Geology Reference
In-Depth Information
Table 14.3
Plot calibrated values of the model erodibility coefficients.
Plot area
(m 2 )
Raindrop impact
erodibility coefficient (J −1 )
Overland flow erodibility
coefficient (mg m −2 s −1 )
Location
Rainfall
Notes
Reynolds Creek,
Idaho a
32.54
Artificial
(single pattern)
1.3/11.8
0.65/5.9
Function of land use
Vale Formoso,
Portugal b
167
Natural
0.13/2
1.3/20
Function of rainfall
event
North Island,
New Zealand c
970
Artificial
2.3/12.2
0.00045-0.00188
Function of season
References : a Wicks et al . (1992); b Bathurst et al . (1996); c Adams & Elliott (2006).
of winter and summer respectively. The raindrop
impact coefficient varies in the two orders of
magnitude range 0.1-10 J −1 but with the lower end
of the range perhaps more relevant to natural
rather than artificial rainfall. The calibrations of
Wicks et al . (1992) and Bathurst et al . (1996) for
the overland flow erodibility coefficient are also
in reasonable agreement. However, considerably
lower values are obtained by Adams and Elliott
(2006), possibly because their plot had a dense
cover of grass. In general the results of the plot
calibrations are consistent with physical reason-
ing (e.g. that loose tilled soil should have a higher
erodibility coefficient than compacted grazed
ground). They therefore provide confidence in the
ability of the model to represent the effect of dif-
ferences in soil conditions and land use on soil
erosion and sediment yield. On the other hand,
they also show that there can be significant vari-
ations in erodibility even at a single site, for
example as a function of season or rainfall/runoff
characteristics, and these variations need to be
accounted for.
Table 14.4 shows the erodibility coefficients
applied in a series of basin simulations. Mostly
these represent the upper and lower uncertainty
bounds, although there is some calibration or
adjustment in the light of measurements for the
Iowa and Rimbaud simulations and the Cobres
values are the same as the Vale Formoso plot
calibrations. The high raindrop impact coeffi-
cient for the Iowa sites may reflect the easily
eroded soil of an arable field area. Otherwise,
there is a tendency for the coefficient to adopt
values from the lower end of the plot calibrated
range, again perhaps a function of natural rainfall
characteristics.
14.7.2
Representation of spatial variability
Tests of the ability of SHETRAN (or other distrib-
uted models) to represent spatial variability in
soil erosion and sediment yield are very limited,
largely because of a lack of the necessary response
data. It is hard enough to find a suitable basin
with sediment transport measurements at its
outlet, let alone on a nested basis! Similarly, the
measurement of spatial variability of erosion over
more than a limited area is currently too time-
consuming to be practical, although remote-
sensing techniques hold considerable hope for
the future. Data currently depend on measure-
ments of caesium-137 activity in the soil for map-
ping patterns of erosion and soil redistribution
(e.g. Ritchie & McHenry, 1990; Walling & Quine,
1992; Higgitt, 1995). A number of studies have
explored the use of such maps for model valida-
tion (e.g. De Roo & Walling, 1994; Sidorchuk &
Golosov, 1996; Ferro et al ., 1998). Data provided
by the technique were therefore used to validate
the capability of SHETRAN for predicting long-
term (30-year) erosion rates and their spatial vari-
ability (Norouzi Banis et al ., 2004). Simulations
on a 20-m grid were carried out for two arable
farm sites (area 3-5 ha) in central England for
which average annual erosion rates had already
Search WWH ::




Custom Search