See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/41198992

Bank retreat study of a meandering river reach case study: River Irwell

Conference Paper · November 2014

Source: OAI

CITATIONS READS

2 523

4 authors, including:

Lindsay Beevers Alessandra Crosato

Heriot-Watt University IHE Delft Institute for Water Education
71 PUBLICATIONS   698 CITATIONS    112 PUBLICATIONS   1,134 CITATIONS   

SEE PROFILE SEE PROFILE

Nigel George Wright

Nottingham Trent University
241 PUBLICATIONS   4,789 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Analysis of river meandering View project

SESAME View project

All content following this page was uploaded by Nigel George Wright on 31 May 2014.

The user has requested enhancement of the downloaded file.

 th th
7 ISE & 8 HIC
Chile, 2009

BANK RETREAT STUDY OF A MEANDERING RIVER REACH

CASE STUDY: RIVER IRWELL

ROXANA DURAN
LINDSAY BEEVERS, ALESSANDRA CROSATO, NIGEL WRIGHT
Water Science and Engineering, UNESCO IHE Institute for Water Education, Westvest
No7, 2611AX Delft, The Netherlands

ABSTRACT

Lack of data is often considered a limitation when undertaking morphological studies.

This research deals with morphological studies of small rivers experiencing bank erosion
processes when only limited data are available. A reach of the meandering gravel-bed
river Irwell (United Kingdom) is taken as a case study in order to analyze the bank retreat
process that is endangering the stability of structures located in the area.
Two models of different complexity are applied. The first is a physic-based one-line
meander model, computing bed topography, flow field and migration rate of meandering
rivers in areas with non-uniform erodibility. The second is a 2D physic-based
morphological model with a simplified bank erosion formulation. The model predictions
show that the morphological development of the river may endanger the existing
structures within the next five years, if no mitigation works are carried out.
The simple meander model produces satisfactory results in much less time using less
data. Moreover it allows taking into account the effect of riparian trees and soil
heterogeneity on bank erosion. Besides, the simple meander model appears appropriate
for probabilistic analysis of future river migration trends. A particular application of this
analysis is the study of climate change effects. The probabilistic analysis will be done in
a second phase of the study.

KEYWORDS: meandering river, bank erosion, morphodynamics

INTRODUCTION

Unstable meandering rivers are dynamic features of the landscape. Erosion and accretion
processes that take place in meandering rivers result in bend growth and channel
migration (Luna et al., 1957) [6]. These morphological changes are often source of
damage to infrastructure and in general, disturb the human activities. In order to prevent
these problems, morphodynamic study based on river modeling is needed. However, the
lack of data is often considered a limitation at the moment of undertaking this type of
studies because some of the existing computational codes require more data than the
available.
The objective of this study is to determine a convenient type of modeling to assess future
river migration trends of small meandering rivers with scarce available data. The study
will be done by comparing two models of different complexity. The first model is a one-
line meander migration model (Crosato, 1990, 2007) [2], and the second model is a 2D
morphological model with simplified bank erosion calculation method (Lesser et al.,
2004) [6]. For this purpose, a reach of the meandering gravel-bed river Irwell, in the
United Kingdom, is taken as a study case in order to analyze the bank retreat process that
is endangering the stability of structures located in the area.

MATERIAL AND METHODS

Several models have been developed to describe the interaction between flow, sediment
transport, grain-size sorting, and bed topography, most of them achieving undoubted
progress. Each of these models approach is founded on certain assumptions that restrict
its applicability e.g. models that assume river banks are non-erodible would be expected
to over-predict the depth of pools (Thorne, 1992) [11]. On the other hand, many practical
options to model bank erosion with a numerical approach have been developed. Simple
methods can be used to model bank retreat rate as long as the details of the erosion
process are not the main interest of the study.
Two models of different complexity are built for a small meandering river in order to
determine which one is more convenient to assess future river migration trends: The first
is a physic-based one-line meander model and a complex 2D model. The models are first
calibrated on the period 2003-2006 and thereafter, they are used to simulate the future
river changes for the period 2006-2010.

One-line meander model

The model combines a steady-state model for water flow and bed topography with a
time-dependent river migration model, which computes the bank retreat rates as a
function of near-bank hydraulic and morphological properties. Bank advance rate is
considered in a simplified way, assuming that it is equal to the bank retreat rate at the
other side of the river. To describe the bed topography, the flow equations are coupled to
sediment transport and sediment balance equations and consider as well the secondary
flow momentum convection in river bends.
The model describing flow and bed topography is a linear version of the more
complex non-linear model developed by Koch and Flokstra (1980) [5] and Struiksma et
al. (1985)[10] for curved channels valid for small to moderate Froude number.
The bank retreat rate is assumed proportional to the near bank tangential velocity excess
and to the near bank water depth excess. The excess velocity term accounts for the effects
of fluvial erosion at the toe of the bank and the excess water depth term accounts for
geomechanical instability.
2D Model
The 2D morphological model (Lesser et al., 2004) [6] is based on the continuity and
momentum equations for water, coupled to the balance and transport equations for
sediment, designed for curved channels. In the equations, the sediment transport direction
is corrected to take into account the transverse bed slope and the spiral flow in river
bends (Struiksma et al., 1985) [10], whereas the spiral flow itself is reproduced in a
parameterized way (Blanckaert and Vriend, 2003) [1].
Bank retreat is treated in a simplified way using the dry-wet cell method by Van der
Wegen et al. (2008) [12] and Roelvink, et al. (2006) [9]. This method assumes that every
time step the erosion of a wet cell, which is determined by the speed at which the eroded
material is removed, is fully or partially assigned to the adjacent dry cell according to a
calibration factor. This bank erosion formulation allows imposing a constant bank slope
to the entire model domain.

Study area description

The study area is located east of Rawtenstall, Lancashire, United Kingdom. The River
Irwell has its origin in the Irwell spring, 427 m above the mean sea level and the total
area of the river basin contributing to the study reach is 120.7 Km². At the study reach the
Irwell shows a meandering planform with fixed banks at the upstream and downstream
boundaries. The migration of the river may endanger the stability of the existing
structures: a weir, the A56 motorway bridge and the railway embankment. The main river
characteristics are listed in Table 1.

Table 1 Main characteristics of the study reach

Length of the study reach (Km) 1.00

Width (m) 11.80
Slope (%) 0.53
Annual mean discharge year 2006 (m³/s) 3.50
Higher observed discharge between 1996 53.00
and 2006 (m³/s)
Sediment type Gravel
Characteristic grain size D50 (mm) 11.20
Grain size D90 (mm) 30.50
Field survey
A field survey was carried out in order to obtain more information about the river. This
information was later used in the model set up and calibration. Structures such as weirs
and areas with river bank protection were identified and included in the model set up.
Location of areas with erosion and accretion processes as well as existing bars were
mapped and later used to calibrate the model. Also a sediment sample was taken to
determine the characteristic grain size which was used in the model set up.

Available data
Topography
Two sources of bed topography data are available: (i) Cross sections extracted from an
existing flow model developed within the Strategic Flood Risk Management Framework
(SFRMF), and (ii) LiDAR data, both of them property of The Environment Agency.
The data extracted from the SFRMF model consists of ten cross sections along the
whole river stretch that is about 800m long. The distance between cross sections is almost
90m which is not sufficient to reconstruct neither the river alignment nor the bed
topography. Consequently, this information was not used to build the morphological
model. However some of the cross sections were used to check bed levels of the LiDAR
topography.
The light detection and ranging (LIDAR) data has a resolution of 2m and the flown
date is April 2003, both characteristics make this data set suitable for building the model.
Since the LIDAR technology uses pulses of laser light that strike the surfaces of the earth
and measure the time of pulse return, a review of the main channel topography is needed
because the laser light may strike the water surface instead of the river bed. Some of the
SFRMF model cross sections are used to check and correct the main channel topography
of the LIDAR data.

Historic river configuration

The historical river configuration data consists of old maps, aerial photographs or satellite
images that reveal the changes of the river morphology during the past years. This
information is needed to calibrate and validate the model; therefore, it should be
subsequent to the topographical information used to build the model. The available
information comprises a satellite image, dated to June 2006 (source: Google Earth), and a
set of photographs of year 2007. The latter is a good reference to estimate bank erosion
rates and identify medium to large morphological changes. However, photographs do not
provide precise or quantitative information.
Hydrology
The main source of hydrological data is the Irwell vale gauging station where the data
available comprises discharges measured every 15 minutes. The measured discharges
correspond to the river Irwell plus inflows from river Ogden that have to be subtracted
since it flows into the river Irwell downstream of the study reach. The discharge
corresponding to the river Irwell is estimated as 75% of the total discharge at Irwell Vale.
A second source of information is the existing flow model developed within the SFRMF
which provides water levels at ten cross sections along the study area. This information is
used to calibrate the model developed for this study.
From the ten cross sections of the SFRMF model, considering a Chezy roughness
coefficient of 22.53 m1/2/s and using a mean slope of 0.0053, the bank-full discharge was
calculated to be 45 m³/s.

Sediment
The particle size of the bed material varies considerably in a lateral direction and along
the river. For these reason, it is desirable to have several samples from the main channel
and from point bars along the river. However, for the case study only one sediment
distribution curve is available, that corresponds to a sample taken at a point bar. This
sample only comprises the gravel and sand phases, while, the sediment in the area
contains also cobbles and boulders. The sediment characteristic sizes are shown in Table
1. The Meyer-Peter and Muller (1948) [8] formula was used to calculate the sediment
transport capacity of the river.

RESULTS

Model calibration
The models have been calibrated on the period 2003 to 2006. The results of calibration
describe the predicted river alignment and bed topography in 2006 which are compared
with the available river configuration information for the same year. Computed River
alignment and bed topography are shown in Figures 1 and 2.
 (a)

(b)

Figure 1. Calibrated one-line meander model. (a) River alignment (b) Bed topography.

River alignment
The results of the one-line meander model are satisfactory (Figure 1) because they match
both bed topography, in terms of location of point bars and pools, and river alignment.
Use of lower erodibility coefficients in areas with herbaceous vegetation and in the
railway embankment contributed to better simulate the river behavior (Gray and Mac
Donald, 1989) [4].

The numerical approach that simulates bank erosion in the 2D Model is based on the
distribution of the erosion between the source wet cell and the adjacent dry cell. Due to
the grid resolution, of about 1m by 1m, the resulting erosion rate is also 1m, where bank
erosion rate is estimated to be 0.60 m/year. Therefore, a distortion of the process is
introduced. Better results are expected from a finer grid size; however, this would require
the reduction of the time step and increase of the cell number, increasing the
computational time. Furthermore, in the 2D model the river becomes wider (Figure 2)
because the bank advance process is not taken into account and because the dry wet cell
method promotes erosion even in straight reaches.
 The model has the shortcoming of assuming that the bank erodibility is constant.
This assumption may apply when the bank material is homogeneous along the river, the
geotechnical properties are uniform and external loads and vegetation are the same along
the river banks. However, these conditions are hardly ever met, as for the river Irwell.
Therefore, the assumption of constant bank erodibility does not reproduce the natural
behavior of the river.

Bed topography
The one-line meander model predicts well the magnitude and location of the bars present
in the river except bars 8 and 9 (Figure 2).
The one-line meander model does not have the option to model weirs, therefore the
effect caused by the weir upstream of the bridge is not considered in the results. The back
water effect of the weir reduces flow velocity and sediment transport capacity so that
sedimentation takes place. As a consequence of flow velocity reduction the bar number 5
of Figure 2 is larger than the model prediction and bar number 9 does not exist.

(a)

(b)

Figure 2. Calibrated 2D model. (a) River banks alignment. (b) Bed topography.
 The best fitted calibrated 2D model partially reproduces the morphological
development of the river. In the studied bend, the resulting point bar (3) is bigger than the
existing one. The erosion of the outer bank is overestimated in (4) which results in
washing away the existing bar (6). Additionally, the bank retreat process (5) that takes
place downstream of the point bar (3) is not represented by the model and the existing
pool is filled with sediment (Figure 2). On the other hand, downstream of the weir, the
model reproduces the bar formation and the development of the pool at the outer bend
(Figure 2).

Future planimetric changes

The one-line meander model predictions show that in 2010 show the river starts to erode
behind the existing channel walls upstream and downstream of the weir. However, the
model does not take into account the effect of the weir that lowers the flow velocity
upstream, and therefore, the meander growth. With regard to the railway embankment
stability, the model indicates that by year 2010 the left river bank might reach the railway
embankment toe.
(a)

(b)

Figure 3. Irwell alignment prediction year 2010. a)One-line meander model. b)2D model
DISCUSSION

From the perspective of the suitability to model small meandering rivers with limited
data, it is clear that the one-line meander model offers important advantages because of
its simple form and because it allows taking into account factors that reduce the river
migration, such as bank protection and vegetation. Even though this model was designed
for sand-bed rivers, in this study, the model was successfully calibrated for a gravel bed
river.
Meander migration is the result of bank retreat and opposite bank advance. In the
case of models that only consider the bank erosion and retreat, like the 2D model used in
the research, the river width tends to increase. Better results are expected from the 2D
model if bank advance is considered in the computation of the morphological changes, at
least in a simplified way.
The one-line meander model uses a few data like a constant channel width, a mean
slope and a characteristic grain size. All these parameters are single values that are
accompanied with great uncertainty that influences the results. When applying a 2D
model, it is possible to use detailed information to describe e.g. topography, sediment and
discharge regime, reducing data uncertainties.
On the other hand, the short calculation time in the one-line meander model, makes it
possible to carry out a probabilistic analysis of the uncertain parameters to observe their
influence on the results. The same type of analysis can be applied to make an assessment
of discharge variation due to climate change. The computational time of the 2D model is
too large to carry out a large number of calculations as it is necessary for probabilistic
analysis.

CONCLUSIONS

This research deals with morphological studies of small rivers experiencing bank erosion
processes when only limited data is available. A reach of the meandering gravel-bed river
Irwell (United Kingdom) is taken as a case study in order to analyze the bank retreat
process that is endangering the stability of structures located in the area. Two models of
different complexity have been applied. The first is a physic-based one-line meander
model, computing bed topography, flow field and migration rate of meandering rivers in
areas with non-uniform erodibility. The second is a 2D physic-based morphological
model with a simplified bank erosion formulation.
According to the one-line meander model, the morphological development of the
river Irwell will endanger the existing structures within the next five years, if no
protection works are undertaken. However, this prediction is accompanied with
uncertainty due to the limitations of the available data and, on the other hand, of model
assumptions.
 In principle, both computational codes, the one-line meander model and the 2D
model, require the same type of information. The main difference lies on the fact that the
one-line meander model requires simple topographic information: river alignment,
characteristic river slope and width, which makes this code suitable to undertake river
morphology modeling with limited information.
The one-line meander model appears to be the most appropriate computational code
to study the morphological development of the case study river. By approaching the
phenomena through the analyses of the physical processes, and by making possible to
model the heterogeneity of bank erodibility, a good representation of the natural river
development is obtained. This is not achieved with the 2D model where a simplified
method for bank erosion calculation and a unique factor for erosion of adjacent dry cells,
implicitly assuming constant bank erodibility, are used.
While the 2D model takes days to run, the one-line meander model makes a similar
calculation in few seconds what makes it appear appropriate for probabilistic analysis of
future river migration trends. This probabilistic analysis will consider the uncertainties
linked to river geometry and sediment size. A particular application of this analysis can
be the study of climate change effects which involves changes in the discharge regime.
The probabilistic analysis will be done in a second phase of the study.

The effects of changing the river width and the sediment grain size will be further
studied. Additionally the effect of climate change on the river morphology will be studied
by considering possible changes on discharge regime.

ACKNOWLEDGEMENTS

This study was carried out with the support of UNESCO-IHE institute for water
education, The Netherlands Fellowship Program and Edenvale Young.

REFERENCES

[1] Blanckaert, K. and H. J. Vriend (2003). "Nonlinear modeling of mean flow

redistribution in curved open channels", Water resources research, AGU. 39(12).
[2] Crosato, A. (1990). “Simulation of meandering river processes”. Communications
on Hydraulic and Geotechnical Engineering, Delft university of Technology, Report
No. 90-3, ISSN 0169-6548.
[3] Crosato, A. (2007). "Effects of smoothing and regridding in numerical meander
migration models", Water resources research, AGU. 43(1): W01401.
[4] Gray, D. H. and A. MacDonald (1989). "The role of vegetation in river bank
erosion", Hydraulic Engineering: 218 - 223.
[5] Koch, F. G. and C. Flokstra (1980). "Bed level computations for curved alluvial
channels", Proc. of the XIX IAHR Congress, New Delhi, India 2: 357.
 [6] Lesser, G. R., J. A. Roelvink, J.A. van Kester and G. S. Stelling, (2004).
"Development and validation of a three-dimensional morphological model", Coastal
Engineering 51(8-9): 883-915.
[7] Luna, B. L. and M. G. Wolman (1957). “River channel pattern: braided,
meandering and straight”, U. S. G. Survey.
[8] Meyer-Peter, E. and R. Müller (1948). “Formulas for bed-load transport”, Proc.,
2nd Meeting, IAHR, Stockholm, Sweden.
[9] Roelvink, D., G. Lesser, and M. van der Wegen. (2006) "Morphological modeling of
the wet-dry interface at various timescales." the 7th Int. Conf. on hydroscience and
engineering, Philadelphia, USA.
[10] Struiksma, N., K. W. Olesen, C. Flokstra, and H. J. De Vriend, (1985). "Bed
deformation in curved alluvial channels." Journal of hydraulic research, IAHR
23(1): 57-79.
[11] Thorne. (1992). "Bend scour and bank erosion on the meandering Red river,
Louisiana." In Lowland Floodplain Rivers: Geomorphological Perspectives, Carling
PA, Petts GE (eds). Wiley: Chichester; 95–115
[12] Van der Wegen, M. Wang, Z.B. Savenije, H.H.G. J.A. Roelvink (2008). “Long-term
morphodynamic evolution and energy dissipation in a coastal plain, tidal
embayment”, J. Geophys. Res., DOI: 10.1029/2007JF000892.

View publication stats