Based on seismic and well logging data, three sets of MTDs, namely MTD1, MTD2, and MTD3, can be identified in the study area ( Figures 3 , 4 ). To enhance the understanding of the internal structural characteristics of MTDs from the bottom to the top, this work selected MTD1 for analysis due to its continuous core sampling. First, the core photographs of MTD1 and its upper and lower surface parts were analyzed ( Supplementary Figures S1 and S2 ). Then, they were drawn so that different core features could be seen, such as sediment gravity flows, bottom currents, as well as hemipelagic and pelagic deposits ( Figure 9 ).

MTD1 is located approximately between 19 mbsf to 42 mbsf ( Figure 3 ). This study, through the analysis of complete and continuous core samples between 18 mbsf to 43 mbsf, has identified a variety of core facies types. Core facies A exhibits the development of slump-folded deformed sediments within the layer extending from 37.28 mbsf to 37.36 mbsf ( Figure 9A ). The scale is not large, with several centimeters in size, and filled with sandy sediments internally. Core Facies C primarily consists of a large number of deformed thin sand layers, with the thickness generally being about 1 cm ( Figure 9B ). The shapes are mainly in the form of elongated strips with irregular extension directions. Core Facies D is primarily distinguished by variations in the orientation of the strata ( Figure 9C ). This core facies exhibits relatively weak internal deformation of the sediments, along with a notably large dip angle of the sedimentary layers, which is also evident in the well logging data ( Figure 3 ). Core Facies E exhibits the most intense deformation, characterized by a significant presence of irregular sedimentary masses ( Figure 9D ). These masses are predominantly composed of dark grey clay, which is distinctly separated from the in-situ sediments and situated at the base of MTD1 ( Figure 3 ).

The above are core facies of different types of MTDs ( Figures 9A–D ). According to core data, MTD1 is characterized by the presence of turbidity currents ( Figure 9E ) and hemipelagic and pelagic deposits ( Figure 9E ). Additionally, bottom currents have been identified in the underlying strata of MTD1 ( Figure 9F ), indicating the complexity of the sedimentary processes. Turbidite deposits are mainly composed of multiple thin sand layers with a thickness of about 0.5 cm, and the thickness of the bottom current deposits is also relatively small, approximately 1 cm ( Figures 9E, F ).

Based on the results obtained from the text above, it can be observed that different data sources have varying resolutions. Seismic data has the lowest vertical resolution, often showing a set of MTDs as a whole with chaotic, low-amplitude reflections, like MTD1 ( Figures 2 , 4B ) (Cardona et al., 2022). Well logging data exhibits a higher resolution, allowing for the differentiation of multiple small-scale MTDs, such as MTD1a, within a given set. This distinction is based on parameters including P-wave velocity, density, and dip logging ( Figures 3 , 10 ) (Sun and Alves, 2020; Crutchley et al., 2022). When combined with core samples, studying the core facies can show how each small-scale MTDs is structured inside ( Supplementary Figures S1 , S2 ; Figures 9 , 11 ) (Tripsanas et al., 2008; Lu et al., 2020). For instance, the MTD1 described in this study is composed of MTD1a, MTD1b, and MTD1c on log-scale data ( Figure 10 ). It is evident that within large sets of MTDs, there are hemipelagic and pelagic deposits, and even turbidite deposits, which are different from what is observed on seismic-scale profiles ( Figure 4B ). According to the core-scale vertical high-resolution core facies, each small-scale MTDs, such as MTD1a, MTD1b, and MTD1c, is composed of various sedimentary systems. Additionally, these deposits exhibit the development of both hemipelagic and pelagic sediments internally. ( Figure 11 ). From this, it can be seen that differences in data resolution can lead to variations in the identification of MTDs, a phenomenon also observed in other study areas (Crutchley et al., 2022; Lu et al., 2025).

Based on the well logging data and core facies, this work has delineated the MTD1a interval, which is approximately located between 19.15 and 20.65 mbsf ( Figure 10 ). On the logs, the P-wave velocity of this layer exhibits a marked increase from approximately 1500 m/s to around 1550 m/s. Concurrently, there is an observed rise in density, accompanied by a sudden alteration in dip (Sun and Alves, 2020; Crutchley et al., 2022). In the cores ( Supplementary Figures S1 , S2 ; Figures 9 , 11 ), MTD1a primarily develops two types of core facies, including Core Facies D and Core Facies K. It is evident that hemipelagic and pelagic deposits develop within MTD1a.

MTD1b is roughly between 24.00 and 31.60 mbsf ( Figure 10 ), where the P-wave velocity remains relatively constant. The density is slightly higher than in the MTD1a, and the dip changes are also very clear. In the cores ( Supplementary Figures S1 , S2 ; Figures 9 , 11 ), MTD1b develops three types of core facies, which are Core Facies C, Core Facies D, and Core Facies K. Particularly, the development of deformed thin sand layers is compelling evidence of the existence of MTDs, which have also been discovered in the Pearl River Mouth Basin in the northern South China Sea and are likely mainly debris flows (Moscardelli and Wood, 2008; Sobiesiak et al., 2018; Lu et al., 2025).

MTD1c is located approximately between 35.40 and 41.90 mbsf, with an overall thickness of about 6.50 m, which is slightly reduced compared to MTD1b ( Figure 10 ). Logs show that MTD1c is characterized by an increase in P-wave velocity and density, which is corresponding with the IODP sites all over the world (Sun and Alves, 2020). The core exhibits a more complex pattern with various deformation features ( Supplementary Figures S1 and S2 , Figures 9 and 11 ). The core deformation is severe, which develops four distinct core facies, including Core Facies A, Core Facies C, Core Facies E, and Core Facies K. Based on the core facies, MTD1c may mainly consist of slumps and debris flows, which is very similar to the Storegga slide (Moscardelli and Wood, 2008; Sobiesiak et al., 2018; Karstens et al., 2023). Furthermore, the phenomenon of sediment deformation can also be observed at outcrops (Shanmugam, 2012).

There is a significant difference in the identification of MTDs based on different resolutions, such as core-log-seismic data. MTDs are not entirely composed of deformed sediments; they also include a large number of hemipelagic and pelagic deposits and even turbidite deposits, providing strong evidence for understanding the sedimentary processes of MTDs.

For a long time, the sedimentary processes and patterns of MTDs have been inconclusive. Due to the lack of complete and continuous core samples, little literature clearly and distinctly points out their sedimentary processes (Bull et al., 2009; Shanmugam, 2022). It is known that MTDs are complex sedimentary bodies resulting from the collapse and seaward transportation of slope sediments, but how are the sets of MTDs observed on seismic profiles formed?

Based on high-resolution core data, patterns for the sedimentary process of MTDs have been proposed. The first pattern, referred to as Pattern 1, is characterized by a slope leading to the deep-sea basin that comprises undisturbed in situ sediments. These sediments are generally classified as hemipelagic and pelagic deposits ( Figure 12A ). Subsequently, the slope becomes unstable and experiences a collapse, resulting in the transport of a significant volume of sediments to the deep-sea basin. These sediments subsequently cover the existing in situ hemipelagic and pelagic deposits, leading to the formation of MTDs, which bear some resemblance to free-slip flows ( Figure 12B ) (Sobiesiak et al., 2018). In this case, if drilling is conducted here, like Site W05, when core samples are collected, hemipelagic and pelagic deposits or turbidite deposits within the MTDs cannot be found. Consequently, this observed pattern does not correspond with the actual geological conditions.

Next is Pattern 2, similar to Pattern 1, where slope sediments become unstable and are transported seaward. Sediments close to the slope cover the in situ hemipelagic and pelagic deposits. In contrast, sediments that are located further from the slope interact with the underlying sediments during transportation, resulting in the formation of a complex sedimentary body. This process is analogous to no-slip flows ( Figure 12C ) (Sobiesiak et al., 2018). This pattern can explain the presence of hemipelagic and pelagic deposits in MTD1 ( Figure 10 ). Moreover, unlike Pattern 2, Pattern 3 spans a longer time frame, encompassing multiple episodes of slumping and sedimentation processes ( Figure 12D ). After the first instability event ends, hemipelagic and pelagic deposits are deposited in the study area, covering the complex sediments. Subsequently, the newly formed slope becomes unstable and collapses once more. This process results in the transportation of new slope sediments seaward, thereby repeating the previously described processes and leading to the formation of another set of small-scale MTDs. This cycle continues, ultimately culminating in the development of large-scale MTDs. This pattern can explain the development of hemipelagic and pelagic deposits and even turbidite deposits within MTD1, and it can also explain why hemipelagic and pelagic deposits are developed within MTD1a, which more closely aligns with real-world scenarios ( Figure 11 ).

Gas hydrates are ice-like solid compounds with a cage-like structure formed by the combination of relatively low molecular weight gases such as methane and water under low temperature, high pressure conditions (Sloan, 2003), and about 95% of them are found in continental margins (Beaudoin et al., 2014). According to well logging data and seismic interpretation, it is evident that Site W05 drilled into tens of meters of turbidite sands beneath the MTDs, which also serves as a reservoir for gas hydrates ( Figure 3 ). The sharp increase in the RES logging and Vp logging curves can be used to identify the gas hydrate layer. Specifically, RES values reach up to 200 ohm·m, while Vp exceeds 3000 m/s, occurring approximately between 127 mbsf and 137 mbsf. Drilling results from multiple wells indicated that a large-scale flat-lying transitional sandy gas hydrate system is developed in the study area. The reservoir mainly composed of coarse-grained channel-levee and lobe bodies, covering an area of several square kilometers (Kuang et al., 2023). The relationship between MTDs and the accumulation of gas hydrates has been widely researched globally (Jung and Vogt, 2004; Boswell et al., 2012). This work has also discovered the phenomenon of multiple large-scale MTDs coexisting with gas hydrates in the deep-water area of the QDNB in the northern SCS ( Figures 2 , 3).

The dissociation of gas hydrates in the continental slope can trigger MTDs, and the formation of MTDs in the deep-sea plain can also influence the accumulation and dissociation of gas hydrates (Maslin et al., 2004; Wang et al., 2024). Specifically, suppose gas hydrate accumulation first and is subsequently covered by MTDs, which is Cycle ① ( Figures 13A, B ). In that case, MTDs can alter the temperature and pressure conditions of the environment where gas hydrates are located. This alteration disrupts the stability zone of gas hydrates, resulting in an upward shift of its base. Consequently, this process may lead to the dissociation of the hydrates ( Figure 13B ) (Wang et al., 2024). As exploration progresses, a large number of active cold seeps and seepage-type gas hydrate systems have been discovered in the QDNB (Feng et al., 2018; Ye et al., 2019). Based on the U/Th and AMS ¹⁴C dating of cold seep carbonates, the dissociation of gas hydrates occurred in three episodes: approximately from 136.3 ka to 114.6 ka, from 48.5 ka to 42.7 ka, and from 11.6 ka to 9.5 ka. These events may be related to three phases of the widespread MTDs development (Wei et al., 2022; Zhang et al., 2023). This also indicates that the accumulation and dissociation of gas hydrates in the study area are dynamic. Therefore, Cycle ① proceeds to the next phase, where deep thermogenic gas and shallow biogenic gas migrate to the gas hydrate stability zone to accumulate and re-form large-scale gas hydrates ( Figure 13C ). Subsequently, another phase of large-scale MTDs deposition leads to the dissociation of gas hydrates ( Figure 13D ), which then repeats the process of free gas and gas hydrate accumulation stage ( Figure 13C ).

Conversely, suppose MTDs form first, which is Cycle ② ( Figure 13B ). In that case, the fine-grained MTDs are denser internally, with lower porosity and permeability, making them an excellent seal for gas hydrates and shallow gas (Wang et al., 2011; Cheng et al., 2021; Crutchley et al., 2021; Wu et al., 2021). When free gas migrates to shallower regions and accumulates, MTDs can act as seals, inhibiting the vertical migration of gas. This process consequently facilitates the formation of extensive lateral gas hydrates ( Figure 13C ) (Kuang et al., 2023). Subsequently, it becomes caught in a dynamic cycle of multi-phase MTDs deposition and gas hydrate dissociation, and then free gas and gas hydrate accumulation ( Figures 13C, D ). Additionally, Cycle ③ directly transitions from gas hydrate accumulation to the deposition of multi-phase MTDs, leading to gas hydrate dissociation ( Figures 13A, B ). However, it ultimately enters the free gas and gas hydrate accumulation stage, remaining in a dynamic process ( Figures 13C, D ). Overall, the formation of gas hydrates is intricately linked to MTDs, exhibiting a dynamic coupling relationship between the two. Nevertheless, the spatio-temporal coupling mechanism between the two still requires in-depth research.

The QDNB, located in the northwestern SCS, has a complex sedimentary process and a variety of sedimentary systems (Gong et al., 2020; Su et al., 2020; Zhong and Peng, 2021; Chen et al., 2022; Li et al., 2024). This study, utilizing comprehensive and continuous core data, a substantial volume of logging-while-drilling data, and high-resolution 3D seismic data, has successfully identified and characterized sediment gravity flows. This includes MTDs, turbidity currents, as well as bottom currents. Additionally, it has examined hemipelagic and pelagic deposits, thereby establishing a sedimentary model for the deep-water area ( Figure 14A ). The formation of MTDs is mainly due to the instability of slope sediments, which are transported to the deep-sea. From a profile, MTDs can include slides, slumps and debris flows (Moscardelli and Wood, 2008; Sobiesiak et al., 2018). The bottom of the MTDs will erode the underlying strata, causing deformation and then mixing with in situ hemipelagic and pelagic deposits ( Figure 14B ). Turbidity currents also mainly originate from the slope, transporting sediments through turbidite channels and deposits. The evidence of bottom current development in the core proves the existence of bottom water in the northern SCS, providing a basis for the study of the deep-water circulation in this area (Yuan, 2002; Liu et al., 2010, 2017; Chen et al., 2019; Wang et al., 2019b). In addition, hemipelagic and pelagic deposits are also developed, mainly located in the slope and deep-sea areas.