Volume 235, Issues 1–4, 20 December 2006, Pages 177–192

Proceedings of the 6th International Congress on Tidal Sedimentology (Tidalites 2004)

Proceedings of the 6th International Congress on Tidal Sedimentology (Tidalites 2004)

Edited By Jesper Bartholdy and Erik P. Kvale

Variability in cohesive sediment settling fluxes: Observations under different estuarine tidal conditions


Abstract

The mass settling flux, which is defined as the product of the concentration and the settling velocity, is of prime importance with respect to both stratified and well mixed estuarine conditions. The determination of these fluxes (for applied modelling purposes) in high energy tidal estuarine environments, is very problematic. This is because the muddy sediments which dominate in estuaries, flocculate producing a variety of sizes and settling velocities, and this flocculation process is not understood well enough to be fully described theoretically. By drawing on examples of floc spectra acquired in-situ using the INSSEV system, this study explains how mass settling fluxes in the near-bed region can vary by three or four orders of magnitude in meso- and macro-tidal estuaries throughout a single tidal cycle. A floc population representative of dilute suspension conditions on a neap tide, indicated only 35% of the floc mass was macroflocs (> 160 μm). However, the macrofloc settling velocity = 2.4 mm s− 1; three times faster than the microflocs, which meant the former fraction contributed 57% of the 205 mg m− 2s− 1 settling flux. Both highly concentrated (4–6 g l− 1) and very turbulent spring tide conditions (τ > 1.6 N m− 2) produced a bi-modal distribution in terms of the floc size and dry mass. With the former, 54% of the mass was contained within the 240–480 μm size fraction, with a further 25% of the dry floc mass in the flocs over 480 μm in diameter. These large flocs had settling velocities between 4–8 mm s− 1, which meant 99.5% of the settling flux (33.5 g m− 2s− 1) was accredited to the macroflocs. The high turbulence environment saw the dry floc mass distribution shift 60:40 in favour of the microflocs. The microfloc settling velocity was 1.45 mm s− 1, 0.35 mm s− 1 faster than the larger macrofloc fraction. In terms of the total mass settling flux, 0.9 g m− 2 s− 1, this translates into the microflocs contributing 70% during high turbulence. At slack water the flux only reached 12 mg m− 2 s− 1 and macrofloc growth was mainly attributed to differential settling. Continuous floc observations made over a complete tidal cycle revealed that the asymmetrical distribution of the tidal energy generated throughout the spring conditions in the Tamar estuary demonstrated a distinct control on the flocculation process. The less turbulent ebb produced 86% of the total tidal cycle mass settling flux, of which only 8% of the settling flux was outside the turbidity maximum. An attempt to simulate these large settling fluxes by using a constant settling rate of 0.5 mm s− 1, under-estimated the tidal cycle settling flux by 78%, with less than 15% of the total flux being estimated during the advection of the turbidity maximum on the ebb. In contrast, using a faster constant settling velocity parameter of 5 mm s− 1, (representative of the macrofloc fraction), resulted in a mass flux over-estimate of 116% for the tidal cycle duration.

Keywords

  • mass settling flux;
  • flocculation;
  • turbulent shear stress;
  • settling velocity;
  • effective density;
  • INSSEV instrument

1. Introduction

Models of sediment transport are widely used as dredging management tools in estuarine locations. However an accurate representation of the vertical sediment settling fluxes is very problematic. This is due to the cohesive nature of the muddy sediments which dominate estuarial locations. Observations of fine sediment suspensions suggest that, except for extremely high energetic conditions, most of the sediment mass occurs as flocs (Kranck and Milligan, 1992). These cohesive sediments can flocculate to form a spectra of aggregates known as flocs. Flocs are less dense, but faster settling than their constituent particles. As flocs grow their effective density generally decrease, but their settling rates rise. Flocculation is a dynamically active process which readily reacts to changes in turbulent hydrodynamic conditions (Manning, 2001 and Manning, 2004a). The flocculation of particles is a function of the mechanisms which bring the particles into contact, e.g. differential settling or turbulence (Manning and Dyer, 1999) and the mechanisms that make them stick together, e.g. salinity related electro-static charging or organic matter content (van Leussen, 1988). In addition to increasing particle collisions, turbulent shear may also break up aggregates (McCave, 1985), and this is further complicated by concentration gradients which can form in the near-bed region throughout a tidal cycle (Dyer et al., 2004).

The processes of aggregation and disaggregation are still not understood well enough to be fully described theoretically, and at the moment predictions tend to rely on empirical generalisations. Until recently, a lack of reliable floc measurements together with those of hydrodynamic turbulence, limited studies of the complex interactions between the factors affecting flocculation (e.g. shear, salinity, organic content and concentration) and floc characteristics. Laboratory experiments do not reliably represent field situations because of the difficulty of reproducing the chemical, physical and biological processes involved, and in-situ measurements have historically been unreliable because of floc disruption when sampling ( van Leussen, 1988). However, the advent of video floc devices (e.g. van Leussen and Cornelisse, 1994, Hill et al., 1998 and Mikkelsen et al., 2004), in particular the unique INSSEV instrument ( Fennessy et al., 1994 and Manning and Dyer, 2002) which was developed at the University of Plymouth (UK), and when combined with an array of miniature high frequency velocimeters, INSSEV has provided a means of accurately determining time series of both the spectral distribution of the floc dry mass and settling velocities, directly from within a turbulent estuarine water column. INSSEV has observed low density (∼ 30 kg m− 3) macroflocs over 1.5 mm in diameter which have displayed settling velocities of 3–25 mm s− 1 ( Manning and Bass, 2004).

In contrast to floc particle sizers, the additional measurement of individual floc settling rates means that INSSEV can provide reliable estimates of floc effective density by using a modified Stokes' Law relationship. A knowledge of floc effective density is very important in the calculation of vertical settling fluxes (Manning, 2004b). As flocs increase in diameter they become more porous (> 90–95%); since their voids are filled with interstitial water; the higher order flocs are less dense than the lower order microflocs. Very few direct quantitative studies have been conducted on floc effective density variations. Floc fragility has precluded the direct measurement of floc density. Also the rheological properties of suspended particulate matter are governed by volume concentrations, as opposed to mass concentrations (Dyer, 1989).

A flux which is of prime importance with respect to both stratified and well mixed estuarine conditions, is the mass settling flux (MSF), which is defined as the product of the suspended particulate matter (SPM) concentration and the settling velocity (Manning, 2004b). By drawing on examples of floc spectra acquired in-situ using the INSSEV system, this paper will illustrate in a semi-quantitative manner how these settling fluxes in the near-bed region can vary by three or four orders of magnitude in meso-/macro-tidal estuaries under different tidal conditions ( Bass et al., 2006 and Manning et al., 2006). The examples presented were obtained from experiments conducted in the Gironde Estuary, France and the Tamar Estuary, England, and cover a variety of conditions including: spring and neap tides, slack water and times when concentrations and turbulent shear stresses were high.

2. Method

2.1. Data acquisition

Near-bed flocculation dynamics during neap tides were studied in the lower reaches of the Gironde estuary during June 1999 as part of the European Commission (EC) funded SWAMIEE project international field experiment, referred to as SWAMGIR1 ( Manning et al., 2004). The sampling location was located 1 km from the shore at Le Verdon (3 km from the mouth) and this was within the vicinity of one of the localised mud deposits situated between Bordeaux and Talais ( Jouanneau and Latouche, 1981). These 2–3 m deep patches of cohesive sediment are significant to the estuary's sediment transport regime, as they move either landward or seaward in response to run-off.

Measurements during spring tide conditions were made in the mesotidal Tamar estuary as part of the EC funded COSINUS project in June and September 1998 (Berlamont, 2002), and an NERC-funded experiment in April 2003 (Bass et al., 2006). The Tamar estuary, which is topographically dendritic in shape, is located in south-west England and has numerous meanders and wide mud flats exposed at low water. The Tamar has a catchment of approximately 800 km2, which is one hundred times smaller than the Gironde estuary. Data acquisition was conducted in a straight channel section in the upper Tamar estuary, within the tidal trajectory of the turbidity maximum (TM). For all experiments the instrumentation and data acquisition systems were similar and are described below.

Flocs, although stable in flowing turbulent water, easily break apart when sampled in response to additional shear created during acquisition (Eisma et al., 1997). Therefore, the floc populations presented in this paper were acquired at a nominal height of 0.5 m above the estuary bed, using the INSSEV — IN-Situ Settling Velocity-instrument (Fennessy et al., 1994). The INSSEV video system was used to measure floc size and settling velocity spectra. A unique feature of INSSEV is that the flocs are measured within a known volume of fluid, which permits the distribution of floc dry mass and floc porosity to be accurately estimated.

The sampling apparatus comprises two inter-connected chambers which allow flocs to be sampled directly from within a turbulent water column, whilst creating minimal disruption to the observed flocs. A high resolution monochromatic Pasecon tube miniature underwater video camera located in an aperture in the lower settling chamber wall viewed the settling flocs. The floc images were recorded by an analogue S-VHS video recorder which produces a practical lower resolution limit of 20 μm.

Detailed measurements of the near-bed hydrodynamical structure were gathered at an acquisition rate of 18 Hz (and low pass filtered at 5 Hz) using the POST — Profile Of Sediment Transport-system (Christie et al., 1997). The system consisted of four 2 cm diameter discus shaped electro-magnetic flow meters (EMCMs), and five optical back-scatter sensors (OBS). The EM sensors were arranged in pairs to measure the streamwise u, cross-channel v, and vertical w components of the flow, at distances 0.25, 0.5 m and 0.75 m above the bed (depending upon the experiment). Changes in water depths were monitored using a Druck 2 bar absolute pressure transducer.

Vertical profiles of salinity, temperature and suspension concentrations were obtained every 15 min using a Seabird Systems SBE 19-03 CTD together with a Downing OBS. Corresponding velocity profiles through the water column were obtained using a Valeport 108 mk III impeller current meter. Peristaltic pumped water samples were obtained from 0.1 and 0.5 m above bed every half hour. Filtered samples were used to verify OBS and INSSEV estimates of suspended mass concentration, and to estimate the percentage of organic matter content by loss-on-ignition tests.

INSSEV was mounted on a metal bedframe, whilst the POST sensors were attached to a vertical pole, which was positioned adjacent to the INSSEV sampling unit. Fig. 1 illustrates a typical instrumentation bed frame configuration awaiting deployment. To prevent vortex shedding occurring at the sampling orifice of the INSSEV stilling chamber, the rig was aligned to within ± 5° with the dominant streamwise flow. This was achieved through the use of either control ropes attached to the outside of the frame and/or a rear fin (depending upon the estuarine location) to control orientation during lowering. A digital 2-axis tilt sensor and a fluxgate compass provided a direct readout of rig positioning on the estuary bed. The rig sat on the bed for the entire duration, sampling data from within an Eulerian reference frame.

Full-size image (102 K)
Fig. 1. 

INSSEV, EMCM, ADV and OBS sensors mounted on the estuarine bed frame — (a) front view and (b) side view.

2.2. Data processing

INSSEV flocs images were measured manually after each deployment from the video monitor, using size calibrations to convert the floc image to real dimensions. The two orthogonal floc dimensions which describe the floc shape (Dx and Dy), were converted into spherical-equivalent floc sizes (D).

By assuming that each floc is settling within the viscous Reynolds region (i.e. when the particle Reynolds number, Re, < 0.5–1), the effective density (ρe) for each floc could be obtained by applying a Stokes' Law relationship:

equation1
View the MathML source
where: Ws is the floc settling velocity, μ the dynamic molecular viscosity and g is gravity. The effective density, also referred to as density contrast, is the difference between the floc bulk density (ρf) and the water density (ρw). The water density was calculated from measured salinity and water temperature data using the International Equation of State of Sea Water, 1980 ( Millero and Poisson, 1981). Stokes' Law assumes that the settling particle is spherical, and observations have shown that flocs are often by no means spherical. Some of the large flocs have a very irregular shape, and are often joined into stringers by threads of biological matter (e.g. Eisma, 1986 and Manning and Dyer, 2002). However, by direct measurement Alldredge and Gotschalk (1988) found that organic aggregates whose shape varied from spherical, including long comets, had settling rates very similar to those of nearly spherical particles. Similarly Gibbs (1985) showed that flocs from Chesapeake Bay had an average height:width ratios of to 1.6:1 which gave a spherical value of 0.91. Therefore, the implementation of a spherical equivalent diameter in Eq. (1) is valid.

Where the Re exceeded 0.5, but was less than 2, the Oseen modification ( Schlichting, 1968), as advocated by Brun-Cottan (1986) and Ten Brinke (1994), was applied to Eq. (1) in order to correct for the increased inertia created during settling. For instances where the Re was greater than 2, the Schiller (1932) formula provides the best results. A series of algorithms was then applied to the INSSEV data from which dry floc mass and settling flux distributions were computed ( Fennessy et al., 1997). This type of flux computational technique has also been applied successfully by: Syvitski et al. (1995), Hill et al. (1998), and Sternberg et al. (1999). Further details of the floc data processing are given in Manning (2004c).

Turbulent shear stress at the EMCM heights was estimated using the turbulent kinetic energy approach (e.g. Kim et al., 2000), which is less sensitive to sensor misalignment errors. Burst-averaged values of turbulent kinetic energy were calculated for each 227.5 s duration EMCM velocity data file.

3. Floc spectra

The first two floc spectra presented below are typically examples of average dilute suspension conditions (i.e. outside of the turbidity maximum zone) experienced during neap and spring tides when the flow was predominantly uniform and the lower region of the water column was generally well-mixed, with very little stratification present. The last three cases (3.3–3.5) will illustrate examples of floc populations present during the more extreme conditions experienced throughout a tidal cycle, including high concentration gradients, slack water and highly turbulent shear stress conditions, all of which can significantly affect the settling fluxes of cohesive sediments.

3.1. Neap tide flocs

Fig. 2 shows an example of a floc sample (G23-8) obtained during neap tides in the macro-tidal Gironde estuary (south-west France). Sample G23-8 was obtained 2 h and 10 min before high water, when the SPM concentration had risen to maximum of 139 mg l− 1 and the turbulent shear stress was 0.46 N m− 2. The floc sizes ranged from 23 μm up to 362 μm, and these settled at velocities spanning three orders of magnitude from 0.04–3.4 mm s− 1. The diagonal lines on Fig. 2a illustrate contours of effective density (calculated by Eq. (1)) and indicate that there are many flocs of different sizes, but with the same density. Also there are flocs present with the same settling velocity, but exhibiting a wide range of sizes and densities. For G23-8, floc effective densities ranged between 40–812 kg m− 3. The shape of the largest macroflocs reflected a comet in appearance, with height to width ratios of 1.5–2.5. An example is shown in Fig. 3 by a macrofloc observed 79 s into the record (settling at 2.85 mm s− 1) with a Dx = 300 μm and Dy = 500 μm, and a corresponding effective density of 48 kg m− 3.

Full-size image (61 K)
Fig. 2. 

Summary of floc characteristics for sample G23-8. Where (a) illustrates the relationships between floc size and settling velocities of individual flocs, with diagonal full lines showing the effective density (kg m− 3); (b) shows the size band distribution of SPM concentration; and (c) shows the mass settling flux spectrum.

Full-size image (26 K)
Fig. 3. 

A large low density “comet-shaped” macrofloc (sample G23-8, 79 s into record).

The flocs represented a total MSF of 205 mg m− 2s− 1. In terms of the SPM distribution (Fig. 2b), 86 microflocs (i.e. D < 160 μm) encompassed 65% of the floc mass, compared to just 20 individual macroflocs constituting the remaining 35%. However this translated into the macroflocs contributing 57% of the flux; a product of a macrofloc (i.e. D > 160 μm) settling velocity (Wsmacro) of 2.4 mm s− 1, which was three times faster than the Wsmicro. Similarly, the primary settling flux mode occurred in both the flocs 120–160 μm (Size Band (SB) 4) and 200–240 μm (SB6) in size; 40 mg m− 2s− 1 in each ( Fig. 2c). The settling characteristics and ranges demonstrated by sample G23-8 are representative of neap tide floc populations typically encountered in dilute suspension conditions outside a turbidity maximum zone in many western European estuaries ( Manning, 2004c).

3.2. Spring tide flocs

This spring tide example of a floc spectrum was measured during the COSINUS experiment in September 1998 in the Tamar Estuary (UK). Fig. 4 shows the record for INSSEV sample T21-1 which was collected on an ebb 2 h 10 min before low water. The particle concentration was much greater than the previous neap tide example, at just over 3 g l− 1, but with a similar shear stress (0.53 N m− 2) at the sampling height. As with the previous neap tide example, this sample was obtained outside of the turbidity maximum zone. Sample T21-1 constituted a total population of 750 individual flocs. There was a negative skewing of the population distribution across the size spectrum, with most aggregates having a near-spherical shape. Settling velocity distributions showed a very linear increase in settling rate, with each larger size band sub-group. The population was dominated by an abundance of macroflocs, with 77.6% of the mass contained by flocs in excess of 200 μm in size. The aggregates composing this section of the floc population became progressively less dense and more structurally porous, as their size increased. This tendency was shown by the respective values of effective density and porosity being 82 kg m− 3 and 93.4% for the SB6 (200–240 μm) flocs. In comparison, the data revealed that the size band mean effective density for the flocs exceeding 640 μm in diameter had reduced to 25 kg m− 3, whilst the porosity had risen to 97.3%. Furthermore, the size spectrum showed that nearly half of the particulate mass was represented by flocs over 320 μm in diameter. It may also be hypothesised that the addition of a total carbohydrate concentration of 9.6 mg l− 1 (Manning et al., this volume), which was 3.8 times higher than experienced during Tamar estuary neap tides, could have had a significant effect by assisting with inter-particular bonding as a result of the high collision frequency.

Full-size image (71 K)
Fig. 4. 

Summary of floc characteristics for sample T21-1. Where (a) illustrates the relationships between floc size and settling velocities of individual flocs, with diagonal full lines showing the effective density (kg m− 3); (b) shows the size band distribution of SPM concentration; and (c) shows the mass settling flux spectrum.

Although the spectrum showed a predominantly unimodal distribution of the SPM% (Fig. 4b), with a primary mode of 20.2% occurring at SB8 (320–400 μm), a much higher mass settling flux existed in the higher size bands despite their individually lower total mass content. For example, the largest size fraction (> 640 μm) had a sub-population of 14 flocs and this constituted 4% of the total particulate mass in suspension, but with the aggregates possessing a mean settling velocity of 6.5 mms− 1, a resultant MSF of 820 mg m− 2s− 1 was attained (for SB12). Comparing this to the primary modal flux of SB8, which was a quarter of the total flux or 2860 mg m− 2s− 1, this fraction's settling flux contribution was 3.5 times greater than that attributed to the higher order SB12 flocs. However this was achieved by SB8 having 9.5 times as many floc comprising its sub-population. The total mass settling flux for this sample was 11.3 g m− 2s− 1. To put this into context, this MSF value was 6.7 times greater than the value computed by the use of an estimated settling velocity of 0.5 mms− 1; a typical parameterised Ws value used in numerical model simulation of sediment transport (J. Spearman, pers. com). Further general comments on applied modelling issues relating to Ws parameterisation and how it influences the settling flux will be made in Section 4.

3.3. Flocs formed in Concentrated Benthic Suspensions (CBS)

CBS layers form in the near-bed region, typically during the advection of a turbidity maximum. A main feature of a CBS is the turbulence damping which can occur within the layer as a result of density stratification; this can have a significant influence on the flocculation. Unlike a fluid mud, which acts independently from the main flow, a CBS is still transported with the dominant current. The solids forming the CBS tend to be kept in suspension by turbulence, although the CBS layer can interact with the turbulent flow field. The CBS behaves as a Newtonian fluid, but with increased viscosity.

INSSEV sample T24-9 was acquired in September 1998 from within a Tamar estuary spring tide CBS layer of 5.6 gl− 1. The high concentration had restricted the turbulent energy to a shear stress of 0.36 Nm− 2 at 0.5 m above the bed. This transformed the flocs into a bi-modal population in terms of the floc size and dry mass distribution (Fig. 5). Although the scatterplot showed a sub-group of slow settling (Ws < 0.3 mm s− 1) flocs in the 30–70 μm size range (Fig. 5a), these only constituted 9% of the total floc population. The majority of the particulate mass, 54%, was contained within the larger sized aggregates of SB7-9 (240–480 μm), with a further 25% of the dry floc mass contained in the three largest size bands (SB10-12). This cluster of large flocs had settling velocities between 4–8 mm s− 1, which produced a resultant mean settling velocity for the fraction > 160 μm of 5.7 mm s− 1, which was an increase of over 2.4 mm s− 1 when compared to the earlier spring sample (i.e. T21-1). These macroflocs were on average 87% porous, with ρe predominantly < 100 kg m− 3.

Full-size image (71 K)
Fig. 5. 

Summary of floc characteristics for sample T24-9. Where (a) illustrates the relationships between floc size and settling velocities of individual flocs, with diagonal full lines showing the effective density (kg m− 3); (b) shows the size band distribution of SPM concentration; and (c) shows the mass settling flux spectrum.

Of the 33.5 g m− 2s− 1 total MSF, 99.5% was accredited to the macroflocs (Fig. 5c). To put these values into perspective, this floc size spectrum which ranged from 30–894 μm represented a mass settling flux three times larger than typical Tamar spring tides outside of the CBS (T21-1) (Fig. 4), and 168 times greater than the Gironde neap tides example (G23-8). This highlights a significant optimisation in the flocculation present in CBS layers conditions. The high percentage of floc mass contained by the macroflocs situated within the TM was also observed by Dyer et al. (2002), and Manning et al. (2006).