Geology Reference
In-Depth Information
other erosion processes such as (ephemeral) gully
erosion, landsliding and river incision and/or
bank erosion and, importantly, also by sediment
deposition. There is therefore no direct relation-
ship between simulated or predicted erosion rates
and river sediment yield. Accounting for sedi-
ment deposition is extremely important, given
that in many cases up to 80% of the eroded sedi-
ment is (re)deposited within a short distance
(<10 km) of the sediment source (Van Rompaey
et al
., 2001; Wilkinson & McElroy, 2007). Rela-
tionships between catchment sediment yields
and soil erosion rates are further complicated by
the fact that, at the catchment scale, internal
mechanisms lead to compensatory effects: a
decreased sediment supply from the slopes due to
the implementation of soil conservation meas-
ures may lead to increased mobilization by the
river of previously deposited sediment through
bank erosion and incision, so that sediment yield
is maintained over a long time period (Trimble,
1999). The poor relationship between soil erosion
rates and river sediment yields is well illustrated
by the fact that the massive increase of soil ero-
sion due to agricultural activities has only led to
an increase of global river sediment fluxes from
the land to the ocean by ca. 20% (Syvitski
et al
.,
2005). One may therefore expect that using an
inverse approach (i.e. estimating soil erosion rates
from total river sediment fluxes) may lead to
errors; a substantial overestimation of soil ero-
sion rates on agricultural land in large catchments
may result if sediment yields are back-converted
to erosion rates using a simple sediment delivery
approach (Lal, 2003).
The above does not imply that coupling river
sediment yields to erosion rates is never possible.
In areas where soil erosion by overland flow is the
dominant sediment production process, river sed-
iment yields can indeed be coupled to soil erosion
rates if a spatially distributed approach is used
and sediment deposition is adequately accounted
for, and if the model allows for the two-dimen-
sional nature of true landscapes (e.g.Van Rompaey
et al
., 2001).
Accounting for the two-dimensional nature of
landscapes is to some extent possible by adapting
a hillslope model so that it can account for flux
divergence (on convexities) and convergence (in
concavities). By doing so it is implicitly assumed
that erosion rates and patterns display continuity
over various landscape elements, and that gully-
ing in hollows can be described using the same
process descriptions as for rill erosion on hills-
lopes. There is some empirical support for this in
the case of ephemeral gullying (Desmet & Govers,
1997). Clearly, such an approach is not possible
when sediment production in concavities is con-
trolled by processes other than those described by
the model, such as permanent gullying driven
by headcut retreat due to surface runoff and/or
groundwater seepage.
Thus, erosion models can be applied at larger
scales than the hillslope provided that appropri-
ate scaling functions are used and the results are
correctly interpreted: water erosion models may
well be used to predict spatial variations in soil
erosion rates by sheet and rill erosion and to some
extent by ephemeral gullying, but not due to
other processes. Coupling simulated soil erosion
rates to temporal variations in sediment yield is
only possible under well-defined conditions: soil
erosion should (by far) be the dominant sediment
production process in the basin, and the model
used needs to be able to simulate sediment trans-
fer and deposition within the catchment. A model
such as WATEM/SEDEM (Van Rompaey
et al
.,
2001) allows this using a steady-state approach
for areas where soil erosion by water is the only
important sediment production process. A prom-
ising approach to develop models incorporating a
wider range of processes is the use of parsimoni-
ous algorithms, as is done in the SEDNET model
(Wilkinson
et al
., 2004).
7.3
Misconceptions About Erosion Models
Soil erosion modelling is an ongoing scientific
activity: there is not only an increasing literature
on the application of soil erosion models, but new
approaches to hillslope erosion modelling con-
tinue to be developed (Wainwright
et al
., 2008;
Heng
et al
., 2009; Wei
et al
., 2009). It is therefore
