Zhang 2021
Zhang 2021
Zhang 2021
To cite this article: Qing Zhang, Gudrun Hillebrand, Thomas Hoffmann & Reinhard Hinkelmann
(2021): Assessing long-term evolution of the fine sediment budget in the Iffezheim reservoir:
temporal upscaling of numerical simulations, International Journal of River Basin Management,
DOI: 10.1080/15715124.2021.1885419
Article views: 51
RESEARCH PAPER
Figure 2. Suspended sediment concentration from ACDP measurement at the gauging station Hügelsheim, which is close to the upstream boundary of the
numerical model.
INTERNATIONAL JOURNAL OF RIVER BASIN MANAGEMENT 3
uniform grain sizes was carried out using 9 sediment frac- Hillebrand et al., 2012b). By analyzing water samples and
tions. Hydro- and morphodynamics were run in fully comparing them to measured back-scatter intensities, distri-
coupled mode, so the bathymetry in the riverbed was butions of SSC at observed cross sections can be derived.
updated after each time step continually within the simu- Cross-sectional measurements of Q and SSC are used to
lation time. Many sensitivity analyses of hydraulic and mor- define the upper boundary of the reservoir (compare section
phodynamic parameters were subsequently conducted and 3). Therefore, we evaluated the point measurement of the
the model was calibrated using field measurements (Hilleb- gauging station at Plittersdorf, which is located close to the
rand et al., 2016; Zhang et al., 2013; Zhang et al., 2015; right bank below the weir. There, SSC is about half the
Zhang et al., 2016a). However, using 24 cores of a 2.4 GHz cross-sectional SSC average at the upstream end of the reser-
reference computer, a computational time of 14.5 h was voir. Thus, this factor is set as the model inflow boundary
required to simulate a period of three months (section 4.1). condition for the numerical simulations.
Thus, long-term (e.g. several decades) simulation of mor- Furthermore, ADCP-based SSC measurements are used
phodynamics of the Iffezheim reservoir with this model to define a concentration profile of suspension at the
was not feasible. Up to now, many research studies have model boundary. Due to the left meander curve of the
also calculated sediment transport, but only for relatively Rhine at the model input, sediment tends to accumulate on
short-term periods, without using a fractional sediment the inner bank side of the curve, which can be observed
transport approach or with just a small number of transport well by ADCP measurements (Figure 2). Therefore, a func-
equations. For the long-term studies of 3D models, when tion with polynomial regression was developed to describe
using a fractional sediment transport approach, we still the concentration profile instead of applying Hunter-Rouse
face limitations regarding computational power. At this equation through mean concentration (Rouse, 1937).
point, upscaling of numerical models becomes necessary. The design capacity of the hydro-power plant was
Merkel (2017) developed the ‘Continuous Vertical Grain 1100 m³/s up to 2013, which corresponds roughly to mean
Sorting Model (CVSM)’ to performance the large scale and flow conditions. For the period when the discharge exceeds
long time simulations. The riverbed evolution and the result- this value the mean flow is about 1550 m³/s. During the
ing flow patterns are investigated with CVSM model instead period under investigation, the maximum value of a flood
of traditional concept of active layer (Hirano, 1971). Zhang discharge was 4170 m³/s, which was measured on 10/08/
et al. (2016b) applied a combination of the coarsening of 2007. The maximum concentration with 369 mg/l was
the grid resolution and the reduction of fractions to reduce measured on the same day.
the computational effort for the Iffezheim model. In this Water levels at the weir (i.e. at the downstream end of the
study, a reduced complexity approach was developed for model) are defined by a control curve, which defines the
long-term numerical simulation in the Upper Rhine. The water level depending on the inflow discharge (Klassen,
purpose of our upscaling approach was to determine the 2013).
long-term riverbed evolution in the Iffezheim reservoir
much faster while limiting the loss of accuracy compared
to the conventional instationary simulations using the 3D 2.2. Bathymetry and dredging data
high-resolution fractional sediment transport model. Due
to numerous impoundments along the river Rhine upstream The Iffezheim reservoir is located between Rhine km-309.1–
of Iffezheim, the bed load transport is not considered to be 334.0 (Figure 1) (Brudy-Zippelius & Schmidt, 2006). In order
relevant and therefore not investigated in this study. to obtain detailed information about the spatio-temporal
evolution of the riverbed, regular echo soundings were car-
ried out by the German Federal Waterways and Shipping
2. Field data and data analysis Administration (WSV) since 2000. Measurements of cross-
sectional soundings every 20 meters within the weir channel
2.1. Discharge and suspended sediment from Rhine km-332.0 to Rhine km-333.9 were conducted
concentration measurements approximately quarterly. High rates of net deposition were
The Plittersdorf gauging station is the closest station observed on the left-hand side of the weir channel, especially
(6.2 km) downstream of the Iffezheim reservoir (Figure 1), after maintenance dredging. Data on the timing and volume
where discharge and suspended sediment concentration of dredged material were supplied by the WSV (Table 1). No
(SSC) were measured daily by the WSV. Suspended sediment long-term net deposition took place on the right-hand side of
concentrations were based on bulk water samples taken once the weir channel (Hillebrand et al., 2012a).
every working day. The suspended sediment concentration Thirty-seven sets of echo sounding data, which were
was determined by filtering the sample and then weighing obtained between July 2000 and February 2011, were used
the solids (Hoffmann et al., 2020; Vollmer & Gölz, 2006). in this research (Figure 3). A section of the weir channel
The data of the gauging station in Plittersdorf used, amongst between Rhine km-332.0 and 333.8 was chosen to compare
others, to determine a relationship between discharge and measurements with the simulation results. This section
SSC, which is needed for the definition of boundary con- reveals the main sediment deposition in the weir channel.
ditions of the applied modelling concept (Hoffmann et al.,
2020) (Section 3.1). Table 1. Dredging periods, dredging quantities and corresponding short-term
In addition, ADCP measurements are used to analyze period.
spatial distributions of flow velocity and suspended sediment Dredging periods Dredging quantities (10³ m³) Time period
concentration for several cross-sections, e.g. at the Hügel- 07/2000 - 03/2001 355 1
sheim gauging station (Figure 2). Echo intensities of the 11/2003 - 03/2004 105 12
01/2005 - 09/2005 290 16,17
ADCP back-scatter signals correlate with SSC (Boldt, 2015;
4 Q. ZHANG ET AL.
Table 2. Data analysis of each short-term period: duration, mean, minimum, maximum inflow, standard deviation σ of inflow, total observed volume change and
average volume change per day along the Rhine between Rhine km-332.0 and 333.8. The periods in which the dredging took place are highlighted in bold. The
periods which are applied in the simulation are highlighted in italic. There is a lack of data in the period 17.
Time periods Duration Qmean Qmin Qmax σ Delta Vtotal Delta Vtot_daily
(approximate dates) (day) (m³/s) (m³/s) (m³/s) (m³/s) (10³ m3) (m³/d)
1 25/07/2000 - 16/12/2000 145 1276 890 2270 253 −146.3 −1009.0
2 17/12/2000 - 04/03/2001 79 1079 772 1950 260 −192.0 −2430.4
3 05/03/2001 - 27/06/2001 115 2155 1520 3260 420 108.6 944.3
4 28/06/2001 - 29/10/2001 124 1415 886 2300 343 10.5 84.7
5 30/10/2001 - 22/01/2002 86 1014 727 2520 338 43.1 501.2
6 23/01/2002 - 08/04/2002 76 1278 777 2370 403 21.6 284.2
7 09/04/2002 - 04/07/2002 87 1400 820 2630 363 26.2 301.1
8 05/07/2002 - 15/10/2002 103 1395 918 2630 351 60.0 582.5
9 16/10/2002 - 13/01/2003 91 1996 1220 3270 563 15.0 164.8
10 14/01/2003 - 28/02/2003 46 1155 863 1640 193 13.1 284.8
11 01/03/2003 - 04/11/2003 249 914 472 1530 226 1250 5020.1
12 05/11/2003 - 09/04/2004 158 951 514 3324 456 −84.4 −534.2
13 10/04/2004 - 22/07/2004 104 1282 901 2677 335 57.0 548.1
14 23/07/2004 - 20/10/2004 90 1006 636 1530 179 −16.1 −178.9
15 21/10/2004 - 11/01/2005 84 889 602 1796 258 24.5 291.7
16 12/01/2005 - 02/05/2005 111 1126 663 2088 361 −203.6 −1834.2
17 03/05/2005 - 14/09/2005 135 817 876 3144 811 −12.1 −89.6
18 15/09/2005 - 16/12/2005 93 716 460 1434 227 −16.0 −172.0
19 17/12/2005 - 06/03/2006 81 660 444 1640 249 27.0 333.3
20 07/03/2006 - 08/08/2006 155 1637 769 3230 576 84.4 544.5
21 09/08/2006 - 18/10/2006 71 1407 918 2600 367 52.0 732.4
22 19/10/2006 - 10/01/2007 85 852 589 1500 231 −1.62 −19.1
23 11/01/2007 - 03/04/2007 83 1235 734 2580 352 34.2 412.0
24 04/04/2007 - 17/07/2007 105 1352 771 2290 413 50.0 476.2
25 18/07/2007 - 27/08/2007 41 1822 1260 4170 565 12.6 307.3
26 28/08/2007 - 10/01/2008 137 1052 553 2390 387 18.8 137.2
27 11/01/2008 - 31/03/2008 81 993 640 1500 214 0.47 5.8
28 01/04/2008 - 24/09/2008 177 1456 999 2940 299 64.5 364.4
29 25/09/2008 - 05/11/2008 42 980 746 1480 204 −1.5 −35.7
30 06/11/2008 - 07/01/2009 64 927 678 1490 197 −7.35 −114.8
31 08/01/2009- 18/02/2009 42 845 607 1330 174 −0.4 −9.5
32 19/02/2009 - 04/05/2009 75 1321 833 1800 249 18.3 244.0
33 05/05/2009 - 30/11/2009 210 1089 467 2650 403 36.7 174.8
34 01/12/2009 - 24/01/2010 55 1205 826 2420 348 16.0 290.9
35 25/01/2010 - 20/07/2010 196 1026 687 2660 353 58.6 299.0
36 21/07/2010 - 01/02/2011 177 1489 772 2820 494 −0.7 −4.0
Sum Mean min max mean mean mean
3853 1200 444 4170 345 39 226
portions of clay and sand. Deposition is concentrated on the boundary conditions. Our approach consists, in general, of
left and middle section of the weir channel. Accordingly, the following four steps.
cohesive sediments are found in this area. On the right Generally, the boundary conditions – observed water dis-
hand of the weir channel and in the upstream part of the charges and observed sediment concentrations at the model
reservoir, coarser sediments are found. This is the location input – represent a very wide range of values. Thus, in a first
and grain size composition of the original river bed before step, the distribution of the range of the values during the
the dam was built. Initial conditions for the grain size distri- investigated period is analysed. They are subsequently subdi-
bution in the model are derived from measured bed grain vided into equidistant bins for the observed daily discharges
size distributions within the weir channel and in the Q and the observed daily sediment concentrations SSC.
upstream region of the reservoir. For discretization of sedi- These bins represent cells in a raster of boundary conditions.
ments, the continuous initial bed sediment sieve curve was The size of the bins depends on the range and the distri-
decomposed into 9 size fractions. Fractionation of suspended bution of the values. If the class size is too small, the calcu-
load and material on the riverbed are used in the numerical lations will take much more time and the preparation of
simulation (Figure 4). the simulation case will take as much expenditure as the
instationary simulation. If the class size is too large, it cannot
cover all the characteristic of the values.
In the second step, the average values of observed Q and
3. Methodology SSC are determined for each cell. Then, these values are used
as representative values for that cell [Qr, SSCr]. The number
3.1. Concept of the daily measurements for each cell is counted.
The basic idea of the upscaling approach is to replace nonsta- In the third step, stationary simulations are performed
tionary simulations of the riverbed volume change by a series applying the 3D high-resolution numerical model SSIIM
of stationary ones. Here, a daily time step was chosen for the using fractional sediment transport with 9 sediment fractions
nonstationary simulations, as also daily measurements were to simulate the daily volume change on the riverbed where the
available (see sec. 2.2). In principal other time steps can also calculated representative values [Qr, SSCr] for each cell are
be chosen. The flowchart in Figure 5 shows the concept of used as the stationary boundary conditions. The simulated
the upscaling approach based on the classification of daily volume changes and the number of daily measurements
6 Q. ZHANG ET AL.
Figure 4. Sediment summation curve of the grain size distribution of the channel bed and cumulative frequency distribution of the grain sizes in the weir channel
based on laser diffraction measurements.
for each cell in the investigated period are multiplied to obtain To compute the water flow, the SSIIM model solves the
the total volume change contributing from that cell. The total Reynolds-averaged Navier-Stokes equations with the k-ε tur-
volume change for a certain period is obtained by summing up bulence model on unstructured grids.
the simulated values of all the cells. The suspended sediment transport was calculated by sol-
In the fourth step, the simulated results are compared ving transient convection–diffusion equation for each grain
with the measurements. size fraction:
∂c ∂c ∂c ∂ ∂c
+ Uj +w = G +S (1)
3.2. Numerical modelling ∂t ∂xj ∂z ∂xj ∂xj
To investigate the long-term sediment transport processes
and predict the morphological changes in terms of depo- With subscript j = 1, 2, 3; where Uj is the Reynolds-averaged
sition and erosion in the Iffezheim reservoir, the 3D compu- water velocity, c is the sediment concentration and w is the
tational fluid dynamics program SSIIM (Sediment particle settling velocity; t is the time and xj is the coordinate
Simulation In Intakes with Multiblock options) has been direction. The turbulent diffusivity Γ was set equal to the
used (Olsen, 2014). The SSIIM program is suitable for repre- eddy viscosity of the k-ε model. For fractional sediment
senting complex natural geometries, whereby the topogra- transport simulation, grain-size classes representing the full
phy of the riverbed can be updated in a user-friendly range from clay to gravel (0.4–90,000 μm) are considered.
manner. This represents a particular advantage when large The sediment continuity equation is solved separately for
quantities of echo sounding data are used for modelling. each fraction.
As a boundary condition the equilibrium sediment con- parameters are used here. The time step of the hydrodynamic
centration ca is calculated for the cell closest to the channel simulations is set to 200 sec and the time step of the morpho-
bed. Using an empirical formula of van Rijn (2007): logical simulation is set to 600 sec. The thickness of the active
layer was calculated according to Hunziker (1995) with a
di · T1.5 value of 0.1 and 10 m for the inactive layer were used
ca,i = 0.015 f silt,i i
(2)
a · D0.3
∗,i (Zhang et al., 2015).
where the dimensionless particle diameter of the fraction i is
given by: 3.3. Approach and data preparation
(rs − rw ) · g 1/3 In this section, the application of the upscaling approach
D∗,i = [ ] · di (3)
rw · n2 based on field data in case of the Iffezheim reservoir is
described.
The sediment particle diameter is denoted as di . In
The approach is applied to 10 short-term and 2 long-term
Equation 2 the parameter a is the reference level set equal
periods. Short-term periods are chosen between two succes-
to half the height of the cell closest to the bed. fsilt,i is the frac-
sive echo sounding measurements with a focus on average
tion of the silt. Ti is the dimensionless bed shear stress par-
flow conditions. Short-term periods with strong flood events,
ameter. ρs and ρw represent the sediment and water
i.e. with high erosion and high deposition rates were
density, respectively. ν is the kinematic viscosity of water
excluded, as our approach is a too strong simplification for
and g is the gravitational acceleration.
extreme events considered for short-term periods. The
In the case of non-uniform sediment transport, inter-
short-term periods 3, 4, 5, 6, 7, 8, 22, 23, 24, 29 met these cri-
actions exist among different size classes of bed material.
teria (Figure 3).
Coarse particles have a higher chance of exposure to the
The longest possible time periods without any dredging
flow, while fine particles are more likely sheltered by coarse
event are chosen for further investigation. As opposed to
particles. To determine Ti for graded beds, the hiding and
short-term simulations, we included flood events – with high
exposure approach introduced by Egiazaroff (1965) for
simulated values for erosion – in long-term simulations to ana-
non-uniform sediment transport was taken into account.
lyse whether the approach is applicable for long-term forecast-
(t0 − ji (di /d50 )tc,d50 ) ing in the future. The first long-term period that was chosen
Ti = (4) (periods 3-8) started directly after dredging in March 2001
(di /d50 )tc,d50
and ended in October 2002. The second long-term period
where τ is the shear stress, τ0 is the effective bed shear stress, began after dredging in December 2005 and ended in February
τ0c,d50 is the critical bed shear stress. The exposure factor ξi is 2011, so it spans the short-term periods 19-36. During this
defined by period, a peak discharge of 4170 m³/s was measured in 2007.
⎡ ⎤2 In the investigated periods, short-term periods varied in
length from one to three months, whereas the long-term
⎢ log19 ⎥
ji = ⎢
⎣ ⎥ (5) periods ranged from two to five years.
di ⎦ Figure 6 gives an example of the application of the
log 19 ·
d50 approach (Section 3.1) for the longest time period (19-36).
The critical shear stress for cohesive beds (in this case Figure 6(a) shows the distribution of the discharge and con-
defined as containing a clay content of ≥ 8%) was calculated centration values. The sediment rating curve is shown for
by using the equation reported by Chien and Wan (1998): comparison and not applied in the modeling. The class
sizes for the discharge and SSC were chosen as ΔQ =
1 r k 500 m³/s and ΔSSC = 50 mg/l. An extreme value of a flood
tcr = 3.2(rs − rw )gd50 + ( b )10 (6)
77.5 rb,0 d50 event was detected far outside the specified classes, which
would lead to a large negative volume change during the cal-
where ρb and ρb,0 are the bulk density of the active layer and culation. The choice of the class sizes is important to ensure
of the consolidated bed, respectively, and k is a calibration that the computation time stays within practical limits. In
coefficient, which is here equal to 2.9×10−2 for the deposition Figure 6(b), all values are classified into cells, which are
dominant case (Hillebrand et al., 2012a). An approach based marked grey when entries occurred. A representative dis-
on a first-order kinetic equation according to Sanford (2008) charge Qr and a representative concentration SSCr were cal-
was implemented in the model, to simulate the effect of a culated for each cell, with which the stationary simulation
consolidation (Hillebrand & Olsen, 2011). was carried out to obtain the daily volume change. Our
The simulation of the morphodynamic processes consists approach covers much more combinations of Q and SSC
of flow field and sediment transport simulation. The numeri- compared to a sediment rating curve.
cal model of the Iffezheim reservoir has been extensively
tested and calibrated in preliminary studies (Zhang et al.,
2013; Zhang et al., 2015). In this study, the instationary 4 Results and dIscussion
boundary conditions for discharge and fractional suspended
4.1 Necessity of a 3D high-resolution model
sediment are replaced by daily stationary ones for each Q-
SSC bin as explained in section 3.1. In order to prepare stable The 3D high-resolution model was calibrated and applied in sev-
initial conditions, a calculation time of two days for water eral studies (Hillebrand & Olsen, 2011; Zhang et al., 2013; Zhang
flow and five days for sediment transport simulations is et al., 2015; Zhang et al., 2016a; Zhang et al., 2016b; Hoffmann
used. The simulations with the stationary boundary con- et al., 2017). Here we just want to point out the necessity of
ditions are running against quasi steady state. The following choosing a 3D model. Figure 7 shows a vertical velocity profiles
8 Q. ZHANG ET AL.
Figure 6. (a) Daily SSC and Q, derived from the gauging station at Plittersdorf, which are used as boundary conditions of the Iffezheim model from December 2005
to February 2011, according to the long-term period 19–36 (Table 2). An extreme value is shown separately. (b) Model raster: subdivision of SSC and Q ranges in
classes. Top in red: cell number. n: total number of days in this cell. Bracket left: Qr. Bracket right: SSCr. Subscript r stands for representative.
and velocity vectors along a cross section in turbine channel The simulation was carried out under unsteady conditions and
(Figure 1), which was simulated for a constant discharge of for the time period from 5th April to 17th July 2007, which is the
1580 m³/s and a constant suspended sediment load concen- investigated short-term period 24 in this study. The simulated
tration of 32 mg/l at the inflow boundary (based on field deposition and erosion patterns are in good agreement with
measurements of June 2010). We see downward velocities on the measurements. From a quantitative point, also a good agree-
the left, while they are upward in the middle and right part of ment was achieved for measured and modelled deposition
the cross section. and a reasonable agreement of model results volumes, while the agreement for the erosion was less good.
and field measurements was obtained (Figure 7). The model For further details, we refer to Zhang (2018).
could demonstrate its capability to simulate secondary flow.
Such a phenomenon cannot be simulated with a 2D or 1D
4.2 3d high-resolution model applied for the
model and has important impacts on erosion and deposition,
upscaling approach
specifically in the turbine channel. Figure 8 shows the compari-
son of bed change patterns in the Iffezheim reservoir between Figure 9 shows a comparison of the simulated volume
the simulated 3D high-resolution model and the measurement. changes with the observation for all applied short- and
INTERNATIONAL JOURNAL OF RIVER BASIN MANAGEMENT 9
Figure 7. Velocity profile and velocity vectors along a cross section in turbine channel, left: field measurements and right: 3D model results.
long-term periods described in section 3.3. The x-axis rep- separately during flood events. By summing up the number
resents the simulated daily total volume changes, where the of the days in the cell of the high discharge, the amount of
volume change of deposition and erosion is included, the erosion is overestimated. Because the model simulate
whereas the y-axis indicates the observed averaged daily the large erosion amount and cannot consider that the
total volume changes, which are listed in the last column supply effect exceed the erosion during higher discharges.
of Table 2. Secondly, Noack et al. (2015) and Hoffmann et al. (2017)
The results of the daily volume change for the short-term analysed uncertainties related to the erosion modelling of
period show large variations; Considering of the difference cohesive sediment. They highlighted that model results are
between the modelled and observed values in percentage, highly sensitive to the implemented calculation approach
they range from −110% (period 23) to 285% (period 4). and various model parameters, such as the distribution of
Best agreement was found with 1% for period 3. This grain size in the riverbed and cohesion of the sediment. Fur-
period has the largest average discharge and a relative long thermore, critical shear stress, which is sensitive to model the
period, which lead to the large deposition due to strong erosion, had a high spatial variability, which was difficult to
upstream supply. During the flood event critical discharges represent in the numerical model.
for the erosion of the channel bed material occurred, how- In comparison, the daily volume change of the two longer
ever, this compensates the deposition amount and results periods show smaller deviations compared to the measure-
in almost no net change of the riverbed. For the short- ments. They were 3% for period 3–8 and 20% for period
term period with lower discharger and without the flood 19-36. The simulation results of the long-term periods
event, which means that the consolidation and deposition show an acceptable agreement with the measurement and
occurred during the low flow phases, the simulated riverbed show less deviations from the measurements than shorter
volume change match well with the observation, such as in ones. This can be explained by the circumstance that long-
periods 22, 29. The model overestimates the volume change, term periods include some averaging behaviour regarding
when high discharge is dominant during the period and the extreme events. For the short-term (month) it would lead
period is sufficiently long, such as period 4. The deposition to deviation due to the uncertainty associated with the impli-
exceeds erosion over time. The model underestimates the cations to model the erosion. For the long-term period
volume change in most of periods. Periods, which are (years), the erosion due to occasionally occurring flood
characterised by average discharges show large simulated event corresponds to small absolute values that do not
erosion amounts, which are not observed, such as in periods impact the overall reservoir budget, which is controlled by
5, 23. In case of increased discharge, the simulated volume deposition.
change shows less deviation to the measurement, such as Finally, this simplified method showed clear savings in
in periods 6, 7, 8, 24. This is because overall deposition computational time, running 97% faster than the instation-
effect became more dominate. ary simulations. In this case only three hours of computation
The results of this approach shows good as well as poor time were needed for a five-year simulation period using the
agreement for short-term periods (months). The variable reference computer mentioned above (Section 1). In con-
agreement is mainly caused by the limitations of the upscal- trast, using the instationary sediment transport calculation
ing approach and the uncertainty of the numerical model would take approximately 85 h of computational time.
regarding the erosion of the channel bed. Firstly, due to Therefore, the developed approach is very suitable for a
the limited temporal resolution of the observed channel long-term simulation of riverbed evolution in the Iffezheim
bed changes the effects of erosion and deposition during a reservoir.
flood event cannot be separated. Observational data may
integrate large amounts of erosion and deposition and there-
5 Conclusion and Outlook
fore show net changes (i.e. the difference between deposition
and erosion during the period of two successive surveys). In In this research, an upscaling approach based on the classifi-
contrast, the amount of erosion and deposition is simulated cation of boundary conditions for discharges and suspended
10 Q. ZHANG ET AL.
Figure 8. (a). Bed level change derived between two subsequent echo soundings from 5th April to 18th July 2007. (b). Simulated bed change patterns in the
Iffezheim reservoir. Light areas present erosion and dark deposition (Zhang et al., 2015).
sediment concentrations was developed for long-term assess- period between July 2000 and January 2011 are the basis of
ment of sediment storage in the Iffezheim reservoir. The the study.
basic idea of the approach was to replace instationary simu- This approach was applied to 10 short-term periods,
lations for modelling the volume change of the riverbed by a between two successive echo sounding measurements, and
precalculated series of simulations with stationary boundary two long-term periods, which included several short-term
conditions. The numerical simulation was carried out using periods. Short-term periods vary between one to three
the 3D SSIIM model based on a high spatial-resolution months, whereas long-term periods vary between two and
model for the Iffezheim reservoir. The measured time series five years. The simulation results for the long-term periods
of discharge and suspended sediment concentrations at the show relatively small deviations from the measurements
gauge station Plittersdorf and the echo sounding bathymetry (3%, 20%), while for the short-term periods small as well
data in the weir channel of the Iffezheim reservoir for the as large deviations have been determined. This approach
Figure 9. Comparison of simulated and observed daily total volume changes for short and long-term periods.
INTERNATIONAL JOURNAL OF RIVER BASIN MANAGEMENT 11
works well with the long-term period during the years, but Hillebrand, G., Klassen, I., & Olsen, N. R. B. (2016). 3D CFD modelling
not for the short-term periods, in which the riverbed erosion of velocities and sediment transport in the Iffezheim hydropower
reservoir. Hydrology Research, 48(1), 147–159. https://doi.org/10.
affected intensively by occurred flow dynamics. 2166/nh.2016.197
Additionally, the upscaling approach exhibited clear sav- Hillebrand, G., Klassen, I., Olsen, N. R. B., & Vollmer, S. (2012a).
ings in computational time compared to the instationary Modelling fractionated sediment transport and deposition in the
simulations (of 97%). The results are promising, suggesting Iffezheim reservoir. 10th International Conference on
that the developed approach is very suitable for long-term Hydroinformatics, 14-18 July 2012 Hamburg, Germany.
Hillebrand, G., & Olsen, N. R. B. (2011). Towards modeling consolida-
simulations of riverbed evolution in the Iffezheim reservoir.
tion of fine sediments upstream of the Iffezheim barrage, Upper
We expect that this approach can also be applied to other Rhine River, Germany. Proceedings of the 7th IAHR Symposium
similar reservoirs. With further extensions this approach on River, Coastal and Estuarine Morphodynamics, 6–8 Sep 2011,
contributes to estimate the development of the fine sediment Peking, China.
budget in the Iffezheim reservoir until 2100 (Zhang, 2018). Hillebrand, G., Otto, W., & Vollmer, S. (2012b). Findings from ADCP-
Measured flow velocities and suspended sediment concentrations at
the Upper Rhine. 2nd IAHR Europe Congress, Munich.
Hillebrand, G., Pohlert, T., & Breitung, V. (2014). Klimaprojektionen
Acknowledgements für den Sedimenthaushalt und Risiken durch kohäsive Sedimente.
The study was carried out in the framework of the project ‘Assessing KLIWAS-48/2014. DOI:10.5675/Kliwas_48/2014_5.01.
long-term evolution of the fine sediment budget in the Upper Rhine’. Hillebrand, G., Seidenkranz, U., & Noack, M. (2015). Untersuchungen
The authors would like to thank the Federal Institute of Hydrology zur Remobilisierbarkeit feiner Sedimente in der Stauhaltung
(BfG), Germany for the financial support and the German Waterways Iffezheim am Oberrhein. 45. IWASA Internationales Wasserbau-
and Shipping Administration for providing the field data used in this Symposium Aachen,169: 1–11.
study. Hirano, M. (1971). River bed degradation with armouring. Proceedings
Japan Soc. of Civil. Engineers, 195.
Hoffmann, T., Hillebrand, G., & Noack, M. (2017). Uncertainty analy-
sis of settling, consolidation and resuspension of cohesive sedi-
Disclosure statement ments in the Upper Rhine. International Journal of River Basin
No potential conflict of interest was reported by the author(s). Management, 1–34(4), 401–411. https://doi.org/10.1080/15715124.
2017.1375509
Hoffmann, T. O., Baulig, Y., Fischer, H., & Blöthe, J. (2020). Scale
breaks of suspended sediment rating in large rivers in Germany
ORCID induced by organic matter. Earth Surface Dynamics, 8(3), 661–678.
https://doi.org/10.5194/esurf-8-661-2020
Thomas Hoffmann http://orcid.org/0000-0002-3853-1830
Huber, J. (2005). Die Umlagerung von Baggergut an der Staustufe
Iffezheim und deren Wirkungen, Karlsruher Vortragsreihe,
University of Karlsruhe.
References Hunziker, R. P. (1995). Franktionsweiser Geschiebetransport.
Mitteilung Nr. 138 der Versuchsanstalt für Wasserbau, Hydrologie
Astor, B., Gehres, N., & Hillebrand, G. (2014). Von der Quelle zur
und Glaziologie, ETH Zürich.
Mündung, eine Sedimentbilanz des Rheins: Korngrößenanalyse
Jia, D. D., Shao, X. J., Wang, H., & Zhou, G. (2010). Three-dimensional
von Schwebstoffproben am Rhein und seinen Nebenflüssen.
modelling of bank erosion and morphological changes in the
Technical Report, BfG-1798, Federal Institute of Hydrology, Koblenz.
Shishou bend of the middle Yangtze river. Advances in Water
Boldt, J. A. (2015). From mobile ADCP to high-resolution SSC: A cross-
Resources, 33(3), 348–360. https://doi.org/10.1016/j.advwatres.2010.
section calibration tool. U.S. Geological Survey, Kentucky Water
01.002
Science Center, Louisville, KY.
Jia, D. D., Shao, X. J., Zhang, X. N., & Ye, Y. T. (2013). Sedimentation
Brudy-Zippelius, T., & Schmidt, A. (2006). Sedimentations-problematik
patterns of fine-grained particles in the dam area of the three Gorges
im Oberwasser der Staustufe Iffezheim - Drei-dimensionale simu-
project: 3D numerical simulation. Journal of Hydraulic Engineering,
lation von Strömung und Schwebstofftransport. Dresdner
139(6), 669–674. https://doi.org/10.1061/(ASCE)HY.1943-7900.
Wasserbauliche Mitteilungen Heft, 32, 149–156. https://hdl.handle.
0000709
net/20.500.11970/103814
Klassen, I. (2013). Numerische Modellierung vom Strömungsverhalten,
Chien, N., & Wan, C. H. (1998). Mechanics of sediment transport.
Transportverhalten feinpartikulärer kohäsiver Feststoffe in
Reston, Virginia: American society of civil engineers.
Kombination mit partikel gebundenen und gelösten Schadstoffen
Egiazaroff, I. V. (1965). Calculation of non-uniform sediment concen-
für den Bereich der Staustufe Iffezheim mit SSIIM 3D. Technical
tration. Journal of the Hydraulics Division, 91(4), 225–247. https://
Report, Karlsruher Institut für Technologie, Karlsruhe.
doi.org/10.1061/JYCEAJ.0001277
Köthe, H., Vollmer, S., Breitung, V., Bergfeld, T., Schöll, F., Krebs, F., &
Faghihirad, S., Lin, B., & Falconer, R. A. (2017). 3D Layer-Integrated
Landwüst, C. V. (2004). Environmental aspects of the sediment
modelling of morphodynamic processes Near river regulated
transfer across the Iffezheim barrage, River Rhine, Germany. 17th
Structures. Water Resource Manage, 31(1), 443–460. https://doi.
world dredging congress and exposition, 27 September–1 October
org/10.1007/s11269-016-1537-x
2004, Hamburg.
Fischer-Antze, T., Olsen, N. R. B., & Gutknecht, D. (2008). Three-
Merkel, U. (2017). C-VSM-II: Large Scale and Long Time Simulations
dimensional CFD modeling of morphological bed changes in the
with Sisyphe s. Continuous Vertical Grain Sorting Model. In:
Danube river. Water Resources Research, 44(9). https://doi.org/10.
Dorfmann, Clements; Zenz, Gerald (Hg.): Proceedings of the
1029/2007WR006402
XXIVth TELEMAC-MASCARET User Conference, 17 to 20
Frings, R. M., Hillebrand, G., Gehres, N., Banhold, K., Schriever, S., &
October 2017, Graz University of Technology, Austria. Graz: Graz
Hoffmann, T. (2019). From source to mouth: Basin-scale morphody-
University of Technology. S. 131-138.
namics of the Rhine river. Earth-Science Reviews, 196, 102830.
Noack, M., Gerbersdorf, S., Hillebrand, G., & Wieprecht, S. (2015).
https://doi.org/10.1016/j.earscirev.2019.04.002
Combining field and Laboratory measurements to determine the
Haun, S., Kjærås, H., Løvfall, S., & Olsen, N. R. B. (2013). Three-dimen-
erosion Risk of cohesive sediments Best. Water, 7(9), 5061–5077.
sional measurements and numerical modelling of suspended sedi-
https://doi.org/10.3390/w7095061
ments in a hydropower reservoir. Journal of Hydrology, 479, 180–
Olsen, N. R. B. (2014). A three-dimensional numerical model for simu-
188. https://doi.org/10.1016/j.jhydrol.2012.11.060
lation of sediment movements in water intakes with multiblock
Hillebrand, G., & Frings, R. M. (2017). Von der Quelle zur Mündung:
option. Version 1 and 2.User’s manual. Department of Hydraulic
Die Sedimentbilanz des Rheins im Zeitraum 1991 - 2010.
and Environmental Engineering.The Norwegian University of
International commission for the hydrology of the Rhine basin,
Science and Technology, Trondheim, Norway.
Report No II-22 of the CHR.
12 Q. ZHANG ET AL.
Olsen, N. R. B., & Hillebrand, G. (2018). Long-time 3D CFD modeling Willis, D. H., & Krishnappan, B. G. (2004). Numerical modelling of
of sedimentation with dredging in a hydropower reservoir. Journal of cohesive sediment transport in rivers. Canadian Journal of Civil
Soils and Sediments, 18(9), 3031–3040. https://doi.org/10.1007/ Engineering, 31(5), 749–758. https://doi.org/10.1139/l04-043
s11368-018-1989-0 Wu, W. M., Rodi, W., & Wenka, T. (2000). 3D numerical modelling of
Pohlert, T., Hillebrand, G., & Breitung, V. (2011). Trends of persistent flow and sediment transport in open channels. Journal of Hydraulic
organic pollutants in the suspended matter of the river Rhine. Engineering, 126(1), 4–15. https://doi.org/10.1061/(ASCE)0733-9429
Hydrological Processes, 25(24), 3803–3817. https://doi.org/10.1002/ (2000)126:1(4)
hyp.8110 Zhang, Q. (2018). Conceptual simplifications for long-term sediment
Rouse, H. (1937). Modern conceptions of the mechanics of fluid turbu- transport simulations - application to Iffezheim reservoir, Germany.
lence. Transactions, ASCE, 102:1965. Doctoral dissertation. Chair of Water Resources Management and
Sanford, L. (2008). Modeling a dynamically varying mixed sediment Modeling of Hydrosystems, Technische Universität Berlin. Shaker
bed with erosion, deposition, bioturbation, consolidation, and Verlag. ISBN 978-3-8440-6335-6.
armoring. Computers & Geosciences, 34(10), 1263-1283. https://doi. Zhang, Q., Hillebrand, G., Hoffmann, T., Speckter, T., Moser, H., &
org/10.1016/j.cageo.2008.02.011 Hinkelmann, R. (2016a). Sensitivity of deposition and erosion to
van Oorschot, M., Kleinhans, M., Buijse, T., Geerling, G., & bed composition in the Iffezheim reservoir, Germany. 13th inter-
Middelkoop, H. (2018). Combined effects of climate change and national symposium on river sedimentation, 19-22 September 2016
dam construction on riverine ecosystems. Ecological Engineering, Stuttgart. Taylor& Francis Group.
120, 329–344. https://doi.org/10.1016/j.ecoleng.2018.05.037 Zhang, Q., Hillebrand, G., Moser, H., & Hinkelmann, R. (2013).
van Rijn, L. C. (2007). Unified view of sediment transport by currents and Sensitivity analysis of flow field simulation in the Iffezheim reservoir
waves. III: Graded beds. Journal of Hydraulic Engineering, 133(7), 761– in Germany with the 3D SSIIM model.35th IAHR World Congress.8-
775. https://doi.org/10.1061/(ASCE)0733-9429(2007)133:7(761) 13 September 2013 Chengdu, China. International Association for
Vente, J. d., Poesen, J., & Verstraeten, G. (2004). Evaluation of reservoir Hydro-Environment Engineering and Research.
sedimentation as a methodology for sediment yield assessment in the Zhang, Q., Hillebrand, G., Moser, H., & Hinkelmann, R. (2015).
Mediterranean: challenges and limitations. Soil conservation and Simulation of non-uniform sediment transport in a German
protection for Europe workshop, 13–15 April 2004 Cinque Terre, Reservoir with the SSIIM Model and sensitivity analysis.36th IAHR
Italy, 139–145. World Congress, 28 June-3 July 2015The Hague, The Netherland.
Vollmer, S., & Gölz, E. (2006). Sediment monitoring and sediment International Association for Hydro-Environment Engineering and
management in the Rhine River. Sediment Dynamics and the Research.
Hydromorphology of Fluvial Systems, July 2006 Dundee, UK. IAHS Zhang, Q., Hillebrand, G., Speckter, T., Moser, H., & Hinkelmann, R.
Publ. 306, 231–240. (2016b). Reduction of a high-resolution numerical flow and sedi-
White, S. (2005). Sediment yield prediction and modelling. Hydrological ment transport model. In River Flow 2016, St. Louis. Taylor &
Processes, 19(15), 3053–3057. https://doi.org/10.1002/hyp.6003 Francis Group.