You are on page 1of 9

E-proceedings of the 36th IAHR World Congress

28 June – 3 July, 2015, The Hague, the


Netherlands

NUMERICAL MODELLING OF RIVER DUNES IN THE WAAL UNDER EXTREME FLOODS

(1) (2) (1) (3) (4)


SANJAY GIRI , SATOMI YAMAGUCHI , MOHAMED NABI , JONATHAN NELSON & YASUYUKI SHIMIZU
(1)
Independent Research Foundation DELTARES, Delft, The Netherlands,
e-mail:Sanjay.Giri@deltares.nl
(2)
Civil Engineering Research Institute, Sapporo, Japan,
e-mail: kawamura-s@ceri.go.jp
(3)
US Geological Survey- Geomorphology & Sediment Transport Laboratory, Denver, USA,
e-mail: jmn@usgs.gov
(4)
Hokkaido University, Sapporo, Japan,
e-mail: yasu@eng.hokudai.ac.jp

ABSTRACT
This The Waal branch of Rhine system is considered to be an important fairway. The flood events and extensive
navigation cause significant morphological changes, which create difficulties for safe and efficient navigation particularly
during low water period. Therefore, it is important to predict the bed level changes, particularly evolution of bed forms
during the flood event including both the high water and low water periods. Besides this, it is also important to
understand the bed form evolution process and resulting flow resistance during extreme floods in order to quantify their
effect on water levels.
Many attempts have been made to improve both understanding and predictive capability of bed form evolution, transition
and associated resistance under varying flow conditions. The interaction among the flow-field, bed forms and sediment
transport is quite complex and difficult to capture in simple models. A number of our recent studies have made some
noticeable efforts to numerically replicate the dune formation and evolution processes. A morphodynamic model was
been developed that successfully reproduces fluid and bed form dynamics in a coupled manner under arbitrary steady or
unsteady flow condition.
This study is the continuation of our previous efforts, in which we have attempted to replicate the flow and bed form
evolution process under a couple of synthetic (extreme) flood scenarios in the river Waal. Within the scope of this study,
at first, we have analyzed the observed data during moderate floods in the Waal. The bed level data were measured
during a moderate flood event in 1997. A vertical two-dimensional morphodynamic model with a free-surface flow
condition, which includes equilibrium, non-equilibrium bed load as well as suspended sediment transport modules, has
been used to replicate the flow and bed form evolution processes under extreme flood conditions. The boundary
condition (unit discharge and water levels) for vertical two-dimensional model was extracted from a calibrated one-
dimensional hydrodynamic model SOBEK, developed for the Rhine branches. A preliminary result shows that bed forms
are present even during extreme flood condition. This study reveals the capability of a numerical model, which can be
applied to predict real-world river dunes and drag evolution with hysteresis effects in physics based manner.

Keywords: River dunes, The Waal, numerical modelling, vertical 2D model, 1D SOBEK model

1. INTRODUCTION
The development and migration of micro-scale bed forms in alluvial rivers during floods is one of the widely concerned
issues in river engineering practice owing to their significance for flood safety issues as they exert substantial hydraulic
resistance that has an effect on flood levels. At the same time, understanding and predicting the evolution and migration
of bed forms and their geometrical characteristics are of great importance for ensuring safe and effective waterways in
lowland rivers, since they adversely affect the navigable depth during low flow period. Consequently, there are basically
two important aspects associated with micro-scale bed forms in Ducth Rhine branches, namely (i) the evolution of bed
forms including their quasi-equilibrium geometric characteristics and the migration, which is important to optimize the
measures to improve the navigability during low flows and, (ii) the bed form induced evolution of hydraulic resistance
and its effect on water level during extreme conditions (e.g. under design discharge in Rhine branches).
There are a number of past studies, which have contributed to address the issues in different ways, e.g. theoretical
analyses based on linear and weakly non-linear stability theories, experimental studies (Yamaguchi & Izumi, 2005;
Fredsøe, 1982; Venditti et al., 2005 etc.) as well as field observations (Itakura et. al, 1986 etc.) were rather common,
which provide the essential knowledge on the related issue. In recent years, the numerical modelling approach, which is
capable of simulating the flow and bed form dynamics in alluvial rivers, is being increasingly applied, particularly to get a
fundamental insight into the physical processes associated with this phenomenon. A number of attempts have been
made so far in order to develop and apply the physics-based numerical modeling technique to replicate the bed form

1
E-proceedings of the 36th IAHR World Congress,
28 June – 3 July, 2015, The Hague, the Netherlands

evolution processes (Nabi et al., 2012, 2013; Niemann et al., 2010; Shimizu et al., 2009; Paarlberg et al, 2009; Giri &
Shimizu, 2006; Tjerry & Fredsøe, 2005 etc.). These efforts demonstrate the recent advances and capabilities of the
computational modelling approach to address such complex issue. In most existing morphodynamic models, particularly
the sediment transport approaches are usually conventional, e.g. consideration of equilibrium bed load transport
formulation, and in some cases consideration of suspended sediment transport with or without advection-diffusion of
sediment concentration in water column. A number of past studies have created an enduring debate on the importance
of sediment transport mechanism for replicating the bed form evolution processes either by using conventional
equilibrium transport formulations with local bed slope effect or by using non-equilibrium transport formulation with pick-
up and deposition functions as separate subsystems, since the cutting edge numerical modelling approach allows to
consider the local variability in a small spatial and temporal scales.
Some of the past studies revealed that very simple models were able to reproduce bed form evolution process. For
example, a bed form growth model, proposed by Jerolmack & Moharig (2005), was able to reproduce the bed form
growth without any hydrodynamics by simply using a sediment mass balance equation with an additional term to correct
a preset boundary shear stress based on the local bed slope gradient as well as the effect of avalanche in sediment
transport formulation. They hypothesized that the detailed flow field was not important for a first order description of the
bed dynamics. However, the physically-based hydrodynamics is of significance to compute bed form evolution and more
importantly associated form drag evolution in a physically consistent way. Furthermore, the micro-scale non-equilibrium
transport, which is partly caused by the near-bed flow variability, appears to be a relevant physical process to be
considered so as to reproduce the morphodynamics of micro-scale bed forms. Some recent studies (Uchida & Fukuoka,
2013; Giri & Shimizu, 2006; Onda & Hosoda, 2004) revealed the importance of hydrodynamics with local variability as
well as the non-equilibrium transport on morphodynamic modeling of bed form evolution and associated form drag.
Zhang et al. (2013) investigated the significance of grain-scale non-equilibrium bed-load and suspended load transports
over a flatbed under unsteady flow condition. They argued that the equilibrium model could be regarded as the special
case of non-equilibrium model, in which the bedload adaptation length is related to grid-cell density, so the equilibrium
transport model is inappropriate to use particularly under unsteady flow condition. We argue that, even under steady
flow condition, the complex near-bed flow with local unsteadiness appears to be leading to the micro-scale non-
equilibrium transport with the adaptation length smaller than the bed form length.
In one of our recent studies (Giri et al., 2014), we attempted to address a fundamental issue associated with the effect of
different sediment transport mechanisms on bed form dynamics by making use of a couple of advanced computational
models, which have been developed by our group of researchers (Giri & Shimizu, 2006; Yamaguchi et al., 2009; Nabi et
al., 2013; Giri & Nabi; 2012). A complex three-dimensional morphodynamic model with advanced sediment transport
modelling approach, which includes particles pick-up, transport by sliding and flying with saltation and/or suspension,
and finally deposition of the particles. The sediment transport was modelled in a Lagrangian framework based on the
motion of the particles, coupled with near-bed hydrodynamic forces. This approach is capable to provide a better insight
into the physics of sediment transport phenomena enabling to simulate details of the sediment motion such as jumping,
rolling, sliding, flying including their effects as a combined system as well as separate effects of each subsystem on
micro-scale and average bed evolution processes. Secondly, the vertical two-dimensional morphodynamic model with a
free-surface flow condition, which includes the equilibrium, non-equilibrium and suspended sediment transport modules,
was used to evaluate the effects of these transport mechanisms. The 2D model was applied to assess the effects of
these transport approaches on bed form evolution process based on the simulations of large scale real-world cases,
namely an experiment in Chiyoda channel in Tokachi River. The model was briefly used for the schematized case of the
river Waal to replicate the bed forms during observed flood event.
This study is the continuation of our previous efforts, in which we have attempted to replicate the bed form evolution
process during observed flood events as well as a case study of dune evolution and stage-discharge relations for a
couple of synthetic (extreme) flood scenarios in the river Waal.

2. RIVER DUNES IN THE WAAL: FIELD OBSERVATION


In the Waal River (The Netherlands), bed level data were measured during two different flood events in 1997 and 1998
respectively (Wilbers, 2004; Sieben, 2006; van Denderen, 2014). The first measurement during two weeks of a moderate
flood in 1997 was carried out by using single-beam echo-soundings for a reach of 10 km in the middle part of the Waal
with sediment bed of 1 mm reach-averaged median diameter. The soundings were clustered per day, comprising four
days during the rising stage, two days during the peak flow and eight days during the falling stage of the flood wave
(Sieben, 2006). The main channel depth varied from 5 to 9 m during the soundings, while the average velocities in the
main channel remained more or less constant (1.0 to 1.2 m/s). The second field measurement was carried out by using
multi-beam soundings of a 2 km reach in the upper part of the river Waal during a month of high flow in 1998. In order
to track the migrating bed forms, bed levels were measured twice a day, during three days at the rising stage, two days
at the peak and five days at the falling stage of the flood (Wilbers, 2004). During the multi-beam measurement, main
channel depth varied from 8 m to 12 m and the averaged velocities were estimated ranging from 1.0 to 2.0 m/s (Sieben,
2006). In this study, we have chosen 1997 flood event to evaluate the model performance, and in turn the model has
been applied to simulate the bed form evolution for a couple of synthetic (extreme) flood waves.

Figure 1 shows the reach in the river Waal, where the bed level measurement was carried out during moderate flood
event of 1997. Figure 2 depicts the observed flow and reach averaged bed form height and length during moderate flood
events in 1997. While Figure 3 shows the observed stage-discharge for different floods in the Waal (the observation

2
E-proceedings of the 36th IAHR World Congress
28 June – 3 July, 2015, The Hague, the Netherlands

station is located at Tiel Waal, which is near the reach of interest for this study). As it can be seen from this figure, the
hysteresis effect is evident during these flood events.

Figure 1. The reach with the bed level measurement and dune features in the Waal (van Denderen, 2014)

Figure 2: Observed dune height (upper plot) and dune length (lower plot) during 1997 flood

3
E-proceedings of the 36th IAHR World Congress,
28 June – 3 July, 2015, The Hague, the Netherlands

Figure 3: Q-h relation at the Tiel Waal for different flood events (van Denderen, 2014)

3. NUMERICAL MODELS

3.1 One-dimensional model (SOBEK)


SOBEK 1D is one of the models available in SOBEK 3, which is one of the in-house software of DELTARES. SOBEK
1D is developed for the simulation of water flows in open channels. It combines functionality of the former SOBEK-
River Estuary and SOBEK-RIVER and is capable of modelling river systems, estuaries, streams and other types of
alluvial channel networks. The software computes efficiently and robustly the one-dimensional shallow water problem in
simple or complex channel networks with more than thousand reaches, cross sections and structures.
The model solves the full Saint-Venant equations using the staggered grid numerical scheme (Stelling and Duinmeijer,
2003; Stelling and Verwey, 2006). The model allows to apply various types of boundary conditions, as well as to define
lateral inflows and outflows using time series or standard formulae. The networks can be branched or looped. The
model is capable of modelling complex cross-sectional profiles consisting of multiple roughness sub-sections, e. g. left
floodplain, right floodplain and main channel.
A hydrodynamic model for the whole Rhine branches has been developed and calibrated. In this study, we have made
use of this model to compute the boundary conditions for the vertical two-dimensional model (mentioned below)
including for the design discharge scenario. The model has particularly been used to deduce the unit discharge and
water level values, since the vertical 2D model uses the unit discharge. The cross-sections in the 1D model were
schematized in a symmetrical (tabulated) shape, which is based on the measured multi-beam sounding (for the main
channel) and DTM for the floodplain. This enables to estimate the flow width (and corresponding unit discharge) at given
water level in an easy manner, which can later be used as a unit discharge boundary condition in vertical 2D model.
3.2 Vertical two-dimensional model
The governing equations for vertical two-dimensional flow are expressed in Cartesian coordinate system as follows:
Continuity equation:

u v
 0 [1]
x y
where, x and y = coordinates in horizontal and vertical direction respectively; u and v = components of velocity in
horizontal and vertical direction respectively.
Momentum equations:
u
t
u
u
x
v
u
y

1 p 

 x x
 u' u' 

y
 
 u ' v'   [2]

v
t
v
u v
x
v
y

1 p 

 y x
 u ' v'  

y

 v' v'  g   [3]

where,  u 'u ' ,  u ' v ' and  v'v' = Reynolds stress tensors,  = fluid density, g = gravitational acceleration and p =
pressure.
Equations [1]-[3] have been transformed from x, y, t Cartesian coordinate system to a moving boundary fitted ξ, η and τ
coordinate system.
Pressure field has been computed as non-hydrostatic. A nonlinear k-ε turbulence closure has been used in this model,
which enables to consider the anisotropy of Reynolds stresses (Kimura and Hosoda, 2003).
The boundary conditions are no slip at the bed and free flow at the surface (free surface flow condition). The following
expression for near-bed region has been used:

4
E-proceedings of the 36th IAHR World Congress
28 June – 3 July, 2015, The Hague, the Netherlands

up 1 yp
 ln [4]
u*  y0
where, u p = velocity at near-bed grid point, u* = local bed shear velocity,  = Karman constant, y 0  k s 30 , k s  2d , in
which ks = roughness height and d = bed particle diameter.
Computation of time-dependent water surface change is of importance for the realistic reproduction of free surface flow
over migrating bed forms. Majority of morphodynamic models developed earlier assumed the rigid lid water surface
condition to achieve numerically stable solution. It is of significance to note that we have succeeded to achieve stable
and reasonable solution under free surface flow condition over migrating bed form. Along with the flux continuity, the
kinematic condition is imposed along the free surface (at y = H) to compute the temporal water surface elevation. The
kinematic condition, which constraints fluid particles to remain on the water surface at any time following the local flow
velocity, is expressed as follows:
H H
u v [5]
t x
H  yb  h [6]
where, yb = bed elevation and h is local flow depth.

The transformed equations have been numerically solved by splitting them into non-advection and pure advection phase.
The non-advection phase has been computed using central difference method. The pressure field has been resolved
using SOR method. The advection phase has been calculated using a high-order Godunov scheme known as CIP
method (Yabe and Takei, 1988; Yabe et al., 1990). In this scheme, at very small time increment, the change in time of
velocity components at a point in space can be split into the time evolution of the inhomogeneous terms and the time
evolution at a point due to the advection of the field (Giri and Shimizu, 2006).
As far as the sediment transport model is concerned, an equilibrium bedload transport formula of Meyer-Peter and
Müller with the inclusion of local bed slope effect, as proposed by Fredsøe (1974) and further used by Niemen et al.
(2011), has been incorporated in our model so as to evaluate the applicability and to compare with the non-equilibrium
formulations. The relationship reads as follows:
1.5
  
qb*    *   *c    [7]
 x 
where qb* = dimensionless bed load transport rate; ƞ = local bed elevation; and µ = a constant based on angle of repose
for sediment (= 0.1 as proposed by Fredsøe, 1974).
Despite the fact that the application of equilibrium transport mechanism with bed slope effect is still questionable
(Yamaguchi et al., 2009) particularly for the case of greater variation in the bed form characteristics (for example the
cases of time varying hydrograph, in which the adjustment of wavelength and height during transient conditions is more
pronounced), this is found to be applicable for the large scale case with mild topographic and flow variability as well as
large ratio between flow depth and bed material as all these conditions are valid for the case of the river Waal.
For the morphodynamic computations, following sediment balance equation is used for the model with equilibrium
transport:

1    t   qx
p
b [8]

Where, λp = porosity.

4. COMPUTATIONAL RESULT AND ANALYSIS

4.1 Modelling river dunes under moderate flood of 1997


At first, the proposed numerical model has been applied to simulate the moderate flood of 1997, for which the bed level
observation was made during the event. Figure 4 shows the result of numerical simulation on bed form evolution and
depth variation. ). In the numerical simulation, the dune height does not appear to be decaying much during falling limb,
attaining almost equilibrium state after the flood till the end of the simulation. While observation showed some decay of
the bed form height during falling stage of the flood. Moreover, the dune height appeared to have been overestimated by
the model. Nevertheless, it should be noted that the methodology that was used to determine the bed form height from
the bed level sounding data based on reached averaged values. While in the simulation, for the sake of simplicity, the
bed form height has been determined as the difference between maximum and minimum bed levels at each time step.
So, the overestimation of the bed form height can partly be attributed to this fact as well. On the other hand, the bed
form length during 1997 flood has been predicted by the model reasonably well. The result shows the wavelength in the
range of 30-50 m before, during and after the flood (as can be seen from Figure 5), which is also the case in field
observation data as shown above in Figure 2.

5
E-proceedings of the 36th IAHR World Congress,
28 June – 3 July, 2015, The Hague, the Netherlands

It should be emphasized that no proper calibration was performed, and all the coefficient and parameters that have been
used in the sediment transport model have not been tuned, since there could be a number of uncertainty in the observed
dune characteristics, which have been in effect deduced as reach-averaged quantity from the bed level measurement
data.

Figure 4: Simulation of bed form evolution and depth variation during 1997 moderate flood

Figure 5: Simulation result of bed form evolution during 1997 flood event

4.2 Simulating bed forms and depth evolution under extreme floods

The simulations of bed form evolution, depth and stage-discharge for observed flood event has provided some
impression about the model performance and capability to assess such phenomenon in physics-based manner.
Consequently, we have attempted to apply this model to replicate few synthetic scenarios with extreme event in order to
assess the bed form evolution as well as corresponding flow depth and stage-discharge relations under such conditions.
Such straightforward evaluation could be of use in case of low land rivers given the fact that the issues, associated with
evolution of bed forms, does not appear to have received enough attention for detailed qualitative and quantitative
analyses of their effects on flood levels and sediment management.

In this study, two synthetic flood scenarios have been considered. The peak flow for both scenarios is same with the
maximum unit discharge of 23 m 2/s, which is rather high for Rhine branches (depicted in Figure 6 and 7). In first
scenario (scenario 1, depicted in Figures 6), the peak flow remains instantaneous. In second scenario (scenario 2,
depicted in Figure 7), the duration of the peak flow has been set to be 10 hours. Despite the initial imposed bed forms,
the computation has been carried out first 100 hours under low constant discharge before the flood waves in order to get
the quasi-equilibrium bed forms.

The computational results on bed form evolution for these synthetic scenarios are depicted in the same figures (Figure 6
and 7). It is to be noted that the depicted dune height is the maximum instantaneous value (difference between
maximum and minimum bed level) within the computational domain. As computational result reveals, the dune height
after the flood wave (at the end of the simulation) is different in these two scenarios despite the fact that the peak value
is the same. This leads to the fact that the duration of the flood wave (flood peak in current simulation) has an effect on
bed form amplitude, which appears to be associated with the adaptation of the bed form to the flow. Moreover, the
relaxation time for attaining the maximum bed form height is different in these scenarios. In both scenarios, the bed form
attains its maximum value during the falling limb of the flood wave; however, the relaxation time to attain maximum bed
for height is much longer in scenario 1 in comparison with scenario 2. In addition, the dune height after the flood is

6
E-proceedings of the 36th IAHR World Congress
28 June – 3 July, 2015, The Hague, the Netherlands

larger in scenario 2, which implies that the maximum equilibrium dune height is not attained in scenario 1, since the
peak is instantaneous. This can well be inferred from the dune height evolution as depicted in these figures.

The simulated average depth variation can be seen in Figure 8. The flow depth during the peak flow in both scenarios is
the same (about 13 m). However, as it can be seen from the result, the depth varies between 12-14 m in scenario 2,
which can be attributed to the larger dune height due to longer duration (10 hours) of the peak flow in this scenario,
leading to the different form drag evolution during peak flow unlike the scenario 1, in which the peak is instantaneous.
Furthermore, the stage-discharge relationship for both scenarios has been depicted in Figure 9 and 10. The hysteresis
effect is evident in both scenarios, and is larger in scenario 2. This implies that the hysteresis effect is associated with
the form drag evolution, exerted by the bed forms.

Figure 6: Scenario 1 with initial dune height of 0.8 m

Figure 7: Scenario 2 with initial dune height of 0.8 m and 10 hours duration of peak flow

Figure 8: Flow depth variation for two scenarios

7
E-proceedings of the 36th IAHR World Congress,
28 June – 3 July, 2015, The Hague, the Netherlands

Figure 9: Stage-discharge relationship for scenario 1

Figure 10: Stage-discharge relationship for scenario 2

5. CONCLUSIONS

A vertical 2D morphodynamic model has been applied for replicating the real-world bed form evolution process
under the observed flood condition in the river Waal. Furthermore, the model has been applied to simulate the flow
and bed form evolution process under a couple of synthetic flood scenarios. The computational result for observed
moderate flood event of 1997 is found to be satisfactory; however, more detailed study should be carried out by
analyzing the hydraulic and morphological data in more elaborative and careful way. Likewise, the simulations
under synthetic (extreme) flood scenarios reveal that the bed form is always present in the river Waal, exerting
additional form resistance to the flow, which is important to consider while designing flood control measures (like
Room for the River measures) as well as for flow and sediment management practices, associated with sustainable
and safe navigation.

It should be noticed that all sediment transport parameters have been kept to be conventional values, and thus the
mode has not been calibrated and tuned to fit the observation. The computed results can be considered to be rather
satisfactory given the scale and complexity of the problem and the data. The model is able to reproduce bed form
evolution as well as hysteresis of stage-discharge in physics based manner. This demonstrates the prospective of
proposed computational model that can be used for the real-world application.

Despite the fact that the application of equilibrium transport mechanism with bed slope effect is still questionable for
the case of greater variation in the bed form characteristics (for example, the cases of time varying hydrograph, in
which the adjustment of bed form wavelength and height during transient conditions is more pronounced), it is found
to be applicable for the large scale case with mild topographic and flow variability as well as large ratio between flow
depth and bed material as all these conditions are valid for the case of the Rhine branches.

ACKNOWLEDGMENTS
The authors would like to express their thankfulness for providing data and information to Arjan Sieben (RWD-WD), R.
P. van Denderen (TU Delft) and Koen Berends (Deltares).

REFERENCES
Giri, S., Nabi, M., Kawamura, S., and Shimizu, Y. (2011). “State-of-the-art modeling of bedform dynamics: Vertical 2D
versus 3D-LES.” River, Coastal and Estuarine Morphodynamics , RCEM 2011, Beijing, China , 1750-1762.

8
E-proceedings of the 36th IAHR World Congress
28 June – 3 July, 2015, The Hague, the Netherlands

Giri S. and Shimizu Y. 2006. Numerical computation of sand dune migration with free surface flow. Water Resour. Res.
42 (10), W10422: doi:10.1029/2005WR004588.
Fredsøe, J. 1974. On the development of dunes in erodible channels. J. Fluid Mech. 64: 1–16.
Itakura et al. 1986. Observation of bed topography during the 1981-flood in the Ishikari River. J. Hydrosci. Hydraul. Eng.
4 (2): 11 – 19.
Jerolmack, D. J. & Mohrig, D. 2005. A unified model for suba-queous bed form dynamics. Water Resour. Res. 41,
W12421: doi:10.1029/2005WR004329.
Nabi, M., de Vriend, H. J., Mosselman, E., Sloff, C. J., and Shimizu, Y. (2012). “Detailed simulation of morphodynamics:
1. Hydrodynamic model.” Water Resour. Res., 48, W12523, doi:10.1029/2012WR011911.
Nabi, M., de Vriend, H. J., Mosselman, E., Sloff, C. J., and Shimizu, Y. (2013a). “Detailed simulation of
morphodynamics: 2. Sediment pickup, transport, and deposition.” Water Resour. Res., 49, 4775–4791,
doi:10.1002/wrcr.20303.
Nabi, M., de Vriend, H. J., Mosselman, E., Sloff, C. J., and Shimizu, Y. (2013b). “Detailed simulation of
morphodynamics: 3. Ripples and dunes.” Water Resour. Res., 49, 5930–5943, doi:10.1002/wrcr.20457.Niemann, et
al. 2011. Sand Dunes in Steady Flow at Low Froude Numbers: Dune Height Evolution and Flow Re-sistance. J.
Hydraul. Eng, ASCE 137 (1): 5-14.
Onda, S., and Hosoda, T. (2004). “Numerical simulation on development process of dunes and flow resistance.” River
Flow 2004 (eds. M.Greco, A. Carravetta and R. Della Morte), Vol. 1, 245–252.
Paarlberg et al. 2009. Modeling river dune evolution using a parameterization of flow separation. J. of Geophys. Res.,
Pt. F: Earth surface 114 (F01014): ISSN 0148-0227.
Shimizu et al. 2009. Numerical simulation of dune–flatbed transition and stage-discharge relationship with hysteresis
effect. Water Resour. Res. 45, W04429: doi:10.1029/2008WR006830.
Sieben, A. 2006. Bed level changes during floods, a case study of the river Waal. RIZA werkdocument 2006.014,
Ministerie van Verkeer en Waterstaat, Directoriaal-Generaal Rijkswa-terstaat, RIZA: 1-8.
Tjerry, S., and Fredsøe, J. (2005). “Calculation of dune morphology.” J. Geophys. Res., 110, F04013.
Uchida, T., and Fukuoka, S. (2013). “Quasi 3D numerical simulation for flow and bed variation with various sand
waves.” Advances in River Sediment Research – Fukuoka et al. (eds) 2013 Taylor and Francis Group, London, ISBN
978-1-138-00062-9, 221-229.
Van Denderen, R. P. (2014). Bed level changes in theWaal during floods : An analysis of bed level measurements. Thesis for
fulfilment of Master of Engineering, TU Delft, The Netherlands.
Venditti, J. G., Church, M. and Bennett, S. J. (2005). Morphodynamics of small-scale superimposed sand waves over
migrating dune bed forms.” Water Resour. Res., 41,W10423, doi:10.1029/2004WR003461.
Wilbers, A.W.E. 2004. The development and hydraulic rough-ness of subaqueous dunes. PhD thesis, Utrecht University,
The Netherlands.
Yamaguchi et al. 2009. Significance of bed slope effect and non-equilibrium sediment transport on bed form evolution
process: A numerical analysis using detailed hydrodynam-ics. Proceedings of the 6th IAHR symposium on River,
Coastal and Estuarine Morphodynamics, Universidad Nacional Del Litoral, Santa Fe, Argentina, 21-25, Septem-ber,
2009: 659-666.
Yamaguchi, S. & Izumi, N. 2005. Weakly nonlinear analysis ofdunes by the use of a sediment transport formula
incorporating the pressure gradient. 4th IAHR Symposium on River, Coastal and Estuarine Morphodynamics, Int.
Assoc. for Hydraul. Res., Urbana, Ill.

You might also like