Geology Reference
In-Depth Information
2
Model Development:
A User's Perspective
R.P.C. MORGAN
National Soil Resources Institute, Cranfield University, Cranfield, Bedfordshire, UK
2.1 Introduction
difficulty that most models are not accompanied
by clear statements of the purposes and conditions
for which they were designed, their limitations or
indicators of the accuracy of their output.
This chapter discusses how the user might deal
with these issues. It does so by proposing that
users should adopt the same procedures in analys-
ing their problem as model developers adopt in
constructing their models. By understanding how
model developers operate and following a com-
mon methodology, users will be better equipped
to decide what questions need to be asked when
selecting a model to meet their specific objectives.
These questions can then be formulated into a set
of design requirements that a model must meet in
order to be suitable. Users will also gain an appre-
ciation of whether they will be able to operate the
model software unaided, or whether they will
need to seek expert advice in how to set up the
model to meet their requirements and interpret
the results. Table 2.1 sets out the steps followed
by model developers and lists the main points
that need to be considered at each stage.
The last 40 years or so have witnessed the devel-
opment of a very large number of erosion models
operating at different scales and different levels of
complexity, with huge variations in the quantity
and type of input data required and, at least
according to the model developers, covering a
wide range of applications. A potential user of
erosion models is therefore faced with a bewilder-
ing choice when attempting to select the best
model for a particular purpose. All too often, the
choice of a model is made more difficult because
the user is unable to define the problem precisely
enough to state what output is required; for exam-
ple, whether knowledge of erosion rates is needed
as a mean annual value or for a specific year, sea-
son, month, day or storm, and if the latter,
whether it is a storm total or a value at the storm
peak which is wanted. The user is sometimes
uncertain whether this information is needed for
a field, a particular hillslope or a catchment.
Perhaps knowledge of actual erosion rates is not
needed at all, and all that is required is an idea of
the location of erosion within the landscape or an
indication of the time of year that it is most likely
to occur. Even when the requirements are clearly
defined, the user is still confronted with the
2.2 Some Fundamentals
Any model is a simplification of reality and, for
some users, this creates an immediate theoretical
issue. How can a problem associated with erosion
in a particular location be predicted by a model
that describes erosion in a generic way? Surely
the only way to deal effectively with a problem in
