Incision of Submarine Channels Over Pockmark Trains in the South China Sea

The genesis of submarine channels is often controlled by gravity flows, but channels can also be formed by oceanographic processes. Using multibeam bathymetry and two‐dimensional seismic data from the western South China Sea, this study reveals how pockmarks can ultimately form channels under the effect of bottom currents and gravity‐driven sedimentary processes. We demonstrate that alongslope and across‐slope channels were initiated by pockmark trains on the seafloor. Discrete pockmarks were elongated due to the erosion of gravity‐driven sedimentary processes and bottom currents, and later coalesced to form immature channels with irregular thalwegs. These gradually evolved into mature channels with continuous overbanks and smooth thalwegs. Submarine channel evolution was significantly influenced by seafloor topography since the Late Miocene. The evolutionary model documented here is a key to understanding how channels are formed in deep‐water environments.

Despite the above, it is still unclear whether pockmarks can evolve into channels and which processes control their morphology. In order to decipher the latter processes, this study aims to: (a) characterize the morphology and internal architecture of channels in a poorly studied part of the South China Sea; (b) reconstruct the initiation and interpret the processes controlling the development of the investigated channels; and (c) reveal the role of pockmarks in submarine channel incision. The Western South China Sea is an ideal region to study the evolution of pockmarks because their origin is well known (Lu et al., 2017), and submarine channels with different orientations and sizes are abundant (Figure 1).

Materials and Methods
High-resolution multibeam bathymetric data and two-dimensional (2D) multi-channel seismic reflection profiles are used in this study.
The multibeam bathymetric data were acquired in 2008 by the Guangzhou Marine Geological Survey (GMGS) using a SeaBeam 2112 system. The dataset covers an area of ∼10,000 km 2 with a water depth range between 300 and 1300 m. These bathymetric data have a horizontal resolution of ∼100 m (cell size) and a vertical resolution of ∼3 m (3‰ of the water depth). The data were imported and analyzed in Global Mapper ® .
Two-dimensional (2D) seismic reflection data were acquired by the China National Petroleum Company (CNPC) in 2005 and processed by the PetroChina Hangzhou Research Institute of Geology. The data were migrated with a common midpoint (CMP) spacing of 12.5 m and a main frequency bandwidth of 30-45 Hz (main frequency: 35 Hz). The vertical resolution of the seismic data approaches 25 m. The 2D seismic data were interpreted on Landmark ® . The ages of main seismic stratigraphic markers were based on Lu et al. (2017).
In order to provide detail about the typical values of current velocity in the western South China Sea, we show in Figure S9 in Supporting Information S1 the average currents along a transect (see location in Figure 1a) in four different years: 2009, 2010, 2011, and 2012. Current measurements were acquired using a vessel-mounted ADCP Ocean Surveyor 38 kHz (OS38) and were published by Yang et al. (2019).

Geological Setting
The South China Sea was formed from Oligocene to the middle Miocene and is the largest (∼3.5 × 10 6 km 2 ) and deepest (>5000 m) marginal sea in the western Pacific Ocean (Li et al., 2014;Zhou et al., 1995). The study area lies southwest of the Xisha Archipelago on a topographic high identified between two drowned carbonate platforms, the Guangle High (GH), and the Zhongjianbei Carbonate Platform (ZCP) (Figure 1b). Pockmarks are abundant and relate to regional hydrothermal activity and gas seepage (Gao et al., 2019;Lu et al., 2017). Sediment cores collected to the west of the study area ( Figure 1a) indicate that bottom sediment is mainly composed of silt, with the particle diameter representing the 50% cumulative percentile value (D50) ranging between 5 and 50 μm (Astakhov, 2004a(Astakhov, , 2004bFigure 1a).
This study focuses on the shallow strata of the western South China Sea, which can be subdivided into three seismic-stratigraphic units: Unit 1 (Quaternary); Unit 2 (Pliocene), and Unit 3 (Late Miocene). The bases of these units correlate with seismic horizons of T20, T30, and T40, respectively (Figures 1d and S1 in Supporting Information S1). The seismic-stratigraphy of the study area is interpreted based on regional correlations with adjacent regions (Lu et al., 2017).

Oceanographic Setting
The South China Sea is a semi-enclosed marginal sea connected to the Pacific Ocean through the Luzon Strait (Liu et al., 2008). At present, the western South China Sea comprises four main water masses: surface water (at a water depth between 0 and 750 m), intermediate water (at water depths between 750 and 1500 m), deep and 3 of 13 bottom waters below 1,500 m (Quan & Xue, 2018;Yin et al., 2021). Quan and Xue (2018) proposed a layered circulation model for the western South China Sea in which current direction between 700 and 1500 m water depth is to the south in the northern part of the study area, but changes to a northward direction in the southern part ( Figure 1a). According to the vessel-mounted ADCP data from Yang et al. (2019), ocean currents close to Figure 1. Location, oceanography, and seismic-stratigraphic markers of the study area. (a) Bathymetric map of the western South China Sea revealing the location of the study area. The purple arrows indicate the circulation direction at a water depth of 700-1500 m based on Quan and Xue (2018). The yellow dots indicate the location of the speed profiles for the ocean currents shown in Figure S9 in Supporting Information S1 that were acquired with a vessel-mounted ADCP (2009)(2010)(2011)(2012) and published by Yang et al. (2019). The red triangles show the location of sediment cores collected for grain size analysis of sea-bottom sediments (Astakhov 2004a(Astakhov , 2004b. XA and ZA indicate the location of the Xisha and Zhongsha Archipelagos, respectively. (b) Multibeam bathymetric map showing the submarine channels and pockmarks analyzed in this study. The bathymetric profiles show the geometry of channel cross-sections. The purple dashed lines indicate the tracks of across-slope and alongslope pockmark trains. GH, Guangle High; ZCP, Zhongjianbei Carbonate Platform. (c) A zoomed-in inset of the seismic profile shown in (d) highlighting the internal architecture of an across-slope channel. The dashed dark-blue line reveals the base and wall of the oldest paleo-channel observed under a modern submarine channel. (d) Two-dimensional seismic profile showing regional stratigraphic units (based on Lu et al., 2017) and main structures around the studied channel system. The dark blue arrows indicate the oblique migration of pockmarks toward the east. Seismic horizons T20, T30, and T40 correlate with the bases of Quaternary, Pliocene, and Late Miocene strata, respectively. the study area show a variable behavior, with their average speed ranging from 10 to 20 cm/s. The measured maximum speed of ocean currents reaches 80 cm/s (Figures 1a and S9 in Supporting Information S1).

Giant Pockmark Field
A giant pockmark field covering an area of more than 9,000 km 2 is recognized on the multibeam bathymetric map in Figure 1b. Pockmarks are widespread and generally arranged in continuous trains of pockmarks (Figure 1b). These pockmark trains are divided into two main categories: alongslope and across-slope. Pockmark trains that are parallel to the regional bathymetric contours, or located around bathymetric highs, are herein named "alongslope pockmark trains." This category includes pockmarks formed around the ZCP and in the eastern part of study area (Figures 1 and 2b). The second category is named "across-slope pockmark trains" and comprises those aligned in a trend perpendicular to the bathymetric contours ( Figure 1b). Across-slope pockmark trains are observed on the slopes south, east, and north of the GH, at a water depth ranging from 750 to 900 m (Figures 1b  and 2a).
Bathymetric data reveal that pockmarks are diverse in their geometry and dimensions ( Figure 2 and S8 in Supporting Information S1). Pockmark depth varies between 50 and 180 m, with a maximum diameter from 1 to 3 km ( Figure S8 in Supporting Information S1). Pockmarks are also variable in plan-view comprising elongated, comet-shaped, circular-and crescent-shaped features (Figures 1b,2b,and 2c). Circular pockmarks are relatively small and isolated when compared to the three other types, with pockmark widths between 0.8 and 1.5 km (Figures 2c and S8 in Supporting Information S1). Crescent pockmarks are shaped as slender curves and distributed in groups; their concave side is aligned in a similar direction ( Figure 2b). Comet-shaped and elongated pockmarks are 1-3 km wide and 80-170 m deep, values that are similar to the width and depth of adjacent channels. They are usually aligned and show a consistent orientation (Figures 2b and 2c).
Seismic profiles reveal that most pockmarks are formed during or after the Pliocene as they occur above or truncate horizon T30 (Figures 1d and 3), with only a few forming before the Pliocene ( Figure S6 in Supporting Information S1). Such a characteristic is further discussed in Section 6 of this paper.

Channel Systems
The studied submarine channels are classified into alongslope and across-slope channels based on their orientation and geometry (Figures 1b and 2). In addition, they can be defined as mature and immature channels based on: (a) the roughness of their thalwegs, and (b) the relative continuity of channel plane morphology ( Figure 2 and Table S1 in Supporting Information S1). In essence, mature channels reveal smoother thalwegs and a more continuous morphology when compared to immature channels ( Figure 2).

Across-Slope Channels
Across-slope channels are perpendicular to the regional bathymetric slopes and chiefly located north and south of the GH (Figures 1b and 2c). For example, a large across-slope channel (D-D′ in Figures 2c and 2d), north of the GH (at ∼16°N) is shown as a ∼38 km long feature with a gentle thalweg dipping toward the NW ( Figure 2c). As the largest mature channel in the study area, channel D-D′ has also the smoothest thalweg and the largest average channel width and depth at, respectively, 2.6 km and 240 m (Figures 2c and 2d). Channel D-D′ is connected to the alongslope channel F-F′ at its southern end ( Figure 2c).
Several immature channels and across-slope pockmark trains are connected to channel D-D′ ( Figure 2c). In the southwestern part of channel D-D′, immature channel E-E′ has a rough thalweg and is connected to channel D-D′ at both its ends (Figures 2c and 2d). Channel E-E′ is significantly shorter (∼20 km) than channel D-D′, and it is also narrower (1.1 km on average) and shallower (94 m on average).
Across-slope channels follow a SE orientation south of the GH, where the slope gradient is ∼0.5° (Figures 1b  and 2a). Channels are roughly parallel to each other and 30-50 km long (Figure 1b). They have rugged thalwegs and discontinuous morphologies (Figure 1b). These across-slope channels have bankfull widths ranging from 1 to 1.5 km, and depths between 50 and 200 m (Figure 2c). Importantly, the across-slope channel G-G′ was formed in a zone with abundant isolated pockmarks and pockmark trains (Figures 1b and 2c). Southwest of channel G-G′, across-slope channels occur on the slope and remain ∼15 km distant from the GH (Figure 1b). To the north of channel G-G′, two across-slope channels of ∼14 and ∼18 km long reveal a relatively flat thalweg and connect to the southern end of channel F-F′ (Figure 2c). These across-slope channels are 1.5 km wide on average and are 50-150 m deep.

Alongslope Channels
Alongslope channels are mainly observed along the southern and western slopes of the ZCP and to the east of the GH (Figures 2b and 2c). To the east of the GH, an alongslope channel (F-F′) is identified as a ∼20 km long feature that runs parallel to the 800 m bathymetric contour (Figures 2a and 2c). Channel F-F′ has an average width of 2.6 km, and its depth ranges from 150 to 180 m (Table S1 in Supporting Information S1). Channel F-F′ has a smooth thalweg and two ∼50 m high topographic highs at both ends ( Figure 2d). These two highs occur at the confluences of channel F-F′ with across-slope channels occurring to the north and south (Figure 2c). A small channel with a sharp bend (∼1.5 km wide, ∼9 km long and with an average depth of 120 m), and trains of elongated pockmarks (∼1.3 km wide and ∼90 m deep), join channel F-F′ in its eastern part (Figures 1b and 2c).
Alongslope channels are the most significant features around the ZCP, being parallel to the platform slopes at a water depth between 1000 and 1200 m (Figures 1 and 2b). Here, the length of alongslope channels ranges from 10 to 25 km, with their width varying between 1 and 2.5 km. Their depth ranges between 50 and 200 m. The channels closest to the ZCP (A-A′) are the shallowest, showing an average depth of 73 m (Figures 1b and S1 in Supporting Information S1). These channels present elevations higher than 150 m within their thalwegs, with slope gradients of 0.5°-0.9° (Figure 2d). Furthermore, the channels closer to the ZCP, for example A-A′, have smoother thalwegs and more continuous plan-view morphologies when compared to more distant channels such as B-B′ and C-C′ (Figures 2b and 2d). Several elongated pockmarks and alongslope pockmark trains occur along, or parallel, to the latter channels (Figure 2b).

Seismic Architecture of Channels and Pockmarks
Seismic reflections are generally continuous and parallel between modern across-slope channels (e.g., D-D′) and horizon T20 (Figures 3b and 3c). In contrast, seismic reflections beneath the modern alongslope channels (e.g., the channel next to A-A′) are significantly truncated (Figure 3a). Chaotic strata with low amplitude are rarely identified in the channel-fill deposits of the modern channel ( Figure 3). Channel D-D′ is remarkably wider and deeper when compared to the other channels imaged in seismic data (Figures 3 and S4 in Supporting Information S1). The inception of some channels, such as D-D′, is recognized between horizons T20 and T30 (Figure 3), with a limited number of channels initiated below horizon T30 ( Figures S4 and S6 in Supporting Information S1). Channel A-A′ is a moat at the foot of the ZCP associated with a contourite drift and shows a typical mounded shape with internal reflections dipping toward the bottom of channel A-A′ (Figure 3a).
There are differences among the seismic cross-sections of across-slope and alongslope channels. Alongslope channels, such as that next to channel A-A′, show distinctive truncations on their banks (Figure 3a). In contrast, across-slope channels such as channel D-D′ are usually located above paleo-channels with chaotic and high amplitude seismic reflections on their bases (Figures 3b and S4 in Supporting Information S1). Seismic reflections on the banks of across-slope channels generally dip toward the channel thalweg (Figures 2b and 2c and S4 in Supporting Information S1). Fluid escape features are identified as convex or chaotic seismic reflections crossing particular seismic reflections (Figure 3). These fluid escape features are sourced from strata older than horizon T30, and truncate the seismic reflections above this same horizon (Figure 3). Most fluid escape features are connected to channels and pockmarks on the modern seafloor (Figures 3 and S4, S5 and S6 in Supporting Information S1).  Table S1 in Supporting Information S1. Some paleo-pockmarks were buried after horizon T30, while a few of the pockmarks at the seafloor show oblique migration since their inception (Figure 1d).

Genesis of Channels and Their Relationship to Pockmark Trains
The studied channels show variable orientations. Alongslope channels such as A-A′ and F-F′ run parallel to the slope contours, whereas across-slope channels (D-D′ and G-G′) developed perpendicularly to the slope topography ( Figure 2). Previous studies interpreted the channels in the study area as moats and furrows formed by contour currents (Yin et al., 2021). However, some of the furrows and channels described in Yin et al. (2021) are perpendicular to the slope contours and, thus, unlikely to be associated with contour currents flowing alongslope. Therefore, other factors probably control their origin in the study area. Oceanographic processes such as internal waves (e.g., internal tides) can flow transversely to the slope, forming intense near-seafloor currents and resuspending sediment, especially inside submarine canyons (Aslam et al., 2018;Puig et al., 2013Puig et al., , 2014. In the northern South China Sea, internal tides have been considered as a process responsible for downslope-migrating sand dunes (Ma et al., 2016). Although internal tides could, in part, contribute to the erosion of the interpreted channels, they are probably not the main factor controlling their inception in the study area.
Interactions between gravity-driven processes and fluid escape in pockmarks can reshape the latter to form comet-shaped pockmarks oriented perpendicularly to the slope (Chen et al., 2019), ultimately forming across-slope channels (Gay et al., 2006;Nakajima et al., 2014;Pilcher & Argent, 2007). Several pockmark trains are perpendicular to the slope gradient north and south of the GH, effectively comprising circular, comet-shaped, and elongated pockmarks (Figure 1b). On the slopes surrounding the GH, active gas seepage through the pockmarks brings deep, unlithified sediment to the seafloor, while the GH comprises an active carbonate factory from where sediments are derived, contributing to the occurrence of gravity flows and slumps (Gay et al., 2006;Lu et al., 2017;Nakajima et al., 2014;Yang et al., 2021). Under the erosion of gravity currents, circular pockmarks were reshaped to form elongated and comet-shaped pockmarks. Furthermore, pockmarks are not only scattered around the investigated channels but also occur inside the channels themselves; hence, irregular depressions in channel thalwegs are the relics of reshaped pockmarks (e.g., G-G′ and E-E′ in Figure 2). Pre-existing pockmark trains affected by gravity currents probably contributed to the formation of across-slope channels on the slopes surrounding the GH. In the study area, the paleo-channels below modern across-slope channels commonly contain channel-fill deposits with chaotic and high amplitude seismic reflections onlapping the bases of paleo-channels (Figures 3b and S4 in Supporting Information S1). These are typical seismic facies indicating the presence of gravity flow deposits (Figures 3b and S4 in Supporting Information S1) .
Contrasting with across-slope channels, there are alongslope channels such as F-F′, A-A′, and C-C′ that run parallel to the bathymetric contours (Figures 2b and 2c). They are likely formed by alongslope currents. Alongslope channels identified near the foot of the GH and ZCP (e.g., F-F′ and A-A′; Figure 2b) are contourite moats and furrows associated with an isolated mounded drift recognized by Yin et al. (2021) (Figure 3a). They are thus related to contour currents flowing along the GH and ZCP, which were strong enough to erode the seafloor and generate erosional truncations on the banks of alongslope channels (Figures 2a and S4 and S6 in Supporting Information S1). Although average bottom currents are relatively weak in the western South China Sea, below 20 cm/s, they can be variable enough to reach a maximum velocity close to 80 cm/s ( Figure S9 in Supporting Information S1; Yang et al., 2019). These periods of intense circulation could be responsible for the observed seafloor erosion (Stow et al., 2013). Andresen et al. (2008) and Kilhams et al. (2011) have suggested that bottom currents can induce the erosion of pockmarks, reshaping and coalescing them along the direction of bottom currents. When this process is maintained for a relatively long time, it results in the formation of alongslope channels similar to those observed on the southwest and southeast flanks of the ZCP (Figures 1b and 2). Bottom current erosion in its broader sense is enhanced on the downstream sides of pockmarks to form asymmetric and elongated features (Figure 2b; Masoumi et al., 2014). The elongated pockmarks are further eroded and coalesce to form channels. In fact, relics of elongated pockmarks are found as asymmetric depressions in some of the channel thalwegs, for example in channel C-C′ (Figure 2d). Micallef et al. (2014) first tested the concept of space-for-time substitution when reconstructing the evolution of submarine canyons and channel systems on continental margins. They suggested that when the established model matches well with the morphological patterns interpreted on geophysical data, time can be substituted by space to reconstruct the evolution of canyons and channels. To illustrate channel development in the study area, we propose a space-for-time substitution model comprising three stages: (a) a channel-inception stage, in which trains of pockmarks provide favorable pathways for eroding gravity flows and bottom currents, (b) an immature stage, during which discrete pockmarks are elongated and coalesce to form immature channels with a rugged thalweg and a discontinuous morphology in plan-view, once again under the erosion of gravity flows or bottom currents, and (c) a mature stage, in which bottom currents and gravity flows are funneled through the channels to smooth their floor and confining banks (Figure 4). Therefore, under the erosion of gravity processes and bottom currents, pockmark trains gradually form immature channels to finally evolve into a complex system of across-slope and alongslope channels (Figure 4).

Evolution of Submarine Channels in the Western South China Sea
In the study area, Lu et al. (2017) proposed that the accumulation and dissociation of gas hydrates significantly contributed to the formation of pockmarks. In parallel, Gao et al. (2019) suggested that pockmarks were formed by hydrothermal fluid flow occurring since the Pliocene. The oldest paleo-channel below channel D-D′ occurs between horizons T20 and T30, suggesting the Pliocene as the time of its inception (Figures 1d and 3). Channel D-D′ is one of the most mature in the study area, and its stratigraphic position correlates with a period of enhanced hydrothermal activity in the Pliocene as identified in Gao et al. (2019). However, there are differences in the timing of inception of channels in the study area, even when considering different reaches of the same channel. Some alongslope channels such as A-A′ have eroded horizon T20, indicating they were formed after the Pliocene (Figures 3a and S2 in Supporting Information S1). Other alongslope channels are identified ubelow horizon T30, on the southeastern flank of ZCP, suggesting an earlier inception ( Figure S6 in Supporting Information S1). Based on the interpreted seismic data, the earliest period of channel inception can be traced to the Late Miocene in the study area.
Seismic reflections on the banks of channel D-D′, above horizon T20, are continuous and parallel, but seismic reflections between horizons T20 and T30 are truncated by paleo-channels or horizon T20, suggesting that erosional processes dominated during channel inception, with the resulting channels becoming filled in their mature stages (Figure 3b). Widespread immature channels such as E-E′, and pockmark trains such as B-B′ formed around mature channels, also show that the investigated system of channels is still evolving (Figure 2). Abundant truncations on the banks of immature channels suggest that erosional processes still dominate their development ( Figure 3). This means that present-day immature channels can still evolve into mature features if gravity flows and bottom currents keep eroding the seafloor pockmarks mapped in this work (Figure 4).
Channel evolution was significantly influenced by seafloor topography, which predominantly controlled the dynamics of ocean currents. Changes in slope gradient not only determine the formation of channels but also control the transition between erosional and aggradational processes in them (Micallef & Mountjoy, 2011). Slope gradients differ to the north, south, and east of the GH; hence, the steepest slope (∼0.8°) north of the GH led to the formation of channel D-D′, which is the widest and more deeply incised of all investigated channels. Mature and immature channels also formed on the slope to the south of the GH, which records a moderate gradient of ∼0.5° (Figures 2 and 4). The slope to the east of the GH does not present any across-slope channels, probably because it is relatively gentle (<0.3°) and, therefore, relatively stable and less likely to be affected by gravity flows (Figures 2 and 4). In addition, it is known that bathymetric obstacles influence the dynamics of bottom currents and control the formation of alongslope channels (Hernández-Molina et al., 2006;Yin et al., 2021). Thus, alongslope channels were preferentially formed around the GH and ZCP (Figure 2). Mulder et al. (2018) demonstrated that sediment supplied by channels (or canyons) onto deep-water depocenters can originate from topographic highs instead of a point source. One gully located on the eastern slope of the GH is connected to channel F-F′ in a zone with a topographic high in the thalweg (Figure 2c). This zone may also contain gravity deposits transported from the GH but, unfortunately, no seismic or sediment core data were available to confirm such an assumption. Furthermore, Wu et al. (2016) revealed an Early Pliocene paleo-slope topography similar to the modern seafloor, and considered it to have had an important morphological control on the development of channel systems. . Schematic diagrams, combined with a three-dimensional morphological map of the study area, summarizing the time-step evolution of channels around the interpreted pockmark field. Stage 1: channel inception is controlled by a pockmark train; Stage 2: under the effect of gravity flows and bottom currents, discrete pockmarks are eroded and coalesce to form an immature channel; Stage 3: gravity flows and bottom currents continue to erode the immature channel, which subsequently evolves into a mature channel with a smooth, continuous thalweg. The purple arrows indicate the direction of gravity flows. The white and yellow arrows indicate the pathways of bottom currents at water depths of ∼800 and ∼1,000 m, respectively. Compared to other well-studied channels in the South China Sea (Chen et al., 2020), the channel system investigated in this work is characterized by its complex morphology, controlled by the effect of multiple depositional and oceanographic processes. Hence, the recognition of a system of across-slope and alongslope channels, initiated from pockmarks and influenced by seafloor topography, has significant implications to the current understanding of how submarine channels are initiated on continental margins across the world.

Conclusions
High-resolution multibeam bathymetry and two-dimensional seismic data enabled us to investigate the morphology of a complex system of channels in the western South China Sea, focusing on its genesis and evolution. The main conclusions of this study are as follows: 1. The studied channel system comprises a large number of across-slope and alongslope channels found within a giant pockmark field, which covers an area of more than 9,000 km 2 at a water depth of 700-1,200 m; 2. The channels analyzed in this study are formed by the incision of gravity flows and bottom currents on seafloor pockmarks, particularly on those arranged as pockmark trains; 3. Based on the space-for-time substitution concept, the evolution of the channels can be summarized in three stages: Stage 1, in which the inception of the studied channels coincides with the erosion of pockmark trains; Stage 2, in which pockmark trains are eroded by gravity flows and bottom currents to form immature channels; Stage 3, during which immature channels evolve into mature channels, with a flatter channel floor, under the effect of continuous erosion; 4. The studied channel system was initiated in the Late Miocene and is still developing at present. Discrete channels were formed at different times and their evolution has been significantly controlled by an ever-evolving seafloor topography

Data Availability Statement
The seismic and bathymetric data supporting this research are owned by the China National Petroleum Corporation (CNPC) and Guangzhou Marine Geological Survey (GMGS), respectively, with commercial restrictions, and are not accessible to the public or research community. The seismic profiles used in this study can be accessed via: https://doi.org/10.5281/zenodo.5756961. The bathymetric data for this research are sourced from Lu et al. (2018) at https://doi.org/10.1190/INT-2017-0222.1.