The study area, as the national shallow sea comprehensive test site of China (Li et al., 2017), is located in the Yellow Sea, north of the coast of Weihai, with Chudao Island at the southern side ( Figure 1 ). Chudao Island is a 0.82 km long and 0.21 km wide bedrock island 3.7 km from the coastline of Shandong peninsula, which is surrounded with rugged coastline and steep bank slope. The average water depth is 30 to 60 m, and presents a generally high-low-high topography from south to north. An east-west directional trench, which is about 2.5 km long and 1 km wide, with maximum water depth 70 m (Zhang et al., 2020a), locates in the north of Chudao Island.

The hydrodynamic in the study area was characterized by the irregular semidiurnal tide and east-west reciprocating flow (Bi et al., 2019). The tide difference has a maximum value of 3.08 m and a mean value of 1.73 m. The northward and eastward wave account for about 58.5% and 37% of the total (Han et al., 2020).

Kelp and scallop raft aquaculture occupy large areas in the northwest of the study area and small areas on the east and west sides of Chudao Island ( Figure 1 ). The breeding cycle of kelp is generally from November to May or June of the following year, which takes up more than half of the whole year. The length of kelp can grow to 4 to 5 m when it is harvested in June and July. The scallop species, Chlamys farreri, which have a shell height of 6 to 7 cm and a shell length of about 7 cm at maturity, is mainly breeding from May to November.

Fieldwork including small spacing sediment sampling and full-coverage bathymetry survey were carried out in November and December 2018, through which the grain size parameter data obtained by laboratory grain-size analysis are used for sediment trend analysis, while the bathymetry data are used to explore the effects of seabed topography on sediment transport trends. A total of 229 surface sediment samples were collected, with grid of 250 m × 500 m ( Figure 1 ), sampling thickness between 20~30 cm, using a clam sampler (type DDC1-2, volume 0.025 m², Beijing Jinyang Wanda Technology Co., Ltd.). The navigation and positioning were executed by Global Navigation Satellite System (H32, plane positioning accuracy ± 8 + 1×10^-6 D, Guangzhou Hi-Target Satellite Navigation Technology Co., Ltd.). The positions were measured twice during sediment sampling process, with one when the vessel arrived at the planned station and the other when the sampler reached the seabed. Repeated sampling was performed if the mass of the sample was less than 300 g. The sea conditions are calm during the sampling period, neither high waves nor strong winds had been recorded in Novermber and the beginning of December of 2018.

The collected sediment samples were brought to the laboratory for grain-size analysis using sieving-densitometer method. For a sediment sample, the part of grain size less than 0.075 mm is used by densitometer method, and the part of grain size larger than 0.075 mm is used by sieving method. The specific steps of the test are referred to Standard for geotechnical testing method (Standard of P. R. China, GB/T 50123-2019, 2019). The mean grain-size, sorting coefficient and skewness were calculated based on the grain size distribution curve as follows (Folk and Ward, 1957):

where M _z, σ _i, and S _ki are the mean grain-size, sorting coefficient and skewness, with φ ₅, φ ₁₆, φ ₅₀, φ ₈₄, φ ₉₅ the grain size corresponding to the cumulative percentage content of 5%, 16%, 50%, 84% and 95%.

Full-coverage bathymetry survey was carried out by a multi-beam echo sounder system (R2 SONIC 2024, R2Sonic Limited Liability Company, USA) with operating frequency 400 kHz, range resolution 1.25 cm and sector angle of 120°~140°. A total of 36 main survey lines with line spacing 100 m and 7 check lines perpendicular to the main survey line were carried out. Global Navigation Satellite System (H32), fiber optic gyrocompass and motion sensor (Octans III, iXSea Ltd., France), sound velocity profiler (MiniSVP, Valeport Ltd., UK) and automatic water gauge (DCX-25PVDF, Druckmesstechnik AG, Switzerland) were applied accordingly to obtain the positioning, attitude, sound velocity profile and sea level data. The CARIS HIPS and SIPS (V10.1, Teledyne Technologies Incorporated, Canada) software were used to process the MBES data to get the water depth and topography of the study area.

McLaren and Bowles (1985) found that there were two cases for the net sediment transport trends, that is, Case 1: finer, better sorted and more negatively skewed; Case2: coarser, better sorted and more positively skewed. GSTA was provided by Gao and Collins (1991) based on McLaren’s model. The first step of GSTA is to define trend vectors for a grid of sampling sites by comparing the grain size parameters of each sample site with its neighbors in any direction. A characteristic distance (D _cr) is specified to identify whether the sites are adjacent. If either a Case 1 or a Case 2 trend is identified between the two sites, a dimensionless trend vector of unit length is defined for two neighboring sites. For each sampling site, there may be more than one unit-length vector.

Then, the trend vectors at each sampling site are obtained by composing the multi-vectors into a single vector:

where n is the number of trend vectors identified for the site, r → ( x , y ) is a trend vector, and R → ( x , y ) is the sum of the trend vectors. Finally, an averaging procedure is applied to remove noise included in the vectors at each site, as:

where R j → ( x , y ) is a summed trend vector obtained on the basis of eq. (4) at a neighboring site, and K is the total number of such sites. The vectors R av → ( x , y ) can be defined as transport vectors to describe the main sediment transport pathways. Since the sampling sites are regular and uniform with a rectangular grid of 250 m × 500 m, the characteristic distance D _cr is set as 250 5 m.

Tidal currents and waves are the main factors controlling sediment transport in coastal waters (George et al., 2018). The sediment grain-size trend is mainly formed under the long-term reciprocating action of periodic tidal currents rather than the random changeable waves (Armstrong et al., 2022). Therefore, this study focuses on the simulation of tidal currents.

The Coastal and Regional Ocean Community model (CROCO) is a three-dimensional nonlinear baroclinic primitive equation model based on the Regional Ocean Modeling System (ROMS) (Huang et al., 2021), which can simulate multiple-scale sea water movement. In the CROCO model, a time-splitting method is used to generate two different time steps, in which the short time steps is used to advance the surface elevation and barotropic momentum, while a large time step is used for tracers and baroclinic momentum. The curvilinear orthogonal coordinate system and Arakawa C staggered grid are adopted in the horizontal direction to increase the resolution of the specific area, while the S coordinate system is used in the vertical direction to improve the resolution at the thermocline and the bottom boundary layer.

The horizontal resolution of the model is 1/1200°, while the vertical water column is meshed into 6 levels. The shoreline with resolution 1′ is from the Global Self-consistent Hierarchical High-resolution Shorelines (GSHHS) provided by the National Geophysical Data Center (NGDC) of U.S. The topography with resolution 1′ is from the Gridded Global Relief Data Collection (ETOPO1) with resolution 1′ provided by NGDC. The initial temperature and salinity data are from the World Ocean Atlas 2009 (WOA09) published by NGDC. The open boundary is forced by ten tidal constituents including M₂, S₂, N₂, K₂, K₁, O₁, P₁, Q₁, M_f, and M_m, in which the harmonic constants are obtained from the Oregon State University Tidal Inversion Software (OTIS). The short time step and the large time step are set as 2 and 120 s, respectively. The model is operated normally after it is integrated forward to a stable state, with the duration covering the whole year of 2018 that consistent with time of the sediment sampling.

As shown in Figure 2 , the water depth of the study area is mainly between 25.0 m and 60.0 m, with an average of 36.7 m and a maximum of 70 m. The seafloor is rugged in the north of the island, while the water depth drastically increases to more than 60.0 m at an east-west trench north of the island. The trench has dense isobaths and steep slope around it, among which the slope is steeper in the north-south direction with a maximum 5.2° than that in the east-west direction. The water depth gradually becomes shallow north of the trench, forming the flattest area within the study area.

The continuous distribution of M _z σ _i, S _ki, which are obtained by interpolating the parameters of discrete point with a kriging method (Oliver and Webster, 1990). The mean grain-size M _z, which is mainly affected by material supply, transport capacity, and the depositional environment, can directly reflect the size of the sediment particle and the strength of hydrodynamics of the environment (Visher, 1969). The sorting coefficient σ _i, which represents the uniformity of sediment, is also an indicator of hydrodynamics and sediment transport trend. The skewness S _ki describes the degree of symmetry of sediment grain size distribution. As shown in Figure 3 , the surficial sediments are mainly muddy silt, with mean grain-size 4.9~6.2 φ, average value 5.03 φ and maximum value 0.79 φ ( Figure 3A ). The sorting coefficient σ _i ranges from 0.014 to 2.206 with an average value of 0.077, most of which are excellent ( Figure 3B ). The skewness S _ki shows a decrease pattern from south to the north, from island to sea, with all positive skew ranging from 0.07 to 0.97 ( Figure 3C ). Generally, the three grain size parameters show an obvious feature of zonal distribution. The area with larger M _z (<2 φ), poor sorting, and strong positive S _ki concentrates around the bedrock island, where rocks and gravels are common seen from the samples. While to the north, M _z gradually transitions to a broad northern area with φ=5, σ _i becomes better away from the island, and S _ki decreases to minimum.

In Figure 4 , the vector direction represents the net sediment transport direction, while the vector length represents the significance of the grain-size trend.

Despite the southward transport trend of small areas in the northwest and northeast, the overall sediment transport trend of the study area is from island shore to sea and from south to north. In addition, there are four convergence centers in the study area and show different transport patterns. The transport trend is more significant on the southern side than that on the northern side of A and B ( Figure 4 ), indicating a clear shore-to-sea transport trend at the east and west of the Chudao Island. A counterclockwise sediment transport trend at C is discovered although C is located at the edge of the study area and the edge effect caused by the lack of information may affect the accuracy of the results. An inconspicuous sediment convergence center D locates at the west of the Chudao Island, where eastward and westward vectors meet.

Since the erosion, transport and deposition of sediments are most closely related to the near-bottom flow velocity, the flow velocity of the bottommost layer (the sixth layer) in the model is particularly exported from the CROCO model to characterize the hydrodynamic factors. Figure 5 shows the comparison of tidal level simulated by numerical model and the value observed by the Chudao tide station (with location illustrated in Figure 1 ), in which the time is from November 11 to December 1, 2018 including an astronomical tide. As can be seen that not only the calculated tide level variation trend but also its values are in good agreement with the observed ones. The correlation coefficient (R ²) of the simulated and observed tide level is 98.42% ( Figure 6 ), which proves the reliability of the presented model.

The overall sediment transport trend in the study area, that is, from the island shore to the sea ( Figure 4 ), is generally consistent with the direction of annual-averaged near-bottom flow ( Figure 7 ). The disturbance of seabed caused by near-bottom water flow is the direct driving force of sediment transport (Gardner et al., 2004; Du et al., 2021), so the direction and velocity of near-bottom flow correspond well with the direction and the significance of sediment transport trends (Gunsolus and Binns, 2018). The locations of the two bottom flow loops at the east and west of the study area ( Figure 7 ) corresponds well with the convergence centers of grain-size trend A and B ( Figure 4 ), indicating that the flow direction controls the direction of sediment transport. Moreover, the average bottom velocity (0.149 m/s) in the west of the study area is greater than that (0.136 m/s) in the east ( Figure 7 ). Correspondingly, the sediment transport trend in the west of the study area is also more significant than that in the east ( Figure 4 ), indicating the controlling effect of flow velocity on the potential quantity of sediment transport. However, it should also be noted that the sediment convergence centers A and B deviate slightly from the positions of the two loops. The convergence center C is located at the junction of the east-west loop, with northwestward flow on its west side and northeastward flow on its east side, which is not completely consistent with the sediment transport directions of west-northwestward on the west side and southwestward on the northeast side. What is more, the bottom flow loop on the east side of Chudao Island does not form a sediment convergence center. This demonstrates that the sediment transport trend may also be influenced by other factors besides the near-bottom flow field.

The hydrodynamic conditions are weak as the average bottom flow velocity of the entire study area is only about 0.1 m/s ( Figure 7 ). The fine-grained sediments are brought from nearshore to offshore areas by the weak bottom currents, most of which are settled in the trench valley, while the coarse-grained sediments are left nearshore. Therefore, the sediment in the southern part of the study area is coarse-grained, while the sediment located in the trench north of Chudao Island is finer-grained. The coarser sediment size in the north of the trench is because the weak bottom flow cannot drive the nearshore fine-grained sediment over the steep slope. This is consistent with the mean grain-size of sediment from south to north showing a “coarse-fine-coarse” striped distribution presented in Figure 3A .

To summarize, the near-bottom flow field is the primary factor that controlling the sediment transport trend in coastal bedrock islands sea area. The greater the bottom flow velocity is, the more obvious the sediment transport trend is. The direction of bottom flow decides the general patterns of sediment transport although small deviations may exist in specific areas. Besides, sediment transport impacted by the near-bottom flow field shapes the sediment distribution characteristics in coastal bedrock islands sea area.

The submarine topography is varied in the study area with steep slope around the Chudao island, then deep trench connected with flat plain from south to north ( Figure 2 ). The near-bottom current basically points in the direction from shallow to deep waters. The general sediment transport trend is from the southern island shore and the northern flat area to the trench, which is consistent with the topography from shallow to deep. In fact, bottom flow ( Figure 7 ), which indicates the sediment transport direction, is also influenced by the topography, because the local flow field is usually constrained by submarine topography and tends to flow from shallow to deep water areas. The existing distribution characteristics of sediment grain size parameters are the result of long-term sediment transport. The mean grain-size, sorting coefficient and skewness ( Figure 3 ) in the study area are zonal distributed, roughly parallel to the isobaths ( Figure 2 ), which also indicates the controlling effect of submarine topography on the long-term sediment transport trend. In addition, the sediment convergence center A ( Figure 4 ) is not coincided with the center of the bottom flow loop, however, it locates northwest of the loop center where densest isobaths and most dramatic topography changes exist. This indicates that rapid decline of seafloor topography can cause certain amount of sediment to be transported down the slope. Actually, previous researches have revealed that the surficial sediment would slowly move down along the seabed slope due to its own gravity (Eschard, 2001; Hsiung and Yu, 2013), the steeper the slope is, the faster and the larger volume of sediments would transport on a long-term scale. It can also be seen that ( Figure 4 ), unlike the general trend of sediment transport from shore to sea, the sediment around Chudao Island is transported in all directions with the island as the center. This is because the water depth around the island decreases drastically, which leads to the sediment around the island transports in the direction of downward slope. Therefore, the main difference in sediment transport trend between coastal bedrock island sea area and traditional bedrock coast is that the presence of the island in makes itself a discrete center of sediment transport, and the sediment around the island transports seaward.

Under the influence of submarine topography, sediment in the coastal bedrock island sea areas mainly transports from shore to sea, making it difficult to silt up near shore. Due to the presence of islands, tides and waves are strongly dissipated before reaching the shoreline (Ogawa et al., 2011). In addition, the bedrock coast is resistant to erosion and has stable geological conditions. All of these conditions meet the requirements of harbor site selection, so the coastal bedrock island sea area is suitable for the construction of harbors, docks and other coastal engineering.

Thus, in the coastal bedrock island sea areas, the topography can either directly transport sediments down its slope or indirectly affect the sediment transport by constraining the near-bottom flow, making the sediment generally transports from shallow to deep waters. In the coastal bedrock island sea area, the sediment around the island is transported seaward with the island as the center, which is the main difference from the sediment transport trend of the traditional bedrock coasts.

Near the considerable aquaculture area ( Figure 1 ), the mean grain-size ( Figure 3A ) of the sediment increase, and the near-bottom flow velocity ( Figure 7 ), indicating that human activities such as aquaculture will also affect the sediment transport trend in coastal bedrock island sea area besides natural factors (Wood and Widdows, 2002). Aquaculture facilities including net cages and rafts is widely deployed in China coasts, particularly in bedrock island sea areas (Yu et al., 2016; Chen et al., 2018) such as our study area. However, these aquaculture facilities have certain impact on sediment transport in shallow waters even to water depth 200 m (Pawar et al., 2001; Srithongouthai and Tada, 2017; Chaves et al., 2020).

The mean grain-size becomes coarse near the aquaculture area rather than the trench ( Figure 3A ), which may be due to the increased sedimentation rate of organic matter induced by aquaculture-produced residual bait and feces. For example, to our common knowledge, the unidirectional eastward near-bottom flow cannot form sediment convergence center at D, however, the sediment convergence center D does exist. This is probably due to the sediment at D is fine, but the sediment east of D is coarsened by the aquaculture and west of D is also coarse, which presents the transport trend from the east and west to the convergence center. In addition, the diffusion of scallop fragments caused by water flow also contributes to the sediment coarsening near the aquaculture area, the time for sediment sampling was from November to December 2018, just when the kelp and scallops in the aquaculture area were ripe and large in size. Similarly, sediment coarsening induced by the dispersal of mussel fragments were also found in Marlborough Bay, New Zealand (Hartstein, 2005), while sedimentation rates were also increased by the mussel and kelp aquaculture activities (Walker and Grant, 2009; Liu et al., 2016). The aggregation of macroalgae and the artificial aquaculture facilities (Gaylord et al., 2007) can change the local hydrodynamics especially reduce the flow velocity to a certain extent (Rosman et al., 2007). For example, the averaged near-bottom flow on the east side of the sediment convergence center C ( Figure 7 ) is only 0.05 m/s, which is 62% lower than that of 0.13 m/s in other surrounding areas. Similarly, the flow velocity at Sungo Bay and Heini Bay off Shandong Peninsula, China, also as bedrock island sea areas, have been observed to be reduced by 54% (Grant and Bacher, 2001) and 21% (Zhang et al., 2020b) as well due to the influence of shellfish and kelp aquaculture.

Reduced flow velocity and weakened sediment transport can exacerbate the enrichment of pollutants and eutrophication of water in aquaculture areas (Chor et al., 2022), resulting in the death of farmed scallops and fish and the frequent occurrence of algal blooms (Cui et al., 2018), harming the economic and environmental benefits. The impact of aquaculture on flow field and sediment transport can be mitigated by reducing stocking densities (Tang et al., 2017; Gretchen, 2018) and netting weights (Liu et al., 2008), allowing for the dilution of enriched pollutants and nutrient salts.

Thus, aquaculture may affect the material source, the deposition and transportation rate of sediments, and thus make influence on the sediment transport of coastal bedrock island sea area. First, the aquaculture waste and the accelerating on sedimentation rate can increase the mean grain-size of sediment. Second, the artificial aquaculture facilities would reduce the movement of seawater, slow down the flow velocity, and thus hinder the initiation, suspension of sediments, resulting in a sediment coarsening and a significant reduction of sediment transport. The accelerated sediment deposition and weakened transport can cause pollutant and nutrient enrichment, which can be mitigated by rational planning of aquaculture density and configuration of aquaculture facilities. This will enable the sustainable development of aquaculture industry while protecting the ecological environment of the coastal bedrock island sea areas.