1. Introduction
Located on the western coast of the Korean Peninsula facing the Yellow Sea, the Chunsu Bay (hereafter, CSB) is a typical shallow embayment. It is a semi-closed north–south elongated bay blocked in the north by two artificial dikes (Seosan A·B district tidal embankments). The CSB is 25 km long and 8.5 km wide with depths shallower than 20 m covering 89.2 % of its area (Fig. 1).
The reclamation project in the Seosan A·B district resulted in the blockage of the northern part of the CSB by constructing two tidal embankments (Fig. 1) in 1983-1985, which led to a reduction in the area of the CSB from 380 to 180 km2. The changes in the CSB included reduced tidal current velocity, sedimentation in the central channel, and frequent occurrence of hypoxia in the summer season leading to collapse of the clam fishery. In addition, freshwater discharge from two lakes rich with agricultural fertilizer significantly degraded the seawater quality to the third level (the worst Korean Environment Agency standard).
In the rapidly changing coastal marine environment, the outcomes of freshwater, contaminants, and other conservative substances are influenced by tidal residual currents, wind-driven currents, and density-driven currents (Ro and Jung, 2010;Jung and Ro, 2010;Jung et al., 2012;2013a). Numerous observational and numerical studies on the circulation of coastal and estuarine water bodies around the world have been conducted to determine the characteristics of tides and tidal currents, and residual wind-driven and density-driven currents (Imasato, 1983;Goodrich et al., 1987;Dube et al., 1995;Guo and Valle-Levinson, 2008;Zhai et al., 2008;Santoro et al., 2013) along with their mechanisms (Kashiwai, 1984;Cai et al., 2003;Foreman et al., 2006).
However, physical oceanographic studies on the CSB in the past 20 years have been scarce compared to studies in the other coastal and estuarine water bodies in Korea. These few studies include a numerical modeling of circulation and diffusion (Yoo, 1992), study of changes in tidal phenomena due to tidal embankment construction (So et al., 1998), and a water-quality model (Choi, 2004) based on a 2D or 3D numerical model in the CSB. Choi (2004) imposed constant discharge rates of 81 and 64 m3/s, respectively, from the two inland lakes, Kanwol (hearafter, KW) and Bunam (hearafter, BN). However, compared to previous research, in the present study, the model was developed with higher spatial resolution, realistic bottom topography, and realistic constrains, such as open boundary conditions, and discharge rates (Yoo, 1992;Park and Oh, 1998;So et al., 1998;Choi, 2004).
Under the large lake water discharge generated by artificial dikes during the summer monsoon season, the CSB becomes a transient estuarine system. We describe the results of numerical modeling experiments in the CSB during the summer season in five separate papers, as follows: in Paper 1, the characteristics of the tidal current and tidal residual current (Jung et al., 2013b); in Paper 2, the impact of the freshwater discharge and formation of stratification (this paper); in Paper 3, the wind-driven current and its influence on the destratification (in preparation); in Paper 4, particle trajectory experiments and calculation of the flushing time; and in Paper 5, formation and maintenance of hypoxia caused by the strong stratification (Jung et al., 2015).
In this study (Paper 2), we focus on the transient features of estuarine dynamics in terms of circulation, salinity structure and vertical stability. The objectives of this study are i) to understand the circulation system of the CSB, in which discharge of a large amount of lake water occurs in summer, ii) to identify and understand diverse physical phenomena under the effects of lake water discharge through the KW/BN dikes (i.e., residual current characteristics, salinity dilution, and formation of stratification).
2. Data and Method
2.1 Observations of Hydrodynamic and Hydrographic Conditions in the CSB, 2010
Intensive fieldwork was carried out in the CSB during the summer of 2010 by conducting in situ measurements with an acoustic Doppler current profiler (ADCP) with time series, T－S profiles, and water-quality parameters including dissolved oxygen, chlorophyll, bottom sediment oxygen demand, and nutrients such as nitrate, phosphate, and silicate. The same observations were made in 2011, except for the ADCP observations (Table 1).
Detailed current profiles were obtained in the center of the CSB, marked by the star in Fig. 1, for a period of 33 days in July 2010 by deploying an upward looking bottom-mounted ADCP (Nortek AWAC, 1 MHz). the maximal current speed exceeded 40 cm/s at the surface, and the dominant direction of the main flow was NNE －SSW. The progressive vector of wind and the residual current were northward throughout the entire water column, and the travel distance in the middle layer was greater than that at the surface layer. The northward residual current was caused by a seasonal southerly wind. A density-driven current formed by the freshwater input also generated a southward residual current at the surface. The freshwater discharged from KW and BN reached the central region of the CSB in 1–2 days. Readers more interested in the variability of the hydrographic and hydrodynamic conditions in the CSB may refer to Jung et al. (2013a).
2.2 Lake Water Discharge Conditions in the CSB, 2010
A large amount of freshwater from the two inland lakes (i.e., KW/BN) was discharged into the CSB during the summer season. The freshwater discharge started before and after low tide in the spring period and it lasted for 3–4 h once each day (Jung et al., 2013b).
A time series record of the KW/BN tidal embankment water discharge rate is shown in Fig. 2. Discharge rates from KW were 400–600 m3/s, and those from BN were 100–300 m3/s. A freshwater volume of 780 × 106 m3 was discharged into the CSB from KW/BN in 2010, which is equivalent to 35 % of the CSB water volume (Jung et al., 2013a;2013b).
2.3 Model Specification and Run Cases
The numerical model for the CSB used in this study was ECOMSED (HydroQual, Inc.). The details of the physical setting, bottom topography, model specification, and numerical schemes (i.e., area and grid size, vertical layers, initial and open boundary conditions, mixing closure, and wet/dry scheme) are described in a previous study of the author (Jung et al., 2013a) and in Fig. 3.
Specifications of the model runs are listed in Table 2. The water discharge record from July to September 2010 was used to impose the freshwater input. After confirming the model performance in an appropriate manner, we started to impose the water discharge condition, as shown in Fig. 2.
2.4 Model Performance and Validation
For the purpose of model calibration and validation, two sets of field data were obtained : conductivity, temperature, and depth (CTD) profiling was obtained for the N–S stations in July 2010 and ADCP data were continuously observed with a sampling interval of 10 min for the month of August 2010. To assess the model skill quantitatively, we adopted the skill score (SS) suggested by Martin and McCutcheon (1999), which is defined as the deviation of the relative error from unity, as in Eq. (1).
where RE = abs (Xmodel−Xobs) / Xobs and X can be any parameter, such as tidal elevation, velocity, temperature, or salinity.
The model results were calibrated and validated using the observed time-series of sea-level and current data (Choi, 2004;Jung et al., 2013a).
3. Results
Various modeling results will be analyzed to show the impact of the freshwater discharges from the two tidal embankments. The tidal dynamics characteristics in terms of ellipse parameters, residual current and stratification will be analyzed.
3.1 Observation results
Fig. 4a shows vertical profiles of salinity in the CSB before and after water discharge ordered by the Bunam–Kanwol dike Authority (BKDA, Korea Rural Community Corporation) in June 2010.
Before the discharge, the profiles are well-mixed and vertically homogeneous. These are in stark contrast with the profiles in July, which show very sharp salinity decreases in the northern stations (Fig. 4b). These figures reveal a transition in the CSB from vertically homogeneous profiles to partially mixed type depending on the condition of lake water discharge. As shown in Fig. 5, when the freshwater discharged from KW/BN impacted the water in the CSB, the salinity field changed rapidly. Salinity dropped to 13.3 psu on July 26, 2010 at the northern most station 1. As shown in Fig. 6a, salinity in the upper 2 m layer was less than 20 psu in the northern area compared to salinity at the bottom along the north–south cross section, which was higher than 30 psu. The depth of the pycnocline ranged between 2–5 m. The buoyancy of the freshwater in the upper layer formed a strongly stratified water column.
As reported by Jung et al. (2013a and 2013b), the observed current speed (Fig. 7) ranged from –30 to 40 cm/s, with standard deviation from 1.7 cm/s at bottom to 18.7 cm/sec at surface. According to the harmonic analysis results, the tidal current direction was NNW–SSE. The magnitudes of the semi-major axes ranged from 9.4 to 14.8 cm/s for harmonic constituent M2 and from 4.4 to 7.0 cm/s for S2, respectively. The magnitudes of the semi-minor axes ranged from 0.1 to 0.5 cm/s for M2 and from 0.4 to 1.4 cm/sec for S2, respectively. More detailed results for tidal and sub-tidal current characteristics caused by the local wind and discharge effect with various analyses (i.e., descriptive statistics, harmonic analysis of tidal constituents, spectra and coherence, complex correlation, progressive vector diagram and cumulative curves) can be found in previous study of the author (Jung et al., 2013a).
3.2 Model Performance and Validation
By comparing the two case model outputs of sea-level elevation (SLE) to local tidal records (Boryung [KHOA], Kojeong and Kanwol; Choi, 2004), SSs of amplitudes and phase lags for the four major tidal constituents M2, S2, K1, and O1 were obtained. Scores for amplitudes were mostly higher than 90 %. The average relative errors were 13.5 % and 10.1 % for amplitude and 3.0 % and 2.8 % for phase lag in CS10T and CS10TD, respectively.
The time series of measured surface current speeds and the model output at Jukdo station are shown in Fig. 7. The SSs of the amplitudes and phases for M2 and S2, the semi-diurnal tides, were higher than 90 % for both CS10T and CS10TD. For CS10T, the SSs of amplitudes and phases for the diurnal tides (K1 and O1) were approximately 70–78 % and 60–90 %, respectively. For CS10TD, they were around 75–88 % and 79–90 %, respectively. The SSss of CS10TD were a refinement of those of CS10T for the four major tidal constituents. The validation of the observational data and model results was in general satisfactory for tidal height and tidal current.
The model output was compared to the transected profiles of salinity and they agreed reasonably well, as shown in Fig. 6. The results of the model temperature and salinity fields showed high SSs, above 85 % for salinity and above 93 % for temperature. The SSs of tidal elevation, current, salinity, and temperature demonstrate that the model performance was satisfactory.
3.3 Simulation of Current Field
Fig. 8 shows the horizontal distribution of the simulated tidal current in the spring tidal phase. In the CS10T run, the maximum ebb current in spring (Fig. 8a-1) showed a high speed, ranging from 2–4 m/s at the entrance of the CSB. However, in the northern part of the CSB, the current was very weak, below 0.5 m/s. Fig. 8a-2 shows that the ebb current pattern for CS10TD was similar to that of CS10T except for the influence of freshwater discharge from BN/KW. The southward ebb current speed was intensified by 0.5–1 m/s. The behavior of the discharged lake water during the flood phase showed a more complicated pattern in that the southward flow was deflected toward the westward bank (Fig. 8b-2).
Fig. 9 shows the tidal residual current (a), density-driven current (b), and total residual current (c), and Table 3 presents the basic statistics of the tidal residual current and the density-driven current at 10 selected points. The total residual current (VelTotalR) is expressed as
where VelTR is the tidal residual current and VelDC is the density-driven current. VelTR (VelDC) is obtained by using the velocity differences between VelCS10T and VelCS10T(HA) (between VelCS10TD and VelCS10T) at all grid points for all the computation steps in such a way that
∀ time step, grid points
where VelCS10T(CS10TD) is the velocity field for CS10T (CS10TD), respectively, and VelCS10T(HA) is the velocity field for the tidal current from the harmonic analysis result of CS10T.
The distribution of the tidal residual current (Fig. 9a) can be summarized as follows; the tidal residual current in the northern CSB ranges from 1 to 7 cm/s, while in the southern region, it has a wider ranging 10 to 30 cm/s. It is noteworthy that there exists an eddying motion near the Jukdo Island in central CSB. Northwestern flows occurred along the coast in the waters near the KW/BN seawall north of the CSB. Overall, mostly northward flows were dominant in the eastern CSB, and southward flows were dominant in the western CSB. Based on the results of the tidal residual current, the flood/ebb imbalance might be attributed to the bottom topography and Coriolis force.
We prepared Fig. 9b by subtracting CS10T from CS10TD to show the sole effect of the freshwater discharge from the north. The freshwater discharge caused buoyant plumes from the two discharge outlets that were initially directed southward but soon deflected toward the west. Eventually, the discharged water should drain out of the CSB by flowing southward in zone 3. The density-driven current, shown in Fig. 9b, flowed mainly southward due to the impact of the freshwater inflow from KW/BN. The density-driven current speed near KW/BN ranges from 10 to 20 cm/s, and it slowed down to a speed of 5 to 10 cm/s from the southern part of Jukdo to the entrance of the bay. At the center of the northern part of the CSB, where the southward flow generated near BN converges with the flow from KW, there was a strong density-driven current with a speed of up to 20 cm/s that rotated in a clockwise direction.
Fig. 9c reveals the effect of the freshwater discharge from the two embankments (KW + BN) on the residual current. The flow pattern can be distinguished in terms of three zones: a near-embankment zone in the north (zone 1), a CSB mouth zone in the south (zone 3), and a middle zone (zone 2) between them. In zone 1, flow from the discharged freshwater was deflected toward the west; in zone 3, three pairs of the main eddy exist as very strong flows; in zone 2, a sort of transition zone, the flow showed a more complicated southward current with eddy activity due presumably to the island effect (Jukdo).
3.4 Simulation of Salinity Field
Fig. 5 shows a time series of salinity from model run CS10TD at four selected stations. The model results agree favorably with the field surveys. The degree of salinity fluctuation was highest at St. 1 and lowest at St. 6. The range of salinity fluctuation at St. 1 reached more than 15 psu, whereas that at St. 6 was less than 3 psu.
Field observations from the north–south direction were made on July 26, 2010 during the discharge period. A vertical transect of the salinity field based on the observational data is shown in Fig. 6a and the result of modeling is shown in Fig. 6b. The vertical transect of the salinity field reveals a two-layer structure. The surface salinity at St. 1 and 2, close to the embankment, was within the range of 14 to 18 psu due to dilution, and the salinity difference between the surface and bottom layers decreased from north to south. The formation of strong haloclines was observed at a water depth between 2 and 5 m in the northern part of the CSB.
Fig. 10 shows the mean salinity field and standard deviation during the 54-day modeling period in the summer season. The mean salinity field of the CSB was below 27 psu in most of its northern part. In zones 2 and 3, the mean salinity ranged from 28 to 31 psu with inclined isohalines that reflect the portion of the discharged freshwater plume deflected toward the west. In Fig. 10b, the fluctuation of salinity is shown in terms of standard deviation. Near the KW/BN embankments, the standard deviation was as large as 5 psu, and it decreased gradually to less than 1 psu near the mouth of the CSB.
To understand the dilution of salinity in the CSB, budget analyses were carried out in terms of volume and salt fluxes (Table 4). About 60 % of the freshwater input from KW/BN was flushed out within the time period of model run CS10TD, 54 days. Volume and salt balances in the CSB region can be calculated by the two boundaries (KW/BN and Southern channel). After 54 days, the net volume gain into the CSB was +8 × 106 m3, accounting for only 0.0036 % of the CSB volume, which would lead to a sea-level change of only 4 cm due to the increase in volume. The net amount of salt loss was 4.6 × 106 kg, which resulted in a salinity decrease of 2.2 psu in the CSB.
3.5 Stratification Strength
To understand the stability structure of the CSB after freshwater discharge, we estimated the Richardson number to assess the relative importance of vertical shear at the bottom and buoyancy input in the upper layer. The Richardson number (Ri) defined by Turner (1973) is shown in Eq. (5):
where N is the Brunt–Väisälä frequency defined as Eq. (6), and ρ is the potential density. Fig. 11 shows a time series of (b) Ri, (c) static stability (S-2), and (d) vertical shear (S-2) for the 10 selected stations along the N–S direction and also shows (a) discharge rate (m3/sec) information and (e) tidal elevation (m) information for the same time period.
The Ri value was estimated to be high, approximately 20 in the northern part of the CSB and between 10 and 20 in the central region during the freshwater discharge. The frequency and percentage of Ri occurrences are presented in Table 5. Ri was more than 10 for 85 % of the time in the northern part of the CSB (Sts. 1 and 2), and it was directly affected by the inflow of freshwater. This indicates that most of the northern part was exposed to very strong stratification, whereas the ratio exposed to strong stratification decreased gradually southward.
The static stability was estimated to be higher than 20 × 10-4 s-2 from the northern to the central part of the CSB owing to an increase in the vertical density gradient in Period-I, when discharge of freshwater occurred. During Period-II, when discharge was stopped, the static stability was low at all selected points, of 5 × 10-4 s-2 or less, at all the ten stations, except in the northern part of the CSB (Sts. 1 and 2), due to vertical mixing.
The vertical shear (d) showed spring tide–neap tidal modulation, similar to the fluctuations in tidal phase (e) through the entire period and had a large value of 10–20 × 10-4 s-2 during the spring tidal phase in the southern part. The vertical shear caused by tides had limited impact in the northern part (Sts. 1 to 3), where current speed was slow, but it increased dramatically during the inflow of freshwater through the KW/BN dikes due to a strong density-driven current.
4. Discussion
4.1 Circulation System
The tidal current in the CSB (Fig. 8) is a mostly north/southward rectilinear flow along isobaths. In the southern part of the CSB (zone 3), the current becomes intensified due to narrowing of the bay mouth. The distribution of the tidal current can be divided into three zones: the southern part of the CSB (zone 3), in which there are high-speed currents, from 2 to 4 m/s; the central part of the CSB (zone 2), in which medium-range current speeds of 0.5 to 1 m/s prevail; and the northern part (zone 1), in which speeds range from 0.1 to 0.7 m/s. In the results of the current field experiment, the tidal current speeds were calculated to be 7 % and 13 % higher than those reported by Choi (2004) and Yoo (1992), respectively. The differences might be due to the water depth and bottom friction coefficient. Other than this, the distribution of tidal current in the CSB is in good agreement with that presented in previous studies (So et al., 1998;Yoo, 1992;Choi, 2004).
The eddy formation in the CSB was consistent with the explanations provided by Maze et al. (1998), who suggested that the formation of eddies was caused by changes in water depth and bottom friction, and by Signell and Harris (2000), who suggested the formation of eddy in the headland area. The tidal residual current was calculated to be in the range of 5 to 40 cm/s, and the speed was approximately 4 to 5 times higher in the southern part (zone 3) than the northern part (zone 1). These speeds of the tidal residual current were faster than those in the previous study of Choi (2004), of 5–10 cm/s. However, a similar distribution of the tidal residual current, including paired cyclonic and anti-cyclonic eddies, had been reported by So et al. (1998). For more detailed descriptions of the characteristics of tidal residual current and relative vorticity, the interested readers may refer to the report of Jung et al. (2013b).
During the summer season, discharges took place once a day for 3–4 h with discharge rates of 400 and 200 m3/s from KW and BN, respectively. Based on the results of the numerical modeling experiments, the inflow of freshwater is expected to have diverse effects in the CSB. The freshwater input in the surface is a major drive for the buoyancy supply and density-driven current (Turner, 1973). In the case of the CSB, the density-driven current (Fig. 9b) reached up to 0.1–0.2 m/s and that rotated in a clockwise direction in the northern part (zone 1) near KW and BN. Therefore, it can be speculated that the freshwater discharged from KW/BN could not easily escape out of the southern part of Jukdo; thus, it might remain in the northern part of the CSB for a considerable period of time.
4.2 Stratification formation and maintenance
The most important effect caused by freshwater discharge is stratification of the water column. Many researchers (Pritchard, 1952;1954;1956;Sharples et al., 1994;Friedrichs and Hamrick, 1996;Geyer, 1993; etc.) indicated that factors such as freshwater inflow and decreasing salinity at the surface layer have caused static stability intensification and stratification formation in the estuarine system. In this study, the surface salinity dropped to 15 –20 psu, which is lower than the bottom salinity by 12–15 psu with a two-layer structure, in which a baroclinic (density-driven) current exists in the northern part (zone 1) of the CSB. The stratification was formed and maintained with a Richardson number (Fig. 11 and Table 5) of 10–20 in the central region and aapproximately 20 in the northern part of the CSB. The range of Ri values obtained in this study are comparable to those of the river estuaries (Goodrich et al., 1987;Geyer, 1993;Peters, 1997;Liu et al., 2008;Jung and Ro, 2010;Ro and Jung, 2010).
Tidal mixing and its competition with buoyancy forces (ultimately derived from freshwater runoff) are known to be important processes in estuaries. According to the review of Uncles (2002), the progression from neap to spring tides caused a severe reduction in stratification from an initial top-to-bottom salinity difference of 18 to < 4, consistent with that of Sharples et al. (1994) in the York River estuary data. The small gradient Richardson numbers were restricted to the weakly stratified bottom layer on the flood portion of neap tides but occurred throughout the water column during the late ebb of springs. In this study, strong stratification occurred in the northern part of the CSB near the embankment during the spring tidal phase, whereas the stratification was strengthened in the central and southern part of CSB during the neap tidal phase, when the vertical shear decreased. Our results are slightly different from those of Sharples et al. (1994), and the reason is characteristic of freshwater inflow. Compared to the case in the York River estuary with continuous inflow of freshwater without distinction of neap or spring tides, our case is the pulse-type discharge of freshwater through the tidal embankment (KW/BN) by human control, which mainly occurred at low tide in the spring period in the CSB.
The stratification process in the CSB can be interpreted as follows. Although there was a partial increase in vertical shear caused by a strong density-driven current in the northern part of the CSB near the embankment during the discharge of freshwater (Period-I), the density gradient between the upper and lower layers increased dramatically, which strengthened static stability. This increase in static stability depressed the turbulence effect, which affected the vertical mixing. Through this process, stratification in the CSB was formed and maintained. After the inflow of freshwater stopped (Period-II), the vertical mixing and advection– diffusion process weakened the static stability, but the stratification was maintained in the northern part of the CSB. One of the reasons for this was that the discharge of freshwater was stopped in the neap tide phase, in which vertical shear decreases dramatically, leading to weakening of vertical mixing. A second reason was that the freshwater remained in the northern part of the CSB, and hence, the static stability was maintained at a high level for a certain period of time. Thus, without any external forces such as wind, the stratification could be maintained in the northern part of the CSB in the summer for several weeks following the inflow of freshwater, irrespective of whether the discharge was continued.
4.3 Environmental implications of the artificial dikes
Many artificial dikes have been built in the semi-enclosed Korean coastal bays to increase the efficiency of flood control, to prevent inundation, for freshwater storage, for land reclamation, and for other recreational activities. CSB was well known as a highly productive coastal bay before the construction of the dikes, and it served as a nursery ground for juvenile fish (Lee et al., 1997). However, the ecosystems in CSB have been disturbed since the dike construction. As described in the Introduction section, the reclamation project in the Seosan AB zone has led to several geographic variations for over 30 years. Following the change in tidal circulation, the tidal current velocity was greatly reduced and freshwater-type phytoplankton temporarily grows during the summer in the uppermost part of the bay because of freshwater discharge from the artificial dikes (Shin et al., 1990;So et al., 1998;Lee et al., 2012). In addition, the subsequent changes included strong stratification by freshwater inflow, and frequent occurrences of hypoxia in the summer season (Jung et al., 2015), which deteriorated the water quality and led to the collapse of the clam fishery.
These problems are closely related to the changes in tidal phenomena due to the KW/BN tidal embankment construction (So et al., 1998) and the stratification formation due to freshwater inflow from the artificial dikes (Jung et al., 2015). The change in the velocity field and vertical structure in physical properties affected various marine environment variables. The aquaculture fishery grounds were damaged and fishery stocks have been declining since the construction of the two dikes (Lee et al., 2012).
This study is only focused on the limited aspect of physical processes of the CSB. Thus, a follow-up study will be carried out to determine the effect of wind on the CSB properties (e.g., destratification and freshwater trap) when the southwest monsoon occurs in the summer season, in which there is inflow of freshwater. Future studies will be required to understand the water quality deterioration and eventual formation of hypoxia at the bottom of the CSB caused by the eutrophic freshwater inflow from KW/BN.
5. Conclusions
A three-dimensional numerical model was established to study the impact of the lake water (KW/BN) discharged by artificial dikes in the CSB during the summer monsoon season. The model results show a realistic overall agreement with the measured tide, current velocities, temperature, and salinity, with over 90 % confidence. This study suggests that the large amount of lake water discharged through the artificial dikes may result in a dramatically changed density field in the CSB, which leads to an estuarine circulation system. The strong stratification suppressed the vertical flux and inhibited the surface aerated water from mixing with the bottom water. This phenomenon would have direct and indirect impacts on the marine environment such as hypoxia/anoxia formation at the bottom.
