Environmental Engineering Reference
In-Depth Information
based models were developed and conditioned
using the data from an intensive monitoring pro-
gramme. However, it is very unlikely that such
extensive datasets will be available for the devel-
opment of all of the possible metamodels required
for the analysis of new catchments. The critical
question becomeswhat is the role of physics-based
models in data-sparse areas?
Computationally intensive physics-basedmod-
els provide the capability to simulate explicitly
the effects of local changes, for example due to
localized tree shelter belts located within a field
or hillslope unit. And, despite significant data
scarcity, these models may still be the best pos-
sible way to upscale local changes to catchment
scale, given that our understanding of the
impacts of land use and landmanagement changes
is largely restricted to the very small scale (such as
changes inwater retention properties, interception
and runoff processes). While continuous hydro-
logical measurements may not be available, use-
ful information about small-scale hydrological
processes can still be obtained from the litera-
ture. The extent to which uncertainty can be
constrained by such data is a key research ques-
tion. We also note that physics-based models
have the power to support the development of
improved conceptual understanding of runoff
processes and the dominant physical controls,
and can thereby provide qualitative insights that
may be of value when considering effects of land
management change.
A methodology is therefore proposed for the
development of detailed physics-based models
for data-sparse land use and land management
types. The key hydrological processes should be
identified from the literature and included in a
physics-based model that has a level of complex-
ity that is appropriate relative to the level of
detailed information available on the system hy-
drological processes. To avoid over-parameteri-
zation, minor processes could be excluded or
treated in a simplified manner. Examining the
model behaviour should also then provide some
further insight into the relative significance of
the model parameters and also inform the meta-
modelling process.
There was quite large uncertainty in these re-
sults, due to parameter uncertainty; however,
median results showed that:
. Removing trees planted within the last decade
causes a 3-7% increase in flow peaks from the
baseline condition.
. Adding tree shelter belts across the lower parts of
all grazed grassland sites causes a 2-11% decrease
in flow peaks from the baseline condition.
. Afforestation of the whole catchment causes
between a 10-54% decrease in flow peaks from
the baseline condition.
. There was no apparent uncertainty in time-to-
peak results, due to the 15-minute resolution of
themodel. The tree removal reduced time-to-peak
by 15minutes, woodland cover increased time-
to-peak by 30minutes, while the shelter belts had
no effect.
Model Regionalization
Introduction
Pontbren is one of only a few catchments in the
UK that are intensely monitored. Clearly, for
more general application, methods are required
to quantify response for other areas, land man-
agement interventions and scales. This raises
issues of the relative lack of detailed supporting
data for other environments, the potential role of
physics-based methods in data-sparse areas, and
the challenge of regionalization. These issues are
being addressed in ongoing research under the
FRMRC2 and FREE programmes. Current prog-
ress is reported below.
Physics-based modelling in data-sparse areas:
the representation of upland peat
management
As we look to extending the upscaling modelling
approach to catchments other than Pontbren,
one of the greatest challenges is how to expand
the existing metamodelling library to account for
the range of land use, land management and soil
types that will be encountered in newcatchments.
For the Pontbren catchment, detailed physics-
Search WWH ::




Custom Search