The samples were pretreated to remove the organic matter, bioclastic, and calcium cement, before grain size measurement. Beckman Coulter LS 13 320 laser diffraction particle size analyzer (measuring range 0.017~2 000 µm, resolution 0.1 Φ) was used for grain size measurement, and 20% of the samples were taken to repeat the test, and the error of repeat test was ≤1%.

The rock chip samples were taken from the YC2 well ( Figure 1 ) located in the shelf-margin of the Qiongdongnan Basin, the northwestern SCS, and collected from the core house of CNOOC-Zhanjiang Branch. The sampling interval is from the upper Yinggehai Formation and Ledong Formation, with a total of 49 sediment samples collected with a sampling interval of 15-20 m. The predominant lithology is argillaceous siltstone, with some silty mudstone and a small amount of fine sandstone.

From bottom to top, the U4 and U3 units are dominated by silty mudstone, interbedded with thin layers of fine-grained sandstone and muddy medium-grained sandstone. The U2 unit is dominated by silty mudstone; while the U1 unit is dominated by mudstone with some conglomerates. The samples were processed using standard grain size analysis methods (Hao et al., 2023; Zheng et al., 2008), and the grain diameters were measured with a laser grain size analyzer, after which the Folk and Ward formulae (Folk and Ward, 1957) were chosen for the calculation of the relevant grain size parameters.

In the frequency distribution curve, the peak distribution range of a sediment grain size is approximately positively correlated with the sedimentary hydrodynamic conditions, while the number of peaks is closely related to the complexity of the depositional environment (Zhang, 2015). The results of sedimentary dynamics research show that the contents of various genesis components can differ depending on factors such as transport mode and transport distance. In general, the frequency curves of single-causal components are single-peaked distributions, and the frequency curves of multiple-causal components are multi-peaked distributions (Sun et al., 2018). The sediment source in well YC2 is mainly the input from the Red River, that is transported into the SCS by the surface circulation influenced by the monsoon (Chen et al., 2019; Zhao et al., 2019; Li et al., 2024). The vertical evolution of various grain size metrics in the YC2 well in the northwestern SCS could provide information for additional depositional hydrodynamics.

The sediments in U4 are mainly dominated by fine silt with good sorting coefficients ( Figure 3 ). The frequency curves are unimodal focused on the interval of 7~8 φ ( Figure 4 ). It is suggested that the sediments in U4 were influenced by relatively stable dynamic conditions during deposition. Additionally, the probability cumulative curves show a three-fold pattern, with a relatively high content of saltating grains ( Figure 5 ). It indicates an extensive influence of alternating fluvial and wave currents with a relatively strong hydrodynamic condition (Visher, 1969).

The sediments in U3 are similar to U4, and are mainly dominated by fine silt, with an average grain size value of 8.56 μm ( Figure 3 ). The sorting factor is about 1.53; it indicates that the sediments in U3 are well sorted. The frequency curves are also unimodal with the peak appearing between 6.5 and 7.5 φ, but one sample is bimodal ( Figure 4 -U3). The probability accumulation curves exhibit a three-fold characteristic, with a junction between rolling and saltating grain transporting at near 4 φ ( Figure 5 ). It indicates that the depositional environment was relatively complex, and the hydrodynamic conditions are strong.

The sediments in U2 are dominated by fine silt, with an increased content of coarser silt compared with those in U3 and U4 ( Figure 3 ). The average value of the sorting coefficients is about 1.51, indicating good sorting. The frequency curves are unimodal ( Figure 4 ), with the peak values concentrated in the interval of 6.5~7.5 φ, indicating that the depositional process may only be affected by a single transporting force. The sediments in U1 show a decrease in clay content but an increase in fine sand content, thus, the sediment grain size has coarsened ( Figures 3 and 4 ). The grain size frequency curves are unimodal with the peak values concentrated between 5.5 and 6.5 φ ( Figure 4 ). The probability accumulation curves of the three samples are four-segmented, and the overall slope is basically above 50° ( Figure 5 ). Among them, the rolling and levitation components have larger slopes and better sorting. The saltating component contains two sub-sets with the coarse intercepts of the rolling and saltating components near 4 φ, and the content of the rolling component is between 5 and 10%. Consequently, this indicates that the U1 has the most complex depositional environment, with the coarsest grain size of sediments implying a more enhanced and complicated hydrodynamic condition.

In summary, the sediments in units U4-U1 are mainly dominated by fine silt. The overall sorting is better, and the frequency curves are mostly unimodal curves ( Figures 4 , 5 ), indicating that the depositional process may be affected by one prominent hydrodynamic force. The probability accumulation curves of U4-U3 are mostly three-segmented, while U1 is four-segmented. The grain sizes of the rolling-saltating intercepts of U4-U1 become coarser and the content of saltating fractions increases; this indicates that the depositional environment became more complicated, and the hydrodynamic conditions appears to be enhanced (Liu et al., 2016; Li et al., 2024). The latter may be related to the influence of regional circulation (Fang et al., 1998; Xia et al., 2001; Liu et al., 2016). The rapid rises and falls of sea level and the enhancement of circulation controlled by the monsoon increased the total amount of sediment transported and imported, and the grain size increased upwards (Miller et al., 2005; Clift et al., 2008; Li et al., 2024).

Three environmentally sensitive components (EM1-EM3) were extracted by the end meta-fraction analysis method (EMMA) where the main grain size ranges were 2.42-8.68 μm (EM1), 5.21-12.7 μm (EM2) and 18.7-40.1 μm (EM3), respectively ( Figure 9A ). Four environmentally sensitive fractions were extracted from YC2 well sediments by the grain size-standard deviation (STD) method, which were 0.46-0.767 μm (A0 fraction), 4.03-6.72 μm (A1 fraction), 27.4-35.3 μm (A2 fraction), and 586-756 μm (A3 fraction, Figure 9B ). In addition, four environmentally sensitive components were extracted by principal component factor analysis (PCA), corresponding to grain sizes of 1.65-11.2 μm (F1-1), 18.7-86.4 μm (F1-2), 0.314-2.42 μm (F2-1), 12.7-16.4 μm (F2-2) and 98.1-144 μm (F3), respectively ( Figure 9C ). Comparison of the environmentally sensitive grain sizes extracted by the three methods revealed that the A1 component (4.03~6.72 μm) extracted by the grain size-standard deviation method, the EM1 (2.42~8.68 μm) and EM2 (5.21~12.7 μm) extracted by the end-member modelling analysis method, and the F1-1 (1.65~11.2 μm) extracted by the principal component factor analysis. These metrics of the grain size showed a consistent trend with depth. In addition, the main control factor F1-2 (18.7~86.4 μm) extracted by principal component factor analysis contains the corresponding grain size ranges of component A2 (27.4~35.3 μm) extracted by the grain size-standard deviation method and EM3 (18.7~40.1 μm) extracted by the end-member modelling analysis method ( Figure 9 ).

The above integrated methods coevally indicate that there are obvious differences among the three methods in the process of grain size parameter selection and calculation. However, some of the sensitive grain sizes extracted by them show good compatibility, and the metrics of these grain sizes show the same trends with depth change ( Figure 9 ). In contrast, the relationship between the main control factors F2 and F3 extracted by the principal component factor analysis method and the A0 and A3 components extracted by the grain size-standard deviation method were not well reflected, which was due to the low content of the corresponding grain sizes represented by these sensitive components ( Figures 3 , 4 and 9 ). The change of the grain size content within this range of grain sizes was not obvious, and it is not appropriate to discuss these components.

The trends of the fine-grained environmentally sensitive components extracted by the three methods of grain size-standard deviation method, end-member modelling analysis and principal component analysis are basically the same with depth. The range of the included grain sizes is also highly overlapped, which can reflect the general trend of the summer monsoon in East Asia, among which 5.21-6.72 μm is the common range of grain sizes shown by the three methods. Therefore, the 5.21~6.72 μm component can be used as a proxy indicator for the East Asian summer wind/monsoon ( Figure 9 ; Table 3 ). Similarly, for the East Asian winter monsoon, the trends of the coarse-grained silt environmentally sensitive components extracted by the three methods with depth are also basically the same. The three grain size ranges are also highly overlapped, and have similar trends with the winter monsoon recorded by the Lingtai Loess and the wind-dust accumulation in the northern SCS (Sun et al., 2010). Therefore, the grain size content change of 27.4~35.3 μm silt at the intersection of the three can be used as a proxy indicator of the Quaternary East Asian winter monsoon in the northwestern SCS ( Figure 9 ; Table 3 ).

Comparing the research results on sensitive grain size in different study area, it can be seen that the range of sensitive fractions extracted by the grain size-standard deviation method mainly relies on estimation. Moreover, this method focuses on the overall size of the grains, and may neglect the variability of the individual grains (Wan et al., 2007b, 2010). Grain size-standard deviation (STD) is a method to visualize the degree of dispersion in the volume percent content of different grain sizes (Shen et al., 2015; Teng et al., 2018). Differences in grain size content are reflected by the size of the standard deviation and reveal changes in the associated hydrodynamic strength and depositional environment. However, the method extracts a single sensitive component and lacks an effective method for determining the number of end-member components, which may lead to ignoring inter-individual differences. In contrast, the end-member modelling analysis method (EMMA) is able to extract the sensitive grain size with actual depositional significance, consider all sediment samples comprehensively, and accurately judge the amount of end-member modelling (Dietze et al., 2012; Yu et al., 2016). In addition, Principal Component Factor Analysis (PCA) accurately identifies the correlation between different grain size components and is more statistically significant (Hao et al., 2023).

To determine the grain-size-sensitive components of SCS sediments and the monsoon evolution, a large number of studies have been done by previous researchers, and it is summarized that the fine-grained components in the range of 0-15 μm are usually considered as a valid indicator of the East Asian summer monsoon (Boulay et al., 2007; Li, 2018; Yang et al., 2008; Zheng et al., 2008) whereas the 16-63.5 μm range can generally indicate the East Asian winter wind intensity (Zheng et al., 2008; Boulay et al., 2006; Yang et al., 2007, 2008) ( Table 3 ). However, there are some differences in the range of sensitive grain sizes extracted from different study areas.

Sediment sources in the SCS region are mainly influenced by the East Asian monsoon. An enhanced summer monsoon leads to increased precipitation and riverine input of fine-grained components, while an enhanced winter monsoon is often accompanied by relatively dry and cold climatic conditions, when more coarse-grained silt and dusty materials are transported to the northwestern SCS (Yang et al., 2008). Therefore, the changes in the contents of fine- and coarse-grained silt components can be used as proxy indicators for the evolution of summer and winter monsoon in East Asia, respectively. Previous researchers have done a lot of studies on the size-sensitive components of SCS sediments and the evolution of the East Asian monsoon ( Figure 10 ; Clift, 2006; Sun et al., 2010; Wan et al., 2010), and concluded that the fine-grained components in the range of 0-15 μm are usually considered to be valid indicators of the East Asian summer monsoon, and the coarse-grained material in the range of 15.5-63.5 μm can generally be used as a valid indicator of the East Asian winter monsoon.

In this study, EM1-EM3 are end-member modelling with actual physical significance, and most of the previous interpretations of the land-sourced detritus in the northwestern SCS are the combined results of river transport (summer monsoon) and wind-dust transport (winter monsoon). The fine-grained fraction is generally interpreted as river mud, the intermediate fraction as river silt, and the coarser silt fraction as wind dust (Wang et al., 1999; Wan et al., 2006). Therefore, EM1 and EM2 obtained from this telogenetic analysis are interpreted as riverine input mud and silt; the grain size of EM3 is close to that of the wind dust fraction in the northwestern SCS (Boulay et al., 2007; Li, 2018; Wan et al., 2007b). However, only a very small fraction of the wind dust material would be deposited in the northwestern SCS (Boulay et al., 2007; Wan et al., 2007b). It is hypothesized that EM3 is a relatively distant source of land-sourced debris or a mixed component of wind-dust transported.

In order to better explore the influence of sediment grain size content changes on understanding the East Asian monsoon history, the three principal control factors F1-F3 extracted by principal component factor analysis were combined to form a new grain size metric (GS). The sum of F1-F3 and their squared loadings was calculated with the following equation:

In addition, EM1/(EM2+EM3) was used to represent the percentage of fine-grained material extracted by the end-member component analysis method in the overall sediment. In order to explore the role of these two metrics in indicating the East Asian monsoon, some alternative metrics for the East Asian monsoon were selected and plotted with reference to previous studies on the East Asian monsoon climate ( Figure 10 ; Wan et al., 2006; Clift, 2006; Wan et al., 2007a, 2007b; Sun et al., 2010).

The GS series can represent the intensity of the East Asian monsoon (Yi et al., 2012). When the East Asian summer wind is strengthened, precipitation increases, river action is enhanced, more fine-grained materials are carried by the river into the sedimentation area, and the GS value increases. When the East Asian winter wind is strengthened, rainfall decreases, wind and dust action is enhanced, more coarse-grained silt is carried into the sedimentation area, and the GS value decreases (Hao et al., 2023).

The GS sequence shows the same trends as the EM1/(EM2+EM3) curves. During U4 (>1.8 Ma period), both the GS series and EM1/(EM2+EM3) show a trend of increasing and then decreasing, indicating that the hydrodynamic strength of this period is first enhanced and then weakened. This is consistent with the trend of the fine-grained component content, reflecting that the East Asian summer monsoon are also enhanced and then weakened during this period. This change is consistent with the East Asian winter monsoon climate recorded in the Lingtai Loess on the Loess Plateau (Sun et al., 2010). During U3 (1.3~1.8 Ma period), both GS series and EM1/(EM2+EM3) are further decreased, indicating that the hydrodynamic component weakened during this period, while the winter monsoon strengthened. During U2 (0.8~1.3 Ma period), both the GS series and EM1/(EM2+EM3) gradually decrease, but the overall change is smaller, which indicates that the hydrodynamic conditions weakened and the winter monsoon enhanced (Clift, 2006; Wan et al., 2007a; Clift et al., 2014). During the 0.8~0.9 Ma period, the slope of the GS series and EM1/(EM2+EM3) curves becomes larger, and the winter monsoon is further enhanced and the summer monsoon is weakened. This is related to an increase in the intensity of the global glacial-interglacial cycle and the sea level rise after the Middle Pleistocene climate transition (Clark et al., 2006; Elderfield et al., 2012; Barker et al., 2021). This trend of change is consistent with the trend of the winter monsoon recorded by the Lingtai Loess and wind-dust accumulation in the northern SCS, and summer monsoon recorded by the clay/feldspar in the southwestern SCS and Hm/G (hematite/acicularite) in the northern SCS (Clift, 2006; Wan et al., 2007a).

During U1 (<0.8 Ma period), temperature fluctuations in the northwestern SCS increased in amplitude and the deposition rate of land-sourced detritus was progressively elevated (Jin and Jian, 2008; Tuo, 2008). During this period, the winter monsoon intensified, the climatic conditions were dry and cold (Clark et al., 2006), resulting in an increase of coarse-grained silt in the sediments ( Figures 3 and 4 ). The GS series and EM1/(EM2+EM3) showed a minimum near 0.6 Ma, when the intensity of the winter monsoon in East Asia reached its maximum ( Figures 9D and 10 ). This phenomenon has a similar trend to the winter monsoon recorded by the Lingtai Loess and the wind-dust accumulation in the northern part of the South China Sea (Wan et al., 2006, 2007b).