Regional Studies in Marine Science 39 (2020) 101467

Contents lists available at ScienceDirect

Regional Studies in Marine Science

journal homepage: www.elsevier.com/locate/rsma

Salt wedge intrusion modeling along the lower reaches of a

Mediterranean river
∗
Konstantinos Zachopoulos, Nikolaos Kokkos, Georgios Sylaios
Laboratory of Ecological Engineering and Technology, Department of Environmental Engineering, Democritus University of Thrace,
Vas. Sofias 12, 67100 Xanthi, Greece

article info a b s t r a c t

Article history: A three-dimensional numerical model was used to simulate the dynamics of a salt-wedge intruding
Received 25 November 2019 along the lower Strymon River mouth (northern Greece). The area is microtidal and under the summer
Received in revised form 22 August 2020 increased freshwater demand for irrigation, Strymon outflux at the mouth reduces to near zero. Forcing
Accepted 12 September 2020
at boundaries was provided a) from the e-HYPE hydrological model, at the upstream river boundary; b)
Available online 15 September 2020
from the TPXO tidal model, imposed at the open sea boundary, c) from CMEMS for water temperature
Keywords: and salinity open sea profiles, and d) from NOAA-GDAS for meteorological forcing. The model was
Strymon River calibrated with field measurements during summer 2003 and produced fairly reliable results of salt
ELCOM model wedge intrusion during summer 2004. The coefficient of determination for salinity reached 0.95; higher
Two-layer flow correlation was exhibited at the river upstream. Model results underestimated slightly the velocity and
Arrested salt wedge salinity along-channel observations. Salt wedge enters the lower river channel in late May 2004, under
Saline water intrusion limited river discharge (< 10 m3 /s) and intrudes up to 4.6 km upstream. The wedge length is controlled
by the low river flow, the microtidal domain and complex bottom topography. Logarithmic and power
law expressions were derived relating saline water intrusion length (L30 ) and Strymon River discharge
from validated model results. Retention time (RT) calculated based on model outputs varies in Strymon
salt wedge between 10 and 20 days. Future work includes the operationalization of Strymon salt wedge
model to produce reliable salinity and velocity forecasts along the lower channel.
© 2020 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND
license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

1. Introduction Hansen and Rattray Jr (1966) classification scheme by mapping

such flows according to the ratio between the freshwater in-
River mouths are ‘‘traditionally the terminal parts of a river flow and tidal dynamics, thus differentiating between perma-
and the regions where lighter river water meets denser sea water nent and time-dependent salt wedges, expressed at spring–neap
and intense mixing occurs’’ (Pritchard, 1967). Mixing, circula- timescales. In this classification scheme, Strymon River exhibits
tion and maintenance of vertical stratification between fresh and similar characteristics to Ebro River, in which the salt wedge is
salt water along the lower river channel is governed mainly by formed due to the weak freshwater forcing. Indeed, in Strymon,
river outflows as well as, forces induced by the action of tides, such regime remains established from early spring to late sum-
waves and wind (Monbet, 1992; Sierra et al., 2002; Hume et al., mer, when low buoyancy inflows prevail; being washed away
2007). In the micro-tidal Mediterranean environment, the low during the winter higher river flows (Haralambidou et al., 2010).
river discharge (during dry months), the lack of strong tidal cur- The dominant pattern is a sharp density interface, governed by
the sharp halocline, configuring the salt wedge. Its mean position
rents and the absence of tide-induced turbulence induces limited
is determined by the balance between the baroclinic pressure
vertical mixing at the fresh-to-salt water interface (Geyer and
gradient due to longitudinal density difference and the opposing
Ralston, 2011). Under such conditions, at the lower reaches of
river streamflow (Sargent and Jirka, 1987). Under steadily lim-
river channels with riverbeds below mean sea level, a bidirec-
ited river discharge conditions, an ‘‘arrested salt wedge’’ arises,
tional, two-layer flow system is established, characterized by an
characterized by an active upper fresh water layer and a stagnant
upper layer of thinner river water flowing downstream towards lower layer with higher salinity (Chanson, 2004). A number of
the river mouth over a lower layer of heavier saltwater advancing Mediterranean rivers meet the appropriate conditions for salt
upstream (Fischer et al., 1979; Ibaňez et al., 1997; Valle-Levinson, wedge formation and upstream intrusion, like the Ebro River in
2010). Geyer and MacCready (2014) expanded the traditional Spain (Ibaňez et al., 1997; Sierra et al., 2004), the Rhone river in
France (Ibaňez et al., 1997), the Neretva (Ljubenkov and Vranješ,
∗ Corresponding author. 2012), the Jadro (Ljubenkov, 2015) and the Rječina rivers in Croa-
E-mail address: [email protected] (G. Sylaios). tia (Krvavica et al., 2017b), and the rivers Louros (Scoullos and

https://doi.org/10.1016/j.rsma.2020.101467
2352-4855/© 2020 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-
nc-nd/4.0/).
K. Zachopoulos, N. Kokkos and G. Sylaios Regional Studies in Marine Science 39 (2020) 101467

Table 1
Characteristics of river mouths experiencing a salt wedge.
River Country River length Mean annual river Intrusion Tidal range Reference
(km) flow (m3 /s) length (km)
Ebro Spain 920 <320 >30 Microtidal Ibaňez et al. (1997); Sierra et al. (2004)
Rhone France 816 ∼1700 >30 Microtidal Ibaňez et al. (1997)
Neretva Croatia 225 355 ∼25 Microtidal Ljubenkov and Vranješ (2012)
Jadro Croatia 4.5 7.9 0.91 Microtidal Ljubenkov (2015)
Rječina Croatia 18 ∼10 <0.8 Microtidal Krvavica et al. (2017b,a)
Louros Greece 80 ∼20 n.a Microtidal Scoullos and Oldfield (1986)
Acheloos Greece 220 ∼140 ∼3 Microtidal Dassenakis et al. (1997)
Strymon Greece 392 ∼59.5 ∼5 Microtidal Haralambidou et al. (2010)
Mississippi U.S.A. 3730 ∼16700 190 Microtidal McAnally and Pritchard (1997)
Hillsborough U.S.A. 97 n.a ∼12 Mesotidal Chen et al. (2000)
Douro Portugal 897 ∼420 ∼6 Mesotidal Vieira and Bordalo (2000)
Yura Japan 146 ∼50 ∼15 Microtidal Kasai et al. (2010)
Fraser Canada 3000 120 >20 Mesotidal MacCready and Geyer (2001)
Merrimack USA 220 220 >6 Mesotidal Ralston et al. (2010)

Oldfield, 1986), Acheloos (Dassenakis et al., 1997) and Strymon in and dynamics of the Strymon River salt wedge, developing under
Greece (Parissis et al., 2001; Haralambidou et al., 2005, 2010). Salt low river flow conditions, using a three-dimensional hydrody-
wedge intrusion also occurs in other rivers outside the Mediter- namic model, calibrated and validated by field measurements.
ranean basin, with limited to higher effect of tidal action such as This research could consist a valuable tool in the hands of local
the Yura River in Japan (Kasai et al., 2010), the Fraser River in west authorities in order to handle the salt water intrusion, reducing
coast of British Columbia in Canada (MacCready and Geyer, 2001), the environmental and economic impacts on agriculture when
the Douro River in Portugal (Vieira and Bordalo, 2000). In the considering the effects of climate change. Understanding the pro-
U.S.A., the Merrimack river in north east coast of Massachusetts cesses controlling the formation and intrusion of this seasonally
(Ralston et al., 2010), the Hillsborough River (Chen et al., 2000) well stratified salt wedge could lead to the better management
and the Mississippi River, where the maximum length of the salt of upstream water resources. At a further step the model could
wedge was reported, reaching as far as 190 km (McAnally and become operational serving as a tool for the day-to-day salt
wedge forecast. The main novelty of this study lies in the use
Pritchard, 1997) (Table 1).
of freely-available hydrologic and hydrodynamic data extracted
In the Mediterranean arid-in-climate micro-tidal environment,
from existing reliable databases (like e-HYPE, CMEMS, TPXO, and
freshwater flow regulation from river damming, river water ab-
NCEP-GDAS), imposed as boundary conditions at the open bound-
straction and the overall water resources mismanagement and
aries of Strymon River lower channel. This tool could be easily
over-use by farmers leads to the summer decline in runoff fa-
replicable to other estuary environments, with only limitation the
voring salt wedge intrusion. As the wedge interacts with bottom accurate bathymetric mapping.
topography, downstream river dredging and sand/silt removal
and other hydraulic modifications affect salt water intrusion char- 2. Materials and methods
acteristics and the salt wedge shape upstream the river channel
(Graas and Savenije, 2008). The salt wedge formation along the 2.1. Study area description
lower river mouth triggers a series of adverse environmental
effects. The saline wedge intrudes not only longitudinally along Strymon is a transboundary river sharing its water between
the river mouth, but also laterally, salinizing coastal groundwater North Macedonia, Bulgaria and Greece. It is 392 km long, draining
and degrading nearby aquifers (Barlow, 2003). The long-term a catchment of 17,330 km2 . The river enters Greece through the
isolation of the lower saline layer prohibits the vertical transfer of Serres plain, and then follows a course of 118 km and eventually
dissolved oxygen to the bottom stagnant layer (Watanabe et al., outflows into the Strymonikos Gulf, in the Northern Aegean Sea
2014). Such event gradually leads to bottom anoxia, deteriorating (Fig. 1). The main controller of river discharge is the man-made
river water quality (D’Adamo and Lukatelich, 1985). Decline in biotope of Kerkini Lake located 77 km upstream of Strymon River
benthic dissolved oxygen level could be associated with changes mouth. Kerkini is a man-controlled reservoir storing the river wa-
in sediment redox and the consequent release of nutrients and ter for irrigation. Strymon has an annually-mean discharge of 59.5
trace metals (McAuliffe et al., 1993). Additionally, during tidal m3 /s; however, during the period of increased irrigation demand
floods the enhanced bottom velocities of the salt wedge induce (May to September) the lowland river flows almost zeros (Sylaios
et al., 2006). Recent findings illustrate that in the latest decades
local resuspension and suspended matter flocculation, favoring
the total freshwater input to the Strymonikos Gulf decreased by
the occurrence of a turbidity maximum at the tip of the wedge
about 30% due to the extended use of river water for irrigation
(Sierra et al., 2002; Haralambidou et al., 2010). Sudden salt water
and the lower precipitation (Sylaios et al., 2006).
intrusions, anoxic events and the presence of turbidity maximum
Such changes in the hydrologic regime of the river have led
at river bottom affect the abundance and distribution of benthic
to remarkable changes in the coastal circulation and renewal of
flora and fauna organisms along the river mouth (Sylaios et al., Strymonikos Gulf while favoring the occurrence of a salt-wedge
2006). intruding upstream the lower river channel (Haralambidou et al.,
Several approaches have been applied to simulate the behavior 2005, 2010). Sea water intrudes easily into the river, since the
and the dynamics of salt wedge in Mediterranean region; regres- riverbed lies below the sea surface, for a distance up to approx-
sion methods, empirical expressions, physical models, 1D, 2D and imately 9 km upstream from the river mouth (Haralambidou
more rare 3D numerical models (Ibaňez et al., 1997; Haralam- et al., 2003). The present study focuses on the lower river system,
bidou et al., 2010; Ljubenkov and Vranješ, 2012; Ljubenkov, 2015; characterized as micro-tidal, with tidal amplitudes ranging from
Krvavica et al., 2017a). Focusing on the Greek region, very limited 0.30 m in spring tide to 0.14 m in neaps. Mean water depth is
publications exist on the hydrodynamic modeling of salt wedge approximately 3.5 m, varying between 1.5 and 5 m, and the mean
propagation along Greek river mouths under limited discharge width of the river channel is about 64 m, varying from 40 to 90
conditions. The present research aims to simulate the behavior m along the channel.
2
K. Zachopoulos, N. Kokkos and G. Sylaios Regional Studies in Marine Science 39 (2020) 101467

aiding model calibration and validation. The total computational

domain consisted in total of 2,081,530 computational cells. The
vertical grid is flexible at sea surface to accommodate the sea
level changes due to tidal oscillations. The river mouth is defined
as the lower open boundary, where the open sea data of CMEMS
cell were spatially and temporarily interpolated and used as
boundary conditions.

2.3. Description of the hydrodynamic numerical model

The Estuary Lake and Coastal Ocean Model (ELCOM) is the

three-dimensional hydrodynamic model applied for the simula-
tion of Strymon River salt wedge intrusion (Hodges, 2000; Hodges
and Dallimore, 2006). The model has been developed by the
Center for Water Research of the University of Western Australia.
The model solves the time-varying Navier–Stokes equations, us-
ing Reynolds averages, as well as the transmission and diffusion
Fig. 1. The hydrographic network of the transboundary Strymon River and the equations using the Boussinesq approach, while eliminating the
study area (in red square) of Strymon River mouth. non-hydrostatic pressure conditions. The most interesting char-
acteristic of ELCOM model is the use of a mixed-layer model,
rather than an eddy-diffusivity model, to quantify vertical mixing.
This mixed-layer model is more appropriate in highly-stratified
flows, as the salt wedges. The mixing process is modeled on a
layer-by-layer basis, by comparing the available mixing energy
from convective overturns, shear production, wind stirring, and
stored mixing energy to the potential energy increase required
to mix a grid cell up into the mixed layer above itself (Hodges,
2000; Hodges and Dallimore, 2006). ELCOM numerical approach
applies a semi-interlace finite volume scheme on an Arakawa C
grid, implementing the TRIM solution (Casulli and Cheng, 1992),
with modifications for scalar conservation, numerical diffusion
and implementation of a mixed-layer turbulence closure. Con-
vective terms are calculated using a third order Euler–Lagrange
scheme, while the ULTIMATE-QUICKEST scheme is introduced for
the advection of scalars (Hodges and Dallimore, 2006).
The model reproduces adequately the dynamics of stratified
water bodies under external environmental forcing, such as tide,
winds, surface heat fluxes and river inflows (Hodges and Dal-
Fig. 2. Strymon River lower channel bathymetry and points of in-situ data
collection for model calibration and validation.
limore, 2006). Robson and Hamilton (2004) used the ELCOM
model to simulate the salt wedge dynamics of Swan River; Ran-
madugala (2004) modeled the wedge in Kelani River and Marti
and Imberger (2015) in the Swan-Caning Estuary. Therefore, this
2.2. Bathymetry and lower river channel discretization
model seems appropriate for the successful modeling of salt
In-situ river depth measurements were carried out during a wedge propagation along the Strymon River mouth.
hydrographic survey along the Strymon River mouth in 2003
and 2004 summer periods (Haralambidou et al., 2010). Mea- 2.4. Initial and boundary conditions
sured depth values were imported into a Geographic Information
System (Quantum GIS) and nearest neighbor interpolation was The 3D numerical model of the Strymon River mouth was
applied leading to the extraction of bathymetry along the lower executed with a time-step of 20 s to simulate in hindcast mode
river channel (Fig. 2). the hydrodynamics of two-years with intense field data (years
The computational grid was created using the Vertical Mapper 2003 and 2004). A ‘‘cold start’’ was initially imposed, with river
GIS toolkit, capable to produce grid files or continuous informa- salinity along the channel arbitrarily set to 0.1 psu and initial
tion surfaces interpolating data points and calculating the values freshwater temperature to 15.3 ◦ C. These conditions imply the
of grid cells between known points. The horizontal computational presence of freshwater throughout the river channel, conditions
domain of the applied three-dimensional model comprised of a representing early spring months (March–April) when the wedge
discretized rectangular grid of 254 × 149 horizontal cells with a is absent. After a warm up period of 2 years, adequate to establish
resolution of 20 m × 20 m (Fig. 3a). Such configuration allowed the initial flow and salinity fields, the model was initiated.
the accurate representation of estuarine complex bathymetry and Haralambidou et al. (2010) performed 11 sampling surveys
topography and the reproduction of the physical interchange and collected vertical profiles of hydrographic (temperature, salin-
characteristics of the river. For the vertical discretization, a grid ity, and velocity) and water quality data (nutrients concentration,
of 55 layers was created, covering the whole water column with dissolved oxygen content, turbidity and pH) during the summer
a thickness of 0.10 m each, aiming to resolve the sharp halo- period of years 2003 and 2004, along the Strymon river lower
cline interface of the wedge (Fig. 3b). This detailed grid was channel, up to 6.1 km upstream. Each survey involved three to
more appropriate to describe in higher accuracy the geometric four successive up-estuary transects along the channel’s center-
characteristics of the salt wedge interface and perform direct line, collecting data with 500 m horizontal spacing and ranging
correlations between model results and in-situ measurements, vertical increments from 0.1 to 0.25 m.
3
K. Zachopoulos, N. Kokkos and G. Sylaios Regional Studies in Marine Science 39 (2020) 101467

Fig. 3. Salinity profiles from ELCOM simulation (red lines) and field observations (black lines) during (a) the flood phase and (b) the ebb phase on August 18, 2004,
at four stations along the Strymon River: (i) at the sea boundary, (ii) 1 km, (iii) 2 km, and (iv) 3.1 km upstream of Strymon River mouth. (For interpretation of the
references to colour in this figure legend, the reader is referred to the web version of this article.)

In-situ data from year 2003 were used for model calibration, database (Lindström et al., 2010; Donnelly et al., 2016) and were
while data from year 2004 for model validation. Daily mean river imposed at the upstream river flow boundary. These data appear
discharge data were retrieved from the e-HYPE numerical model in agreement with the sparse river flow data of Haralambidou
4
K. Zachopoulos, N. Kokkos and G. Sylaios Regional Studies in Marine Science 39 (2020) 101467

Fig. 4. Velocity profiles from ELCOM simulation (red lines) and field observations (black lines) during (a) the flood phase and (b) the ebb phase on August 18, 2004,
at three stations along the Strymon River: (i) at the sea boundary, (ii) 1 km, and (iii) 3.1 km upstream of Strymon River mouth. Negative velocity values indicate
water flowing seawards while positive represent an upstream flow. (For interpretation of the references to colour in this figure legend, the reader is referred to the
web version of this article.)

et al. (2010) study (R2 = 0.76). Moreover, temperature and Table 2

Dates and observation points used for the ELCOM model validation.
salinity in-situ data of the sampling point 6.1 km upstream the
Dates Observation points
river mouth were linearly interpolated to fit the vertical grid
16 Jul 2004 a, b, c, d
(∆z = 0.1 m) of the model and used as boundary conditions in 25 Jul 2004 a, b, c, d, e
the upper open boundary. Tidal data from TPXO model (http:// 31 Jul 2004 a, b, c, d, e
18 Aug 2004 a, b, c, d
volkov.oce.orst.edu/tides/global.html) and temperature and salin- 28 Aug 2004 a, b, c
ity profiles from the Copernicus Marine Environmental Service a) near river mouth, b) 1 km upstream, c) 2 km upstream, d) 3.1 km upstream
(CMEMS, http://marine.copernicus.eu/) were imposed at the open and e) 4.1 km upstream.

oceanic boundary of the river mouth. Meteorological data (wind

speed and direction) were derived from the NOAA Global Data
2.5. Model calibration and validation procedure and criteria
Assimilation System (GDAS, https://www.ncdc.noaa.gov/data-acc
ess/model-data/model-datasets/global-data-assimilation-system- Model calibration involved the fine-tuning of the drag bottom
gdas) database and were imposed over the sea surface of the coefficient (CD ) and the horizontal eddy diffusivity coefficient (Nh )
channel. in 2003. More precisely, using the ‘‘trial and error’’ approach at
5
K. Zachopoulos, N. Kokkos and G. Sylaios Regional Studies in Marine Science 39 (2020) 101467

Fig. 5. Variability of (a) the salt wedge length (L30 ), in relation to (b) the tidal range (in m) at the open boundary, (c) the Strymon River discharge (in m3 /s) at the
upstream boundary, (d) the temperature and salinity change at the upstream, and (e) at the open ocean boundaries.

five points along the river channel, the model salinity vertical following equation:
profile results were related to the respective field measurements √
∑n
for summer 2003 (Haralambidou et al., 2010) (Fig. 2 and Table 2). i=1 (oi − mi )2
RMSE = (3)
The best fitted value for the horizontal eddy diffusivity coefficient n
was Nh = 0.6 m2 /s, for the drag bottom coefficient was CD =
0.002. 2.6. Model output post-processing
To assess the validity of model results with in-situ data during
model calibration and validation, a set of criteria was established Reported model outputs refer to the salt wedge intrusion from
and imposed on the modeled and observed salinity profile data: April to November 2004. The salt-wedge intrusion length L30 (in
(a) The slope (γ ), produced by the linear regression on the km), defined as the upstream distance of the 30 psu isohaline
scatter plot between measured and simulated values. The form from the seaward boundary and the salt wedge height H15,0 (in
of that line is: m), defined as the vertical distance from seabed of the 15 psu
isoline at the estuarine mouth, were determined (Seim et al.,
oi = γ · mi (1) 2009).

where mi and oi are the modeled and the observed values, re- 3. Results
spectively.
(b) The coefficient of determination (r2 ), assessing the inter- 3.1. Model validation
pretive power of the linear model, expressed as:
⎛ ⎞2 The model produced fairly reliable results describing the hy-
∑n
i=1 (o i − o i ) (m i − m i ) drodynamics of the river and the salt wedge intrusion during
r2 = ⎝ √ ⎠ (2) summer 2004. Evaluation criteria illustrated relatively satisfac-
∑n 2 ∑n
i=1 (o i − o i ) i=1 (mi − m)2
tory model performance at five different points along the river,
focusing at the salinity distribution in the water column. The
where n is the number of observations, oi and mi are the mea- salinity vertical profiles of the model were correlated with the
sured and simulated values, and oi and mi are the means of profiles of in-situ salinity measurements, in five points along the
measured and calculated values, respectively. river at the respective tidal phase of the sampling times (Table 3).
(c) The root of Mean Square Error (RMSE), expressing the Salinity r2 reached 0.95, when all records are considered, with
deviation between the simulated and the measured values. Such lower correlation at the river mouth. Upstream the river, correla-
deviations are called residuals, and RMSE serves to concentrate tion increases due to the formulation of the salt wedge. Maximum
them on a single predictive measure. The RMSE is given by the correlation is achieved at 3.1 km upstream the river mouth (r2 =
6
K. Zachopoulos, N. Kokkos and G. Sylaios Regional Studies in Marine Science 39 (2020) 101467

0.97; RMSE = 3.98 psu; slope = 1.09). Slope values (>1) indicate
slight model underestimation throughout the channel.
Indicative profiles of modeled/observed salinity (August 18,
2004; river discharge Q = 4 m3 /s; tidal range A = 0.30 m) are
shown in Fig. 3. The height of the 15 psu isohaline, varies by ± 0.5
m between tidal phases and upstream locations. The distribution
of salinity in the water column and along the river channel
has been described with high accuracy by the hydrodynamic
simulation. Minor declinations between simulation and in-situ
measurements are observed; the higher discrepancy recorded at
the surface layer (less than 5 psu). In the surface, the salinity
estimated by the model is around 5 psu and in observations
varies from 0 to 3 psu. In the bottom layers of the river mouth
observation point, salinity gets the maximum value (up to 37
psu). During the flood tidal phase, the height of the salt wedge
increases significantly (up to 30 cm), indicating the entry of sea
water upstream the river. Moreover, during tidal ebb stratifica-
tion becomes stronger and the height of the salt wedge decreases
by approximately 30 cm. Further upstream, the effect of the tide
appears negligible and stratification remains stable within the Fig. 6. Relation between salt-wedge length (L30 ) and river discharge (Q).
tidal cycle. Relatively high correlation is also shown comparing
in-situ flow observations to hydrodynamic model results. Fig. 4
presents indicative water flow data and model results for the 3.3. Strymon salt wedge length
same tidal cycles in August 18, 2004. Positive values represent
upstream directed velocities while negative designate the water L30 -index describes the length of the wedge in relation to the
flow from river to the sea. The model performed better in flood longitudinal mixing processes along the sharp wedge interface.
tidal phases, producing a surface freshwater layer (around 1 m The length of the salt wedge is strongly related to the river
thin) moving seawards with flow speed up to 0.18 m/s. Lower discharge and tidal dynamics. Fig. 5 presents the temporal change
layers propagate to the opposite direction with maximum veloc- in L30 in relation to the tidal and river flow conditions. During
ity of 0.17 m/s. The denser sea water enters the river from the the summer 2004 simulation the mean L30 -value reaches 3.6 km
lower layers pushing the salt wedge upstream the river, while upstream. However, a sharp rise in L30 was observed from July
fresh water flows to the opposite direction of the salt wedge. 30, 2004 to August 12, 2004 up to the maximum intrusion length
During ebb, the model produces seaward movement throughout of 4.6 km upstream. The salt wedge, driven by the tidal springs
the water column, having velocities up to 0.21 m/s at the surface prevailing on August 5, 2004, reached the highest L30 intrusion
layer. Overall, focusing on model result presented in Figs. 3 and length under low river flow conditions of about 3 m3 /s.
4, the upper layer of the river (about 1 m) with the lower salinity Minimum intrusion lengths (L30 ) in the summer were sim-
moves seaward and the lower layers move upstream, indicating ulated on June 19, 2004, July 01, 2004 and August 18, 2004,
the salt wedge intrusion. The model tends to underestimate (0.2 coinciding with a slight increase in freshwater discharge (> 3
to 0.4 m/s) the water velocity values at the river surface in both m3 /s) and tidal neaps (A ∼ 0.14 m). In these dates, L30 was limited
flood and ebb tidal phases. Moreover, slight underestimation in to 2.2 km upstream the river mouth (Fig. 5). In early September
bottom layer velocities (up to 0.2 m/s) is observed. The variation 2004 the saline intrusion length (L30 ) gradually declines as the
salt wedge is gradually flushed out of the river mouth. However,
in river discharge between the E-Hype model and field observa-
part of the salt wedge was separated from its main core and was
tions (Haralambidou et al., 2010); R2 = 0.76) could explain model
entrapped at the deeper parts of the riverbed. Such entrapment
flow underestimation, especially in the surface layers.
and presence of residual salt-wedge water was also observed
by Haralambidou et al. (2010), approximately 4 km upstream,
3.2. Strymon salt wedge formation attributed to complex bottom topography. Model results showed
that this water mass became eventually completely mixed during
Model results analysis was performed on the simulations cov- the late days of October 2004.
ering the summer 2004 (June to September). Salinity increases up The relationship between L30 produced by the validated model
to 37 psu at Strymon River mouth in early spring, attributed to and the variability in Strymon River discharge is best-approached
the seasonal thinning of surface Black Sea Water, the entry and by a logarithmic function (Fig. 6):
the vertical mixing with the more saline Levantine Intermediate
log L30 = 8.14407 + 0.04026 × Q − 0.01955 × Q 2 (4)
Water at the shallow parts of Strymonikos Gulf. This gradual
salinity rise in Strymonikos Gulf induces a higher horizontal This function describes better the L30 variability (R2 = 0.90)
density gradient, promoting salt intrusion along the river (Sylaios along the lower Strymon channel and removes the wedge under
et al., 2006). Salt water enters the river channel in late May when increased river discharge conditions (>16 m3 /s).
river discharge is diminished. Intrusion occurs during the flood Testing the power curve law, reported by many previous in-
tidal phase on May 30, 2004, acting against the river discharge of vestigators, a simpler function was derived, as:
10 m3 /s. In June the salt wedge establishes along the river chan-
L30 = 5120 × Q − 0.68 (5)
nel, intruding gradually further upstream during tidal springs. In
less than 10 days, saline water of 30 psu appears approximately The produced power curve is less successful in L30 -
3.6 km upstream the river due to the low river discharge (< 10 representation (R2 = 0.77). The power index (−0.68) appears
m3 /s). significantly lower than those reported by Krvavica and Ružić
7
K. Zachopoulos, N. Kokkos and G. Sylaios Regional Studies in Marine Science 39 (2020) 101467

Table 3
Statistical parameters for salinity validation along the Strymon River.
At the river 1 km 2 km 3.1 km 4.1 km
mouth upstream upstream upstream upstream
Number of observations 178 194 197 249 42
Slope (γ ) 1.07 1.13 1.12 1.09 1.13
R-Square 0.93 0.96 0.94 0.97 0.96
RMSE (psu) 5.88 4.42 5.30 3.98 4.23

Fig. 7. Longitudinal distribution of (a, b) water salinity, and (c, d) water flow speed during the spring tide on July 16, 2004 (Q = 2 m3 /s; A = 0.28 m). Graphs on
the left represent the flood tide and on the right the ebb tidal phase. The river boundary is located to the right of figure.

(2020) (exponent −2.03) and Geyer and Ralston (2011) (expo- upper layers of the salt wedge, indicating a slight retreat of salt
nents from −2.5 to −2.0), since these were derived for higher water from the salt wedge to the sea. The speed diminishes to
river flows (>250 m3 /s), signifying the importance of river dis- near zero values at the bottom layer, retaining the upstream flow
charge in highly-stratified flows. Ralston et al. (2010) reported an direction (Fig. 7b).
exponent of −0.19 for the mesotidal Merrimack River. Poggioli In neap tide (July 10, 2004; Q ∼ 2 m/s; A = 0.10 m), limited sea
and Horner-Devine (2015) noted that such simple power-law water volume enters the river channel, resulting in the reduction
models describe the physics of salt wedges in the absence of tidal of salt wedge height (H15,0 =∼1.3 m). Mixing is inhibited along
and topographic variation. the river mouth and the wedge seems better formulated with
Further, when the salt intrusion length L30 depends not only a sharp interface. In flood phase (Fig. 8a), higher flow speeds
on river discharge (Q), but also on the tidal variability (SWL), a directed from sea to the river are observed at the interface; at
more complex function was fitted: the river mouth (up to 0.085 m/s) and at 3.2 km upstream (up to
0.1 m/s). During ebb almost zero velocity values are reported at
L30 = 4647.04 × Q −0.53 + 43.14 × SWL (6)
the interface and flow reversal prevails at the area near the open
having R = 0.83 (n = 43). This function is directly comparable to
2 boundary (Fig. 8b).
that reported by Krvavica et al. (2016) for the microtidal Rječina
River Estuary in Croatia. 4. Discussion

3.4. Salt wedge intra-tidal and fortnight dynamics Strymon River mouth is a micro-tidal system characterized by
the presence of a salt-wedge propagating upstream during the
Salt wedge intrusion in Strymon River was studied in two tidal summer months when low freshwater flow prevails. Similar salt-
periods (spring and neap) and their two tidal phases (flood and wedge patterns have been reported along other Mediterranean
ebb) under low river discharge (Q ∼ 2 m3 /s). Fig. 7 shows the rivers. Indicatively, in Ebro river (Spain) the length of the salt
distribution of salinity and the total speed modeled in tidal spring wedge is observed up to 32 km upstream (Ibaňez et al. 1997;
along the longitudinal Strymon River transect (July 16, 2004; Q = (Sierra et al., 2002), in Neretva river salt water is observed over
2 m3 /s; A = 0.28 m). During the flood phase (Fig. 7a), sea water 23 km upstream the river mouth (Ljubenkov and Vranješ, 2012),
(∼37 psu) enters the river mouth, pushing the wedge upstream, in Jadro salt water intrusion appears over 1 km (Ljubenkov, 2015)
inducing increased vertical shear. The main body of the wedge and in Rječina intrudes just under 1 km (Krvavica et al., 2017b).
flows upstream reaching values up to 0.25 m/s, while the surface A three-dimensional hydrodynamic model (ELCOM) was con-
layer moves from the river to the sea (speed up to 0.20 m/s). figured to describe the hydrodynamic characteristics of the last
During ebb, the surface flow approximates 0.2 m/s at the mouth, 8 km of Strymon River flow. The model was calibrated and
moving seaward. Significant speed values are also observed at the validated based on previous field surveys (June to August 2003
8
K. Zachopoulos, N. Kokkos and G. Sylaios Regional Studies in Marine Science 39 (2020) 101467

Fig. 8. Longitudinal distribution of (a, b) water salinity, and (c, d) water flow speed during the neap tide on July 10, 2004 (Q = 2 m3 /s; A = 0.10 m). Graphs on
the left represent the flood tide and on the right the ebb tidal phase. The river boundary is located to the right of figure.

an amplitude of 0.30 m. The salt-wedge length (L30 ) fluctuated up

to 4.6 km and the salt-wedge height at the mouth (H15,0 ) up to
0.9 m between the maximum flood and the maximum ebb tidal
phase. Neap cycles illustrated negligible intra-tidal changes in the
salt wedge geometry.
Another factor that affects the salt wedge shape is the river-
bed morphology, where mixing processes are primarily driven
by bottom shear, with interfacial stress playing a secondary role
(Nepf and Geyer, 1996; Zhou, 1998; Trowbridge et al., 1999). In
Strymon River, two shallow submerged sills are formed at 3.6 km
and 4.6 km upstream of the river mouth, inhibiting salt wedge
propagation and limiting its intrusion up to the 4.6 km-limit.
Submerged sills blocking the salt wedge propagation were also
found in the Rječina River Estuary (Krvavica et al., 2017b).
To understand the water column characteristics of the lower
Strymon River channel, the nS indicator was produced based on
the results of the numerical model, expressed as (Tuin, 1991):
δS
Fig. 9. Mean variability of the stratification indicator nS along Strymon River ns = ′
(7)
lower channel on August 9, 2019. Sm
where δ S = Sbott − Ssurf , S’m = 0.5 × (Sbott + Ssurf ), with Ssurf
and Sbott the salinity at the surface and bottom of the water
and 2004) (Haralambidou et al., 2010). Model findings suggest column, respectively. The stratification indicator nS takes near
that salt wedge dynamics along the river channel are principally zero values at the riverine part of the channel due to the pres-
driven by Strymon River discharge. The salt wedge is observed ence of well-mixed water column. Along the remaining channel,
from June to September, intruding up to 4.6 km when the river from the seaward boundary until the salt wedge limit of 4.6 km
discharge in the river mouth area is low (Q < 12 m3 /s). As shown upstream, nS was consistently found above 1.0, ranging between
by the models (Eqs. (4)–(6)) introduced to describe the salt wedge 1.5 at the river mouth to 2.0 at the upper saline intrusion limit,
intrusion length (L30 ), river discharge plays a major role governing
indicating the presence of a stationary, well-stratified salt wedge
its dynamics, although during the summer period, when Strymon
(Fig. 9). In accordance to MacCready and Geyer (2001) in the
River flow is constantly low, the impact of tidal action in the
Fraser river, flood is the most productive period for vertical salt
length of wedge intrusion is observed. Similarly, in the Neretva
river the salt wedge penetrates upstream up to 23 km in low flux leading to the elongation of the isohalines. Additionally,
river flow during summer months and flushes out when the river results from both hydrodynamic model and Haralambidou et al.
discharge exceeds 500 m3 /s (Ljubenkov and Vranješ, 2012). (2010) study illustrate that the minimum value in the stratifica-
Additionally, the intra-tidal dynamics in combination with the tion index nS occurred at the commencement of the flood tidal
low river flow affect the geometry (length and thickness) of the phase, while higher stratification is observed at the end of flood
salt wedge and the river column stratification. The most signifi- to the commencement of ebb period.
cant change in salt wedge interface is observed under spring tide Additionally, the strong vertical stratification in the river was
and low river discharge rates, when the sea level oscillates with also indicated by estimating the buoyancy frequency (N2 ), by
9
K. Zachopoulos, N. Kokkos and G. Sylaios Regional Studies in Marine Science 39 (2020) 101467

Brunt-Vaisala:
g ∂ρ
N2 = − (8)
ρ0 ∂ z
where ρ0 is the reference density of sea water. In Strymon, the
surface-to-bottom density varies from 1,000 to 1,026 kg m−3 ,
thus the squared buoyancy frequency ranged typically from 0.04
to 0.08 s−2 , with a typical depth-averaged N2 -value of 0.07 s−2 .
However, close to the wedge interface, the local N2 -value reaches
0.25 s−2 . According to Geyer et al. (2008), salt wedge estuar-
ies may have values up to 0.3 s−2 . Similar squared buoyancy
frequency values are observed in Hudson river estuary (Peters,
1997).
To explore further the bidirectional flow dynamics, the non-
dimensional freshwater Froude number (Ff ) was introduced, as:

Q
Ff ≡ √ (9)
b0 g0′ h30

expressing the barotropic flow induced in the river channel due

to sea-surface elevation at the riverine boundary. Q is river dis- Fig. 10. Model linking the non-dimensional intrusion length L∗ to the freshwater
Froude number Ff , based on the results of the hydrodynamic simulation.
charge, b0 is the channel width at the ocean boundary, g0′ is the
reduced gravitational acceleration, considering buoyancy (g0′ =
g∆ρ/ρ0 ), with ∆ρ being the difference between layer densities,
and h0 is the water depth at the ocean boundary. Considering
Strymon River mouth as a sloped, prismatic channel, having hori-
zontal bed slope α = 0.062 and along-channel convergence ratio
RC = b0 /bR = 2.25 (bR is the channel width at the upstream river
boundary), the non-dimensional salt wedge intrusion length L∗ is
defined, as:
CD L30
L∗ = (10)
h0
where CD is the drag bottom coefficient, defined in calibration
as CD = 0.002. Through the theoretical analysis of Poggioli and
Horner-Devine (2015), the relative influence of seabed frictional
effects to topographic and bathymetric along-channel changes Fig. 11. Distribution of retention time along the Strymon River lower channel
on the salt wedge intrusion length L∗ is addressed by the CD /α as computed by ELCOM on September 25, 2004 (Q = 9 m3 /s; A = 0.15 m).
(CD /α for Strymon is 0.032) and RC -ratios. Fig. 10 illustrates the
variability of the non-dimensional salt wedge intrusion length
over the densimetric Froude number, on a daily basis for year
exerting limited force to the salt wedge. The retention time of the
2004 (June to September), based on the ELCOM model results.
salt wedge water varies from 10 to 20 days with the maximum
The L∗ vs Ff -curve proves the previously reported theoretical value observed at the deeper parts of the river mouth (Fig. 11).
considerations that channel convergence has a significant effect The slow water renewal in these areas is directly related to the
on Strymon salt wedge intrusion length. Based on this analysis, low concentration of dissolved oxygen in the water column and
a function linking the non-dimensional intrusion length and the the riverbed, highlighted on August 22, 2004 by in-situ observa-
freshwater Froude number was fitted on model results, as: tions (Haralambidou et al., 2010). The dissolved oxygen content
L∗ = 2.65 × (1 + Ff )−10.41 (11) recorded at the deeper Strymon River parts was lower than 4
mg/l, affected by the slow water renewal. This low dissolved
having high reliability (R2 = 0.90). oxygen value favors the remineralization of phosphorus and trace
Water renewal along the river channel depends highly on the metals from riverbed to the water column (Haralambidou et al.,
river discharge and the tidal force acting along the river mouth. 2010).
Furthermore, it is also affected by seabed morphology. More
specifically, the existence of two submerged sills and the presence 5. Conclusions
of two deeper channel parts influence significantly the water
retention time. Retention time (RT) is an important parameter The simulation of salt water intrusion, the salt wedge forma-
for the environmental management of the study area. According tion and establishment along the channel were studied applying
to Dyer (1997), retention time is defined as the time taken to a three-dimensional model along the Strymon River mouth. The
replace the existing freshwater in the estuary at a rate equal model was calibrated and validated based on sampling campaigns
to river discharge. The model calculates the time each water over the spring/summer months of 2003 and 2004. Model results
parcel spends within the computational grid. RT provides an allowed the study of salinity and velocity profiles, the strength
understanding of the dynamics of chemical substances dissolved of stratification along-channel and the salt wedge morphology
in the water, especially at the bottom stagnant layer, assessing the during different hydrographic conditions. Overall, a stationary,
negative consequences for the water quality. In the Strymon River well-stratified salt wedge is observed in summer months, when
mouth RT varies from 1 to 20 days. Under low river discharge, river discharge diminishes as freshwater is used upstream for
the fresh water flows on the surface layer towards the open sea agriculture.
10
K. Zachopoulos, N. Kokkos and G. Sylaios Regional Studies in Marine Science 39 (2020) 101467

The three-dimensional model applied herein produced fairly Geyer, W.R., MacCready, P., 2014. The estuarine circulation. Annu. Rev. Fluid
accurate results and captured satisfactorily the shape of the ar- Mech. 46, 175–197. http://dx.doi.org/10.1146/annurev-fluid-010313-141302.
rested salt-wedge in the irregular geometry of Strymon River Geyer, W.R., Ralston, D., 2011. The dynamics of strongly stratified estuaries.
In: Treatise on Estuarine and Coastal Science, first ed. Academic Press,
mouth. Hydrodynamic model results explained adequately envi- Amsterdam, pp. 37–51.
ronmental issues reacted to salt wedge intrusion, such as bottom Geyer, W.R., Scully, M.E., Ralston, D.K., 2008. Quantifying vertical mixing in
anoxia and nutrients and metals remineralization along the river. estuaries. Environ. Fluid Mech. 8 (5), 495–509. http://dx.doi.org/10.1007/
The spring/summer low river discharge, acting in parallel to the s10652-008-9107-2.
tidal action favored the salt wedge intrusion up to 4.6 km up- Graas, S., Savenije, H.H.G., 2008. Salt intrusion in the Pungue estuary, Mozam-
bique: effect of sand banks as a natural temporary salt intrusion barrier.
stream. The morphology of the river topography and bathymetry Hydrol. Earth Syst. Sci. Discuss. 5, 2523–2542. http://dx.doi.org/10.5194/
is also a key factor controlling the shape and the length of the hessd-5-2523-2008.
salt wedge, while entrapping water at deeper points. Landward Hansen, D.V., Rattray Jr, M., 1966. Gravitational circulation in straits and
width constriction allows the saline intrusion to penetrate further estuaries. J. Mar. Res. 23, 104–122.
up-estuary than it would have otherwise. Model results were Haralambidou, K.I., Sylaios, G.K., Tsihrintzis, V.A., 2003. Testing alternatives for
salt wedge management in an estuary with the use of monitoring and
used to test theoretical considerations developed for sloped and
mathematical model. Global Nest: Int. J. 2, 107–118. http://dx.doi.org/10.
convergent channels by previous researchers. 30955/gnj.000295.
These results highlight the need for further studies, especially Haralambidou, K., Sylaios, G., Tsihrintzis, V.A., 2010. Salt-wedge propagation in a
in the field of operational numerical modeling and forecasting Mediterranean micro-tidal river mouth. Estuar. Coast. Shelf Sci. 90, 174–184.
to produce real-time reliable salinity and water flow profiles. http://dx.doi.org/10.1016/j.ecss.2010.08.010.
Haralambidou, K.I., Tsihrintzis, V.A., Sylaios, G.K., Akratos, C., 2005. Seasonal and
Such operational system will have the capacity to respond in any
spatial characteristics of water quality in the estuary of Strymon River. J.
change of external conditions affecting the system, thus leading Mar. Environ. Eng. 7, 231–239.
to proper river management decisions. Hodges, B.R., 2000. Numerical Techniques in CWR-ELCOM (code release v. 1).
CWR Manuscript WP, p. 1422.
CRediT authorship contribution statement Hodges, B., Dallimore, C., 2006. Estuary, Lake and Coastal Ocean Model: ELCOM
V2. 2 Science Manual. Centre for Water Research, University of Western
Australia.
Konstantinos Zachopoulos: Data curation, Formal analysis,
Hume, T.M., Snelder, T., Weatherhead, M., Liefting, R., 2007. A controlling factor
Investigation, Methodology, Project administration, Visualization, approach to estuary classification. Ocean Coast. Manag. 50, 905–929. http:
Writing - original draft. Nikolaos Kokkos: Data curation, For- //dx.doi.org/10.1016/j.ocecoaman.2007.05.009.
mal analysis, Investigation, Methodology, Software, Visualiza- Ibaňez, C., Pont, D., Prat, N., 1997. Characterization of the Ebre and Rhone
tion, Writing - original draft. Georgios Sylaios: Conceptualiza- estuaries: A basis for defining and classifying salt-wedge estuaries 42,
89–101. http://dx.doi.org/10.4319/lo.1997.42.1.0089.
tion, Methodology, Resources, Supervision, Validation, Writing -
Kasai, A., Kurikawa, Y., Ueno, M., Robert, D., Yamashita, Y., 2010. Salt-wedge
review & editing. intrusion of seawater and its implication for phytoplankton dynamics in the
Yura Estuary, Japan. Estuar. Coast. Shelf Sci. 86, 408–414. http://dx.doi.org/
Declaration of competing interest 10.1016/j.ecss.2009.06.001.
Krvavica, N., Kožar, I., Travaš, V., Ožanić, N., 2017a. Numerical modelling of
The authors declare that they have no known competing finan- two-layer shallow water flow in microtidal salt-wedge estuaries: Finite
volume solver and field validation. J. Hydrol. Hydromech. 65 (1), 49–59.
cial interests or personal relationships that could have appeared http://dx.doi.org/10.1515/johh-2016-0039.
to influence the work reported in this paper. Krvavica, N., Ružić, I., 2020. Assessment of sea-level rise impacts on salt-wedge
intrusion in idealized and Neretva River Estuary. Estuar. Coast. Shelf Sci. 234,
Acknowledgments 106638. http://dx.doi.org/10.1016/j.ecss.2020.106638.
Krvavica, N., Travaš, V., Ožanić, N., 2016. A field study of interfacial friction and
entrainment in a microtidal salt-wedge estuary. Environ. Fluid Mech. 16 (6),
The research leading to these results received funding from the
1223–1246. http://dx.doi.org/10.2112/JCOASTRES-D-16-00053.1.
European Union Horizon 2020 Program (H2020-BG-12-2016-2) Krvavica, N., Travaš, V., Ožanić, N., 2017b. Salt-wedge response to variable river
under grant agreement 727277 - ODYSSEA (Towards an inte- flow and sea-level rise in the Microtidal Rječina River Estuary, Croatia. J.
grated Mediterranean Sea Observing System). Coast. Res. 33, 802–814. http://dx.doi.org/10.2112/jcoastres-d-16-00053.1.
Lindström, G., Pers, C., Rosberg, J., Strömqvist, J., Arheimer, B., 2010. Development
and testing of the HYPE (Hydrological Predictions for the Environment)
References
water quality model for different spatial scales. Hydrol. Res. 41, 295–319.
http://dx.doi.org/10.2166/nh.2010.007.
Barlow, P., 2003. Saltwater intrusion from the Delaware River during drought-
Ljubenkov, I., 2015. Hydrodynamic modeling of stratified estuary: case study
implications for the effect of sea-level rise on coastal aquifers. Ground water
of the Jadro River (Croatia), 63, 29–37, http://dx.doi.org/10.1515/johh-2015-
in freshwater–saltwater environments of the Atlantic coast. US Geol. Surv.,
0001.
Reston 4, 6–48.
Ljubenkov, I., Vranješ, M., 2012. Numerical model of stratified flow - Case
Casulli, V., Cheng, R.T., 1992. Semi-implicit finite difference methods for three-
study of the Neretva riverbed salination (2004). Građevinar 64 (2), 101–112.
dimensional shallow water flow 15, 629-648. http://dx.doi.org/10.1002/fld.
http://dx.doi.org/10.14256/JCE.639.2011.
1650150602.
MacCready, P., Geyer, W.R., 2001. Estuarine salt flux through an isohaline
Chanson, H., 2004. Environmental Hydraulics for Open Channel Flows. Elsevier
surface. J. Geophys. Res.: Oceans 106, 11629–11637. http://dx.doi.org/10.
Butterworth-Heinemann, Burlington.
Chen, X., Flannery, M.S., Moore, D.L., 2000. Response times of salinity in relation 1029/2001jc900006.
to changes in freshwater inflows in the lower hillsborough river, Florida. Marti, C., Imberger, J., 2015. A real-time management system for the Swan-
Estuaries 23, 735–742. http://dx.doi.org/10.2307/1352899. Canning River Basin and its receiving waters, Western Australia, Australia,
D’Adamo, N., Lukatelich, R.J., 1985. Water Quality of the Murray River Estuary. E-proceedings of the 36th IAHR World Congress.
Centre for Water Research, University of Western Australia, Nedlands, W.A. McAnally, W.H., Pritchard, D.W., 1997. Salinity control in Mississippi river under
Dassenakis, M., Scoullos, M., Gaitis, A., 1997. Trace metals transport and be- drought flows 123, 34–40. http://dx.doi.org/10.1061/(ASCE)0733-950X(1997)
haviour in the mediterranean estuary of Acheloos river. Mar. Pollut. Bull. 34, 123:1(34).
103–111. http://dx.doi.org/10.1016/s0025-326x(96)00062-8. McAuliffe, T., Lukatelich, R., Hill, N., 1993. The role of sediments in phosphorus
Donnelly, C., Andersson, J.C.M., Arheimer, B., 2016. Using flow signatures and cycling in the Swan River Estuary, Swan River Trust—The Future, Workshop
catchment similarities to evaluate the E-HYPE multi-basin model across Proceedings, pp. 75–87.
Europe. Hydrol. Sci. J. 61, 255–273. http://dx.doi.org/10.1080/02626667.2015. Monbet, Y., 1992. Control of phytoplankton biomass in estuaries: A comparative
1027710. analysis of microtidal and macrotidal estuaries 15, 563–571. http://dx.doi.
Dyer, K.R., 1997. Estuaries: A Physical Introduction, second ed. Wiley. org/10.2307/1352398.
Fischer, H.B., List, J.E., Koh, C.R., Imberger, J., Brooks, N.H., 1979. Mixing in Inland Nepf, H.M., Geyer, W.R., 1996. Intratidal variations in stratification and mixing in
and Coastal Waters. Academic Press. the Hudson estuary 101, 12079–12086. http://dx.doi.org/10.1029/96jc00630.

11
K. Zachopoulos, N. Kokkos and G. Sylaios Regional Studies in Marine Science 39 (2020) 101467

Parissis, A., Sylaios, G., Tsihrintzis, V., 2001. A numerical model for the study of Sierra, J.P., Sánchez-Arcilla, A., Figueras, P.A., Gonzalez Del Rio, J., Rass-
salt intrusion at Strymon River mouth, Northern Greece. In: Proceedings of mussen, E.K., Mösso, C., 2004. Effects of discharge reductions on salt wedge
the First International Congress on Ecological Protection of Planet Earth, pp. dynamics of the Ebro River. River Res. Appl. 20, 61–77.
5–8. Sierra, J.P., Sánchez-Arcilla, A., González Del Río, J., Flos, J., Movellán, E., Mösso, C.,
Peters, H., 1997. Observations of stratified turbulent mixing in an estuary: Neap- Martínez, R., Rodilla, M., Falco, S., Romero, I., 2002. Spatial distribution of
to-spring variations during high river flow. Estuar. Coast. Shelf Sci. 45 (1), nutrients in the Ebro estuary and plume. Cont. Shelf Res. 22, 361–378.
69–88. http://dx.doi.org/10.1006/ecss.1996.0180. http://dx.doi.org/10.1016/S0278-4343(01)00061-9.
Poggioli, A.R., Horner-Devine, A.R., 2015. The sensitivity of salt wedge estuaries Sylaios, G., Koutrakis, E., Kallianiotis, A., 2006. Hydrographic variability, nutri-
to channel geometry. J. Phys. Oceanogr. 45 (12), 3169–3183. http://dx.doi. ent distribution and water mass dynamics in Strymonikos Gulf (Northern
org/10.1175/JPO-D-14-0218.1. Greece). Cont. Shelf Res. 26, 217–235. http://dx.doi.org/10.1016/j.csr.2005.11.
Pritchard, D.W., 1967. Observations of circulation in coastal plain estuaries. In: 002.
Lauff, G.H. (Ed.), Estuaries. American Association for the Advancement of Trowbridge, J.H., Geyer, W.R., Bowen, M.M., Williams, A.J., 1999. Near-bottom
Science Publication, Washington DC, pp. 37–44. turbulence measurements in a partially mixed estuary: Turbulent en-
Ralston, D.K., Geyer, W.R., Lerczak, J.A., Scully, M., 2010. Turbulent mixing in ergy balance, velocity structure, and along-channel momentum balance. J.
a strongly forced salt wedge estuary. J. Geophys. Res.: Oceans 115 (C12), Phys. Oceanogr. 29, 3056–3072. http://dx.doi.org/10.1175/1520-0485(1999)
C12024. http://dx.doi.org/10.1029/2009JC006061. 029<3056:Nbtmia>2.0.Co;23056:Nbtmia>2.0.Co;2.
Ranmadugala, S.B.H., 2004. Influence of Seasonal Sea Level Variability on Tuin, H., 1991. Guidelines on the Study of Seawater Intrusion Into Rivers. Unesco.
Salt-Water Intrusion in Kelani River Basin. Valle-Levinson, A., 2010. Contemporary Issues in Estuarine Physics. Cambridge
Robson, B.J., Hamilton, D.P., 2004. Three-dimensional modelling of a Microcystis University Press, Cambridge.
bloom event in the Swan River estuary, Western Australia. Ecol. Model. 174, Vieira, M.E.C., Bordalo, A.A., 2000. The Douro estuary (Portugal): a mesotidal salt
203–222. http://dx.doi.org/10.1016/j.ecolmodel.2004.01.006. wedge. Oceanol. Acta 23, 585–594. http://dx.doi.org/10.1016/s0399-1784(00)
Sargent, F.E., Jirka, G.H., 1987. Experiments on saline wedge 113, 1307–1323. 01107-5.
http://dx.doi.org/10.1061/(ASCE)0733-9429(1987)113:10(1307). Watanabe, K., Kasai, A., Antonio, E.S., Suzuki, K., Ueno, M., Yamashita, Y., 2014.
Scoullos, M.J., Oldfield, F., 1986. Trace metal and magnetic studies of sediments Influence of salt-wedge intrusion on ecological processes at lower trophic
in greek estuaries and enclosed gulfs. Mar. Chem. 18, 249–268. http://dx.doi. levels in the Yura Estuary, Japan. Estuar. Coast. Shelf Sci. 139, 67–77. http:
org/10.1016/0304-4203(86)90012-5. //dx.doi.org/10.1016/j.ecss.2013.12.018.
Seim, H.E., Blanton, J.O., Elston, S.A., 2009. The effect of secondary circulation Zhou, M., 1998. Influence of bottom stress on the two-layer flow induced by
on the salt distribution in a sinuous coastal plain estuary: Satilla River, GA, gravity currents in estuaries. Estuar. Coast. Shelf Sci. 46, 811–825. http:
USA. Cont. Shelf Res. 29, 15–28. http://dx.doi.org/10.1016/j.csr.2008.03.018. //dx.doi.org/10.1006/ecss.1998.0306.

12