Process realism: flow paths and storage - Runoff Prediction in Ungauged Basins: Synthesis across Processes, Places and Scales - page 67

the aquitard, bedrock topography, and the presence of
fractures that may conduct large volumes of water. Setting
model structure and parameters in a realistic way, based
on information on flow paths, can be either quantitative
(e.g., by using data on depth to the bedrock) or qualitative
(e.g., by obtaining a conceptual model of the flow system,
see e.g., Blöschl et al., 2008 ; Clark et al., 2011 ). In both
instances, setting model structure and parameters in a
realistic way has implications for predicting all the runoff
signatures (annual and seasonal runoff, flow duration
curve, low flows, floods, hydrographs), although the
importance of the flow path information varies between
signatures. For runoff hydrographs, the flashiness of the
catchment depends on the magnitude of storage and the
responsiveness of flow paths, as represented by the struc-
ture and parameters of the model. For low flows, infor-
mation on the depth at which flow processes occur
(whether they occur on the surface, shallow subsurface or
deep subsurface), the magnitude of deep storage (either in
the fractured rocks or in porous aquifers feeding a stream
during low flow periods) and aquifer characteristics may
all assist in choosing model structure and parameters.
Similar things apply to floods, although the focus is usu-
ally on shallower flow paths. For seasonal runoff predic-
tions, the magnitude of subsurface storage is mainly of
interest, which may also indirectly affect the annual runoff
(see Chapter 5 ). For annual runoff it may also be very
important whether a catchment loses or gains water
through subsurface flow paths, i.e., whether there are any
interbasin transfers.

Figure 4.13. Hydrogeological response units in the Wattenbach (left)
and Weerbach (right) catchments, Austria. From Rogger et al.
( 2012a ).

Process realism, however, is not an absolute quantity. It
depends on the runoff signatures being predicted.
Depending on the signatures, information about flow paths
and storage can be used in a number of ways to inform
runoff predictions in ungauged basins.

4.5.2 Statistical methods
The alternative group of methods for predicting runoff in
ungauged basins is statistical methods (see Sections 5.3 ,
6.3 etc.). In most of these methods, relationships between
catchment characteristics and runoff signature are estab-
lished from regional data and then used for the predictions.
Although these relationships are usually considered black-
box models, much can be gained by interpreting the rela-
tionships in a process-based, if simplified and aggregated,
way, in order for them to be realistic, and provide the right
predictions for the right reasons. An understanding of the
flow paths and storage can be beneficial for statistical
methods for a number of reasons: (i) Flow path and storage
understanding can assist in the selection of catchment
characteristics. These should be selected not only on the
basis of the goodness of fit of the relationship between
regional runoff signatures and catchment characteristics
but also on the basis of what they represent hydrologically.
They are simple hydrological models themselves, i.e., they
represent the processes at an aggregate level. For example,
one may have the choice to use catchment characteristics

4.5.1 Process-based (rainfall runoff) methods

Runoff signatures can be predicted by process-based
methods from rainfall through some kind of rainfall

-

runoff
model, as discussed in Sections 5 .4 , 6.4 etc. In conceptual
rainfall

-

runoff models, where the flow paths are repre-
sented in a simplified way, understanding of the flow
system may assist in the selection of the model structure.
The catchment may consist of highly permeable fractured
rocks with aquifers that have a lot of storage, deep flow
paths and long residence time. Alternatively the permea-
bility may be low with shallow soils inducing shallow flow
paths, or porous aquifers. The stream may be connected to
or disconnected from the subsurface system. Depending on
the flow system, the model structure may look different in
each of these cases. The parameters related to storage and
response times may also be different. In physics-based
runoff models, information on flow paths and storage is
beneficial during the parameter-setting stage, e.g., depth to

-