Environmental Engineering Reference
In-Depth Information
There is a lake upstreamof gauge 8. Its response is
simulated using the weir equation Q
Application to Pontbren
The time-series data used for this study are
15-minute resolution rainfall, daily MetOffice
Rainfall and Evaporation Calculation System
(MORECS) evapotranspiration data, and 15-min-
ute resolution streamflow data from six gauges
(gauges 2, 5, 6, 7, 9 and 10: see Fig. 3.1). Gauge
10 data are used for assessing lowflow (
<
1.5m
3
/s)
performance only because the rating curve for
higher flows is poorly defined. The contributing
areas at eachof the sixgauges aregiven inTable3.2.
The period used for the modelling demonstration
is 1 January 2007 to 1 April 2007, for which the
best-quality and most complete data exist.
The catchment is discretized into 100
100m
runoff-generating elements. A semi-distributed
modelling toolkit (Wagener et al. 2004; Orellana
et al. 2008) is used that requires specification of
a conceptual runoff-generating model and a rout-
ing model for each element. A probability distrib-
uted soil moisture model together with two
parallel linear routing stores (Fig. 3.12) is selected,
as this is perceived to be widely applicable in
the UK (Lamb and Kay 2004; Calver et al. 2005;
Lee et al. 2006).
kH
m
,where
H is the water level in the lake above the outlet's
lowest point, and the coefficients kandmare related
to the outlet geometry (Montes 1998), which, fol-
lowing channel measurements (McIntyre and
Marshall 2008), is assumed to be parabolic.
Each runoff-producing element is prescribed a
parameter set according to BFI-based regionaliza-
tion described above. Element runoff is routed
down the streamnetworkusing a constant celerity
approach, i.e. the water moves with constant ve-
locity (Beven 1979). Several samples of peak flow
arrival time observed at gauge 6 are used to esti-
mate the best of the sampled celerity values.
Figure 3.13 shows discharge predictions for
January 2007, which is representative of the
3-month evaluation period. Here, the dark-grey
areas represent the 90th percentile on the dis-
charge prediction; and the black dots are observed
data points. The light-grey area represents the
90th percentile of prior uncertainty, when there
is no distinction due to soil type and land use, and
parameters are assigned uniformly across the
catchment. The dashed lines show the range of
flows within which the streamflow gauge was
calibrated and considered accurate (McIntyre and
Marshall 2008), so that the data points lying in the
range could be considered as being more reliable
than the points lying outside. Note that gauge
10 rating curve is estimated using flows up to
1.5m
3
/s only. The Nash-Sutcliffe statistics for
expected values are summarized in Table 3.3. The
ΒΌ
Table 3.2
Contributed areas for the considered gauges
Gauge number
2
5
6
7
9
10
Contributing
area, km
2
1.3
2.4
3.2
5.8
4.1
12.5
Precipitation
Evapotranspiratio
Fast flow store
k
f
C
max
Water excess
Fast flow
b
Runoff
1
Cumulative distribution
of storage capacity
Slow flow store
k
s
1-
Slow flow
Fig. 3.12
Rainfall-runoff conceptual
model and associated parameters.
