Geology Reference
In-Depth Information
Table 5.3 Parameter values used for the EUROSEM simulations. The 'best estimate' reflects parameter values
derived from measurements and those estimated from the EUROSEM manual (Morgan et al ., 1998b). A 'Y'
indicates that the parameter was sampled for the GLUE analysis, an 'N' indicates that the parameter was held
constant at its best estimate. Minimum and maximum values were derived from field measurements and from
the EUROSEM manual and were used to construct the uniform distributions from which parameter values
were sampled for the GLUE analysis.
Parameter
Best estimate
Sampled
Minimum
Maximum
Length of element (m)
360
Y
300
900
Width of element (m)
125
N
Slope (%)
0.08
N
Manning's n inter-rill
0.02
Y
0.01
0.035
Saturated hydraulic conductivity (mm hr −1 )
0.61
Y
0.1
1
Capillary drive (mm)
806
Y
0.1
1000
Porosity (v v −1 )
0.44
Y
0.3
0.6
Initial moisture content (% saturated moisture content)
0.32
Y
0.1
1
Saturated moisture content (% porosity)
0.34
Y
0.5
1
Maximum interception depth (mm)
0.01
N
Random roughness (mm)
20.44
Y
1
30
Inter-rill slope (%)
0.08
N
Vegetation cover (%)
15
N
Plant angle (degrees)
55
N
Plant basal area (%)
5
N
Plant height (m)
0.03
N
D50 ( μ m)
30
Y
20
250
Erodibility (g J −1 )
1.9
Y
0.1
5
Cohesion (kPa)
13.7
Y
10
50
5.3.2 Model parameterization
sampled as uniform distributions, with minima
and maxima reflecting our a priori uncertainty
about feasible values. All other parameters were
set at default 'best estimate' values. Rainfall was
described using 1-minute data derived from the
rain-gauge record.
EUROSEM was evaluated separately for two
events: 17 November 2006 (event 1) and 2
February 2007 (event 2). The two events were
chosen to represent contrasting storm types, a
single-peaked event (event 1) and a double-peaked
event (event 2). For each of the events the soils
were near saturation due to preceding rainfall,
and vegetation cover was poor. Model parameter-
ization for GLUE simulations requires the defini-
tion of ranges, which are then sampled (see
Section 5.3.3). For the purposes of this study these
definitions were based on a combination of field
measured values and those suggested for similar
sites in the EUROSEM user guide (Morgan et al .,
1998b) (Table 5.3). Parameters which previous
EUROSEM model studies had shown to produce
a sensitive response in simulated model outputs
(Veihe & Quinton, 2000; Veihe et al ., 2000) were
5.3.3
The Generalised Likelihood Uncertainty
Estimation (GLUE) methodology
GLUE was set up in this paper as summarized
in Table 5.4 (following Krueger et al ., 2009). In
brief, initial (prior) distributions of parameters
(noted above) are determined, either from meas-
ured data, literature data or, as in this study, our
own expert opinion. These distributions are then
randomly sampled and a model simulation is
run, varying all chosen (sensitive) parameters at
the same time. The model results are compared
with the measured data (discharge and sediment)
Search WWH ::




Custom Search