The Roles of Different Mechanisms Related to the Tide-induced Fronts in the Yellow Sea in Summer

The Yellow Sea (YS) is a shallow, semi-enclosed shelf sea surrounded by the west coast of the Korean Peninsula and the east coast of mainland China (Fig. 1a). It is connected to the East China Sea (ECS) in the south. In the warm season, one of the characteristic compositions of the region is the Yellow Sea Cold Water Mass (YSCWM), which is a basin-scale water mass of low temperature, exerting significant influences on the three-dimensional circulation of the YS (Hu et al., 1991; Su and Huang, 1995). The tidal effect is another important feature of the YS. Propagating from the outer ocean into the YS, the tides have large amplitudes and produce strong residual currents through an interaction with the sea-floor topography (Lie, 1989; Lee and Beardsley, 1999; Fang et al., 2004; Xia et al., 2006). Therefore, conspicuous surface cold patches (SCPs) and corresponding tidal fronts are often found around the edge of YSCWM. Based on a joint Chinese-Korean investigation in the Yellow Sea in July 1997 (Zou et al., 2001), it was found that the cores of cold water patches off Subei Bank and Mokpo are about 3^°C-4^°C lower than the surrounding water. The cold water off Shandong Peninsula is observed to be quite stable in summer, and the central water is usually 4^°C -5^°C lower than the water nearby (Xia and Guo, 1983). Three striking surface thermal fronts——the Subei Bank Front (SBF), Shandong Peninsula Front (SPF), and Mokpo Front (MKF)——were observed. Sketch maps of these three fronts are given in Fig. 1a. Furthermore, previous model results suggested that the SBF, SPF and MKF are the most conspicuous fronts in summer in the YS (Ma et al., 2004; Moon et al., 2009; Lu et al., 2010). Several possible physical mechanisms, such as upwelling and tidal mixing, have been suggested to explain the tidal fronts in the YS, but previous studies have been unable to reach agreement on what roles these two mechanisms play, and how important they are, in the maintenance processes of the above three fronts.

Figure 1.  (a) Bathymetry (contours; units: m) of the study area [red box in (b); (30^°-38^°N, 119^°-128^°E)]. Blue lines show two representative sections along 34.1^°N (Sec. SB-MK) and 37^°N (Sec. SP). Red lines represent a sketch map of the three striking fronts (i.e., the SBF, SPF, and MKF). (b) Three-model nested system, in which the middle model covers most of the North Indian and West Pacific oceans (black box), and the inner- model includes most of marginal seas of China (blue box).

Tide-induced upwelling, which is generally considered a part of the YSCWM-related circulation, supplies cold water from the deep layer. Based on hydrographic and satellite observations, (Zhao, 1987) suggested that upwelling occurs near the frontal area in the western YS along the boundaries of the YSCWM. On the other hand, some earlier studies considered strong tidal mixing to play an important role in the formation of tidal fronts in the western YS. (Moon et al., 2009) suggested that tidal mixing occurs in the bottom boundary layer of sloping topography in the western YS, and generates horizontal tidal mixing fronts. (Liu et al., 2003) identified vertical velocities near the fronts along two sections using a numerical model, suggesting upwelling motions appear around all the fronts in the YS. In the theoretical studies on cross-frontal secondary circulation by (Garrett and Loder, 1981) and (Dong et al., 2004), the tidal front circulation was generally composed of multiple cells, and upward branches were highly related to fronts. However, hydrographic observations have revealed that the strong tidal stirring effect contributes to the cold coastal water around the front off Mokpo, and the tidal mixing in the area seems to have been enhanced by the presence of many small islands (Lie, 1986).

It seemed that a uniform mechanism for tidal fronts could not be obtained, and studies on the topic are also limited by a lack of observational data or by insufficient model resolutions. Moreover, it is also very hard to distinguish between the effects of vertical mixing and vertical convection, because when upward motions bring the cold water to the surface, the mixing effects of entrainment and encroachment always occur simultaneously. The mechanisms of tidal mixing and tide-induced upwelling are not independent; they are highly related. According to a model study by (Lu et al., 2010), different tidal mixing effects between shallow and deep water lead to conspicuous fronts. The baroclinic pressure gradient force, which stems from the intense density differences across the fronts, triggers a secondary circulation in the frontal zone, and the upwelling appears as a branch of the circulation on the mixed side of the fronts. In most previous observational studies, it is even harder to tell which is more important to the front merely by analyzing the thermal fields. Besides these two major factors, strong tidal currents may also have influences on the fronts by their intense horizontal advection effects. (Garrett and Loucks, 1976) suggested that the centrifugal forces related to strong tidal currents flowing around convex coastlines may cause upwelling and subsequent fronts. A similar mechanism was also applied by (Xia and Guo, 1983) to investigate the front off Shandong Peninsula.

In this paper we seek to clarify which mechanism is more important in the formation of specific fronts and explore the interaction between different mechanisms by analyzing results from a three-dimensional numerical model with a complete set of physical processes, focusing on the tidal forcing. The remainder of the paper is organized as follows. The numerical model and its validation are described in section 2. The simulated results of tide-driven circulation and distributions of the SBF, SPF and MKF are shown in section 3.1, followed by the mechanisms of the three fronts, including the effects of tide-induced upwelling and tidal mixing in section 3.2. In section 3.3, we attempt a quantitative evaluation of the frontal mechanism from the thermal balance terms. Finally, conclusions are given in section 4.

The ocean model used in this study is the Hybrid Coordinate Ocean Model (HYCOM). In this model, the vertical coordinates are isopycnal in the open and stratified ocean, and in order to adapt to topography in coastal regions, a smooth transition to terrain-following coordinates (sigma-coordinate) is used. In the mixed layer or weakly stratified seas, z-coordinates are used (Bleck, 2002; Chassignet et al., 2003; Chassignet et al., 2007). These features make HYCOM suitable for simulating the YS in summer, including the stratified central YS, the steep topography of the coastal region, and the strong influence of tidal mixing.

A three-level one-way nested system based on HYCOM, comprising an outer, middle and inner model, was set up as shown in Fig. 1b. The outer domain covers most of the Indian and Pacific with a horizontal resolution of approximately (1/3)^° by (1/3)^°. The middle model domain is 30^°S-51^°N, 30^°-180^°E), with a horizontal resolution varying from 22 km near the equator to 13 km near the northern boundaries. The inner model domain is 11^°-42^°N, 105^°-135^°E), covering most of the Chinese marginal seas, the Japanese Sea, the Northern South China Sea, and the Philippine Sea. The model's horizontal resolution is about 5 km, which is sufficient to detect and analyze fronts in our study area.

HYCOM provides subroutines to read bathymetry from the GEBCO (the General Bathymetric Chart of the Oceans) dataset at one arc-minute grid. To obtain higher resolution sea-floor topography, we interpolated the GEBCO dataset onto 30 arc-second grid. The coastline was smoothed slightly in order to reduce computation cost and instability. The outer, middle and inner models all use 28 hybrid layers, with a minimum thickness of 2 m in the top layers. The 28 target densities of isopycnal coordinates are: 0.1, 0.2, 0.3, 0.4, 0.5, 22.0, 22.50, 23.00, 23.5, 24.0, 24.45, 24.90, 25.30, 25.65, 26.00, 26.30, 26.60, 26.84, 27.04, 27.22, 27.38, 27.52, 27.64, 27.74, 27.78, 27.82, 27.84, and 27.86. The top five target densities were purposefully set to small values to ensure that they exist as z-level coordinates at the surface and provide reasonable vertical resolutions in the surface mixed layer (Wan et al., 2008; Xie et al., 2011). In this model, the vertical mixing coefficients K_m (vertical eddy viscosity) and K_h (vertical thermal diffusion) are parameterized by the KPP scheme (Large et al., 1994).

The outer model was initialized from zero velocity with temperature and salinity from the Levitus climatology (Antonov et al., 2006; Locarnini et al., 2006), and spun-up for 10 years with climatology forcing. The simulation used realistic forcing fields from the European Centre for Medium-Range Weather Forecasts 40-year Reanalysis (ERA40) (Uppala et al., 2005) starting from January 1960. Meanwhile, the initial state of the middle model was interpolated from an equilibrium state of the outer model with the same forcing. After a 30-yr spin-up, the inner-model started in January 1991, and was driven by the high frequency atmospheric forcing of ERA-Interim data (Dee et al., 2011). Model results from a 10-yr period (1998-2007) were chosen to study.

A monthly climatology was estimated by applying the ERA40 river run-off data with a correction implemented in the tropics (Troccoli and Kållberg, 2004) to the Total Runoff Integrating Pathways (TRIP) (Oki and Sud, 1998) hydrological model. Tide is included in the inner model, with a barotropic pressure forcing applied at the open boundary from eight constituents (K1, O1, P1, Q1, M2, N2, S2 and K2). The tidal heights are taken from the Finite Element Solution global atlas (FES2004) (Lyard et al., 2006).

Two experiments ("Expt. T" and "Expt. NoT") were designed to explore the role of tidal forcing. Results of the 10-yr climatology mean for August were used in Expt. T. In Expt. NoT, the tidal forcing was excluded, while other physical processes remained unchanged in the model. In this study, the thermal front intensity was quantitatively calculated by the SST gradient in each grid. The gradient magnitude (GM; units: ^°C km^-1) was estimated by the following formula:

2.2.1 Tide verification

Considering the important role that tides play in the YS, we first compared model co-tidal charts with satellite observational results (Fang et al., 2004) (figures not shown). Co-amplitude and co-phase lag contour distributions of the four principal constituents (i.e., M2, S2, K1 and O1) all showed good agreement with observations (Fig. 2). The positions of amphidromic points were also consistent with previous studies. Both the M2 and S2 tides in the YS have two amphidromic points, respectively located northeast of Chengshantou and off Haizhou Bay (Figs. 2a and b). K1 and O1 have one amphidromic point to the east of Subei Bank (Figs. 2c and d). Compared with diurnal tides (K1 and O1), the semidiurnal tidal system (M2 and S2), especially the M2 constituent, is much stronger in the YS. The amplitude of M2 can reach approximately 2 m near the western coast of Korea.

Figure 2.  Numerical simulation co-tidal charts for four tidal constituents: (a) M2 tidal constituent; (b) S2 tidal constituent; (c) K1 tidal constituent; and (d) O1 tidal constituent. Solid and dashed lines show distributions of phase lag in degrees and are referred to in Beijing standard time (UTC + 8) and amplitude (cm), respectively.

2.2.2 Satellite data

Two sets of remotely-sensed satellite SST data were chosen to validate the model results in the YS. In this study, we used 4-km resolution Advanced Very High Resolution Radiometer (AVHRR) Pathfinder Version 5.0 (PFV5.0) data, obtained from the U. S. National Oceanographic Data Center and the Group for High Resolution SST (GHRSST) (http://pathfinder.nodc.noaa.gov) described in (Casey et al., 2010). We chose the "cloud-screen" version, which is a version under stricter quality control and with the cloud effect eliminated. The other dataset used was the Operational Sea Surface Temperature and Sea Ice Analysis (OSTIA) dataset produced by the U. K. Meteorological Office, which uses satellite data together with in-situ observations to determine SST, and is produced daily at a resolution of (1/20)^° (Donlon et al., 2012). We averaged the daily data to a monthly mean dataset for analysis. Climatological means of August SST over the same periods as the model from 1998 to 2007 were used for the validation.

In the YS, both AVHRR and OSTIA data detect four cold SST patches located off Subei Bank, off the eastern tip of Shandong Peninsula (Chengshanjiao), and off the two tips of the Korean Peninsula, marked by ellipses in Figs. 3a and b. The model shows good agreement with the satellite-observed SST, except that simulated SST is 1^°C-2^°C warmer off Subei Bank, and modeled cold SST patches are not as conspicuous as observed. However, the intensity of the modeled cold SST patches off Chengshanjiao, and off Mokpo, is larger than in the satellite observations. Three fronts (SPF, SBF and MKF) correspond to the cold SST areas shown in Fig. 3c.

Figure 3.  Distribution of climatological monthly mean SST (^°C) in August based on satellite data: (a) OSTIA; (b) AVHRR. The surface cold areas are marked by ellipses. (c) Modeled monthly mean SST in August.

To examine the tidal effects on SST, a comparison between Expt. T and Expt. NoT is shown in Fig. 4. As can be seen, SST changes substantially with tidal forcing included (Fig. 4a vs. Fig. 4b). In Expt. T, surface waters off Shandong Peninsula and Mokpo are much cooler than in Expt. NoT; meanwhile, the strength of corresponding fronts are much larger than those of Expt. NoT (Fig. 4c vs. Fig. 4d). Tides also enhance the cyclonic circulation significantly, which basically fringe the YSCWM. However, the absence of tides does not affect the SBF, only with its strength decreased. The cyclonic circulation in the upper layers is mainly a quasi-geostrophic flow along a tidal-induced temperature front, and it is also strengthened by the tide's residual currents. This cyclonic gyre is consistent with previous results (Moon et al., 2009). In Expt. T, the YS coastal currents bring the cold water southward around the eastern tip of Shandong Peninsula. However, southward currents are weaker when tides are switched off, which also demonstrates a tidal influence on the formation of the SPF (Fig. 4). The narrow belt-shaped warm water zones along the western coast of Korea and south of Shandong Peninsula still remain when tidal forcing is switched off. It should be noted that, because near-shore water depth in these areas is very shallow, the water column will be mixed homogeneously and heated by solar radiation. Fronts in these areas are very sensitive to the underlying topography.

Figure 4.  Model results of mean surface circulation superimposed on the (a, b) SST field (^°C) and (c, d) SST gradient (^°C km^-1) for the result of (a, c) Expt. T and (b, d) Expt. NoT in August. Black lines in (c, d) indicate depth contours. The positions of two cross sections across the fronts are marked by the blue lines.

Two cross sections lying across the fronts are selected to exhibit the vertical thermal structures. For convenience, the two sections are named Sec. SP (37^°N) and Sec. SB-MK (34.1^°N). Figure 5 shows the vertical profiles of temperature along these two transects. In Expt. T (Figs. 5a and b), the isotherms are almost horizontally distributed in the fully stratified deep water, and become vertically mixed in the coastal shallow waters. In the slope area between Shandong Peninsula and the central YS, the isotherms are tilted upwards, forming surface cooling. Another conspicuous surface cold area is located off Mokpo (around 126^°E). As for the SPF (Fig. 5b), cold surface water appears on the shallow, mixed side of the front. In Expt. NoT, the tilted isotherms are barely seen from thermal fields in both sections, except for the SBF where weak surface cooling still exists. A possible reason is that the prevailing wind blows from the southeast, almost parallel to the Jiangsu coastline. Correspondingly, the associated Ekman transport is contributing to the upwelling off Subei Bank (Lu et al., 2007).

Figure 5.  Modeled temperature (^°C) of (a, b) Expt. T and (c, d) Expt. NoT along (a, c) Sec. SB-MK and (b, d) Sec. SP. See Fig. 4 for section positions.

In section 3.1, we described how the modeled results indicate that tide plays an important role in the fronts of the YS. (Simpson and Hunter, 1974) proposed a criterion to quantify the location of tidal fronts based on mechanical energy balance, by defining a stratification parameter (Simpson Hunter parameter):

where H is the water depth in meters and U is the amplitude of depth-mean tidal current. If the location of some given value of stratification parameter corresponds well with the fronts, this critical value can be used for frontal detection. Following (Zhao, 1986), we take 1.9 as the threshold value to determine the position of fronts, where the effect of stratification is in balance with tidal mixing. In other words, location of tidal fronts varies with the balance between the potential energy and turbulence energy. In our study, the locations of contour lines log (H/U³) = 1.9 are in good agreement with the frontal positions in the numerical model (shown in Fig. 6).

Figure 6.  Location of tidal front based on Simpson-Hunter (SH) index (solid line represents SH = 1.9) and model results of SST gradient (colored shading; units: ^°C km^-1).

As previously mentioned, upwelling and tidal mixing are two major mechanisms in the formation of fronts in the YS. Therefore, in the next two sections, we explore the intrinsic linkages between tides and fronts from these two aspects.

3.2.1Tide-induced upwelling in the YS

Figure 7 shows the vertical profiles of vertical velocity and u-direction velocity along Sec. SB-MK and Sec. SP. Generally speaking, the vertical velocities in the ocean are in the order of 10^-5 m s^-1, and such small velocities are very hard to measure directly and precisely. Numerical models can provide us with a quantitative view of vertical velocity. Upwelling off Subei Bank and off Mokpo can clearly be seen in Fig. 7a, both of which correspond to the SST frontal region. Over the sloping topography off Subei Bank, upwelling generally appears on the mixed side of the front. On the offshore side of the front, downward motions are also found, with their maximum magnitude located in the bottom layer. Combining with u-direction velocity along Sec. SB-MK (Fig. 7e), a cross-frontal circulation is formed. For the upwelling off Mokpo, its strong intensity extends from the bottom to the surface and has a broad extent from around 125^° to 126^°E.

Figure 7.  Sections of (a-d) vertical velocity (negative values represent upwards velocity; units: m s^-1) and (e-f) u-direction velocity (units: cm s^-1) of Expt. T (left) and Expt. NoT (right) along (a, b, e, f) Sec. SB-MK and (c, d, g, h) Sec. SP. A schematic diagram of secondary circulation is shown as arrows in (e) and (g) for Expt. T.

Unlike at Subei Bank, it is noted that strong downward and upward motions alternately occur on the stratified side of Shandong Peninsula (Fig. 7c), forming an opposite cross-frontal secondary circulation associated with surface convergence and bottom divergence (Fig. 7g). The surface convergence can bring water with different temperature together and plays an important role in the formation of the SPF. When tide is excluded, the upwelling is much weaker than in Expt. T, and secondary circulation is hardly seen from the results of Expt. NoT.

3.2.2 Comparison of the tidal mixing effect

The vertical mixing driven by tides in the ocean circulation model is parameterized by the vertical mixing coefficient. In our model, the K-Profile Parameterization (KPP) was chosen as the vertical mixing algorithm (Large et al., 1994; Large et al., 1997). The KPP model provides mixing from surface to bottom, smoothly matching the large surface boundary layer diffusivity/viscosity profiles to the weak diapycnal diffusivity/viscosity profiles of the interior ocean.

To examine the effects of tidal mixing, we compare the vertical diffusion coefficient of temperature (K_z) between Expt. T and Expt. NoT (Fig. 8). The coefficients of K_z in the surface layer are very large in both experiments due to strong surface solar radiation, and they are barely affected by tides. The most obvious differences exist in the bottom layer of the YSCWM, where tidal mixing is much stronger in Expt. T. With tidal forcing, the mixing effect could also extend along the bottom of sloping topography off Subei Bank and Shandong Peninsula (Figs. 8a and c). Although the directions of cross-frontal circulation are opposite in the two sections, the entrainment and encroachment effect due to tide-induced vertical motion and U-velocity shear (Fig. 7e and g) nevertheless enhances the vertical mixing along the western slope of the YS. The cold water is stirred up from these intense mixing paths, cooling the surface water. Therefore, besides upwelling, tidal mixing also contributes markedly to the formation of the SBF.

Figure 8.  Vertical diffusion coefficients (K_z) for temperature along (a, c) Sec. SB-MK and (b, d) Sec. SP for (a, b) Expt. T and (c, d) Expt. NoT.

Mixed layer depth (MLD) is another important factor in assessing the intensity of vertical mixing. Figures 9a, b, e and f show the MLD and model depth of Expt. T and Expt. NoT. Along Sec. SB-MK, where solar radiation is dominant in the shallow area and contributes to strong vertical mixing, the MLD is almost equal to the water depth in Expt. T. In the absence of tidal forcing, the MLD is somewhat shallower compared with that in Expt. T; in the deep and stratified region, the MLD is small in both cases with and without tidal effects. The same situation also happens along Sec.SP, where a deeper MLD appears in shallow water in the case of tide, and in Expt. NoT the MLD decreases significantly. Therefore, the ratio of MLD to water depth (mixing ratio) can better represent the degree of mixing. Figure 9c demonstrates that frontal positions usually show good agreement with the maxima of the mixing ratio gradient, except for the MKF where intense upwelling occurs. Along Sec. SP, the frontal positions also show quite good consistency with the peak value of the mixing ratio, proving that the SPF also results from the balance of tidal mixing and stratification (Fig. 9g).

Figure 9.  Mixed layer depth (MLD) and mixing ratio (MLD/depth) along (a-d) Sec. SB-MK and (e-h) Sec. SP of Expt. T (left column) and Expt. NoT (right column).

So, the effects of tidal mixing on surface fronts can be summarized as follows. Beneath the stratified water of the interior region of the YSCWM, tidal mixing occurs in the bottom boundary layer and extends upwards by stirring effects until it is blocked by the thermocline. The bottom tidal mixing effect makes the temperature and density of water below the thermocline nearly vertically homogeneous. The upwelling, which is usually the upward branch of cross-frontal circulation, can take cold water upward. So, even if the upwelling is not strong enough to get to the sea surface, when cold water pumped up from the bottom meets a certain type of topography, e.g., the slope off Shandong Peninsula and Subei Bank, the high mixing rate path transfers the cooling to the surface through the effect of vertical diffusion along the bottom of the slope.

To further identify the dynamic mechanism of front formation, in this section we examine the heat budget in the potential temperature equation. Temperature equations of HYCOM are created in the x,y,s,t space, where s represents the generalized vertical coordinate. In this structure, the vertically-integrated HYCOM equation for the conservation of potential temperature within a model layer (Bleck, 2002) is recorded below in Eq. (3):

where (∆ p)_i and θ_i are layer thickness and potential temperature within layer i, respectively; ∇_s represents the horizontal gradient operator under the vertical coordinate of s; represents the time derivative of s; and K _Hθ and K _Vθ are horizontal and vertical diffusion coefficients based on the parameterization of HYCOM (Smolarkiewicz and Grabowski, 1990; Large et al., 1994).

The evolution of potential temperature is affected by horizontal advection (HADV; top line, right side), temperature flux associated with mass flux across model interfaces (ZADV; second line), horizontal diffusion (HDIF; third line), vertical diffusion (ZDIF; fourth line), and net radiation flux (FLUX; bottom line) terms. The time mean values of terms in the surface layer temperature equation along Sec. SB-MK and Sec. SP are shown in Fig. 10. As a steady state, the long-term mean temperature changes (DT/dt) are almost close to zero. Generally, the main contributions to surface temperature in the frontal area are net heat flux (FLUX), vertical diffusion (ZDIF), and horizontal advection (HADV). In Sec. SB-MK of Expt. T (Fig. 10a), negative FLUX occurs within Subei Bank, where outward radiation exceeds the inward amount because of high SST. To the west of the SBF, the ZDIF is positive and is in balance with negative FLUX. The frontal position of Subei Bank happens to be the zone where the direction of net heat flux reverses. To the east of this front, the balance is between negative ZDIF and the sum of positive HADV and FLUX. The peak value of FLUX corresponds to surface cooling where colder water leads to a larger air-sea temperature difference, and thus to larger surface heat flux. Usually accompanied by a peak value of negative ZDIF, positive HADV is a result of strong outflow of cold water and always counterbalances other cooling terms. (Lee and Beardsley, 1999) suggested that strong tidal currents in frontal areas produce high vertical shearing strain of velocity, which enhances the turbulence mixing. This is a possible explanation for the interaction between positive HADV and negative ZDIF. However, the terms that balance DT/dt are different for the MKF. Negative vertical advection ZADV becomes part of the term balances, indicating that upwelling off Mokpo is strong enough to reach the surface, which is consistent with the results in Fig. 7. In contrast, when tide is excluded, ZADV is very small, indicating upwelling almost disappears (Fig. 10b). The patterns of the SBF are similar in both experiments, only with smaller values in Expt. NoT due to much weaker mixing. As for Sec. SP, Figs. 10c and d show the amplitude of ZDIF in Expt. T are larger than Expt. NoT, indicating the cooling effect of the SPF is mainly from vertical diffusion. Meanwhile, the amplitudes of horizontal advection induced by coastal currents increase simultaneously when tides are considered, suggesting that the coastal currents of the YS are much stronger and bring more cold water to the frontal zone of Shandong Peninsula.

Figure 10.  Term balance analysis of the potential temperature equation along (a, c) Sec. SB-MK and (b, d) Sec. SP for (a, b) Expt. T and (c, d) Expt. NoT. Blue solid lines are frontal locations based on corresponding experiments. Units: m s^-2.

The intention of this study was to clarify the different mechanisms related to the tidal fronts of the YS in summer by analyzing results from a three-dimensional numerical model. Tide plays a significant role in the formation and maintenance of three striking fronts; namely, SBF, SPF, and MKF.

A significant feature of the thermal field in the YS in summer is that several surface cold patches are scattered around the YSCWM, which is confirmed by both satellite observations and numerical modeling. Following (Simpson and Hunter, 1974), the parameter of log(H/U³) was calculated to predict the location of tidal fronts, and this value agrees well with the frontal distribution. By comparative experiments with and without tidal forcing, we proved that tide-induced upwelling and tidal mixing are two major mechanisms underpinning frontal formation in the YS. To further investigate the dynamic mechanism of these fronts, we also examined term balances in the potential temperature equation. We found that the upwelling is conspicuous in the frontal region off Subei Bank and Mokpo, while the SPF appears to be less influenced by upwelling. Both fronts off Subei Bank and Shandong Peninsula are characterized by a typical tidal mixing front, which is in the transition zone of deep stratified water and shallow mixed water. Large values of vertical diffusion coefficient also extend along the bottom of sloping topography off Subei Bank and Shandong Peninsula, which proves the importance of vertical mixing on the fronts in these two areas.

In summary, we can conclude that, for the MKF, upwelling is the dominant force, while tidal mixing is the main contributor to the SPF. As for the SBF, the combined effect of tide-induced upwelling and tidal mixing determines the frontal formation.