Environmental Engineering Reference
In-Depth Information
Many ecological situations depend on very specific local conditions and are hence
difficult to reproduce: an example for this would be a pronounced spatial heteroge-
neous distribution which is partially amplified through biotic activities. Or as in
standard models of plankton algae population dynamics: it is difficult to meet all
responses to the influencing conditions and thus to simulate specific site conditions.
However, the quality of a model could be assessed by comparing the frequency of
certain situations (e.g. population outbreaks or occurrences of the maxima of a
particular species) in a larger time span and analyse whether repeated model applica-
tions with randomly varying input would lead to comparable distributions. Thus, for
many complex situations it is possible to develop reasonable models which might not
be able to directly describe a particular development of a process but at least capture
and formalize certain general characteristics. In this case, validation procedures
which operate on higher-level characteristics can be of great help (see e.g. Sect 12.4).
23.4.1 Scenario Calculations and the Validity Window
In situations where it is not possible to predict precise developments it can be useful
to investigate scenarios (see Chap. 22). In the context of ecological modelling this
means to consider a set of assumptions and then use a model to calculate the
implications of the considered assumptions. In other chapters you find interesting
examples for scenario calculations, e.g. in Chap. 2 a model shows the implications
of carabid beetle dispersal. This model analyses the colonization success of beetles
for several different scenario settings with different stepping stones in an adverse
landscape surrounding (Fig. 2.6). Another example relates to the nutrient retention
capacity simulated under the assumption of different agricultural processing inten-
sities (Fig. 2.7, nitrogen leaching in industrial agriculture and in green agriculture
simulated with the WASMOD modelling system).
A seemingly negative result of a scenario calculation would be to find out that
certain expected model results are not compatible or consistent with particular input
conditions. This frequently limits the window of expectation in an interesting way,
as it helps us to understand more details of the investigated ecological processes and
relates them to a larger context.
In other cases of model validation we can find out how far the conditions can be
varied before the output becomes ecologically unreasonable. For temporal and spatial
extrapolations it is important to note that, again, a
ceteris paribus
condition is used.
This means that all basic relations and processes which are not explicitly changed in the
scenario setup are kept unchanged, and that all external influences which are consid-
ered only implicitly, remain the same, hence ignoring potential changes in processes,
reactions of organisms or structural relationships. It is in the responsibility of the
modeller to consider and document this thoroughly when extrapolating the results.
For complex ecological situations it is interesting to see that to a surprisingly
large extent the degree of freedom is relatively limited. Selecting biologically
