Geology Reference
In-Depth Information
respectively, when average annual values were
compared, showing that, in terms of predicting
long-term average values, the USLE outperformed
more recent, sophisticated models. When looking
at individual annual erosion values, one should
expect the difference between the different mod-
els to reverse. However, Tiwari
et al
. (2000) found
that this was not the case, as model efficiencies
were 0.58, 0.60 and 0.40 for USLE, RUSLE and
WEPP respectively. Thus, at higher temporal reso-
lution the USLE performed similarly to the
RUSLE and somewhat better than WEPP.
With respect to event-based erosion modelling,
nearly all studies that have been carried out hith-
erto demonstrate that the application of erosion
models to single events often results in very large
errors (e.g. De Roo & Jetten, 1999), suggesting that
even dynamic, process-based models should not
be used or evaluated at the time-scale of a single
event. However, they offer a key advantage in
comparison with statistical models when it comes
to predicting the dynamics of erosion processes, in
that they can generate distributions of individual
events; it is then not only possible to compare the
predicted, average values, but also the simulated
distribution of events, so that insight can be gained
into what type of event causes most erosion. We
may then anticipate how this distribution may
shift with a changing climate or a changing agri-
cultural system (Baffaut
et al
., 1998).
deposition rates calculated from radionuclide
inventories (mostly
137
Cs) or soil truncation.
Integrative measures of soil redistribution pro-
vide valuable information about the intensity of
soil redistribution rates, but both rates and pat-
terns may be affected by other processes than
sheet and rill erosion. On arable land, tillage ero-
sion, ephemeral gully erosion and soil loss by root
crop harvesting can all significantly contribute to
soil redistribution, and their impact may even
dominate the overall soil redistribution pattern.
Data from integrative tracers can therefore not be
used directly to validate a model that was specifi-
cally developed to simulate sheet and rill erosion
such as the USLE or WEPP. While in past studies,
137
Cs-derived erosion rates and water erosion
model predictions were sometimes directly com-
pared since researchers were not aware of the
importance of tillage erosion or root crop harvest-
ing (e.g. Busacca
et al
., 1993), recent literature
shows that researchers have become increasingly
aware of the fact that, given the importance of
other soil redistribution processes, all processes
contributing to soil redistribution need to be
accounted for when interpreting the results of
tracer and/or soil truncation studies (e.g. Belyaev
et al
., 2005).
7.2.3
The spatial scale
There are several reasons why spatial scale is
important for erosion model applications. Firstly,
different processes dominate erosion at different
scales. For instance, in Europe tillage erosion is
often the dominant erosion process at the field
scale (Govers
et al
., 1994). Yet, at the catchment
scale tillage does not cause any net soil move-
ment as all soil that is eroded by tillage is rede-
posited within the same field. A possible solution
to this is to view the landscape as a hierarchically
structured set of response units, each with their
own dominant processes described by relevant
submodels (Cammeraat, 2002). A second reason
why scale issues are important is that model
parameters are often obtained from measurements
at a scale that is different from the one at which
the model is finally used: model parameters, such
7.2.2 The processes
Obviously, a model should only be used to simu-
late those processes that it describes. Most water
erosion models describe so-called 'sheet and rill'
erosion, thereby excluding the effects of soil
redistribution by other processes. A valid test of
such a model can therefore only be carried out
using data on sheet and rill erosion rates. The lat-
ter appears to be obvious, yet misapplications
occur mainly when direct data are not available
and sheet and rill erosion rates are equated to
total soil redistribution rates which are measured
by looking at an integrative measure of soil
redistribution. The latter may, for instance, be
the degree of soil redistribution or soil erosion/
