Geology Reference
In-Depth Information
Table 5.1
Published evaluations of the European Soil Erosion model.
Author
Country
Scale and data source
Evaluation methodology
Results
Quinton (1997)
UK
Plot scale (0.1 ha), event-
based data (six hydrographs
and sedigraphs) from
arable agriculture.
Four parameters sampled from measured
distributions and model run for all
combinations of parameters. Evaluated
based on observed data coincidence with
95% confidence intervals.
Observed data within 95% confidence limits, but
confidence limits wide.
Quinton &
Morgan (1998)
US
4.78 ha watershed in Ohio,
event totals.
7 training storms and 4 test storms.
Comparison of event totals and a visual
comparison of hydrographs.
Performed best on higher magnitude events, and
poorly on dynamic data. Hampered by low
slopes (<1%) and poorly described geometry.
It later became apparent that there were
obstructions in the channel of this watershed
that had not been documented prior to the
simulation study.
Folly
et al
. (1999)
Netherlands
45 ha catchment, loess soil,
mixed agriculture and
natural rainfall.
Calibrated against five storms of varying
magnitude and evaluated against five
more storms.
Performed better with short duration events. Lack
of crusting routines identified as problematic.
Veihe
et al
.
(2001)
Costa Rica
2.5 × 1.5 m rainfall simulator
plots on Dystrustept with
maize, grass and no cover,
and Usthotent with no
cover.
Calibrated against one plot for each land use
and model applied to four remaining plots
in each land use. Uncertainty considered
by modifying key parameters within
measured ranges.
Better performance on bare plots. Poor evaluation
results on vegetated surfaces. Attributed to
spatial variability in hydraulic characteristics
and differences between calibration and
evaluation plots.
Veihe
et al
.
(2001)
Nicaragua
22.8 × 1.5 m natural rainfall
plots on bare Cambisol.
Model calibrated against total runoff and
total soil loss data for events in 1993 and
applied to events in 1994 and 1995.
Good performance on annual totals.
Veihe
et al
.
(2001)
Mexico
1 × 1 m rainfall simulator
plots on Bare Vitrand.
Calibrated on three plots and evaluated
against results from four remaining plots.
Total discharge and soil loss simulated reasonably
well. Time to start of erosion poorly simulated
and attributed to lack of soil crusting algorithm.
Cai
et al
. (2005)
China
10 × 2 m rainfall simulator plots
on sandy clay loam soil with
four treatments: unculti-
vated; contour only; contour
+ hedgerows; and contour +
hedgerows + fertilizer.
Calibrated against one experimental run for
each treatment and evaluated against two
other runs.
Problems with timings and rates of erosion, but
excellent fit to total runoff and erosion.
Mati
et al
. (2006)
Kenya
10 × 2 m erosion plots within
the Embori (barley) and
Mukogodo (bare and grass)
catchments.
Model evaluation carried out with no
calibration. Ten events at Embori and
24 plot events at Mukodoogo.
Good correlations between observed and predicted
values at Embori. Model could not reproduce
runoff and erosion at Mukogodo under different
vegetation covers.
