Water Resources Research - 2006 - Bungartz - Fluvial Suspended Sediment Dynamics Implications for Particulate Organic

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 12

WATER RESOURCES RESEARCH, VOL. 42, W10424, doi:10.

1029/2005WR004486, 2006

Fluvial suspended sediment dynamics:


Implications for particulate organic carbon transport modeling
Heinz Bungartz,1 Angela Krüger,1 and Christof Engelhardt1
Received 4 August 2005; revised 2 May 2006; accepted 7 July 2006; published 20 October 2006.
[1] The effect of fluvial suspended sediment aggregation on the hydrodynamic transport
properties of particulate organic carbon (POC) was investigated using a combination of
field measurement and numerical modeling. Field-based settling tube experiments were
conducted to obtain particle settling velocity frequency distributions of a 4-km-long
section of a lake outlet river at two distinct discharge levels (6.8 and 29.7 m3 s1). Results
of field studies conducted in April 2001 and April 2002 showed that the lacustrine settling
velocity frequency distributions of suspended particulate matter (SPM) and POC were
altered owing to the formation of riverine aggregates having comparatively higher settling
velocities. Data suggest that aggregation promoted deposition fluxes of POC; these fluxes
were in turn closely tied to the processes that controlled deposition of SPM. Data
interpretation was aided by the employment of a hydrodynamic sediment transport model.
The approach used to integrate aggregation in the sediment transport simulations
represents the interaction of multiple settling velocity mass fractions. Both field data and
modeling demonstrated the effect of aggregation on the increase in deposition fluxes of
POC.
Citation: Bungartz, H., A. Krüger, and C. Engelhardt (2006), Fluvial suspended sediment dynamics: Implications for particulate
organic carbon transport modeling, Water Resour. Res., 42, W10424, doi:10.1029/2005WR004486.

1. Introduction edged that the SPM load in rivers mostly consists of


[2] Production, fluxes, and transformation of particulate composite particles or flocs/aggregates, formed by allochth-
organic carbon (POC) and the impact of these processes on onous and autochthonous material, which could be from the
the local habitat are key concerns in understanding aquatic pelagic, benthic or terrestrial environment [Eisma, 1993].
ecosystems [Ittekkot, 1988; Minshall et al., 2000]. In rivers, These flocs represent a complex matrix of microbial com-
POC originates from organic-rich surficial soil, riparian munities, organic and inorganic particles, and interfloc
vegetation, and autochthonous production, which is influ- spaces or pores [Droppo, 2001]. Aggregate formation is
enced by the trophic state and season. Also important is the mediated by a number of physical, biological and chemical
allochthonous input of algae from lakes and tributaries, and variables. Aggregation is a dynamic process, indicating that
the contribution of macrophyte debris from the littoral zone. hydrodynamic aggregate transport properties change in
[3] Much of the literature regarding carbon transport has response to the ambient conditions. However, the extent
focused on the mechanics controlling the production, stor- to which POC fluxes are quantitatively linked to the
age and export of POC from river basins [Barth et al., 1998; aggregate dynamics in the freshwater fluvial environment,
Gomez et al., 2003]. Karickhoff [1981], Voice and Weber is poorly understood because of few experimental studies.
[1983], and Sobek et al. [2004] emphasize the central role of [5] Previous studies on aggregation phenomena and the
POC in organic contaminant association and transport. processes governing them, have been focused on marine,
Moreover, particulate organic matter often has a high estuarine and lacustrine environments [Kranck, 1973; Eisma
nutritional value and thus can be a major food source for and Li, 1993; Grossart and Simon, 1997]. In marine
both pelagic and benthic consumers [Ramseier et al., 2000]. systems the role of large aggregates (marine snow) in the
Of particular interest, along with lateral carbon flows, are vertical flux of organic matter into the oceans’ interior is
the in-river vertical fluxes of POC which couple the water widely accepted [Jackson and Burd, 1998].
column with the streambed. [6] In fluvial systems the basic nature of cohesive par-
[4] On the basis of the determination of carbon content in ticles, their constituents and hydrodynamic transport prop-
suspended particulate matter (SPM), POC yields are com- erties have only recently been investigated [Leppard and
monly estimated as a proportion of the SPM discharge. Droppo, 2005]. Correlative microscopy observations were
Thus the movement of POC is directly linked to the used by Droppo [2001, 2004] to evaluate the internal
movement of suspended sediments. It is widely acknowl- structure of riverine flocs and its potential effect on floc
behavior. The works of Tsai et al. [1987] and Van Leussen
[1997] suggest that turbulence is the most relevant physical
1
Department of Ecohydraulics, Leibniz Institute of Freshwater Ecology mechanism for the formation of aggregates in riverine
and Inland Fisheries Berlin, Berlin, Germany. transport. Suspended sediment concentration, and concen-
tration of dissolved and particulate organic carbon might be
Copyright 2006 by the American Geophysical Union.
0043-1397/06/2005WR004486

W10424 1 of 12
19447973, 2006, 10, Downloaded from https://agupubs.onlinelibrary.wiley.com/doi/10.1029/2005WR004486 by Sichuan University, Wiley Online Library on [04/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
W10424 BUNGARTZ ET AL.: SUSPENDED SEDIMENT DYNAMICS W10424

Figure 1. Map of the study reach of the Spree River showing the locations of the transect stations T1,
T2, and T3/T3*; T3 and T3* denote the different outflow cross sections of the study reach at discharge q1
and q2, respectively.

other important factors controlling aggregation [Burban et al., fluvial system, we examined changes in the settling velocity
1990; Droppo and Ongley, 1994]. frequency distribution of SPM and POC in a lake outlet
[7] POC incorporates many forms of organic material river (Spree River, Germany). We now present data from
(such as detritus, macrophytes, microorganisms and their field measurements of settling velocity, and results of sim-
secretions, and transparent exopolymer particles) which are ulations using a hydrodynamic sediment transport model.
extremely cohesive and may thus help promote aggregation We applied the model to transport and deposition of three
and aggregate stabilization. However, studies on the impact interacting settling velocity fractions of sediment-associated
of biological factors (such as the influence of sticky carbon. The objectives of our study were (1) to demonstrate
polymers produced by diatoms or bacteria) on particle the influence of enhanced turbulence in the river (compared
aggregation and their role in floc stabilisation and sedimen- to the lake) on aggregation and associated changes in the
tation have been almost solely orientated to marine environ- lacustrine settling velocity distribution, (2) to predict POC
ments [Alldredge et al., 1993; Passow et al., 2001] and vertical fluxes and to investigate the extent to which these
engineering systems [Bura et al., 1998]. fluxes are tied to aggregation processes, and (3) to reveal the
[8] Despite the common presence of aggregates within potential error due to particle aggregation in predicting POC
fluvial systems, their mechanisms of formation are not well transport. For the both latter objectives, we carried out two-
understood. The most significant impact of flocculation, in dimensional vertical plane transport simulations. Results
terms of sediment transport, is that it alters the downward were evaluated with field data of total SPM and POC
fluxes of sediment by changing the hydrodynamics of the concentration and settling velocity frequency distributions
sediment. Aggregation and break up of flocs essentially for river discharge levels of 6.8 and 29.7 m3 s1.
alter the particle size, density, and surface area, with
concomitant changes in particle settling velocity [Droppo 2. Study Site
et al., 2000; Droppo, 2004]. Turbulent shear is believed to
affect aggregation processes, and consequently aggregate [11] The Spree River is a typical lowland river in northern
sizes and settling velocities, in two opposing ways. Turbu- Germany with a catchment area of about 10,000 km2. The
lence is to some extent conducive to particle aggregation river originates in the Lusatian mountains (Saxonia, Ger-
resulting in larger particles, however increasing turbulent many) at an elevation of 580 m, and flows for 400 km
shear rate presumably increases the chance of floc destruc- through several shallow lakes to Berlin. A detailed descrip-
tion, thus limiting floc size [Manning and Dyer, 1999]. This tion of the Spree River, and its tributaries and lakes, is given
dynamics, in turn (because settling velocity is a fundamental by Köhler [1994].
input value in modeling these processes), impacts the ability [12] We investigated a 4-km-long sixth-order section of
to predict suspended sediment and sediment-associated the river, the ‘‘Krumme Spree’’ (14°000E, 52°070N). The
carbon transport. inflow cross section of the study reach (transect station T1)
[9] Even though it is clear that settling velocity changes was located about 500 m downstream from the outlet of
dynamically due to particle formation processes, aggrega- Lake Neuendorfer See (Figure 1). The enhanced turbulence
tion is neglected in virtually all of existing sediment (compared to the lake) on the chosen river section provided
transport models developed for rivers, and the composite optimal conditions for study of the influence of changes in
structures are usually assumed to behave as discrete indi- turbulent shear on aggregation processes and associated
vidual particles. Conventional models which omit the changes in the settling velocity distribution.
effects of particle dynamics provide erroneous predictions [13] As a result of channel straightening between 1906
of POC transport [Ongley et al., 1992; Nicholas and Walling, and 1912, the Krumme Spree now has a trapezoidal
1996]. channel. Mean slope is approximately 0.01%, the mean
[10] To establish the role of aggregation in suspended width 25 m, and the mean depth 1.5 to 2.5 m, depending on
sediment dynamics and its effect on POC transport in a river discharge. Between 1997 and 2002, river discharge
varied from 2 m3 s1 to 32 m3 s1 (mean 12.4 m3 s1), with

2 of 12
19447973, 2006, 10, Downloaded from https://agupubs.onlinelibrary.wiley.com/doi/10.1029/2005WR004486 by Sichuan University, Wiley Online Library on [04/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
W10424 BUNGARTZ ET AL.: SUSPENDED SEDIMENT DYNAMICS W10424

mean flow velocities between 0.08 m s1 and 0.67 m s1 3.3. Particulate Organic Carbon Concentration
(mean 0.29 m s1). [18] Particulate organic carbon (POC) was measured
[14] At transect T1, concentration and composition of using the same samples used for SPM determination. What-
SPM, content of POC, and aggregate settling characteristics man glass microfibre filters GF/F were washed with 50 ml
were predominantly controlled by the particle input received pure water, combusted at 500°C for 5 h and preweighed.
from the lake. The seasonal variability of SPM concentra- After filtration, filters were dried to constant weight at
tion was greatly affected by phytoplankton dynamics. From 105°C and weighed again. POC was determined using the
April 2001 until May 2002, when field studies were con- CHN Analyser Vario EL (Elementar Analysensysteme,
ducted, SPM concentration at transect T1 varied between Hanau). POC was measured as mg C L1. In the combus-
3.1 and 11.7 mg dry weight L1. The lake, as a point source tion reactor, oxygen gas and other oxidizing and catalyzing
of suspended organic carbon, imposes on this outlet stream reagents support the completeness of high-temperature
its own patterns of mostly autochthonous organic carbon combustion of organic carbon compounds to carbon diox-
production. Expressed as a percentage of SPM, contents of ide. CO2 was detected and quantified by the thermal
POC at transect T1 showed a large organic rich fraction conductivity detector.
(annual mean 17% of dry SPM with only small seasonal
fluctuations) [Rother and Köhler, 2005]. In the lake, typical 3.4. Settling Velocity Determinations
frequency distributions are dominated by material with [19] In order to determine settling velocity frequency
settling velocities <1.25 cm h1 (75 – 80% of SPM); mate- distributions of suspended particles, a modified bottom
rial with settling velocities >90 cm h1 contribute only 2 to withdrawal method was employed [McCave and Syvitski,
4% to SPM concentration. In the river, under equilibrium 1991; Prochnow et al., 1996]. Field settling tubes (Perspex
conditions, only proportions between 38 and 60% are in columns with an inner diameter of 4.7 cm, height 30 cm)
the lower than 1.25 cm h1 fraction, whereas the fraction were filled with river water 0.3 m below surface, and
>90 cm h1 ranges from 6 to 20% [Kozerski et al., 1991]. stored vertically in a large container filled with river water.
The distinct differences in settling velocity distribution [20] Frequency distributions of SPM were obtained from
indicate how the lacustrine hydrodynamics of suspended sediment accumulation – time curves, the so-called Oden
sediment may be altered by riverine turbulence structures. curves. Following the theory of Oden [1915], in such a
system, after time t, the material that has settled to the
3. Field Work: Materials and Methods bottom of a tube with the height H, can be divided into two
fractions: (1) a portion of the sediment with settling velocity
3.1. Experimental Design vs(t) < H/t and (2) particles with vs(t)  H/t. According to
[ 15 ] Field settling experiments were conducted on these considerations, a water sample is filled into the
24 April 2001 and 23 April 2002, at discharge levels of settling tube equipped with a tap at the bottom. At times
6.8 m3 s1 (= q1) and 29.7 m3 s1 (= q2), respectively. tk (0 < t1 < . . . < tN) samples are withdrawn and filtered in
Longitudinal changes in total SPM concentration and set- order to determine the mass of particles Bk = Bk(tk) settled
tling velocity frequency distribution of SPM were deter- during the time tk. The discrete Oden curve is then estab-
mined for both discharge levels. To examine how POC lished by pairs (tk, Bk).
pathways are tied to SPM aggregate dynamics, percentages [21] In our experiments, at each of three withdrawal times
of POC in total SPM and in SPM settling velocity fractions t1 = 1 h, t2 = 6 hrs, and t3 = 24 hrs, samples from one rack
were measured, resulting in settling velocity frequency (with four tubes to provide replicate samples for each
distribution of POC. withdrawal time tk), were taken and then filtered to determine
[16] The distances between the inflow cross section (T1) the sediment content in the lowest 5 cm of each tube. As a
of the study reach and the sampling sites T2 and T3/T3* result of this quasi in situ measurement, the SPM load was
were 1050 m and 3830 m/2200 m, respectively (Figure 1). divided into three fractions, each assigned to an interval of
T3 and T3* denote the different outflow cross sections of settling velocities (first fraction < 1.25 cm h1, 1.25 cm h1 <
the study reach at discharge q1 and q2, respectively. The second fraction < 30.0 cm h1, third fraction >30.0 cm h1).
distance T3* = 2200 m at discharge q2 was chosen to be More detailed information on the measuring procedure and
shorter than that at discharge q1 (T3 = 3830 m) because data processing are given by Prochnow et al. [1996]. Settling
under conditions of higher turbulence/discharge an equilib- velocity distributions of POC have been calculated by using
rium between the settling velocity fractions is achieved measured POC contents of total SPM and of SPM settling
faster [Bungartz and Wanner, 2004]. velocity fractions.
3.2. Suspended Particulate Matter Concentration 3.5. Flow Characteristics
[17] Water samples were collected in Lagrangian (time of [22] Estimates of effective river bed roughness heights and
travel) sampling mode between the transect stations T1 and shear velocities in the river midstream region were obtained
T3 at 7 sites (discharge level q1), and between T1 and T3* by fitting the log law of the wall to measured vertical profiles
at 4 sites (discharge level q2). At each site, three replicate of the streamwise flow velocity. For these measurements, we
grab samples were collected midstream at a water depth of used a portable one-dimensional electromagnetic current
about 0.3 m. To determine suspended particulate matter meter (NAUTILUS C2000, OTT/Germany).
(SPM) concentration (mg dry weight L1), samples were
filtered through preweighed Sartorius 0.45 mm cellulose 4. Field Work: Results and Discussion
acetate filters, dried to constant weight at 105°C, and
weighed again. [23] Distributions of particle settling velocity were mea-
sured in both hydraulic situations at the transect stations T1,

3 of 12
19447973, 2006, 10, Downloaded from https://agupubs.onlinelibrary.wiley.com/doi/10.1029/2005WR004486 by Sichuan University, Wiley Online Library on [04/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
W10424 BUNGARTZ ET AL.: SUSPENDED SEDIMENT DYNAMICS W10424

SPM appears to be already altered by river turbulence


structures in the section from the lake outlet to transect
T1 (Figure 3). It can be hypothesized that the typical
lacustrine sediment settling characteristics are modified
more quickly (in comparison to discharge q1) due to higher
fluid shear. An equilibrium in the floc formation processes
is consequently reached faster, due to a short response time
of aggregation (the time needed to establish an approxi-
mately constant relation between settling velocity fractions)
and further changes of the distribution (such as the decrease
of the <1.25 cm h1 fraction) are relatively small over the
traveling distance to transect T3*. The median settling
velocity of SPM increases from 0.94 cm h1 at T1 to
1.02 cm h1 at T3*.
[25] Comparison of settling velocity distributions of POC
and SPM for both discharge situations (see Figure 3)
indicates similar distribution patterns at all transect stations.
Mean contents of POC in total SPM of 14.8% and 17.7%
were determined for the discharge levels q1 and q2, respec-
tively. Percentages of POC present in SPM and in settling
velocity fractions are summarized in Table 1. An interesting
result concerns the formation of aggregates with high
settling velocities; the percent POC in the >30 cm h1
fraction increased from 10% (at transect T1) to 16.3% at the
outflow cross section (T3) at low discharge q1, and from 9.4
to 13.3% for high discharge q2. Data in Table 1 also show
that POC content in the settling velocity fractions differs
Figure 2. Longitudinal profiles of suspended particulate
matter (SPM) and particulate organic carbon (POC)
concentration at discharge levels of 6.8 m3 s1 (q1) and
29.7 m3 s1 (q2). Each point represents the mean value with
the standard error of the mean (n = 3 –5).

T2, and T3/T3* with the modified bottom withdrawal


method described above. Longitudinal profiles of total
SPM and POC concentration (Figure 2) showed that at
low discharge q1, there were losses of 25% and 30% for
SPM and POC respectively, probably attributable to simple
sedimentation along the river reach, (SPM (mg dry weight
(dw) L1) = 6.66  0.33 X, R2 = 0.38, and POC (mg C
L1) = 1.04  0.066 X, R2 = 0.33; distance X in kilo-
meters). At high discharge q2, there was a slight increase of
SPM and POC concentration (Figure 2), possibly due to net
resuspension. Concentration data, however, showed no
significant dependency on traveling distance.
[24] The results of the settling column experiments show
that settling velocity distribution and its longitudinal alter-
ation is dependent on aggregation processes controlled by
river discharge/turbulence (Figure 3). At the low discharge
q1, the settling velocity distribution of SPM at transect T1
still exhibits characteristics typical for the distribution in
the lake (about 80% of the total mass are in the lower than
1.25 cm h1 fraction). From transect T1, this distribution
changes slowly over the traveling distance to transect T3.
While the relative contribution of particles with settling
velocities <1.25 cm h1 to the total mass declines to 60% at
transect T3, the contribution of particles having settling
velocities >30 cm h1 increases from 14 to 31%. This
process results in an increase in the SPM median settling Figure 3. SPM and POC settling velocity fractions and
velocity from 0.44 cm h1 at T1 to 0.89 cm h1 at T3. At frequency distributions determined at the transect stations
the high discharge q2, the settling velocity distribution of T1, T2, and T3/T3* at discharge levels of 6.8 m3 s1 (q1)
and 29.7 m3 s1 (q2).
4 of 12
19447973, 2006, 10, Downloaded from https://agupubs.onlinelibrary.wiley.com/doi/10.1029/2005WR004486 by Sichuan University, Wiley Online Library on [04/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
W10424 BUNGARTZ ET AL.: SUSPENDED SEDIMENT DYNAMICS W10424

Table 1. Percentages of Particulate Organic Carbon in Total Suspended Particulate Matter and in Three Settling Velocity Fractions at the
Transect Stations T1, T2, and T3/T3* at Discharge Levels of 6.8 m3 s1 (q1) and 29.7 m3 s1 (q2)
Discharge q1 Discharge q2

Transect Station Total SPM First Fraction Second Fraction Third Fraction Total SPM First Fraction Second Fraction Third Fraction

T1 14.5 14.6 21.3 10.0 17.9 18.9 19.7 9.4


T2 13.7 12.3 13.6 17.4 17.3 19.3 18.6 10.7
T3/T3* 16.1 15.0 22.5 16.3 17.8 19.1 19.1 13.3

from that in total SPM at both discharge levels. The varying tions) with corresponding multiple deposition behavior.
POC content in the >30 cm h1 fraction over the traveling Aggregation and breakup are represented indirectly by mass
distance indicates POC transfer between fractions owing to fluxes between the SPM/POC fractions, thus simulating
aggregation processes. This change in the proportions of only the effects rather than the processes of aggregation.
POC in SPM fractions (as a results of nonequilibrium The model has previously been calibrated with regard to
transport) may influence transport and deposition of organic morphological data and the specific hydraulics of the
carbon. An overview of further settling characteristics given examined river reach. Details of longitudinal river topogra-
in Table 2, reveals a strong consistency of SPM and POC phy, results of flow velocity calculations, and results of
data at both discharge levels. model application to transport and deposition of tracer
[26] The similarity of the settling velocity distributions of particles are presented by Bungartz and Wanner [2004].
SPM and POC (Figure 3) suggests that both distributions, Finally, the ability of the model to reliably simulate natural
and their alteration over the traveling distance, were con- SPM transport and settling was validated.
trolled by the same process of aggregation and breakup of [29] We applied a two-dimensional vertical plane version
flocs. The POC is assumed to promote particle aggregation, of the numerical model SEDIFLOW. The governing equa-
and the increasing contents of POC in the >30 cm h1 tions used to describe 2-D turbulent open channel flow are
fraction might promote floc stabilization. Considering also based on the Reynolds-averaged Navier-Stokes equations
the comparable longitudinal decline in SPM and POC for incompressible flows. The standard k-e model [Rodi,
concentration at discharge q1 (Figure 2), it can be suspected 1993] was employed to calculate the vertical turbulent
that riverine turbulence provides aggregation mechanisms transfer coefficients for momentum and particulate matter.
which continuously generate particles that settle toward the [30] The selected model type reflects the downstream
river bed, thus largely accounting for the losses of SPM and transport of settling velocity mass fractions in the midstream
POC. part of the river by vertical plane simulations. Vertical
structures of flow and SPM/POC concentration, as well as
5. Modeling: Model Description the interaction between the pelagic and the benthic zone
controlled by the prevailing hydrodynamic conditions, are
[27] In order to reliably simulate cohesive sediment calculated in detail. Transverse variations and particle
transport, it is necessary to add to the hydrodynamic model retention due to the presence of dead zones or vegetation
base, appropriate algorithms describing the properties and were neglected in the model. The focus and scope of this
behavior of cohesive sediment, and its response to hydro- paper preclude describing the numerical model in full; a
dynamic stresses as well as to the interactive biological and more detailed description of the mathematical background
chemical factors influencing floc development. has been provided by Bungartz et al. [2000]. However, for a
[28] In the present study, we used a hydrodynamic model better understanding it is helpful to outline here the basic
called SEDIFLOW [Prochnow et al., 1991]. In this model, idea of the approach for integration of particle aggregation
an approach is employed to incorporate particle aggregation in the sediment transport model. Detailed information on
as an additional element. The aggregation module repre-
sents the interaction of multiple SPM/POC classes (frac-

Table 2. Particle Settling Characteristics at the Transect Stations T1 and T3/T3* at Discharge Levels of 6.8 m3 s1 (q1) and 29.7 m3 s1
(q2)a
Median Averaged
Median Settling Averaged Averaged Settling
Settling Velocity at Settling Settling Velo- Velocity in
Velocity at T1 T3/T3* Velocity at T1 city at T3/T3* Equilibrium
(Measured), (Measured), (Measured), (Measured), (Simulated),
Mean SPM Mean POC cm h1 cm h1 cm h1 cm h1 cm h1
Discharge Concentration, Concentration,
Level mg dw L1 mg C L1 SPM POC SPM POC SPM POC SPM POC SPM POC

q1 6.02 0.89 0.44 0.38 0.89 1.03 8.67 7.01 17.38 18.08 17.5 17.2
q2 6.55 1.16 0.94 0.87 1.02 0.96 11.11 8.28 14.44 12.06 14.28 11.37
a
Mean concentration equals longitudinally averaged concentration.

5 of 12
19447973, 2006, 10, Downloaded from https://agupubs.onlinelibrary.wiley.com/doi/10.1029/2005WR004486 by Sichuan University, Wiley Online Library on [04/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
W10424 BUNGARTZ ET AL.: SUSPENDED SEDIMENT DYNAMICS W10424

Table 3. Hydraulic Conditions at Discharge Levels q1 and q2a


Discharge q, Mean Water Mean Flow Mean Bed Longitudinal Response Time
m3 s1 Depth Midstream, m Velocity, m s1 Shear Stress, N m2 Mixing, m2 s1 of Aggregation, hours

6.8 (q1) 2.62 0.18 0.097 1.8 3.1


29.7 (q2) 3.25 0.75 1.76 9.55 0.45
a
Mean values equal longitudinally averaged values.

model parameters and on initial and boundary conditions [33] In the numerical procedure, the effect of particle
are given by Bungartz and Wanner [2004]. formation processes on the settling velocity distribution is
[31] The vertical plane transport of k interacting settling specified by the fluxes Fij. These fluxes were adjusted to get
velocity mass fractions ci is governed by the system of accurate simulations of the settling velocity distribution at
convection-diffusion equations the transect station T3/T3*. Simulation results can be used
 
to quantify the influence of the factors which are believed to
@ci @ci @ci @ Dx @ci regulate particle formation. Using appropriate initial con-
þu þv  ditions (concentration data and settling velocity distribu-
@t @x @y @x sci @x
  tions of SPM and POC measured at the inflow cross
@ Dy @ci i
 þ ci  vs ¼ Fi ; i ¼ 1; . . . ; k ð1Þ section) for the numerical calculations, after the response
@y sci @y
time of aggregation, a roughly constant relation between
concentrations ci and cj is established downstream (Kij =
where t is time (s), x and y are the streamwise and vertical
cjeq/cieq, where cieq, cjeq are the equilibrium concentrations).
Cartesian coordinates (m), respectively, ci is the SPM
It should be stressed that in this model concept, particle
concentration (mg dw L1) of the ith fraction, or POC
interaction is allowed merely between either SPM or POC
concentration (mg C L1) of the ith fraction, u and v are the
fractions. Dynamics in the distribution of POC to sediment
components of flow velocity (m s1) in x, y directions,
settling velocity fractions can be modeled indirectly by
respectively, vis is the settling velocity (m s1) of the ith
different partitioning coefficients Kij for SPM and POC
SPM/POC fraction in quiescent fluid (settling velocity of
fractions.
ith SPM fraction and ith POC fraction is identical), Dx and
Dy are the coefficients of turbulent momentum exchange
(m2 s1) in x and y directions, respectively, and sci is the 6. Modeling: Results and Discussion
turbulent Schmidt number, which relates the coefficients of 6.1. Flow Field Calculations
sediment mixing to the coefficients of turbulent momentum [34] Using field data from previous depth measurements
exchange. and numerical simulations of particulate tracer experiments
[32] The source term Fi (mg dw L1 s1 or mg C L1 (April 1999) at discharge levels of 7.9 and 14.8 m3 s1
1
s ) on the right hand side of equation (1) [Wanner and Pusch, 2000; Bungartz and Wanner, 2004], we
determined the parameters necessary to operate the model
X
k
(for example for the coefficient of longitudinal mixing, a di-
Fi ¼ Fsi þ Fij ð2Þ
j¼1
mensionless value Dx/hu* 70 was identified numerically;
here h is the mean water depth, u* is the mean bed shear
describes external SPM/POC sources and the interaction velocity). In order to characterize the discharge situations q1
between settling velocity fractions. This term models the and q2, a summary of hydraulic parameters is presented in
effects of particle aggregation and breakup on the Table 3. Hydraulic predictions (steady state flow fields) for
concentration of settling velocity fractions, where Fsi is a the two discharge levels are carried forward to the partic-
vector, describing external sources/sinks (mg dw L1 s1 or ulate transport component of the model, which is able to
mg C L1 s1), and Fij is a coupling matrix, describing simulate both the transport of suspended sediment and POC,
SPM/POC fluxes (mg dw L1 s1 or mg C L1 s1) and the depositional behavior for a range of SPM/POC
between k considered SPM/POC fractions, where the mass settling velocity fractions.
transfer between the ith and jth fraction is governed by the 6.2. Model Simulations of SPM and POC Transport
exchange terms in the Spree River
[35] In order to simulate fraction behavior, a specific
Fij ¼ Qði; jÞcj  Qðj; iÞci :
settling velocity vis was assigned to each particle fraction
ci (i = 1,2,3) as follows: v1s = 0.625 cm h1, v2s = 15.6 cm h1,
Here Q(i,j) is the reciprocal value of the response time of and v3s = 50 cm h1. As mentioned above, ci can repre-
aggregation (s1), and Q(j,i) is defined by sent both SPM and POC settling velocity mass fractions.
  While for fractions c1 and c2, the mean value of the respec-
Qðj; iÞ ¼ Qði; jÞ  Kij s1 ; ð3Þ tive interval was used, for c3 the value was an estimate,
because the measuring procedure precluded determination
where Kij are constant partitioning coefficients, describing of an upper limit for the settling velocity. Using the hydrau-
an equilibrium between the settling velocity fractions that is lic predictions resulting from the hydrodynamics module,
assumed to occur under uniform conditions related to the the transport of SPM and POC fractions downstream
mechanisms controlling aggregate formation along the river from transect T1 was simulated with and without particle
reach.
6 of 12
19447973, 2006, 10, Downloaded from https://agupubs.onlinelibrary.wiley.com/doi/10.1029/2005WR004486 by Sichuan University, Wiley Online Library on [04/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
W10424 BUNGARTZ ET AL.: SUSPENDED SEDIMENT DYNAMICS W10424

Figure 4. Comparison of predicted SPM and POC behavior along the sampling reach, with and without
aggregation, at discharge level q1 (all values at 30 cm below surface). (a and b) Settling velocity
fractions. Symbols for first fraction are as follows: solid circles, measured; thin dashed line,
nonaggregated; thick dashed line, aggregated. Symbols for second fraction are as follows: open circles,
measured; thin dotted line, nonaggregated; thick dotted line, aggregated. Symbols for third fraction are as
follows: squares, measured; thin dot-dashed line, nonaggregated; thick dot-dashed line, aggregated.
(c and d) Total concentrations. Symbols are as follows: diamonds, measured values; thin solid line,
calculated as the sum of the fractions, nonaggregated; thick solid line, calculated as the sum of the
fractions, aggregated. (e and f) Mean settling velocity vsm calculated from equation (4) for nonaggregated
(thin solid line) and aggregated (thick solid line). In all graphs, lines represent calculated values, and
symbols represent measured values.

interaction. The measured settling velocity distributions material, deposition and resuspension can occur simulta-
at T1 were used to specify initial sediment concentrations ci neously [Lick, 1982]. However, other investigators argue
at the inflow boundary. In the interacting fractions mode, that simultaneous erosion and deposition do not occur
the values of the partitioning coefficients Kij in equation (3) during cohesive sediment settling [Lau and Krishnappan,
were tuned so that the downstream relationships between 1994]. In our model concept, we assume a spatial variability
SPM/POC fractions (at transect T3/T3*) were accurately of bed shear, and of bed composition and stability, at which
reflected by the simulations. Using these coefficients, proper opposite sediment fluxes at the riverbed are allowed. The
values for the response time of aggregation 1/Q(i,j) were estimation of the resuspension rate is the key problem in
identified by numerical experiments. In the noninteracting modeling nonequilibrium suspended sediment transport,
mode, all fractions behaved independently (Fij = 0 in (2)). particularly when more than one fraction is considered
External sources were neglected (Fsi = 0). [Celik and Rodi, 1988]. Depositional history, consolidation,
[36] Measurements at discharge level q1 showed that the and biostabilization of settled material are factors that
flow field was in an apparently depositional mode, in which essentially affect resuspension of sediment containing a
both SPM and POC concentrations declined along the study significant amount of cohesive particles. Because of these
reach. Therefore, in our simulations, we made the simpli- effects, fine sediment is increasingly difficult to resuspend
fying assumption that deposition occurs, but not resuspen- after deposition [Eisma, 1993].
sion. In contrast, at the considerably higher discharge q2, [38] The features of downstream transport and deposition/
measurements suggested a net resuspension rate resulting in resuspension for aggregated and nonaggregated fractions
a slight increase of SPM and POC. are shown in Figures 4 and 5 for discharge levels q1 and q2,
[37] In such situations, where the bed shear stress exceeds respectively. At discharge q1, predicted total SPM/POC
a critical value high enough to mobilize consolidated concentrations (as the sum of the concentrations ci) reflected

7 of 12
19447973, 2006, 10, Downloaded from https://agupubs.onlinelibrary.wiley.com/doi/10.1029/2005WR004486 by Sichuan University, Wiley Online Library on [04/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
W10424 BUNGARTZ ET AL.: SUSPENDED SEDIMENT DYNAMICS W10424

Figure 5. Comparison of predicted SPM and POC behavior along the sampling reach with and without
aggregation at discharge level q2 (all values at 30 cm below surface). (a and b) Settling velocity fractions.
Symbols for first fraction are as follows: circles, measured; thin dashed line, nonaggregated; thick dashed
line, aggregated. Symbols for the second fraction are as follows: circles, measured; thin dotted line,
nonaggregated; thick dotted line, aggregated. Symbols for the third fraction are as follows: squares,
measured, thin dot-dashed line, nonaggregated, thick dot-dashed line, aggregated. (c and d) Total
concentrations. Symbols are as follows: diamonds, measured values; thin solid line, calculated as the sum
of the fractions, nonaggregated, thick solid line, calculated as the sum of the fractions, aggregated. (e and f )
Mean settling velocity vsm calculated from equation (4) for nonaggregated (thin solid line) and aggre-
gated (thick solid line). In all graphs, lines represent calculated values, and symbols represent measured
values.

the decline of measured values only when particle interac- it is apparent that the mass weighted averaged settling
tion was allowed (Figures 4c and 4d). Modeling which did velocity vm
s , calculated using the relation
not take particle formation processes into account, resulted
in lower sediment deposition fluxes than the modeling that X
3 X
3
considered aggregation due to particle interaction. This vm
s ¼ ci  vis = ci ; ð4Þ
i¼1 i¼1
result is consistent with the results presented by Ongley et
al. [1992] and Nicholas and Walling [1996]. For SPM,
longitudinally averaged deposition rates of 1.0 104 and behaves very differently when aggregation processes are
2.5 104 mg dw m2 s1 were predicted for nonaggre- assumed. In the case of nonaggregating fractions, vm s
gated and aggregated fractions respectively; corresponding decreases downstream; however, when the aggregation is
rates for POC were 1.4 105 and 4.1 105 mg C m2 s1. allowed for, vm s increases markedly for a distance of about
Consequently, deposition flux of POC for aggregated frac- 2000 m, then approaches a constant (equilibrium) value (see
tions was predicted to be about 3 times higher than for Table 2). The dynamic behavior of vm s within the distance
nonaggregated fractions. Figures 4a and 4b illustrate the of 2000 m corresponds to a response time of aggregation
very different downstream behavior of the three fractions 1/Q(i,j) = 3.1 h. This response time was identified numeri-
when particle interaction was considered. Simulations dem- cally as the most suitable value in fitting model predictions
onstrate that the predicted higher deposition rates were to SPM experimental data.
caused by the formation of aggregates having relatively [39] For discharge q1, any model parameters (in particu-
high settling velocities. Furthermore, from Figures 4e and 4f lar the partitioning coefficients Kij and the response time of
aggregation) used for simulation of SPM fraction behavior
8 of 12
19447973, 2006, 10, Downloaded from https://agupubs.onlinelibrary.wiley.com/doi/10.1029/2005WR004486 by Sichuan University, Wiley Online Library on [04/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
W10424 BUNGARTZ ET AL.: SUSPENDED SEDIMENT DYNAMICS W10424

were also applied to simulate the transport of POC fractions. particular when attached to surfaces like biofilms or macro-
As shown in Figure 4, these parameters proved to be aggregates), were shown to produce slime films [Decho,
suitable, not only for SPM, but also to reflect the POC 1990]. Bacterial activity may release large amounts of very
transport by the numerical model. This finding gives reason small sticky fibrillar material providing bridging mechanism
to the assumption of a close linkage between SPM aggre- between floc components, and thus may directly control the
gate dynamics and alterations in the settling velocity distri- formation and properties of aggregates [Liss et al., 1996].
bution of POC. [43] Overall, field data and model predictions highlight
[40] Simulation results of the transport of SPM/POC the significance of aggregation for transport, deposition,
fractions for the high discharge q2 are depicted in Figure 5. and retention of POC in rivers. Transport of terrestrial
The increasing SPM values suggest that under the prevailing carbon into rivers, and eventually into the ocean, con-
hydraulic conditions, a net resuspension occurs. Even though stitutes an important link in the global carbon cycle, but
it is quite possible that resuspension of particles from all frac- the POC flux is likely tempered by sedimentation losses.
tions occurs, our numerical experiments provided the best fit About 45% of the organic carbon entering the rivers
to experimental data when only material of the <1.25 cm h1 globally, is transported to the ocean, 25 –30% is oxidized
fraction was allowed to resuspend. Rates of 7.6 104 mg within the system, and 25– 30% is stored in the system as
dw m2 s1 and 1.2 104 mg C m2 s1 were identified sediment [Sarin et al., 2002]. Beside the global aspect of
numerically for the resuspension of SPM and POC respec- organic carbon fluxes, riverine POC transport is important
tively. Consequently, the percent POC present in resuspended in distributing energy and associated nutrients at multiple
sediment is about 15.8%. Accounting for the model predicted spatial and temporal scales, and is therefore a vital
deposition rates, net resuspension rates were 4.3 104 mg measure of trophic linkages between river segments.
dw m2 s1 and 0.67 104 mg C m2 s1 for SPM and Explicit information regarding POC transport distance is
POC respectively. Predicted total SPM and POC concen- required to quantitatively asses the strength of longitudi-
trations show only small deviations from the measured nal linkages in river ecosystems, and to estimate fluxes of
values, and virtually no difference regardless of aggregation organic matter between suspended and benthic POC pools
(Figures 5c and 5d). This is accounted for by the opposite [Minshall et al., 2000]. Our results reveal that vertical
trend of the lower than 1.25 cm h1 settling fraction, and the fluxes of organic carbon can be highly variable tempo-
fraction with settling velocities higher than 30 cm h1 rally, due primarily to temporal variations in river turbu-
(Figures 5a and 5b). lence (discharge). Model calculated deposition rates also
[41] However, as already demonstrated for discharge q1, show that POC deposition in fluvial systems may poten-
simulations reveal a different transport behavior of the tially be enhanced by particle aggregation.
fractions when sediment was nonaggregated and aggregated. [44] Comparing the alteration of settling velocity distri-
Independent of discharge level, simulation show that in the bution along the examined river reach, a different fraction
case of noninteracting sediment, the shares of the fraction behavior is shown at both discharges (Figure 3). Response
with low settling velocities (<1.25 cm h1) are overesti- time of aggregation became considerably shorter at higher
mated by the model, whereas the shares of the third fraction flow velocity/fluid shear (Table 3). For approximately
(settling velocity of >30.0 cm h1) are underestimated. uniform flow conditions along a river reach, we assume
Figures 5e and 5f depict the changes in the mean settling that depending on the prevailing shear characteristics, a
velocity vm s at discharge q2. Owing to the higher turbulence definite aggregate settling velocity distribution develops.
intensity, an equilibrium between the fractions (indicated by Under the influence of turbulence forces, there is a contin-
constant vm s ) is already established after a response time of uous process of aggregation and breakup, resulting in a
aggregation of about 0.44 h. This measure corresponds to a dynamic equilibrium of the aggregates (strength, size,
traveling distance of 1200 m. Deviating from the proce- density and settling velocity) after a particular response
dure applied to discharge q1, partitioning coefficients used time of aggregation [Van Leussen, 1997; Manning and
for SPM simulations had to be modified in order to properly Dyer, 1999]. It is hereby assumed that larger flocs are
reflect POC field data by the numerical model. Although formed in zones of lower shear with settling velocities high
SPM and POC fractions exhibit a consistent qualitative enough to settle toward the river bed. There, the strongest
longitudinal behavior (Figures 5a and 5b), a disparity in lift and shear forces occur, and only those flocs that are
the partitioning coefficients Kij = ceqj/ceqi (ratios of equilib- strong enough, settle through the region of maximum shear
rium concentrations in (3), governing the mass fluxes Fi,j) stresses to the bed. If the floc integrity is less than the
of SPM and POC necessitates the use of different model ambient turbulent shear stress, the floc will fracture, and the
parameters. resulting smaller units (with lower settling velocities) are
[42] We can only speculate about the reasons for the lifted back to participate again in the aggregation process
different association of POC with settling velocity fractions. [Metha and Partheniades, 1975]. Thus turbulent bed shear
Diverse hydraulic conditions differently influence the for- stress can impose a maximum floc size on the suspended
mation of aggregates, and hence the distribution of POC to floc population.
SPM settling velocity fractions. Seasonality can also control [45] The ‘‘Kolmogorov microscale of turbulence’’ Lk =
aggregate development and stability. For instance, the (n 3 e1)1/4, as a measure for the smallest eddies in a
occurrence of exopolymeric substances, which are known turbulent patch in a river (where n (m2 s1) is kinematic
to be conducive to aggregation due to their cohesiveness, is viscosity and e (m2 s3) is dissipation rate of kinetic
closely related to growth stage and species composition of energy), is assumed to be a good reference for the maximum
phytoplankton [Grossart and Simon, 1997; Passow, 2002]. possible floc size, which in turn may influence the settling
However, not only phytoplankton, but also bacteria, (in velocity spectrum of suspended particles [Van Leussen,

9 of 12
19447973, 2006, 10, Downloaded from https://agupubs.onlinelibrary.wiley.com/doi/10.1029/2005WR004486 by Sichuan University, Wiley Online Library on [04/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
W10424 BUNGARTZ ET AL.: SUSPENDED SEDIMENT DYNAMICS W10424

are also well reflected by the model at the transect T2. In


Figure 6, model-calculated mass fluxes Fi,j between the ith
and jth SPM/POC fraction are depicted for discharge
level q1. Simulations show high transfer rates from the
lower than 1.25 cm h1 fraction directly to the >30 cm h1
fraction within the distance of 1000 m. In contrast, the mass
transfer from the <1.25 cm h1 fraction to the second
fraction (1.25 cm h1 < 2nd fraction < 30.0 cm h1) is
comparatively low over the total distance, whereas the mass
transfer flux from the second fraction to the >30 cm h1
fraction is more intensive within the first 500 m.
Corresponding to the similar longitudinal behavior of
SPM and POC fractions, fluxes Fi,j shown in Figure 6 also
exhibit this similarity.
[48] The response times of aggregation and median
settling velocities under equilibrium conditions for the
examined river reach at four turbulence levels, represented
by their mean bed shear stresses (data from this study and
from Bungartz and Wanner [2004]), are given in Figure 7.
The numerically identified response time of aggregation
declines with increasing turbulence, that is to say the higher
the turbulence, the faster an equilibrium between the frac-
tions is established. Although our analysis is based on only
a few data sets, the behavior of median settling velocity is in
agreement with the hypotheses that settling velocity is
largely controlled by turbulent shear. Manning [2004]
Figure 6. Model calculated mass fluxes Fi,j between showed that the settling velocity of flocs formed in an es-
the ith and jth SPM/POC fraction at discharge level of tuary, rise with increasing shear stress until a limiting value
6.8 m3 s1 (q1). was reached. Accordingly, turbulence is to some extent con-
ducive to particle aggregation, resulting in larger particles
with higher settling velocity (increasing median values up
1997]. This described quantitative effect of turbulence on to a particular bed shear stress, as shown in Figure 7a).
the aggregation process and deposition behavior of cohesive
sediment is also referred to as ‘floc recycling’ and may have
implications on the modeling of these processes [Droppo,
2004].
[46] Our simulations confirm that the bed shear stress of
0.097 N m2 at discharge q1 is below a threshold value
(determined by the strength of the flocs) for deposition,
resulting in a net deposition rate for SPM and sediment-
associated POC. Beyond a critical bed shear stress for
erosion (discharge q2, bed shear stress of 1.76 N m2),
only extremely shear resistant flocs with high settling
velocities are still deposited, but simultaneous resuspension
of other fractions leads to a net resuspension rate. Consid-
ering that under the conditions of high discharge q2, flocs
will be destroyed in the zone of high shear, the numerically
identified net resuspension rates for SPM and POC will
comprise not only resuspended bed material but also frag-
ments of flocs which get broken up as they approach the
bed. However, our model is not designed to differentiate
between these two constituents. Simulations only reflect the
fraction behavior near the water surface, which to a great
extent is controlled by the breakup of flocs under shear
stresses applied near the bed.
[47] As we noted above, the mass fluxes Fij were adjusted
to get accurate simulations of the settling velocity distribu-
tion at the transect station T3/T3*. Starting with the mea-
sured initial settling velocity distribution at the inflow
transect T1, the modified distribution occurred after a Figure 7. (a) Median settling velocity and (b) response
certain traveling distance, which was dependent on the time of aggregation versus mean bed shear stress, obtained
response time of aggregation. Model predictions show that from two studies at the river reach downstream from the
measured absolute fraction concentrations (Figures 4 and 5) outlet of Lake Neuendorfer See.
10 of 12
19447973, 2006, 10, Downloaded from https://agupubs.onlinelibrary.wiley.com/doi/10.1029/2005WR004486 by Sichuan University, Wiley Online Library on [04/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
W10424 BUNGARTZ ET AL.: SUSPENDED SEDIMENT DYNAMICS W10424

A further increase in shear (beyond a critical value) may re- [55] Acknowledgments. We would like to thank Antje Lüder, Marcus
Schulz, and Axel Siegemund for their assistance with field work. Grit
sult in the breakup of flocs in fragments with lower settling Siegert is thanked for performing POC analysis. Two anonymous reviewers
velocity (indicated by the low median value at high shear are thanked for their constructive criticism and many helpful suggestions
stress in Figure 7a). for improving the manuscript. We are indebted to Sarah Poynton for final
corrections to the manuscript.

7. Conclusions References
[49] Aggregate dynamics have significant consequences Alldredge, A. L., U. Passow, and B. E. Logan (1993), The abundance and
for the transport of suspended sediment and sediment- significance of a class of large, transparent organic particles in the ocean,
Deep Sea Res., Part I, 40, 1131 – 1140.
associated organic carbon in rivers. The present study has Barth, J. A. C., J. Veizer, and B. Mayer (1998), Origin of particulate organic
employed both field data on the behavior of SPM and POC carbon in the upper St. Lawrence: Isotopic constraints, Earth Planet. Sci.
settling velocity fractions, and numerical modeling, to Lett., 162, 111 – 121.
establish the role of aggregation in the continual alteration Bungartz, H., and S. C. Wanner (2004), Significance of particle interaction
of hydrodynamic transport properties of SPM and its to the modelling of cohesive sediment transport in rivers, Hydrol. Pro-
cesses, 18, 1685 – 1702.
implications for POC transport. Bungartz, H., B. Shteinman, M. Thiele, and A. Parparov (2000), Modelling
[50] Field experiments showed the effect of different of flow and suspended sediment transport in the Jordan River, Arch.
turbulent shear on aggregate formation, and its effect on Hydrobiol., 55, 283 – 299.
particle settling velocity frequency distribution, in a lake Bura, R., M. Cheung, B. Liao, J. Finlayson, B. C. Lee, I. G. Droppo,
G. G. Leppard, and S. N. Liss (1998), Composition of extracellular poly-
outlet river at two markedly different levels of river dis- meric substances in the activated sludge floc matrix, Water Sci. Technol.,
charge. Our data suggest that aggregation results in the 37, 325 – 333.
formation of particles with settling velocities high enough to Burban, P.-Y., Y.-J. Xu, J. McNeil, and W. Lick (1990), Settling speeds of
settle toward the river bed. Furthermore, field data indicate a flocs in freshwater and seawater, J. Geophys. Res., 95, 18,213 – 18,220.
Celik, I., and W. Rodi (1988), Modeling suspended sediment transport in
close interrelationship of particle formation processes and non-equilibrium situations, J. Hydraul. Eng., 114, 1157 – 1191.
POC deposition fluxes. Decho, A. W. (1990), Microbial exopolymer secretions in ocean environ-
[51] A hydrodynamic and sediment transport model has ments – their role (s) in food webs and marine processes, Oceanogr. Mar.
been applied to simulate transport and sedimentation Biol., 28, 73 – 153.
behavior of SPM/POC settling velocity fractions. The ag- Droppo, I. G. (2001), Rethinking what constitutes suspended sediment,
Hydrol. Processes, 15, 1551 – 1564.
gregation module we used represents the interaction of Droppo, I. G. (2004), Structural controls on floc strength and transport,
multiple SPM/POC fractions, with corresponding multiple Can. J. Civ. Eng., 31, 569 – 578.
deposition behavior. The mass fluxes between the three Droppo, I. G., and E. D. Ongley (1994), Flocculation of suspended sedi-
settling velocity fractions could be quantified by comparing ment in rivers of southeastern Canada, Water Res., 28, 1799 – 1809.
Droppo, I. G., D. E. Walling, and E. D. Ongley (2000), The influence of
field data of settling velocity frequency distributions with floc size, density and porosity on sediment and contaminant transport, in
predicted numerical results. Simulation results show the The Role of Erosion and Sediment Transport in Nutrient and Contami-
capability of the model to reliably predict SPM/POC trans- nant Transfer, edited by M. Stone, IAHS Publ., 263, 141 – 147.
port and vertical exchange processes at the river bed. Eisma, D. (1993), Suspended Matter in the Aquatic Environment, Springer,
[52] However, because of the close linkage between floc New York.
Eisma, D., and A. Li (1993), Changes in suspended-matter floc size
strength and sediment transport, there is a great need for during the tidal cycle in the Dollard estuary, Neth. J. Sea Res., 31,
more intensive study of floc dynamics under varying shear 107 – 117.
rates during their journey through the water column Gomez, B., N. A. Trustrum, D. M. Hicks, K. M. Rogers, M. J. Page, and
[Droppo, 2004]. Although the effect of shear on floc K. R. Tate (2003), Production, storage, and output of particulate organic
carbon: Waipaoa River basin, New Zealand, Water Resour. Res., 39(6),
formation is quite well understood, in contrast, the breakup 1161, doi:10.1029/2002WR001619.
of floc under applied shear stress is an almost unexplored Grossart, H.-P., and M. Simon (1997), Formation of macroscopic organic
field of flocculation research. aggregates (lake snow) in a large lake: The significance of transparent
[53] Generally, the predictive capability of models of exopolymer particles, phytoplankton and zooplankton, Limnol. Ocea-
nogr., 42, 1651 – 1659.
cohesive sediment and sediment-associated material trans- Ittekkot, V. (1988), Global trends in the nature of organic matter in river
port is limited. Because of the inherent nature of cohesive suspensions, Nature, 332, 436 – 438.
sediment and site specificities, an adequate database against Jackson, G. A., and A. B. Burd (1998), Aggregation in the marine envir-
which cohesive sediment models could be validated does onment, Environ. Sci. Technol., 32, 2805 – 2814.
not exist [Teisson, 1997]. Nevertheless, existing models Karickhoff, S. W. (1981), Semi-empirical estimation of sorption of hydro-
phobic pollutants on natural sediments and soils, Chemosphere, 10,
represent valuable exploratory tools allowing the examina- 833 – 846.
tion of the interaction between model components, as well Köhler, J. (1994), Origin and succession of phytoplankton in a river – lake
as the sensitivity of processes to model parameters. system (Spree, Germany), Hydrobiologia, 289, 73 – 83.
[54] The results of the present study contribute to a small Kozerski, H. P., J. Köhler, and G. Schellenberger (1991), Transport of
particulate nutrients and pollutants in the lowland river Spree, in Sedi-
but growing set of supporting data on the effect of aggre- ment and Stream Water Quality in a Changing Environment: Trends and
gation on the transport properties of POC in rivers. Our Explanation, edited by N. E. Peters and D. E. Walling, IAHS Publ., 203,
preliminary results will hopefully encourage others to tackle 359 – 366.
a number of critical issues on river carbon transport, such as Kranck, K. (1973), Flocculation and suspended sediment in the sea, Nature,
improvement of estimation of organic carbon flux in rivers, 246, 348 – 350.
Lau, Y. L., and B. G. Krishnappan (1994), Does reentrainment occur during
in-river POC deposition and storage, remobilization of POC cohesive sediment settling, J. Hydraul. Eng., 120, 236 – 244.
deposits, as well as the model-based identification of in- Leppard, G. G., and I. G. Droppo (2005), Overview of flocculation pro-
river processes such as biological utilization or phytoplank- cesses in freshwater ecosystems, in Flocculation in Natural and Engi-
ton dynamics. neered Environmental Systems, edited by I. G. Droppo et al., pp. 25 – 46,
CRC Press, Boca Raton, Fla.

11 of 12
19447973, 2006, 10, Downloaded from https://agupubs.onlinelibrary.wiley.com/doi/10.1029/2005WR004486 by Sichuan University, Wiley Online Library on [04/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
W10424 BUNGARTZ ET AL.: SUSPENDED SEDIMENT DYNAMICS W10424

Lick, W. (1982), Entrainment, deposition, and transport of fine-grained Prochnow, D., C. Engelhardt, and H. Bungartz (1996), On the settling
sediments in lakes, Hydrobiologia, 91, 31 – 40. velocity distribution of suspended sediments in the Spree River, Arch.
Liss, S. N., I. G. Droppo, D. T. Flannigan, and G. G. Leppard (1996), Floc Hydrobiol., 47, 389 – 400.
architecture in wastewater and natural riverine systems, Environ. Sci. Ramseier, R. O., C. Garrity, D. G. Parsons, and P. Koeller (2000), Influence
Technol., 30, 680 – 686. of particulate organic carbon sedimentation within the seasonal sea-ice
Manning, A. J. (2004), The observed effects of turbulence on estuarine regime on the catch distribution of Northern Shrimp (Pandalus borealis),
flocculation, J. Coast. Res., 41, 90 – 104. J. Northw. Atl. Fish. Sci., 27, 35 – 44.
Manning, A. J., and K. R. Dyer (1999), A laboratory examination of floc Rodi, W. (1993), Turbulence Models and Their Application in Hydraulics—A
characteristics with regard to turbulent shearing, Mar. Geol., 160, 147 – State-of-the Art Review, Int. Assoc. for Hydraul. Res., Delft, Netherlands.
170. Rother, A., and J. Köhler (2005), Formation, transport and retention of
McCave, I. N., and J. P. M. Syvitski (1991), Principles and methods of aggregates in a river-lake-system (Spree, Germany), Int. Rev. Hydrobiol.,
geological particle size analysis, in Principles, Methods, and Application 90(3), 241 – 253, doi:10.1002/iroh.200410777.
of Particle Size Analysis, edited by J. P. M. Syvitski, pp. 3 – 21, Cam- Sarin, M. M., A. K. Sudheer, and K. Balakrishna (2002), Significance of
bridge Univ. Press, New York. riverine carbon transport: A case study of a large tropical river, Godavari
Metha, A. J., and E. Partheniades (1975), An investigation of the deposi- (India), Sci. China, Ser. C, 45, suppl. S, 97 – 108.
tional properties of flocculated fine sediment, J. Hydrol. Res., 92(C13), Sobek, A., O. Gustafsson, S. Hajdu, and U. Larsson (2004), Particle-water
361 – 381. partitioning of PCBs in the photic zone: A 25 – month study in the open
Minshall, G. W., S. A. Thomas, J. D. Newbold, M. T. Monaghan, and C. E. Baltic Sea, Environ. Sci. Technol., 38, 1375 – 1382.
Cushing (2000), Physical factors influencing fine organic particle trans- Teisson, C. (1997), A review of cohesive transport models, in Cohesive
port and deposition in streams, J. N. Am. Benthol. Soc., 19, 1 – 16. Sediments, edited by N. Burt et al., pp. 367 – 381, John Wiley, Hoboken,
Nicholas, A. P., and D. E. Walling (1996), The significance of particle N. J.
aggregation in the overbank deposition of suspended sediments on river Tsai, C. H., S. Iacobellis, and W. Lick (1987), Flocculation of fine-grained
floodplains, J. Hydrol., 186, 275 – 293. lake sediments due to a uniform shear stress, J. Great Lakes Res., 13,
Oden, S. (1915), Eine neue Methode zur mechanischen Bodenanalyse, Int. 135 – 146.
Mitt. Bodenk., 5, 257 – 311. Van Leussen, W. (1997), The Kolmogorov microscale as a limiting value
Ongley, E. D., B. G. Krishnappan, I. G. Droppo, S. S. Rao, and R. J. for the floc sizes of suspended fine-grained sediments in estuaries, in
Maguire (1992), Cohesive sediment transport: Emerging issues for toxic Cohesive Sediments, edited by N. Burt et al., pp. 45 – 62, John Wiley,
chemical management, Hydrobiologia, 235/236, 177 – 187. Hoboken, N. J.
Passow, U. (2002), Transparent exopolymer particles (TEP) in aquatic en- Voice, T. C., and W. J. Weber (1983), Sorption of hydrophobic compounds
vironments, Prog. Oceanogr., 55, 287 – 333. by sediments, soils and suspended-solids. 1. Theory and Background,
Passow, U., R. F. Shipe, A. Murray, D. K. Pak, M. A. Brzezinski, and A. L. Water Res., 17, 1433 – 1441.
Alldredge (2001), The origin of transparent exopolymer particles (TEP) Wanner, S. C., and M. Pusch (2000), Use of fluorescently labelled Lyco-
and their role in the sedimentation of particulate matter, Cont. Shelf Res., podium spores as a tracer for suspended particles in a lowland river, J. N.
21, 327 – 346. Am. Benthol. Soc., 19, 648 – 658.
Prochnow, D., H. Bungartz, and C. Engelhardt (1991), Modeling and si-
mulation of contaminant transport and sedimentation processes in fluvial
systems, in Proceedings of the 13th IMACS World Congress on Compu- 

tation and Applied Mathematics, edited by R. Vichnevetsky and J. J. H. H. Bungartz, C. Engelhardt, and A. Krüger, Department of Ecohydraulics,
Miller, pp. 1970 – 1971, Elsevier, New York. Leibniz Institute of Freshwater Ecology and Inland Fisheries Berlin, PO Box
850 119, D-12561 Berlin, Germany. (bungartz@igb-berlin.de)

12 of 12

You might also like