Geology Reference
In-Depth Information
a considerable amount of knowledge can be trans-
ferred from our work with traditional models (e.g.
cover factors).
However, if the user wants to examine other
non-fluvial processes then there are associated
parameters that need to be calibrated. For
instance, if the user wishes to model soil creep,
then there are soil creep rates and other parame-
ters (e.g. the dependence of the creep rate on the
hillslope gradient) to be calibrated. Each new
physical process involves a new set of parameters.
However, a problem is that for these latter, non-
fluvial, processes the LEMs are ahead of the avail-
able field data, and applicable parameter databases
that may be used or adapted do not, in general,
exist. Sometimes the models can be fitted to nat-
ural landforms to infer parameters, although
there may be questions about the reliability of
using natural materials as analogues of manmade
materials. It is to these calibration issues that
this section is (briefly) devoted.
In many of the applications that the authors
have been involved in by the time there was an
interest in doing LEM simulations, there had
already been a considerable amount of erosion
modelling done with traditional models. This
meant that there was a set of parameters for a tra-
ditional erosion model that had already been fit-
ted to the site. Accordingly the LEM could be
fitted to the simulations of these pre-existing
models.
Briefly, the calibration process is as follows.
The LEM is run for a year. The LEM parameters
are adjusted so that the erosion from LEM matches
the pre-existing model. This use of the pre-exist-
ing model is advantageous for two reasons. Firstly,
it makes use of any pre-existing knowledge base at
the site. Secondly, it demonstrates that the LEM
generates results that are consistent with a pre-
existing model under the same experimental con-
ditions (i.e. when the landform is not significantly
evolving, assuming that a year is a short time in
the evolution of the landform). This second result
cannot be underestimated because it highlights
that any inconsistency between failure mecha-
nisms in the LEM and traditional simulations is a
result of the landform evolution, not differences
25
20
15
10
5
0
-0.1
0
0.1
0.2
0.3
0.4
0.5
0.6
α
Fig. 18.7
The net erosion at the bottom of a concave
slope (site 1 in the text) for a range of concavities
a
(based on data in Hancock
et al
., 2003).
held constant (i.e. the height of the above-ground
structure was held constant) it was possible to
double the average slope of the hillslope, but it
was still possible for a steeper concave hillslope
to have less erosion than a flatter planar slope.
The reason for this was partly the better perform-
ance of the concave hillslope and partly because
the steeper hillslope was of shorter length because
the height is constant.
18.4
Calibration of Landform
Evolution Models
The calibration of an LEM for a specific site
involves many of the same problems as faced
when calibrating a traditional erosion model (see
Chapter 3). For the fluvial erosion component of
the model there are the soil cover and treatment
effects, the grading of the soil material and the
erosivity of the rainfall. The similarity between
LEMs and physically-based erosion models is that
the underlying physics is generally the same
(although the details will depend on the LEM), so
