The modern Yellow River Delta is a river-dominated system that started forming in 1855, at that time the Yellow River entered the Bohai Sea (Figure 1). In the past 170 years, a fan-shaped delta has taken shape owing to frequent channel avulsion. The current active Yellow River Delta lobe started to develop in 1976, when the previous active channel toward the north was abandoned (Wang et al., 2017). The river channel avulsed again in 1996, shifting north (Figure 1).

The Yellow River Delta is dominated by micro-tidal tides with a mean range of 0.73–1.77 m (Yang et al., 2011a). Spatially, the northern part of the present delta lobe has a diurnal tidal regime, whereas the southern region has a mixed semi-diurnal tidal regime (Ji et al., 2020). Field observations were conducted on the northern Qingshuigou tidal flat (QSG) and southern Kendong-dyke tidal flat (KDD) (Figure 1). The elevation was approximately −0.2 m above mean sea level for both (Figures 1C, D). The tidal flat profiles were convex-up at the QSG site and concave-up at the KDD site (Figure 1). The surface sediments were composed of silt and sand with median diameters of ~60 μm and ~66 μm at the QSG and KDD, respectively (Figure 2). The two sites were submerged during high tides, with a maximum water depth of 0.9 m, but exposed during low tides. Both tidal flats were bare and without vegetation (Figure 3A).

Hydrodynamic measurements were performed from 10:00 on November 19 to 10:00 on November 23, 2019. The fluvial sediment flux from the Yellow River, monitored at the Lijin Hydrometric Station, was low during this period (Figure 4). Thus, the direct influence of riverine sediment on tidal flat evolution in the study area may be negligible. Current velocity, suspended sediment concentration (SSC), wave activity, and bed-level changes were collected, using two tripod systems mounted on the bed of the two sites (Figure 3A). Both tripods were equipped with an acoustic doppler velocimeter (ADV) (Nortek AS, Norway), an optical backscattering sensor (OBS) (Campbell Scientific., USA), and a tide and wave logger (RBR) (RBR Limited, Canada). The ADVs, OBSs, and RBRs were placed 30 cm, 10 cm, and 10 cm above the bed surface, respectively. The ADVs were set up in 64 Hz mode, with a burst interval of 5 min recorded both the near-bed 3D velocity and bed-level change (Andersen et al., 2007; Zhu et al., 2016; Xie et al., 2018a). RBRs were deployed to measure the wave activity and water depth. The optical backscatter signal of the OBSs was used to estimate the SSC, with sampling every 5 min. The calibration of the optical backscatter signal into the SSC is shown in Figure 3B.

Field estimations of the current-induced and wave-induced bed shear stresses are made using ADV and RBR. These estimates are useful for understanding sediment dynamics and sedimentary processes in situ using tidal currents and waves.

The measured current velocities (u, v, w) can be decomposed into terms of the mean (U, V, W) and fluctuating components (u’, v’, w’). Several studies have estimated the bed shear stress from the second components, including the prevalent turbulent kinetic energy (TKE) method, where the TKE (J/m³) is estimated from the fluctuating components:

Where ρ w is the water density at 20°C (ρ = 1030 kg/m³ in this study) (Andersen et al., 2007; Xie et al., 2018b),

The current-induced bed shear stress (τ_c, N/m²) is estimated using:

and C is a constant (C = 0.19) (Stapleton and Huntley, 1995; Pope et al., 2006).

Furthermore, the peak wave orbital velocity ( U ^ δ ) m/s) is related to the wave-induced bed shear stress (τ_w, N/m²) by the wave friction coefficient f_w (van Rijn, 1993):

where the peak wave orbital velocity ( U ^ δ ) and peak orbital excursion ( A ^ δ )can be calculated using linear wave theory (Stokes, 1847):

where H is the wave height (m), T is the wave period (s), h is the water depth (m), L is the wavelength (m) (L = (gT²/2π)tanh(kh)), k is the wave number (k = 2π/L), ω is the angular velocity (rad/s), and g is the gravitational acceleration (g = 9.8 m/s²).

The wave friction coefficient (f_w ) is determined by the hydraulic regime (Soulsby, 1997):

Where R e w   is the wave Reynolds number ( R e w =   U ^ δ A ^ δ ν ), r is the relative roughness ( r =   A ^ δ k s ), k_s is the Nikuradse roughness ( k s = 2.5 d 50 ), d₅₀ is the median diameter of the surficial sediment, and ν is the kinematic viscosity of water (ν = 1.5×10⁻⁶ m²/s).

Then, the combined wave-current bed shear stress (τ_cw, N/m²) can be estimated using the following hydrodynamic formulation (Soulsby, 1997):

Where τ c   is the current-generated bed shear stress, and τ w is the wave-induced bed shear stress.

The root-mean-square value τ rms is a good average measure of the bed shear stress, particularly useful in random waves. So τ rms is used here to describe the bed shear stress and it can be estimated using the following hydrodynamic formulation (Soulsby and Clarke, 2005):

The Yellow River Delta is in a micro-tidal environment, and the tidal range was less than 1.0 m at both the QSG and KDD sites (Figures 4B and C). The inundation period at the QSG site was much longer than that at the KDD site. The QSG tidal flat was always submerged during the tidal cycle, whereas the KDD tidal flat was submerged for less than 7 h per day (Figures 5A, 6A).

Wave activities at the KDD site were stronger than those at the QSG site during the observation period (Figures 4B, C). The averaged significant wave height was 0.10 m at KDD while it was only 0.03 m at QSG. At the KDD site, the wave heights were higher when the tidal flats were submerged; however, this was not the case at the QSG site (Figures 4B, C). These differences were ascribed to different coastal landscapes: the QSG site was located in a sheltered bay, whereas the KDD site was open to the sea (Figure 1).

The ADV data indicated that the velocity in the vertical direction was always negligible compared to that in the horizontal direction. Therefore, we assumed that the current was two-dimensional. The current on the tidal flat was more complex than that of a simple bidirectional flow. For simplicity, the velocity was projected along two directions: the eastern direction (positive for the east) and northern direction (positive for the north). At the QSG site, which is parallel with the coast, the velocity component along the eastern direction was relatively small (< 0.05 m/s) (Figure 5B). During the flood period, the direction of the current was between 170°N and 210°N. During the ebb, it was between 330°N and 10°N. At the KDD site, the current direction during the flood and ebb periods was between 310°N and 10°N, and 100°N and 150°N, respectively (Figure 6B). The current direction was symmetric with respect to the highest water level at the KDD site (Figure 6B), whereas it showed no clear correlation with the tidal oscillation at the QSG site (Figure 5B).

For each tidal cycle, the current velocity, at both the QSG and KDD sites, was higher during the flood than during the ebb (Table 1). The average velocity at the QSG site was higher than at the KDD site. At the QSG site, the average velocity was 0.09 m/s for the tidal cycle. The velocity varied between 0.07 m/s and 0.16 m/s during the flood period, and between 0.08 m/s and 0.10 m/s during the ebb period. The maximum flood velocity occurred when the water depth was ~0.30 m, and the maximum ebb velocity occurred at a higher elevation of ~0.40 m (Figure 5B). The velocity decreased to zero in the slack water. At the KDD site, the velocity was relatively low (<0.07 m/s) for most of the tidal period. During the flood, the velocity ranged from 0.03–0.07 m/s. During the ebb, it was between 0.02 m/s and 0.03 m/s. The maximum flood velocity appeared at the beginning of the tidal cycle and then decreased to zero during high slack water (Figure 6B). The maximum ebb velocity was reached when the water depth was less than 0.20 m (Figure 6B).

There was strong fluctuation in the water level and velocity at the QSG site, indicating the currents flowed forward and back frequently at a short timescale (hours). By contrast, the currents were bidirectional flows at the KDD site.

The bed shear stress was calculated under the combined effect of waves (τ_c) and currents (τ_w) for both sites (Figures 5C, 6C). On average, the bed shear stress (τ_rms) at the QSG site (0.27 N/m²) was much smaller than that at the KDD site (0.43 N/m²) (Table 1). The wave-induced bed shear stress (τ_w) had a stronger influence on the KDD site than the sheltered QSG site. At the QSG site, the τ_c ranged from 0.01–0.61 N/m² with an average value of 0.10 N/m² and τ_w ranged from 0.03–0.85 N/m² with an average value of 0.20 N/m². The τ_rms showed weak correlation with tidal oscillations. At the KDD site, the average τ_w was high, reaching 0.64 N/m². The τ_c ranged from 0.01–0.30 N/m², with an average value of 0.07 N/m². In general, the τ_rms of the KDD site was higher during the flood period than during the ebb period. Peak τ_rms occurred at the beginning of the flood as well as when the ebb water depth was less than 0.20 m, which showed a clear correlation with the current velocity (Figure 6C).

The SSC at QSG was almost the same as that at KDD. The SSC ranged from 0.05–0.53 g/L with an average value of 0.21 g/L at the QSG (Figure 5A), and at the KDD it was from 0.13–0.63 g/L with an average value of 0.24 g/L (Figure 5B). However, for each tidal cycle, the SSC trend differed remarkably between the two sites. At the QSG site, the SSC was higher than 0.20 g/L for most of the study period. The peak SSCs were in phase with high water levels (>0.30 m) (Figure 5A). During Tide 3 and Tide 4, high SSC values occurred at the beginning of the flood periods. The mean SSC during the flood and the ebb tides were 0.21 g/L and 0.22 g/L, respectively (Table 1). At the KDD site, the SSC was also always higher than 0.20 g/L, but inundation was much shorter than that of the QSG site (Figure 6A). The mean SSC during the flood period and the ebb period were both 0.24 g/L (Table 1). The SSC had two peaks within a tidal cycle: before the high slack water in the flood, and near the end of the ebb. The peak SSC corresponded to high velocities, possibly owing to local erosion and sediment resuspension.

At the QSG site, the SSC showed no clear correlation with the bed shear stress (Figure 7A). The high SSC appeared at the beginning of flood periods or during high slack water when the bed shear stress was relatively low. The SSC at the KDD site followed the same trend as the bed shear stress (Figure 7B). The peak SSC clearly corresponded to high bed shear stress but with a short delay in time due to the horizontal advection.

The sediment flux was calculated multiplying the velocity by the water depth and the SSC. As the velocity was projected along the eastern and northern directions, the sediment flux was also computed in these two directions (Figures 5D, 6D). The sediment flux at the QSG site was greater than at of the KDD site.

At the QSG site, the sediment flux in the eastern direction was smaller than that in the northern direction (Figure 5D). During floods, the tidal mean sediment flux was westward in the range of 0.54–9.00 kg/m with an average value of 3.44 kg/m, and the sediment flux in the other direction was southward and greater, with an average value of 106.58 kg/m. During ebbs, the eastern sediment flux was −0.97 kg/m on average, and the northern sediment flux was 36.13 kg/m (Table 2).

At the KDD site, the sediment flux was correlated with the velocity (Figure 6D). During the flood period, the tidal mean sediment fluxes in the eastern and northern directions were -3.98 kg/m and 27.33 kg/m, respectively. During the ebb periods, the average eastern sediment flux was 16.48 kg/m, and the northern sediment flux was −6.25 kg/m (Table 2).

At the QSG site, the average net sediment flux per tidal cycle in the eastern direction was −4.23 kg/m, and −65.39 kg/m in the northern direction (Table 2), indicating that a large amount of sediment was delivered to the QSG site (Figure 8A). At the KDD site, the net sediment flux per tidal cycle in the eastern and northern directions were 13.19 kg/m and 20.77 kg/m, respectively, which indicated a moderate amount of sediment was removed from the KDD site.

The bed level change at the QSG and KDD sites displayed a clear correlation with the sediment flux. At the QSG site, the bed elevation increased by 8 mm, suggesting net deposition over the tidal cycles (Figure 5E). However, at the KDD site, the bed level decreased by 10 mm, suggesting bed erosion (Figure 6E).

The two sites in the Yellow River Delta exhibited very different types of behavior regarding the hydrodynamics and sedimentary processes. The northern QSG tidal flat was deposited, whereas the southern KDD site eroded during the observation (Figures 5E, 6E). The two sites had similar sediment compositions (Figure 2), suggesting that the sediment source of each site was likely to be the same, both from the adjacent estuary and riverine inputs (Herrling and Winter, 2018; Pearson et al., 2020). It could be concluded that tidal and wave conditions were responsible for this difference. The northern QSG site was in a diurnal tidal environment, whereas the southern KDD site was in a mixed semi-diurnal tidal environment (Figure 9; Ji et al., 2020). As a result, the inundation period of the QSG site was much longer than that of the KDD site (Figure 4). Longer inundation times were associated with greater water depths, which promoted sedimentation on the tidal flats (Yang et al., 2008; Davis et al., 2017). Moreover, more powerful waves at the KDD site under a shallow water environment resulted in comparatively more erosion than at the QSG site (Shi et al., 2019).

Notably, at the QSG site, a high SSC occurred at the beginning of the flood period and at high slack water (Figure 5), which indicated that a large amount of sediment was transported from the adjacent sea to the tidal flat (Shi et al., 2012; Nowacki and Ganju, 2021). In contrast, the KDD site showed large temporal variations in SSC, which correlated highly with the large bed shear stress at the KDD site (Figure 6). A high-SSC environment and long slack water period provided favorable conditions for sediment settling at the QSG site, whereas a large bed shear stress led to sediment resuspension at the KDD site. Previous studies indicated that longshore currents flowed from north to south during flood tides while from south to north during ebb tides in the Yellow River Delta (Bi et al., 2010; Fan et al., 2020). The direction of the residual water mass and sediment transport was northeast at KDD and was southwest at QSG during fair weather conditions (Fan et al., 2020; Ji et al., 2022), which agreed with the results of the net sediment fluxes at the two sites in this study.

In this study, we proposed an integrated conceptual model to describe the sedimentary processes at the two sites (Figure 10). At the QSG site with diurnal tides, the sediment from the adjacent estuarine water was delivered to the tidal flat; additionally, it was resuspended due to high bed shear stress, leading to rich sediment supply and a high SSC environment. Combined with a relatively high-water level, most of the sediment settled and deposited (Figure 10A). At the KDD site, with mixed semi-diurnal tides, relatively strong wave activity, combined with shallow water, resulted in high bed shear stress and sediment erosion from the bed. Unlike the QSG site, which was always submerged, the KDD site was predominantly eroded during a short period of inundation when the bed shear stress was high. The eroded sediment was transported to the subtidal area and the adjacent estuarine water, leading to erosion at the KDD site (Figure 10B).

Previous studies have reported that estuarine tidal flat evolution was jointly influenced by the fluvial sediment supply, tidal currents, waves, and vegetation (Temmerman et al., 2003; Mariotti and Fagherazzi, 2013; Xie et al., 2021). The fluvial sediment supply affected tidal flat evolution by controlling sediment availability (Yang et al., 2011b; Nittrouer et al., 2021; Ji et al., 2022). A larger riverine sediment input enhanced sediment availability, leading to delta progradation and associated tidal flat accretion (Blum and Roberts, 2012; Bi et al., 2021). In this study, the fluvial sediment supply was not a determinant factor for the asymmetric evolution in the Yellow River Delta. The river channel of the delta was along the northwest-southeast direction before 1996 and was closer to the KDD site (Figure 1). After 1996, the river shifted to the present channel position, which was in the northeast–southwest direction and closer to the QSG site. If the fluvial sediment supply was the dominant factor, the KDD site would have accumulated and deposited sediment between 1976 and 1996. However, coastline changes indicated that the southern tidal flat of the delta was eroded and retreated between 1976 and 1996 (Figure 8). These results suggested that the fluvial sediment supply was not directly responsible for the contrasting evolution of the Yellow River Delta. Previous studies have reported that local sediment availability had a direct positive impact on tidal flat evolution (Kirwan and Megonigal, 2013; Schuerch et al., 2014; Maan et al., 2019). Besides riverine sediment supply, local sediment availability for a tidal flat was also influenced by local hydrodynamics, plant growth, and human modification (Friedrichs, 2011; Roman, 2017; Xie et al., 2018b). In this study, we chose to perform the measurements in November, when the fluvial sediment flux from the Yellow River was low. In this study, we chose to perform the measurements in November, when the fluvial sediment flux from the Yellow River was low. During the study period, the different tidal conditions in the Yellow River Delta, were the principal factor in controlling the local sediment availability. The diurnal-tidal environment in the northern part of the delta enhanced sediment import from the adjacent estuarine water and increased the inundation time of tidal flats (Figure 10A). The tidal conditions provided favorable environment for sediment settling, which helped capture sediment on tidal flats (van Proosdij et al., 2006; Xie et al., 2018b). The mixed semi-diurnal tidal environment in the southern part of the delta led to a shorter inundation period compared with the northern part, thus tidal flats were unlikely to trap sediment (Figure 10B). The difference in tidal conditions might explain why the asymmetric evolution in the Yellow River Delta did not respond to the river channel shift.

The second influential factor in tidal flat evolution was wave activity. The bed shear stress (τ_rms) was important to sedimentary processes of tidal flats. Wave-driven sediment resuspension and transport, interacted with sediment transport driven by currents, always controlled sediment erosion and deposition on tidal flats (van Rijn, 1993; Green and Coco, 2014). The KDD tidal flat, which was open to the sea, was more likely to experience stronger wave energy than the sheltered QSG tidal flat. Moreover, waves played a vital role in river delta progradation and channel avulsion (Nienhuis et al., 2016; Gao et al., 2020). In some cases, waves controlled channel orientation and, therefore, asymmetric growth by wave-driven bypassing sediment flux (Nienhuis et al., 2016).

The third significant factor was vegetation. Vegetation, which attenuated wave energy and enhanced sediment trapping, also played an important role in the tidal flat evolution (Yang et al., 2008). The vegetation at the QSG site was Spartina alterniflora which had been proven to be effective in trapping sediment on tidal flats (Yuan et al., 2011; Angelini et al., 2018), whereas the KDD site was almost bare without vegetation (Figure 11). Different vegetation condition was also the impact factor in determining the contrasting evolution of the Yellow River Delta.

Coastal regions are highly sensitive to change, caused by natural variations as well as human modifications. This includes the increasing changes in climate, specifically rising sea-levels and extreme weather events. Regional water and sediment management are urgently required to minimize the effects of such risks (Kirwan and Megonigal, 2013; Syvitski et al., 2022). Tidal flats are important components of river deltas and usually act as sediment sources for both salt marshes and estuarine waters (Friedrichs, 2011; Donatelli et al., 2018). For effective coastal protection and management, a better understanding of sedimentary processes on tidal flats, and their determinants, is required (Schuerch et al., 2014). The determinants of these tidal flat sedimentary processes vary in different estuarine environments. For example, storm-associated waves and storm surges dominated tidal flat sedimentation in an open-coast tidal flat in the south-western Korea (Yang et al., 2005). However, for the Kapellebank tidal flat in the Netherlands, storm events were infrequent and therefore did not affect the long-term morphodynamics of the tidal flat; while small wind waves had a significant stabilizing effect on tidal flat evolution (Maan et al., 2018). Field measurements also confirmed the ability of the tidal flat system to quickly recover from storms (Xie et al., 2017). Flow asymmetry and mud properties were also reported to influence tidal flat sedimentation (van Maren and Winterwerp, 2013). Moreover, local sediment availability was considered to have strong impacts on the intertidal environment in the Delaware bay and the Yangtze Estuary (Stammermann and Piasecki, 2012; Xie et al., 2018b).

The sediment composition and mud properties were almost identical at the two sites studied. The influence of the fluvial sediment supply could be omitted during the study period. The results of this study suggested that tidal conditions, waves, and local sediment availability were the main determinants of the asymmetric evolution in the Yellow River Delta. The different tidal condition was the most important impact factor, because it formed long inundation period and reduced the wave effect at the QSG site. The tidal environment made the QSG site quite suitable for sediment settling and deposition. Richer local sediment availability in the northern part might also have a positive effect on the depositional process at the QSG site compared to that at the KDD site. Owing to the large decrease in fluvial sediment supply, tidal flats in the Yellow River Delta experienced severe erosion and retreat (Wang et al., 2017). Moreover, it should be noted that the above sedimentary patterns of the two contrasting tidal flats were concluded under fair weather conditions, which appeared different during storms. Storms greatly affected longshore currents, changed the direction and enhanced the magnitude of residual water mass and sediment flux in the Yellow River Delta. For example, the residual currents were transported toward the northeast at KDD under fair weather conditions. Affected by the northerly storm, the residual transport of water flowed southward (Fan et al., 2020). Moreover, the trends of sediment transport under the combined influence of river-tide-wave dynamics, also differed from those under river-tide dynamics (Ji et al., 2022), indicating that storms could sharply alter local hydrodynamics, sediment availability, and sedimentary processes of tidal flats within a short period of time.

Considering the different tidal conditions in the northern and southern parts of the Yellow River Delta, solutions to protect estuarine tidal flats might be different. In the northern part, if sediment availability could be enhanced, additional sediment could be transported to tidal flats and settle easily due to the diurnal tide environment. For the southern part, the inundation period was quite short, and the wave effect was relatively high due to the shallow water. Vegetation restoration, which could attenuate waves and increase the capability to trap more sediment, would be a good solution to avoid continuing erosion in this area. The southern part of the Yellow River Delta has already implemented a similar solution with the establishment of Suaeda salsa to protect tidal flats (Figure 11).