Five core samples of mudstone have been collected from potential Eocene source rocks, including the LS-1, LS-2, and LS-3 members from the WZ22-3-1, WZ10-7-1, and WZ10-3-30 wells in the study area (Table 1). In addition, four oil samples collected from the WS22-9-3D, WS16-1-HD, and WS16-1W-7 wells have been analyzed using gas chromatography-mass spectrometry. Additionally, twenty-two published samples, including source rocks and oils, have been used for chemometric analysis (Huang et al., 2017), and the selected parameters are present in Table 2.

The core samples were cleaned first with distilled water and then crushed into powder. The powdered samples were subjected to IFP Rock-Eval 6 analyzer to obtain geochemical parameters such as total organic matter (TOC), hydrogen index (HI) and hydrocarbon generation potential (S₁ + S₂), following the same protocols described previously (Wang et al., 2018b). Initially, power samples were heated for 300°C for 3 min to yield the Rock-Eval S₁ that stands for the amount of free hydrocarbon, and subsequently heated to 650°C at a heating rate of 25°C/min to obtain the Rock-Eval S₂ that stands for the hydrocarbons generated from pyrolytic degradation of the kerogen. T_max is the temperature at which maximum kerogen pyrolysis occurs. Finally, the powders were further heated from 300 to 850°C at a rate of 20°C/min to produce the residual organic and inorganic carbon content. ∼20 g of dry rock powder were extracted with 9:1 of Dichloromethane: Methanol (DCM: MeOH) through Accelerated Solvent Extraction. All extracts and crude oils were separated into saturated, aromatic, and resin fractions using silica gel column chromatography.

Gas chromatography-mass spectrometry (GC-MS) analysis of oils and source rock extracts were analyzed using a Thermo Fisher Trace 1,300 gas chromatography coupled to an ISQ 7000 mass spectrometer equipped with a DB-5 capillary column (30 m × 0.25 mm × 0.25 μm). Pure helium gas (99.999%) was used as the carrier gas. The GC oven temperature was initially held at 50°C for 2 min, programmed to 120°C at a rate of 20°C/min, then to 250°C at 4°C/min and to 310°C at 3°C/min, and held at 310°C for 30 min. The mass spectrometer ionization energy was 70 eV and the ion source temperature was 230°C. Full-scan and selected ion recording were used for analysis, with the full-scan detection range of 50–550 (m/z). The peak area was integrated by Xcalibur software (Thermo Scientific). The biomarker compounds were identified in the studied rock and oil samples in accordance with the relative retention time, mass spectral feature, peak sequence, and identification standards from prior studies (Huang et al., 2017; Wang et al., 2018a; Gan et al., 2019).

In this paper, three chemometric methods (e.g., HCA, PCA and MDS) were applied to yield a detailed oil-oil and oil-source rock correlations in the Wushi Sag using eight biomarker ratios (labelled by “#” in Table 2), similar to previous studies (Peters et al., 2013; Wang et al., 2018a; Wang et al., 2018b). The selected biomarker parameters, including Pr/Ph, Ts/Tm, Ol/C₃₀H, Ga/C₃₀ hopane, C₂₇(%), C₂₈(%), C₂₉(%), and C₃₀-4MSI index (Table 2), are generally suffered less from secondary processes, such as biodegradation, maturity, and migration. The HCA and PCA were performed using Pirouette® software (Infometrix, Inc.), whereas MDS was computed through in-house software (Wang et al., 2016).

Three sets of source rocks are present in the Wushi Sag: the first, second, and third members of the Liushagang Formation, i.e., LS-1, LS-2, and LS-3 (Liu et al., 2018; Li et al., 2021). The detailed source rock geochemistry can be referred to as Huang et al. (2017) and Li et al. (2021). In this paper, five-rock samples have also been analyzed using Rock-Eval pyrolysis (Table 1), in addition to published data collected from an open-source database. As depicted in Figure 3A, the studied rock sample from the LS-1 member is a fair source rock, whereas rock samples from the LS-2 and LS-3 members show very good to excellent hydrocarbon generation potential. A plot of hydrogen index versus T_max for the LS-1 samples demonstrates that the organic matter of studied source rocks is mainly Type II₂ kerogen, whereas the LS-2 and LS-3 samples primarily contain Type II₁ and Type I kerogens, respectively (Figure 3B).

The thermal maturity of the studied rock samples was evaluated based on Rock-Eval T_max and biomarker parameters. T_max values for the Wushi samples range from 429 to 434°C, which indicates immature to early mature organic matter (Tissot and Welte, 1984). This is also supported by the maturity-related biomarker ratios. The ratios of C₃₂ 22S/(22S + 22R) homohopanes and C₂₉ 20S/(20R + 20S) steranes of the studied samples are 0.44–0.58 and 0.17–0.30, respectively (Table 2), suggesting that the samples have entered the oil generation stage (Seifert and Moldowan, 1980; 1986).

The distribution characteristics of n-alkanes and isoprenoid hydrocarbons in the Liushagang Formation can be used to interpret sources and depositional conditions of organic matter. The distribution of n-alkanes for the studied mudstone samples typically ranges from n-C₁₅ to n-C₃₁ (Figure 4). All three samples have bimodal distribution, suggesting a mixture of one mature source and one immature source in different proportions. The pristane/phytane (Pr/Ph) ratio is commonly used as an index of the depositional environment (Peters et al., 2005). Generally, low Pr/Ph ratios (<1) indicate the anoxic condition (Didyk et al., 1978), medium Pr/Ph ratios varied between 1 and 3 are likely related to suboxic conditions (Hunt, 1995), and high Pr/Ph ratios (>3) suggest oxic condition (Hunt, 1995; Harris et al., 2004). The Pr/Ph ratios of the Liushagang mudstone samples are in the range of 0.44–2.68, suggesting that samples from the Liushagang Formation may be deposited under anoxic to suboxic conditions.

The representative sterane chromatograms (m/z 217) for studied rock samples are illustrated in Figure 4. The relative content of C₂₇, C₂₈, and C₂₉ ααα 20R steranes of the studied samples are in the range of 11–35%, 14–17%, and 49–75%, respectively (Table 2). This high C₂₉ steranes contents may imply a strong contribution of terrigenous organic matter input (Huang and Meinschein, 1979). In addition, the relative abundance of C₂₇ steranes for samples of the LS-1 member is lower than those of the LS-2 and LS-3 members, suggesting they may be less algal organic matter input to LS-1 member compared to the LS-2 and LS-3 members. This is also confirmed by the C₃₀ 4-methylsteranes to C₂₉ regular steranes ratios (C₃₀-4MSI index) that is widely used as an indicator for certain dinoflagellates blooming in freshwater lakes (Brassell et al., 1988; Goodwin et al., 1988; Peters et al., 2005). The C₃₀-4MSI ratios of samples from the LS-1 member range from 0.40 to 0.53, whereas those of the LS-2 and LS-3 samples range from 0.66 to 3.03. These results may suggest a higher contribution of dinoflagellates to the organic matter in the deposition of the LS-2 and LS-3 members than that to the LS-1 member. This is also consistent with previous studies (Huang et al., 2013; Huang et al., 2017).

Terpane mass chromatograms (m/z 191) of the representative source rocks are characterized by a dominant C₃₀ hopane with a higher abundance of pentacyclic terpanes relative to tricyclic terpanes (Figure 4). The C₃₅/C₃₄ ratios of the studied source rocks range from 0.43-0.84, probably suggesting anoxic to suboxic depositional conditions (Peters and Moldowan, 1991). A high Ga/C₃₀ hopane ratio is commonly associated with a stratified and reducing water column (Fu et al., 1986; Sinninghe Damsté et al., 1995). Ga/C₃₀ hopane ratios of the samples are very low, ranging from 0.04-0.10, indicating a lack of water column stratification.

A complete set of acyclic isoprenoids (e.g., Pr and Ph) and low molecular weight n-alkanes are present in the analyzed oils, probably indicating no distinct signs of biodegradation (Peters and Moldowan, 1993). Additionally, low Pr/n-C₁₇ and Ph/n-C₁₈ ratios of all oils also show that the studied oil samples were not biodegraded (Volkman et al., 1983). The mass chromatogram of the representative oil sample is shown in Figure 5A. The Pr/Ph ratios of the studied oils range from 1.17 to 1.46, likely suggesting suboxic conditions (Hunt, 1995).

The m/z 217 mass chromatogram for the representative oil is shown in Figure 5B. The relative contents of C₂₇, C₂₈, and C₂₉ ααα 20R steranes for the oils range from 31–37%, 13–16%, and 46–56%, respectively (Table 2). These data may indicate a substantial contribution of terrigenous organic matter input (Huang and Meinschein, 1979).

The m/z 191 mass chromatogram for the representative oil is shown in Figure 5C. These studied oils have C₃₅/C₃₄ ratios ranging from 0.45 to 0.53, indicating that oxidation is mainly occurring on the source (Peters and Moldowan, 1991). The Ga/C₃₀ hopane ratios of the oil samples are in the range of 0.11–0.14, probably indicating a lack of water column stratification (Sinninghe Damsté et al., 1995).

Hierarchical cluster analysis (HCA) and principal component analysis (PCA) are valuable tools to identify genetic relationships among crude oils and source rocks based on age- and source-related biomarker ratios (Peters et al., 2007; Peters et al., 2013; Wang et al., 2018b). Similarly, multidimensional scaling (MDS) is also a reliable method for a detailed oil-oil and oil-source rock correlation (Wang et al., 2016; Wang et al., 2018a; Wang et al., 2020a). The Wushi oil samples were divided into two groups, as shown in Figure 6.

Group A oils are primarily produced from the LS-3 reservoirs (Table 2). This group of oils have relatively high Pr/Ph ratios, ranging from 2.12-2.54, and low Ga/C₃₀ hopane ratios of 0.03–0.09, indicating the source rocks of Group B oils were probably deposited in more oxic condition (Table 2). The relative abundance of C₂₉ ααα (20R) steranes and C₃₀-4MSI ratios are in the range of 45–56% and 2.33–2.90%, respectively (Table 2). These data may indicate source rocks of Group B oils may have received more contribution of marine organic matter when compared to the source rocks of Group B oils. The ratios of C₂₉ 20S/(20S + 20R) range from 0.41 to 0.49 (Table 2), indicating that Group A oils are mature and have relatively lower maturity than that of Group B oils.

Group B oils principally consist of oils from the LS-1 member. The Pr/Ph ratios are relatively low, ranging from 1.16 to 1.46 (Table 2) and suggesting that the source rocks of Group A oils were deposited under more anoxic depositional environments. Group A oils have a relatively low abundance of C₂₇ ααα (20R) steranes and a high abundance of C₂₉ ααα (20R) steranes. Relatively low C₃₀-4MSI ratios range from 0.50–0.79 (Table 2), probably indicating that the source rocks may have received more contributions from terrigenous organic matter. The maturity parameter of C₂₉ 20S/(20S + 20R) has relatively high values, ranging from 0.53–0.64 (Table 2), which is indicative of relatively high maturity.

Similarly, we have also identified genetic affinities between oils and source rocks using chemometric methods. Figures 7, 8 show the PCA and MDS scenario of multi-parameter oil-source rock correlation for the Wushi Sag. The biomarker parameters used for chemometric methods are the same as those in the HCA and PCA of Figure 6, marked by stars in Table 2. The three-dimensional view of PCA indicates two genetic oil families (Figure 7), inconsistent with the results discussed above. Group A oils have a strong affinity to the LS-2 and LS-3 members. The LS-2 mudstone member is related to group B oils. This conclusion is also confirmed by the MDS plot (Figure 8).

Oil and gas exploration activities have been primarily in oils derived from the second member of the Liushagang Formation (LS-2) and structure traps in the Wushi Sag over the last three decades (Liu et al., 2008). Recently discovered oil-bearing structures have stimulated a strong interest in the central of the sag (Yang et al., 2016). Future exploration and understanding of the petroleum system can be significantly improved by investigation of the genetic relationships between the discovered crude oils and source rocks.

As described previously, two categories of oils at different maturity levels have been identified using chemometric methods. Group A oils represent mixtures generated from the LS-2 and LS-3 members, whereas group B oils were derived from the LS-3 member. Group A oils with relatively lower maturation were distributed in eastern Wushi Sag (Figure 9) and primarily produced from the LS-3 reservoir of the WS162- and WS172- wells (Table 1). The corresponding maturity of Group A oils is consistent with the maturity of the nearby source rocks in the study area (Figure 9). The group of oil is close to the WS172 N Fault. The fault was in a shallower depth and had relatively weak activities during the Paleogene period (Yuan et al., 2020). Furthermore, the combination of oil-source rock correlation results and low permeability reservoirs (Gao et al., 2019) in the study area probably suggests that the migration and accumulation of Group A oils mainly occurred within the LS-2 and LS-3 members, forming a self-generation and self-storage petroleum reservoir.

Group B oils with relatively high maturation are distributed in southeast Wushi Sag and are mainly produced from the shallower LS-1 reservoir of the WS22-9 and WS16-1 wells (Table 1). The corresponding maturity of Group B oils is essentially equivalent to the maturity of the nearby source rocks from the LS-2 and LS-3 members (Figures 9, 10). Similarly, the distribution of oil is in the vicinity of the WS16-1 Fault, which holds across the entire source rock members of the study area (Figure 11) and is a good pathway connecting oil and source (Yuan et al., 2020). This evidence implies that group B oils were probably migrated vertically through faults, contributing to the shallower distribution of crude oil.

In summary, the results in the present study provide some inspiration for future petroleum exploration. For example, the oil-source rock correlation results suggest that the contribution of the LS-3 member to the Wushi oils has been underestimated in the past, and thus greater emphasis should be deserved in future exploration. In addition, exploration of the eastern Wushi Sag, where close to the distribution of Group A oils, should probably be focused on the relatively deep members of the Liushagang Formation, considering a combination of fault features and reservoir characteristics. In contrast, the exploration of southeast Wushi Sag, wherein the proximity of Group B oils should be paid more attention to the shallow layer of the Liushagang Formation.