Satilla salt balance

This article appeared in a journal published by Elsevier. The attached
copy is furnished to the author for internal non-commercial research
and education use, including for instruction at the authors institution
and sharing with colleagues.
Other uses, including reproduction and distribution, or selling or
licensing copies, or posting to personal, institutional or third party
websites are prohibited.
In most cases authors are permitted to post their version of the
article (e.g. in Word or Tex form) to their personal website or
institutional repository. Authors requiring further information
regarding Elsevier’s archiving and manuscript policies are
encouraged to visit:
http://www.elsevier.com/copyright
Author's personal copy
ARTICLE IN PRESS
Continental Shelf Research 29 (2009) 15– 28
Contents lists available at ScienceDirect
Continental Shelf Research
journal homepage: www.elsevier.com/locate/csr
The effect of secondary circulation on the salt distribution in a sinuous
coastal plain estuary: Satilla River, GA, USA
H.E. Seim a,, J.O. Blanton, S.A. Elston c
a
b
c
Department of Marine Science, University of North Carolina at Chapel Hill, CB# 3300, 340 Chapman Hall, Chapel Hill, NC 27599, USA
Skidaway Institute of Oceanography, 10 Ocean Science Circle, Savannah, GA 31411, USA
Spatial Information Science and Engineering, University of Maine, 5711 Boardman Hall, # 348 Orono, ME 04469, USA
a r t i c l e in fo
abstract
Article history:
Received 12 February 2007
Received in revised form
29 February 2008
Accepted 13 March 2008
Available online 27 March 2008
An analysis of observational data suggests salt exchange in a sinuous coastal plain estuary is
significantly impacted by counter-rotating residual horizontal eddies formed by channel curvature in
meandering channels. The parts of adjacent eddies that advect material downstream follow the deep
part of the channel where the flow continually criss-crosses from one side of the channel to the other
and follows a relatively unimpeded trajectory to the sea. On the other hand, the parts of adjacent eddies
that advect material upstream cross the channel at a different location where it encounters a series of
shoals. In this case, the resulting upstream transport of salt is relatively inefficient and retards the rate
at which salt can disperse upstream into the estuary. The strength of these circulations is modulated by
the spring/neap cycle, allowing for a stronger gravitational mode of exchange to develop near neap
tides, but has minimal impact on the length of the salt intrusion. It is suggested that the impeded
upstream salt transport accounts for the observation that an impulse of river discharge advects a given
isohaline 10 km downstream in 20 days, but that after the impulse, 70 days are required to return the
isohaline to a similar position, counter to the notion of a simple dependence of intrusion length on river
discharge.
& 2008 Elsevier Ltd. All rights reserved.
Keywords:
Channel meanders
Secondary circulation
Axial salinity distribution
Lateral salinity distribution
Neap–spring cycle
Salt exchange
Stratification
1. Introduction
The Satilla River estuary is a relatively pristine salt-marsh
estuary along the coastline of Georgia, USA (Fig. 1). Lying at the
center of the South Atlantic Bight (SAB), the Satilla experiences
some of the largest tides in the southeast US. The sinuous nature
of the Satilla River estuary is characteristic of many of the
estuaries along the SAB. The M 2 tide accounts for about 80% of the
tidal energy in the Satilla (Seim et al., 2006). The tidal regime is
consistent with a strongly convergent estuarine geometry. On
most bends, the momentum core shifts from the inside to the
outside of the bend. A strong (order 0:1 m s1 ) depth-averaged
residual flow is produced at the bends forming counter-rotating
eddies that meet at the apex of the bends. This type of residual
flow is thought to be typical of sinuous, meandering estuaries
(Dronkers, 2005).
Less well known is the structure and variability of the mass
field in these systems. Importantly, a secondary circulation that
takes the form of a helical flow pattern is generated in open
Corresponding author Tel.: +1 919 962 2083.
E-mail addresses: [email protected] (H.E. Seim), [email protected]
(J.O. Blanton), [email protected] (S.A. Elston).
0278-4343/$ - see front matter & 2008 Elsevier Ltd. All rights reserved.
doi:10.1016/j.csr.2008.03.018
channel flow around bends in unstratified systems (Smith and
McLean, 1984). In stratified systems, the helical flow can be
suppressed if aFr 2 ðB=RÞo1, where a is a measure of vertical shear
of the streamwise velocity us, typically about 0.5, Fr ¼ U 22 =g 0 h is a
Froude number, g 0 is reduced gravity, h is water depth, B is channel
width and R is the radius of curvature of the channel (Seim and
Gregg, 1997). The strength and structure of the secondary
circulation in estuaries responds to the neap–spring cycle and to
river discharge (Chant, 2002). A helical flow pattern can be
established during spring tide with flow to the outside of the
curving channel at the surface and flow to the inside near bottom.
This pattern can change to a two-cell structure, one cell stacked
over the other, during smaller tides (Chant, 2002; Elston, 2005)
with secondary flow due to channel curvature confined to the
upper cell. Because the secondary circulation has the potential to
overturn the water column its presence or absence may
significantly impact mixing and exchange in an estuary.
The impact of curvature-induced secondary circulation on the
salt balance in an estuary remains largely unexplored. Most
observations and models of exchange in straight estuaries indicate
the importance to the salt balance of runoff and baroclinic
pressure gradients (e.g. Uncles et al., 1986; Wong, 1994). Landward flow is found in the thalweg and seaward flow in the
shallows. The opposite configuration is found in barotropic
Author's personal copy
ARTICLE IN PRESS
16
H.E. Seim et al. / Continental Shelf Research 29 (2009) 15–28
Satilla River
31.05
Latitude (°N)
31.00
FL
ATL A N TI C OCEA N
SC
GA
7
5
2
3
1
4
LT2
6
4a
LT1
>5 m
30.95
2m
5m
0m
2m
-2 - 0 m
0 MLW
<-2 m
30.90
-81.70
-81.60
-81.50
Longitude (°W)
-81.40
Fig. 1. Satilla River morphology showing the large intertidal areas typical of estuaries along the southeastern US. The locations of current, sub-surface pressure and salinity
monitors during SAT1 and SAT2 are shown as white circles. Distances of stations from the ocean are given in Table 1.
models (Li and O’Donnell, 1997). There have been studies to
explore how these two modes operate together in shallow tidal
channels (Li et al., 1998; Ralston and Stacey, 2005). In these
simplified models the details of exchange are governed by the
2
Ekman number Ek ¼ Az =ðfH Þ (Valle Levinson et al., 2003) which
measures the relative importance of rotation based on the
strength of vertical mixing (Az ), depth (H), and Coriolis effect (f).
Using typical ranges of vertically averaged axial velocity (U)
throughout
the estuary, C d ¼ 0:002 (Seim et al., 2002), and with
pffiffiffiffiffiffiffiffiffi
Az ¼ 0:05 C d U , Az varies from 0:009 m2 s1 at neap to
0:015 m2 s1 at spring. The corresponding values for Ek are 1:5
at neap and 2:4 at spring, indicating that frictional forces can be
expected to dominate over Coriolis effects in the Satilla River
estuary.
These formulations do not explicitly include the role of
secondary circulation, however. Secondary circulation may decrease dispersion of salt landward (Smith, 1976; West et al., 1990).
Some evidence of the impact of secondary circulation on the salt
balance has been seen. Focusing attention on a single channel
bend, the along-channel density gradient in the thalweg was
found to be weakest during spring tide (Seim et al., 2002),
suggesting that the salinity intrusion could be greatest at
spring tide. At neap, vertical stratification was strong enough
to raise the gradient Richardson number well above 0.25 during
ebb, consistent with tidal straining. The transition was related
to activation and suppression of helical circulation around the
bend.
In contrast to the focus on tidal circulation by Seim et al.
(2006), this paper describes the subtidal circulation, the subtidal
salt field variations, and the intrusion of salt into the estuary. We
provide a conceptual model to explain the observed variability.
The elements of the conceptual model include a link between
channel curvature and the secondary circulation is generates,
which causes a tilt in the free surface at the bends, and the
residual tidal eddies reported briefly by (Seim et al., 2006), which
can be considered a tertiary flow. The overturning circulation at
the bends promotes a horizontal salinity contrast that is carried
along-estuary by the residual tidal eddies. The geometry is such
that the seaward flow follows the deep channel of the estuary and
proceeds largely unobstructed, whereas the landward flow
follows a shallower channel and is prone to strong mixing over
shoals. This circulation appears to weaken the dependency of the
salt intrusion on fluctuations in river discharge because it allows a
pulse of river discharge to rapidly push saline water seaward.
However, salt moves landward significantly slower.
After outlining the field program and analysis performed, we
describe the observed circulation features including the spring–neap differences in the structure and strength of the observed
residual tidal eddies. Next we describe details in the salinity field
and how its longitudinal and lateral distribution changes in
response to secondary circulation and the neap–spring cycle.
Finally, we discuss how the observed exchange flow in counterrotating eddies affect how salinity intrudes into the estuary in
response to fluctuations in river discharge.
2. Data description and analyses
We undertook two 2-month long field efforts in the Satilla
River estuary during 1999, designed to sample high (spring) and
low (autumn) river discharge conditions. In each case two
sampling schemes were employed: a series of along-channel
moorings, deployed for the full 2-month period, and several
intensive small-boat sampling exercises. The small-boat sampling
strategy was designed for limited reaches of the estuary to yield a
picture of the three-dimensional variability of the circulation and
density field over a tidal cycle. This effort covered four adjacent
sections of the estuary (Fig. 1), once at spring tide and once at
neap tides. We refer to the these intensive efforts as roving
samples. These eight intensive sampling periods were split to
coincide with neap and spring tides of the high and low river
discharge mooring programs. Additional temperature, salinity,
and pressure data were collected at two long-term moorings, one
upriver from the mouth of the Satilla River at LT1 and one near
river kilometer 38 at LT2.
2.1. Salinity, pressure, and velocity moorings
The mooring array used Teledyne RD Instruments, Inc. (RDI)
ADCPs and InterOcean S4 current meters to measure velocity and
Seabird Electronics, Inc. (SBE) sensors to measure temperature,
salinity, pressure, and optical backscatter (OBS) (Table 1). As a
result, in some locations full-depth velocity profiles are available,
while at other locations currents are observed at a single depth.
Moorings consisted of a pyramid-shaped stainless steel frame
placed at each location on which a subsurface (0.5 mab) SBE
SeaCAT CTD and either an InterOcean S4 current meter or an
upward-looking RDI Workhorse ADCP were mounted. A SBE
MicroCAT CTD was tethered with 40 ft of stainless steel cable to
each pyramid to measure surface temperature and salinity.
Author's personal copy
ARTICLE IN PRESS
H.E. Seim et al. / Continental Shelf Research 29 (2009) 15–28
Table 1
Stations of moored instrumentation for spring and autumn data sets
Station
2
LT1
3
4a
4
5
6
LT2
7
Distance (km)
Depth (m)
HW width (m)
LW width (m)
Parameters
4
8.9
3800
3000
P; CB
10
3.4
5400
1500
P
13
9.3
3800
1100
P; CB
18
5.0
6000
900
P; CB
18
7.9
6000
900
P; CP
25
9.6
6000
500
P; CP
30
8.5
2300
200
P; CP
39
4.3
1400
300
P
WOC+4
10.2
3600
170
P; CP
Station 7 is in White Oak Creek (WOC). Distance is measured from the mouth of
the estuary. Station depth was calculated from bottom pressure using UNESCO
(1985) protocol. LW and HW widths were calculated from regional GIS data by
Alice Chalmers, University of Georgia. Parameter symbols are: P ¼ subsurface
pressure from SeaBird sensors; CB ¼ bottom current from InterOcean S-4; CP ¼
current profile from RDI ADCP workhorse.
Temperature, salinity, pressure, and OBS were recorded at 6 min
intervals by the SeaCAT and MicroCAT CTDs while the ADCPs
measured depth, velocity, and acoustic backscatter at 12 min
intervals starting at 1.24 mab to 0.5 m below the water surface.
Shipboard DGPS and in situ fathometry were used to determine
the exact locations and depths of the mooring placements. See
Elston (2005) for additional information on the instrument
configuration at each mooring.
Two long-term monitoring stations were in place during this
study. Data at LT1 will be shown for a 9-month time series
(March–December 1999) of temperature, salinity, and pressure at
a 6-min interval. Data collected at LT1 were used to monitor and
relate changes in temperature, salinity, and pressure over the
monthly, seasonal, and annual cycles to changes in freshwater
discharge. LT1 pressure data were used to adjust the depth
component of the moored CTDs and moored ADCPs to the mean
lower low water (MLLW) datum. LT2, located farther upriver,
functioned in a similar manner, and a 2-month record (September–November 1999) will be shown.
Salinity data from the moored SeaCAT and MicroCAT CTDs
were of excellent quality except for Station 4 during the autumn
1999 mooring deployment. Antifoulants significantly reduced biofouling during mooring deployments. Due to rough weather and a
highly corrosive environment, several of the surface-tethered
MicroCAT CTDs broke free approximately 20 days after the initial
deployment thus compromising most of our surface salinity
measurements at the moorings. Temperature and pressure data
for the SeaCAT CTDs and the MicroCAT CTDs were of excellent
quality for both the spring and autumn 1999 mooring deployments.
The salinity fluctuations recorded during autumn 1999 at
Station 4 (main channel) had tidal fluctuations that were too
small, suggesting some clogging of the conductivity sensor. In
order to better simulate the salinity for Station 4, the data at
Station 4 were substituted with Station 5 salinity to which a
constant value of 7.5 PSU was added, based on mean downstream
increase in salinity from Station 5 to Station 4.
17
fathometers and a DGPS unit, a side-towed downward-looking
600 kHz RDI Broad Band ADCP, mounted on a floating sled to
resolve three-dimensional currents in the axial, lateral, and
vertical directions, and a frame-suspended Falmouth Scientific
Instruments (FSI) CTD for stop and drop CTD casts at predetermined side channel locations and at predetermined locations over
and adjacent to the mooring field. While the stop and drop CTD
data collected had discrete (rather than continuous) spatial
coverage, it nicely complemented the surface temperature and
salinity data from the SBE-21 unit by documenting cross-channel
changes in vertical stratification.
Roving ADCP observations were analyzed spatially on a
latitude/longitude grid following Seim et al. (2006) and on
latitude/depth grids following Elston (2005). For the latitude/
longitude grid analysis, the processed, 15-s averaged roving ADCP
observations were analyzed on a 200 m grid. At each location in
the sampling grid, a simple least-squares fit was used to estimate
the mean, tidal current ellipse parameters at semi-diurnal and
quatra-diurnal frequencies and an error term. The fit was
performed for a grid cell if there were a minimum of seven
clusters of observation times over the tidal cycle. Often there was
more than one 15-s average in a grid cell from a circuit, in which
case multiple current estimates were included. Typically there
were 9 or 10 sets of current observations per grid point over the
13 h sampling period. Root-mean-square errors average 1:6 cm s1
during spring tides and 1:3 cm s1 during neap tides. Differences
in the spatial coverage between spring and neap tide surveys
occurred because some shallows became impassable at spring
tide low water. The most dramatic example was near mooring 3,
where we were unable to adequately sample the shallower
channel to the west of the marsh island.
For the latitude/depth grid analysis, raw, 2-s averaged
irregularly spaced roving ADCP observations were reduced to
the MLLW datum and interpolated onto a regular grid with 8-m
horizontal and 0.5-m vertical bins. Data in a given cross-section
were not demeaned, so that absolute residual values could be
calculated. Next, at each survey location, the laterally gridded
cross-section was rotated into along and cross-channel components by using the tidal spatially averaged depth-averaged
transport. Last, for each cell in the regular latitude/depth grid
with at least five velocity observations over the tidal cycle, a
simple least-squares fit was used to estimate the mean, tidal
current ellipse parameters for the semidiurnal and quatradiurnal
frequencies, and an error term. The rms values of the errors are,
respectively, on average 0:9 cm s1 at spring and 0:4 cm s1 at
neap tides. Because of the difference in the spatial area covered
between spring and neap tides and the minimum five observations required per sample cell fit, the lateral coverage in a given
section varied between surveys. The intertidal area was excluded
from the resulting tidally averaged cross-sections, and the number
of fit boxes at spring tide was reduced from that at neap tide by
approximately 12%.
3. Results
2.2. Roving vessel observations
3.1. Distribution of axial flow along the estuary
Roving surveys were conducted while underway on a 8-m
estuarine research vessel. In the ‘‘roving survey’’ mode, highresolution surface temperature and salinity, velocity, depth, and
position were collected continuously at a sampling interval of 1, 5,
and 15 s over a tidal cycle in a predetermined zigzag pattern
covering a 4-km reach. The surveys were conducted at an average
boat speed of 3 m s1 , a speed that allowed coverage of 4-km
zones along the axis in 1:5 h. Data were acquired using a surface
pumped SBE Thermosalinograph (SBE-21) system linked to two
With few exceptions, the near-bottom axial subtidal flow in the
thalweg of the estuary was consistently seaward, in both the fall
deployment (shown in Fig. 2 because of its superior coverage) and
in the spring deployment (not shown). To make ADCP measurements consistent with S-4 measurements, we compare flow about
1 m above bottom. Stations 2–6 were seaward except for the
disturbances associated with an offshore tropical cyclone. Station
2 was located in a channel with relatively small curvature. The
Author's personal copy
ARTICLE IN PRESS
18
H.E. Seim et al. / Continental Shelf Research 29 (2009) 15–28
SAT2 40hlp Bottom Currents (Flood > 0)
0.2
S1
N1
S2
N2
S3
m/s
0.1
0
−0.1
Sta 2
Sta 3
Sta 4
Sta
2554 (Side) 260
Sta 5
Sta 6
−0.2
250
265
270
275
Year Day from 1 January 1999
280
285
290
285
290
SAT2 40hlp Bottom Currents (Flood > 0)
0.2
S1
N1
S2
S3
N2
m/s
0.1
0
−0.1
−0.2
250
Sta 3
Sta 4 (Side)
255
260
265
270
275
Year Day from 1 January 1999
280
Fig. 2. Subtidal axial flow ðflood40Þ. All stations are compared in the upper plot. Only Stations 3 and 4a are compared in the bottom plot which shows intermittent
landward flow at Station 3 and a consistent landward flow at Station 4a.
other three stations were positioned where seaward flow occurred
from residual tidal eddies (described below). The strength of the
subtidal flow was modulated by the spring/neap cycle, but did not
reverse direction. The two exceptions occurred at Stations 3 and
4a. Subtidal flow at Station 3, opposite the cusp of a large
meander, had episodes of landward flow near the time of neap
tide, as previously observed (Seim et al., 2002; Blanton et al.,
2003). Station 4a, located in a side channel, had consistently
landward flow. The fluctuations in flow at Station 4a were
remarkably well correlated with those at Station 3 (Fig. 2a).
The magnitude of the observed subtidal velocities significantly
exceeds that due to river discharge. A rough estimate of the
momentum carrying cross-sectional near station 4, about midestuary, is 3.5 m deep times 1200 m across or A ¼ 4200 m2.
Discharge during the fall period shown in Fig. 2 was Q ¼
50 m3 s1 or less, and therefore the downstream velocity due to
river flow is 0:01 m s1 or less. Even during the spring freshet
when discharge peaked at 150 m3 s1 , Q =A is only 0:03 m s1 and
its signature is difficult to perceive in the observed subtidal
velocities (not shown).
Station 3 is located near the outside shore of a channel
meander where channel curvature is maximum. Given the link
between channel bends and the position of residual tidal eddies
(see below), the landward flow present at this station suggests
minimal influence of eddies at this location.
3.2. Lateral flow differences
Tidally averaged maps of the vertically averaged flow field (Fig.
3) show an array of residual eddies situated between the channel
meanders (Seim et al., 2006). The eddies are situated on either
side of prominent cusps on the inside bend of the channel
meanders. The counter-rotating flow of these eddies has been
cited as the cause of the sharp cusps on the inside bend of
meanders in many estuaries (Dronkers, 2005).
The tidally averaged axial flow through several cross-sections
conform to the presence of the tidal eddies and illustrate the
combined effect of density-driven along-channel exchange and
channel curvature on the cross-sectional distribution of the flow
(Fig. 4). While recent models (e.g. Valle Levinson et al., 2003)
indicate landward flow in the main (deeper or ebb) channel and
seaward flow on the side (shallower or flood) channel, this picture
is altered by the curvature-driven residual flow between channel
meanders. The most seaward cross-section (A36, lower right) was
situated in the large meander in the vicinity Station 3 and its
surface flow was within a clockwise eddy (partially shown in
Fig. 3) downstream of the cusp. This flow overlayed landward
subsurface flow which was strengthened substantially during
neap where it spread across much of the channel.
To the west around the bend (A45, middle right) a counterclockwise eddy promoted seaward flow near the inside of the
bend and augmented the landward flow towards the outside of
the bend, here filling most of the water column in the deeper
channel. Note the spreading of landward flow into shallower areas
at neap.
The third panel (A910, upper right) is located where the
channel begins to straighten and bottom depths are shallower and
more uniform. At spring tide, the flow was clearly separated into
two parts: landward flow throughout the water column on the
north side of the channel and seaward flow on the south side. The
landward flow spread laterally at neap to occupy the lower half of
the water column on the south side of the channel where there
was seaward flow present at spring tide.
The last two cross-sections (Fig. 4) are located near the cusps of
the next two meanders farther upstream. The first section (B56,
Author's personal copy
ARTICLE IN PRESS
H.E. Seim et al. / Continental Shelf Research 29 (2009) 15–28
19
Spring tide residual
31.01
Latitude
31
30.99
30.98
30.97
0.1 m/s
-81.64
-81.62
-81.6
-81.58
Longitude
-81.56
-81.54
-81.52
-81.56
-81.54
-81.52
Neap tide residuals
31.01
Latitude
31
30.99
30.98
30.97
0.1 m/s
-81.64
-81.62
-81.6
-81.58
Longitude
Fig. 3. Vertically and tidally averaged currents from the roving ADCP during spring (top) and neap (bottom) tide. Note the diverging flow of residual tidal eddies on either
side of the meander cusps and converging flow at the cusps (figure from Seim et al., 2006).
lower left) was situated near the downstream edge of a clockwise
rotating residual eddy. The flow on the south side of the channel
was landward. This result is supported by current data at Station
4a which also indicated persistent landward flow in this channel.
Seaward flow was measured in the deeper channel on the north
side, consistent with the time series measurements at Station 4
(Fig. 2). At neap, seaward flow seemed to overtop the landward
flow in the southern channel.
The final section (B1011, upper left) is close to the next cusp
where seaward flow along the south side to the west of the cusp
crossed over from the south to the north shore. During spring tide,
landward flow is confined to the northern half of the section.
During neap, landward flow occupied the entire lower layer across
much of the channel and seaward flow occupied almost the entire
channel in the surface layer.
The structure of the tidally averaged exchange flow appears
consistent with the residual tidal eddies depicted in Fig. 3.
Rather than conform simply to landward/seaward exchange in
deep/shallow channels (Valle Levinson et al., 2003), the lateral
structure of the exchange flow appears to be governed to a large
extent by the residual tidal eddies (Dronkers, 2005). The
additional vertical stratification present at neap tides allows the
landward flow to spread laterally along the bottom so that it
occupies more of the cross-section than it does at spring tide but
is not capable of altering the depth-averaged flow pattern shown
in Fig. 3.
3.3. Axial distribution of salinity
Salinity was monitored at several stations (Fig. 5) along the
Satilla estuary. Freshwater discharge to the estuary occurred as
three main events. The first reached 150 m3 s1 in February,
consistent with the climatological maximum in winter. The other
two were related to the passage of two tropical cyclones on the
continental shelf.
The salinity response to discharge is clearly seen in the alongchannel salinity time series (bottom panel of Fig. 5). Note the
response of salinity to the discharge curve: salinity decreased
rapidly by up to 15 PSU over a 20-day period in response to the
February discharge event while the recovery period lasted 70
days. A similar rapid decrease in salinity was seen as a response to
the October discharge event. During both discharge events the
salinity response was quite pronounced at Stations 3–7 but was
much less obvious at Station 2. Station 2 was located in an area of
more open water (St. Andrews Sound) and suggests that a
fundamental change in the dynamics may occur somewhere
between Stations 2 and 3. The muted response of the salinity at
LT1 to the July discharge event may indicate the change in
dynamical regime is landward of its location (between Stations 2
and 3).
The axial–vertical distribution of salinity, as seen in high water
surveys from a wide range of conditions, shows changes in the salt
regime typical of the response of a partially mixed estuary to
Author's personal copy
ARTICLE IN PRESS
20
H.E. Seim et al. / Continental Shelf Research 29 (2009) 15–28
Section A910: Spring
0
Depth (m)
-2
A36
A45
A910
-4
-6
-8
B1011
-10
B56
0
30.99
Section A910: Neap
Depth (m)
-2
-4
-6
-8
0
Depth (m)
-2
30.99
12
-4
14
16
-2
Depth (m)
-8
30.98
Section B1011: Neap
-4
-6
0
-8
-2
-10
-4
30.99
Section A45: Neap
0
-6
Depth (m)
-2
-8
0.001
-10
30.98
-4
-6
-8
9
10
11
12
13
0.001
14
-10
30.99
13
Section B56: Spring
19
21
0
Depth (m)
Depth (m)
17
-2
-4
-6
-4
-6
-8
-8
-10
-10
30.99
30.97
Section A36: Neap
Section B56: Neap
0
-2
Depth (m)
-2
Depth (m)
15
Section A36: Spring
-2
0
20
0
-6
0
18
Section A45: Spring
-10
Depth (m)
0.001
-10
Section B1011: Spring
-4
-6
-8
12
14
0.001
-10
30.99
30.97
13
-6
-8
0.001
-10
-4
15
16
17
13
15
17
19
21
Fig. 4. Tidally averaged axial velocity cross-sections combined with tidally averaged vertical and surface distributions of salinity, each grouped by spring and neap tide
pairs. The map (upper left) shows the location of each cross-section, in each case north is to the right of the section plots. The light shade pattern denotes seaward flow; the
heavy dotted pattern denotes landward flow. Each tick in latitude is 0:001 . Some artistic license has been used to fill in data near the bottom where ADCP data are
unreliable. Contour interval ¼ 0:1 m s1. Salinity colorbars differ for each cross-section, but are common for each spring–neap pair. Circles mark locations of moorings.
changes in freshwater discharge (Fig. 6). At 25 km from the
ocean the salinity in the estuary varied between 0 and 22 PSU, and
surface-to-bottom salinity difference at high water varied be-
tween 1 and 3 PSU throughout the year. It is significant that
stratification exists at the end of flood, implying that tidal
straining during each flood tidal cycle is unable to mix away the
Author's personal copy
ARTICLE IN PRESS
H.E. Seim et al. / Continental Shelf Research 29 (2009) 15–28
SATILLA - Discharge - 1999
160
140
Discharge (m3 s-1)
120
100
80
60
40
20
0
Jan Feb Mar Apr May Jun
Jul
Aug Sep Oct Nov Dec Jan
SATILLA - Salinity - January to December 1999
35
20d
70d
LT1
30
2
LT1
Salinity (PSU)
25
20
2
3
4
3 Alternate LCW
5
15
4 Baileys Cut (west)
6
10
5 Crows Harbor
5
6 Ceylon
7 White Oak Creek
6 LT2 - Woodbine
0
Jan Feb Mar Apr May Jun
Jul
Aug Sep Oct Nov Dec Jan
1999
Fig. 5. Evolution of the salinity regime during 1999 in the Satilla River estuary. The
top panel is river discharge measured 140 km inland at Atkinson, GA. The bottom
panel graphs the salinity data from several recording salinometers during the two
monitoring periods as well as a long-term monitor at Station LT1. See Fig. 1 for
station locations. Note that the salinity decreased in response to the spring
discharge event over a 20-day interval, while the recovery period to higher salinity
lasted 70 days, closely in step with the discharge curve.
stratification, and is borne out in the individual profiles collected
in the deep channels as part of the roving surveys.
The spring and autumn salinity regimes are compared by
plotting against river discharge the distance of the 10-PSU
isohaline from the ocean and the axial salinity gradient at three
station pairs (Figs. 7 and 8). During spring, the 10-PSU line
responded to the freshet by moving 7 km seaward. Most of this
excursion was accomplished in about 3 days, and 10-PSU water
remained about 16 km from the ocean for the next 25 days, even
though discharge decreased almost five-fold. The landward
excursion during Days 68–72 was not accompanied by any
obvious event in water level or wind. However, the neap tide
during this interval was one of the smallest of the study period.
The maximum axial salinity gradient (ds=dx) was located
between Stations 3 and 4, about 15 km from the ocean throughout
the spring observation period. Its strength was remarkably
constant at around 1:6 103 PSU m1 with a slight increase
around the time of maximum discharge. The corresponding
gradients defined by the surrounding stations remained between
0.5 and 1:0 103 PSU m1 . Only the most seaward pair (Stations
21
2 and 3) showed a response to changing discharge. There was no
obvious correlation of either the position of the 10-PSU isohaline
or the axial salinity gradient with the spring–neap cycle (Fig. 7).
The autumn regime (Fig. 8) began with extremely low
discharge, a reflection of summertime drought conditions. It
persisted into the fall sampling period for about 17 days before the
freshet associated with Hurricane Floyd that began on Day 270.
This freshet was only one-third the volume rate of the spring
discharge. The isohaline excursion at Day 259 was associated with
the passage of Hurricane Floyd, which raised water level 0.4 m in
the estuary. During the time before the freshet, the position of the
10-PSU isohaline hovered around 30 km inland. During and after
the freshet, the position moved 10-km seaward with a final
location about 20 km from the ocean, compared with an average
position of about 16 km during the spring freshet.
The maximum ds=dx observed in autumn was positioned
between Stations 4 and 5, about 6 km farther landward than in the
spring and was about 1:0 103 PSU m1 before the freshet.
Afterwards, ds=dx increased to 1:5 103 PSU m1 which was
about equal to the maximum gradient observed in spring. The
strength remained at this level even as discharge began to
decrease to the same magnitude observed in late spring. The
gradient between the seaward pair also increased in response to
the freshet. Again, there is no obvious response of the salt
intrusion to the spring/neap cycle.
In summary, the maximum observed ds=dx was about 1:5 103 PSU m1 for discharges between 50 m3 s1 (autumn) and
150 m3 s1 (spring). Freshets during both seasons, even though
they were of different magnitudes, moved the 10-PSU isohaline
about 10 km seaward. The gradient stayed remarkably constant in
spring even during the low discharge periods at the beginning and
end of the spring observations. At the beginning of the autumn
study period, there was a smaller salinity gradient of 1:0 103 PSU m1 but ds=dx increased to 1:5 103 PSU m1 during
and after even this small freshet. A consistent and surprising
result is that the near bottom salinity intrusion does not exhibit a
response to the spring/neap cycle of variation in subtidal
circulation.
3.4. Cross-sectional distribution of salinity
Though not as complete as for the velocity field, a representation of the tidally averaged salinity was constructed based on
vertical salinity profiles near the channel sides and continuous
surface salinity across the channel (Elston, 2005). The tidally
averaged salinity is superimposed on the tidally averaged alongchannel velocity sections in Fig. 4 to demonstrate the relationship
between the flow field and mass field. During neap tides, top-tobottom difference in salinity in the thalweg was approximately
4 PSU. Moreover, the water was stratified even in the shallows by
3 PSU over depths less than 4 m. During spring tide, weaker
stratification of about 1:5 PSU was found in the thalweg. The
difference decreased to almost zero in the shallows, but there was
a significant lateral salinity change of 2 PSU across the channel.
Thus, the spring tide regime was characterized by lateral salinity
gradients with vertical gradients confined to the main channel.
During neap, the salinity regime shifts; significant lateral salinity
gradients across the channel remain but strong vertical gradients
are present in both deep and shallow parts of a cross-section.
The cross-channel distribution of salinity is influenced by the
curvature-induced secondary circulation. At neap tides the
seaward flow, always surface trapped, carries the freshest water
in the cross-section, switching from side to side of the channel as
it moves downstream (Fig. 4). During spring tides surface water is
significantly saltier, little vertical stratification exists, and lateral
Author's personal copy
ARTICLE IN PRESS
22
H.E. Seim et al. / Continental Shelf Research 29 (2009) 15–28
Fig. 6. Axial sections of salinity at high water in the Satilla River estuary during 1999. Section dates from top to bottom: 21 January, 05 February, 09 March, 09 September,
and 18 October.
salinity differences persist. Unlike neap tides when salinity
differences of 5 are common, salinity variability in a cross-section
at spring tide is considerably less.
The best spatial representation of the salinity field can be
created for the surface from the roving samples. Removing the
mean along-channel salinity gradient from the surface salinity
accents cross-channel differences (Fig. 9). While the data are
noisy, tidally averaged cross-channel differences in salinity reach
as high as 1 at spring tide and 2 at neap. The cross-channel
salinity gradient changes sign, depending upon the sense of
curvature in a meander. Minimum surface salinity is most often
observed in the main channel, consistent with the circulation
pattern represented by the counter-rotating residual eddies.
Maximum surface salinity is observed in the shallower side
channels though the sampling of this branch of the circulation is
more limited because the shallows prevented sampling throughout the tidal cycle at these sites.
The vertical structure of the salinity is most readily appreciated
as a series of profile plots which show the tidal cycle average
salinity profile for each station (Fig. 10). At each cross-section
there are profile stations on either side of the meandering
channel, where vertical profiles were collected over a tidal cycle.
The southern station is plotted in red, the northern station in blue.
Note that the range of salinity shown in each subplot increases in
the seaward direction. It is uncommon to observe in the
instantaneous profiles the same salinity profile at the 2 sites in
a cross-section; in a tidal cycle average this results in a nearly
constant offset of up to 3 PSU over the entire water column as
typical. The offset between stations in a given cross-section tends
to be the smallest at the surface during spring tides, indicating the
surface plots (Fig. 9) may under-represent the lateral structure
during large tidal range.
Greater salinity can be either on the north or south side of the
channel and can be related to the residual eddy field induced by
the channel bends (Fig. 3). The sections with landward residual
flow (flood) channels to the south (A36, A45, B56, and B78)
consistently indicate greater salinity in the side channels. Those
sections with seaward residual flow (ebb) channels to the south
(A910, C34 and C910) tend to show lower salinity in the main
channel. The occurrence of higher salinity near the bottom in the
main channels is consistent with curvature-induced secondary
circulation. Stations B1011 and C12 appear to be on top of the
transition zone between clockwise and counterclockwise rotating
residual eddies as they show both types of behavior.
Vertical stratification is much more pronounced at neap tides,
even in water as shallow at 3 m. For the sections in domain A
vertical gradients in tidally averaged profiles at neap exceed 1 m1
and display little indication of any boundary layers; individual
profiles exhibit in some instances thin bottom boundary layers
(o2 m) and remarkable salinity structure—in some cases 8 PSU
change in 5 m of water. Vertical stratification is greatly reduced at
spring tides; individual profiles are characterized by large regions
of well-mixed fluid and well mixed profiles are common for those
collected in less than 5 m depth. There is also a clear tendency for
stratification to decrease moving upstream. It is worth noting that
the tidal current amplitude increases in the upstream direction
Author's personal copy
ARTICLE IN PRESS
H.E. Seim et al. / Continental Shelf Research 29 (2009) 15–28
23
Satilla River Discharge − Spring 1999
m3 s−1
150
100
50
0
20
30
40
50
60
70
80
10 − PSU Intrusion Distance
30
km
25
S
N
20
S
N
N
S
N
S
60
70
N
S
N
50
Year Day 1999
60
70
15
20
30
40
50
80
PSU m−1 × 103
Axial Salinity Gradient
2
1.5
1
0.5
N
0
N
S
20
30
S
40
S
80
Fig. 7. The intrusion distance of the 10-PSU isohaline and the axial salinity gradient compared with spring discharge of the Satilla River. The dotted line in the second graph
is based on the relationship of Monismith et al. (2002). The bold line for the gradient is calculated from salinity at Stations 3 and 4. The dash-dash and dash-dot lines are
calculated from salinity data at Stations 2 and 3 and Stations 4 and 5, respectively. See Fig. 1 for station locations.
Satilla River Discharge − Autumn 1999
m3 s−1
60
40
20
0
250
255
260
265
270
275
280
285
290
285
290
285
290
10 − PSU Intrusion Distance
35
km
30
25
S
N
S
N
S
20
250
255
260
265
270
275
280
PSU m−1 × 103
Axial Salinity Gradient
2
1.5
1
0.5
0
250
S
N
255
260
S
265
N
270
275
Year Day 1999
S
280
Fig. 8. The intrusion distance of the 10-PSU isohaline and the axial salinity gradient compared with autumn discharge of the Satilla River. The dotted line in the second
graph is based on the relationship of Monismith et al. (2002). The bold line for the gradient is calculated from salinity at Stations 4 and 5. The dash-dash and dash-dot lines
are calculated from salinity data at Stations 2 and 4 and Stations 5 and 6, respectively. See Fig. 1 for station locations.
Author's personal copy
ARTICLE IN PRESS
24
H.E. Seim et al. / Continental Shelf Research 29 (2009) 15–28
detrended salinity − spring
31
30.995
30.99
30.985
30.98
30.975
30.97
30.965
−81.59
−1.5
−81.58
−81.57
−1
−81.56
−0.5
−81.55
−81.54
0
−81.53
0.5
−81.52
1
−81.51
1.5
detrended salinity − neap
31
30.995
30.99
30.985
30.98
30.975
30.97
30.965
−81.59
−1.5
−81.58
−1
−81.57
−81.56
−0.5
−81.55
0
−81.54
0.5
−81.53
−81.52
1
−81.51
1.5
Fig. 9. Detrended surface salinity from roving surveys at spring (top) and neap (bottom) tide. Higher salinity tends to be found opposite the cusps in the upstream part of
the diverging flow of counter-rotating eddies (Fig. 3). The high-salinity zone in the lower figure (downstream of the right-hand cusp) is in a flood-dominated tidal channel
separated from the main channel by a shoal, a zone that was only accessible throughout the tidal cycle at neap tide.
over the area under study (Seim et al., 2006) and that the
decreasing stratification may reflect the increased tidal energy
available for mixing.
When including all cross-sections and calculating the average
horizontal and vertical density difference over all tidal cycles
sampled, this difference amounts to 1 kg m3 or greater throughout the first 25-km of the estuary. The mean differences are the
greatest during neap tides in the lower reaches of the estuary that
were sampled. The ratio of horizontal to vertical density
differences is near unity on neap tides but doubles on spring
tide. These observations indicate a persistent cross-channel
density structure that is increasingly pronounced at spring tides.
4. Discussion
Data presented here suggest that channel curvature can
strongly influence the circulation field, mass field, salt flux, and
Author's personal copy
ARTICLE IN PRESS
H.E. Seim et al. / Continental Shelf Research 29 (2009) 15–28
25
c910
31
a45
30.99
b1011
a36
a910
c34
30.98
c12
b78
30.97
b56
30.96
30.95
81.6
81.59
81.58
spring c12
81.57
81.56
81.55
spring b78
81.54
81.53
spring a910
spring a36
0
0
0
0
0
1
1
1
1
1
2
2
2
2
2
3
3
3
3
3
4
4
4
4
4
5
5
5
5
5
6
6
6
6
6
7
7
7
7
8
20.5
21
21.5
8
24
25
26
8
12
27
spring c34
7
8
14
8
16
spring b1011
15
20
0
0
0
0
1
1
1
1
2
2
2
2
3
3
3
3
4
4
4
4
5
5
5
5
6
6
6
6
7
7
7
7
8
23.5
8
10
8
12
24
24.5
12
14
8
14
15
16
15
20
neap a910
neap b78
0
0
1
1
1
2
2
2
3
3
3
4
4
4
5
5
5
6
6
6
7
7
7
8
12
8
0
14
20
neap b56
0
0
1
1
1
2
2
2
3
3
3
4
4
4
5
5
5
6
6
6
7
7
7
8
10
8
12
14
15
20
neap a45
0
12
neap a36
8
15
16
neap b1011
20
spring a45
spring b56
SPRING
81.61
NEAP
81.62
spring c910
8
14
16
15
20
Fig. 10. Tidally averaged salinity profiles collected during the roving sampling from the north (blue) and south (red) side of the estuary. The panels are arranged from left to
right to match the map layout (top) from spring tide (middle) and neap tide (bottom) conditions. For each panel, the x-axis is salinity (PSU) and y-axis is depth (m).
salt balance in an estuary. A thorough investigation of the nature
of each of these influences is beyond the scope of this effort, but a
cursory examination of the dynamics involved in each of these
process is undertaken below.
Considering first the circulation field, the cross-channel surface
gradients associated with the curving flow drive a tidally averaged
barotropic along-channel residual circulation between the bends
that creates the counter-rotating eddies between channel
bends. The magnitude of the surface slope can be estimated as a
steady balance between the cross-channel surface slope and the
centrifugal acceleration:
g
qZ u2
¼
;
qy
R
DZ B u2
g R
Author's personal copy
ARTICLE IN PRESS
26
H.E. Seim et al. / Continental Shelf Research 29 (2009) 15–28
where R is the radius of curvature of the bend and Z is the surface
elevation. Using a channel width B of 1000 m, R ¼ 1000 m, and
U ¼ 0:5 m s (a reasonable rms value) gives DZ ¼ 4 cm. The
magnitude of cross-channel setup is large enough to be observed
with conventional sensors and should be readily confirmed.
The residual along-channel current driven by the crosschannel setup can be considered a balance between alongchannel surface slope and bottom friction:
g
qZ 1
¼ t ;
qx h b
tb C d u 2
Assuming the sections between bends are 2 km long, the mean
depth h ¼ 4 m and a drag coefficient of C d ¼ 0:002 (Seim et al.,
2002) suggests the residual flows should be roughly 10 cm/s,
consistent with the observed values. Obviously, the magnitude of
the flow will be modulated by variations in tidal amplitude, but
this should be readily re-produced in numerical simulations that
adequately resolve the cross-channel flow and the channel
geometry. Thus, the flow features of the residual eddies require
explicit recognition of channel curvature (centrifugal accelerations) and cross-channel variations in flow.
With regard to the impact of the curving flow on the mass field,
we observed persistent cross-channel salinity differences of
2–4 PSU and persistent vertical stratification. The latter is
surprising considering the shallowness of the system and the
strength of the tidal currents. One possibility is that the crosschannel salinity differences and cross-channel flow induced by
the channel bends help maintain the vertical stratification. We
estimate the importance of cross-channel processes through
examination of a simplified stratification equation (neglects
vertical advection and lateral dispersion of stratification)
q 2
g qr qu
g qr qv
q
qN2
N ¼ Kz
qt
r qx qz
r qy qz qz
qz
where N2 ¼ ðg=rÞqr=qz is the buoyancy frequency and K z is the
vertical eddy diffusivity assumed to be equal to the vertical eddy
viscosity. We observe persistent stratification, suggesting a quasisteady balance between production and destruction; assuming
the first term vanishes, the destruction of stratification embodied
in the last term may be balanced be either one or both of the
remaining terms on the left hand side. The magnitude of each
term is estimated from observations as follows: tidally averaged
u; v 0:1 m s1 ;
qr=qx ¼ 103 kg m4 ;
N 2 ¼ 102 s2 ;
Kz 102 m2 s1 . Assuming a vertical length scale of 5 m, the alongchannel production term is 2 107 N m3 whereas the destruction term is 4 106 N m3 . The scaling suggests gravitational
circulation alone is unable to maintain the observed stratification.
Assuming a cross-channel length scale of 500 m, qr=qyDr=Dy 2 kg m3 per 500 m ¼ 4 103 kg m4, the cross-channel production term is 106 N m3 , the same order of magnitude as the
destruction term and a factor of five greater than the alongchannel production term. Hence it appears that the variability and
structure of the mass field is largely controlled by the channel
curvature.
4.1. Effects of residual eddies on the axial salinity gradient
In the residual eddies the landward flows are found to be
correlated with higher salinity and seaward flows with lower
salinity that should provide an efficient mechanism for moving
salt upstream. The cross-channel salinity difference of 2–4 PSU
combined with the residual currents, estimated at 0:1 m s1 ,
suggest upstream salt flux u0 s0 driven by this mechanism alone
should be roughly 0:220:4 PSU m s1.
The changing orientation of the bends generates a pattern of
counter-rotating residual eddies and the communication between
eddies requires lower and higher salinity water to switch to
opposite sides of the channel. The communication between eddies
occurs at the bends themselves. The existing observations allow a
qualitative description of this flow feature, but do not lend
themselves to calculation of the salt flux on the bends; we
therefore make some inferences about the overall characteristics
of the salt balance through the use of simple formulations then
relate these to the existing observations.
4.2. Longitudinal dispersion
Steady-state theory for estuarine salt exchange holds that salt
advected seaward by river discharge is balanced by upstream
dispersive processes; a wide variety of decompositions of the
processes responsible for the upstream flux have been proposed.
There is also recognition of the time-dependence of the salt
balance and its implications for estimating the importance of the
various processes (Banas et al., 2004). We here take a very simple
approach to calculate a representative axial dispersion coefficient
under the range of axial salinity gradients observed in the Satilla;
obviously, much more sophisticated approaches are possible. The
stability of the salinity intrusion length over month-long time
scales during decreasing discharge is assumed to justify a steady
balance, and we choose to define a single horizontal dispersion
coefficient to represent all processes responsible for the upstream
movement of salt. If the influence of time dependence is similar to
that seen in Banas et al. (2004) then we would expect the range of
estimates of diffusivity based on a steady-state balance to be
greater than estimates made that include time dependence. This
balance is succinctly expressed by the following simple onedimensional model:
Kx ¼
RS
(1)
AqS=qx
where R is river discharge, S is the salinity in the domain, and A is
the cross-sectional area. We use data from sections in the ‘‘A’’ and
‘‘B’’ domains (Fig. 4), collected when R ¼ 20 m3 s1 , and apply
Eq. (1) to the results (Table 2). The ‘‘C’’ sections were collected
under extremely low discharge (R ¼ 5 m3 s1 ), and the axial
salinity gradient was the same order as the neap-tide value
during springtime sampling, producing K x 30 m2 s1 .
The axial salinity gradient (qS=qx) was smaller at spring tide,
resulting in K x 100 m2 s1 . During neap, K x 50 m2 s1 , roughly
one-half the value at spring tide. The range observed here
(302120 m2 s1 ) is consistent with other studies (see, for
example, West et al., 1990; Ralston and Stacey, 2005).
Table 2
Dispersion estimates based on Eq. (1)
Section X (km) Area ðm2 Þ Salinity
Spring Neap Spring
A34
13
A910
B56
15
17
B1011
C12
20
20
C910
24
K x ðm2 s1 Þ
Sx ðm1 Þ
Neap
Spring Neap
5200
17.9
14.4
5:45 104 1:0 103 120
3600
16.8
13.5
12.4
13.1
8:2 104
1:1 103 80
60
2900
11.0
18.8
9.9
–
1:0 103
–
–
14.6
–
30
50
Sections in the A and B domain were done during a discharge of 20 m3 s1 , while
those in the C domain were done during a discharge of 5 m3 s1 . ‘‘X’’ is distance
from the ocean. Sections ‘‘A’’ and ‘‘B’’ were done at spring and neap tide between
10 and 20 March 1999. The ‘‘C’’ section was done during a spring tide on 10
September 1999. Hurricane Floyd prohibited obtaining neap tide sections for
comparison. See Fig. 4 for location of sections.
Author's personal copy
ARTICLE IN PRESS
H.E. Seim et al. / Continental Shelf Research 29 (2009) 15–28
27
An estimate of the total upstream salt flux is simply
u0 s0 ¼ K x qS=qx. For the range of axial salinity gradients observed,
estimates are 0:01520:18 PSU m s1 . Given the magnitude of the
salt flux driven by the residual eddies alone (estimated as
0:220:4 PSU m s1 above) it would appear that the upstream salt
flux is limited by processes acting at the bends in the crossover
regions.
A possible explanation for the blockage of upstream salt flux at
the bends is connected to the morphology surrounding the cusps
of meanders, analogous to that found in meandering rivers (Ritter
et al., 2002). The bends exhibit a specific geometry of detached
point bars downstream of the cusp of the bend that merge with
mid-channel bars which separate the channel into a main and side
channel. Point bars at the cusps are displaced seaward, and these
bars constrict the width and depth of the flood channels. This
geometry favors the transfer of lower salinity water between
eddies because it is surface trapped and because it follows the
main channel of the estuary. It discourages transfer of higher
salinity water between eddies because the flow is bottom trapped
and must leap-frog over the point bars and because it follows the
shallower side channel that will induce more vertical mixing.
The blockage of upstream flow occurs on a scale presumably
limited to the eddy size (12 the meander wave length), in this case
2 km along the axis of the estuary. A portion of the salt is simply
recirculated on this scale and lowers the amount of salt that can
be passed farther upstream. We suggest that recirculation of salt
in the residual eddies accounts for the relatively slow increase of
salinity after an episode of river discharge. In other words, while
an event of river discharge rapidly pushes salinity downstream,
the return of salinity as discharge slacks is significantly slower in
meandering channels of estuaries where channel curvature
induces the recirculating eddies (Figs. 7 and 8).
circulation in channels of such strong curvature is robust enough
to set up spatially trapped eddies that efficiently transport water
and salt seaward and less-efficiently transport water and salt
landward.
Related to the lack of response of salt intrusion to the
spring–neap cycle, the time scales of adjustment for the
Satilla—estimated above as 20 days in response to an increase in
discharge and 70 days to return to pre-freshet conditions—are long
given the scales of the system. MacCready (1999) suggests that in
the diffusive limit the adjustment time varies at K x =u2 ¼ 106 s, of
order 10 days for the Satilla. Other estimators yield similar values.
The slow response of the system to change is consistent with the
notion that poor communication between the recirculation features
in the main body of the estuary limits the landward flux.
Lastly, it appears that the series of channel bands in the Satilla
River estuary act to trap the salinity gradient. The region of
maximum salinity gradient was consistently found in the region
of the bends over a range of discharge with weaker gradients both
landward and seaward. Interpreting the along-channel variations
in the salinity gradient in terms of varying horizontal dispersion
suggests the upstream movement of salt is a minimum in the
region of the bends. Seaward of the bends the channel broadens
rapidly and the topography becomes more complicated. This
region was much less responsive to increases in river discharge
and more akin to an open sea with more random dispersive
processes. Landward of the bends the channel narrows and
deepens and is characterized by a single depth maximum. Both
horizontal and vertical salinity gradients weaken, more in line
with predictions based on the geometric scales. This form of
estuary geometry is reasonably common in undisturbed (i.e. not
dredged) estuarine systems along the southeast US seaboard and
may act as a natural buffer to salinity intrusion.
4.3. Salt intrusion
5. Summary and conclusions
The observed ranges of salt intrusion as defined by X s for the
10-PSU isohaline are similar whether for the spring (Fig. 7, high
discharge) or fall (Fig. 8, low discharge). While the salt intrusion
distance (X s ) is theoretically related to Q by X s Q a where a ¼ 13
(Hetland and Geyer, 2004), data in San Francisco Bay yielded a
smaller exponent of a ¼ 17 (Monismith et al., 2002) suggesting a
significantly weaker dependency of salt intrusion on river
discharge. We have added to the intrusion sub-panels the
ð1=7Þ
equation X s ¼ kQ
with k ¼ 40 for the Satilla. This constant
results from a regression of log X on log Q for all collected mooring
data and represents the approximate distance in kilometers of salt
intrusion when Q approaches zero.
There is a reasonably close correspondence to the equation
when the 10-PSU isohaline is pushed seaward during a discharge
impulse. But the relationship overestimates the return landward
of this isohaline, suggesting that the circulation impedes its rate of
return landward. Thus, the asymmetry is described by the
observation that pulses of discharge move salt rapidly seaward,
but that the salt moves landward at a relatively slow rate when
discharge decreases.
There is no clear response of intrusion distance to the
neap–spring cycle. Tidally averaged cross-sectional flow (Fig. 4)
does indicate a stronger gravitational mode at neap tide. Note the
single event around Days 69–73 when discharge was steady and
low (Fig. 7). The 10-PSU isohaline travelled landward about 5 km
during the smallest neap tide of this study period when the
secondary circulation weakened to a level insufficient to provide
overturning. Thus, at times when secondary circulation is
sufficiently small, we expect the gravitational mode to set up
and efficiently push salt landward. But as a rule, the secondary
Our findings lead to a conceptual model that takes into account
the dynamics induced by channel curvature found in sinuous
channels of tidally forced shallow coastal plain estuaries. Tidal
flow in the Satilla around curving channels raises the water level
opposite the cusps of channel meanders and generates a series of
counter-rotating subtidal tidal eddies with convergent flow at the
cusps and divergent flow opposite (Fig. 3). The secondary
circulation at the bends tends to separate density classes across
the estuary; together with the counter-rotating eddies between
bends these flows promote and maintain lateral density gradients
in the central estuary. The geometry of the system is such that
where the system is wide enough to host multiple channels, the
surface-trapped (because of its reduced density) subtidal seaward
flow follows the deep channel and moves towards the sea
relatively unimpeded. The bottom-trapped (because it is dense
fluid) landward flow follows the shallower channel and is prone to
strong mixing at the cusps of the bends where its channel
shallows owing to mid-channel shoals that nearly attach to the
meander cusps.
The circulation of the conceptual model described above is
enhanced at spring tide when secondary circulation is strongest.
Weakened residual eddies and secondary circulation at neap tide
allows the setup of a stronger gravitational mode of estuarine
circulation that is surprisingly well developed even in the
shallows. Scaling the stratification equation indicates that such
strong stratification cannot be maintained by the gravitational
circulation alone, but requires the cross-channel production of
stratification largely controlled by channel curvature. A stronger
gravitational mode at neap is evidenced by the spread of landward
flow beyond the confines of the deeper parts of the channel. Even
Author's personal copy
ARTICLE IN PRESS
28
H.E. Seim et al. / Continental Shelf Research 29 (2009) 15–28
during prolonged periods of low river discharge, high salinity
water resists travelling inland over the normal energy range of
spring–neap tides (Figs. 7 and 8). Only during the weakest neap
tides is there some evidence that salt travels farther inland (Fig. 7)
in an enhanced gravitational flow. Thus, the role of secondary
(cross-channel overturning) and tertiary flow (bend-scale horizontal eddies) in this meandering shallow mesotidal estuary is to
retard the normal tidal mixing of higher salinity oceanic water
landward.
Several important questions remain. While we have shown
that the general regime of the salinity gradients flip from a
vertically stratified system at neap to a horizontally stratified
system at spring, our observations are inadequate for determining
which mode is more important for maintaining the salt balance.
To a large part, the strong horizontal gradients observed at spring
are a product of circulation in the strongly curving channels of the
Satilla River. Due to the limited dynamic range of river discharge
that we observed, we are unable to assess how changes in river
discharge affect the ‘‘flip-flop’’ in the fortnightly change in the
stratified regime. We would need more detailed surveys covering
a larger range of river discharge.
Acknowledgments
We wish to thank several people who made indispensable
contributions to this study. Julie Amft, Cheryl Burden Ross, Trent
Moore, and Guoqing Lin played critical roles in the design,
planning and execution of the field program; it could not have
happened without them. Anna Boyette at Skidaway Institute of
Oceanography contributed to the preparation of figures. The crew
of the R/V BLUE FIN (Jay Fripp, Raymond Sweatte, and Mike
Richter) of the Skidaway Institute of Oceanography gave us
dedicated enthusiasm and support in carrying out the field
experiments.
We gratefully acknowledge the following agencies who
supported the work described in this paper: the Georgia Coastal
Zone Management Program (Grant no. RR100-279/9262764),
National Science Foundation (LMER Grant no. DEB-9412089, LTER
Grant no. OCE-9982133, and SEI þ II Grant no. 0429644), NOAA
Coastal Ocean Program (Grant to South Carolina SeaGrant
Consortium entitled ‘‘Tidal Circulation and Salt Transport in a
Tidal Creek-Salt Marsh Complex’’), the Office of Naval Research
(the SEACOOS Program, N00014-02-1-0972) and the Department
of Energy grant to the Savannah River National Laboratory, SR06COL073.
References
Banas, N., Hickey, B., MacCready, P., 2004. Dynamics of Willapa Bay, Washington: a
highly unsteady, partially mixed estuary. Journal of Physical Oceanography 34,
2413–2427.
Blanton, J., Seim, H., Alexander, C., Amft, J., Kineke, G., 2003. Transport of salt and
suspended sediments in a curving channel of a coastal plain estuary: Satilla
River, GA. Estuarine, Coastal and Shelf Science 57, 993–1006.
Chant, R., 2002. Secondary circulation in a region of flow curvature: relationship
with tidal forcing and river discharge. Journal of Geophysical Research 107
(C9), 3131.
Dronkers, J., 2005. Dynamics of coastal systems. Advanced Series on Ocean
Engineering, vol. 25. World Scientific, Singapore.
Elston, S., 2005. Secondary circulation in a sinuous coastal plain estuary. Ph.D.
Thesis, Georgia Institute of Technology, 294pp.
Hetland, R., Geyer, W., 2004. An idealized study of the structure of long partially
mixed estuaries. Journal of Physical Oceanography 34, 2677–2691.
Li, C., O’Donnell, J., 1997. Tidally driven residual circulation in shallow estuaries
with lateral variation. Journal of Geophysical Research 102, 27915–27929.
Li, C., Valle-Levinson, A., Wong, K.C., Lwiza, K.M., 1998. Separating baroclinic flow
from tidally induced flow in estuaries. Journal of Geophysical Research 103,
10405–10417.
MacCready, P., 1999. Estuarine adjustment to changes in river flow and tidal
mixing. Journal of Physical Oceanography 29, 708–726.
Monismith, S., Kimmerer, W., Burau, J., Stacey, M., 2002. Structure and flowinduced variability on the subtidal salinity field in northern San Francisco Bay.
Journal of Physical Oceanography 32, 3003–3019.
Ralston, D., Stacey, M., 2005. Longitudinal dispersion and lateral circulation in the
intertidal zone. Journal of Geophysical Research 110, C07015.
Ritter, D., Kochel, R., Miller, J., 2002. Process Geomorphology, fourth ed. Waveland
Press, Long Grove, Illinois, USA.
Seim, H., Blanton, J., Gross, T., 2002. Direct stress measurements in a shallow
sinuous estuary. Continental Shelf Research 22, 1565–1578.
Seim, H., Blanton, J., Elston, S., 2006. Tidal circulation and energy dissipation in a
shallow, sinuous estuary. Ocean Dynamics 56, 360–375.
Smith, R., 1976. Longitudinal dispersion of a buoyant contaminant in a shallow
channel. Journal of Fluid Mechanics 78, 677–688.
Smith, J., McLean, S., 1984. A model for flow in meandering streams. Water
Resources Research 20, 1301–1315.
Uncles, R., Elliott, R.C.A., Weston, S., 1986. Observed and computed lateral
circulation patterns in a partly mixed estuary. Estuarine, Coastal and Shelf
Science 22, 439–457.
Valle Levinson, A., Reyes, C., Sanay, R., 2003. Effects of bathymetry, friction and
rotation on estuary-ocean exchange. Journal of Physical Oceanography 33,
2375–2393.
West, J., Uncles, R., Stephens, J., Shiong, K., 1990. Longitudinal dispersion processes
in the upper tamar estuary. Estuaries 13, 118–124.
Wong, K.-C., 1994. On the nature of transverse variability in a coastal plain estuary.
Journal of Geophysical Research 99, 14209–14222.