Está en la página 1de 18

Earth Surface Processes and Landforms

Climate
flood risk
Earth Surf.change
Process.and
Landforms
32, 429446 (2007)
Published online 5 September 2006 in Wiley InterScience
(www.interscience.wiley.com) DOI: 10.1002/esp.1404

429

Interactions between sediment delivery, channel


change, climate change and flood risk in a
temperate upland environment
S. N. Lane,1* V. Tayefi,2 S. C. Reid,3 D. Yu1 and R. J. Hardy1
1
2
3

Department of Geography, University of Durham, Science Laboratories, South Road, Durham, DH1 3LE, UK
Scott Wilson Water, Unit 33, Mansfield i-centre, Oakham Business Park, Hamilton Way, Mansfield, Nottinghamshire, NG15 5BR, UK
School of Geography, University of Leeds, Woodhouse Lane, Leeds, LS2 9JT, UK

*Correspondence to: S. N. Lane,


Department of Geography,
University of Durham, Science
Laboratories, South Road,
Durham, DH1 3LE, UK.
E-mail: s.n.lane@durham.ac.uk

Received 12 July 2005;


Revised 19 April 2006;
Accepted 8 May 2006

Abstract
This paper uses numerical simulation of flood inundation based on a coupled one-dimensional
two-dimensional treatment to explore the impacts upon flood extent of both long-term
climate changes, predicted to the 2050s and 2080s, and short-term river channel changes
in response to sediment delivery, for a temperate upland gravel-bed river. Results show that
16 months of measured in-channel sedimentation in an upland gravel-bed river cause about
half of the increase in inundation extent that was simulated to arise from climate change.
Consideration of the joint impacts of climate change and sedimentation emphasized the nonlinear nature of system response, and the possibly severe and synergistic effects that come
from combined direct effects of climate change and sediment delivery. Such effects are likely
to be exacerbated further as a result of the impacts of climate change upon coarse sediment
delivery. In generic terms, these processes are commonly overlooked in flood risk mapping
exercises and are likely to be important in any river system where there are high rates of
sediment delivery and long-term transfer of sediment to floodplain storage (i.e. alluviation
involving active channel aggradation and migration). Similarly, attempts to reduce channel
migration through river bank stabilization are likely to exacerbate this process as without
bank erosion, channel capacity cannot be maintained. Finally, many flood risk mapping
studies rely upon calibration based upon combining contemporary bed surveys with historical flood outlines, and this will lead to underestimation of the magnitude and frequency of
floodplain inundation in an aggrading system for a flood of a given magnitude. Copyright
2006 John Wiley & Sons, Ltd.
Keywords: flood risk mapping; flood inundation; sediment delivery; gravel-bed rivers

Introduction
One of the key elements that can affect the hydrological response of a catchment is climate change. This can directly
affect runoff generation through precipitation changes which, in turn, result in changes in flow characteristics. Inundation extent is a function of flow discharge and so climate change may have a direct impact on flood inundation extent.
There is evidence that climate change could be responsible for increases in the magnitude of peak flows (e.g.
Middelkoop et al., 2001; Milley et al., 2002) and recent major flooding in the UK has raised the concern that climate
change is causing increases in flood frequency and magnitude (e.g. Hunt, 2002). Although the evidence that the
magnitude and frequency of high flows is increasing is not clear (Robson, 2002), observations that: (a) climate
variability is seen to have a very strong impact on flood records; (b) rainfall has become more variable; and (c) rainfall
intensity and the frequency of high intensity rainfall may have increased in some areas, including the UK (Osborn and
Hulme, 2002), suggest that future flood risk must take note of potential climate change.
However, the magnitude of a peak flow is not the only control on flood risk: a flood begins when main channel
water levels are sufficient to exceed local bank height. Thus, flood risk is driven by changes in river channel stage,
which may be impacted upon by changes in both flow magnitude and river channel conveyance. In the literature, the
Copyright 2006 John Wiley & Sons, Ltd.

Earth Surf. Process. Landforms 32, 429446 (2007)


DOI: 10.1002/esp

430

S. N. Lane et al.

geomorphological impacts of river flows and flooding and the influence of different land management practices
(e.g. land-use change, clear-cutting, urbanization) on downstream flooding have been hypothesized and researched but
are still the subject of debate (see OConnell et al. (2005) for review). Much less attention has been given to the
impact of river channel configuration (i.e. within-channel morphology) on flood risk and inundation extent (Stover and
Montgomery, 2001) and how sediment delivery impacts on conveyance. This is perhaps surprising given evidence that
the sensitivity of alluviation to climate in temperate upland environments appears to have increased significantly over
the last 4000 years as a result of anthropogenic land-use change (Macklin and Lewin, 2003). In a study of the
Skokomish river, Washington State, USA, Stover and Montgomery (2001) used historical data series to show that
whilst peak discharge was either reducing slightly or not changing, there was evidence of significant stage increases
associated with a given discharge. James (1999) described systematic shifts in stagedischarge relationships arising
from in-channel sedimentation from a Californian case-study. Korup et al. (2004) demonstrated the importance of high
sediment delivery rates for floodplain inundation in the Poerua and Waitangitaona Rivers, New Zealand: exceptional
landslide events were required to cause floodplain inundation as otherwise flows were sufficient to erode into the
valley floors and remain in-bank. Finally, Pinter and Heine (2005) used equal discharge analysis (after Blench, 1969)
to show that stages have systematically risen for equal discharge volumes in the Lower Missouri river. Floods that
were completely in-bank during the early part of the twentieth century were now found to lead to flood inundation,
and the most extreme floods now have a stage that is up to 37 m higher than at the start of the record. In two of the
five stations considered, they found that this was due to constriction in channel cross-sectional areas.
The above examples suggest that, in addition to understanding the impact of climate changes upon flood inundation,
it is also important to consider the effects of changes in channel geometry, especially in systems undergoing long-term
aggradation. The small number of publications that consider this issue, as well as the general lack of its consideration
in flood risk mapping studies, is an interesting issue. This may be due to the traditional view of sediment delivery as
some kind of local (in time and space) disturbance to an equilibrium channel morphology, where cross-sectional areas
adjust to some characteristic bankfull flow. When sediment is delivered, it is thought to be just a matter of time before
channel capacity re-establishes itself through transport of that sediment downstream through the river system. However, this overlooks extensive geomorphological evidence of alluviation, potentially rapidly over short time periods
(e.g. Macklin and Lewin, 2003), including aggradation. For example, Harvey (2002) reports that the long-term
response of a major input of sediment to Bowderdale and Langdale Becks in the Howgill Fells was partial stabilization
of associated sedimentation zones and not downstream progradation.
This paper aims to understand how future climate change might interact with river channel change and so impact
flood risk, for an upland gravel-bed riverfloodplain system. Equal discharge analysis (e.g. Blench, 1969; Pinter and
Heine, 2005), where stage changes through time associated with a given discharge are used to detect conveyance
changes, is not adopted here. Such an approach is difficult except in systems with a very long history of measurement
as: (a) extreme events do not happen very often; (b) many of our instrumental records are short and/or incomplete; and
(c) discharge is often inferred from a stage record, especially in smaller tributaries that are not well-suited to more
advanced discharge measurement methods, making such an analysis circular. Instead, we use a hydrodynamic modelling approach. We measure changes in channel geometry over a short timescale (16 months), in a reasonably well
understood reach of aggrading gravel-bed river. These are then applied to a baseline hydrograph, as well as hydrographs
scaled for climate change, based upon the predicted monthly percentage precipitation changes simulated by HADRM3
for periods 2050s and 2080s (Hulme et al., 2002).

Methodology
Study site
The test site in this study is based upon a 6 km upper reach of the River Wharfe, UK (Figure 1). This comprises a
typical upland gravel-bed river with a range of 10 to 15 m in width. It is generally single thread, with individual
meander bend series linked by relatively straight reaches. The floodplain is relatively wide (Figure 1), has a gentle
slope (c. 00040) but is divided by a mixture of dry-stone walls, hedges and fences. Observations suggest that the drystone walls are largely impervious to flow except where there are gates, and exert an important impact on flood
routing. Coulthard et al. (2005) produce Holocene timescale estimates of the aggradation/erosion characteristics of
this reach (which corresponds to their Reach 2), in comparison with other reaches of the Yorkshire Dales system. As
with the upper reaches of the Swale and the Nidd, this reach is modelled to have undergone rapid erosion followed by
gradual aggradation, commonly over hundreds of years. The timescale for the focus of this paper (present to 100 years
into the future) is shorter than that in Coulthard et al. (2005) and the shorter term fluctuations in aggradation and
Copyright 2006 John Wiley & Sons, Ltd.

Earth Surf. Process. Landforms 32, 429446 (2007)


DOI: 10.1002/esp

Climate change and flood risk

431

Figure 1. The catchment and reach under study: the Upper Wharfe, Yorkshire Dales National Park, North Yorkshire. Sediment
sensors were installed in the Oughtershaw and Greenfield subcatchments, and at Beckermonds, Deepdale, Hubberholme, Buckden
and Starbotton.
Copyright 2006 John Wiley & Sons, Ltd.

Earth Surf. Process. Landforms 32, 429446 (2007)


DOI: 10.1002/esp

432

S. N. Lane et al.

erosion associated with our timescale will determine whether or not the reach is aggradational. Thus, we include
surveyed results to assess current sedimentation characteristics.

Flood inundation modelling


This research is underpinned by a key development in the modelling of floodplain inundation with complex topological structure (Lane, 2005; Yu and Lane, 2006a,b; Tayefi et al., in press). A range of studies have demonstrated that
floodplain topographic features have a controlling role in flow distribution on floodplains (see Zanobetti et al., 1968;
Cunge, 1975; Hardy et al., 1999; Marks and Bates, 2000; Bates and De Roo, 2000; Horritt and Bates, 2001a,b;
Nicholas and Mitchell, 2003). This has resulted in the development of two-dimensional (2D) floodplain flow models
in which the depth-averaged form of the NavierStokes equations is simplified through ignoring the associated inertial
terms (e.g. Bates and De Roo, 2000; Horritt and Bates, 2001a,b; Bradbrook et al., 2004). Commonly, the channel
flows are represented using one-dimensional (1D) models. Strictly speaking, the simplification of the depth-averaged
equations restricts the validity of the modelling process to situations where the flood inundation depths are low. These
coupled 1D2D models have allowed a much more ready incorporation of topographic data into the flood modelling
process. However, they have still represented subgrid-scale topographic features using friction rules (e.g. the Manning
friction law), often outside the range of conditions for which the rules were developed and overlooking the arbitrary
nature of the original friction laws as they were developed (e.g. see Manning, 1891; Lane, 2005). Yu and Lane (2006a)
show that these models have very low sensitivity to friction parameters, amplifying the need to use very high values
of roughness in order to control model response. Thus, Yu and Lane (2006b) develop a subgrid-scale treatment
(FLOODMAP) in which topography is represented through blockage and porosity terms rather than roughness
parameterization, and show that this results in a significant improvement in model performance for an urban area,
including better time-step control and the recovery of sensitivity to roughness perturbation. Tayefi et al. (in press)
show for the case study presented here that this model development is equally necessary for rural floodplains with
complex topological structures. They compared three options for modelling floodplain inundation of upland floodplains
with complex topographic and topological structure: (1) lateral extension of cross-sections across the floodplain; (2)
representation of the floodplain as a series of connected storage cells; and (3) a coupled one-dimensional model of the
main channel and two-dimensional diffusion wave model of the floodplain. Model assessment was based upon an
event that occurred on 4 February 2004. This showed that a significantly better level of agreement was obtained with
the coupled 1D2D model than with either the extended section or storage cell approaches, and so the 1D2D model
is used in this study.
Tayefi et al. (in press) detail the 1D2D modelling approach and only a summary is provided here. The main
channel flow is modelled using HEC-RAS, a hydraulic model developed by the Hydrologic Engineering Centre (HEC)
of the US Army Corps of Engineers. This model allows description of the river channel as a series of discrete crosssections perpendicular to the flow direction. The finite difference procedure for solving the flow equations then defines
the topography of the system domain according to the cross-sections. Water surface elevation and other hydraulic
characteristics of flow are calculated at each cross-section and all hydraulic structures defined in the domain (e.g.
lateral weirs).
The floodplain was discretized to an 8 m resolution, following recommendations for complex floodplains in Yu and
Lane (2006a,b). Fluxes between cells were set to zero where cells fell adjacent to impervious boundaries. Flow
exchanges between the main river and floodplain were calculated based on water levels in the adjacent cells and some
form of boundary condition, in this case a weir equation. We assume that the weir segment and water level over the
weir are horizontal, allowing weir flow to be computed using the standard equation for a broad-crested weir:
Q = CLH3/2

(1)

where Q is discharge, C is the weir flow coefficient, L is the length of spillway crest and H is the upstream energy
head above the spillway crest (i.e. the evacuating weir segment). It should be emphasized that this type of boundary
condition will exert a crucial control on the magnitude of the flux from river to floodplain. Whilst there has been much
research into the parameterization of fluxes from river to floodplain (e.g. Knight and Shiono, 1996), much less of this
research has been incorporated into models that couple 1D river and 2D floodplain representations and this remains an
area of significant uncertainty.
Flows from floodplain to river are based upon evaluating the water surface elevation difference between 2D
diffusion wave model predictions on the floodplain and the 1D model predictions in the river using the 2D diffusion
wave approximation. The models are loosely coupled, i.e. the 1D model was used to generate the boundary conditions
necessary for the 2D model, and it was assumed that return flow to the channel would have a minimal impact upon
Copyright 2006 John Wiley & Sons, Ltd.

Earth Surf. Process. Landforms 32, 429446 (2007)


DOI: 10.1002/esp

Climate change and flood risk

433

downstream flooding. The latter is justified by the relatively small floodplain length being modelled as compared with
the duration of the event.

Geometric data
Both the 1D and 2D treatments require geometric data, although they use these data in different ways. The channel
topography was described by cross-sections surveyed using a combination of total station and real-time, kinematic,
Global Positioning System (GPS) surveys. Both of these provided data precise to better than 002 m. A total of 58
sections were surveyed over the 6 km reach, with the greatest survey density within sinuous reaches (at approximately
one-quarter of the meander bend wavelength) and lower survey densities on straighter reaches (maximum spacing
250 m). These surveys were collected on two occasions: December 2002 and March 2004. The GPS system was used
to guarantee cross-section relocation to exactly the same end points in each survey. We then used both the 2002 and
2004 data with a February 2004 flood event (see below) in order to assess the impacts of geometrical change upon
flood inundation. We also scaled the 2004 event from a 1-in-05-year event to a 1-in-2-year event (also see below) in
order to assess the interactions between a larger event and the same two geometrical configurations.
The floodplain topography was described using remotely sensed Light Detection and Range (LiDAR) data. These
data were filtered for trees and hedgerows (assumed to be pervious to flow at the scale of model application, but
compare with Cobby et al., 2003) using automated image processing techniques. Field tests suggested that the
processed data were precise to better than 020 m. These data were supported by Ordnance Survey Landline data,
held in the same co-ordinate system as that used for both the cross-section and LiDAR data surveys. These data
needed additional field-based classification into whether or not the boundaries were pervious, and over what spatial
extent. Parameterization of Equation 1 required additional field survey of individual gates and their width and elevation characteristics, and this was also undertaken using the GPS survey.

Boundary conditions
Two flood events were modelled in this research: one with an estimated 1-in-05-year return period and one with an
estimated 1-in-2-year return period. The 1-in-05-year return period flood was based upon stage data measured during
an event that occurred on 4 February 2004. The stage record was upstream of the study site in a relatively stable
bedrock cross-section. Tayefi et al. (in press) explain how these data were used with a Flood Estimation Handbook
(Institute of Hydrology, 1999) rainfallrunoff model to provide the discharge data required at the upstream end of the
domain. The 1-in-2-year event was based upon an analysis of measured maximum stages at the upstream end of the
domain to identify a representative 1-in-2-year stage. This was then used to scale the measured 1-in-05-year flood
event linearly. At the upstream end of this reach, both of these flows were well in-bank. Here, the section was
relatively simple and there were minimal backwater effects, meaning that the stagedischarge relationship is likely to
be linear in the zone to which the scaling was applied. The downstream boundary conditions were set as normal depth
using the local channel bed slope, but the reach was extended beyond the zone of interest such that this boundary
condition had no impact on model predictions. Sensitivity tests, in which both n and channel bed slope were varied,
showed that this was the case.
Detailed sensitivity analysis was used to identify the most sensitive model parameters in relation to main channel
water levels. This showed that Mannings n was by far the most sensitive model parameter (Tayefi et al., in press) and
this was the focus of model calibration. It was found that n = 0055 gave the optimum prediction of water levels based
upon 14 locations of estimated maximum water surface elevation, reconstructed from wrack lines measured immediately after the flood event. The 1D model was then used to determine the stage boundary condition for application to
the lateral weirs (using Equation 1) alongside the main river to derive the fluxes required for the 2D model. Flux could
be bidirectional (i.e. river to floodplain and/or floodplain to river), determined by the difference between water levels
in the river and on the floodplain, and dependent upon the channel adjacent bank height.

System drivers: sediment delivery


As we wish to situate these results within broader geomorphological processes operating in the catchment, we also
sought to collect geomorphologically relevant data over a longer time period. First, in order to assess the medium-term
patterns of channel change in the river system, we: (a) undertook repeat survey of the cross-sections that we model in
this study; and (b) synthesized together, as far as possible, the cross-section data collected in previous studies of the
river dating back to the 1980s (Reid et al., in press, a). The repeat section survey was limited to the upper 50 per cent
of the reach, extending from Hubberholme to Heber, until December 2002 but extended to the full, modelled reach,
Copyright 2006 John Wiley & Sons, Ltd.

Earth Surf. Process. Landforms 32, 429446 (2007)


DOI: 10.1002/esp

434

S. N. Lane et al.

extending from Hubberholme to Starbotton, for the period December 2002 to March 2004. Second, we installed a
series of sediment sensors (see Figure 1), designed to record the onset and duration of coarse sediment movement
during transport-effective events, in order to understand the spatial patterns of sediment delivery and deposition (Reid
et al., in press, a). The sensors comprise pressure plates, 0150 m 0130 m in area, installed flush with the bed and
connected to loggers buried within the bed. The detection of sediment movement rests upon grains rolling or saltating
along the bed (i.e. there is no detection of particles in suspension) and if they impact upon the pressure plate, a pulse
is recorded by the logger. The plates are calibrated to be sensitive to any grain coarser than 0020 m in diameter. The
main appeal of using these is that we were able to install a relatively large number of them (ten), distributed across the
catchment, in order to explore possible coarse sediment sources. There are three main disadvantages to using them.
First, each sensor occupies only a small proportion of the potentially active channel width and we must assume that
the plate location is representative of transport activity across the entire channel width. This is less likely to be the
case at lower transport rates when there is a greater probability that a narrower part of the bed is experiencing active
sediment transport. Second, for saltating particles, as saltation step lengths (not particle step lengths as conventionally
explored when particle transport distances are determined at event timescales) increase, so there is a greater probability of overpassing of the plate. This probability should be a negative function of grain size and a positive function of
discharge. These two issues mean that care must be shown in using these sensors to infer absolute coarse sediment
transport rates. Rather, they are likely to give an index of relative transport variability as well as very useful information on the onset and termination of transport (Reid et al., in press, a). The cross-section and sediment sensor data are
used to contextualize the channel change findings for the December 2002 to March 2004 period.

System drivers: climate change


By accepting the conclusion that flooding in the future is likely to be affected by climate change, through changes in
rainfall event characteristics, future climate conditions must be modelled. Here, we use a simple method to translate
specified changes in climatic inputs into changes in hydrological responses. This involved (i) the application of a
hydrological model using current (i.e. observed) climate data (see Tayefi et al. (in press); (ii) the definition of climate
scenarios; (iii) perturbation of the original (i.e. observed) input climate data accordingly; and (iv) running the hydrological model under future climatic condition. At present, the prime source of information regarding potential climate
changes are Global Climate Models (GCMs) whose results can be interpreted at the scale of individual grid boxes
(typically 300 km). In turn, these can be used in combination with regional topographic and climatic characteristics to
drive regional climate models like HADCM3.
GCMs attempt to represent the climate and physical process of the atmosphere, ocean, cryosphere and land surface
using a three-dimensional grid across the globe (Prudhomme et al., 2002). Although the.y are often run at fine
temporal resolutions (e.g. 1530 minute time-steps), the most reliable output variables are at the daily scale (for largescale circulation indices) and the monthly scale (for weather variables such as temperatures and precipitation)
(Prudhomme et al., 2002). GCMs simulate current and future climate elements, according to different global emission
scenarios. The emission scenarios are based upon varying assumptions about future population and economic growth,
technological advancement and social attitudes towards energy use (Reynard et al., 2001). None of these scenarios
includes targeted global or national strategies but they assume different development paths for the world. The most
commonly applied emission scenarios are the SRES scenarios developed by the Intergovernmental Panel for Climate
Change (IPCC) and labelled Low Emissions, Medium-Low-Emissions, Medium-High-Emissions and High-Emissions
(IPCC, 2001).
The period 19611990 has been chosen internationally as the baseline for predicting future climate changes (IPCC,
2001; Hulme et al., 2002). Future climate change results are mostly presented for three different time periods: the
2020s, the 2050s and the 2080s. The 2020s, 2050s and 2080s present the periods 20112040, 20412070 and 2071
2100, respectively. HADCM3 has been run for all four of the SRES scenarios and for three future time slices. The data
sets available from the UK Climate Impacts Programme 2002 (Hulme et al., 2002) include 14 climate variables at a
spatial resolution of 50 km in the form of monthly averages for the 2020s, 2050s and 2080s time slices. Due to the
high degree of variability exhibited in the UK climate, average results for a 30-year period are generally considered
more reliable than results for a single year (Reynard et al., 2001). In this paper, in terms of climate change, we focus
only on precipitation changes and then only on 2050 and 2080 simulations. UKCIP02 scenarios have been produced
for all four of the SRES greenhouse gas emission scenarios. However, only A2, the medium-high emissions scenario,
is considered here as it is the only one to have been fully simulated by HADRM3.
As discussed, the HADRM3 precipitation predictions are only reliable at the monthly timescale. Given the temporal
resolution required for hydrological modelling they must be further downscaled. The monthly percentage precipitation
changes simulated by HADRM3 for periods 2050s and 2080s under emission A2, for a 50-km grid within which the
Copyright 2006 John Wiley & Sons, Ltd.

Earth Surf. Process. Landforms 32, 429446 (2007)


DOI: 10.1002/esp

Climate change and flood risk

435

Table I. Percentage changes in precipitation relative to the 19611990 baseline, as simulated by HADRM3 for the 2050s and
2080s using emission A2 (Hulme et al., 2002)
Month

January
February
March
April
May
Jun
July
August
September
October
November
December

Percentage change in precipitation


2050s

2080s

111
85
45
13
86
157
207
206
138
37
55
107

195
150
80
23
152
277
364
362
244
64
96
188

study site, are shown in Table I. As Table I shows, the UKCIP02 scenarios indicate an increase in the seasonality of
precipitation for North Yorkshire in the future with a progressive increase in winter rainfall and decrease in summer
rainfall (Hulme et al., 2002). In order to generate flows scaled for climate change effects, we scaled the 15-minute
precipitation data used to generate the 4 February 2004 peak discharge using a proportional approach. From Table I,
the proportional upscaling was 85 and 150 per cent for scenarios in the 2050s and 2080s, respectively. Due to its
simplicity, this type of change has been frequently applied in literature (e.g. Arnell, 2003; Prudhomme et al., 2003).
The scaled precipitation data were applied to the Flood Estimation Handbook (Institute of Hydrology, 1999) rainfall
runoff model assuming no change in the associated parameter values. The rainfallrunoff method includes parameters
such as standard percentage runoff (SPR) and base flow, which depend on parameters that may be climate-dependent
(e.g. due to antecedent condition effects). We are currently exploring continuous hydrological simulation for the
catchment to address this issue. However, the estimated increases in the 2004 peak flow for the 2050s and 2080s
scenarios of 82 per cent and 147 per cent matched the findings of Dugdale (2003) who applied a modified version of
TOPMODEL (Lane et al., 2004) to a subcatchment of the study area and found similar percentage increases under a
proportional downscaling of climate parameters.

Results
Short-term channel morphological changes and sediment delivery.
In order to contextualize the subsequent geometrical changes and flood inundation response, this section summarizes
the short-term changes in channel morphology and sediment delivery that occurred during and immediately before the
study period. The cross-section surveys suggested that, over the period December 2001 to March 2004, the upper part
of the study reach was aggradational, with a reach- and width-averaged mean rise in bed levels of 022 0021 m, and
this fits with much longer timescale simulations of this reach which also suggest aggradation (Coulthard et al., 2005).
Over 80 per cent of the net in-channel aggradation measured during the period December 2003 to March 2004
occurred in the February 2004 flow event modelled here. This change was not distributed evenly throughout the study
period (Figure 2). The bed-level changes were aggradational except for the period December 2002 to March 2003,
which was relatively dry for a winter period and, unlike all of the periods shown, experienced no bankfull flood
events. The study reach was surveyed in full from December 2002 onwards and, during this period, the reach- and
width-averaged mean rise in bed levels was 012 0014 m. This corresponds to the aggradation that occurred
between the two geometrical boundary conditions applied to the flood inundation model.
Figure 3 shows the spatial distribution of the measured mean bed level changes in terms of erosion and deposition.
This shows substantial spatial variability associated with these mean changes. It is tempting to infer the passage of
waves of deposition and erosion from these data, but our observations suggest that this is problematic, as these
patterns hide substantial transfer of material into store for periods greater than the measurement period, as well as the
Copyright 2006 John Wiley & Sons, Ltd.

Earth Surf. Process. Landforms 32, 429446 (2007)


DOI: 10.1002/esp

436

S. N. Lane et al.

Figure 2. Changes in reach- and width-averaged mean bed level from December 2001 to March 2004. Vertical error bars show
estimated mean bed level uncertainty.

Figure 3. Spatial distribution of measured erosion and deposition for the five study periods. Each epoch is centred on a
horizontal grid line with the measured change corresponding to erosion when above the grid line and deposition when below: 25
is the no change line for the December 2001 to March 2002 epoch; 50 is the no change line for the March 2002 to December
2002 epoch; etc. The vertical grid line is the downstream end of the gravel trap.

release of material that has been in store for considerable time. However, the observations do suggest that there is
substantial spatial variability superimposed upon the net aggradational trend, and that this will have implications for
which reaches deliver water to the floodplain and which parts of the floodplain are inundated. Situating this reach in a
longer time frame of analysis is more difficult as the cross-section measurements have been undertaken for a number
of purposes (including flood risk assessment and geomorphological analysis) and by a number of different surveyors.
The cross-sections were rarely co-located and cross-sections were not always surveyed into a reliable fixed datum.
However, Figure 4 shows a schematic of river channel response dating back to the 1980s for the upper 50 per cent of
the reach. Between 1982 and 1990 a gravel trap was installed towards the upstream end of the reach and extensive
Copyright 2006 John Wiley & Sons, Ltd.

Earth Surf. Process. Landforms 32, 429446 (2007)


DOI: 10.1002/esp

Climate change and flood risk

437

Figure 4. Schematic maps of change inferred from available cross-sectional data for the period 1982 to 1990 and 1990 to 2000.
Tributary streams are shown, some of which connect with the river, some of which do not. Lighter shading is depositional, darker
shading is erosional.

shoals of gravel were removed and the channel realigned. In both periods shown in Figure 4, the gravel trap reach was
depositional and, through time, this deposition appears to have extended upstream: much of the river downstream of
the sudden reduction in valley gradient at Hubberholme (Figure 4a) was depositional. Downstream of the gravel trap
reach, there was evidence of erosion (Figure 4a) that appeared to extend downstream during the 1990s (Figure 4b),
although our cross-section surveys suggested that deposition had begun to occur in parts of this zone from 2000
onwards. This emphasizes the difficulty of generating clear conclusions about secular erosion and deposition trends
without higher frequency data (in space and time) before 2000. From the Buckden Bridge bend, patterns of erosion
and deposition appear to be relatively stable over the time period of available data: the bend itself is generally
depositional; there is then a straight reach that is generally erosional; and finally there is a weakly sinuous and
depositional reach until a sudden realigment of the river at the downstream end of the upper reach, where there is a
small amount of erosion evident.
Our general interpretation of these results is that the short-term (2000 to present) response of the study reach is
expressed in terms of aggradation. In general, from the restricted data that are available between 1980 and 2000, this
short-term pattern mirrors longer term change. Indeed, the rationale for gravel trap introduction during the 1980s was
to manage gravel accumulation which was blamed for a high frequency of flood events in the system: sediment
accumulation was estimated to be between 043 m and 14 m in depth in the gravel trap reach between 1966 and 1981
(Stewart, 1984), and this was linked to a frequency of flood inundation of between 20 and 40 times per year
Copyright 2006 John Wiley & Sons, Ltd.

Earth Surf. Process. Landforms 32, 429446 (2007)


DOI: 10.1002/esp

438

S. N. Lane et al.

Figure 5. Downstream changes in sediment sensor results for the period March 2003 to March 2004. The relative intensity index
is the number of clasts per second scaled by the channel width (see Reid et al., in press, a). The maximum recorded intensity is the
largest number of clasts recorded per second at each sensor.

(Yorkshire Water Authority, 1983; Hey and Winterbottom, 1990). After installation, the gravel trap appears to have
acted as a sediment transfer discontinuity for a short period of time. However, it filled extremely rapidly (within two
years of becoming operational) and it was deemed too expensive to empty it again. In July 2002 it was removed and
it is possible to detect the effects of this in the March 2002 to December 2002 survey (Figure 3, erosion upstream of
gravel trap associated with gravel removal) and establishment of a clear zone of gravel accumulation downstream
from the trap in the subsequent survey periods.
The second major source of evidence that can be used to understand the geomorphological setting of the study site
comes from the sediment sensors. Based upon data given in Reid et al. (in press, a), Figure 5 shows the relative intensity
index for sensed sediment based upon Hubberholme, Buckden Bridge and Starbotton (see Figure 1b for locations). This
shows a dramatic reduction in the amount of sediment that is sensed between Hubberholme and Starbotton (approaching three orders of magnitude) and this is likely to be a realistic trend, even given the sensor uncertainties described
above. However, with distance downstream, the maximum intensity during the largest event recorded (4 February
2004) increases. This helps us to explain why the study tends to be depositional. There is significant delivery
of sediment from the catchment upstream of Hubberholme, and this appears to be associated with a very large number
of transporting events, not least because the main river and tributaries are much steeper and often bedrock, allowing
much more ready transfer of coarse sediment (Reid et al., in press, a). Only the largest events are capable of
transporting material within and through the Hubberholme to Starbotton reach. Little sediment was observed to move
beyond Starbotton during the study period (Reid et al., in press, a) which can be attributed to an alluvial fan that
extends across the valley at this point. The net effect of high sediment delivery to Hubberholme and low sediment
transfer beyond Starbotton is the substantial bed aggradation recorded in our cross-section surveys, and it does seem
to be the case that the alluvial fan at Starbotton is contributing to aggradation upstream (Coulthard et al., 2002).

Impact of channel geometry changes


Given the above evidence of bed aggradation, this section assesses the effects of measured aggradation upon the
inundation associated with a 1-in-05-year event that occurred on 4 February 2004 and the upscaling of the 1-in-05year event to a 1-in-2-year event. The effects of the geometrical changes were a reduction in the magnitude of the
bankfull discharge, not because of a reduction in flow magnitude, but because of a rise in bed level. For those sections
that were bankfull in both the December 2002 and March 2004 cases, the average bankfull discharge fell from 306 to
287 m3 s1 (i.e. a 61 per cent decrease), associated with an average rise of 012 m in bed levels along the reach,
although there was substantial spatial variation in both the magnitude of bankfull discharge and the bankfull discharge
reduction. Figure 6 shows the inundated area through time for the 2002 and 2004 geometrical scenarios, using the
1-in-05-year and 1-in-2-year floods. It is clear that the return period of the two floods has a major impact upon the
maximum inundated area. However, the 16 months of measured sedimentation resulted in increases in inundation
area of 57 per cent and 71 per cent for the 1-in-05-year and 1-in-2-year floods, respectively. Visual comparison of
the inundated areas obtained for the time at which the peak flow passed the downstream end of the reach, using the
Copyright 2006 John Wiley & Sons, Ltd.

Earth Surf. Process. Landforms 32, 429446 (2007)


DOI: 10.1002/esp

Climate change and flood risk

439

Figure 6. Inundation through time for the December 2002 and March 2004 bed geometries and two different flood magnitudes.

two geometry scenarios and the two flood return periods (Figure 7), shows that these increases are predominantly
associated with the upstream part of the study site. The average out-of-bank flow duration increased from 302 minutes
to 330 minutes (i.e. 95 per cent) and the downstream distribution of durations for the 1-in-05-year event and the two
geometries is shown in Figure 8. It is possible to associate the changes in bed level shown in Figure 3 to the duration
changes shown in Figure 8. The relevant change series is the sum of the top three curves in Figure 3. Figure 3 shows
substantial aggradation, that begins about 300 m downstream and peaks at about 600 m downstream. In Figure 8, from
about 300 m downstream, there is a rapid rise in inundation duration to a peak at about 400 m downstream. The
difference between the location of the peak increase in inundation duration and the peak aggradation is interesting and
implies that the onset of upstream aggradation causes flow to go out-of-bank further upstream. The increased flux to
the floodplain reduces the flux in the river such that whilst the in-channel aggradation continues to increase in magnitude
downstream of 400 m, this is countered by there being smaller flows in the channel. This type of process occurs over
a larger scale between 300 m downstream and 1600 m downstream, where the increase in duration of inundation with
the March 2004 geometry to 800 m downstream is countered by a decrease in duration of inundation after 800 m
downstream: the increased flux to the floodplain in the upstream reach is countered by reduced flows further downstream. The key point emerging from this discussion is that the effects of in-channel aggradation can lead to quite
complex changes in flood inundation as a result of a changing partitioning of flow between channel and floodplain.
Under both geometric scenarios and flood events, changes in the downstream inundation extent were partly limited
by high levels of valley confinement. Thus, whether or not geometrical changes impact upon inundated area depends
upon both the magnitude of the flood event and the geometrical change in relation to the degree of valley confinement.
However, as the geometrical changes lead to a reduction in bankfull discharges, they will also increase the frequency
of onset of flooding, so reducing the return period associated with a given inundation area. Similarly, flows with return
periods close to bankfull will be most sensitive to geometrical changes although, for the two geometries simulated
here, the peak flow magnitudes were somewhat larger than the mean bankfull discharge.

Impact of predicted climate changes


Figure 9 shows the time-series of inundated area for the present event, the 2050s and the 2080s, for the 1-in-05-year
event. The peak inundation extent is estimated to increase by 122 per cent for the 2050s and 216 per cent for the
2080s. The inundation maps (Figure 10) show that the patterns of inundation remain similar and that the prime sensitivity
of the system is to changes in the volume and duration of out-of-bank flow, which results in the increase in inundated area.

Combined geometrical and climate change


Finally, we combined the 1-in-05-year event with the 2002 to 2004, 2050 and 2080 simulations. The effects of
the December 2002 to March 2004 sedimentation in isolation were a 57 per cent increase in inundated area by 2004.
The 2050s climate scenario resulted in an increase of 122 per cent and the 2080s in an increase of 147 per cent.
Copyright 2006 John Wiley & Sons, Ltd.

Earth Surf. Process. Landforms 32, 429446 (2007)


DOI: 10.1002/esp

440

S. N. Lane et al.

Figure 7. Predictions of areas inundated in the 1 in 05 and 1 in 2 year events simulated in this study, with 2002 and 2004
geometries respectively, both at the time of peak flow at the downstream boundary.

When the 2002 to 2004 sedimentation was combined with the 2050s climate scenario, the increase was 382 per cent
and with the 2080s, the increase was 521 per cent. This shows non-linear interaction between the two drivers of the
flood system, not surprising as the lateral inundation extent will be controlled by the number of cross-sections
experiencing out-of-bank flow and the magnitude and duration of those out-of-bank flows, i.e. a threshold process.

Discussion
The main finding that emerges from this research is that relatively short duration channel configuration changes, in this
case over a 16-month period, can lead to substantial changes in inundated areas for 1-in-05 and 1-in-2-year floods,
Copyright 2006 John Wiley & Sons, Ltd.

Earth Surf. Process. Landforms 32, 429446 (2007)


DOI: 10.1002/esp

Climate change and flood risk

441

Figure 8. Duration for which out-of-bank flow was exceeded as a function of distance measured upstream from the downstream
end of the reach.

Figure 9. Inundation through time for the February 2004 event as baseline and with scaling to the 2050s and 2080s.

in an upland environment associated with coarse sediment transfer. For the 1-in-05-year event, the increase in
inundated area was almost one-half of the increase in inundated area estimated as resulting from simulated climate
change to the 2050s. Key to this increase was coarse sediment aggradation associated with a depositional reach of
river (Figure 2), on which was superimposed quite significant spatial variability in the deposition process (Figure 3).
The non-linear nature of the onset of floodplain inundation (i.e. associated with the onset of overbank flow) meant that
when these geometrical changes were combined with forecast climate changes, the increases in inundated area were
amplified over the increases associated with each change in isolation. In other words, in this case study, in-channel
sedimentation increases the sensitivity of flood inundation to future climate changes. The comparison of the 1-in-05
and 1-in-2-year flood events emphasizes that when this kind of analysis is couched in terms of event return periods,
this appears to be a much more important impact upon the inundated area (Figure 6) than either short-term
geomorphological response or climate change. However, return periods can be misleading, not simply because of the
reliability in their determination, but also because they are commonly based upon the peak magnitude of flow, which
may be only weakly related to the magnitude and duration of out-of-bank flow. The kinds of geomorphological
changes described here do not impact upon discharge return periods, but they do impact upon the return periods of
floodplain inundation, such that there could be a major increase in the magnitude and frequency of inundation events
in any river, or river reach, that is aggradational, with no change in the magnitude and frequency of imposed flows.
Some distinction is required here between changes in inundated area and changes in inundation depth. Once a flood is
large enough to occupy the full valley width, bed aggradation will only impact upon the inundation depth. For larger
Copyright 2006 John Wiley & Sons, Ltd.

Earth Surf. Process. Landforms 32, 429446 (2007)


DOI: 10.1002/esp

442

S. N. Lane et al.

Figure 10. Predicted inundation patterns for a 1 in 05 year event for the 2050s (a) and the 2080s (b), both at the time of peak
flow at the downstream boundary.

floods, for which all sections are at bankfull, aggradation will result primarily in increases in the duration of outof-bank flow, which will be reflected in increases in inundation depth and duration. For the size of floods shown in
Figure 7 (i.e. 1-in-05-year, 1-in-2-year), aggradation effects will be manifest as changes in both return periods of
out-of-bank flow and increases in inundated area.
There are interesting parallels here with the common distinction between climate variability and climate change.
Robson (2002) notes that some current hydrological changes can be attributed to hydrological variability and that
periods longer than 40 years are needed to distinguish between climate variability and climate change. Under climate
variability, a climate may differ from one period to the next but under climate change, a long-term alteration in the
climate is occurring. However, what may appear to be variability in the short term may be diagnostic of change in the
longer term. This is the sense in which the measured aggradation since 2000 (Figure 2) combined with the probable
tendency towards aggradation within this system that can be traced back to the 1980s (Figure 4) may be: (1) a shortterm response to temporarily enhanced sediment delivery associated with a period of more intense storm activity,
particularly given empirical (Reid et al., in press, a) and modelling evidence (Reid et al., in press, b) of the sensitivity
of sediment delivery to extreme rainfall events that do not require high levels of antecedent catchment saturation;
(2) part of a long-term trend towards higher sediment delivery rates due to a long-term increase in storm activity
linked to climate change; and/or (3) associated with a reach of river that is aggradational as a result of larger-scale
geomorphological controls.
Coulthard et al. (2002) and Reid et al. (in press, a) describe the large-scale geomorphological setting that leads to
this reach being aggradational, but it is quite possible that variability and change in sediment delivery rates are also a
contributing factor. We know that UK upland rivers can go through distinct cycles of alluviation in response to both
short- and long-term climatic fluctuations (e.g. Macklin et al., 1992) and that, at present, upland landscapes are more
sensitive to climatic fluctuations as a result of historical land-use change (Macklin and Lewin, 2003). For instance,
Coulthard and Macklin (2001) used modelling to show a severely non-linear interaction between climate change and
deforestation in a similar upland river catchment. S. C. Reid et al. (unpublished work) applied the same climate
change scenarios used in this paper to a continuous simulation, coupled hydrologysediment delivery model (Reid
et al., in press, b). For a subcatchment of this study reach (Buckden Beck, Figure 1) they estimated that, by the 2030s,
the volume of sediment delivered annually would increase by between 7 and 40 per cent according to the nature of
the changes in precipitation (essentially, the extent to which the precipitation increases is distributed evenly across all
rainfall events as opposed to being associated with a smaller number of additional extreme events). By the 2080s, the
increase was estimated to be between 28 per cent and 68 per cent. It appears that the indirect impacts of the climate
Copyright 2006 John Wiley & Sons, Ltd.

Earth Surf. Process. Landforms 32, 429446 (2007)


DOI: 10.1002/esp

Climate change and flood risk

443

change signal upon sediment delivery will act so as to reinforce the importance of short-term aggradation upon the
magnitude and frequency of floodplain inundation, and that short-term variability in aggradation rates may be superimposed on top of a longer-term, aggradational trend. This emphasizes that the changing flood risk identified here will
be strongly impacted upon by climate change, not only because of the direct impacts of climate upon the magnitude
and frequency of high flow events, but also because of the indirect effects of changing climate upon coarse sediment
delivery which in turn will impact upon coarse sediment delivery and hence aggradation rates. There has been much
debate over the extent to which land management impacts on catchment hydrology might be contributing to changing
flood magnitude and frequency (see review in OConnell et al., 2005). There has been much less acknowledgement
of the extent to which land management practices have sensitized river basins to climate change (Coulthard et al.,
2002; Macklin and Lewin, 2003) and in so doing are indirectly exacerbating flood magnitude and frequency. This
case study emphasizes the need to broaden the current focus of debate on the impacts of climate change upon river
flows to include a much stronger engagement with catchment sediment delivery processes and how these interact with
climate change.
The observations from this case study point to two important management implications. The first is in terms of
practical aspects of flood risk mapping. Common practice, at least in the UK, involves specially commissioned
geometrical surveys, application to one-dimensional hydraulic models, and model calibration based upon adjustment
of model parameters, notably channel friction (commonly Mannings n), to predict the water level recorded in historical events. If there has been bed aggradation since that historical event, then a lower value of n will be required to
reproduce that event than if the historical flood had been combined with historical geometrical survey. Thus, when
modelling future events, models calibrated in this way will underestimate water levels, and hence inundated area,
inundation depth and flood risk, as a result of having a calibrated value of n that is too low. The converse would occur
in systems that are degradational.
The second management issue is in relation to river management more generally, especially in upland environments
and where large-scale geomorphological controls have a major impact upon the sediment transfer process. In this
study, despite the delivery of large volumes of sediment being linked to major flood flows, the sudden reduction in
main valley slope at Hubberholme (Figure 1) coupled to the valley constriction at Starbotton (Figure 1) associated
with a major alluvial fan, creates a zone of natural sediment deposition. The extent to which aggradation is a more
generic characteristic of upland temperate environments, and hence the representativeness of this case study, needs
further consideration. However, this case study does raise a more generic question in relation to the appropriateness of
common assumptions made in river engineering with regard to river management. Our field observations show that,
for this case study, the natural river channel response to coarse sediment delivery is: (a) localized sedimentation in
point bars, commonly attached to either meander bends or the inside edge of weakly sinuous reaches; coupled to (b)
erosion of mixed fine and coarse material from channel banks on the channel margin opposite the zone of deposition.
We can hypothesize that this transfer of coarse sediment to storage, through channel migration, should be an integral
contribution to the observed rapid downstream fining, as well as a mechanism that leads to long-term valley fill.
Increasing sediment throughput through river realignment will simply transfer the sediment aggradation downstream,
eventually introducing coarser material into river reaches where it would not normally be found.
Where river bank stabilization has been advocated to protect farmland, properties and footpaths, the erosion
required to maintain channel capacity will be prevented. There is still a general view in some river management
communities that bank erosion (as opposed to artificially enhanced bank erosion, such as due to stock poaching) is a
problem that has to be prevented, rather than a natural geomorphological response to high rates of sediment delivery.
Consultant reports from the late 1990s for the reach of river considered in this paper advocated focusing management
on stretches where eroding banks should be stabilized (e.g. RKL Arup, 1999a,b). This assumes that a channel can be
engineered to be robust, to transfer sediment through it and so to absorb changes in sediment supply with only minor
adjustment in its form. However, channel morphology and stability in fact reflect the net sediment budget (e.g. Hooke,
2003), which in turn reflects: (i) the net erosion and deposition observed; and (ii) the supply and connectivity of
sediment from the hillslopes and upstream reaches. Where river management restricts (a) lateral movement of the river
and (b) transfer of sediment into floodplain storage, the channel bed becomes responsive (e.g. Schumm, 1979; Harvey,
2001; Hooke, 2003), and one response is aggradation. This will reduce channel conveyance and lead to the kinds of
impacts described above in relation to flood risk. Thus, in addition to naturally high rates of sediment delivery, this
kind of river management exacerbates flood risk. The Holocene record (e.g. Macklin and Lewin, 2003) tells us that
these systems can experience large increases in sediment delivery over relatively short time periods and it is this
historical record that emphasizes the need to incorporate changes in conveyance due to sediment aggradation more
explicitly into flood risk assessment. Whilst this research is based upon an upland case study, the observations will
apply to flood risk issues in any upland or piedmont system, where there is coarse sediment delivery and accumulation, but also where active river management precludes the possibility of coarse sediment transfer into floodplain
Copyright 2006 John Wiley & Sons, Ltd.

Earth Surf. Process. Landforms 32, 429446 (2007)


DOI: 10.1002/esp

444

S. N. Lane et al.

storage. Research by Pinter and Heine (2005) shows that river engineering on a river with a sand and silt bed may also
lead to systematic increases in stage for a given flow, and this may make these kinds of issues of relevance in other
river systems.

Conclusions
This paper has shown the importance of considering the effects of coarse sediment delivery and the associated bed
aggradation in the case study of a temperate upland environment. Whilst based on a case study, it has some very
important implications for climate change studies, flood risk assessment and river management more generally. In this
study, for a flood with a 1-in-05-year event, short-term bed aggradation (over a 16-month period) was shown to result
in about half the inundated area increase that results from applying climate changes estimated for the 2050s to the
same event. Consideration of the joint impacts of climate change and bed aggradation emphasized the non-linear
nature of system response, and the possibly severe and synergistic effects that come from combined direct effects of
climate change and sediment delivery. Such effects are likely to be exacerbated further as a result of the impacts of
climate change upon coarse sediment delivery. When couched in conventional flood frequency terms, these changes
do not result in a change in the return period of extreme flow events, but rather a reduction in the return period of the
flow events that lead to floodplain inundation. Thus, analyses that focus upon the changing magnitude and frequency
of flow events will overlook a key influence upon the return period and magnitude of flood inundation events. This is
particularly important as future climate changes will impact upon sediment delivery rates (S. C. Reid et al., unpublished work) as well as upon flows. Currently, the effects of bed aggradation are commonly overlooked in flood risk
mapping exercises that require calibration based upon combining contemporary bed surveys with historical flood
outlines, and this will lead to underestimation of the magnitude and frequency of floodplain inundation in an aggrading
system. Finally, the research emphasizes the importance of considering the natural process by which a river adjusts
through lateral migration to high rates of coarse sediment delivery, as river management activities that seek to slow
this process may exacerbate the effects of coarse sediment delivery upon river conveyance and hence upon flood risk.
Ultimately, and despite repeated calls (e.g. Sear et al., 1995), we still do not properly address geomorphological
processes in flood risk management studies and we desperately need tools that can predict the medium-term (annual to
decadal) response of river beds and river banks to sediment delivery in order to assess flood risk impacts. This should
include consideration of those types of system where aggradation issues are relevant. Whilst the prime candidates may
seem to be those with gravel, or coarser, bed material, aggradation-related shifts in stagedischarge curves have also
been observed in some systems with sand/silt boundaries (e.g. Pinter and Heine, 2005).

Acknowledgements
This research was funded by a PhD scholarship awarded to V.T. by the Iranian Science, Research and Technology Ministry and by
NERC Connect Grant NER/D/S/2000/01269 awarded to S.N.L., Professor A. McDonald (A.McD) and Professor M. J. Kirkby, and
by Environment Agency R&D award E1-108 awarded to A.McD and S.N.L. Downscaled climate data were provided by the UK
Climate Impacts Programme funded by the UKs Department of the Environment, Farming and Rural Affairs. The Design and
Imaging Unit, Durham, assisted with figure production. Tom Coulthard and an anonymous reviewer provided very valuable reviews
of this paper.

References
Arnell NW. 2003. Relative effects of multi-decadal climatic variability and changes in the mean and variability of climate due to global
warming: future streamflows in Britain. Journal of Hydrology 270: 195213.
Bates PD, De Roo APJ. 2000. A simple raster-based model for flood inundation simulation. Journal of Hydrology 236: 5477.
Blench T. 1969. Mobile-Bed Fluviology: A Regime Theory Treatment of Canals and Rivers for Engineers and Hydrologists. University of
Alberta Press: Edmonton.
Bradbrook KF, Lane SN, Waller SG, Bates PD. 2004. Two dimensional diffusion wave modelling of flood inundation using a simplified
channel representation. International Journal of River Basin Management 3: 113.
Cobby DM, Mason DC, Horritt MS, Bates PD. 2003. Two-dimensional hydraulic flood modelling using a finite-element mesh decomposed
according to vegetation and topographic features derived from airborne scanning laser altimetry. Hydrological Processes 17: 19792000.
Coulthard TJ, Macklin MG. 2001. How sensitive are river systems to climate and land-use changes? A model based evaluation. Journal of
Quaternary Science 16: 347351.
Coulthard TJ, Macklin MG, Kirkby MJ. 2002. A cellular model of Holocene upland river basin and alluvial fan evolution. Earth Surface
Processes and Landforms 27: 269288.
Copyright 2006 John Wiley & Sons, Ltd.

Earth Surf. Process. Landforms 32, 429446 (2007)


DOI: 10.1002/esp

Climate change and flood risk

445

Coulthard TJ, Lewin J, Macklin MG. 2005. Modelling differential catchment response to environmental change. Geomorphology 69: 222241.
Cunge JA. 1975. Two-dimensional modelling of floodplains. In Unsteady Flow in Open Channels, Vol. II, Mahmood K, Yevjevich V (eds).
Water Resources Publications; Fort Collins, Colorado.
Dugdale LJ. 2003. A hydrological modelling study to investigate the potential effect of climate change on the flood regime of a small upland
catchment: How certain can we be? MSc dissertation, University of Leeds.
Hardy RJ, Bates PD, Anderson MG. 1999. The importance of spatial resolution in hydraulic models for floodplain environments. Journal of
Hydrology 216: 124136.
Harvey AM. 2001. Coupling between hillslopes and channels in upland fluvial systems: implications for landscape sensitivity, illustrated
from the Howgill Fells, northwest England. Catena 42: 225250.
Harvey AM. 2002. Effective timescales of coupling within fluvial systems. Geomorphology 44: 175201.
Hey RD, Winterbottom AN. 1990. River engineering in National parks: the case of River Wharfe, U.K. Regulated Rivers: Research and
Management 5: 3544.
Hooke J. 2003. Coarse sediment connectivity in river channel systems: a conceptual framework and methodology. Geomorphology 56: 79
94.
Horritt MS, Bates PD. 2001a. Effects of spatial resolution on a raster based model of flood flow. Journal of Hydrology 253: 239249.
Horritt MS, Bates PD. 2001b. Predicting floodplain inundation: raster-based modelling versus the finite-element approach. Hydrological
Processes 15: 825842.
Hulme M, Jenkins GJ, Lu X, Turnpenny JR, Mitchell TD, Jones RG, Lowe J, Murphy JM, Hassell D, Boorman P, McDonald R, Hill S.
2002. Climate Change Scenarios for the United Kingdom: The UKCIP02 Scientific Report. Tyndall Centre for Climate Change Research,
School of Environmental Sciences, University of East Anglia.
Hunt JCR. 2002. Floods in a changing climate: a review. Philosophical Transactions of the Royal Society of London A 360: 15311543.
Institute of Hydrology. 1999. Flood Estimation Handbook, Volume 4, Restatement and appliation of the Flood Studies Report rainfall-runoff
method. Institute of Hydrology: Wallingford.
IPCC. 2001. Climate Change 2001: The Scientific Basis, Summary for Policymakers. Cambridge University Press, Cambridge.
James A. 1999. Time and the persistence of alluvium: River engineering, fluvial geomorphology, and mining sediment in California.
Geomorphology 31: 265290.
Knight DW, Shiono K. 1996. River channel and floodplain hydraulics. In Floodplain Processes, Anderson MG, Walling DE, Bates PD (eds).
Wiley: Chichester; 139181.
Korup O, McSaveney MJ, Davies TRH. 2004. Sediment generation and delivery from large historic landslides in the Southern Alps, New
Zealand. Geomorphology 61: 189207.
Lane SN. 2005. Roughness time for a re-evaluation? Earth Surface Processes and Landforms 30: 251253.
Lane SN, Brookes CJ, Kirkby MJ, Holden J. 2004. A network index based version of TOPMODEL for use with high resolution digital
topographic data. Hydrological Processes 18: 191201.
Macklin MG, Lewin J. 2003. River sediments, great floods and centennial-scale Holocene climate change. Journal of Quaternary Science
18: 101105.
Macklin MG, Rumsby BT, Heap T. 1992. Flood alluviation and entrenchment: Holocene valley floor development and transformation in the
British uplands. Geological Society of America Bulletin 104: 631 643.
Manning R. 1891. On the flow of water in open channels and pipes. Transactions of the Institute of Civil Engineers 20: 161207.
Marks K, Bates PD. 2000. Integration of high-resolution topographic data with floodplain flow models. Hydrological Processes 14: 2109
2122.
Middelkoop H, Daamen K, Gellens D, Grabs W, Kwadijk JCJ, Lang H, Parmet BWAH, Schdler B, Schulla J, Wilke K. 2001. Impact of
climate change on hydrological regimes and water resources management in the Rhine basin. Climatic Change 49: 105128.
Milley PCD, Wetherald RT, Dunne KA, Delworth TL. 2002. Increasing risk of great floods in a changing climate. Nature 415: 514517.
Nicholas AP, Mitchell CA. 2003. Numerical simulation of overbank processes in topographically complex floodplain environments.
Hydrological Processes 17: 727746.
OConnell PE, Beven KJ, Carney JN, Clements RO, Ewen J, Fowler H, Harris GL, Hollis J, Morris J, ODonnell GM, Packman JC, Parkin
A, Quinn PF, Rose SC, Shepher M, Tellier S. 2005. Review of impacts of rural land use and management on flood generation: Impact
study report. Department of the Environment, Farming and Rural Affairs R&D Technical Report, FD2114/TR.
Osborne TJ, Hulme M. 2002. Evidence for trends in heavy rainfall events over the UK. Philosophical Transactions of the Royal Society of
London A 360: 13131325.
Pinter N, Heine RA. 2005. Hydrodynamic and morphodynamic response to river engineering documented by fixed-discharge analysis,
Lower Missouri River, USA. Journal of Hydrology 302: 7091.
Prudhomme C, Reynard N, Crooks S. 2002. Downscaling of global climate models for flood frequency analysis: where are we now?
Hydrological Processes 16: 11371150.
Prudhomme C, Jakob D, Svensson C. 2003. Uncertainty and climate change impacts on the flood regime of small UK catchments, Journal
of Hydrology 277: 123.
Reid SC, Lane SN, Berney JM, Holden J. (in press, a). The timing and magnitude of coarse sediment transport events within an upland
gravel-bed river. Geomorphology.
Reid SC, Lane SN, Montgomery DR, Brookes CJ. (in press, b). Does hydrological connectivity improve identification of coarse sediment
delivery in upland environments? Geomorphology.
Reynard NS, Prudhomme C, Crooks SM. 2001. The flood characteristics of large UK rivers: Potential effects of changing climate and land
use. Climatic Change 48: 343359.
Copyright 2006 John Wiley & Sons, Ltd.

Earth Surf. Process. Landforms 32, 429446 (2007)


DOI: 10.1002/esp

446

S. N. Lane et al.

RKL Arup. 1999a. Dynamic Assessment of Unstable Reaches of the Upper Wharfe. Environment Agency Final Report.
RKL Arup. 1999b. Upper Wharfedale Best Practice Project Erosion Risk Assessment. Environment Agency Report.
Robson AJ. 2002. Evidence for trends in UK flooding. Philosophical Transactions of the Royal Society of London A 360: 13271343.
Schumm SA. 1979. Geomorphic thresholds: the concept and its applications, Transactions of the Institute of British Geographers 4: 485
515.
Sear DA, Newson MD, Brookes A. 1995. Sediment-related river maintenance: the role of fluvial geomorphology. Earth Surface Processes
and Landforms 20: 583708.
Stewart L. 1984. River Wharfe land drainage proposals from above Buckden Bridge down to Kettlewell and their impact upon fisheries and
angling. Report to Yorkshire Water Authority, Leeds.
Stover SC, Montgomery DR. 2001. Channel change and flooding, Skokomish River, Washington. Journal of Hydrology 243: 27286.
Tayefi V, Lane SN, Hardy RJ, Yu D. (in press). A comparison of 1D and 2D approaches to modelling flood inundation over complex upland
floodplains. Paper submitted to Hydrological Processes.
Yorkshire Water Authority. 1983. River Wharfe: Buckden Scheme. Report by Regional Land Drainage Committee, Yorkshire Water
Authority.
Yu D, Lane SN. 2006a. Urban fluvial flood modelling using a two-dimensional diffusion wave treatment: 1. Mesh resolution effects.
Hydrological Processes 20: 15411565.
Yu D, Lane SN. 2006b. Urban fluvial flood modelling using a two-dimensional diffusion wave treatment: 2. Development of a sub grid-scale
treatment. Hydrological Processes 20: 15671583.
Zanobetti D, Lorgere H, Preissmanm A, Cunge JA. 1968. Le modele mathematique du delta du Mekong. La Houille Blanche, Nos. 1, 4
and 5.

Copyright 2006 John Wiley & Sons, Ltd.

Earth Surf. Process. Landforms 32, 429446 (2007)


DOI: 10.1002/esp

También podría gustarte