Geography Reference
In-Depth Information
P
E
P
E
P
E
P
E
P
E
Q
se
Q
se
Q
se
Q
se
Q
ie
S
us
S
us
sat.
area
Q
se
S
b
S
b
S
b
S
b
S
S
S
sat
S
sat
S
sat
Q
ss
Q
ss
Q
ss
Q
ss
S
gw
S
gw
Q
bf
Q
bf
S
1
S
2
S
3
S
4
M4 series
Figure 10.18. Schematic describing the increasing complexity of the models used (S1
S4). M4 provides an example of how a single-bucket
configuration (S4 in this case) is represented in the multiple-bucket form. From
Farmer et al. (2003)
.
-
be invoked: similarity between ungauged and gauged
catchments in order to transpose the model structure in
space, and similarity between landscape units in order to
assist in defining the model structure within the catchment
of interest (
Figure 10.8
).
2002
; Fenicia et al.,
2008a
). These examples suggest that
model structures needed to capture essential processes can
differ along a climate gradient and one model may not work
equally well everywhere. A suite of such lumped conceptual
models, ranging in complexity from simple, single bucket
models to configurations involving multiple buckets, is
presented in
Figure 10.18
. The model structure found suit-
able in the gauged catchment is then applied in the similar
ungauged catchment. This is the most common approach of
selecting conceptual model structures that goes beyond an
a-priori perception of processes.
Similarity between ungauged and gauged catchments
The idea of this possibility is to select a gauged catchment
that is hydrologically similar to the ungauged catchment of
interest. This similarity can be established by the methods
discussed in
Section 10.2
, e.g., by the clustering methods of
Figure 10.10
and assuming that the cluster applies to a
contiguous region. For the gauged catchment a suitable
structure of the conceptual model can then be identified on
the basis of runoff, and other hydrological data that may be
available in that catchment through a diagnostic framework.
Numerous studies have demonstrated the usefulness of
multiple data sources for model structure selection in
gauged basins (e.g., Wagener et al.,
2001
; Blöschl et al.,
2008
; Clark et al.,
2011
) and in particular tracer data (Son
and Sivapalan,
2007
; Fenicia et al.,
2008a
,
b
; Hellebrand
et al.,
2011
; Birkel et al.,
2011
; also see
Chapter 4
). In each
case, one aims to account for the dominant processes unique
to each locality. For example, it is likely that in some arid
environments subsurface drainage may not be well
developed, especially in places where there is not a peren-
nial ecosystem that facilitates the creation of subsurface
drainage. In such environments, rapid subsurface flow
may be small or absent and the dominant mechanisms are
likely to be infiltration excess overland flow and deep
percolation. Evaporation will make up for a large fraction
of the water balance and precipitation is often highly vari-
able. Distributed models may be necessary to capture the
spatially heterogeneous runoff generation and routing pro-
cesses (
Reszler et al., 2008
). In wet environments, on the
other hand, high levels of saturation may make it difficult to
separate between different runoff components, and simple
bucket models may perform very well (e.g., Atkinson et al.,
Similarity between landscape units
Model structure
conceptualisation may also be guided by the topographic
organisation and other spatial characteristics of the land-
scape reflective of the functioning of the catchment, as
they can reveal the nature of dominant hydrological pro-
cesses operating in the landscape. For the case of typical
Western European catchments in a temperate climate,
Savenije (
2010
) proposed a model structure on the basis
of three landscape units: wetland, hillslope and plateau. In
contrast to most conceptual models, the runoff processes of
these landscape units act in parallel, whereby it is assumed
that they have direct pathways to the drainage system: the
plateau through the groundwater system, the hillslope
through rapid subsurface flow (and to a smaller extent the
groundwater) and the wetlands (or riparian zones) through
saturation overland flow. To identify and quantify these
landscape units, the method uses an independent landscape
classification based on topographical information: the
height above the nearest drain (HAND, Rennó et al.,
2008) and the slope of the terrain. The advantage of this
method is that landscape information is used to select
relatively simple model structures, targeted to specific run-
off mechanisms. In the wetland zone, slopes are modest
and the groundwater level is close to the surface. The
model structure represents the saturation excess overland
flow (SOF) process as a function of soil moisture. On the
Search WWH ::
Custom Search