Seasonal variability of cohesive sediment aggregation in the bach dang–cam estuary, haiphong (vietnam

Three stations situated at key spots of the estuarine system were monitored during 24-h surveys corresponding to one spring tidal cycle: one station was located upstream on the Cam River

Seasonal variability of cohesive sediment aggregation

in the Bach Dang–Cam Estuary, Haiphong (Vietnam)

Jean-Pierre Lefebvre&Sylvain Ouillon&Vu Duy Vinh&

Robert Arfi&Jean-Yves Panché&Xavier Mari&

Chu Van Thuoc&Jean-Pascal Torréton

Received: 6 April 2011 / Accepted: 19 December 2011 / Published online: 15 January 2012

# Springer-Verlag 2012

Abstract In the Bach Dang–Cam Estuary, northern Vietnam,

mechanisms governing cohesive sediment aggregation were

investigated in situ in 2008–2009 As part of the Red River

delta, this estuary exhibits a marked contrast in hydrological

conditions between the monsoon and dry seasons The impact

on flocculation processes was assessed by means of surveys of

water discharge, suspended particulate matter concentration

and floc size distributions (FSDs) conducted during a tidal

cycle at three selected sites along the estuary A method was

developed for calculating the relative volume concentration

for the modes of various size classes from FSDs provided by

the LISST 100X (Sequoia Scientific Inc.) It was found that all

FSDs comprised four modes identified as particles/flocculi,

fine and coarse microflocs, and macroflocs Under the

influ-ence of the instantaneous turbulent kinetic energy, their

pro-portions varied but without significant modification of their

median diameters In particular, when the turbulence level

corresponded to a Kolmogorov microscale of less than

∼235 μm, a major breakup of flocs resulted in the formation

of particles/flocculi and fine microflocs Fluctuations in tur-bulence level were governed by seasonal variations in fresh-water discharge and by the tidal cycle During the wet season, strong freshwater input induced a high turbulent energy level that tended to generate sediment transfer from the coarser size classes (macroflocs, coarse microflocs) to finer ones (particles/flocculi and fine microflocs), and to promote a transport of sediment seawards During the dry season, the influence of tides predominated The turbulent energy level was then only episodically sufficiently high to generate transfer of sediment between floc size classes At low turbulent energy, modifications in the proportions of floc size classes were due to differential settling Tidal pumping produced a net upstream transport of sediment Associated with the settling of sediment trapped in a near-bed layer at low turbulent energy, this causes the silting up of the waterways leading to the harbour of Haiphong

Responsible guest editor: D Doxaran

J.-P Lefebvre (*):S Ouillon

IRD, Université de Toulouse, UPS (OMP), UMR 5566 LEGOS,

14 av Edouard Belin,

31400 Toulouse, France

e-mail: jean-pierre.lefebvre@ird.fr

V D Vinh:C Van Thuoc

Institute of Marine Environment and Resources (IMER),

Vietnam Academy of Science and Technology (VAST),

246 Danang Street,

Haiphong City, Vietnam

R Arfi

IRD, Université Aix-Marseille 2, UMR 6535 LOPB,

Centre d’Océanologie de Marseille,

Luminy,

13288 Marseille cedex 09, France

J.-Y Panché IRD, US 191 IMAGO,

BP A5,

98848 Nouméa cedex, New Caledonia

X Mari:J.-P Torréton

IRD, Université Montpellier II, UMR 5119 ECOSYM,

cc 093, Place Bataillon,

34095 Montpellier, France

Present Address:

X Mari Institute of Biotechnology, Environmental Biotechnology Laboratory,

18 Hoang Quoc Viet Street, Cau Giay, Hanoi, Vietnam

DOI 10.1007/s00367-011-0273-8

In Vietnam, the silting up of the Red River delta constitutes

a main concern for the authorities due to its particularly

negative impact on traffic in the country’s second biggest

harbour, Haiphong During the 1980s, the construction of

two dams across the Red River induced a reduction of

sediment inputs to the coast, which caused fast and locally

intense erosion in the bay of Haiphong Although the impact

of dam constructions has long been investigated worldwide

in terms of total suspended particulate matter concentration

(SPMC; e.g Uncles et al 1988; Vörösmarty et al 2003;

Mantovanelli et al.2004; Wolanski et al.2006; Scully and

Friedrichs 2007; Winterwerp 2011), less is known about

interrelations with floc size distributions (FSDs)

Flocculation processes depend on various factors

includ-ing the electric charge on particles (ζ potential), organic

matter content, suspended matter availability, and turbulent

shear rate (e.g Lunven and Gentien 2000; Lunau et al

2006; Mietta et al.2009) In estuaries, the high variability

in forcing (Verney et al.2009) and the impact of biological

factors make the behaviour of cohesive sediments even

more complex and still not well understood (Winterwerp

2011) Organic bindings such as those due to transparent

exopolymeric particles (TEPs; e.g Passow et al.2001; Wetz

et al 2009; Mari et al 2011) or dissolved exopolymeric

substances (e.g Bhaskar et al.2005) can generate

macro-flocs of various sizes and strengths The abundance of

mineral particles can affect the process of aggregation,

ex-cept for SPMC in the range 50–250 mg L−1(cf Milligan

and Hill1998) Turbulence affects flocculation processes by

increasing the collision frequency between aggregates, and

also by generating a shear stress at the surface of aggregates

that limits their size increase in the same order of magnitude

as the smallest turbulent eddies (van Leussen1997; Jarvis et

al.2005) Sudden disaggregation of flocs beyond a

thresh-old of turbulent intensity was found by Chen et al (2005) in

the Scheldt Estuary

Turbulence level and differential settling of aggregates of

various sizes and densities are usually thought to dominate

the aggregation and breakup processes (e.g van Leussen

1994; Winterwerp 2006; Kumar et al 2010) Lick et al

(1993) proposed a model where the median apparent

diam-eter of aggregates Dv is related to the product of the

turbulence-induced shear rate (G) and SPMC through a

power law Dv0 α (G·SPMC)β, where α and β are constants

Manning and Dyer (1999) compared this equation to a more

sophisticated formula and proved that it was reasonably

accurate

The governing action of tides on flocculation and

differ-ential settling has been identified in numerous estuaries of

temperate regions In the Tamar Estuary (UK), for example,

the balancing of re-suspension and differential settling at

high tide slack water is revealed by a close relationship between SPMC and Dv(Uncles et al.2010) In this estuary, moderate levels of turbulence promote collisions between flocs, and transfers from microflocs to macroflocs enhanced

by organic bindings During spring tide, a highly concen-trated benthic suspension layer is generated near the bed that contributes to dampening the turbulence within that layer The generation of coarser macroflocs results in a marked bimodality in floc size distribution (Manning2004)

In shallow estuaries, an asymmetry between ebb tide and flood tide caused by nonlinear tidal interaction and by astronomic tides can be observed Since the celerity of the tidal wave increases with increasing water depth, the tide propagates faster at high water than at low water This causes the shape of the tidal wave to distort as it moves landwards; the rise of the tide becomes faster than its fall and, consequently, the peak current is faster at flood tide than at ebb tide An increase in tidal level counterbalances the liquid budget, so that the overall balance of the tidal flow

is essentially nil This asymmetry results in a larger sedi-ment transport upstream at flood tide than downstream at ebb tide (Dronkers1986) This landward transport of sedi-ment caused by ‘tidal pumping’ has been described by Geyer et al (2001) for the Hudson River Estuary during spring tides At the site of the present study near Haiphong, the Bach Dang–Cam Estuary is under the influence of a diurnal tidal regime; therefore, this mechanism is likely to

be enhanced (Hoitink et al.2003)

In monsoon-dominated rivers, the freshwater discharge exhibits a marked seasonality that affects turbulence, salt stratification in the water column and tidal wave propaga-tion in the estuary SPMC can be related to turbulence through bed erosion, and to inputs from the catchment basins The balance between tidal propagation and freshwa-ter discharge defreshwa-termines the amplitude and direction of the current flow, and the level and advection of turbulence in the water column, both impacting on the transport and settling

of sediment (Dyer 1995) Water column stratification can constitute an impediment to the vertical advection of turbu-lence (Geyer 1993; Uncles and Stephens 1993; Jay and Musiak 1994; Peters 1997; Scully and Friedrichs 2003), and can prevent the advection of sediment across the fresh-water–saltwater interface, resulting in the trapping of sedi-ment in the upper freshwater or lower saltwater layers As an example, the Mekong and Red River estuaries experience similar seasonal forcing Solid discharge by the Mekong River (170×106metric tons year−1) is similar to that of the Red River (Milliman and Meade1983), both being charac-terized by silt-dominated material In the southern branch of the Mekong delta, the salt wedge is observed near the mouth

of the estuary during the wet season During the dry season, the tidal asymmetry increases and the salt wedge propagates further into the estuary The sediment load budget indicates

a tidally averaged flux to the sea of at least 95% during the

wet season (Wolanski et al.1996), and tidal pumping from

the coastal area to the estuary during the dry season, coupled

with a balancing of settling out at slack tide and

re-entrainment at higher current speeds In the Mekong River

delta, non-biological flocs are found only in brackish water

and remain non-flocculated in the freshwater layer This

results in a seaward transport of particles as ‘wash load’ in

the near-surface freshwater layer, and a tidally varying

transport of near-bed flocculated sediments No significant

difference in floc size was observed between the wet and dry

seasons, and examination of the aggregates confirmed their

non-biological origins (Wolanski et al.1998)

In the silt-dominated Fly River Estuary, Papua New

Guinea, smaller flocs have been found in the near-surface

layer and larger flocs in the near-bed saltwater layer

Nevertheless, the impact of turbulence on floc size was

identical in the freshwater and saltwater layers (Wolanski

and Gibbs1995) In the Yangtze River Estuary, flocculation

triggered by biological processes has been observed in

freshwater and brackish water (Guo and He 2011) The

varying hydrodynamics in the estuary generated strong

spatiotemporal fluctuations in FSDs, associated with strong

variations in macroflocs (defined as D≥D75in Guo and He

2011, where D is diameter), moderate variations in coarse

microflocs (D50<D≤D75), and a lack of significant

varia-tions in flocculi and fine microflocs (D≤D25) Due to the

complexity of flow in the Yangtze River Estuary, no clear

relation was obtained between vertical variations in current

flow and FSDs In contrast to the Mekong delta, however,

larger flocs were found episodically near the surface, where the turbulence was weak

The present study aimed at filling the gap in existing knowledge on the hydrosedimentary functioning of the northern tributary of the Red River delta Special attention was paid to the mechanisms responsible for the ongoing silting up of the waterways leading to the harbour of Haiphong Because the size and settling velocity of suspended aggregates are salient parameters for sediment transport, the focus was on the response of floc size distributions to various controlling factors characteristic of this monsoon estuarine setting, notably the impact of the seasonally fluctuating freshwater discharge on the tidal propagation in terms of turbulence level, the presence/absence of a saltwater layer, and tidal pumping

Study area and physical setting

The Red River (Song Hong) is situated at 20–25°30′N and 100–107°10′E and has a total catchment area of 169,000 km2 It bifurcates into numerous distributaries feed-ing the Red River delta, and enters the Gulf of Tonkin through six main mouths (Fig 1) Milliman and Meade (1983) estimated that the Red River brought approx 1%

of the world’s solid discharge to the ocean (160×106metric tons year−1), ranking it as 9th worldwide However, the authors recognized that this estimate was based on an in-complete database Moreover, the Hoa Binh dam con-structed in the late 1980s has proved to efficiently trap

Fig 1 The Red River delta,

Vietnam

sediments (Le et al.2007) Data from long-term monitoring

(1960–2008) indicate that the annual suspended sediment flux

was on average 113×106metric tons before dam construction,

and since then has decreased to on average 49×106metric

tons (Dang et al.2010) This sediment reaches the Gulf of

Tonkin through several distributaries, including the Cam

River that is connected to the easternmost branch of the delta

The Red River delta is under the influence of a tropical

monsoon climate Annual rainfall in the region is close to

200 cm, of which nearly 90% falls during the summer

monsoon The wind direction is dominantly from the south

in April–September (wet season), and from the northeast in

October–March (dry season) Typical of a

monsoon-dominated river, the Red River discharge is highly seasonal

Based on data for the years 1956–1998, discharge averaged

1,200 and 14,000 m3s−1at the Son Tay station (Fig.1) in

the dry and wet seasons respectively (van Maren and

Hoekstra2004) Approximately 90% of the total sediment

load is transported during the wet season (June–October)

Nowadays, inter-annual suspended sediment transport (Son

Tay) varies by as much as factor 4 (van Houwelingen2000)

The tides in the Gulf of Tonkin are largely diurnal, due to

resonance of the O1 and K1 diurnal components In the

vicinity of the Bach Dang–Cam Estuary, the diurnal tidal

regime shows a maximum amplitude of 4 m Because of

sheltering by the island of Tonkin (Hainan), wave action is

reduced in the northern coastal sector of the Red River delta,

more under the influence of tidal currents

The Bach Dang–Cam Estuary is located on the

eastern-most branch of the Red River, and fed by two relatively

wide main tributaries with shallow lateral shoals and deep

narrow channels: the Bach Dang and Cam

Materials and methods Surveys

Two field campaigns were conducted during the wet season

of 2008 (July) and the dry season of 2009 (March) Three stations situated at key spots of the estuarine system were monitored during 24-h surveys corresponding to one spring tidal cycle: one station was located upstream on the Cam River, another was upstream on the Bach Dang River, and yet another was close to the mouth of the estuary near the Dinh Vu industrial area These are hereafter referred to as the ‘Cam’, ‘Bach Dang’ and ‘Dinh Vu’ stations respectively (Fig 2) The tidal amplitude was approx 2 m during both campaigns, similar to the mean annual tidal amplitude During the dry season, bed samples were taken at each of the three stations by means of a clamshell-style dredge The deflocculated grain size distributions were assessed in the laboratory by use of a laser particle size analyzer in the range 0.05 to 878 μm (Mastersizer S, MALVERN Instruments)

During each survey, key physical parameters were mon-itored every 3 h from aboard a 12-m flat-bottom vessel Instantaneous cross-sectional velocity profiles were assessed using a 600 kHz acoustic Doppler current profiler (ADCP RDI Workhorse in bottom tracking mode) config-ured for a 0.5 m bin size Immediately after completing a cross section, the ship was anchored at the location corresponding to the maximum depth of the cross section (determined by echo sounding), for vertical profiling and sampling (cf below) Discharge at each cross section was estimated by WinRiver II software (RD Instruments)

Fig 2 The Bach Dang–Cam

Estuary, with the locations of

the three sampling stations

Water discharge and hydrodynamic parameters

The averaged river flow <Q> over a tidal cycle was estimated

from the integration of the instantaneous discharge Q(t) in a

24-h series of N measurements (N09) according to:

Q

h i ¼ 1

tN t1

X

N 1

i¼1

Qiþ Qiþ1

This steady component is defined as the fluvial

compo-nent of the discharge, and the fluctuating part, Q′0Q–<Q>,

as the tidal component The tidal asymmetry was defined as

the ratio of the duration of the observed flood tide (Q<0) to

the duration of the entire tidal cycle, expressed in

percent-age In order to precisely estimate the moments for observed

slack tides, the data were interpolated by cubic spline

(Fig.3)

During the wet season, the wind speed was obtained from

hourly recordings by the Vietnamese Meteorological

Ser-vice in the city of Haiphong During the dry season field

campaign, a Davis weather station was deployed on the roof

of the marine station of the Institute of Marine Environment

and Resources (IMER) at Do Son (Fig.2), at the entrance of

the bay Air temperature (under shelter), as well as wind

speed and direction were recorded every 30 minutes

The turbulent kinetic energy (TKE) dissipation rate

(ε, m2 s−3) integrated over the water column was

expressed as a function of wind and averaged current

velocity (van der Lee et al 2009):

"¼ kb

uav

h þ ks

w3

where kb and ks are the bottom and surface drag coef-ficients respectively, uav the depth-averaged water veloc-ity, w the wind velocveloc-ity, and ψ the ratio between the water and air density

The Kolmogorov microscale (lk, μm) yields an estimate

of the smallest turbulent eddies, and was calculated from the kinematic viscosity of water (ν) and from ε integrated over the water column (van Leussen1997):

lk¼ n

3

"

 1=4

ð3Þ The turbulence-induced shear rate G (s−1) is given by:

G¼ n

At each station, two vertical profiles of temperature, salinity and turbidity (optical backscattering sensor at

l 0880 nm, Seapoint turbidimeter) were recorded by means of a Seabird SBE19+ CTD probe Due to strong variations in salinity, a precise synchronization of the different sensors and pressure-to-depth conversion was carefully implemented

Following Simpson et al (1990), water column stratifi-cation was estimated in terms of the potential energy anom-aly ϕ (J m−3), which represents the amount of energy needed

to mix a unit volume of water column This parameter accounts not only for saltwater input but also for other factors (e.g heat flux, wind, rain) commonly influencing stratification:

ϕðtÞ ¼g h

Xh z¼0

ρwðtÞ ρwðz; tÞ

where ρw is the water density, g the acceleration due to gravity, and h the water depth, ρwbeing the value averaged over the water column (ρwðtÞ ¼1

h P h z¼0

ρwðz; tÞΔz)

Suspended particulate matter concentration Optical backscattering sensors have been widely used to assess total SPMC based on turbidity measurements (e.g Creed et al 2001; Fugate and Friedrichs 2002; Hoitink

2004; Voulgaris and Meyers 2004; Jouon et al 2008) In this study, each turbidity depth profile was measured in the main channel a few minutes after the velocity cross section; this slight delay is negligible compared to the tidal cycle of

24 h and the slow variations in river discharge

Water samples were collected at 3-h intervals 1.5 m be-low the surface and 1.5 m above the bed using Niskin bottles SPMCs (mg L−1) were determined by filtering 150–500 mL (depending on turbidity) subsamples through

Fig 3 Tidal asymmetry calculated as the ratio of the flood tide

duration and the tidal cycle duration (durations estimated from the

intercept of extrapolated discharge (line) with the zero ordinate;

ex-trapolated discharge obtained by fitting a spline curve to measurements

of discharge, circles)

pre-weighed polycarbonate Nucleopore filters (porosity

0.4 μm), as recommended by Fargion and Mueller (2000)

Filters were rinsed three times with 5.0 mL distilled

water, dried for 24 h at 75°C in an oven, and then

stored in a desiccator until weighing on a high-precision

(5 μg) electrobalance Data for duplicate or triplicate

samples were averaged in each case

The voltage delivered by the turbidity sensor was

converted by a Seapoint routine into turbidity (in FTU) using

laboratory-determined calibration parameters (Wass et al

1997; Bunt et al.1999) Since this conversion assumes that,

in the absence of reflecting particles, turbidity is equal to zero,

SPMC (mg L−1) was calculated from the relationship:

where m is a proportionality factor

At each station and separately for each campaign,

second-order linear regression analyses were conducted

on datasets for near-surface and near-bottom layers

Pooling these data per station and season revealed

coef-ficients of determination that were sufficiently high to

justify using one average conversion factor per station

and campaign

Suspended particulate matter discharge

In each case, the velocity profile corresponding to the

loca-tion of the turbidity profile was extracted from the

cross-sectional set It was assumed that the CTD profile was

representative of the same location, i.e any drift of the ship

was considered negligible, and the two scale depths were

matched between the surface and the bottom The velocity

profiles achieved with a bin width of 0.5 m were

interpolat-ed at the depths of CTD profiling Sinterpolat-ediment flux fs(z,t)

(g m−2s−1) was calculated as:

fsðz; tÞ ¼ u z; tð Þ
SPMC z; tð Þ ð7Þ

This comprises the advective sediment flux and the tidal

pumping of sediment By expressing both SPMC(z,t) and u

(z,t) as the sum of their tidally averaged components and the

deviation from the tidally averaged values, the tidally

aver-aged sediment flux becomes:

< fs>¼< < u > þuð 0Þ
< SPMC > þSPMCð 0Þ > ð8Þ

where the brackets < > indicate time-averaging over one

tidal cycle, and the prime indicates the deviation from the

tidally averaged value

Since << u >
SPMC0>¼ < u >
< SPMC0>¼ 0

a n d < u0
< SPMC >>¼< u0>
< SPMC >¼ 0 , E q

(8) becomes:

< fs>¼< u >
< SPMC > þ < u0
SPMC0> ð9Þ

These two components are defined as (cf Geyer et al

2001) the advective component of the tidally averaged sediment flux < qaðzÞ >¼< u z; tð Þ >
< SPMC z; tð Þ > , and the tidally driven component of the tidally averaged flux

< qpðzÞ >¼< u0ðz; tÞ
SPMC0ðz; tÞ > in the vicinity of the deepest location of the channel

The discharge and sediment transport per unit area S at the sampling station, q(t) (m3s−1) and qs(t) (g s−1) respec-tively, were calculated as:

qðtÞ ¼S h

Xh z¼0

and

qsðtÞ ¼S h

Xh z¼0 uðz; tÞSPMCðz; tÞΔz ð11Þ

The total sediment load over the whole cross section of the river, Qs(t), was obtained by assuming that the ratio qs(t)/q(t) did not vary significantly across the whole cross section:

QsðtÞ QðtÞ ¼

qsðtÞ

The sediment load averaged over one tidal cycle was computed according to:

Qs

h i ¼ 1

tN t1 X

N 1 i¼1

QsðiÞþ Qsðiþ1Þ

2 ðtiþ1 tiÞ ð13Þ

Floc size distribution Immediately after each CTD profile, a depth profile of FSD and concentration was conducted using an in situ laser scattering and transmissometry instrument (LISST 100X, Sequoia Scientific Inc.; e.g Traykovski et al.1999; Agrawal and Pottsmith 2000; Mikkelsen and Pejrup 2000; Jouon et

al 2008) The LISST of type C enables measurement of volumetric particulate concentration in 32 logarithmically spaced size classes ranging from 2.5 to 500 μm, with atten-uation at l0660 nm In view of the high turbidity in the study area, an optical path reduction module of 90% was employed, and the measurements corrected accordingly The mean apparent diameter Dvwas calculated for every FSD Dv was determined as the apparent diameter corresponding to 50% of the cumulative volume concentration

of aggregates

Expressed on a log normal scale for the apparent diame-ter, each FSD was decomposed into a mixture of 25 irreg-ularly spaced Gaussian curves, using the expectation-maximization (EM) algorithm of Tsui (2009) based on a maximum likelihood criterion A non-supervised spectrum analysis was applied: the Gaussian curves were sorted by

increasing modal diameter, and they were then progressively

merged as partial components until the mid-height position

met the boundary condition for a given component:

individ-ual clay/silt particles and flocculi (<30 μm), fine (<105 μm)

and coarse (<300 μm) microflocs, and macroflocs

(≥300 μm) As the size distribution range of the LISST

100X (type C) is truncated at 500 μm, the macrofloc mode

is not fully defined and the windowed mode is used as an

indicator (Fig 4) For each mode, two parameters were

calculated: its apparent median diameter Dv, and its relative

volume concentration (RVC, %) defined as the ratio of its

volume concentration to the cumulative volume

concentra-tion of all modes

Results

The main results obtained during the wet and dry seasons at

the three stations and averaged/integrated over the tidal

cycle and water column are summarized in Table1

Sediments

Bottom sediments

Grain size distributions of bottom sediments were similar at

the two upstream stations (Bach Dang River and Cam

River), characterized by relatively high clay to fine silt

contents (Fig 5) By contrast, bottom sediments at the

coastal Dinh Vu station were considerably more enriched

in fine sand (Wentworth scale)

Suspended particulate matter The factor m used to convert turbidity into SPMC had an average value (pooling all seasons and stations) of 1.54 (1.62 wet season, 1.47 dry season; Table2) The normalized bias for determination of SPMC averaged 2.8% and its standard deviation 5.4% in the dry season, and less in the wet season

During the wet season, highest SPMCs averaged over the tidal cycle reached 214 and 200 mg L−1at the Dinh Vu and Cam stations respectively, contrasting with only 128 mg L−1

at the Bach Dang station During the dry season, the corresponding values showed reductions by factors 2.6 and 2.9 at the Bach Dang and Cam stations respectively, and by factor 4.8 at the Dinh Vu station The relative variation in SPMC, defined as (SPMCmax−SPMCmin)/mean SPMC, did not differ markedly between the two seasons, but always remained minimal at the Cam station

At the Cam station during the wet season, SPMCs were homogeneous throughout the water column Similarly low SPMCs were recorded at the beginning of ebb tide and at flood tide They increased slightly at flood tide and more significantly at ebb tide Maximum SPMC was attained in the second half of ebb tide (Fig.6) During the dry season, SPMCs were lower at the beginning of flood tide than at ebb tide At flood tide and during the first half of ebb tide, a low SPMC near-bed layer was observed

At the Bach Dang station during the wet season, SPMCs were homogeneous throughout the water column, except at the end of ebb tide and beginning of flood tide when a low SPMC near-bed layer appeared (Fig.6) Lowest SPMC was recorded at the beginning of ebb tide and highest SPMC near the bed at the beginning of flood tide During the dry season SPMCs were very low, with a slight increase in the second half of ebb tide and at flood tide, associated with the formation of a turbid near-bed layer

At the Dinh Vu station during the wet season, SPMC reached its maximum value at mid-flood tide and, associated with peak discharge, at ebb tide (Fig 6) Thereafter, the values decreased, associated with the formation of a turbid near-bed layer that persisted until slack water of low tide During the dry season, the pattern was similar but more marked than that at Bach Dang: the overall turbidity was higher at flood tide than at ebb tide, accompanied by the appearance of a high-turbidity near-bed layer at flood tide Minimum SPMC values were recorded at the beginning of ebb tide

Median apparent diameter

An increase in Dvaveraged over one tidal cycle and over the water column was observed between the wet and dry sea-sons at each station, although less markedly at the Bach

Fig 4 Decomposition of an FSD into a particle/flocculus mode, a

microfloc mode comprising fine and coarse components, and part of a

macrofloc mode truncated by the LISST (dashed lines)

Dang station (Table2) There was a consistent linear

rela-tionship (R20 0.82) between <lk> and <Dv> (data not

shown)

During the wet season at the Cam station, Dvvalues were

rather homogeneous throughout the water column (Fig.7)

Dvwas larger at flood tide than at ebb tide From the end of

ebb tide until low tide slack water, the smallest Dvvalues

were found near the bed The pattern was reversed at the

Bach Dang station, with larger Dvat ebb tide than at flood

tide Larger Dvwere recorded near the surface at high tide

slack water and at late ebb tide A decrease in Dv was

observed at mid-ebb tide At the Dinh Vu station at flood tide, the water column was characterized by small Dvin the upper layer and larger Dv near the bed At ebb tide, the distribution of moderate Dvvalues was more homogeneous

in the water column, with values decreasing at mid-ebb tide During the dry season at the Cam station, a near-bed layer about 2 m thick was observed during the whole tidal cycle, except at mid-ebb and mid-flood tide (Fig.7) In this layer, the Dvvalues were smaller than in the upper part of the water column, with the exception of large Dv being recorded at flood tide Large Dvwere also found near the surface at mid-ebb tide At the Bach Dang station, large Dv

were observed near the surface at low tide slack water, and near both the bed and the surface at high tide slack water Dv

values were larger at flood tide than at ebb tide, the smallest values occurring in the lower part of the water column from high tide slack water to mid-ebb tide Water column distri-butions of Dvwere most homogeneous at ebb and mid-flood tide At the Dinh Vu station, the patterns were rather similar at ebb tide and flood tide, with large Dvnear the bed and small Dvnear the surface at slack tide, and homogeneous distribution of Dvvalues in the water column at mid-tide

Table 1 Physical parameters per station and field campaign: discharge

(Q, liquid), sediment load (Q s , solid), advective (q a ) and tidal pumping

(q p ) sediment fluxes (calculated positively downstream), tidal

asym-metry factor, potential energy anomaly (ϕ), salinity (Sal.), Kolmogorov

microscale (l k ), suspended particulate matter concentration (SPMC),

G·SPMC scale and microfloc RVC (relative volume concentration) The < > brackets indicate averaging over a tidal cycle, and the over-lines integration along the water column Minimum and maximum values are indicated in some cases

Wet season Dry season Wet season Dry season Wet season Dry season

qa

h i qp



 g mð 2s 1Þ 119.8–30.8 9.2–1.1 51.2–5.4 12.4–1.8 7.9–3.4 1.64–4.0

<ϕ> (J m −3 ) 0.2 (0.1–0.4) 3.5 (0.2–17.8) 4.9 (0.0–28.3) 31.9 (3.2–68.6) 4.3 (0.0–18.7) 23.4 (1.8–59.1)

Sal:

 

0.0 (0.0–0.2) 1.2 (0.1–7.0) 0.7 (0.1–3.0) 5.8 (1.9–13.9) 2.0 (0.1–7.6) 15.0 (7.0–23.7)

<lk> (μm) 323 (201–483) 390 (266–779) 406 (276–551) 482 (337–736) 293 (181–459) 438 (256–950)

SPMC

 

mg L 1

200 (133–287) 70 (22–91) 128 (41–179) 50 (21–105) 214 (56–372) 45 (23–72)

G
SPMC

10 3mg L 1s 1

3 (0.7–7) 0.7 (0.04–1) 0.9 (0.2–2) 0.3 (0.08–0.8) 4 (0.3–8) 0.4 (0.05–0.7)

Dv



RVC of microfloc mode (%) 9.0 (5.1–13.1) 9.5 (5.9–19.4) 12.4 (3.7–52.3) 7.5 (4.6–15.0) 23.4 (11.8–56.3) 8.7 (5.6–16.1)

Fig 5 Deflocculated size distributions of bed samples obtained during

the dry season at the Cam, Bach Dang and Dinh Vu stations

Wet season

Dry season

Bach Dang 1.59 1.51

Table 2 Proportionality factor

m obtained per station and campaign, where turbidity (FTU) 0 m SPMC (mg L −1 )

Floc size distribution

The Dvof the four components obtained from the analysis

of all measured FSDs combined were very consistent:

<10 μm, 30–50 μm, 130–170 μm, and >350 μm The first component is here defined as being constituted of individual particles and flocculi Following Dyer and Manning (1999), the second and third components are defined as belonging to Fig 6 SPMC during one tidal cycle at the Cam (top), Bach Dang (middle) and Dinh Vu (bottom) stations during wet (left) and dry (right) seasons

Fig 7 D during one tidal cycle at the Cam (top), Bach Dang (middle) and Dinh Vu (bottom) stations during wet (left) and dry (right) seasons

fine and coarse microflocs; the fact that fine aggregates are

likely to withstand higher levels of shear stress than are

larger aggregates legitimates the merging of these two

com-ponents into a distinct microfloc mode The last component

is defined as macroflocs, consistent with, for example,

Burban et al (1990) who defined marine snow as being

larger than 500 μm

The median apparent diameter for the particle/flocculus

mode exhibited a high stability at the three sites and during

the two seasons (mean Dv0 9.2 μm, σ20 0.60 μm) In the

present paper, the focus on the microfloc mode circumvents

the bias associated with median size being calculated over a

range of multimodal LISST spectra, or with a major part of

the spectrum extending beyond the maximum size detected

by the LISST

At the Cam station during the wet season, the RVC of coarse

microflocs exceeded that of fine microflocs at flood tide and at

the beginning of ebb tide, while the total RVC (fine+coarse

components of microflocs) remained almost constant (∼5%;

Fig.8) The proportion was reversed for the remainder of ebb

tide, with microfloc RVC increasing from 5% to 11–13% at late

ebb tide During the dry season, the coarse microflocs occupied

a larger volume than the fine microflocs throughout the tidal

cycle Microfloc RVC initially increased and then decreased at

ebb tide, ranging from 5% to 19% At flood tide, the values

remained constrained between 7% and 8%, exceeding those

recorded at flood tide during the wet season

At the Bach Dang station during the wet season, the

coarse microfloc RVC always exceeded that of fine

micro-flocs, except for one occasion at flood tide when the latter

reached 35% At ebb tide, the weak variation in microfloc

RVC (6–8%) was due to slight fluctuations in the coarse

microfloc component, the fine microfloc RVC remaining

nearly constant At flood tide the microfloc RVC increased,

whereby the fine component reached 35% and the coarse

component 18% at mid-flood tide At this stage, the FSDs

differed strongly between the 3-m-thick near-surface

fresh-water layer and the near-bed saltfresh-water layer, whereby fine

microflocs and particles/flocculi predominated in the

for-mer Thereafter, both the fine and coarse microfloc

compo-nents decreased to reach 1 and 5% respectively at the end of

flood tide During the dry season, the volume of coarse

microflocs always exceeded that of fine microflocs The

variation in microfloc RVC at ebb tide was similar to that

observed during the wet season, although more pronounced:

this encompassed an increase from 5% to 15%, followed by

a decrease to 6%, also mostly due to coarse microflocs Near

low tide slack water, there appeared a 4-m-thick near-bed

layer in which the proportion of macroflocs was higher than

in the remainder of the water column At flood tide, a similar

but less marked variation in microfloc RVC occurred, with a

maximum of approx 9% at mid-flood tide The fine

micro-floc RVC varied only weakly (approx 2%)

At the Dinh Vu station during the wet season, a strong increase in particles/flocculi (5%) together with fine and coarse microflocs (16 and 18% respectively) occurred at the beginning of ebb tide The particle/flocculus mode and the fine microfloc component were more pronounced in the 2.5-m-thick near-surface freshwater layer, their RVCs reached 3 and 38% respectively at mid-ebb tide The coarse microfloc RVC varied only slightly (14–18%) during most

of the ebb tide phase, with two exceptions: the values decreased at maximum discharge (9%) and at the end of ebb tide (9%) At flood tide there was an initial increase in the RVCs of microflocs—from 4% to 9% and from 8% to 12% for the fine and coarse components respectively—then

a decrease of the two modes

During the dry season, variations in microfloc RVC remained limited (between 6 and 10%) at the Dinh Vu station Although one order of magnitude smaller, the RVCs

of particles/flocculi had patterns similar to those of fine and coarse microflocs (Fig 8) At ebb tide, the fine microfloc RVC decreased steadily from 4% to 1% A decrease in coarse microflocs was recorded at flood tide simultaneously with a maximum of discharge, followed by an increasing and then decreasing pattern At mid-ebb tide, the macrofloc RVC decreased with depth, reaching a minimum at about

1 m above the bottom Near low tide slack water, there was a strong presence of particles/flocculi and microflocs; coarse microflocs were found mainly in a 2.5-m-thick near-bed saltwater layer At flood tide, the fine microfloc RVC remained constant (2%) but that of coarse microflocs de-creased steadily from 8% to 5% At high tide slack water, a 4-m-thick near-surface freshwater layer exhibited a high macrofloc RVC The difference between the RVCs of micro-flocs in the freshwater and saltwater layers was less marked than at low tide slack water

Hydrology Discharge Discharge exhibited a marked tidal influence at all stations during both seasons (Fig.9) The current flow averaged over one tidal cycle showed high seasonal variations The Cam station is under a dominant riverine influence but is also impacted by the tide, and would be classified as an upper estuarine site in both seasons following the schemes of Dionne (1963) and Dalrymple et al (1992)

Fig 8 Salinity (PSU) and volume concentration ratios for (from left to right) modes of particles/flocculi, fine and coarse components of microflocs, and macroflocs at (from top to bottom) mid-ebb tide, low tide slack water, mid-flood tide and high tide slack water at the coastal Dinh Vu station during the dry season Horizontal solid line Freshwa-ter–saltwater interface

b