1 INTRODUCTION

A river plume is a complex system that is influenced by river discharge, tides, circulation, wind, complex bathymetry, and presence of artificial structures (Pu et al., 2016). Horner-Devine et al. (2009) proposed four dynamic regimes: the source (i.e., estuary), near-field plume, recirculating plume, and far-field plume. In this study, we focused on the mixing in the near-field plume. The near-field region is the jet-like area of initial plume expansion where the momentum of the plume layer dominates its buoyancy, and results in intense mixing (Horner-Devine et al., 2015). Compared to estuaries, characteristics of near-field plume are three-dimensional because river plume is not constrained to river channels. The near-field river plume can be characterized by supercritical Froude numbers, intense mixing, and rapid water mass modification (Stacey et al., 2011). The turbulence in the near-field plume is closely linked to plume spreading (e.g., Chen et al., 2009; MacDonald and Chen, 2012) because it functions as the source of energy and matter for the far-field region (Fong and Geyer, 2001; Lentz, 2004; Horner-Devine et al., 2009).

The Changjiang (Yangtze) River discharges a substantial amount of fresh water (9.24×10¹¹ m³/a, Shen et al., 2003), dissolved inorganic nitrogen (6.1×10¹⁰ mol/a, Liu et al., 2003), and dissolved organic carbon (1.18×10⁷ t/a, Dagg et al., 2004) into the Yellow and East China Seas (Fig. 1). These discharges have caused an increase in eutrophication and harmful algal blooms, as well as hypoxia events (Shen et al., 2011; Kang et al., 2016; Kong et al., 2016). The temporal and spatial variations of these detrimental events are correlated to the mixing and spreading of the Changjiang River plume (i.e. Rabouille et al., 2008; Zhang et al., 2019).

The discharge of the Changjiang River has a distinct seasonal variation that ranges from 1×10⁴ to 6×10⁴ m³/s (Zhu et al., 2001), depending on the year and the season. The Changjiang estuary is a mesotidal estuary with a tidal range between 4 m and 5 m during the spring tide and 2 m during the neap tide (Zhu et al., 2015). Currents off the Changjiang River include the north intrusion of the bottom Taiwan Warm Current inshore branch (Wang et al., 2019) and the Yellow Sea Coastal Currents along the Jiangsu Province (Xuan et al., 2016).

Mixing and stratification of the Changjiang River plume are rather complicated. The combination of discharge variability and tidal currents cause changes in mixing intensities over time at the same location (Shen and Zhang, 1992; Ding et al., 2018). Additionally, tidal forcing enhances vertical mixing in the near-field plume, which can affect the exchange of water masses in river plumes (Wang et al., 2015).

Early works studied the spreading of the Changjiang River far-field plume (Beardsley et al., 1985; Bang and Lie, 1999; Chang and Isobe, 2003; Chen et al., 2008; Shi and Lu, 2011; Rong and Li, 2012; Wu et al., 2014; Ge et al., 2015; Zhu et al., 2015; Yuan et al., 2016). The far-field plume is usually called Changjiang Diluted Water (CDW) and has a salinity less than 31; the near-field Changjiang plume has a salinity less than 26 (Mao et al., 1963), which is consistent with the definition of near-field plume conditions (Hetland, 2005). As compared to the substantial research on spreading in the Changjiang River far-field plume, fewer observation studies focus on the mixing process in the Changjiang near-field plume. Numerical models are used to study mixing in the Changjiang River plume as an alternative, since there is a high cost for direct observation. The mixing between CDW and surrounding water was first studied by Mao et al. (1963). They defined the near-field (salinity less than 26) and far-field plumes (salinity higher than 32) in the Changjiang River. Furthermore, vertical and lateral diffusions have been estimated using a simplified turbulent diffusion equation. Shen and Zhang (1992) determined that the main mechanisms that control the mixing of salt water and fresh water in the Changjiang estuary are the bottom shearing turbulent diffusion caused by flood current and the advection during ebb current. Using numerical experiments, Wu et al. (2011) proposed that tidal forcing increased the vertical mixing, resulting in a strong horizontal salinity gradient at the northern side of the mouth of the Changjiang River along the Jiangsu coast. An additional three-dimensional hydrodynamic model, the Coupled Hydrodynamical-Ecological Model for Regional and Shelf Seas (COHERENS), was employed by Shi and Lu (2011) to understand mixing, stratification, and circulation in the Changjiang River plume. However, numerical models cannot accurately estimate mixing near estuaries owing to complicated bathymetry and physical environments. More observations are needed to quantify turbulent mixing in the near-field plume of the Changjiang River.

In addition to numerical studies, other studies have been performed to characterize mixing in the Changjiang River plume using the gradient Richardson number and the Froude number. The gradient Richardson number, , is the ratio of the stabilizing forces of density stratification (∂ρ/∂z) to the destabilizing influences of velocity shear (∂u/∂z), where g is the gravity gravitational acceleration and ρ₀ is density. Ri_g is used to characterize mixing, with a critical value of 1/4 (Howard, 1961; Miles, 1961). Competition between shear and stratification results in an upper layer Froude number, (, where reduced gravity is g'=gΔρ/ρ, and u₁ and h₁ represent the upper layer velocity and depth, respectively) that remains O(1) (Armi and Farmer, 1986; MacDonald and Geyer, 2005).

Early studies were mainly based on parameter estimates that controlled stratification, such as Ri_g and F₁, because of a lack of direct microstructure observations in the Changjiang River plume. Mao et al. (1993) defined the front of the Changjiang River plume by calculating F₁, and proposed that the front induced strong mixing in the river plume. Shen and Zhang (1992) calculated the stratification coefficient (buoyancy frequency, ), to study the main mechanisms that control the mixing in the Changjiang estuary. Ni et al. (2012) studied the variation of Ri_g with the tidal phase.

A number of studies have directly measured the turbulence structure in other river plumes. Wright and Coleman (1971), who measured the currents and density structure of the flow out of South Pass, which is one of the Mississippi River Delta Head of Passes, first described the characteristic features of the supercritical outflow region. Stacey et al. (1999) used an Acoustic Doppler Current Profiler (ADCP) variance method to investigate the evolution of turbulence in the partially mixed San Francisco Bay. MacDonald and Geyer (2004) proposed control volume approaches to quantify indirectly the turbulence in the Fraser River. Nash et al. (2009) studied the estuary Richardson number, Ri_E (Fischer, 1972), using downward looking ADCP in the Columbia River, which suggested that turbulent mixing can be predicted through Ri_E.

Modern numerical models do not adequately resolve processes within the supercritical near-field plume. Because the water leaving the river mouth may be modified tremendously within the near-field, both numerical and analytical predictions of far-field plume structure could be inaccurate if they are based directly on estuary outflow properties. However, only rare examples can be found in the Changjiang River plume. Recently, Matsuno et al. (2006) presented the results of microstructure measurements of the turbulent kinetic energy (TKE) dissipation rate (ε) in the far-field Changjiang River plume. It is difficult to find work that studies turbulent mixing in the near-fi eld Changjiang River plume, although it is important to understand this behavior. For that reason, observation of the turbulent energy dissipation rate (ε) in the near-field plume of the Changjiang River is required. We have obtained high quality microstructure data that can be used to study the distribution and variation of TKE dissipation. This paper presents the quantitative results of the observed TKE dissipation rate in the near-field plume and discusses its variation with tide.

Our paper is organized as follows: first, we describe our fieldwork measurements and methods (Section 2); we then describe the variation of TKE dissipation rate in a tidal cycle (Section 3). In Section 4, we discuss the potential mechanism that affects the TKE dissipation rate. Conclusions are presented in Section 5.

2 MATERIAL AND METHOD 2.1 Measurement

To measure TKE dissipation in the near-field Changjiang River plume, we used the newly developed Turbulence Ocean Microstructure Acquisition Profiler (TurboMAP). TurboMAP is a quasi-free-falling vertical profiler that measures hydrographic parameters (conductivity, temperature, and pressure) and turbulent parameters (∂u/∂z, where u and z are the turbulent velocity fluctuations and depth, respectively). The turbulent velocity fluctuations are measured using two standard shear probes, which are cross-installed. All sensors are set onto a parabolically shaped cap at the head of the instrument to ensure the sensors are undisturbed during deployment. TurboMAP descended freely with a sinking velocity of approximately 0.5 m/s. A 5-mm diameter data transmission line covered with Kevlar rope was used to send the data to a PC onboard the research vessel and bring the instrument back to the surface after each downcast measurement. The TurboMAP equipment is described in detail by Wolk et al. (2002). The turbulent sensors sampled at a rate of 512 Hz.

Observations used in this study included microstructure data measured by TurboMAP, hydrographic data measured by conductivity temperature depth (CTD) sensors, and current data measured by ADCP. We took observations at station A, which is located off the mouth of the Changjiang River (31.00°N, 122.83°E, Fig. 1 station A). There are two crossed shear probes on TurboMAP, resulting in two profiles for each deployment. Measurements using TurboMAP were repeated twice every hour, yielding 52 downcasts for a diurnal observation from October 31, 2014 to November 1, 2014. The other four TurboMAP casts were taken on June 9, 2014 at stations CJ1a, CJ2, CJ2a, and CJ3, respectively. Analysis of the microstructure data and estimation of ε followed the procedures described by Wolk et al. (2002). CTD casts were conducted every 3 h following the measurement of TurboMAP (Table 1).

During the observation period, the downward looking bottom track ADCP (RDI-workhorse: 300 kHz) was mounted on the bottom of the vessel to obtain the background structure of the current and the fine-scale velocity shear with 2 m vertical bins. The blank area of ADCP is 3.1 m using this set up. For this reason, the first available layer of current in this study represented data at 5.1 m below the sea surface. The horizontal velocity was sampled every 2 s and the temporal mean for every minute was used to produce the dataset of the current velocity.

To study the structure of the Changjian River plume, CTD sensors (shown in Fig. 1) were cast along 31°N outside the mouth of the Changjiang River (CJ section), from 122.5°E to 126°E with a spatial interval of 0.5°. The depth changes from 17 m to 69 m in this crosssection. The stations locations are listed in Table 2.

2.2 Method

The rate of dissipation of kinetic energy in isotropic turbulence is defined as follows (Wolk et al., 2002):

where ν is the kinematic molecular viscosity taken constant at 10^-6 m²/s and the over line indicates the spatial average. ∂u/∂z represents either one of the two horizontal components of shear. The shear variance was calculated by integrating the spectrum of the shear signal ψ(k), where k is the vertical wavenumber and ψ(k) is the spectrum of the vertical shear (Osborn, 1980). The lower integration limit, k₁, is set to 1 cpm, and the upper limit, k_max, is variable following Shang et al. (2017). The shear spectrum ψ(k) is computed from the shear signal with half-overlapping 2-m depth segments, so the vertical resolution of ε is 1 m.

An estimation of the turbulent diffusivity in the river plume is necessary to determine the role of the river plume in the environment of the East China Sea. Measurements of dissipation rate were performed to estimate the magnitude of the turbulent diffusivity between the river plume and the ambient water. Dissipation rate ε refers to the energy dissipation, and diffusivity Kz refers to the mixing of mass. In a river plume, the turbulent diffusivity Kz is estimated using the squared buoyancy frequency N² and the observed dissipation ε, following Kz=γε/N² (Osborn, 1980) with the assumption that γ=0.2. N was calculated using the density gradient obtained from low-resolution CTD sensors mounted on the TurboMAP (Wolk et al., 2002).

Measurements of the microstructure velocity shear are susceptible to vibrations of the instrument and tethered rope. Furthermore, the vertical profile of the dissipation rate itself may have large variability over short time periods. Therefore, we repeated the deployments of the instruments twice for each cast. Several minutes were required for a single deployment and recovery. Therefore, the time difference between each pair of profiles was within 10 min. The two repeated profiles at each station generally showed similar shapes; the significant peaks of ε mostly coincided between the two casts. For similar types of ε from two contiguous deployments, we took the mean value from two shear probes of two deployments as the representative ε profile. However, if the structures of ε were not similar to each other, we checked the spectrum of the shear and used the mean of those ε values with a spectrum matching Nasmyth's spectrum. There were seven out of 52 deployments that produced different structures of ε from different shear probes. Because we took repeated downcasts, the seven downcasts all had high quality results. Thus, reliable and high-quality data were obtained from all time points.

3 RESULT 3.1 Structure of the Changjiang River plumes

The TKE dissipation distribution in the river plume varies with position along the plume (MacDonald et al., 2007; McCabe et al., 2009; Nash et al., 2009). To study the distribution of TKE dissipation in the nearfield plume, we investigated the basic hydrological structure of the Changjiang River plume. Based on salinity distribution of the CJ section, station A was located in the near-field plume. As shown in Fig. 2, the halocline was nearly horizontal near the coast (CJ1 to CJ2), but sloping at the plume front (CJ2 to CJ3). The surface salinity changed rapidly from 22 to 29 between stations CJ1 and CJ2. The thickness of the near-field plume was relatively thin during the observation period. The isohaline showed a lift-off at station CJ2. A lift-off point occurs in a river plume with supercritical state and enhanced shear (Stacey et al., 2011). The lift-off point region is characterized by strong stratification and shear, resulting in rapid mixing of the discharging water and ambient ocean water (Wright and Coleman, 1971). Rapidly thinning and mixing fresh water lay above the saline ambient water in the near-field region. The thickness of the near-field plume changes with tidal phase and river discharge, and the front also shifts with changes in those environmental factors (Geyer and Ralston, 2011). River discharge during the observation period from the Datong gauging station was approximately 1.97×10⁴ m³/s. River discharge was relatively low as compared with higher discharge (6.0×10 m³/s) during flood season (Zhu et al., 2001), which indicated CJ2 station is always influenced by the near-field plume of the Changjiang River.

The salinity was less than 30 at station CJ2 and increased to 33 at CJ3, which indicated that the water mass below the Changjiang River plume was north of the intrusion of the TWC. The high salinity water at stations CJ7 to CJ8 represents the Yellow Sea Warm Current (Lie et al., 2015), which occurs in winter.

3.2 Measurement of the turbulent energy

Mixing in the Changjiang River plume influences the exchange of different water masses, making it important to study the intensity and variation of mixing in the Changjiang River plume. The turbulent energy dissipation rate ε measured by TurboMAP was calculated by integrating the small-scale shear spectrum. We normalized the observed shear power spectrum using the Kolmogorov scale, which is on an order of 10^-3 m. This was compared with the empirical universal spectrum, which was measured by Nasmyth (1970), as shown in Fig. 3 (black solid line). The shape of the shear spectra was approximately fitted with the Nasmyth spectrum in most cases. The spectra in Fig. 3a1, b1, & d1 fit the Nasmyth spectrum for both shear probes (black solid line). However, the shape of the spectra of shear probe 2 (red line) in Fig. 3c1 was bent as compared with the Nasmyth spectrum in the low wave number range. The spectra of shear probe 1 in Fig. 3c1 fit the Nasmyth spectrum well, which is presumably caused by damping of shear probe sensitivity or from damage to the shear probes. The resulting ε is shown in Fig. 3a2, b2, c2, & d2. Considering that ocean turbulence is isotropic, the measured vertical structures of ε from different shear probes should be similar to each other in one deployment. The profile of ε was calculated using the mean ε of both shear probes from the repeated deployments for those with a well-defined spectrum, as shown in Fig. 3a2, b2, & d2. Since the spectra of shear probe 2 in Fig. 3c1 was bent, ε cannot be used. We used ε from shear probe 1 and then repeated deployment to represent the dissipation of this deployment. To achieve sufficient quality control, 26 profiles were gathered to study the tidal variation of the mixing in the near-field Changjiang River plume, and four profiles were collected to study the spatial variation.

The spatial distribution of ε within the Changjiang River plume was observed at the section along 31°N (Fig. 4). Station CJ1a was located within the near-field plume (on the shore side of station A) and stations CJ1a, CJ2, and CJ2a were located on the offshore side of station A. The upper layer ε at station CJ1a reached as large as 10^-4 W/kg, which is similar to the values located at 6 m, 10 m, 9 m, and 12 m, for stations CJ1a, CJ2, CJ2a, and CJ3, respectively. The value of ε at the bottom of the mixing layer at station CJ1a was 10-7 W/ kg, and ε values were 10^-8.1, 10^-7.9, and 10^-8.2 W/kg for stations CJ2, CJ2a, and CJ3, respectively. This indicates that the mixing became weak with the spreading of the plume. The dissipation rate increased to 10^-6 W/kg with the increase of depth at stations CJ1a, CJ2, and CJ3, which may be caused by the intrusion of the TWC inshore branch from the bottom. The TWC layer at station CJ2 did not increase significantly because the cast was too shallow to measure the enhanced mixing by the TWC inshore branch.

Salinity contour in Fig. 4 showed that station A is located in the near-field plume with salinity less than 26 (Mao et al., 1963; Hetland, 2005). The up layer dissipation rate ε in the near-field plume from our observation at CJ1a, CJ2 and CJ2a (Fig. 4) reached 10^-4 W/kg, and decreased with distance from the river mouth; similar to what was observed by MacDonald et al. (2007). The observed up layer ε at CJ3 was 10^-7 W/kg. The dissipation rate changed by three orders of magnitude from the near-field plume (CJ1a) to the far-field plume (CJ3). The vertical distribution of dissipation rates decreased to less than 10^-8 W/kg below the river plume and increased to 10^-6 W/kg at the layer where the TWC intruded at all the four stations (Fig. 4). The mixing of the Changjiang River plume is complex, and we aim to clarify the variation of mixing in the near-field plume.

3.3 Tidal variation of the turbulent energy dissipation rate ε

River plumes are influenced by the interactions of tides, wind, and freshwater discharge. Fresh water discharge is the fundamental forcing that distinguishes river plumes from other areas. Freshwater-saltwater interaction caused by freshwater input is the fundamental characteristic of river plumes. Barotropic pressure caused by tidal forcing produced oscillatory flow on semidiurnal, diurnal, and longer periods. Semidiurnal and diurnal tidal energies influence the tidal mixing during the flood-ebb and spring-neap cycles, which are the dominant sources of turbulent kinetic energy (Stacey et al., 2011). Wind stress is another important energy source of surface turbulent kinetic energy (Wang et al., 2015).

Hydrologic conditions in the near-field are significantly influenced by tidal phase and river discharge. The observation was carried out during the dry season in November with relatively low river discharge (19 700 m³/s). The daily discharge data showed that discharge increased by 1 300 m³/s during our observation. Wind speed was steady during the observation, and ranged from 4 m/s to 7.2 m/s (Fig. 5a). The observation period covered two tidal cycles with tidal ranges between 1.1 m and 2.0 m, as shown in Fig. 5a. The tidal heights presented in Fig. 5a were measured using an echo sounder installed on the bottom of the ship, and measurements were collected hourly.

Strong salinity variation was observed. The halocline rose and became thinner as the flood progressed. This suggests that during flood tide, local stratification associated with intrusion of the salt wedge dominated mixing at the interface. Fresh water was detected in the upper layer with salinity less than 26, which was considered the threshold of near-field water. The halocline stayed between 4 m and 8 m where most shear enhanced turbulence occurred in the river plume. The intensity of the halocline varied with tidal phase as shown in Fig. 5b. The halocline decayed with the flood tide at station A, because the flooding water forced the fresh water back to the river mouth. Station A was controlled by the ambient water at the bottom, and the halocline became weak. During the ebb tide, the Changjiang River discharged a significant amount of fresh water into the East China Sea, and the fresh water enhanced the halocline in the upper layer. A high salinity core occurred under the halocline during the flood tide. The observed salinity was greater than 32, indicating that the source of the high salinity water was the TWC (Wang et al., 2019).

With the start of the second ebb tide, the upper halocline intensified dramatically. The surface salinity was as low as 23, which was significantly less saline than during first ebb. The isohaline with salinity of 28 was deeper than the first tidal cycle. The enhanced halocline could be induced by stronger upstream river discharges, which can also result in intense shear turbulent mixing as showed in Fig. 6.

The asymmetry in salinity observed at station A shows characteristics associated with tidal asymmetry (Jay and Smith, 1990; Simpson et al., 1990). Interaction between tidal flow and river discharge intends to establish stratification during ebb tide and destroy it during flooding. Tidal straining and internal tidal asymmetry are widely accepted mechanisms for tidal asymmetry. Simpson et al. (1990) explained the internal structure of Liverpool Bay using tidal straining, and Jay and Musiak (1994) used internal asymmetry to explain the observed strong mean flow and internal tides in the Columbia. Similar concepts have been applied to rivers such as the Hudson (Geyer et al., 2000) and numerous other river plumes. Turbulence in a river plume is mainly produced by shear at the density interface at tidal frequencies, which may destabilize the stable stratification induced by the river discharge. This mixing varies on tidal timescales, due to the tidal forcing of the interfacial shear between plume and ambient water. The turbulence dissipation along this interface originates from the destabilization induced by velocity shear.

Observations of the TKE dissipation rate were conducted in the near-field plume to study the tidal variation of turbulent mixing in the Changjiang River plume over 25 h (two semidiurnal tidal cycles). TKE ε was clearly related to the tidal phase as shown in Fig. 6. Our plume observations proved the occurrence of tidal asymmetries in mixing, and we noted that dissipation rates were stronger during ebb tide than flood tide. This is a result of increased shear associated with the river flow and tidal current-induced shear, which is similar to that observed in the Fraser River (Geyer and Smith, 1987; Nash et al., 2009). The magnitude of dissipation rate ε was larger than 10^-7 W/kg in the Changjiang River plume (Fig. 6). In addition, a strong dissipation rate ε was observed during ebb tide in the upper layer, which reached 10^-4 W/kg. The variation is similar to observations in the Merrimack River plume (Wang et al., 2015) and Columbia River plume (Nash et al., 2009). Comparing this to the salinity plot (Fig. 5b), we found that strong dissipation coincided with fresh water input. When the halocline was weaker and shallower, the magnitude of ε was smaller. The dissipation rate decreased under the river plume, resulting in a value of ε less than 10^-8 W/kg between 10 m and 14 m, which coincided with salinity distribution. Otherwise, the enhanced dissipation rate was observed during flood tides when a salt wedge intruded from the TWC layer bringing TWC water to the Changjiang River mouth. The intense dissipation rate can also reach 10^-5 W/kg. Therefore, the vertical structure of ε was large in the surface and TWC layers, and less at the halocline in the near-field plume.

4 DISCUSSION 4.1 The observed stratification N² and turbulent diffusivity Kz

In river plumes, TKE is produced by variations in velocity, which is known as shear production. Part of the energy is transferred to the Kolmogrov scale at which molecular viscosity can effectively dissipate the energy. The rate at which energy is dissipated by viscosity is the TKE dissipation rate (Stacey et al., 2011). To understand the vertical transport of mass in stratified turbulence, turbulent diffusivity for mass, Kz, is defined as the rate at which the transport occurs. Moreover, the diffusivity influences the rate at which heat, mass, contaminants, and biota are distributed throughout the turbulent fluid (Barry et al., 2001).

Time series of the square of the buoyancy frequency N², and the vertical turbulent diffusivity Kz are shown in Fig. 7 at station A. The surface layer of the river plume at station A was well mixed as indicated by a large Kz. The magnitude of Kz varied in time with values greater than 10^-4 m²/s and increased up to 10^-2 m²/s near the surface between t=4–11 h. The observed ε was large because of shear-induced turbulence during the first ebb between t=4–11 h, and N² was relatively small, which resulted in increased diffusion. Diffusion in the Changjiang River is similar to that observed by MacDonald and Geyer (2004) in the Fraser River. The magnitude of N² was small in the surface layer within the river plume and increased through the halocline, resulting in a small ε value of less than 10^-8 W/kg, as shown in Fig. 6. In the second flood beginning at t=11 h, there was enhanced stratification with fresher water at the surface layer, and both ε (Fig. 6) and Kz (Fig. 7) were reduced. The N² in the river plume was strong and reached as high as 10^-2 1/s², indicating that the diffusion was constrained by the enhanced stratification. High stratification was also observed at the TWC layer below 12 m during the flood tide. The high stratification indicated the intrusion of TWC water from the bottom. Higher Kz was also observed at the TWC layer in the morning (t=5–6 h) and evening (t=21–22 h). The enhanced diffusion indicated that the TWC water that intruded during flood tide mixed with upper layer water during ebb tide, which corresponded with a dramatically deeper halocline (Fig. 5). A large Kz was identified at the TWC layer, reflecting the large ε value shown in Fig. 6 and weak stratification. As shown in Figs. 5b & 7a, the halocline deepened during low tide at t=10 h, which indicated that the halocline was affected by internal waves. The displacement of vertical halocline was ~4 m and occurred over a period of ~4 h (Figs. 5b & 7a). The displacement could be caused by the passage of an internal wave of elevation with amplitude of ~4–5 m (Liu et al., 2009). Lee et al. (2006) reported a series of internal waves in the East China Sea, which were associated with the low barotropic tide. The deepened pycnocline here was also observed during low tide, which supports our hypothesis that the passage of an internal wave could be responsible for this phenomenon. The internal wave was accompanied by a significant enhanced vertical shear (Fig. 8a), low Ri_g zone (Fig. 8b), and intense turbulence dissipation (Fig. 6).

4.2 The competition between stratification and shear

The enhanced ε and N² overlapped with each other, as shown in Figs. 6 & 7a, and were both induced by the Changjiang River plume, whereas diffusion was constrained by stratification. The dissipation, stratification, and diffusivity showed tidal variations with a time lag. The influence of the tide on river plume turbulence is clearly shown from in ε, N², and Kz plots. The competition between shear and stratification plays a crucial role in plume dynamics in opposing ways: shear acts to enhance turbulence, while stratification acts to weaken mixing and transform TKE to potential energy. This competition is parameterized as gradient Richardson number, Ri_g=N²/S² (Turner, 1973), where N is the density stratification and S is the velocity shear. Ri_g provides instability (Ri_g < 1/4) (Miles, 1961). Strongly stratified shear flow in river plumes with Ri_g < 1/4 usually generates shear instability and breaks down into turbulence. In this process, kinetic energy from the mean flow is converted to viscous dissipation and buoyancy flux. Turbulence mixing continues until the system returns to a stable Ri_g.

The structure of ε and Kz depend on the competition between stratification, which acts to reduce mixing and TKE production, and shear intensification associated with the river discharge (Stacey et al., a necessary but insufficient condition for turbulent 2011). Shear intensification produces rapid acceleration in the near-field plume, resulting in a supercritical state and enhanced shear. During deployments of the microstructure profiler, the ADCP was used to obtain the fine-scale velocity shear. The contour of the square of vertical shear (S²) is shown in Fig. 8a. The shear components were calculated following Liu et al. (2009). We calculated smoothed velocity shear as S²= (∂u/∂z)²+(∂v/∂z)². The smoothed u and v were calculated using the second order Butterworth low-pass filter with a cut-off wavenumber of k_lp= 0.13 cpm. Because of the occurrence of the blank zone in the ADCP, the first layer of shear is at 6.1 m where the halocline and intense mixing remains. As the bin size of ADCP is 2 m, the ε and N² profiles were interpolated into 2-m intervals to match S² and obtain Ri_g.

In the river plume, strong shear was observed during ebb tide at t=6–14 h, which was consistent with a high dissipation rate. Velocity shear varied by orders of magnitude due to the variation in tidal current. A low shear layer occurred below the river plume where there was low dissipation and high stratification (Fig. 8). The shear at the TWC layer showed significant tidal variation, indicating that stronger shear occurred at ebb tide. The occurrence of strong shear did not coincide with high dissipation (Fig. 6) because of the variation in stratification, and the shear induced by intrusion of the TWC was not strong enough to overcome the stratification. The time series of S² (Fig. 8) and N² (Fig. 7) showed tidal asymmetries, as N² increased during ebbs and induced a highly sheared transition to flood.

In the upper layer where the river plume exists, the contour of the variation of Ri_g at station A (Fig. 8b) shows that there was a high potential for shear instability (Ri_g < 1/4) during the ebb tide at t=7–12 h (approximately 5-h long). This segment of low Ri_g occurred concurrently with a strong displacement in the halocline (Fig. 5b), which was assumed to be caused by the high discharge of fresh water and induced soliton through the observational period. Water below the river plume had a relatively high Ri_g, which is consistent with the structure of N², ε, and S². High shear instability was observed at the TWC layers, which showed similar patterns as Ri_g (Fig. 8b). Ri_g decreased during ebbs owing to enhanced shear from tidal currents. In addition, at all heights from the observed bottom to the lower boundary of the halocline, Ri_g was always greater than 1/4. The distribution of ε at the TWC layer was not consistent with the shear, because the strongest shear was trapped by high stratification. Early studies showed that shear instability developed in river plumes with strong gradients, where Ri_g is proportional to N but inversely proportional to S. Hence, if N and S are intensified or weakened by the same factor, Ri_g stays the same while the susceptibility to turbulence is increased. To study the mixing turbulence, we need to consider dissipation, diffusivity, and the gradient Richardson number.

5 CONCLUSION

The hydrographical condition in the near-field Changjiang River plume is complicated because it is occupied by a fresh river plume in the upper layer and saline TWC water intruding from the bottom during flood tide. Many biological organisms are sensitive to mixing between different water masses, because it provides a sufficient influx of nutrients to biological communities (Kudela and Peterson, 2009), and influences biological productivity through changing nutrients and light (Lohan and Bruland, 2006). Nevertheless, few observations have been performed to quantify turbulent mixing because of measurement difficulties. We used a newly developed microstructure profiler (TurboMAP) to investigate turbulent mixing and hydrographic parameters in the near-field Changjiang River plume.

Surface dissipation rate ε in near-field plume reached 10^-4 W/kg, and it decreased with distance from the river mouth. The observed ε was 10^-7 W/kg, which indicated the dissipation rate changed by three orders of magnitude from near-field to far-field plume. The dissipation rate decreased to less than 10^-8 W/kg below the river plume and increased to 10^-6 W/kg at the layer where the TWC intruded. In addition to the spatial variation of dissipation, variation of the mixing rate with time is important to study. We investigated the temporal distribution of turbulent mixing in the near-field plume. The strongest turbulent mixing was observed in the river plume during strong ebb tide, and the relevant dissipation and diffusion were observed to be more than 10^-4 W/kg and 10^-2 m²/s, respectively. Minimal mixing was also observed in the plume during flood tide and low discharge periods, and ε and Kz were 10^-7.5 W/kg and 10^-5.5 m²/s, respectively. The tidal cycle variations had a three orders of magnitude change of ε and Kz in the Changjiang River near-field plume, which is consistent with turbulent observation in other river plumes (MacDonald and Geyer, 2004; Nash et al., 2009; Wang et al., 2015). The depth of the halocline changed with tidal cycle, and stratification varied by one order of magnitude; less than ε and Kz. Stratification in the TWC layer followed the distribution of the halocline, which is opposite of the dissipation structure. The TWC layer mixing was influenced by a combination of the intrusion of the TWC and the river plume. Tidal current led to the intrusion and mixing between two water masses. The observed ε and Kz were as strong as in the river plume, but occurred for a relatively shorter time during ebb tide.

Velocity shear acts to enhance turbulence, while stratification weakens turbulent mixing. The competition between shear and stratification is estimated as Ri_g. To study the dynamics of turbulent mixing in the near-field plume, we deployed a shipmounted ADCP to obtain the shear distribution. The structure of velocity shear was similar to the dissipation rate in both the river plume and the TWC layer. However, the velocity shear in the TWC layer did not match the stratification structure. In high dissipation rate areas, the gradient Richardson number was less than the critical value (Ri_g < 1/4). The structure of Ri_g was also consistent with the shear and dissipation distribution, indicating that turbulent mixing in the near-field of the Changjiang River is controlled by a combination of shear induced by discharged river flow and tidal current.

6 DATA AVAILABILITY STATEMENT

The datasets generated during and/or analyzed during the current study are available from the corresponding author on reasonable request.

7 ACKNOWLEDGMENT

We thank Dr. NAN Feng from Institute of Oceanology, Chinese Academy of Sciences for the useful discussions on this topic.