Water and Salt Transports in the Hengsha Channel of Changjiang Estuary

Abstract

In a multilevel bifurcated estuary, the channels between the bifurcated branches play important roles in the exchanges of water and salt. In the Changjiang Estuary, the Hengsha Channel (HC) connects the North Channel (NC) and the North Passage (NP). In this paper, based on a two-way nesting unstructured quadrilateral grid, finite-differencing, three-dimensional estuarine and coastal ocean model, the tidal and seasonal variations in the water and salt transports in the HC were simulated, and their dynamic mechanism was analyzed. The residual water and salt transports in the HC both flow southward from the NC to the NP. In wet season, the residual water transport in the HC is 677 m³/s during neap tide and 245 m³/s during spring tide, and the residual salt transport is 0. In dry season, the residual water and salt transports in the HC are 1278 m³/s and 0.38 t/s during neap tide, respectively, and 1328 m³/s and 12.61 t/s during spring tide. Affected by the northerly wind and the southeastward baroclinic gradient force, the water and salt fluxes in dry season are much larger than those in wet season. The dynamic mechanism responsible for the water transport in the HC was numerically simulated and analyzed.

1. Introduction

Estuaries, which represent the regions of transition between rivers and the ocean, are commonly characterized by intensive human activities and the interaction of multiple dynamic factors [1,2,3]. Rivers not only discharge fresh water into the sea but also carry a lot of nutrients. Tide is the most energetic source of hydropower in estuaries. The tidal range can introduce significant tide current, which oscillates the estuarine fronts and mix dissolved materials. The fresh water carried by the river and the high salt water of the open sea are strongly mixed in this area, making the estuary produce obvious density gradients in the horizontal and vertical directions, accompanied by strong temporal and spatial changes. This can produce estuary circulation under the baroclinic pressure gradient [4,5,6]. Compared with tidal currents, estuarine circulation essentially determines the transport of salt and other substances on a long-term scale. Estuary circulation can significantly change the vertical structure of water flow and can have an impact on water mixing, estuary sediment flocculation and ecological environment [7,8,9,10,11,12,13]. The main driving forces of estuary circulation are the barotropic and baroclinic gradient forces. The barotropic gradient force is generated by the slope of the water level on the surface. The water level, generated by the tide and runoff, fluctuates with the tide cycle and seasonal river discharge. During the process of flood tide, the water level is high outside the mouth and low inside the mouth, and the water level slopes toward the land; during the process of ebb tide, the water level is low outside the mouth and high inside the mouth, and the water level slopes toward the sea, which is the same as the water level slope caused by runoff. The baroclinic gradient force is produced by the difference of density in the horizontal direction and is always landward in the estuary, increasing with depth. On the average scale of the tidal cycle, the barotropic and baroclinic gradient forces interact to produce estuary circulation [14,15,16]. As some of the most typical natural phenomena in estuaries, saltwater intrusion changes the estuary circulation by affecting the density field. The main influence factors on saltwater intrusion are tide and runoff, and other factors include wind stress, topography and shelf circulation. How dynamic factors affect saltwater intrusion, such as vertical mixing, horizontal advection, barotropic force and baroclinic force, has been extensively studied in theory [17,18,19,20]. In a single-channel estuary, saltwater intrusion usually occurs in the along-channel direction. However, in a bifurcated estuary, the water and salt transport between different channels also triggers lateral water and salt transport. The material transportation and hydrodynamics in estuarine bifurcated channels have been extensively studied: for example, the Sixianjiao between the Xijiang River and the Beijiang River in the Pearl River Estuary [21,22]; the Chunkhuri, Pankhali and Batiaghata Channels between the Pasur River and the Shibsa River [23,24,25]; the Suisan Cutoff between the Crizzlt bay and the Honker bay in the northern San Francisco Bay [26,27]; and the Connecting Channels in the Weatern Scheldt Estuary [28,29].

The Changjiang, also known as the Yangtze River, is the largest river in China. It discharges large amounts of fresh water into the East China Sea. Its river discharge has obvious seasonal variation, reaching a maximum monthly mean of 49,500 m³/s in July and a minimum of 10,500 m³/s in January [30]. It is a mesotidal estuary with a mean tidal and experiences multilevel bifurcations. It is divided into the South Branch and the North Branch by the Chongming Island. Then, the Changxing Island and Hengsha Island bifurcates the South Branch into the South Channel and the North Channel (NC). Finally, the Jiuduan Sandbank bifurcates the South Channel into the South Passage and the North Passage (NP). In addition, there are also a number of inlets connecting adjacent channels: for example, the northern inlet of the NC on the eastside of the Chongming Island, and the Hengsha Channel (HC) between the Changxing Island and Hengsha Island connecting the NC and NP (Figure 1). Lateral saltwater intrusion caused by water and salt exchange between different branches in the Changjiang Estuary has attracted extensive attention. The most famous is the saltwater spillover from the NB into the SB [31,32].

The NC is the main channel of the Changjiang Estuary. Its water diversion ratio varies with the river discharge and tide from 65% to 75% [33,34,35]. The Qingcaosha Reservoir, the largest estuarine reservoir in the world, was built within the upper reaches of the NC in 2010 and supplies approximately 50% of the freshwater for Shanghai [36,37,38]. Saltwater intrusion in the NC originates from multiple sources. First, saltwater intrudes from the open sea along the downstream NC; this intrusion is the weakest among the four outlets due to its higher water diversion ratio. Under extreme conditions, such as persistent strong northerly winds, the NC also experiences severe saltwater intrusion [38,39]. Second, saltwater spills over from the North Branch into the upstream South Branch during spring tide and is then transported downstream by runoff [40]. Finally, saltwater originates from the northern outlet of the NC [41]. Lateral saltwater intrusion is affected by various factors, such as tides, wind and diversion ratio [31,32,33,42,43].

The HC, which was created by a great flood in 1954, is the only independent northwest–southeast channel in the Changjiang Estuary [44]. As the channel connecting the NC to the NP, the HC is an invaluable waterway for shipping activities. However, the topography of the HC has changed greatly in recent years due to intensive human activities, such as the Deep Waterway Project, the construction of the Qingcaosha Reservoir and the implementation of reclamation projects on both sides of the HC [45,46]. Moreover, the tidal and seasonal variations in the water and salt transports within the HC have not been studied to date, and the dynamic mechanism responsible for the transport fluxes has not been quantitatively analyzed.

Accordingly, based on the results of a numerical model, this paper quantitatively studies the water and salt transports in the HC. The remainder of this paper is organized as follows: The Methods section introduces the measured data, the numerical model used herein, an analysis of the measurements, the model validation and the setting of the numerical experiments. The Results section presents the residual water and salt transports in the HC under actual conditions during the observation period, as well as those in the wet and dry seasons. The implications of the numerical experiments and analyses of the impacts of river discharge, tides, wind, and the interaction between river discharge and tides on the water transports in the HC are presented in the Discussion section. Finally, the Conclusions section summarizes the findings of this work.

2. Methods

2.1. Observations

Field observations were conducted in and around the HC from 11 to 21 May 2019. Three sites, A, B and C, were located along the HC from north to south (Figure 1). Site A was at the northern end of the HC at its intersection with the NC, site B was in the middle of the HC, and site C was at the southern end of the HC at its intersection with the NP. Tripods were placed to acquire continuous observations at each site. The observations lasted 9 days at site A and 2 days each during neap tide (12–15 May) and spring tide (19–22 May) at sites B and C, respectively. An acoustic Doppler current profiler (ADCP), conductivity-temperature-depth (CTD) and ocean-bottom seismometer (OBS) were installed on the tripods to collect data on the water level, salinity, and flow velocity and direction. During spring tide and neap tide, three research vessels conducted profile observations for at least 26 h for two full semidiurnal tides. The water samples were calibrated so that all instruments operated normally.

The river discharge measured at Datong Station was 31,000 m³/s at the beginning of the observation period on 11 May 2019 and 37,000 m³/s at the end (Figure 2a). The river discharge in this period was slightly lower than that in the dry season under the climatological state. From 11 to 18 May, the wind was mostly southeasterly, but it changed to northerly from 18 to 22 May. Strong winds were recorded during both the neap tide and the spring tide, with maximum wind speeds of 13.3 and 17.8 m/s, respectively.

At measurement sites A, B and C, two complete flood and ebb currents were observed during neap tide and spring tide, respectively. The ebb current durations were longer than the flood current durations. Under the influence of bottom friction, the highest flood and ebb current velocities appeared in the surface layer, while the lowest current velocities appeared in the bottom layer (Figure 3). At site A, during the neap tide, the former and latter ebb current durations were 6.30 and 7.67 h, and the corresponding flood current durations were 5.63 and 5.67 h. The maximum flood and ebb current velocities were 1.22 and 0.93 m/s, respectively (Figure 3a), and the residual unit width water fluxes in the surface and bottom layers (one-sixth of the water depth) were 0.27 and 0.24 m³/s (northeast), respectively. During the spring tide, the former and latter ebb current durations were 7.87 and 7.73 h, and the corresponding flood current durations were 4.40 and 4.77 h. The maximum flood and ebb current velocities were 1.51 and 1.14 m/s, respectively (Figure 3b), and the residual unit width water fluxes in the surface and bottom layers were 0.31 and 0.26 m³/s (northeast), respectively.

At site B, during the neap tide, the former and latter ebb current durations were 6.57 and 7.77 h, and the corresponding flood current durations were 5.87 and 5.43 h. The maximum flood and ebb current velocities were 1.33 and 1.04 m/s, respectively (Figure 3c), and the residual unit width water fluxes in the surface and bottom layers were 0.24 and 0.22 m³/s (north), respectively. During the spring tide, the former and latter ebb current durations were 8.00 and 7.60 h, and the corresponding flood current durations were 4.40 and 4.60 h. The maximum flood and ebb current velocities were 1.81 and 1.55 m/s, respectively (Figure 3d), and the residual unit width water fluxes in the surface and bottom layers were 0.23 m³/s (south) and 0.07 m³/s (southeast), respectively.

At site C, during the neap tide, the former and latter ebb current durations were 6.67 and 7.67 h, and the corresponding flood current durations were 5.80 and 5.51 h. The maximum flood and ebb current velocities were 0.86 and 1.18 m/s, respectively (Figure 3e), and the residual unit width water fluxes in the surface and bottom layers were 0.48 and 0.36 m³/s (southeast), respectively. During the spring tide, the former and latter ebb current durations were 8.00 and 7.60 h, and the corresponding flood current durations were 4.43 and 4.60 h. The maximum flood and ebb current velocities were 2.03 and 1.60 m/s, respectively (Figure 3f), and the residual unit width water fluxes in the surface and bottom layers were 0.33 m³/s (south) and 0.24 m³/s (east), respectively.

Water flows from the NP to the NC during flood tide and flows in the opposite direction during ebb tide. During the observation period, no obvious salinity gradient was observed near the HC. Therefore, the water flow was controlled mainly by the barotropic gradient force and was also affected by the surface wind stress and bottom friction. During ebb tide, driven by the barotropic gradient force, the ebb current velocity increased from north to south. By contrast, the flood current velocity presented a different trend between spring tide (decreasing from north to south) and neap tide (increasing from north to south).

2.2. Numerical Model Configuration

We used a two-way nesting unstructured quadrilateral grid, finite-differencing, three-dimensional estuarine and coastal ocean model (UnFECOM) developed by Ding et al. [47]. The Mellor-Yamada level 2.5 order turbulence closure module with the stability parameters from Kantha and Clayson was included [48,49]. A nonoscillatory advection scheme with 3rd-order spatial interpolation at the middle temporal level coupled with a total variation diminishing limiter (HSIMT) was used to solve the advection terms in the transport equations [50]. At the encrypted inner boundary, the HSIMT parabolic interpolation scheme and full-weighting operator scheme were adopted to maximize stability and conservation [51].

The model domain covered the entire Changjiang Estuary, Hangzhou Bay and adjacent seas. In the north–south direction, the domain extended to 33.7° N and 27.5° N; to the east, the model domain extended to 124.9° E, and Datong Station was the boundary in the west. The total number of cells was 225 * 337 in the horizontal direction, and 10 uniform sigma layers were adopted in the vertical direction. The resolution was approximately from 300 to 500 m around the estuary and 10,000 m at the open boundaries (Figure 4a). The area near the HC was partially encrypted at a ratio of 1:3 with a resolution of approximately 100–150 m (Figure 4c).

The open boundaries were driven by 16 astronomical constituents, namely, M2, S2, N2, K2, K1, O1, P1, Q1, U2, V2, T2, L2, 2N2, J1, M1 and OO1, which were derived from the NaoTide database (http://www.miz.nao.ac.jp, accessed on 20 December 2020). The river boundary was specified in the form of the volume flux measured at Datong Station or as a constant value. The initial elevation and water current were set to zero. The wind data were downloaded from ERA5 (https://cds.climate.copernicus.eu, accessed on 22 December 2020). The initial salinity field was taken from the Ocean Atlas in Huanghai Sea and East China Sea (Hydrology) [52].

2.3. Model Validation

The numerical model employed herein was validated for the current and salinity [47]. To further demonstrate the model capacity, the measured data were used to further validate the model. Due to the increased river discharge during the observation period, the salinity approached zero. Therefore, only the water level, current velocity and current direction were used for the model validation.

The numerical model was cold-started on 1 April 2019. The daily river discharge at Datong Station was adopted. The correlation coefficient (CC), root-mean-square error (RMSE) and skill score (SS) were used to quantify the validation:

$$\mathrm{CC}=\frac{{\displaystyle \sum }\left(X_{mod}-\overline{X_{mod}}\right)\left(X_{obs}-\overline{X_{obs}}\right)}{{\left[{\displaystyle \sum }{\left(X_{mod}-\overline{X_{mod}}\right)}^2{\displaystyle \sum }{\left(X_{obs}-\overline{X_{obs}}\right)}^2\right]}^{1/2}};$$

(1)

$$\mathrm{RMSE}={\left[{\displaystyle \sum }{\left(X_{mod}-X_{obs}\right)}^2/N\right]}^{1/2};$$

(2)

$$\mathrm{SS}=1-\frac{{\displaystyle \sum }{\left(X_{mod}-X_{obs}\right)}^2}{{\displaystyle \sum }{\left(X_{obs}-\overline{X_{obs}}\right)}^2},$$

(3)

where $X$ and $\overline{X}$ are the variable of interest and the mean value, respectively. The SS was introduced as a statistical metric to describe the degree to which the observed deviations from the observed mean correspond to the simulated derivations from the observed mean (Murphy, 1988). In this article, we classified the model results into four categories to evaluate model performance according to the SS: >0.65, excellent; 0.5–0.65, very good; 0.2–0.5, good; and <0.2, poor.

The elevation results are compared in Figure 5, and the skill assessments are summarized in

Table 1. CCs, RMSEs and SSs of the elevation at measurement sites A, B and C.

	A		B		C
	Neap	Spring	Neap	Spring	Neap	Spring
CC	0.96	0.99	0.96	0.99	0.96	0.99
RMSE	0.16	0.25	0.20	0.23	0.21	0.23
SS	0.96	0.97	0.92	0.98	0.92	0.98

. The water level elevations at the measurement sites were well simulated, with high CCs (>0.96 during the neap tide and >0.99 during the spring tide), low RMSEs (<0.21 m during the neap tide and <0.25 m during the spring tide), and excellent SSs (>0.92 during the neap tide and >0.97 during the spring tide). In general, the simulation results during the spring tide were better than those during the neap tide.

Comparisons of the current velocity and direction are presented in Figure 6, and the skill assessments of the current velocity are summarized in

Table 2. CCs, RMSEs and SSs of the current velocities at the measurement sites (value in the surface layer/value in the bottom layer).

	A		B		C
	Neap	Spring	Neap	Spring	Neap	Spring
CC	0.68/0.63	0.58/0.33	0.62/0.65	0.56/0.58	0.99/0.92	0.88/0.83
RMSE	0.22/0.17	0.33/0.35	0.32/0.21	0.45/0.28	0.14/0.08	0.25/0.21
SS	0.82/0.84	0.75/0.66	0.76/0.84	0.68/0.77	0.95/0.97	0.91/0.88

. Except for site A during the spring tide, the CCs exceeded 0.56. Furthermore, the SSs were above 0.65 at each site and greater than 0.9 at site C during neap tide, meaning that the model performance was very good.

2.4. Residual Transport of Water and Salt

To quantitatively analyze the subtidal movements of water and salt in the HC, the instantaneous water transport per unit width through the water column is defined as:

$$\overrightarrow{Q}={{\displaystyle \int }}_{-1}^{0}D\overrightarrow{V}d\sigma .$$

(4)

The residual transports per unit width of water $(\overrightarrow{R_w})$ and salt $(\overrightarrow{R_s})$ are calculated as:

$$\overrightarrow{R_w}=\frac{1}{T}{{\displaystyle \int }}_0^T\overrightarrow{Q}dt,$$

(5)

$$\overrightarrow{R_s}=\frac{1}{T}{{\displaystyle \int }}_0^T{{\displaystyle \int }}_{-1}^{0}D\overrightarrow{V}sd\sigma dt,$$

(6)

where $\sigma$ is the relative depth (0 at the surface and −1 at the bottom), $\overrightarrow{V}$ is the current velocity vector, D is the total water depth, and T is the tidally averaged period. In this article, T was defined as the semidiurnal tidal cycle to remove the semidiurnal and diurnal tidal signals.

2.5. Numerical Experiments

Eleven numerical experiments were set up (

Table 3. Numerical experiment setting.

	Tide	River Discharge	Wind	Salinity at Open Boundaries
EX0	open	realistic	realistic	climatic
EX1	open	49,900 m3/s	climatic (wet season)	climatic (wet season)
EX2	open	12,428 m3/s	climatic (dry season)	climatic (dry season)
EX3	open	close	close	close
EX4	close	12,428 m3/s	close	close
EX5	close	49,900 m3/s	close	close
EX6	close	close	climatic (dry season)	close
EX7	close	close	climatic (wet season)	close
EX8	open	20,000 m3/s	close	close
EX9	open	30,000 m3/s	close	close
EX10	open	40,000 m3/s	close	close
EX11	open	49,900 m3/s	close	close

). Based on the real river discharge and wind during the observation period in May 2019, EX0 simulates the hydrodynamics and water transport in the HC during the measurement period. EX1 and EX2 simulate the hydrodynamics and salinity fields in July (wet season) and February (dry season). The river discharge in July and February takes monthly mean values of 49,900 and 12,428 m³/s, respectively, from 1950 to 2019. The wind also takes the monthly mean value: in July, the southeasterly wind speed is approximately 2.5 m/s near the HC, and in February, the northerly wind speed is approximately 5 m/s. EX3–EX7 simulate a single factor to investigate the hydrodynamics and discuss the dynamic mechanism responsible for the water transport in the HC. EX3 considers only the tide and ignore the river discharge, wind and salinity. EX4 and EX5 consider only the river discharge of 12,428 m³/s in February (dry season) and 49,900 m³/s in July (wet season), respectively, and ignore the tide, wind and salinity. EX6 and EX7 investigate the hydrodynamics under only wind action in the dry season and wet season. In EX8–EX11, the effects of the nonlinear interactions between the river discharge and tide on the hydrodynamics are evaluated without wind; in these four experiments, the river discharge is successively set to 20,000, 30,000, 40,000 and 49,900 m³/s. Each experiment is run for 30 days to stabilize the model, and the next month’s data are taken as the output for further analysis and discussion. The simulation of salinity starts two days after the model has been running.

3. Results

3.1. In the Realistic Case

Under realistic conditions, EX0 was conducted to simulate the hydrodynamics and salinity field in the Changjiang Estuary during the observation period. The intrusion of saltwater around the river mouth was stronger during spring tide than during neap tide. Therefore, we present the vertically averaged salinity at the flood slack during spring tide (Figure 7), showing that no saltwater intrusion occurred around the HC due to the increased river discharge in the measurement period.

The surface currents at the maximum flood and maximum ebb (reference site is site B) during the neap tide and the spring tide are depicted in Figure 8. The flood current was northwesterly, and the ebb current was southeasterly. The flood and ebb currents in the HC were controlled by the barotropic gradient force generated by the difference in the water level elevation between the NP and NC. During the neap tide, the flood current velocity was larger than the ebb current velocity. Furthermore, the flood and ebb current velocities were distinctly larger during the spring tide than during the neap tide.

During flood tide, water entered the NC through the HC, where the flow direction was reversed during ebb tide. The water flux across the section could reflect the mean current velocity, and its peaks corresponded to the maximum flood current and maximum ebb currents. During the neap tide, the time of the maximum water flux in the HC was later than that across the sections in the NC and NP (Figure 9a); during the spring tide, the time was earlier than that across the sections in the NC and NP (Figure 9b).

The residual water transports during the neap tide and spring tide (four tidal cycles, ~52 and ~49.5 h, respectively) are shown in Figure 10. Previous studies have pointed out that under the influence of runoff, the residual water transports in the estuary are seaward. There is no doubt that the residual water transports in the NC and NP are seaward. The residual water transport was larger in the NC, the main runoff channel; however, the HC is oriented northwest–southeast, meaning it intersects the NC and NP at a certain angle. Therefore, even if the river discharge reached 30,000 m³/s during the observation period, the influence of wind on the residual water transport was not negligible. The residual water transports flowed from southeast to northwest (almost landward) under the southeasterly wind during the neap tide, whereas during the spring tide, the wind turned northerly, and the residual water transports flowed southeastward (almost seaward). Taking the ebb current direction as positive, the residual water transports in the HC were −1092 and 332 m³/s during the neap tide and spring tide, respectively.

3.2. In the Climatic Case

Under the interaction among river discharge, tides and wind, the water transport is seaward during the wet season (EX1) and dry season (EX2) in the Changjiang Estuary (Figure 11). In EX1, although the wind blows southeasterly in the wet season, the residual water transport is still southeastward (Figure 11a,b). During neap tide, the residual water transports across Sec.3 in the NC and Sec.5 in the NP are 24,471 and 13,213 m³/s, respectively, and the residual water transport across Sec.1 in the HC is 677 m³/s, accounting for 2.8% and 5.1% of the residual water transports in the NC and NP, respectively. During spring tide, the residual water transports across Sec.3 in the NC and Sec.5 in the NP are 24,072 and 12,536 m³/s, respectively, and the residual water transport across Sec.1 in the HC is 245 m³/s, accounting for 1.0% and 2.0% of the residual water transports in the NC and NP, respectively. The southeastward residual water transport in the HC during neap tide is stronger than that in spring tide and has a greater impact on the NP. Moreover, the river discharge in the wet season is much larger than that in the observation period in May 2019; therefore, there is no salt transport around the HC.

In EX2, the diminished river discharge in the dry season significantly reduces the seaward residual water transport (Figure 11c,d). During neap tide, the residual water transports across Sec.3 in the NC and Sec.5 in the NP are 5381 and 4264 m³/s, respectively, and the residual water transport across Sec.1 in the HC is 1278 m³/s, accounting for 23.8% and 30.0% of the residual water transports in the NC and NP, respectively. During spring tide, the residual water transports across Sec.3 in the NC and Sec.5 in the NP are 6776 and 5719 m³/s, respectively, and the residual water transport across Sec.1 in the HC is 1328 m³/s, accounting for 19.6% and 23.23% of the residual water transports in the NC and NP, respectively. Compared with those during neap tide, the residual water transports in the NC and NP during spring tide are higher, and the southeastward residual water transport in the HC is slightly greater (Figure 11c,d). Hence, the HC is an important channel for water transport between the NC and NP in the dry season as well, and its role cannot be ignored.

Saltwater intrusion around the HC occurs in the dry season; therefore, here, we present only the salinity simulation results of EX2 (Figure 12 and Figure 13). To show the different stages of saltwater intrusion in more detail, in addition to that, during spring tide and neap tide, the intrusion characteristics of saltwater at middle tide after neap tide and middle tide after spring tide are also presented. During neap tide, the salinity is below 1.0 psu on the north side of the HC and above 1.0 psu on the south side (Figure 12a). When the Deep Waterway Project was built in the NP, it was dredged to a water depth of 12.5 m. Due to the deeper water depth and weaker vertical tidal mixing in the NP, the salinity front moves landward under the action of baroclinic pressure, which increases with water depth; as a result, the salinity is more than 3 psu at the bottom along Sec.6 (Figure 13a). The most serious saltwater intrusion occurs during middle tide after neap tide (Figure 12b). On the one hand, the weaker tidal mixing in this period and the previous neap tide is beneficial for the salinity front to move upstream under the action of baroclinic pressure. On the other hand, the highly saline water that enters the NC from the northern outlet of the NC moves upstream along the north side of the NC. The bottom salinity in the northern part of Sec.6 is more than 10 psu (Figure 13b). Although the salinity in the NP also increases, it reaches only 6.0 psu, far less than that in the NC. Thus, the salinity in the HC exhibits a large horizontal gradient and stratification. During spring tide, the enhanced tidal mixing weakens the baroclinic effect, which decreases the intrusion of saltwater compared to that during neap tide and middle tide after neap tide. However, strong tidal currents still trigger severe saltwater intrusion. The salinity reaches approximately 6 psu at the northern entrance of the HC and more than 3 psu at the southern entrance (Figure 12c). A stronger salinity front is also produced within the HC (Figure 13c). During middle tide after spring tide, the tidal energy decreases, and the baroclinic effect weakens, resulting in significantly weaker saltwater intrusion. Consequently, the salinity around the HC drops to ~1 psu (Figure 12d). Along Sec.6, the salinity in the HC is slightly higher during middle tide after spring tide than during neap tide and is well mixed. The salinity in the NP is obviously lower during middle tide after spring tide than during neap tide (Figure 13d).

The strong salinity front in the HC produces a strong southeastward baroclinic gradient force during middle tide after neap tide and during spring tide (Figure 12b,c); this force obviously enhances the water and salt transports during ebb tide and weakens them during flood tide (Figure 14). By contrast, during neap tide, the salinity gradient in the HC is weaker, and the water and salt transports during ebb tide are still distinctly larger than those during flood tide because the northerly wind is more influential during neap tide than during spring tide [38,39].

The residual salt transport distributions around the HC under the four tidal patterns in EX2 are shown in Figure 15. During neap tide and middle tide after neap tide, salt is transported mainly from the NC, flowing landward, and some of this salt enters the HC and then flows seaward in the NP. During spring and middle tide after spring tide, salt is transported mainly from upstream of the NC, where it originates from saltwater spilling over from the North Branch into the South Branch, and some salt enters the HC and then flows seaward in the NP [30,32,43]. Therefore, the HC is an important salt exchange route between the NC and NP, especially during spring tide and middle tide after neap tide. During neap tide, the residual salt transport fluxes are −15.21 and −13.7 t/s across Sec.4 in the NC and Sec.5 in the NP, respectively, and the flux is 0.38 t/s across Sec.1 in the HC (Figure 15a), which accounts for only 2.5% and 2.8% of the residual salt transport fluxes in the NC and NP, respectively. During middle tide after neap tide, the residual salt transport fluxes are −32.28 and 18.08 t/s across Sec.4 in the NC and Sec.5 in the NP, respectively, and the flux is 16.23 t/s across Sec.1 in the HC (Figure 15b), accounting for 50.3% and 89.8% of the residual salt transport fluxes in the NC and NP, respectively. These findings reveal that salt being transported in the NC is derived mainly from the northern outlet of the NC and then crosses the NC to the south, whereupon one branch moves downstream and another enters the HC. During spring tide, the residual salt transport fluxes are 29.27 and 22.23 t/s across Sec.4 in the NC and Sec.5 in the NP, respectively, and the flux is 12.61 t/s across Sec.1 in the HC (Figure 15c), which accounts for 43.1% and 56.7% of the residual salt transport fluxes in the NC and NP, respectively. During middle tide after spring tide, the residual salt transport fluxes are 15.27 and 8.20 t/s across Sec.4 in the NC and Sec.5 in the NP, respectively, and the flux is 1.78 t/s across Sec.1 in the HC (Figure 15d), accounting for only 11.6% and 21.8% of the residual salt transport fluxes in the NC and NP, respectively.

4. Discussion

In this section, the dynamic mechanism responsible for the water transports in the HC was numerically investigated by considering only tides (EX3), the river discharge in February (EX4) and July (EX5), the wind speeds in February (EX6) and July (EX7), and the interaction between river discharge magnitudes of 20,000, 30,000, 40,000 and 50,000 m³/s and tides (EX8–EX11).

4.1. Tides

In EX3, tides drive the water from the NP into the NC through the HC, resulting in northward residual water transport in the HC (Figure 16a,b). During neap tide, the water transport in the HC is 5.94 × 10⁸ m³ during flood tide and 5.26 × 10⁸ m³ during ebb tide. During spring tide, the water transport in the HC is 1.05 × 10⁹ m³ during flood tide and 8.23 × 10⁸ m³ during ebb tide, which are increased by 76% and 56% compared with those during neap tide, respectively. The residual water transports in the HC during spring tide and neap tide are −367 and −1295 m³/s, respectively, indicating that the residual water transport is much larger during spring tide than during neap tide.

4.2. River Discharge

In EX4 and EX5, only a river discharge of 12,428 m³/s in February and 49,900 m³/s in July is considered. The water transports in the NC, South Channel and NP are seaward, and some of the water transport flows into the HC and then into the NP (Figure 16c,d). In EX4, the residual water transport is 4864 m³/s across Sec.3 in the NC and 1777 m³/s across Sec.1 in the HC, where the latter accounts for 36.5% of the former. In EX5, the residual water transport is 21,950 m³/s across Sec.3 in the NC and 4846 m³/s across Sec.1 in the HC, where the latter accounts for 22.1% of the former. Compared with the dry season, when the river discharge increases by 301%, the water flux in the NC increases by 351.1%, which is larger than the increase in river discharge. The water flux in the HC increases by only 172.7%, which is much smaller than the increase in river discharge. The water transport in the HC flows southeastward due to these variations in river discharge and northwestward due to the tidal behavior, and the magnitude of the former is larger than that of the latter.

4.3. Wind

In EX6, the northerly wind is ~5 m/s over the open sea and 2.3 m/s near the HC. Under landward Ekman transport, the residual water transport flows landward in the NC; some water is transported into the HC and then seaward flows in the NP (Figure 16e). The drag effect of the local wind also drives the southeastward water transport in the HC, and the water flux reaches 1421 m³/s. In EX7, the southeasterly wind is ~2.5 m/s over the open sea and 3.5 m/s near the HC. Under the wind force, the residual water transport is landward in the South Passage and NP and is seaward in the NC (Figure 16f). The water flux is −2168 m³/s in the HC, partly due to the local wind drag effect.

4.4. Tide and River Discharge Interactions

The flood and ebb tides in the HC are controlled mainly by the barotropic gradient force, which is determined predominantly by the river discharge and tides, which interact nonlinearly. EX8–EX11 simulate the residual water transports in the HC under the interactions of tides and different river discharge intensities. When the river discharge increases, the water fluxes across the sections in the NC, South Channel, NP and South Passage during ebb tide all increase, while those during flood tide decrease, and there is a strong linear relationship (Figure 17). However, the water flux in the HC has a different response to the river discharge during spring tide and neap tide. During neap tide, the change trend is the same as that in the NC, South Channel, NP and South Passage (Figure 17a); during spring tide, as the river discharge increases, the water flux during ebb tide increases, and the water flux during flood tide also displays an increasing trend (Figure 17b). In EX8–EX11, the residual water transports across the HC are (in order) 1035, 1132, 1171 and 1276 m³/s during neap tide and 141, 89, 175 and 300 m³/s during spring tide, respectively. The influence of the increasing river discharge on the residual water transport in the HC is relatively small in the cases with no wind or other factors. Under only the interaction between river discharge and tides, the residual water transport in the HC flows from the NC to the NP and is much greater during neap tide than during spring tide.

5. Conclusions

Based on a two-way nesting unstructured quadrilateral grid, finite-differencing, three-dimensional estuarine and coastal ocean model, the tidal and seasonal variations in the water and salt transports in the HC were simulated, and the dynamic mechanism responsible for these transports was analyzed. Comparing the water level elevation, current velocity and current direction measurements in May 2020 indicates that the numerical model can successfully simulate the hydrodynamic processes within the Changjiang Estuary.

The residual water and salt transports in the HC flow southeast from the NC to the NP during both the wet season and the dry season. In the wet season, the residual water transport in the HC is 677 m³/s during neap tide, accounting for 2.8% and 5.1% of the residual water transports in the NC and NP, respectively, and is 245 m³/s during spring tide, accounting for 1.0% and 2.0% of those in the NC and NP, respectively. By contrast, the residual salt transport is zero. In the dry season, the residual water transport in the HC is 1278 m³/s, accounting for 23.8% and 30.0% of the residual water transports in the NC and NP, respectively, and the residual salt transport is 0.38 t/s, accounting for 2.5% and 2.8% of residual salt transports in the NC and NP, respectively, during neap tide. During spring tide, the residual water transport in the HC is 1328 m³/s, accounting for 19.6% and 23.23% of the residual water transports in the NC and NP, respectively, and the residual salt transport is 12.61 t/s, accounting for 43.1% and 56.7% of the residual salt transports in the NC and NP, respectively. Affected by the northerly wind and the southeastward baroclinic gradient force, the residual water and salt transports in the HC in the dry season are much greater than those in the wet season.

Numerical experiments on the dynamic mechanism of the water flux in the HC show that the water fluxes in the HC are −367 and −1295 m³/s during spring tide and neap tide, respectively, when considering only tides. By contrast, considering only the river discharge, the water flux in the HC in the dry season is 1777 m³/s, accounting for 36.5% of that in the NP, and that in the wet season is 4846 m³/s, accounting for 22.1% of that in the NP. Considering only the wind, the water flux in the HC is 1421 m³/s in the dry season and −2168 m³/s in the wet season. Considering only the interaction between tides and river discharge, the water transport flows from the NC to the NP, and the water flux during neap tide is much larger than that during spring tide.

The research results reported in this paper show that the HC serves as an important transport route for water flowing from the NC to the NP. Moreover, the HC acts as an important route for salt transport during the dry season, and therefore, its role cannot be ignored.