Environmental Engineering Reference
In-Depth Information
sults of a CAESAR run. Some outcomes can be
relatively easily validated. The flow model, for
example, can be compared against measured flood
outlines or inundation maps generated by 1-D
hydrodynamic models such as HEC-RAS or iSIS.
Conversely, suitable data to validate model pre-
dictions of erosion, deposition and channel plan-
form evolution are more difficult to find.
A limitation of fluvial models in general, which
is certainly applicable to CAESAR, is that the
heterogeneity of natural environments presents
a major problem. For example, spatial and tempo-
ral variability in bed roughness, climate-related
fluctuations in catchment runoff and sediment
yield, and changes to vegetation and land use all
influence the behaviour of a river system. Our
inability to predict runoff and sediment responses
to future anthropogenic impacts introduces un-
certainties that are particularly difficult, if not
impossible, to represent in the input parameters
for the model, or indeed to replicate determinis-
tically within the numerical simulation.
A further limitation common to all sediment
models is the considerable uncertainty that sur-
rounds the sediment transport rules used to drive
the model. In CAESAR they are far from ideal, and
no generally applicable equation from which reli-
able rules can be derived has yet been developed -
for further discussion see Coulthard et al. (in
press). At present it would, therefore, be unwise
to put too much faith in the absolute sediment
fluxes generated by CAESAR. Rather, the value of
the model lies in its use for revealing relative
changes, for example when seeking to establish
whether increases in flood frequency or magni-
tude are likely to cause more erosion or more
deposition in a project reach. Indeed, experience
of using CAESAR has shown that its greatest
strength is in simulating system-scale patterns of
erosion and deposition. For example, it can con-
sistently differentiate between those river reaches
that are more likely to be eroding/incising and
those
model can be calibrated. However, if this is not
available nearby rainfall data can be used, and
there are ranges of example settings from which
the hydrological model can be parameterized.
CAESAR requires a raster DEM (not a Triangu-
lar Irregular Network - TIN) of the catchment, and
editing and correcting the DEM is an important
part of preparing for a CAESAR simulation. The
model has been applied with DEMs having grid
cell sizes ranging from1m to 100m. The choice of
grid cell size is important, as this allows compro-
mises to be made between the area that can be
modelled, the spatial resolution, and the run time.
CAESAR can run with up to 2 million grid cells,
but is best suited to arrays with 250,000 to 500,000
cells. DEMs often contain errors that can cause the
model significant problems, and it is therefore
recommended that DEMs are preprocessed to re-
move any pits that act as internal sinks, and to
ensure that the drainage network follows a
straightforward descent to the exit point, which
must be located at the bottom righthand corner of
the DEM. This can be carried out simply using the
freely available ARC-HYDRO extensions toolkit
for ARC-GIS 8.x and 9.
In reachmode, rainfall data are not required but
it is instead necessary to specify the water and
sediment inputs for the reach.
CAESAR also requires information on the char-
acteristic grain size distributions in the catch-
ment. The model can accept up to nine different
grain size fractions and account for sediment mov-
ing by both bedload and suspended load transport
mechanisms.
Limitations and uncertainty
CAESAR is an experimental tool for scientific
research and hypothesis testing. It is based on
physical processes but uses approximatedflowand
sediment transport rules to replace the complex
governing equations of fluid flow and sediment
transport. The accuracy of these approximations,
and how they are affected by their application in a
2-D model, are largely unknown. Uncertainty in
the veracity of CAESAR's computational engine is
compounded by difficulties in validating the re-
that
are probably depositing/laterally
unstable.
Finally, the ethos of CAESAR's development is
very much one of openness. The code is open
source, model applications are freely available and
