Environmental Engineering Reference
In-Depth Information
100
90
80
70
60
50
40
30
20
10
0
0
10
20
30
40
Years
0.05 0.1 0.15 0.2
Fig. 5.3 The results of varying the hunting mortality in the deer model, shown for
4 hunting mortality rates, 0.05-0.2. The population starts at the unhunted
equilibrium. Tip for plotting graphs—keep them simple and clear, with only a few
lines so that they are legible in black and white.
values. For example, in the deer model, the harvest rate should vary between zero
and a rate high enough to cause extinction of the population. This ensures that you
observe model behaviour over a wide range of possible scenarios, giving you a good
intuitive understanding of the model. Plot the results up (Figure 5.3).
Already these analyses can tell you something about the sustainability of the sys-
tem—how it responds to hunting, for example, or to price changes in the hunter
model. They also confirm that you have properly tested your model, because you
should be able to predict the shapes of the curves you produce, based on experience
and the shape of the underlying functions.
However, in reality parameters do not vary independently, one at a time. In fact
it is likely that variation is correlated in space and time—if there is a bad winter it
is likely to affect all age classes and all the populations in the region, and to impact
on both survival and fecundity. One simple approach to this issue is to simulate
baseline, best case and worst case scenarios using the feasible range of values
that each parameter can take. The baseline scenario represents your best guess
at the parameter values, the best case has all the key input parameters set at the end
of the plausible range that gives the best outcome for your parameter of interest,
and the worst case has the input parameters set at the other end of the range.
This is useful for exploring the effects of our ignorance about parameter values
(i.e. observation error) on model outcome.
A more sophisticated alternative is to investigate the sensitivity of the model
to variation in a single parameter (say adult survival) in the context of the other
parameters, by picking values for each parameter independently at random from
their plausible range (Kremer 1983) and repeating this randomisation many times.
Search WWH ::




Custom Search