Estuarine, Coastal and Shelf Science 57 (2003) 679–688

Suspended sediment concentrations in the Tamar estuary

G.R. Tattersalla,*, A.J. Elliotta, N.M. Lynnb
a
Centre for Applied Oceanography, Marine Science Laboratories, Menai Bridge, Anglesey, LL59 5EY, UK
b
Nuclear Department, HMS Sultan, Gosport, Hampshire, PO12 3BY, UK

Received 12 November 2001; accepted 1 November 2002

Abstract

Suspended sediment concentrations, obtained from the lower Tamar estuary in South West England using optical back scatter
sensors, showed a depth-averaged background concentration of 0.02 kg m
3 throughout most of the spring–neap cycle. On spring
tides the depth-averaged concentration increased to 0.25–0.40 kg m
3 either side of low water; however, the concentration maxima
did not correspond to the time of maximum tidal flow, which suggests the influence of sediment advection. The observations were
simulated using two-dimensional depth-averaged models of tidal currents and suspended sediment concentrations. Harmonic
constants generated by the tidal model were used to estimate the advective terms and the bed shear stress in the sediment transport
model. The sediment model included three size fractions which represented the low settling velocity wash load (2 lm), the cohesive
(25 lm) and the non-cohesive (75 lm) suspended loads.
During spring tides the simulated fine bed sediment (25 and 75 lm fractions) was resuspended in the upper model region and
advected down estuary on the ebb tide. The sediment transport model reproduced the observed low water concentration increases to
within a factor of two to five. The quantity of eroded sediment increased from medium tides to spring tides, as maximum bed shear
stress increased, and formed a mobile pool of suspended sediment. After spring tides less sediment was remobilised on successive
phases of the tide and accumulation occurred. The finer silty material (25 lm) was deposited in the shallower upper model region
whereas the sand sized particles (75 lm) accumulated in the deeper parts of the estuary, which was in general agreement with
published bed composition data. At neap tides, in accordance with the observations, the simulations showed no evidence of
sediment resuspension.
Ó 2003 Elsevier Science B.V. All rights reserved.

Keywords: sediment transport; advection; erosion; resuspension; model; Tamar

1. Introduction A turbidity maximum is present in the upper reaches

The Tamar estuary (Fig. 1) is a drowned river val-
ley in South West England, which forms a natural
boundary between Devon and Cornwall. The estuary is
31.7 km long and consists of the main river and two tidal
sub-estuaries: the Tavy and the Lynher. The river enters
the sea through the Narrows, a channel that is about
30 m deep, and then through Plymouth Sound. The
Tamar has a mean tidal range of 3.5 m and a mean
annual discharge of 27 m3 s
1.
* Corresponding author. The Centre for Environment, Fisheries and
Aquaculture Science, Pakefield Road, Lowestoft, Suffolk NR33 0HT,
UK. Tel.: +44-1502-524331; fax: +44-1502-513865.
E-mail address: g.r.tattersall@cefas.co.uk (G.R. Tattersall).
0272-7714/03/$ - see front matter Ó 2003 Elsevier Science B.V. All rights reserved.
doi:10.1016/S0272-7714(02)00408-0
of the estuary during summer (Uncles & Stephens, 1989,
1993), which migrates down estuary into the middle
reaches during winter due to the increased river dis-
charge. At spring tides the turbidity maximum is clearly
defined and the concentration reaches a peak value of
2–3 kg m
3, but at neap tides it is more diffuse with
maximum values of about 0.2 kg m
3 (Uncles, Barton,
& Stephens, 1994; Uncles, Stephens, & Barton, 1992).
Accordingly, the turbidity maximum has been the focus
of studies over the last two decades, while suspended
sediment transport processes in the lower estuary have
received less attention. Sediment accumulation is partic-
ularly important in the lower reaches where dredging
is required to maintain access to the dockyards at
Devonport (Fig. 1).
680 G.R. Tattersall et al. / Estuarine, Coastal and Shelf Science 57 (2003) 679–688

Fig. 1. The Tamar estuary showing the model area (dashed line) and the instrument locations: tide gauges (), bed-mounted ADCP (+), current
meter and optical back scatter sensor mooring (), and vertical profile mooring buoy (). Mean low water spring is shown by the thin solid line.
Insets: South West England (top left); location of the Tamar estuary (top right).

In summer there is a steady down estuary decrease in
the proportion of silty bed particles from 90 to 95% in
the turbidity maximum zone (1–6 km from Weir Head)
to 60% at the Narrows (Stephens, Uncles, Barton, &
Fitzpatrick, 1992). In winter the proportion of silt
decreases in the upper estuary as the turbidity maximum
is advected down estuary. There is a steady increase in
silt content from 50% at the head of the estuary to 90–
95% at a distance of 12 km from Weir Head (Fig. 1)
which decreases again to 60% at the Narrows. The mean
grain size of the bed sediment is 20–30 lm in the
turbidity maximum zone (Uncles et al., 1992) which
agrees with the mean suspended sediment grain size
determined using laser diffraction by West, Oduyemi,
Bale, and Morris (1990). At slack water the mean sus-
pended sediment particle size increases to 200–300 lm
due to flocculation.
Uncles and Stephens (1989) developed a one-dimen-
sional (1-D) axial model of the suspended sediment
distribution in the Tamar and showed that the formation
and position of the turbidity maximum was essentially
due to the combination of two processes: tidal resus-
pension of bed sediment and the magnitude of the river
discharge. In contrast, the turbidity maximum is usually
associated with the freshwater–saltwater interface (Festa
& Hansen, 1978). Uncles and Stephens (1989) found
that the predicted maximum was more diffuse than that
observed and hypothesised that this was probably due to
 G.R. Tattersall et al. / Estuarine, Coastal and Shelf Science 57 (2003) 679–688 681

the absence of gravitational circulation which was not
included in the model.
In common with the Tamar, the Weser estuary in
Germany exhibits a turbidity maximum in its low sa-
linity reaches and Lang et al. (1989) developed a 3-D
sediment transport model to reproduce the observed
twin peaked signal which is characteristic of the com-
bined effect of local resuspension and advection. The
model was able to simulate both peaks on the flood
tide but only the advected peak on the ebb tide. High
frequency fluctuations in the bed shear stress on the ebb
tide, which were not reproduced by the model, were
thought to be responsible for this additional sediment
resuspension. Teisson (1991) noted that although compu-
tation techniques exist, our ability to monitor changes in
three dimensions and over a sufficient period of time
limits the development of 3-D models. Realistic simu-
lations are hampered by our knowledge of erosion, de-
position and consolidation in the laboratory and the field.
The goal of the present study was to adapt tidal and
suspended sediment models developed for use in the
Firth of Forth (Clarke & Elliott, 1998) for application
to the Tamar estuary and to use the models to explain
the suspended particulate matter (SPM) variability ob-
served within the tidal cycle and its transition in charac-
ter from spring to neap tides.
where Cz is the concentration at a height z above the bed
and Ca is the concentration at a reference level, z ¼ a.
The power law exponent, the Rouse parameter, is pro-
portional to the ratio of the settling velocity, ws, and the
friction velocity, u , with the constant of proportionality
given by the product of von Karman’s constant, j ¼ 0:4,
and b. A value of unity is commonly used for b, how-
ever, empirical values range from b ¼ 1 for fine sand
to b ¼ 10 for medium sand (Dyer, 1986). A value of
b ¼ 1 was chosen to represent the fine sediment found
in this region of the Tamar.
The modified Rouse equation was fitted to each of
the 51 SPM profiles from the Hamoaze; Fig. 2 shows
some examples at 30 min intervals on 18 July 1996. The
closest agreement between the Rouse profile and the
observed concentrations was achieved at slack water
(about 13:30 hours) and maximum velocity (16:00–16:15
hours) when the water column was not accelerating.
During the accelerating and decelerating phases of the
tide, correspondence was poorer because the water
column was not in steady state. The settling velocity was
determined from the gradient of a best fit exponential

2. The observations

2.1. Vertical profiles from the Hamoaze

Vertical profiles of velocity and SPM concentration

were collected in the Hamoaze (Fig. 1) on three separate
occasions. The surveys were located at a buoy near
Weston Mill Lake where the mean depth was approx-
imately 17.5 m. A direct reading current meter was low-
ered to the bed in 2 m increments, the velocity being
recorded for 30 s at each depth. The profiles were
repeated every 20–30 min for 6–7 h (half a tidal cycle).
In addition, a CTD fitted with an optical back scatter
(OBS) sensor was lowered to the bed at the beginning
and end of each velocity profile.
The tidal and sediment transport models that will
be described in a later section use vertically integrated
equations and thus a depth-averaged velocity and SPM
concentration were required from the observations for
comparison with the model output. The sediment pro-
file described by Rouse (Dyer, 1986) is often used to re-
present the vertical distribution of suspended sediment.
The profile represents steady state conditions when
downward settling is balanced by upward diffusion of
sediment. The modified Rouse profile can be written as

bjws =u
a
Cz ¼ Ca ð1Þ Fig. 2. SPM concentration (+) and fitted Rouse profiles (dashed
z curve) at half hour intervals from the Hamoaze on 18 July 1996.
682 G.R. Tattersall et al. / Estuarine, Coastal and Shelf Science 57 (2003) 679–688
Fig. 3. Settling velocity (solid curve) and reference concentration
(dashed curve) calculated from the Rouse profiles fitted to the
Hamoaze data.
profile and Ca was calculated at the reference level which
was taken to be equal to the roughness length, i.e.
a ¼ z0 . For fine sediment, z0 lies in the range 2  10
4–
7  10
4 m (Dyer, 1986), so a roughness length of z0 ¼
4  10
4 m was selected for the Tamar. Fig. 3 shows the
settling velocity and reference concentration for the
18 July 1996 survey. The friction velocity, also required
to evaluate the Rouse parameter, was estimated from
the velocity data by calculating the best fit logarithmic
profile.
2.2. Time series from Saltash
Two recording current meters (Aanderaa RCM-7)
were suspended from the stern of a boat moored about
1 km north of Saltash for a period of 6 months (Fig. 1).
The mean depth at the mooring was approximately 8 m
and the current meters were suspended at depths of 2
and 5 m below the water surface. The boat rose and fell
with the tide and was free to rotate about the mooring
buoy. (When processing the data, account was taken
of the vertical movement of the sensors due to the rise
and fall of the surface vessel within the tidal cycle.) The
lower instrument was fitted with an OBS sensor and
spot readings of suspended sediment concentration were
recorded every 10 min at the start of each velocity aver-
aging period. The OBS sensors were calibrated by the
Tidal Waters Research Group at the University of
Birmingham using pumped suspended sediment samples
and recently deposited bed sediment. A linear relation-
ship was used to relate the voltage output of the sensor
to the suspended sediment concentration (Hills, per-
sonal communication). However, it should be noted that
Clifford, Richards, Brown, and Lane (1995) suggested
a linear relationship may not be appropriate and a
separate regression may be required for both the top
and bottom 10–20% of the sensor range. Moreover,
Osborne, Vincent, and Greenwood (1994) showed that
optical sensors are more sensitive to the smaller grain
sizes. The combination of these two factors may result in
larger calibration errors in the Saltash data at low
concentrations.
An expression for the depth-averaged concentration,
C, was obtained by integrating the modified Rouse
equation from the bed ðz ¼ 0Þ to the surface ðz ¼ hÞ,
where C9 is the measured concentration at a height z9
above the bed:

bjws =u
 C9 z9
C¼ ð2Þ
1
f h
The depth-averaged concentration was evaluated using
Eq. (2) with h and z9 taken from the results of the tidal
model and C9 was the observed concentration at Saltash.
The friction velocity was estimated by rearranging the
logarithmic velocity equation to give
v0j
u ¼ ð3Þ
lnðz0=z0 Þ
where v0 is the recorded velocity at a height z0 above the
bed (determined from the tidal model output). The set-
tling velocity was approximated using a step function
to distinguish between the background wash load (10
4
m s
1) and the higher settling velocity (2  10
3 m s
1)
observed when the concentration exceeded 0.02 kg m
3.
The higher settling velocity represented resuspended
bed sediment as determined from the Rouse profiles fitted
to the Hamoaze data (Fig. 3).
3. Model simulations
3.1. Hydrodynamic model
A 2-D tidal model of the Tamar estuary was de-
veloped from code that had been applied to the Firth
of Forth (Elliott & Clarke, 1998). The model used the
standard shallow water non-linear equations with a
numerical scheme that was centred in both time and
space, and included an algorithm to enable drying of
shallow areas. Tidal analysis of a 6-month tide gauge
record from Mill Bay (Fig. 1) provided harmonic con-
stants for the major constituents which were then used
to force the elevation across the open boundary at
Plymouth Sound. The Tamar grid had a resolution of
about 120 m  185 m which required a time step of less
than 2 s, and tidal analysis of the velocity output from
each grid cell provided harmonic constants for use in the
sediment transport model. The model was validated
against tide gauge, current meter and bed-mounted
ADCP data. Surface elevations, which were recorded at
Mill Bay and Weirquay (Fig. 1), were simulated with an
accuracy of 0.15 m while tidal current speeds at Saltash
and in the Narrows were reproduced to within 0.1 m s
1.
 G.R. Tattersall et al. / Estuarine, Coastal and Shelf Science 57 (2003) 679–688 683

At spring tides the simulated currents were accurate
to within 5% of the observed values and the times of
both peak flow and slack water were realistically repro-
duced. Simulations of drogue trajectories from the es-
tuary showed equally good agreement.
3.2. Sediment transport model
The sediment model included flocculation which en-
ables more sediment to settle out of the water column at
slack water than by Stokes Law alone. Flocculation
increases the mean settling velocity as larger particles
are formed, which is usually approximated in sediment
transport modelling by a power law relationship with
the concentration such that

The sediment model was developed to examine trans- n

wsf ¼ KC ð6Þ
port processes in the Firth of Forth (Clarke & Elliott,
where wsf is the floc settling velocity and K and n are
1998) and adapted for use in the Tamar. The depth-
empirical constants. In the Tamar it was assumed that
averaged velocity and bed shear stress were calculated
settling velocity increased linearly with concentration,
from the harmonic constants generated by the hydro-
i.e. n ¼ 1, and the constant of proportionality was set
dynamic model. A longer time step ðDT ¼ 10 sÞ was
to K ¼ 2  10
6 (Uncles et al., 1992). At each time step,
permitted because the sediment model was restricted by
the floc settling velocity was compared with the 25 lm
the stability criterion associated with the tidal currents
grain settling velocity determined by Stokes Law and the
not with the phase speed of the surface tide. The 2-D,
higher value used to calculate the deposition rate.
depth-averaged, advection–diffusion equation is written
     
q hC  q huC 
q hvC
þ þ 3.3. Initial conditions and boundary inputs
qt qx qy

 

q qC q qC The mean sediment grain size in the Tamar estuary
¼ hKX þ hKY þ E9
D9 ð4aÞ
qx qx qy qy is in the range 20–30 lm (Uncles et al., 1992), so 25 lm
where E9 is the erosion rate (Partheniades, 1965) and D9 was used to classify the cohesive sediment fraction in the
is the deposition rate (Krone, 1962), given by model. A 2 lm grain size was chosen to represent the
"
 # low settling velocity wash load and the non-cohesive fine
2
u sand (Stephens et al., 1992) was given a nominal 75 lm
E9 ¼ M
1 ; ð4bÞ
ue grain size. The bed sediment layer was composed of the
two larger fractions in the ratio of 65% cohesive (25 lm)
"
2 # and 35% non-cohesive (75 lm) sediment throughout the
u model region.
D9 ¼ ws Cb 1
ð4cÞ
ud A number of parameters required by sediment
transport models are poorly defined, site specific and
In Eq. (4a) h is the instantaneous depth (mean depth not easily determined from field data. In the literature,
plus elevation), u and v are the depth-averaged velocities the erosion threshold is generally in the range 0.1–
and KX and KY are the diffusion coefficients in the x and 1.5 N m
2 (Uncles & Stephens, 1989) and the Uncles
y directions. The erosion rate (4b) depends upon the et al.’s (1992) 1-D Tamar model used a critical erosion
erodability, M, and the critical erosion friction velocity, shear stress that increased linearly down estuary from
u*e. The deposition rate (4c) is described by the settling 0.14 N m
2 at Weir Head to 1.6 N m
2 at the Narrows.
velocity, the critical deposition friction velocity, u*d, and The model described in this paper had an equivalent
Cb, a near-bed concentration. critical erosion friction velocity gradient, which in-
In a depth-integrated model it is unrealistic to sup- creased from 0.025 m s
1 at Halton Quay to 0.040 m s
1
pose that the depth-averaged concentration represents at the Narrows. Uncles et al. (1992) used an erodability
the sediment that could settle out of the water column of M ¼ 3  10
5 kg m
2 s
1 but typical values for the
in a given time step. Hence, Cb is the concentration of erodability are generally higher, in the range 2  10
4–
sediment that is within a height wsDT above the bed. 3  10
3 kg m
2 s
1. An erodability of M ¼ 3:5  10
5
Clarke and Elliott (1998) calculated the near-bed SPM kg m
2 s
1 gave a better fit to the data presented in this
concentration in terms of the depth-averaged concentra- paper and was used in the 2-D sediment transport model
tion by integrating the modified Rouse profile (1) from described here. More consensus exists in the literature
the bed to wsDT, which gave on a deposition threshold, which is usually in the range

bjws =u 0.06–0.1 N m
2. The calculations described here had
 h
Cb ¼ C ð5Þ a uniform critical deposition friction velocity of ud ¼
ws DT
0:008 m s
1 (equivalent to 0.06 N m
2).
Eq. (5) was used to estimate the near-bed concentra- The initial wash load concentration was determined
tion required by Eq. (4c) from the depth-averaged from 10 CTD/OBS sections collected on five occasions
concentration. between May and September 1996 (Hills, personal
684 G.R. Tattersall et al. / Estuarine, Coastal and Shelf Science 57 (2003) 679–688

communication). Each CTD section consisted of 16

profiles taken at approximately 1 km intervals from
Weirquay to the Narrows on the down estuary leg and
then repeated when returning up estuary. The SPM
concentration showed an up estuary increase from
0.005 kg m
3 in Plymouth Sound to Weirquay, which
was extrapolated to Halton Quay and used to initialise
the model. The greatest of the 10 concentrations ob-
tained for Halton Quay was 0.070 kg m
3, which was
used for the 2 lm river boundary condition. The Lynher
and Tavy river boundary concentrations were 0.020
kg m
3 for the wash load (2 lm) and the input concen-
tration for the two larger fractions (25 and 75 lm) were
set to 10
5 kg m
3 at all four open boundaries. The
freshwater inputs at the Tamar, Tavy and Lynher
river boundaries were 20, 6 and 4 m3 s
1, respectively.

3.4. Suspended sediment simulations

Fig. 4 shows the simulated depth-averaged SPM

concentration compared with a 6-day section of the
Saltash time series from July 1997; Fig. 4c–e show
the contribution of each individual size fraction to the
total simulated concentration. The model reproduced
the background concentration and the ebb tide peaks in
a manner that was consistent with the data, which sug-
gests that no sediment was resuspended or advected past
the mooring at Saltash during neap tides. At spring tides
good correspondence, in terms of the timing and the rate
of ebb tide increase, was achieved between the simulated
suspended sediment and the observed SPM concentra-
tion. The instrument was sensitive to about 1 kg m
3,
above which it became saturated; this often occurred at
low water spring tides. A period when the instrument
was saturated or when it was on the bed is shown as zero
concentration in Figs. 4–6.
At the start of the flood tide the OBS sensor recorded
a sharp increase in SPM concentration followed by a
more gradual decrease. It was not possible to reproduce
the flood tide peak because the critical bed shear stress
in the model increased linearly down estuary. Thus, the
farther down estuary the mobile sediment was depos-
ited, the greater the shear stress that was required
to resuspend the sediment on the next phase of the tide.
To simulate the flood peak, it would be necessary to in-
corporate an erosion shear stress that increased verti-
cally through the bed to represent consolidation. For
example, Piasecki (1998) used a depth-dependent critical
bed shear stress to model suspended sediment and
radionuclides in the lower Fox river, Wisconsin.
3.5. Local erosion at Saltash
So far no distinction has been made between erosion
and resuspension. Erosion is the removal of sediment
from the bed that has either undergone some consol-
Fig. 4. (a) Model elevation at the Saltash mooring for the period
17–22 July 1997. Depth-averaged SPM concentration (solid curve) at
Saltash compared with (b) the total simulated suspended sediment
concentration (dashed curve), (c) the 2 lm wash load, (d) 25 lm
cohesive fraction and (e) 75 lm non-cohesive fraction. Periods when
the instrument saturated or was on the bed are shown as zero
concentration.
idation or been strengthened by biological reworking.
In contrast, resuspension describes the lifting of bed
sediment into the water column where consolidation has
not taken place and thus occurs at lower shear stresses.
The threshold required to resuspend sediment is less
than that necessary to remove the underlying consoli-
dated bed.
The rapid increase in observed SPM concentration at
the beginning of the flood tide which was not simulated
by the model (Fig. 4) could be either the resuspension
of sediment deposited during low water slack or the
erosion of the consolidated bed. To test the hypothesis
that the sediment could be eroded at Saltash, the model
was configured to ignore the advection and diffusion
terms of the sediment balance. This reduces Eq. (4) to
  "
 # "
2 #
q hC u
2
u
¼M
1
ws Cb 1
ð7Þ
qt ue ud
 G.R. Tattersall et al. / Estuarine, Coastal and Shelf Science 57 (2003) 679–688
Fig. 5. Local erosion model simulation. (a) Model elevation for the
Saltash mooring site for the period 16–21 July 1997. (b) Simulated
suspended sediment concentration (dashed curve) compared with the
depth-averaged SPM concentration (solid curve) estimated from the
Saltash data (16–21 July 1997). (c) As (a) for the period 22–27 July 1997.
(d) As (b) for the period 22–27 July 1997. Periods when the instru-
ment saturated or was on the bed are shown as zero concentration.
The local resuspension model was tuned to find the
critical erosion friction velocity that would allow the
simulated resuspension event to coincide with the ob-
served increase in SPM concentration at Saltash.
A critical erosion friction velocity of ue ¼ 0:02 m s
1
and an erodability of M ¼ 3:5  10
3 kg m
2 s
1 gave
a good correspondence at spring tides. Fig. 5 shows a
12-day simulation of the local resuspension at Saltash.
Over spring tides, 21–23 July 1997, the onset of the
simulated flood tide resuspension coincided with the
observed increase in depth-averaged SPM concentra-
tion. In contrast, at medium tides the observed flood
tide peaks were smaller and the onset occurred earlier in
the tidal cycle than suggested by the simulated increase
in concentration. On the ebb tide there was no observed
increase in SPM concentration corresponding with the
simulated resuspension peak which suggests that local
erosion of the bed was not the source of the observed
SPM concentration peaks at Saltash.
685
Fig. 6. Resuspension model simulation. (a) Model elevation for the
Saltash mooring site for the period 13–18 September 1997. (b) Sim-
ulated suspended sediment concentration (dashed curve) compared
with the depth-averaged SPM concentration (solid curve) estimated
from the Saltash data (13–18 September 1997). (c) As (a) for the period
19–24 September 1997. (d) As (b) for the period 19–24 September 1997.
Periods when the instrument saturated or was on the bed are shown
as zero concentration.
Within an individual tidal cycle, there was a rapid
increase in the simulated SPM concentration due to the
resuspension of the bed sediment on the ebb (Fig. 5).
Once the bed source was depleted the concentration
continued to rise, but more slowly because the volume of
the model grid cell decreased towards low water. The
suspended sediment started to settle when the shear
stress was less than the threshold of deposition, which is
evident as a rapid decrease in concentration immediately
after low water. This was followed by a return to the
background concentration because the majority of the
sediment had settled out of the water column. The rapid
increase at the start of the flood was the resuspension
of this sediment. Once again, the bed source was soon
depleted and the concentration then decreased gradually
due to the increasing volume of the grid cell. At the end
of the flood tide, the friction velocity fell below the
deposition threshold and a rapid decrease in concen-
tration was simulated when the suspended sediment
686 G.R. Tattersall et al. / Estuarine, Coastal and Shelf Science 57 (2003) 679–688
settled out of the water column at high water slack. The
gradual concentration increase/decrease in a particular
cell, simulated by the model when advection–diffusion
effects are neglected, illustrates one of the problems of
cohesive sediment transport modelling. Sanford and
Halka (1993) examined this paradigm where sediment is
neither settling nor resuspended into the water column.
For this period of the tidal cycle the bed does not inter-
act with the water column and the mass of suspended
sediment remains constant. Some sediment transport
models do not include a critical deposition shear stress
but assume that deposition is a continual process, e.g.
Lang et al. (1989).
3.6. Resuspension at Saltash
An alternative hypothesis is that sediment is eroded
at a location up estuary of Saltash and advected down
estuary on the ebb tide past the instrument mooring
at spring tides. Some of the sediment settles out of the
water column at low water slack, is resuspended on the
flood tide and advected back up estuary along with
the sediment that remains in suspension. Tidal asym-
metry in the region of Saltash would leave some of the
settled sediment in situ on the flood tide which may be
carried farther down estuary on the following ebb due to
the higher maximum bed stress. Clifton and Hamilton
(1979) determined an annual sediment accumulation
rate of 1–2 cm on the mud flats of St. John’s Lake and the
mechanism described above could account for the sup-
ply of sediment to the Hamoaze and the lower estuary.
The sediment advected back up estuary would be pumped
into the middle reaches because tidal asymmetry is domi-
nated by the flood tide in that region of the Tamar.
To test this hypothesis the model was initiated with
an arbitrary thin layer of bed sediment (0.02 m) in the
upper model region but with no bed sediment down
estuary of the Saltash mooring. The bed sediment was
eroded according to the linear erosion threshold gra-
dient equivalent to that used in the Uncles et al.’s (1992)
model. Sediment that settled outside the source zone
could be more readily be resuspended where ue ¼
0:02 m s
1 and M ¼ 17:5  10
5 kg m
2 s
1 (five times
the erodability of the source region). Fig. 6 shows the
simulated SPM concentration against the September
1997 Saltash data set. The maximum concentration in-
creased from medium tides up to spring tides and then
decreased again. On each tide there were two concen-
tration maxima, separated by a minimum at low water,
which were in phase with the observations and the
simulated concentration returned to background levels
at high water. The high background concentration
present in the data may be due to an increase in water
column turbidity between the July and September de-
ployments or an instrument calibration error. Wells
and Kim (1991) observed a background concentration
increase in the Neuse river, North Carolina, due to or-
ganic matter in the water column. Increasing the wash
load input at the model boundaries could not reproduce
the high background concentration (0.04 kg m
3) and
thus it is probable that the apparent increase was an
instrument calibration error. The wash load was not
included in this simulation so an offset of 0.04 kg m
3
was applied to the model output to aid the comparison
between the simulated resuspended sediment and the
high concentration peaks observed in the Saltash record.
3.7. Sediment redistribution
The simulated change in bed elevation and compo-
sition after eight springs–neaps cycles (approximately
116 days) are shown in Fig. 7. The model was initiated
with sufficient bed sediment to prevent scour occurring
during the simulation and Fig. 7a shows the change in
bed elevation at high water neap tides. The main depo-
sitional areas were the shallower regions at the sides of
the estuary, the sub-estuaries and the Hamoaze, whereas
erosion occurred in the central channel north of Saltash,
in the Narrows and in Plymouth Sound. Admiralty
charts show that the Hamoaze is continually dredged to
maintain the minimum depth for shipping, while the
simulated erosion sites in the Narrows and in Plymouth
Sound are rocky outcrops where the bed is clear of
sediment.
Two sizes of sediment were included in the simulation
to allow an estimate of the change in bed composition.
The model was initiated with 65% cohesive sediment
and 35% non-cohesive sediment everywhere and Fig. 7b
shows the percentage of cohesive bed sediment after
eight springs–neaps cycles. An increase in the propor-
tion of silty (25 lm) sediment indicates either net accre-
tion of cohesive sediment or removal of non-cohesive
sediment; conversely an increase in the proportion of the
non-cohesive (75 lm) fraction demonstrates the removal
of fine sediment or a build up of sand. Comparison of
Fig. 7a and b suggest that the accretion in the upper
reaches of the rivers Tamar, Tavy and Lynher is due to
an increase in the proportion of fine sediment, while the
accretion in the Hamoaze is due to an increase in sand.
In the model simulations, more sand than silt was re-
moved from the main erosion area in the central channel
north of Saltash. In general there was an accretion of silt
in the shallower upper reaches of the estuary and an
increase in the proportion of sandy bed sediment in the
deeper Hamoaze. This trend is in broad agreement with
Stephens et al. (1992) who showed that silt content
increased with distance from the Narrows.
4. Summary
The SPM concentrations observed at Saltash showed
that there was little or no sediment resuspension in the
 G.R. Tattersall et al. / Estuarine, Coastal and Shelf Science 57 (2003) 679–688 687

Fig. 7. (a) Change in bed elevation and (b) percentage of cohesive bed sediment after eight springs–neaps cycles (approximately 16.5 weeks). The
initial bed composition was 65% cohesive sediment (25 lm) and 35% non-cohesive sediment (75 lm). Hatched areas represent regions of the estuary
that have dried.

lower estuary at neap tides. At spring tides sediment was
advected down estuary on the ebb tide and back up
estuary on the flood. Some sediment was deposited at
low water over a single tidal cycle and most was re-
suspended on the following flood. The observed concen-
trations were simulated to within a factor of two to five
by a source of bed sediment located up estuary of the
Saltash mooring which eroded steadily from medium
tides to spring tides. The sediment supply increased to-
wards spring tides and decreases afterwards. In conse-
quence, the eroded sediment became a mobile pool of
sediment that was advected past the Saltash mooring
just before low water. Some of the sediment settled out of
the water column at low water slack but was resuspended
again on the flood. The mobile sediment was advected
farther up estuary than the ebb tide erosion site because
sediment motion was induced earlier in the tidal cycle
due to the lower threshold required for resuspension
than for erosion. The mobile sediment was advected
down estuary on the following ebb tide when the volume
was increased by erosion from the bed.
Over the springs–neaps cycle there was a gradual
increase in eroded sediment supplied to the mobile pool.
The pool increased up to the time of the maximum
spring tide because all deposited sediment was resus-
pended on the following phase of the tide. After the
spring tide maximum not all the deposited sediment was
resuspended; that which remained was consolidated into
the bed. Sediment that was deposited to the south of the
Saltash mooring at low water was preferentially moved
into the Hamoaze due to ebb tide asymmetry, which
supplied sediment to the lower model region. Con-
versely, sediment deposited in the northern part of the
model region was pumped up estuary into the turbidity
maximum zone due to a longer slack period at high
water and the greater flood tide asymmetry in that
region of the estuary.
Uncles and Stephens (1993) observed that sediment
was moved down estuary from the turbidity maximum
zone in winter and spring when river flow is high and
returned to the upper estuary in summer and autumn
when runoff is low. The mechanism described in this
paper details the pumping of sediment from the lower
estuary to the middle estuary when tidal processes dom-
inate over the down estuary river transport, e.g. during
the summer months. A comparatively small amount of
688 G.R. Tattersall et al. / Estuarine, Coastal and Shelf Science 57 (2003) 679–688
sediment was supplied to the lower part of the estuary
and could account for the annual increase determined
by Clifton and Hamilton (1979).
Acknowledgements
This work was supported by the UK Defence Pro-
curement Agency. The authors thank Drs J.R. West and
A.M. Hills of the Tidal Waters Research Group at the
University of Birmingham for the OBS sensor calibra-
tion data and the CTD/OBS sections.
References
Clarke, S., & Elliott, A. J. (1998). Modelling suspended sediment
concentrations in the Firth of Forth. Estuarine, Coastal and Shelf
Science 47, 235–250.
Clifford, N. J., Richards, K. S., Brown, R. A., & Lane, S. N. (1995).
Laboratory and field assessment of an infrared turbidity probe and
its response to particle size and variation in suspended sediment
concentration. Hydrological Sciences Journal 40, 771–791.
Clifton, R. J., & Hamilton, E. I. (1979). Lead-210 chronology in rela-
tion to levels of elements in dated sediment core profiles. Estuarine
and Coastal Marine Science 8, 259–269.
Dyer, K. R. (1986). Coastal and estuarine sediment dynamics (342 pp.).
Chichester: Wiley-Interscience.
Elliott, A. J., & Clarke, S. (1998). Shallow water tides in the Firth of
Forth. Hydrographic Journal 87, 19–24.
Festa, J. F., & Hansen, D. V. (1978). Turbidity maxima in partially
mixed estuaries: a two-dimensional numerical model. Estuarine and
Coastal Marine Science 7, 347–359.
Krone, R. B. (1962). Flume studies of the transport of sediment in
estuarial shoaling processes (110 pp.). Berkeley: University of
California Hydraulic Engineering Laboratory.
Lang, G., Schubert, R., Markofsky, M., Fanger, H.-U., Grabemann,
I., Krasemann, H. L., Neumann, L. J. R., & Riethmuller, R.
(1989). Data interpretation and numerical modeling of the mud
and suspended sediment experiment 1985. Journal of Geophysical
Research 94, 14381–14393.
Osborne, P. D., Vincent, C. E., & Greenwood, B. (1994). Measurement
of suspended sand concentrations in the nearshore: field inter-
comparison of optical and acoustic backscatter sensors. Continental
Shelf Research 14, 159–174.
Partheniades, E. (1965). Erosion and deposition of cohesive soils.
Journal of the Hydraulics Division, ASCE 91, 105–139.
Piasecki, M. (1998). Transport of radionuclides incorporating co-
hesive/non-cohesive sediments. Journal of Marine Environmental
Engineering 4, 331–365.
Sanford, L. P., & Halka, J. P. (1993). Assessing the paradigm of
mutually exclusive erosion and deposition of mud, with examples
from upper Chesapeake Bay. Marine Geology 114, 37–57.
Stephens, J. A., Uncles, R. J., Barton, M. L., & Fitzpatrick, F. (1992).
Bulk properties of intertidal sediments in a muddy macrotidal
estuary. Marine Geology 103, 445–460.
Teisson, C. (1991). Cohesive suspended sediment transport: feasibil-
ity and limitations of numerical modelling. Journal of Hydraulic
Research 29, 755–769.
Uncles, R. J., & Stephens, J. A. (1989). Distributions of suspended
sediment at high water in a macrotidal estuary. Journal of
Geophysical Research 94, 14395–14405.
Uncles, R. J., & Stephens, J. A. (1993). Nature of the turbidity
maximum in the Tamar Estuary, U.K. Estuarine, Coastal and Shelf
Science 36, 413–431.
Uncles, R. J., Barton, M. L., & Stephens, J. A. (1994). Seasonal
variability of fine-sediment concentrations in the turbidity max-
imum region of the Tamar estuary. Estuarine, Coastal and Shelf
Science 38, 19–39.
Uncles, R. J., Stephens, J. A., & Barton, M. L. (1992). Observations of
fine-sediment concentrations and transport in the turbidity
maximum region of an estuary. In D. Prandle (Ed.), Dynamics
and exchanges in estuaries and the coastal zone, Coastal and estuarine
studies Vol. 40 (pp. 255–276). Berlin: Springer.
Wells, J. T., & Kim, S.-Y. (1991). The relationship between beam
transmission and concentration of suspended particulate material
in the Neuse river estuary, North Carolina. Estuaries 14, 395–403.
West, J. R., Oduyemi, K. O. K., Bale, A. J., & Morris, A. W. (1990).
The field measurement of sediment transport parameters in
estuaries. Estuarine, Coastal and Shelf Science 30, 167–183.