Passey et al. (1990) proposed a method for calculating the TOC by overlaying a properly scaled porosity curve (generally the sonic transit time curve) over a resistivity curve and baselined them in clay-rich, organics-poor, and fine-grained rock intervals (Liu et al., 2020). The mathematical expressions that were used by Passey for the calculation of ΔlogR summarized in Eqs 1, 2 and widely used for evaluating the unconventional resources (Mahmoud et al., 2017). Δ log ⁡ R = log R / R b + 0.02 Δ t − Δ t b (1)

Where ΔlogR is the curve separation measured in logarithmic resistivity cycles; R is the resistivity measured in Ω·m using the logging tool; Δt is the measured sonic transit time in μs/ft; and Rb is the resistivity corresponding to the Δtb value when the curves are baselined in non-source,clay-rich rocks. The LOM can be obtained from core sample analysis, such as the vitrinite refiectance or Tmax (Passeyet al., 1990).

The ΔlogR separation in the source rock and non-source rock intervals can be distinguished using the natural gamma-ray curve (Passeyet al., 1990). For clay-rich rocks, they proposed an empiricalequation for calculating the TOC from ΔlogR: T O C = 10 2.297 − 0.1688 × L O M × Δ log ⁡ R (2) where TOC is the total organic carbon content measured in wt%; and LOM is the maturity.

The V-△logR technique was proposed and improved by Liu et al. (2011b); Liu et al. (2014); Liu et al. (2019). This model includes three revisions to the ΔlogR model. 1) Replace the fixed overlay-coeffcient (0.02) with a variable one, and determine it according to the actual geological data (TOC and well logs). 2) Remove the maturity (LOM) from the model. 3) Determine the baseline from the ΔlogRV curve, instead of determining it from the sonic and resistivity curves. Δ log ⁡ R V = k × log ⁡ R + 1 − k × P (3) T O C = α × Δ log ⁡ R V − b a s e l i n e + b (4)

Here, ΔlogRv is the curve separation measured in logarithmic resistivity cycles; R is the deep lateral resistivity (Ω·m); p is the sonic transit time, bulk density, or neutron log. k is a variable that ranges from 0 to 1; baseline is the baseline value of the ΔlogRv curve; b is the background value of the TOC content; and α is a coeffcient calculated using a mathematical tool.

The Songliao basin is located in the northeastern part of China (Figure 1A), and the Shuangcheng area is located in the eastern fault depression zone in the northern part of the Songliao Basin (Figure 1B). The study area described in terms of structural elements in the Shuangcheng sag which is sub-element within the Shuangcheng area. The strata with clastic rocks in the Shuangcheng sag include the Yingcheng Formation and the Denglouku Formation, The fourth member of the Yingcheng Formation develops black shale considered to be rich potential source rocks for hydrocarbons (Figure 2). Since 2017, a suite of organic-rich shale has been discovered in the southern sag of the Shuangcheng region (Huang et al., 2015; Yin et al., 2018). This set of shale has strong hydrocarbon-generating capacity, because it was discovered in the sandstone of the Denglouku Formation, and the shale is oil, and the natural productivity of a single well can reach up to 91 t/d, The shale rich in organic matter is developed in the fourth member of the Yingcheng Formation. The shale is deposited in the freshwater environment of continental semi-deep lake-deep lake, with a thickness of 20 m–120 m. Previously, this set of shale was mainly studied as a source rock, and it was not used as a shale oil exploration object until 2019.

The total organic carbon, Rock eval pyrolysis and vitrinite reflectance (Ro) data involved in this study were collected from Key Research Laboratory on Tight Sandstone and Shale Oil Reservoir Forming. The data were obtained from core samples of 11 Wells in the Shuangcheng area which are used to determine the origin, content, and maturity level of the organic matter present in a shale. The depth of the data ranges from 1,241.00 m to 1796.35 m, and the well locations are shown in Figure 3C. In order to avoid the volatilization of light hydrocarbons in shale, Rock eval pyrolysis and TOC tests of core samples were carried out soon after drilling and coring. Analysis data from two typical exploration wells were shown on Supplementary Table S1. These data revealed that the TOC of shale in Shuangcheng area ranged from 0.78% to 6.38%, with an average of 2.86% (Figure 4A). S1 ranged from 0.05 mg/g-1.65 mg/g (average 0.58 mg/g), and S2 ranged from 0.36mg/g-16.36 mg/g (average 6.35 mg/g). The measured Ro is between 0.82% and 1.31%, indicating that shale has entered a mature evolution stage, and Ro increases regularly with the increase of depth (Figure 4C). Figure 5 shows geochemical analysis data and conventional logging curves of typical exploration wells in Shuangcheng area. Shale segments correspond to higher gamma values (typically greater than 140API), which can be used to distinguish shale from non-shale.

The acoustic time of shale is relatively high, mainly between 70 μs/ft-100 μs/ft. The density of shale is mainly between 2.46 g/cm³ and 2.52 g/cm³, which is slightly less than that of sandstone. However, the density curve was greatly affected by hole collapse (e.g., 1,366 m–1378 m), so it was not used to predict the total organic carbon. The deep lateral resistivity of shale is lower than that of sandstone, but the resistivity of shale does not show the obvious positive anomaly as Marine shale, and even shows a slight negative anomaly, even when the shale is mature and contains liquid hydrocarbons (e.g., 1,460 m–1480 m chloroform bitumen A 0.049%–0.361%).

Two typical wells (well S59 and well S70) in the targeted area with core TOC measurements were randomly selected to establish and vertify the prediction model of TOC, and the remaining 9 wells were used for model validation. The key point to establish the model is to obtain the parameters of the V-ΔlogR methods described in process 3.3. Therefore, we determine the parameters in Eq. 3 of the V-ΔlogR model separately. Combined with the TOC evaluation standard of Lacustrine shale (SY/T 5735-1995), the value b should not exceed the lower limit of organic matter abundance of source rock (< 0.4%). As the value b has little influence on the prediction accuracy of TOC of shale. We mainly discuss the coefficients k, baseline and α in the V-ΔlogR model.

In the model validation process, coefficient A was calculated using the value of the adjacent well, while K was calculated using the logging curve of the well (Supplementary Table S2). The prediction model was applied to nine retention wells. The model of well S59 was applied to well Sx67, S71, and Sx74 in the same eastern slope zone. The prediction model established in S70 well in the western slope zone was applied to S66 well, S661 well, S68 well, S72 well, S76 well and S83 well, which also belong to the western slope zone. The cross plot of TOC calculated by V-△logR model and TOC predicted by core for all Wells in the study area shows that TOC calculated by model and TOC predicted by core are distributed around Y=X, with a determination coefficient of 0.83 and an average relative error of 11.06% (Figure 7).

In Supplementary Table S2, the determination coefficients of the predicted TOC and core TOC of the single well model ranged from 0.57 to 0.93, and the relative average errors ranged from 7.13% to 25.31%. In addition to the S74 well prediction error is relatively large (R2 = 0.57 and MRE = 25.31%), the overall anastomosis effect is also good. In general, the established prediction model can accurately predict the TOC of surrounding wells, and the TOC of prediction model is consistent with the TOC of core in well profile.

Shale grading evaluation usually considers reservoir oil-bearing property evaluation and recoverable factors (Lu et al., 2012). Since there is little difference in the content of mineral components in shales in the study area, this shale grading evaluation is mainly carried out based on the difference in oil content of shales. Rock pyrolysis index S1 represents the amount of hydrocarbons that have been generated and remain in the shale. Although S1 lacks heavy components and some light components are lost during sample collection, it is still one of the more intuitive and reliable indicators to characterize shale oil content (Lu et al., 2012; Li et al., 2015). Given the difficulty of accurately predicting S1 from well logging data, a feasible approach is to establish relationships with more readily available indicators (including TOC, organic matter type and Ro) that are closely related to S1, and then use these easily available indicators to characterize shale oil content and to carry out shale grading evaluation. For example, Lu et al. (2012) carried out shale grading evaluation through the relationship between TOC and S1.

The shale maturity parameter (Ro) was also used in grading evaluation in addition to the relationship between S1 and TOC. By discussing the relationship between S1 and TOC and Ro, shale in Yingcheng Formation can be divided into three categories. Type I (TOC > 3.5%, Ro > 0.9%) shale is characterized by a slight increase in S1 or a steady increase in TOC (Figure 8A), and most samples have high S1 values (Y is generally greater than 0.7 mg/g). Samples in this interval have high organic matter abundance and hydrocarbon generation ability. When TOC is greater than 3.5%, S1/TOC shows a downward trend (Figure 7B), indicating that the hydrocarbon generation content in these shale has exceeded the adsorption capacity, and the excess hydrocarbon is expelled (Li et al., 2019), which also indicates that the oil has relatively good mobility. In Figure 8A, some samples with TOC greater than 3.5% and still with low S1 values were expelled from type I shale. Because of their low maturity (Ro is typically less than 0.9%), they do not have sufficient hydrocarbon and have not yet reached the hydrocarbon expulsion threshold (Ro=0.9%) shown in Figures 8C, D. The S1 of type II shale (TOC between 2% and 3.5%) increases rapidly with the increase of TOC, indicating that the oil content (hydrocarbon generation capacity) of shale increases with the increase of TOC. Class II shale also include shale with TOC > 3.5 and Ro less than 0.9. The saturation degree of adsorbed hydrocarbon of type II shale should be lower than that of type I shale and the mobility should be lower than that of type I shale. The S1 of Class III (TOC < 2.0%) shale increases slowly with the increase of TOC, and shalewith different evolution degrees have lower S1 values, mainly because they have lower organic matter abundance and hydrocarbon generation capacity.

Based on the calculation of TOC of 11 wells in the study area, combined with the shale grading limits in Figure 7A, the distribution of different grades of shale is identified on the well profile, and the thickness is calculated. As shown in the typical interwell profile in Figure 9, the development frequency of type I shale is relatively low, and the top, middle and bottom of the fourth member of the Yingcheng Formation are all developed. The continuous thickness of single layer is generally no more than 5 m, and the thickness ratio revealed by drilling (cumulative thickness of type I shale/total thickness of shale) is generally 10%–20%. The thickness of class II shale monolayer is generally 10 m–20 m, and the cumulative thickness of a single well is 40%–60%. The monolayer thickness of shale of Class III is generally less than 5 m, and the cumulative thickness of a single well is generally 20%–30%.

By statistical analysis of the relationship between the thickness of different grades of shale and the total thickness of shale, we established a method conducive to predict the thickness of different grades of shale. As shown in Figure 10, the determination coefficient between the thickness of different grades of shale and the total shale thickness exceeds 0.91, indicating that the thickness of different grades of shale can be predicted by the total shale thickness. The quantitative relationship between the two was established based on the zonal zone. For areas lacking drilling, the thickness of shale in each grade was predicted based on the total shale thickness predicted by the combination of well vibration (Sun et al., 2019) (Figure 11). The results show that the thickness of class I shale is generally 5 m–15 m, class II shale is generally 20 m–80 m, and class III shale is generally 10 m–40 m.