Reducing uncertainty in flood prediction: the representation of vegetation in hydraulic models

The summer 2007 flooding in England was the country's largest peacetime emergency since
World War II, with 13 deaths, over 55,000 homes & businesses flooded & an associated
insurance cost of over £3 billion. Prior to 2007 floods, the UK had experienced a number
of significant flood events over the recent past which have included amongst others; 1)
the Easter 1998 floods of Northampton & surrounding towns in the Midlands when 4,200
homes were flooded in a 1:50 year event &; 2) the winter 2005 floods of Carlisle, a 1:200
year event, when 3 people lost their lives & 1,800 properties were flooded. Following the
2007 floods the Government commissioned the Pitt Review to discover the lessons that
needed to be learnt to manage future flood risk. The key observation reported within the
Pitt Review relevant to this application is that practices which were undertaken to manage
the river corridor; namely dredging, debris removal & notably vegetation clearance, were no
longer being performed as frequently, in order to maintain the ecological diversity of the river
following the Water Framework Directive. This has substantially reduced the capacity of the
river channel & has thus increased the potential of flooding. This is set within the context
of the risk of flooding within the UK increasing into the future, with climate change models
(UKCIP09) predicting that winters will be ~25% wetter, with an increase in extreme rainfall
events.

Flood defences in the UK are managed by the Environment Agency. In order to manage
these resources we require knowledge of the capacity of river channels & associated
floodplains. Aquatic vegetation is present in many UK rivers & this reduces the capacity of
the channel that causes a reduction in flow velocity, which in turn produces higher water
levels per unit discharge, thus increasing the risk of flooding. Therefore, there is a need
to develop our understanding of how vegetation partitions discharge between changes
in velocity & depth & how, in turn, this impacts upon the discharge carrying capacity of a
channel, namely conveyance, to better manage flood prediction & prevention within the UK.
This proposal argues that we can now measure topography to a high resolution & precision
& incorporate it into flood models explicitly. This is not the case for vegetation, & there
remains a lack of understanding of how to represent the influence of vegetation on fluvial
system function. Indeed, the vast majority of uncertainty in flood model predictions stem from
the influence of vegetation on conveyance. In order to move away from an empirical based
approach to the parameterisation of vegetation resistance, a new understanding of the
flow & turbulence production is necessary to be able to re-formulated a dynamic vegetation
roughness treatment for flood models & thus reduce the uncertainty in flood predictions. This
will be achieved by undertaking high resolution experiments in the laboratory in conjunction
with the development of a new three dimensional model that is capable of predicting both
the flow & the plant movement. The model will be validated using the experimental data
& then the two data sets will be combined to enable a new formulation of the drag caused
by the vegetation. This new understanding of the influence of vegetation of drag will be
incorporated into an industry standard flood prediction model. An existing flood example
will be used to develop & test the model as this will allow us to; 1) assess how well this new
modeling approach improves model predictions &; 2) disentangle parameterization & data
error in flood models & enable us to assess what uncertainty needs to be addressed next
generation of predictive flood models.

We anticipate a number of impact areas for the grant & have identified key beneficiaries
within the UK for this work, which include; the Environment Agency who is responsible
for managing water levels & flood defences in UK waterways & our adaption plans to the
impacts of future environmental change; environmental engineers who need to develop
management strategies for sustainability & safety of engineering structures such as flood
defences, &; the large number of academic users interested in complex feedback linkages of
environmental systems.

The results of this blue-skies study will ultimately have impact on the modelling of flow in all
channels & the principles that underpin the modelling will be transferable to some of Britain's
most sensitive localities. Potential model improvement outlined in this proposal will be of
interest to a range of numerical modellers. Durham University, via its well regarded MSc
programme in Risk & Environmental Hazards is responsible for training the next generation
of environmental risk managers & hence knowledge from this project will be made available
to this group via Hardy's involvement on this course.

One of the most important aspects of engagement as regards this project is the provision
of data in a user-friendly format such that it can be viewed, analysed & integrated across
a wide range of platforms. To this end on completion of the project all raw data will have
been post-processed & made available to download from the project website & the NERC
GeoSceince Databases, for the users mentioned above. The results from the work will
be written up in international journals (e.g. Journal of Fluid Mechanics, Water Resources
Research & Journal of Geophysical Research) & presented at a range of international
conferences (e.g. American Geophysical Union Fall Meeting & IAHR River Flow). Final
project derived drag equations will be made available on the project website & through the
NSF funded CSDMS network (csdms.colorado.edu).

We have included a separate pathway to impact document with letters of support from
Industry & Centre for Ecology & Hydrology.