Remote sensing data is obtained from the Geospatial Data Cloud Platform (http://www.gscloud.cn/), with Landsat TM/OLI_TIRS satellite images are selected as the data source, including images obtained from Landsat 5 and Landsat 8 land satellites. When selecting images, the cloud coverage is ensured to be no more than 10%. Images from the same year are selected with a time difference of no more than 1 month ( Table 1 ). ENVI is a remote sensing image processing software developed by using the interactive data language (IDL). It allows for comprehensive and efficient extraction of relevant information from remote sensing images and provides a complete set of tools for processing and analyzing remote sensing data (Khosravi et al., 2020; Song, 2021). The remote sensing images are processed by using ENVI for tasks such as cropping, radiometric correction, and atmospheric correction. The Normalized Water Index (NDWI) (McFeeters, 1996) can highlight water information in images, but does not consider the impact of soil background. The Modified Normalized Water Index (MNDWI) (Xu, 2005) is effective in reducing the impact of non-water information and reducing errors in shoreline extraction. Its calculation formula is as follows:

Where Green is the reflectance of the green band and M I R is the reflectance of the mid-infrared band. The image is then binarized to obtain a black and white raster map, which is imported into ArcGIS for vectorization to obtain the water edge line. Finally, visual interpretation is used to obtain the coastline ( Figure 2 ), The process diagram for obtaining the coastline is shown in Figure 3 . In total, 200 sampling points are taken along the coastline in the study area. We use Google Earth satellite imagery to determine the marine/land attributes of each sampling point and validate the accuracy of the extracted coastline. The spatial resolution of Google Earth satellite imagery is typically higher than that of Landsat satellite imagery, ranging from 1m to 15m. After calculation, the accuracy rates of the three shoreline land-sea attribute classifications are 86.4%, 91.2%, and 89.6%, respectively. This indicates a good accuracy in coastline extraction.

The Digital Shoreline Analysis System (DSAS) is an application software for the ArcGIS system platform. It can systematically calculate the rate of coastline change and quantitatively analyze long-term and short-term coastline changes (Liu et al., 2020). Extracted shorelines are imported into a personal geodatabase, and buffers are established around the shoreline. Choose a suitable baseline based on the morphology of the coastline. Generate tangents from the baseline towards the coastline, and adjust overlapping and intersecting tangents. We analyzed the erosion and siltation of the Yellow River Delta coastline from 2009 to 2019 by calculating the Net Shoreline Movement (NSM), End Point Rate (EPR), and Linear Regression Rate (LRR) based on the coastlines in 2009, 2014, and 2019.

Net Shoreline Movement (NSM) refers to the calculation of the distance separating the most recent coastline from the ancient coastline The calculation formula is as follows:

Where d y o u n g represents the distance from the baseline to the youngest coastline and d o l d represents the distance from the baseline to the oldest coastline.

The End Point Rate (EPR) is determined by dividing the Net Shoreline Movement (NSM) by the time interval between the two coastlines. The calculation formula is as follows:

Where t y o u n g represents the time of the youngest coastline and t o l d represents the time of the oldest coastline.

The Linear Regression Rate (LRR) is calculated based on least squares fitting through the intersection of transects and coastlines and requires coastlines at three or more time points (Himmelstoss et al., 2018). The calculation formula is as follows:

Where y is the spatial position of the coastline, x is the year of the coastline, ∑ i = 1 n ( x i − x ¯ ) ( y i − y ¯ ) is the fitted constant intercept, and y ¯ − a x ¯ is the regression slope, i.e., LRR.

Using the DSAS, we selected the current and abandoned estuary shorelines of the Yellow River Delta in 2009, 2014, and 2019. A total of 772 cross-sections were generated from north to south along the baseline at intervals of 100m ( Figure 4 ). The calculated LRR and EPR are shown in Figure 5 .

From 2009 to 2014, the coastline near Shenxian Gou to Gudong Seawall (cross-section A and B) was basically eroded, with an average erosion rate of 206m/a. The current estuary area (cross-section C and D) showed an overall trend of silting towards the sea. The coastline advanced an average of 404m, with a maximum shoreline movement rate of 731m/a. The abandoned old estuary area was generally in a state of erosion, with an average erosion of 955m and a maximum shoreline movement rate of 847m/a. The coastline at the estuary (cross-section number C, D and E) changed dramatically, with a maximum change of 3954m. The change rate in other areas was basically stable and did not exceed 408m/a. From 2014 to 2019, the current estuary area was mainly in a state of siltation. Cross-sections B changed from an eroded state to a silted state, with a maximum shoreline movement rate of 233m/a. Cross-sections D changed from a silted state to an eroded state, with a maximum shoreline movement rate of 256m/a. The coastline changes in the abandoned old estuary area were relatively gentle, with only a small part of the coastline showing erosion. The erosion and siltation range was less than that from 2009 to 2014. The average coastline movement was 483m and the maximum shoreline movement rate was 443m/a. From 2009 to 2014, the current estuary increased by an average of 7.7km²/a per year, and from 2014 to 2019 it increased by an average of 7.22km²/a. From 2009 to 2019, the coastline in the study area exhibited an average movement distance of 1285 m (NSM) and an annual change rate of 127.7 m/a (LRR). Among them, the maximum growth and annual change rate of the shoreline were respectively 6.59km (NSM) and 668m/a (LRR), while the maximum erosion and annual change rate were respectively 4.7km (NSM) and 485m/a (LRR). The average increase in the area of the current estuary was 7.68km²/a, while the average decrease in the area of the old estuary was 4.91km²/a.

The dynamic evolution of the delta coastline is determined by the interaction between river inputs and ocean dynamics, with the inflow of seawater and sediment being the main source of shaping the morphology of the delta coastline (Ji et al., 2018). Figure 6 displays the changes in water and sediment discharge from the Yellow River between 2009 and 2019, and the data was obtained from the Sediment Bulletin of the Yellow River Conservancy Commission.

The water and sediment discharge of the Yellow River is continuously changing due to the influence of climate change and human activities. Between 2009 and 2014, the direction of siltation at the estuary shifted towards the north and the extent of shoreline siltation gradually expanded, due to water and sediment management measures. From 2016 to 2017, the interruption of water and sediment regulation resulted in a decrease in the discharge of water and sediment into the ocean. As a result, the primary factor driving changes in the shoreline was replaced by waves. In 2018, water and sediment regulation was resumed, resulting in the increase of the siltation rate at the current estuary and made water and sediment discharge to be the major factor once again for affecting shoreline changes. From 2014 to 2019, the implementation of water and sediment management resulted in a pronounced accumulation of sediment in the study region. In the old estuary area, ocean dynamics were the dominant force, resulting in continuous shoreline erosion. During this period, although sediment discharge could accelerate shoreline evolution, its impact on the shoreline was gradually weakened due to ocean dynamic. The coast of the Yellow River Delta is impacted by the interaction of water and sediment discharged into the ocean, as well as the dynamic of the surrounding sea.

Delft3D is a powerful computer software package developed by the Delft Hydraulics Research Institute (WL Delft Hydraulics), designed for conducting water environment simulation research in various scenarios. The Delft3D-FLOW model is based on Navier-Stokes equations and numerical solution by adopting Alternating Direction Implicit method (ADI).

(1) Continuity equation

Where U and V are the average velocities in the ξ and η directions in the curved coordinate system, respectively; Q is the change in volume due to water inflow, outflow, evaporation or precipitation per unit area; P and E represent precipitation and evaporation, respectively; q i n and q o u t represent the inflow and outflow of water per unit volume, respectively; G ξ ξ and G η η represent the conversion factors from the curved coordinate system ξ and η to the x and y directions in the rectangular coordinate system; H is the total water depth; σ represents the σ coordinate system, which is defined as follows: σ = z − δ H z is the vertical coordinate in physical space; δ is the free surface elevation above the reference plane (at z = 0).

(2) Dynamic equation

Where ω is the vertical velocity in the sigma coordinate system; f is the Coriolis force parameter; u and v are the flow velocities in the ξ and η directions; ρ 0 is the reference density of water; P ξ and P η are the hydrostatic pressure gradients in the ξ and η directions, respectively; F ξ and F η are the turbulent momentum fluxes in the ξ and η directions, respectively; v V is the vertical turbulent viscosity coefficient; M ξ and M η are source/sink terms for momentum in the ξ and η directions, respectively.

Using the Delft3D model, hydrodynamic models were constructed under the shoreline conditions of 2009, 2014, and 2019, respectively. The model uses an orthogonal curvilinear grid. The grid range is between 117° E~122° E, 37° N~41° N, covering the entire Bohai Sea and the western region of the Yellow Sea. In the Yellow River Delta area, the grid is refined and aligned with the shoreline to optimize its utilization and reduce computational time. The model grid is shown in Figure 7 . The offshore bathymetry data used in the model is obtained from the ETOPO database, which provides bathymetry data for a large regional area with a resolution of 1′. The bathymetry data for the study area is derived from nautical charts with a precision of 500 m. The offshore boundary is defined by water level boundaries, driven by eight constituents including M₂, S₂, N₂, K₂, K₁, O₁, P₁, and Q₁. To simulate the Yellow River estuary, a runoff boundary is implemented by using daily discharge data obtained from the Lijin station of the Yellow River Conservancy Commission, which operates under the Ministry of Water Resources, China. The model adopts cold start method with a Manning coefficient of 0.014 and a time step of 2 minutes. To establish a sediment transport model, the model includes the sediment type of cohesive sediment. Excluding sediment input from external sources, all sediment input comes from the Yellow River. The sediment data is sourced from the Lijin station’s sediment concentration and sediment report. The settling velocity is 0.2 mm/s, the critical shear stress for settling is 0.16 N/m², and the critical shear stress for erosion is 0.15 N/m².

The tide and current data are selected from three stations named H1 (119°45’5”, 38°32’21”), H2 (119°59’16”, 38°14’11”), and H3 (120°12’44”, 37°56’7”) in the Bohai Sea area. The sediment data are measured at two stations near the Yellow River Delta at P1 (119°6.376′, 37°30.466′) and P2 (119°6.43′, 37°18.979′). The simulation time ranges from October 1st to October 12th, 2018. To verify the accuracy of the simulation, three parameters are used for statistical analysis: Skill Score (SS), Root Mean Square Error (RMSE), and Correlation Coefficient (CC). Their calculation formulas are as follows:

Where X mod and X o b s are the model simulation values and field measured values, respectively; X ¯ mod and X ¯ o b s are the arithmetic means of the simulation and measured values, respectively; N is the number of statistical variables. An SS value greater than 0.65 suggests a very high level of accuracy, while a range of 0.5 to 0.65 indicates good accuracy. SS values falling between 0.2 and 0.5 represent fair accuracy, and an SS value below 0.2 indicates poor accuracy. After verification, the simulated data fit well with SS above 0.91 for tide verification and the RMSE above 0.74. The RMSE for current was less than 0.09m/s and the RMSE for flow direction was less than 37°.The overall trends of flow velocity magnitude,direction and suspended sediment concentration were consistent ( Figure 8 ). This indicated that the model had high reliability and could be used for our study.

This study set up 26 representational measurement points in the Yellow River Delta and Laizhou Bay to intuitively analyze the impact of shoreline changes on ocean dynamics (Seen in Figure 9 ). Points 1-9 are close to the shoreline while points 10-26 gradually extend towards the open sea. We selected the same time during high tide and low tide to compare flow velocity changes under different shoreline conditions ( Figure 10 ) and calculated change rates ( Table 2 ).

Overall, the results of flow velocity changes are mostly negative, with velocity in high tide period generally greater than that in low tide period. The measurement points that located near the Yellow River estuary, specifically between point 4 and point 9, are susceptible to changes in the shoreline and river discharge. The changes of flow velocity magnitude are relatively significant. Among them, points 4 and point 5 are located in the current estuary, with flow velocity above 0.5 m/s. From 2009 to 2019, the current estuarine coastline gradually extended outward, resulting in shallower water depths and a continuous decrease in flow velocity, with a decrease range of approximately 0.03 to 0.09 m/s. Points 7 and point 8 are located in the abandoned estuary. From 2009 to 2019, the abandoned estuarine coastline underwent erosion, resulting in a deepening of the seabed. The flow velocity initially increased significantly by approximately 0.08 to 0.11 m/s, and then gradually decreased by 0.03 to 0.06 m/s. From 2009 to 2014, there was significant siltation along the northern coastline of Dongying, which resulted in a decrease in flow velocity at point 1 and point 2, with the maximum value of 0.3 m/s. Measuring points from 24 to 26 are located near the interior of the Bohai Sea and are relatively far from the coastline. They are minimally influenced by the changes of the Yellow River estuary coastline, with flow velocity variations below 0.03 m/s. Due to the topography, the measuring points within Laizhou Bay generally exhibit higher flow velocity, exceeding 0.4 m/s. Due to the siltation at the Yellow River estuary, the width of Laizhou Bay gradually increases. This leads to a decrease in flow velocity. In Laizhou Bay, the flow velocity decreased rapidly from 2009 to 2014, with a maximum change rate of 13.61%. From 2014 to 2019, the flow velocity showed a slow decrease tendency. From 2009 to 2014, changes in the coastline of the Yellow River Delta had a more pronounced impact on the speed of high tide in comparison to low tides. From 2014 to 2019, shoreline changes were relatively small and flow velocity changes were also relatively small, with a maximum change rate of 7.44% and a minimum of 0.05%. Overall, shoreline changes can have an impact on tidal currents within nearshore areas. The more dramatic the shoreline changed and the closer to the coast, the greater the impact showed.

Due to friction, seabed topography, boundary conditions and other reasons, residual currents caused by nonlinearity in tidal currents are called tidal-induced residual currents. Nonlinear effects are stronger closer to the coast (Wang, 2012). Eulerian tidal residual current refers to the average vector at a fixed location over one or more tidal cycles and is expressed as:

Where u E u l e r is Eulerian tidal residual current; T is statistical time; u is tidal current velocity at spatial point.

Based on simulation results for one spring tide cycle, we calculated surface residual currents in three years and obtained differences between residual currents for each year as shown in Figure 11 .

Sediment from the Yellow River subsides in the estuary channel, intertidal zone, and coastal area and is transported by residual currents to extend the shoreline of the current estuary towards the sea. Residual currents are influenced by many factors such as water depth, tides, shoreline and wind field, among which the most important factors are shoreline and water depth. The Bohai Sea region experiences minimal residual current, whereas Laizhou Bay and the Yellow River estuary area encounter strong residual current, averaging of 3-20cm/s. The residual currents primarily flow offshore, and the prominent coastline of the Yellow River estuary affects the magnitude and direction of the residual currents. To the north of the Yellow River estuary, the residual currents exhibit a northwest direction, while south of the estuary, the residual currents flow in a southeast direction. The surface residual currents are generally greater than those in other layers. From 2009 to 2019, the residual currents near the current estuary gradually increased by approximately 2 cm/s, while the residual currents near the abandoned old estuary decreased by approximately 1 cm/s. Along the coast of Laizhou Bay, there was an average decrease of 3-6 cm/s in the residual currents. However, in the southeast direction near Diaolongkou, the residual currents exhibited an increase tendency. The residual currents near the estuary and coastal areas are notably impacted by variations in the coastline. They generally exhibit speeds ranging from 10 cm/s to 25 cm/s and tend to maintain relatively stable flow directions. Residual current variations in the offshore areas are relatively small, with an average increase of 0.5 cm/s.

The coastline of 2019 is used for simulation. The model simulation is conducted for a period of one year. By adjusting the acceleration factor, the sedimentation and erosion changes are calculated for one year, two years, five years, and ten years, respectively. The results are shown in Figure 12 . The simulation results for a period of one to two years indicate that the current estuary area is showing signs of sediment accumulation. Due to the water flow from the Yellow River, the sedimentation at the estuary is relatively small. There is only slight erosion in the abandoned estuary, with an erosion rate of less than 0.15 m/a. The calculation results for erosion and siltation over a period of five to ten years indicate that there is significant siltation in the nearshore area within 15.45 km of the current estuary. The sediment thickness can reach up to 8 m after ten years. Erosion has intensified at the abandoned estuary, and erosion phenomena have appeared within a 17.7km away from the Yellow River Delta. The maximum erosion thickness can reach up to 2 m. Under the dynamics of tidal currents, the sediment carried by the Yellow River mainly accumulates at the current estuary. The shorter the distance to the estuary, the greater the degree of sedimentation. The abandoned old estuary area is experiencing erosion in the coastal area, with the greatest erosion intensity observed at the estuary. To the north of the Yellow River Delta, there is a counterclockwise circulation, while to the south, there is a clockwise circulation. Influenced by the direction of residual currents, sediment deposition is occurring in the northeast direction, gradually shifting towards the northern region.