are not sure what we mean, do a Web
search for 'yellow smiley face'. We do not need hair, ears,
eyebrows, eyelashes or anything else to recognize this as
a face. Indeed some real faces do not have those features
(or at least they cannot be seen) so adding them to your
model as a necessary condition for recognition as a face,
reduces the generality of your model. Children are very
good at abstraction as the four year old's image of a person
in Figure 2.1 indicates: a single shape for the body, stick
arms and legs, button eyes and nose and smiley mouth.
Nothing else is needed as this is very clearly an abstraction
of the human body. An element of bias is added as for this
child the belly button is also an important component of
the human form, hence it is in the model!
Arm yourself with a spreadsheet and turn your
abstraction into numbers and simple equations. Play,
examine, delete, add, think and play some more with
the numbers and the equations. What can you learn
about the system? What still confuses? Experience of
this kind will help develop intuition and insight where
it is lacking. We present you with a series of modelling
problems on the web site that complements this topic
and going over them repeatedly will help further. The key
to successful modelling is to be able to abstract carefully
so that your model is on the one hand simple but on
the other hand realistic enough to offer a solution to the
problem at hand. Considering a cow as spherical may be
appropriate for understanding some elements of how a
cow works (Harte, 1985), but will not be all that helpful
in understanding its locomotion!

Olive Mulligan, aged 4

Figure 2.1 Children are often very good at abstraction because
they tend not to see things in the complicated ways that adults
do (or to have complex preconceptions about them). This is a
four year old's abstraction of a human - clearly recognizable, if
not detailed (Courtesy of Olive Mulligan [aged 4]).

You are not new to modelling - everyone does it!
All scientists use some form of conceptual or mental
model of the data they work with. Even data are, in fact,
models; they are simplified representations of (unob-
servable) processes, time and space, compared with the
reality, all sensors form a model of reality. For example,
a temperature sensor measures change in the level of a
column of mercury as this level is linearly related to a
change in temperature. The changing level of mercury is
an empirical model for a temperature change. (Consider
how different a digital thermometer actually is from an
analogue one using mercury.) Your whole perception of
reality is a model, not the reality itself. You are armed
with a series of sensors for light in the visible spectrum
(eyes) and certain wavelengths of sound (ears), which are
only fractions of what can be sensed. Other animals have
different perceptions of the same environmental charac-
teristics because they have different sensors, but also a
different mental model and context for decoding those
signals. There is thus little difference between modelling
and other scientific endeavours (and indeed life itself).

2.1.4 Researchingenvironmental systems

According to some, we have crossed a geological bound-
ary from the Holocene to the Anthropocene (Crutzen,