Environmental Engineering Reference
In-Depth Information
Methodology
This approach has been trialled for the devel-
opment of a detailed physical model of upland peat
management, for specific application to the Hod-
der catchment in northeast England. In the UK,
approximately 1.5 million ha of the country's 2.9
million ha of peat has been drained (Milne and
Brown 1997). Drainage of peats is typically via a
series of open ditches, alternatively running par-
allel to the slope, in a herring-bone pattern or
randomly positioned (Robinson 1990). This par-
ticular type of drainage is often referred to as
'gripping' (Stewart 1991). Peats are drained with
the aim of improving vegetation and therefore the
production of livestock and game (Stewart 1983).
The rationale is that drainage will remove excess
runoff and lower the water table. Recent work
has evaluated the effects of grip drainage and
grip-blocking, using surrogate data from research
sites elsewhere (Wallage 2007). Current work is
developing metamodels
The parameter space restriction methodology is
as follows. Suppose a catchment is discretized
into a large number of runoff-generating ele-
ments, and catchment response is viewed as the
integration of all the individual elemental re-
sponses. Potentially, the catchment model needs
a separate set of parameters for each element.
Here, it is assumed that all elements with the
same BFI
HOST
have the same set of parameter
values. Therefore, the number of different param-
eter sets needed for runoff generation modelling
is far less than the number of elements and can-
not exceed 29 - the number of soil types in the
HOST classification.
For each model element, a hydrological model
can be run, and then the BFI can be estimated
from the simulated runoff (Gustard et al. 1992).
The simulated BFI is compared to the expected
BFI
HOST
value for the associated soil class, ac-
counting for the standard deviation of BFI
HOST
due to natural variablity within a soil class as
defined by Boorman et al. (1995). This compari-
sonisusedtoidentifybehaviouralandnon-be-
havioural parameter sets, and the associated
likelihood.
Modelling aspects of land use change involves
identifying suitable changes to parameter distri-
butions for the affected elements. Two examples
are discussed here - afforestation and intensive
grazing. There is also evidence that base flow
proportion increases under forest (Wheater et al.
2008), but currently only limited quantitative in-
formation about this change is available. There-
fore, afforestation is assumed to lead to higher
BFI, while keeping the same HOST soil type.
Changes in interception losses associated with
afforestation may be estimated using a standard
model using parameter distributions from the lit-
erature (David et al. 2005; Jewitt 2005). The second
land management change considered is increasing
stocking density, leading to soil structural degra-
dation. Following the approach of Hollis (Pack-
man et al. 2004), degraded soil is assigned an
appropriate analogue HOST class to represent the
change.
for
catchment-scale
application.
Regionalization approaches and
preliminary results
As discussed above, an important research objec-
tive is to develop a regionalization scheme that
may be applied throughout theUK to predict flows
in ungauged catchments and to explore impacts
of local land management changes to catchment
properties. As a first step towards this, the re-
sponse index approach to regionalization is
adopted (Bardossy 2007; Yadav et al. 2007), in
which indices are used to condition prior model
parameter uncertainty into posterior distributions
so that probabilistic predictions of land manage-
ment impacts can be made. While the methodol-
ogy allows all available runoff response indices to
be introduced, we use the Base Flow Index, derived
fromtheHOSTclassification ofUK soils (BFI
HOST
)
with the hypothesis that this on its own is usefully
informative (Lamb and Kay 2004; Lee et al. 2006;
Young 2006; Yadav et al. 2007). The strategy is
to use BFI
HOST
to constrain model parameters
(Bulygina et al. 2009). This then provides a basis
for consideration of modifications to HOST to
allow for changing land management practices.
