Misapplications and Misconceptions of Erosion Models - Erosion Modelling - page 128

factors controlling deposition rates and sediment
enrichment has important practical applications
as well, since it will, for instance, help us to
decide which measures in a catchment will be
most effective in reducing not only total sedi-
ment yield, but also the movement of different
sediment size classes, even if we cannot always
make a correct quantitative prediction.
The focus of erosion modelling on simulating
correctly the total quantity of soil movement has
also led to neglect of what is actually eroding - the
soil. Soils are an important part of ecosystems, pro-
viding services to human society. The potential of
soils to provide these services depends on their
quality, which is undoubtedly affected by erosion.
The most important service provided by agricul-
tural soils is the production of crops, and one would
therefore expect that over the years a large observa-
tional database would have been established on the
relationship between erosion and crop productiv-
ity. Surprisingly, this is not the case: although some
data have been collected, many of them were
obtained using rapid desurfacing, a method that is
not capable of simulating the gradual removal of
soil by erosion and leads to strong overestimation
of erosion effects on agricultural productivity
(Bakker et al ., 2004). Other data were collected
under more realistic conditions, but do not acc-
ount for the confounding effect of landscape posi-
tion: eroded soils are located in specific landscape
positions (steep slopes, convexities) that may be
inherently less productive than concavities and
footslopes (Stone et al ., 1985; Bakker et al ., 2004).
This important hiatus can be filled by setting
up specific activities whereby the erosion-
productivity link is studied using an integrated
approach under various agro-ecological condi-
tions. The latter is a necessity, as the effect of ero-
sion on crop productivity is known to depend on
soil properties (Dercon et al ., 2006), but will also
depend on the agricultural system. Under high-
input agriculture the negative effects of erosion
on nutrient supply are compensated for by fertili-
zation, so that the effect of erosion on crop yield
is ultimately controlled by the effect of erosion
on the ability of the soil to store water and make
it available for plant growth (Bakker et al ., 2005).

However, in low-input agriculture, the effects of
nutrient losses may indeed have strong effects on
crop productivity (Dercon et al ., 2006).
Erosion-soil interactions are not limited to
soil productivity. Soil properties can dramatically
change due to erosion effects: for instance, if rock
fragments are present, erosion rates may be dra-
matically reduced on a decadal time-scale due to
the concentration of protective rock fragments at
the surface. As the reduction of erosion rates will
be greatest in locations where initial erosion rates
are highest, this will also lead to significant
changes in erosion patterns in the landscape
(Govers et al ., 2006). Erosion will also lead to the
redistribution of soil organic matter which has
profound implications for the carbon exchange
between the soil and the atmosphere (Van Oost
et al ., 2005b). Such long-term interactions need
to be explicitly accounted for in models designed
to simulate erosion over longer time spans.
Further model development efforts will not
result in the ultimate universal soil erosion
model. Depending on the research objectives, dif-
ferent approaches are necessary. Modelling with-
in-field carbon or nutrient redistribution by
erosion over long time-scales does not require a
dynamic model, but can be studied using a steady-
state approach (Van Oost et al ., 2003). On the
other hand, studying sediment delivery to water-
courses where event size may be crucial requires
the use of a high temporal resolution model (Van
Oost et al ., 2004; Fiener et al ., 2008). Thus, sev-
eral types of models will also co-exist in the
future in order to address different issues.

7.4 Conclusions

Misapplications and misconceptions of erosion
models can be discussed from different perspec-
tives. Here we took a rather pragmatic viewpoint
of this issue and discussed how misapplications of
models may be related to applications of models
at inappropriate spatial and temporal scales and/
or to a spatial domain where processes other than
those represented in the model contribute signifi-
cantly to sediment redistribution. Upscaling a