Geology Reference
In-Depth Information
consider the integrated effect of a range of erosion
processes at the catchment scale is reviewed by
de Vente & Poesen (2005). These models apply
factor scores to each process to quantify an index
which is related to an erosion risk class or sedi-
ment yield and, therefore, unlike SHETRAN, are
not physically-based.
A full comparison of SHETRAN with its sister
model, MIKE SHE, is not presented in Table 14.2
as the models have a common heritage. Both
models simulate basin hydrology and sediment
yield, although some of the components have a
different basis. The principal difference between
them is that SHETRAN has perhaps seen greater
process development (e.g. landslide erosion, bank
erosion) and has more generally been used in
research studies (but also for applied work), while
MIKE SHE has been developed for industry users
and is widely used for applied work (but also for
research studies).
Advances in model design over the last two
decades have tended to be incremental rather
than fundamental. For example, the more recent
models represent the same broad erosional proc-
esses but tend to include greater process detail,
such as macropore flow, depression storage and
representation of microtopography in InHM, and
soil crusting in EUROSEM and LISEM. More fun-
damentally, InHM's use of a finite element
scheme for spatial discretization enables linear
features such as roads to be represented (Heppner
et al
., 2007). Such features are often important
sediment sources (e.g. Reid & Dunne, 1984;
Grayson
et al
., 1993) but are not easily incorpo-
rated in the finite difference or grid square
schemes used by earlier models. Another advance
is that, compared with the early developments,
most models now make use of Geographical
Information Systems (GIS), either directly in their
design or in the process of parameterization.
Despite model developments, many of the dif-
ficulties recognized in the application of the ear-
lier physically-based models remain unsolved or
only partially solved:
●
the data assembly and detailed considerations
required for model parameterization are, as pre-
sented by Heppner
et al
. (2007) for InHM, remi-
niscent of the early SHE applications (e.g.
Bathurst, 1986; Jain
et al
., 1992). Remote sensing
data hold considerable promise for parameteriza-
tion in the future (e.g. King
et al
., 2005) but are
not yet able to characterize crucial soil properties
and conditions (e.g. the hydraulic property curves
or moisture content);
●
most of the models represent the ease with
which the soil can be eroded using an erodibility
coefficient but at present these coefficients can-
not be evaluated from a directly measurable soil
property and must therefore be calibrated;
●
the uncertainties associated with parameteriz-
ing physically-based, spatially-distributed models
(such as the potential for multiple acceptable
parameterizations) are only partially addressed by
the use of uncertainty envelopes and effective
parameter values, as discussed above;
●
spatially-distributed models are still only rarely
tested for their ability to represent basin internal
as well as outlet responses, leaving open the pos-
sibility that an apparently good outlet simulation
may be based on compensating internal errors and
limiting confidence in the ability of the model to
be extrapolated to wider ranges of conditions; and
●
in any model where the erosion and sediment
transport simulation is driven by the output from a
hydrology or flow hydraulics simulation, the accu-
racy of the erosion simulation is heavily dependent
on the accuracy of the hydrology simulation.
In conclusion, SHETRAN still compares favou-
rably with other popular models, especially in its
ability to simulate the impacts of land use and cli-
mate change at a range of basin scales and as a
function of several broad erosion processes. Of the
more recently developed models, InHM is probably
the closest to SHETRAN in its design (e.g. for inte-
grated surface/subsurface response) but also incor-
porates important new developments (e.g. a finite
element discretization scheme).
14.7
Experience of Model Application
The practical problem under consideration is the
use of the model to support the development of
strategies for minimizing the adverse effects of
