Flow and sediment processes in a cutoff meander of the Danube Delta during
episodic flooding
Laura Jugaru Tiron a,b,
, Jérôme Le Coz c
, Mireille Provansal b
, Nicolae Panin a
, Guillaume Raccasi b
,
Guillaume Dramais c
, Philippe Dussouillez b
a
INCD GEOECOMAR, 23­25 Dimitrie Onciul Street, RO-024053, Bucharest, Romania
b
UMR 6635, CEREGE-CNRS, Europôle Méditerranéen de l'Arbois, 13545, BP 80, Aix-en-Provence, Cedex 04, France
c
Cemagref, Hydrology­Hydraulics Research Unit, 3bis quai Chauveau - CP 220, 69336 Lyon, Cedex 9, France
a b s t r a c ta r t i c l e i n f o
Article history:
Received 12 May 2008
Received in revised form 22 October 2008
Accepted 23 October 2008
Available online 5 November 2008
Keywords:
River morphodynamics
Sediment fluxes
Acoustic Doppler current profiler (aDcp)
Former meander
Danube Delta
St. George branch
This article analyzes the water and suspended solid fluxes through a straightened meander of the southern
branch of the Danube Delta (the St. George branch) during episodic flooding. The Mahmudia study site
corresponds to a vast natural meander which was cut off in 1984­1988 by an artificial canal opened to
shipping. The meander correction accelerated fluxes through the artificial canal and dramatically enhanced
deposition in the former meander. After his formation, the cutoff meander acted as sediment storage
locations, essentially removing channel and point bar sediments from the active sediment budget of the main
channel. Increases in slope and stream power in reaches upstream and downstream have also occurred, but
to a lesser degree. During the one-hundred-year recurrent flood in April 2006, bathymetry, flow velocity and
discharge data were acquired across several sections of both natural and artificial channels with an acoustic
Doppler current profiler (aDcp Workhorse Sentinel 600 kHz, Teledyne RDI) in order to investigate the
distribution of the flow and sediment and his impact on sedimentation in a channelized reach and its
adjacent cutoff. The contrasting hydro-sedimentary processes at work in both channels and bifurcation/
confluence nodal points are analyzed from the measured flux distribution, morphological profiles and
velocity and concentration patterns. In the cutoff, a diminishing of the intensity of the flow velocity (c. 50%)
and of the SSC was observed correlated with the aggradation of the river bed. In the bifurcation/confluence
nodal points and in the artificial canal were observed the most intensive hydrodynamic activity (high flow
velocity, SSC concentration, degradation of the river bad). Both the event-scale and long-term morphological
trends of the alluvial system are discussed analyzing the boundary shear stress and SSC variability. Excess
boundary shear stress in the sub-reaches directly affected by cutoffs resulted in scour that increased
downstream bed material load. These high sediment loads play a key role in driving morphological
adjustments towards equilibrium in the cutoff channel.
The approach followed in this paper combines detailed episodic in-situ aDcp measurements and robust
numerical 1D modeling in order to provide a practical comprehension of the relevant morphodynamical
processes. The 1D model reproduces robustly the continuity of hydrodynamical variables along the
streamwise axes of the two-channel network. The simulated are used in the paper for highlighting reachscale
morphological processes, at both event and long-term scales.
 2008 Elsevier B.V. All rights reserved.
1. Introduction
Large river deltas are complex alluvial systems of outstanding
social, ecological, and economical importance. Most of them are deeply
influenced by human activity: dam construction (Stanley and Warne
(1993), Stanley (1996) (Nile River); Panin et al. (1992) and Jugaru et al.
(2006) (Danube River)); Batalla et al. (2004) (Ebro River); meander
cutoff (Biedenharn et al. (2000), Kesel (2003), Harmar et al. (2005)
(Mississippi River), Popa (1997) (Danube River), Kiss et al. (2007) (Tisza
River)). Generally speaking, channel mobility is found to be reduced
effect of the damming, in association with a decrease in sediment
inputs (Hooke, 1986; Stanley and Warne, 1993 and Panin, 2003).
Moreover, river training operations (such as construction levees and
meander cutoff) initiated for navigational and flood defense purposes
often lead to dramatic changes in the streamwise profiles (Lane, 1947;
Keller, 1972; Hooke, 1986, Kesel, 2003; Kiss et al., 2007). Meander
correction affects both the hydraulic and morphodynamical behavior
Geomorphology 106 (2009) 186­197
 Corresponding author. INCD GEOECOMAR, 23­25 Dimitrie Onciul Street, RO-024053,
Bucharest, Romania. Tel./fax: +40 21 252 25 94.
E-mail addresses: jugarul@geoecomar.ro (L. Jugaru Tiron),
jerome.lecoz@cemagref.fr (J. Le Coz), provansal@cerege.fr (M. Provansal),
panin@geoecomar.ro (N. Panin), raccasi@cerege.fr (G. Raccasi),
guillaume.dramais@cemagref.fr (G. Dramais), dussouillez@cerege.fr (P. Dussouillez).
0169-555X/$ ­ see front matter  2008 Elsevier B.V. All rights reserved.
doi:10.1016/j.geomorph.2008.10.016
Contents lists available at ScienceDirect
Geomorphology
journal homepage: www.elsevier.com/locate/geomorph
of the modified branches that sedimentation occurs in time, while
newly built canals usually experience degradation (Ichim and Radoane,
1986).
Main climatic parameters like mean annual temperature and
annual precipitation play only a relatively minor role in determining
the processes of erosion and deposition in rivers but the intensity of
precipitation and its seasonal distribution are much more important
(Vandenberghe, 2003). Extreme events, at a recurrence interval of
years or tens of years leave the largest imprints in sediment
production.
This paper reports and analyzes new data on the hydrological and
sedimentary processes at work during a morphogenic flood in a large
modified meander of the St. George branch, the southern branch of
the Danube Delta. The 100-year recurrent flood that occurred in 2006
offered an exceptional opportunity for scanning different cross
sections of the Mahmudia River system by means of the emerging
Doppler profiler (aDcp) technology in order to analyze the impact on
sedimentation and dynamic processes in the study area. Such
experimental data, coupled with one-dimensional hydrodynamical
modeling, contribute to the understanding of morphological consequences
of the rectification.
In a first section, the work context is presented in regard to the
river system, the Mahmudia meander and the hydrological conditions
during the field observation campaign. The next section is dedicated
to experimental and numerical methodology, i.e., hydroacoustic
measurements, water sampling procedures, and the one-dimensional
hydrodynamical model set up for further analysis of the experimental
data. Then a third section gathers the results relative to the water and
sediment fluxes through the whole system, the streamwise variability
of morphological and hydrodynamical parameters, and local hydrosedimentary
processes. In the last section, morphodynamical trends
and sediment balances are discussed at the local, meander, and delta
scales.
2. Regional setting
2.1. Fluvial background and anthropogenic forcing
The Danube is a major river with a drainage basin of 817000 km2
(Ielenicz et al., 2006), a total length of 2860 km, and a total mean
annual discharge of 6500 m3
s-1
. The present suspended sediment
discharge in the Black Sea is about 25­30×106
t y-1
(Panin and Jipa,
2002). The Danube is the main contributor of fresh water and
sediment into the Black Sea, through the largest deltaic system in
Europe. The Danube Delta (5600 km2
) consists of three branches.
Unlike the northern branch, which is situated along the Ukrainian
border and has mostly remained natural, the other two branches have
been fitted for navigation to a large extent through meander
correction (Fig. 1A).
While the Danube sediment discharge decreased in the last
century mainly due to damming (Iron Gates I and Iron Gates II in
1970's), the water discharge increased (Panin, 2003). Also the 25­30%
sediment discharge reduction was not as extreme as in the case of the
Nile (98%, Sestini, 1992) or Ebro (95%, Jimenez and Sanchez-Arcilla,
1993). The increase in water discharge can be attributed to climatic
changes, and partly after damming to marsh reduction. Major rivers
draining the Carpathian and Balkan mountains are dammed, and
discharge is therefore insignificant (Giosan et al., 1999). The relatively
small reduction in the sediment discharge after damming, was due to
increased erosion of the river bottom and islets in the lower course of
the river (Mihailescu, 1983), and to meander cut-offs in 1990's, along
the distributaries in delta area (Panin, 2003).
The most southern branch, the St. George branch, starts out from a
bifurcation 108 km upstream of the outlet to the Black Sea. It carries
23% of the Danube water discharge and 21% of the Danube sediment
discharge (Bondar and Panin, 2000). The width of the distributary
varies from 150 to 550 m, and the water depth varies between 3 and
Fig. 1. (A) General map of the Danube Delta. (B) Mahmudia former meander with the aDcp profiles position.
187L. Jugaru Tiron et al. / Geomorphology 106 (2009) 186­197
27 m under the local low water mark (Bondar and Panin, 2000). The
St. George channel is highly sinuous and still morphologically
dynamic, as attested by migrating meanders (Jugaru et al., 2006)
and by the fast evolution of its mouth: the progradation of a deltaic
lobe and the displacements of the barrier-like Sakhalin Island during
the twentieth century (Panin, 2003).
In the last decades, a number of changes in the St. George
branch behavior in response to anthropogenic forcing have been
observed. Following the building of the Iron Gates I dam and the
hydrotechnical regulation works along the Danube tributaries, the
suspended sediment discharge decreased by c. 10% at Isaccea
station (KM 54, upstream of the delta head). The construction of
the Iron Gates II dam, at Ostrovul Mare (KM 846), induced a
catastrophic decrease in the suspended sediment discharge: the
sediment load measured at Isaccea dropped by an additional 50%
(Panin and Jipa, 2002). During the years 1984­1988, six free
meanders were modified in order to improve navigation. The total
length of the St. George branch was diminished to 32 km from the
previous 108 km. The man-made intersecting channels were
initially 7­8 m deep and 75­100 m wide (Popa, 1997). Such river
training works caused a change in the distribution of the solid as
well as the streamflow in the whole delta, enhancing fluxes
through the St. George branch into the sea.
2.2. The Mahmudia experimental site
The former Mahmudia free meander is located in the middle part
of the St. George branch, between KM 84 and KM 64 (Fig. 1A). This
meander results from the river impingement against the Mahmudia
hills that deflected the St. George channel into a large meander loop
named the Mahmudia-Uzlina meander bend (Panin, 2003). This is the
largest and the most complex meander in the St. George branch. Its
wavelength is 5.1 km, the radius of curvature is 3.93 km and the
degree of sinuosity is 4.1.
The cutoff of the natural channel by a navigational canal since
1984­1988 (Fig. 1B) caused dramatic changes in the distribution of
river flow velocities, discharge, and sediment fluxes (Ichim and
Radoane, 1986; Popa, 1997; Panin, 2003). As a consequence, a number
of morphological responses occurred during the period 1990­2000
when the former meander experienced aggradation up to 11 m
(i.e. 1.8 m y-1
) (Popa, 1997) and decrease in width (an average of
-2.7 m y-1
, with a maximum of -7.5 m y-1
(about 33%) at KM 81).
Sediment infilling is manifested by the issue of the island located at
KM 74. The island developed from the left bank and blocked the
entrance of the Gârla Filatului canal; another island appeared at KM 65
in 1990 and increased in area twofold during 1990­2000 (Jugaru et al.,
2006).
2.3. Hydraulic conditions during the field observations (100-year
recurrent flood of 2006)
Along the Lower Danube section, the flooding events occur
when the water discharge are over 10,000 m3
s- 1
. A statistical
analysis of the data set for 160 years (1840­2000), concerning the
mean annual water discharge of the Danube River, show that, at the
delta apex, 89 flooding events were recorded, of which one flooding
event with the largest ever recorded mean annual water discharge
of 16,000 m3
s- 1
in 1897 (also the largest daily water discharge of
20,940 m3
s- 1
(Bondar, 2004).
The period during which the measurements were made (28­30
April, 2006) corresponds to a 100-year recurrent flood that started at
the beginning of April 2006 during the high spring waters and lasted
for a period of several months. During the measurements, the flow
rate of the Danube was 16,500 m3
s-1
at its entrance into the delta at
Ceatal Izmail (Fig. 1A) and 4600 m3
s-1
at the bifurcation of the St.
George branch at Ceatal St. George (Fig. 1A). During the three days
when measurements took place, the water flow discharge remained
almost constant (Fig. 2).
3. Materials and methods
3.1. Hydroacoustic measurements (aDcp)
The application of aDcp (acoustic Doppler current profilers) to river
discharge measurements has recently gained considerable popularity
because of their efficiency and reliability (for a review see, e.g., Yorke
and Oberg, 2002). Based on the signal processing of the echoes of
ultrasonic pulses, the deployment of an aDcp mounted on a boat
yields the river channel bathymetry and the tree-dimensional velocity
field from which the total discharge can be computed. The frequency
of ultrasonic pulses operated by aDcp directly affects the measuring
range (increasing with lower frequency) and the uncertainty
associated with flow velocity measurements (also increasing with
lower frequency). In this study case, with flow depth up to 20 m, a
600 kHz aDcp was a good practical compromise in order to avoid
range limitation while obtaining acceptable flow and SSC uncertainty
with accurate enough spatial resolution. The data analyzed in this
study were acquired with an aDcp Workhorse Sentinel 600 kHz
(Teledyne RDI) mounted on a powered boat. During the field
campaign, transverse aDcp profiles were systematically completed
at relevant sections of the Mahmudia River system (Fig. 1B): around
the upstream bifurcation (sector A), around the downstream confluence
(sector E), along the cutoff meander (sectors B­C­D), and
along the straight outlet reach downstream of the confluence (sector
F). In addition to the bottom-track reference used for aDcp velocity and
path measurements, the data were continuously georeferenced by
a vessel-mounted DGPS (ensemble pivot with mobile Trimble
monofrequency ProXR). The bin size was set to 50 cm, and the blank
distance (i.e., the unmeasured top layer height) was 80 cm.
3.2. Suspended sediment concentrations
For each bin and each beam, the aDcp records the acoustic
backscatter intensity as the acoustic energy (in dB) returned from
scattering particles. This intensity is an accepted indicator of
suspended-load concentration (see, e.g., Reichel and Nachtnebel,
1994; Gartner, 2004; Dinehart and Burau, 2005; Kostaschuk et al.,
2005), although it depends on the properties of the water (temperature,
absorption), of the acoustic device itself (frequency, measuring
mode, hardware), as well as of the suspended particles (granulometry,
concentration, mechanical properties). Consequently, calibration
methods for converting backscatter intensity into suspended sediment
concentration (SSC) require extensive sediment sampling
operations (Rotaru et al., 2006). Here, the state-of-the-art, pragmatic
SediView methodology (Land and Jones, 2001) was followed: a
Fig. 2. Danube daily discharges at the bifurcation of the St. George branch (Ceatal St.
George gauging station).
188 L. Jugaru Tiron et al. / Geomorphology 106 (2009) 186­197
manual and iterative calibration of hydroacoustic parameters against
SSC samples was achieved using the SediView software dedicated to
the processing of Teledyne RDI aDcp data. Once calibrated, hydroacoustic
measurements potentially offer the convenient, nonintrusive,
synoptical surveying of bathymetry, water velocity, and SSC, i.e., all
required information to assess suspended sediment fluxes across a
river section.
Consequently, stationary aDcp in situ measurements were coupled
with water sampling with a 5 L horizontal Niskin-type bottle. In
several selected cross sections, coupled aDcp and sample data were
acquired in three separate verticals at depths of 3, 6, 9, and 12 m. The
sampled SSC (expressed in mg l-1
) was measured in the laboratory by
means of a HACH DR/2000 digital spectrophotometer (using the 630
method). Thus samples were used to calibrate aDcp concentration
profiles. The resulting calibration data cloud (Fig. 3) shows that an
acceptable agreement between backscatter data and SSC was reached,
even if deviations up to 30% appear for two samples. Because of
technical restrictions, the number of samples is not high enough to
fully understand the reasons for such residuals. However, as errors do
not show any clear trend with the sampling depth nor the SSC value,
this calibration operation was assumed to be reliable enough for
further assessment of the SSC distribution across sections and SSC flux
estimates.
3.3. One-dimensional hydrodynamical model (MAGE)
A one-dimensional hydrodynamical model was set up using the
MAGE code (Cemagref). MAGE solves the unsteady one-dimensional
shallow water equations in looped channel networks, i.e. including
diffluences and confluences. Computational cross-sections every
100 m were built from bathymetry profiles extracted from the aDcp
transects. Assuming constant water slope, the aDcp bathymetry
profiles were vertically shifted to fit the 1.25 m water fall that
occurred between sections A1 and F2. This water fall was computed
from observed flood levels and topography data extracted from
topographical maps of the Danube Delta (Topographical Military
Direction, 1980) (scale 1:25000). As no data in the floodplain were
available, only the main channel flow was simulated. The downstream
and upstream boundary conditions are the observed water level and a
constant discharge of 3000 m3
s-1
, respectively. A homogeneous
Manning coefficient of 0.06 leads to a simulated water fall of 1.20 m
between A1 and F2. This quite high flow resistance value likely
involves the effect of sinuosity and overbank flow processes on energy
losses.
4. Results
4.1. Overview of water and suspended sediment fluxes
By comparing the total suspension fluxes through each section
(Fig. 4), we can drawa first balance of the ways of transport and of their
relationship to the hydrodynamics and morphology of the channel.
Most of the sedimentary flux (73%) passes through the artificial
canal (between A3 and E3). In the former meander, the sediment
transit is weak and progressively decreases with the distance (Fig. 4).
The high SSC values reappear downstream the confluence (E1­F2).
This heterogeneity and its spatial variations enable us to design a box
model of the transport of sediments for the investigated sector.
At the nodal point of bifurcation, the water as well as suspended
sediment flux balances are conservative. The water discharge A2+
A3 (510+2436=2946 m3
s-1
) exceeds the discharge through A1
(2905 m3
s-1
), while the loss of sediments between cross sections
A2+A3 (73+246=319 kg s-1
) and section A1 (338 kg s-1
) is
19 kg s-1
, i.e., a relative variation lower than 6% (not significant in
comparison with measurement uncertainties).
The water discharges and the sediment fluxes progressively decrease
along the cutoff meander from 510 m3
s-1
(A2) to 406 m3
s-1
(E2) and
from 73 kg s-1
(A2) to 8 kg s-1
(E2), respectively, indicating an obvious
siltation in the main channel and in the floodplain as well. Sediment flux
"losses" are significant between sections B/C1 (70/27 kg s-1
) and D/E2
(22/8 kg s-1
) only. All in all, the flowand sediment flux issuing from the
meander lost 104 m3
s-1
and 65 kg s-1
in comparison to the flow
entering the meander. These gross values suggest a complex balance
between local supplies and losses toward the most important canal in
the left bank, the Uzlina canal (the mean values of the water (m3
s-1
)
and sediments (kg s-1
) discharges entering the Uzlina lake through the
canal are 16.4 m3
s-1
(or 1.5%) and 3.0 kg s-1
(or 1%), values for the
period 1981­1989).
In the artificial canal, the water discharge balance is rather
conservative (+ 9%), whereas the output sediment flux increases by
90 kg s-1
(+ 36.5%) between sections A3 and E3. Hence, the canal
potentially is an area with significant erosion.
The confluence is a complex sector because profile E1 exhibits a
sedimentary flow (422 kg s-1
) larger than the total sum of the
potential supplies (E2+E3=344 kg s-1
), while the water discharge
Fig. 3. Results of the hydroacoustic calibration of aDcp backscatter data against SSC
sampling.
Fig. 4. A box model for SS fluxes, streamflow, and SSC by section through the Mahmudia river network.
189L. Jugaru Tiron et al. / Geomorphology 106 (2009) 186­197
balance is almost conservative. The flow decrease between sections
E1, F1 and F2 (3241 m3
s-1
to 2514 m3
s-1
and 2476 m3
s-1
) is due
to an export of 700 m3
s-1
toward the Murighiol canal in the
right bank. The decrease in the sedimentary transit (422 kg s-1
to
236 kg s-1
) between profiles E1 and F2 can be explained by a
progressive decrease in energy but also by the overflowing that causes
a progressive decrease in discharge and concentration, from the
sedimentation in the floodplain.
4.2. Morphological profiles
Raw bathymetry data extracted from the aDcp measurements are
not georeferenced, and the elevations are relative to the local freesurface
level. For further morphological analysis, the cross-sectional
profiles were georeferenced by DGPS tracking; the free-surface stage
was estimated according to the 1.25 m water fall observed between
sections A1 and F2. Hence, the margin of error in bed elevation
estimates typically is 2­3%. Because of unmeasured areas on the sides
of channels, the total cross-sectional width typically is underestimated
by 2­5%.
Regarding the natural former meander, the mean talweg slope
(A2­E2) is very low (0.089 m km- 1
). The upstream reach (A1­A2,
downstream of the bifurcation) corresponds to an important
reduction of the depth (Fig. 5). The talweg line remains located on
the right bank (which was concave before the cutoff) and varies
from -16 m (A1) to -9 m (A2) in depth (i.e., a counter-slope of
-8.75 m km- 1
); in the left bank between B and C1 (the convex
bank of the sinuosity), another counter-sloping area attests to an
accumulation of sediments. Downstream, the talweg depth
increases locally in the concave bank of the apex (-15.5 m on B,
-10.6 m on C2). Close to the confluence upstream, profile E2 shows
a deepening toward the right bank (-13.2 m). It could be attributed
either to the original concave pool of the bend or to the water swirls
coming out from the canal; between E2 and E1 a steep talweg slope
(22.9 m km- 1
) is observed.
Theentranceinthe artificialcanal (A1­A3) corresponds to anincrease
in depth (from -16 to -0.8 m, or a local talweg slope of 7.4 m km-1
).
Between A3 and E3, the slope of the canal decreases (0.3 m km-1
).
However, the confluence (E3­E1) is characterized by a counter-slope of
-8.5 m km-1
, which may correspond to a sedimentary deposition
caused by the junction of both flows.
In the natural channel downstream of the confluence, the talweg
goes from -18.8 m (F1) to -11.3 m (F2) in depth, with a counterslope
of -8 m km-1
, which likely corresponds to deposition between
the two sections.
The observed morphology clearly records the impact of the
artificial canal on the whole alluvial system. The natural channel
adapts to the flow decrease, but its response varies streamwise
because of sediment infilling. On the contrary, the artificial canal is
characterized by an increase in depth that creates a well-calibrated
shape suitable for maintaining high energy. The intersections of the
cutoff channel with the artificial canal at the bifurcation and at the
confluence are accumulation sectors with counter-slopes, either in
relation to flow/velocity decrease (A2) or to the junction of
contrasting flows (E1).
4.3. Flow and sediment processes
The analysis of the aDcp velocity data provides evidence as to the
highly heterogeneous hydrodynamical conditions in the different
reaches of the Mahmudia alluvial system (Table 1).
Upstream of the bifurcation, the flow (2905 m3
s-1
) is homogeneously
distributed across the approach section A1 (Fig. 6A). The
average velocity is 0.9 m s.-1
. Then the bifurcation splits the flow
into two hydrodynamically contrasted branches: the main flow
(2436 m3
s-1
or 82%) is captured by the artificial canal (A3) where
velocity increases (an average of 1.4 m s-1
).
The former meander (A2) receives only 18% of the upstream
flow (510 m3
s-1
), with slightly reduced velocities (an average of
0.8 m s-1
). Farther downstream, the former meander is characterized
Fig. 5. The talweg and water stage profiles along the natural channel (A1­F2) and the artificial canal (A3­E3).
Table 1
Hydrodynamical and sedimentary parameters in investigated cross-sections.
Section A B C D E F
A1 A2 A3 B C1 C2 D E1 E2 E3 F1 F2
Width (m) 362.0 99.0 100.0 90.0 408.0 299.0 320.0 183.0 347.0 145.0 207.0 260.0
Maximum depth (m) 16.3 9.9 20.8 15.5 4.1 10.6 7.1 16.9 13.2 22.2 18.8 11.3
Maximum velocity (m s- 1
) 1.8 1.3 2.2 1.2 0.7 0.7 0.7 2.5 0.5 2.3 2 1.8
Mean velocity (m s-1
) 0.9 0.8 1.4 0.7 0.3 0.3 0.2 1.2 0.1 1.2 0.9 1.1
Area (m2
) 3312.0 650.0 1790.0 938.0 1438.0 1481.0 1747.0 2708.0 2626.0 2143.0 2673.0 2276.0
Liquid discharge (m3
s-1
) 2905.0 510.0 2436.0 642.0 497.0 544.0 388.0 3241.0 406.0 2655.0 2514.0 2495.0
Flux (kg s-1
) 338.0 73.0 246.0 70.0 27.0 27.0 22.0 422.0 8.0 336.0 268.0 237.0
Mean SSC (mg l-1
) 120.0 140.0 100.0 110.0 50.0 50.0 60.0 130.0 20.0 130.0 110.0 90.0
190 L. Jugaru Tiron et al. / Geomorphology 106 (2009) 186­197
by a progressive decrease in average velocity (0.7 to 0.1 m s-1
from B to
E2).
The flow rate also decreases (from 642 to 406 m3
s-1
from B to E2)
because of a waste of water toward small channels starting from the
left bank. Around the first apex (B), the channel dissymmetry
produces highest velocities near the concave bank of the meander.
In C1 and C2 sections, the velocities are homogeneously distributed
again, as the channel is much less sinuous.
At the confluence (Fig. 6B), the E2 profile is still characterized by
low and homogeneous velocities (average 0.1 m s-1
) and by a low
discharge (406 m3
s-1
). It sharply differs from the upstream profile of
the artificial canal (E3) where a high discharge arrives (2655 m3
s-1
)
with high velocity (average 1.2 m s-1
); E3 is an asymmetrical profile
(vertical fast curls close to the right bank between 1.8 and 2.4 m s-1
,
lower velocities toward the left bank and close to the bottom, ranging
from 0.8 to 1.5 m s-1
). Downstream of the confluence (E1), the flow is
concentrated in the central part of the channel. This velocity
asymmetry stems from the sharp contrast between input flows from
both upstream channels. The resulting total discharge is 3241 m3
s-1
.
Downstream of the confluence, the influence of the fast waters
coming out from the artificial canal is rapidly diminished. The velocities
progressively increase from 0.9 m s-1
in F1 to 1.1 m s-1
in F2.
However, a nucleus of higher velocity (between 1.1 and 1.4 m s-1
)
persists on the F1 right bank, determined by the water supply of the
artificial canal, while the velocity distribution remains uniform for F2
(Fig. 8).
The aDcp discharge and velocity measurements allow the evaluation
of the one-dimensional numerical simulationperformance (MAGE). The
simulated distribution of the total discharge (3000 m3
s-1
) between
the artificial cutoff canal (2520 m3
s-1
) and the former meander
(480 m3
s-1
) is in acceptable agreement with the discharges measured
by aDcp (Table 1). Simulated mean velocities also closely follow the
measured trends (Fig. 7).
However, results in the upstream part of the former meander
suggest that some extra investigation of sections would have been
necessary there in order to describe the channel more accurately:
measurements are lacking to decide whether the simulated velocity
profile is correct or some hydraulic controls are missing. In the last
two sections (F1 and F2), downstream simulated velocities are biased
high because of significant overbank flow: flow rate in the main
channel is 3000 m3
s-1
in the model, whereas about 500 m3
s-1
are
missing in the aDcp discharges. In spite of these limitations, the
Fig. 6. aDcp depth-averaged velocity and river-bank contours within sectors of (A) bifurcation and (B) confluence. Channel contours were digitized from a satellite image dating from
June 2000 for different hydrological conditions than during the measurement period.
Fig. 7. Cross section averaged velocity: aDcp measurements and simulation results.
191L. Jugaru Tiron et al. / Geomorphology 106 (2009) 186­197
simulation results are reliable enough for further interpretation of the
hydrodynamical behavior of the system.
4.4. Local hydrosedimentary processes
The distribution of local velocities and SS concentrations varies
with every investigated section, depending on the local morphology.
In particular, the sections situated near the bifurcation (A1/A3) in
the apex of the former meander (B) and those close to the confluence
(E1, E3, and F1) are characterized by high local velocities (from 1.2 to
2.5 m s- 1
) and by the presence of gravitationally related sedimentary
flows associated with SSC plumes (Figs. 8 and 9). The
concentrations range between minimal values of 10­20 mg l- 1
and maximal values exceeding 180 mg l- 1
.
For the symmetrical sections A3, E3, E1, and F1, the concentrations
are classically higher near the bed and the banks and lower near the
free-surface and the central part of the channel. Section E1 situated
upstream of the confluence records flows with very contrasting
concentrations.
Going into the details of three-dimensional velocity and concentration
data, some vertical plumes with high SSC suggest that
gravitationally related sedimentary flows are eroding sediments
from the bottom layer. Such hydrosedimentary patterns particularly
appear near the less steep slopes of convex banks (sections A1, B, and
F1) and also in deep and calibrated sections (E3 and E1). They may be
related to the development of the bedforms. Popa et al., (unpublished
report) and Panin et al. (1992) reported the existence of dunes with
heights ranging from 0.5 to 2 m and a wavelength of 20­40 m; they
can be entirely mobilized during extreme floods like the 2006 great
flood. A longitudinal aDcp profile of 2.4 km was acquired during the
2006 campaign along the former meander between KM 83.5 and KM
81.5 (Fig.10). These bathymetry data show the structure of high dunes
ranging from 0.5 to 2 m and from 4 to 13 m in waveheight,
respectively. It also illustrates the fact that between two aDcp cross
sections some hydraulic control sections can be missing for the onedimensional
numerical model.
The following discussion will focus on potential coherence
between our results and the other available data concerning the
geomorphological evolution of the studied area. The deposition/
erosion reaches suggested by boundary shear stress and SSC
variability as well as the morphological evolution will be analyzed.
A balance of the total SSC flux through the St. George branch is
drawn.
5. Discussion
5.1. Event-scale morphodynamical analysis
Local morphodynamical processes during the 2006 flood are
analyzed through two geomorphological parameters: the specific
stream power indicating the stream ability to adjust its channel
morphology (Biedenharn et al., 2000); and the boundary shear stress
Fig. 8. Distribution of local velocity magnitude within cross-sections of the Mahmudia river network (A1/A3/B/E1/E3/F1).
192 L. Jugaru Tiron et al. / Geomorphology 106 (2009) 186­197
Fig. 10. Longitudinal talweg profile in the former meander between KM 83.5 and KM 81.5.
Fig. 9. Distribution of local SSC within cross sections of the Mahmudia River network (A1/A3/B/E1/E3/F1).
193L. Jugaru Tiron et al. / Geomorphology 106 (2009) 186­197
indicating the flow capacity to mobilize sediment from the bed as
suspended-load or bedload.
Morphodynamical activity is generally assessed by the specific
stream power () as an indicator for the river transport capacity
(Bagnold, 1966):
 = X= B W m
-2
h i
1
where B is the bankful channel width and  is the stream power that
writes:
X = gQS W m
-1
h i
2
where the representative discharge Q [m3
s-1
] is usually taken as the
bankful discharge Qbf. In the present analysis, the discharge gauged by
aDcp for each channel is taken as the representative discharge. As for
bed shear stress estimates (see below), the energy slope S is
approximated by the observed mean water slope. The specific stream
power is actually a complex index, especially when applied to
meander systems, where the water slope is influenced by the
sinuosity and variations of the width of the channel (Schmitt et al.,
2001).
Sectors with high specific stream power (above 29 W m-2
) are to
be found in the artificial canal (A3, E3, Table 2) because of low width
and steep slope. In the former meander, stream power values are
logically much lower because of the lower slope and longer length
(A2/E2: below 4 W m-2
). In the upstream (A1) and downstream (E1/
F2) single channels,  values are quite high (from 14.1 W m-2
in A1 to
27.2 W m-2
in E1 and around 20 W m-2
in F1/F2).
The mean bed shear stress (0) corresponds to the unit tractive
force exerted on the bed:
0 = gRS N m
-2
h i
3
where  is the fluid density (1000 kg m-3
for sediment-free water), g
is the gravitational acceleration (9.81 m s-2
), R is the hydraulic radius
[m], S is the water energy slope [m m-1
] (i.e., S=0.056 m km-1
in the
former meander (A2­E2) and S=0.174 m km-1
in the main channel
(A1/F2)).
Experimental estimates of 0 are generally in good agreement
with the MAGE simulation results (Fig. 11). The numerical results
offer a more continuous view of the variability in the erosion
capacity along both channels. However, the dramatic increase of
simulated bed shear stress values in the very downstream part of the
reach (F2) seems unrealistic because of the ignored overbank flow
effects.
Compared with the value in the approach section A1, the shear
stress is significantly higher in the artificial canal (A3/E3) and in the
downstream channel (E1/F1). In the former meander, shear stress
values are dramatically lower (A2/E2). The apex at section B
corresponds to the most active reach in the former meander. Thus,
simulated bed shear stress values indicate high erosion capacity in the
artificial canal and lower and decreasing erosion capacity in the
former meander (deposition reaches).
This analysis of potential erosion/deposition reaches is coherent
with the values of mean cross sections averaged SSC (Fig.12) as well as
the SS flux analysis supported by the box model (Fig. 4).
The decrease in mean SSC up to section C1 (50 mg l-1
) confirms
that sediment deposition mainly occurs in the first half of the former
meander. In the artificial canal, the increase in SSC between A3 and E3
(from 100 to 130 mg l-1
) confirms the existence of erosion processes
already evidenced as upwelling sediment-laden plumes originating
from the bed. These vertical sediment plumes are associated with high
values of 0 for sections A1, B, E1, E3, and F1, which allows for the
resuspension of the bed material.
5.2. Long-term morphodynamical analysis
The measurements performed in April 2006 represent an
opportunity for relating the multiannual morphological balances to
the sedimentary flows occurring during a morphogenic flood event.
Slope adjustments are often the dominant morphological response
when alluvial channels are cut off that facilitated the accretion of the
former meanders (Biedenharn et al., 2000). Shields et al. (1984)
explained that the accretion of the alluvial plug (i.e. the terrestrial unit
formed at the mouth of the former channel) are controlled by the
geometry of the former channel, particularly by the ratio between the
former channel length and the new channel length, and the upstream
angle between them. Most authors suggest that the older the cut-off
channel, and the closer it is to the main channel, the thicker the
overbank deposits will be (Piegay et al., 2000).
Lane (1947) distinguished two phases: 1. immediate response
occurring a short time after the cutoff is implemented; 2. subsequent
response taking place gradually and over a very long period.
For the Mahmudia cutoff meander, the first phase was monitored
by Popa (1997) who carried out a sequence of bathymetric measurements
during a period of six years immediately after the building of
the artificial canal. The measurements were made with a portable
computerized system (Biotron S.A Company, Bucharest) that performed
simultaneous measurements of the flow velocity and depth.
The velocity was measured with an electromagnetic currentmeter,
while depth was measured by a differential pressure sensor. The
measurements were carried out every year in September or October,
Table 2
Values of  and 0 in the Mahmudia meander cross-sections.
Sections A B C D E F
A1 A2 A3 B C1 C2 D E1 E2 E3 F1 F2
 [W m-2
] 14.1 2.2 34.4 3.3 0.6 0.8 0.6 27.2 0.4 29.8 19.7 18.6
0 [N m-2
] 15.5 2.7 24.1 4.9 1.9 2.6 2.8 22.0 4.1 23.4 21.1 16.4
Fig. 11. Mean bed shear stress from 1D simulation and estimates with Eq. (3).
194 L. Jugaru Tiron et al. / Geomorphology 106 (2009) 186­197
when the Danube discharge usually reaches the lowest values. The
bathymetry profiles were situated at the bifurcation of the Mahmudia
meander, close to our three aDcp profiles (A1/A2/A3).
The monitoring of the bifurcation cross sections (Fig. 13)
shows that the channel morphology adjusted suddenly during
the studied period with the intense plugging of the former
meander inlet (aggradation rates up to 11 m, i.e., 1.8 m y- 1
) and
the intense scouring of the artificial canal (degradation rates up
to 13 m, i.e., 2.1 m y- 1
). Dramatic deposition in the former meander
even caused the development of islands rapidly fixed by the
riparian forest. In addition, Jugaru et al. (2006) showed that the
meander mobility in the horizontal plane has been fixed since
the 1990.
The comparison of bathymetry profiles measured by Popa (1990­
1996) and our measurements of 2006 makes sense as the hydrologic
regime was similar between 1990 and 2000 (homogeneity of the
maximal monthly discharges, cf. Fig. 13). The comparison shows that
during the last few years the morphodynamical processes have
dropped in intensity: the sedimentation rate in the former meander
was only 0.1 m y-1
during 1996­2006, whereas the erosion rate in
the artificial canal was 0.2 m y-1
. Such a contrast in evolution rates is
much more significant than potential errors stemming from the
different measuring procedures followed by Popa (1997) and us.
This clearly indicates that after a fast relaxing evolution, the
morphology of the alluvial system has reached its new dynamic
equilibrium.
5.3. Sediment balance in St. George branch
The SSC concentrations found in the investigated sections during
the flood (10­180 mg l-1
) are quite low in comparison with data
recorded in the world major rivers. Meybeck et al. (2003) reported the
following values for daily SSC: 849 mg l-1
for the Mississippi River
(USA), 962 mg l-1
for the Mekong River (Thailand), 8240 mg l-1
for the
Rio Grande River (USA), and even 15 760 mg l-1
for the Colorado River
(USA). A study on the Rhône River during a flood period indicates high
SSC values of 36890.05 mg l-1
(Eyrolle et al., 2006).
The volume of the deposited sediments within the former
meander between the A2 and E2 sections can be estimated to be
0.0035×106
m3
d-1
, for a discharge water of 4600 m3
s-1
(at Ceatal St.
George).
An estimation of the suspended solid discharge at the mouth of the
St. George branch and to compare it with the previous estimates is
calculated. Panin and Jipa (2002) and Panin (2003) calculated the
amount of solid fluxes for the period following the building of the dam
(1983­2000). They estimated the suspended sediment output at the
mouth of the St. George branch to be 24.2% of the total Danube output
and to lie between 3.8×106
and 6×106
m3
y-1
. For the same period and
Fig. 12. Cross-section averaged suspended sediment concentration (SSC) from aDcp measurements.
Fig. 13. River-bed evolution (1990­1996) in the Mahmudia modified system (Popa, 1997) and maximal monthly discharges between 1990 and 2004 at Ceatal St. George gauging
station.
195L. Jugaru Tiron et al. / Geomorphology 106 (2009) 186­197
the same station, Bondar (unpublished data) mentioned a value of
2.99×106
m3
y-1
.
For the three days during our aDcp measurements, we estimated
the solid volume going out of the Mahmudia system to be between
0.043×106
and 0.068×106
m3
. These values are higher than the
data reported by Panin and Jipa (2002) and Panin (2003) (between
0.012­0.016×106
m3
d- 1
). Deviations between estimates are
significant, but unfortunately the same order of magnitude as the
expected errors associated with the estimating method each author
used.
Further research could deal with sediment storage in the former
meanders located downstream. The St. George branch loses alluvial
material load along its course by progressive sedimentation and
laterally through channels, canals, crevasses, and overflow. The
storage amounts to 0.005×106
m3
d-1
(43%, mean value for
the years 1981­1989), according to Bondar and Panin (2000). This
high value can be explained by the discontinuity in sediment
transport capacity from training works achieved in that specific
period of time, as studied at the Mahmudia site in this paper.
6. Conclusion
This paper reports the flow and sedimentary analysis of a cutoff
meander in the Danube Delta, the Mahmudia cutoff meander
system. The aDcp and water sampling survey performed during the
2006 100-year recurrent flood offered data on the distribution of
water and suspended-load fluxes through both natural and manmade
channels. Since the meander correction, morphological
trends were assessed through the bed shear stress estimates and
through the available morphological monitoring. The impact of the
2006 100-year recurrent flood is observed in the velocity and SSC
distributions on the investigated cross-sections. The water flow
acceleration in the artificial canal (from 1.8 m3
s- 1
upstream the
bifurcation to 2.3 m3
s- 1
) enhances incision processes in the
artificial canal (gravitationally related sedimentary flow), while the
liquid and solid fluxes through the former meander are greatly
reduced (from 338 kg s- 1
upstream the bifurcation to 73 kg s- 1
even 8 kg s- 1
, with sediment deposition processes), especially
downstream of the first meander apex. The former main channel
undergoes very visible infilling processes (observed through the
development of islands fixed by the riparian forest) correlated with
the decrease in hydraulic energy. Since 1996, the morphology of the
alluvial system seems to have nearly reached a new state of
dynamical equilibrium (diminishing of the sedimentation rate in
the former meander (0.1 m y- 1
) and of the erosion rate in the
artificial canal (0.2 m y- 1
)).
Similar measurements for other hydrological events and even for
other deltaic meanders would be valuable to assess the generality of
our conclusions. In particular, knowing how the observed processes
scale with the multiannual flow rate in the St. George branch would be
useful. Our process-oriented observations can contribute to further
comprehensive studies with extended time and space scales in order
to establish a sediment balance for the whole St. George branch or the
whole Danube Delta.
Acknowledgements
This paper was accomplished within a bilateral Brancusi "Fonctionnement
des systmes hydro-sédimentaires de deux zones deltaques:
le Rhône et le Danube", project holding between GeoEcoMar
(Bucharest, Romania) and CEREGE (Aix-en-Provence, France) and
within a National Project, "The Black Sea: Biotical, paleoecological and
paleoclimatic evolution since the last 7000 years" (PNCDI II IDEI, no.
364/2007).
The authors thank the team of Geomorphology from CEREGE for
their technical and logistic support and for their participation in the
campaign for data acquisition. Many thanks as well as to the Cemagref
(Lyon, France) for the logistic support and data processing.
References
Bagnold, R.A., 1966. An approach to the sediment transport problem from general
physics. U.S. Geol. Surv. Prof. Pap. 422-I, 1­37.
Batalla, R.J., Gómez, C.M., Kondolf, G.M., 2004. Reservoir-induced hydrological changes
in the Ebro River basin (NE Spain). J. Hydrol. 290 (1­2), 117­136.
Biedenharn, D.S., Thorne, C.R., Watson, C.C., 2000. Recent morphological evolution of
the lower Mississippi River. Geomorphology 34, 227­249.
Bondar, C., 2004. Studiul privind consecintele asupra mediului, precum si cele
economice si sociale in Rezervatia Biosferei "Delta Dunarii" Romania. determinate
de constructia canalului Bastroe din Rezervatia Biosferei Dunarea-Ucraina. Report
GeoEcoMar, Bucharest, Romania.
Bondar, C., Panin, N., 2000. The Danube Delta hydrologic database and modeling. GeoEco-Mar.
5­6, 5­53.
Dinehart, R.L., Burau, J.R., 2005. Repeated surveys by acoustic Doppler current profiler
for flow and sediment dynamics in a tidal river. J. Hydrol. 314, 1­21.
Eyrolle, F., Duffa, C., Antonelli, C., Rolland, B., Leprieur, F., 2006. Radiological consequences
of the extreme flooding on the lower course of the Rhone valley (December 2003,
South East France). Geomorphology 366, 427­438.
Gartner, J.W., 2004. Estimating suspended solids concentrations from backscatter
intensity measured by acoustic Doppler current profiler in San Francisco Bay,
California. Mar. Geol. 211, 169­187.
Giosan, L., Bokuniewicz, H., Panin, N., Postolache, I.,1999. Longshore Sediment Transport
Pattern Along Romanian Danube Delta Coast. J. Coast. Res. 15 (4), 859­871.
Harmar, O.P., Clifford, N.J., Thorne, C.R., Biedenharn, D.S., 2005. Morphological changes
of the Lower Mississippi River: geomorphological response to engineering
intervention. River Res. Applic 21, 1107­1131.
Hooke, J.M., 1986. Changes in Meander Morphology. International Geomorphology, Part I.
Wiley, Chichester, pp. 591­600.
Ichim, I., Radoane, M., 1986. Efectele barajelor in dinamica reliefului. Abordare
geomorfologica. Editura Academiei. Bucuresti.
Ielenicz, M., Negut, S., Ene, M., Savulescu, I., Vlad, L.B., Neacsu, M.C., 2006. Atlas
Geografic General. Editura Universitara. ISBN: 978-973-749-076.
Jimenez, J.A., Sanchez-Arcilla, A.,1993. Medium-term coastal response at the Ebro Delta,
Spain. Mar. Geol. 114, 105­118.
Jugaru, L., Provansal, M., Panin, N., Dussouillez, P., 2006. Apports des Systmes
d'Information Géographiques  la perception des changements morphodynamiques
(1970­2000) dans le delta du Danube. Le cas du bras de Saint-George. Geo-Eco-Mar.
12, 29­42.
Keller, E.A., 1972. Development of alluvial stream channels. Bull. Geol. Soc. Am. 83,
1531­1536.
Kesel, R.H., 2003. Human modifications to the sediment regime of the Lower Mississippi
River flood plain. Geomorphology 56, 325­334.
Kiss, T., Fiala, K., Sipos, G., 2007. Alterations of channel parameters in response to river
regulation works since 1840 on the Lower Tisza River (Hungary). Geomorphology
98 (1­2), 96­110.
Kostaschuk, R., Best, J., Villard, P., Peakall, J., Franklin, M., 2005. Measuring flow velocity
and sediment transport with an acoustic Doppler current profiler. Geomorphology
68, 25­37.
Land, J.M., Jones, P.D., 2001. Acoustic measurement of sediment fluxes in rivers and
near-shore waters. Seventh Federal Interagency Sedimentation Conference, Reno,
Nevada.
Lane, E.W., 1947. The effect of cutting off bends in rivers. Studies in Engineering
University of Iowa City, Iowa City, pp. 239­240.
Meybeck, M., Laroche, L., Dürr, H.H., Syvitski, J.M.P., 2003. Global variability of daily total
suspended solids and their fluxes in rivers. Glob. Planet. Change 39, 65­93.
Mihailescu, N., 1983. Lower Danube Recent Alluvia Sediments and Sedimentary Factors,
An. Inst. Geol. Geof., Bucuresti, pp. 287­296.
Panin, N., 2003. The Danube Delta. Geomorphology and Holocene evolution: a
synthesis. Géomorphologie: relief, processes. Environnement 4, 247­262.
Panin, N., Jipa, D., 2002. Danube river sediment input and its interactionwith the North --
western Black Sea. Estuarine. Costal and Shelf Science 54, 551­562.
Panin, N., Popa, A., Gaita, C., Farnoaga, D., 1992. Dinamica sedimentelor pe bratele
Dunarii in delta. Modificarile determinate de amenajarile hidrotehnice. Simpozionul
"Provenienta si efluenta aluviunilor" 4/1992, Piatra Neamt, Romania.
Piégay, H., Bornette, G., Citterio, A., Hérouin, E., Moulin, B., Statiotis, C., 2000. Channel
instability as a control on silting dynamics and vegetation patterns within perifuvial
aquatic zones. Hydrol. Process. 14, 3011­3029.
Popa, A., 1997. Environment changes in the Danube Delta caused by the hydrotechnical
works on the St. George branch. Geo-Eco-Marina 2, 135­147.
Popa, A., Panin, N., Farnoaga, D., Mihaila, E., unpublished report. Observatii asupra
modificarilor determinate de amenajarile hidrotehnice de pe bratul Sf. Gheorghe, in
transportul de apa si de sedimente.
Reichel, G., Nachtnebel, H.P., 1994. Suspended sediment monitoring in a fluvial
environment: advantages and limitations applying an acoustic Doppler current
profiler. Water Res. 28 (4), 751­761.
Rotaru, E., Le Coz, J., Drobot, R., Adler, M.J., Dramais, G., 2006. ADcp measurements of
suspended sediment fluxes in Banat rivers. Romania. Balwois, Ohrid, Macedonia
23­26, 13 May 2006.
Schmitt, L., Maire, G., Humbert, J., 2001. La puissance fluviale: définition, intért et
limites pour une typologie hydro-géomorphologique de rivire. Z. Geomorphol. N. F.
45 (2), 201­224.
196 L. Jugaru Tiron et al. / Geomorphology 106 (2009) 186­197
Sestini, G., 1992. Implications of climatic changes for the Nile Delta. In: Jeftic, L.,
Milliman, J.D., Sestini, G. (Eds.), Climatic Change and the Mediterranean. Edward
Arnold, New York, pp. 535­601.
Shields, F.D., Nunnally, N.R., Asce, A.M., 1984. Environmental aspects of clearing and
snagging. J. Environ. Eng. 110 (1), 152­165.
Stanley, D.J., 1996. Nile delta: extreme case of sediment entrapment on a delta plain and
consequent coastal land loss. Marine Geology 129 Letter Section, pp. 189­195.
Stanley, D.J., Warne, A.G., 1993. Nile delta: recent geological evolution and human
impact. Science 260, 228­231.
Topographical Military Direction, 1980. Topographical maps, 1:25000, sheets L-35-107C-a,
L-35-107-C-b, L-35-107-C-c, L-35-107-C-d, Bucharest, Romania.
Vandenberghe, J., 2003. Climate forcing of fluvial system development: an evolution of
ideas. Quat. Sci. Rev. 22, 2053­2060.
Yorke, T.H., Oberg, K.A., 2002. Measuring river velocity and discharge with acoustic
Doppler profilers. Flow Meas. Instrum. 13, 191­195.
197L. Jugaru Tiron et al. / Geomorphology 106 (2009) 186­197