Environmental Engineering Reference
In-Depth Information
A way forward
(12 February 2005), it shows the rate of change in
peak discharge with a change in local-scale
surface roughness (Strickler coefficient) for the
Dunsop catchment, northwest England. The spa-
tial pattern that can be seen in the vulnerability
map is primarily the result of the combined
effects of the spatial patterns for rainfall, soils
and land cover/management, plus the effects of
topography and the travel times from runoff sites
to the catchment outlet. The value plotted for
each grid square is for a patch of land covering 5%
of the catchment area (nearest 5% to the centre of
the square).
Sensitivities are, in effect, linearized responses
to change, so although they can be used to
estimate whether, say, a stream bank will be
overtopped as a result of a change in land use/
management, they cannot be used to estimate
the consequences of overtopping. Other tracking
methods may be useful in studying the effects
of non-linearity, including methods based on un-
conventional types of numerical modelling that
involve atomizing and tracking mass and infor-
mation (e.g. Ewen 2000). When using convention-
al models, some form of particle tracking can be
used, with information running piggyback on the
particles, or packets of water can be labelled and
tracked (O'Connell et al. 2007; O'Donnell 2008).
Tracking is straightforward in simple models that
use linear routing. In the geographical information
system (GIS) model of Liu et al. (2003), for exam-
ple, the outlet hydrograph is calculated by super-
position as a sum of contributions from each cell,
so the spatial decomposition of the hydrograph is
trivial. Because sensitivity and vulnerabilitymaps
are simply forms of spatial decomposition of im-
pact, their generation would also be trivial. The
challenge with all these tracking approaches is in
proving they track information well enough such
that the resulting maps are useful. In a drainage
network, for example, some information moves
with the water molecule velocity, some with
the mean bulk velocity, and some with the
kinematic wave velocity, and these velocities do
not behave in a simple manner even when the
geometry of the channels and network is simple
(Henderson 1966).
If catchmentmodelling is toplay its full rolewithin
a DPSIR approach, the main requirement is some
form of Source-Pathway-Receptor (SPR) model
that links small-scale interventions in land use/
management (source) to large-scale impacts down-
stream (receptor), via the drainage network (path-
way). In the DPSIR, this SPR model will provide
the link between impacts, responses, and state.
Through the use of tracking, the results from the
SPRmodelling can be presented and interpreted in
a form that is immediately suitable as the basis
for decision-making (e.g. vulnerability maps).
A suitable SPRmodel, with integrated tracking,
is being developed in the modelling projects
shown in Figure 2.4, and includes methods for
estimating and tracking uncertainty. The main
elements of the SPRmodel are the metamodelling
being developed by Imperial College (see detailed
description in Chapter 3) and the tracking being
developed at Newcastle University. When forced
with synthetic rainfall representing present or
future climatic conditions, the SPR model pro-
duces vulnerability maps that show the link
between the flood frequency curve and potential/
proposed interventions in land use management.
These maps can then be used within the other
components of a DPSIR system, as a basis for
decision support. The tracking uses a detailed
drainage network model (currently, a non-inertia
St Venant approach) allied with impact sensitivity
tracking using adjoint modelling. The develop-
ment and testing of the SPR modelling is relying
heavily on themultiscale field experiments shown
in Figure 2.4 and an existing rainfall generation
model (Burton et al. 2008).
Discussion and Conclusions
The UK's rural landscape has been undergoing
continuous change for more than 50 years, and
will continue to undergo change in the future in
response to climatic and socioeconomic drivers.
Moreover, climatic change is expected to enhance
flood hazard in the coming decades. A broad
