Analysis of Enrichment-Controlling Factors and Seismic Prediction of Gas-Rich Zones in Tight Sandstone Gas Reservoirs: A Case Study from the Second Member of the Qianfoya Formation in Bazhong Area, Sichuan Basin

Abstract

The Qianfoya Formation in the Bazhong area of the Sichuan Basin comprises a typical tight sandstone formation characterized by low porosity, low permeability, and strong reservoir heterogeneity. Effective reservoirs in this region are relatively thin, and seismic responses between gas-bearing layers and dry layers differ only slightly. These geological complexities create substantial challenges for predicting tight sandstone reservoirs and gas-rich zones, which necessitate targeted and effective geological and geophysical prediction research. Therefore, taking the Second Member of the Qianfoya Formation (Qian-2 Member) as the research object, this study utilizes multiwell logging, mud logging, and 3D seismic data sets. We first analyze the geological characteristics of the target interval to clarify the primary controlling factors of tight sandstone gas reservoir enrichment. Based on these factors, we establish an effective geophysical prediction workflow tailored for gas-bearing tight sandstone reservoirs. This method is further applied to predict gas-rich zones within the Qian-2 Member. The study concludes that (1) the main controlling factors for tight gas reservoir enrichment in the Qian-2 Member are sedimentary facies distribution, reservoir-cap rock conditions, and fracture distribution. (2) Prestack simultaneous inversion integrated with rock physics analysis enables effective sand-body identification. Rock physics analysis identifies P-wave impedance (Zp) and S-wave impedance (Zs) as the sensitive parameter pair for sand bodies. A sand-body factor prediction model is then constructed via coordinate rotation. Sensitive parameters are derived from prestack simultaneous inversion, and the sand-body factor is calculated from the prediction model. High positive values of the sand-body factor are used to predict sand-body distribution. (3) Based on near- and far-offset seismic comparison, the dominant amplitude–frequency method is used to extract the gas-bearing prediction attributes, whose positive anomalies effectively identify gas-bearing intervals. (4) A multifactor prediction framework combining sand bodies, gas-bearing properties, and fractures is adopted for gas-rich zones prediction. The results show that gas-rich zones in the Qian-2 Member are concentrated in the middle and upper submembers, and the predicted favorable zones are mainly distributed around faults. (5) This study establishes a workflow for predicting gas-rich zones of tight sandstone gas reservoirs via multifactor geological superposition analysis. Well-seismic calibration verifies its applicability, providing a methodological foundation and practical reference for the comprehensive prediction of favorable targets in similar tight sandstone gas reservoirs.

1. Introduction

Tight sandstone gas, a prevalent unconventional hydrocarbon resource, is widely distributed globally and has substantial reserves. − With the continuous advancement of global oil and gas exploration, tight sandstone gas reservoirs have increasingly become a critical sector driving the growth of unconventional hydrocarbon production. , According to international standards, tight sandstone gas is defined as sandstone reservoirs with porosity less than 10% and permeability lower than 0.1 × 10^–3 μm². , Tight gas reservoirs are extensively developed in petroliferous basins worldwide. Statistics indicate that approximately 70 petroliferous basins globally host tight sandstone gas accumulations, with the total tight gas resource volume reaching 210 × 10¹² m³. , The combined tight gas resources of the Asia-Pacific, North American, and Latin American regions account for more than 60% of the global total. − In recent years, the exploration and development of tight sandstone gas reservoirs have achieved remarkable progress in China, with large-scale production obtained in the Sichuan, Ordos, and Songliao Basins. − The rapid growth in reserves and production of tight gas reservoirs has become a “highlight type” in China’s efforts to increase oil and gas reserves and production. , Tight sandstone gas reservoirs are characterized by thin single-layer thickness of effective reservoirs, strong reservoir heterogeneity, significant vertical and lateral variations in physical properties, and complex gas–water relationships. These features bring great challenges to the geophysical prediction of effective reservoir distribution and gas-bearing distribution. − Therefore, developing suitable geophysical exploration techniques tailored to the characteristics of tight sandstone gas reservoirs is of great theoretical significance and practical value in improving the efficiency of exploration for favorable targets.

Domestic and foreign scholars have conducted extensive research on prediction methods for tight sandstone reservoirs, with these methods continuously updated and improved in recent years. In terms of rock physics analysis, Tenchov et al. conducted a dispersive rock physics analysis of tight sandstone using high-temperature and high-pressure core experiments, emphasizing that fluid identification relying solely on elastic impedance is susceptible to lithological interference. Introducing frequency dimension constraints can help alleviate the ambiguity associated with fluid prediction. Pang et al. constructed lithology indicator factors based on P-wave impedance and the P-to-S-wave velocity ratio, effectively addressing the challenge of identifying deep-water fan reservoirs amid interference from calcareous sandstone. In terms of inversion methods, the Bayesian inversion framework combines prior information with the likelihood function to produce relatively stable estimates of elastic parameter volumes. , Buland et al. applied Bayesian linear inversion to AVO analysis and provided a probabilistic interpretation scheme for lithology and fluid prediction. In recent years, machine learning methods have attracted widespread attention for seismic lithofacies prediction. Algorithms such as support vector machines, random forests, and deep neural networks have been applied to lithology classification and gas-bearing identification via multiattribute fusion. Das et al. demonstrated the nonlinear mapping capability by using convolutional neural networks to predict petrophysical properties from prestack seismic data. In anisotropy research, Gong et al. derived the mapping relationship between reservoir characteristic parameters and anisotropic parameters via rock physics analysis and established an elastic impedance inversion workflow for anisotropic parameters targeting tight sandstone reservoirs. These methodological advances provide technical support for the effective prediction of tight sandstone reservoir targets.

The Sichuan Basin represents a principal petroliferous province within China. , Current exploration and production efforts targeting tight sandstone gas reservoirs in this basin predominantly focus on the Upper Triassic Xujiahe Formation and the Middle Jurassic Shaximiao Formation. − By contrast, studies on the Middle Jurassic Qianfoya Formation primarily focus on analyses of sedimentary and reservoir characteristics, hydrocarbon enrichment laws, exploration potential, and geophysical prediction. Cheng et al. calibrated well-logging data against core samples. For reservoirs with different lithologies and types, they constructed a porosity calculation model based on a petrophysical volume model and multiparameter mathematical statistical analysis. They also established a permeability prediction model constrained by core-derived porosity–permeability correlations. These models significantly enhance the accuracy of porosity and permeability calculations and effectively address challenges in reservoir parameter estimation for the Qianfoya Formation. Shen et al. developed a hydrocarbon indicator via AVO technology to predict the spatial distribution of gas-bearing intervals in the target formation and further analyzed the correlation between fracture density and gas enrichment, examining both sectional and planar perspectives. Cheng and Peng integrated sedimentological, geochemical, and petrophysical analyses to systematically reveal the coupled controlling mechanisms of continental shale reservoirs in the Qianfoya Formation. Their research concluded that sedimentary facies dominate organic matter accumulation, lithofacies govern pore development, and diagenesis controls pore preservation. Jiang et al. analyzed the influence of fractures on hydrocarbon enrichment and preservation according to the development and distribution characteristics of natural fractures in the Qianfoya Formation. Nevertheless, existing research on gas-rich zones in tight sandstones of the Qianfoya Formation remains insufficient. Therefore, it is necessary to develop targeted prediction approaches adapted to the local geological conditions of the Bazhong area to delineate favorable gas-bearing zones. In the aforementioned studies, analyses and discussions are mostly carried out for a single research domain or a single prediction method. Systematic analytical ideas that integrate geology and geophysics and implement hierarchical target prediction based on multiple factors are rarely considered. This points out the direction and approach for the present study.

Taking the Qian-2 Member in the Bazhong area of the Sichuan Basin as the study area, this paper adopts core, logging, and seismic data. Based on the identification of the main controlling factors for gas reservoir enrichment in the Qian-2 Member, this paper proposes a comprehensive prediction method for gas-rich targets in tight sandstone gas reservoirs using multifactor analysis. The core of this method lies in three key controlling enrichment factors: sedimentary sand bodies, gas-bearing potential, and fractures. It follows a hierarchical, comprehensive prediction workflow: first, identify sand bodies; then, detect gas-bearing layers within sand bodies; and finally, incorporate the influence of fractures. The gas-rich target zones are delineated by intersecting and superimposing attribute characteristics of sedimentary sand bodies, gas-bearing potential, and fracture distributions. This study provides a methodological foundation and practical reference for comprehensively predicting favorable targets in tight sandstone gas reservoirs in typical regions.

2. Geological Setting

The Sichuan Basin, a large superimposed petroliferous basin situated on the Yangtze Plate in southwest China, − has experienced multiple evolutionary stages, including the Early Paleozoic cratonic depression, Late Paleozoic cratonic rift, and Meso-Cenozoic foreland depression. Geographically, the study area lies in the eastern part of the Bazhong area in northeastern Sichuan, covering an area of approximately 406 km², with six wells penetrating the target strata within the survey extent. Tectonically, it is located on the gentle fold belt of northern Sichuan, bounded to the north by the Micang-Dabashan foreland thrust belt. , More specifically, the study area occupies the southern segment of the Tongnanba Anticline Belt, − a transitional zone between the Tongjiang Sag and the gentle structural belt of central Sichuan (Figure a).

1.

(a) Tectonic unit division of the northern Sichuan Basin and location of the Bazhong area (adapted with permission from ref . Copyright 2015 Elsevier). (b) Stratigraphic column of the Jurassic strata within the study area (adapted with permission from ref . Copyright 2022 Frontiers).

Two major fault sets striking NW and nearly SN are extensively developed within the continental strata of the Bazhong area. Controlled by compressional tectonics from the Dabashan Orogen, these faults penetrate downward into the gypsum-salt rocks of the fourth and fifth members of the Jialingjiang Formation and disappear upward within the lower Shaximiao Formation. Faults developed above and below the Qianfoya Formation across the study area and its surrounding regions dominantly trend NNW. , These faults are predominantly reverse faults characterized by steep dips, extensive distribution, and small vertical throws, with limited lateral extents on the plane view. Their strikes align with the axes of regional folds, and numerous small anticlines and fault noses have formed along the fault trace.

The Qianfoya Formation belongs to the Middle Jurassic series and conformably overlies the Ziliujing Formation, with its top boundary in conformable contact with the overlying Lower Shaximiao Formation. Its lithology is dominated by gray and light gray fine sandstones and light gray siltstone, subordinately interbedded with light gray argillaceous siltstone (Figure b). The top and bottom intervals of the formation are mainly composed of gray and dark-gray shale. Vertically, the Qianfoya Formation is subdivided into three members from base to top: Qian-1 Member, Qian-2 Member, and Qian-3 Member (also referred to as J₂q₁, J₂q₂, and J₂q₃). The Qian-2 Member is further divided into three submembers, namely, the lower, middle, and upper submembers (also referred to as J₂q₂1, J₂J₂ ², and J₂J₂ ³). This study focuses on the Qian-2 member, which has an average stratum thickness of 93.5 m. This interval features a large cumulative sand thickness and favorable reservoir physical properties, making it the primary exploration target within the Qianfoya Formation in the study area.

3. Data Sets and Methods

3.1. Test Analysis and Geophysical Data

3.1.1. Well-Logging Data

Six wells were used in the study area. The well data include logging curves, logging interpretations, mud logging data, and comprehensive stratigraphic columns. The sampling interval is 0.125 m. Before application, well-logging correction and preprocessing are performed based on a classified statistical analysis. These corrections contain depth correction, borehole correction, multiwell consistency analysis, and standardization correction. Well-logging data are mainly used for petrophysical analysis and well-seismic calibration.

3.1.2. Seismic Data

Prestack common reflection point (CRP) gathers and poststack seismic profiles are available within the study area. The acquisition azimuth of the survey is due north, covering an area of approximately 406 km², with a bin size of 25 m × 25 m and a sampling interval of 2 ms. Both prestack and poststack data have a dominant frequency of around 42 Hz. The signal-to-noise ratio of poststack data ranges from 17 to 21 dB, whereas that of the prestack CRP gathers varies from 6 to 9 dB. Therefore, the prestack CRP gathers require processing optimization, including automatic gain control, band-pass filtering, common-offset stacking, and offset-to-angle conversion. Angle gathers are obtained and can be applied to prestack simultaneous inversion and stacking across different angle ranges for attribute extraction. The optimized angle gathers achieve a signal-to-noise ratio of approximately 14–18 dB while maintaining the original dominant frequency.

3.2. Geophysical Reservoir Characteristic Methods

Basis on an analysis of the enrichment-controlling factors of tight sandstone reservoirs, a comprehensive geophysical prediction was conducted for the gas accumulation elements and the distribution of gas-rich zones in the target reservoirs. This included rock physics analysis, well-seismic response characteristic analysis, prestack simultaneous inversion, and extraction and interpretation of seismic attributes, forming a set of prediction methods and processes for gas-rich zones in tight sandstone reservoirs (Figure ).

2.

Flowchart of the prediction workflow for gas-rich zones in tight sandstone reservoirs.

3.2.1. Rock Physics Analysis

Two elastic parameters are chosen for the cross plots. By analyzing the sample distribution on cross plots, we evaluated the discriminative capability for sandstones and mudstones. The optimal elastic parameter pair for sandstone identification is determined, and a relevant cross-plot template is generated. The sandstone–mudstone boundary is then extracted from the template, and a prediction model for the sand-body indicator factor is built via the coordinate rotation method.

3.2.2. Prestack Simultaneous Inversion

Based on prestack angle-gather data, prestack simultaneous inversion is implemented after constructing the initial model, extracting angle-dependent wavelets, and performing a multiwell mudstone background-trend analysis to obtain elastic parameter volumes that are sensitive to sandstones.

3.2.3. Sand-Body Distribution Prediction

Two sandstone-sensitive elastic parameters are fused into a single sand-body indicator factor attribute by using the sand-body indicator factor prediction model. This attribute is used to predict the spatial distribution of the sand bodies.

3.2.4. Gas-Bearing Distribution Prediction

By comparing and analyzing the spectral characteristics and near- and far-offset profile features of reservoirs with different fluid contents, an identification pattern for gas-bearing layers and dry layers is established. On this basis, near- and far-offset seismic data are subjected to frequency-division processing and analysis to extract dominant amplitude–frequency attribute values that reflect Amplitude Variation with Offset (AVO) responses, thereby predicting gas-bearing distribution.

3.2.5. Fracture-Distribution Prediction

Based on an understanding of fracture characteristics, coherence attributes are extracted from poststack seismic data to predict fracture distribution.

3.2.6. Gas-Rich Zones Integrated Prediction

Following the principle of facies-controlled reservoir prediction and combining sedimentary distribution patterns, we performed hierarchical intersection superposition of sand-body, gas-bearing-property, and fracture-distribution ranges. This approach predicts favorable zones of gas-bearing reservoirs.

4. Results

4.1. Analysis of Enrichment-Controlling Factors for Gas Reservoirs

4.1.1. Sedimentary Facies Distribution

During the deposition of the Qianfoya Formation, the study area and its surrounding regions underwent slow and steady tectonic subsidence. The depositional setting was characterized by broad sedimentary coverage, gentle terrain gradients, and shallow water depths, resulting in the dominant development of two subfacies within the Qianfoya Formation: the shallow-water delta front and lakeshore subfacies. Multiphase subaqueous distributary channel microfacies prevail across most of the area, while beach-bar and interdistributary bay microfacies locally occur in partial zones.

The Qian-2 member in the study area is dominated entirely by delta-front deposits. Subaqueous distributary channels represent the favorable microfacies for sand-body development, with sand bodies distributed regularly in nearly NS trending strips and exhibiting stable lateral continuity on the plane. Previous studies have demonstrated that sand bodies within subaqueous distributary channels feature good lateral continuity, large single-layer thickness, relatively coarse grain size, and favorable porosity–permeability properties, which provide high-quality reservoir space for natural gas accumulation.

Nevertheless, significant differences exist in sediment provenance and spatial distribution characteristics among the three submembers of the Qian-2 Member. The lower submember receives sediments from both NE and NW provenances and is dominated by subaqueous distributary channel deposits. Multiple distributary channels converge, resulting in widely distributed sand bodies, whereas interdistributary deposits are poorly developed, with local beach-bar deposits occurring sporadically (Figure a). The middle submember is supplied by a single NE provenance, where subaqueous distributary channel and interdistributary bay deposits are widely developed. Zonal channel bodies in the eastern part intersect with northern channels, forming thick sand accumulations surrounding well YL173 (Figure b). The upper submember also features NE-derived sediment supply, with laterally extensive subaqueous distributary channels and well-developed muddy interdistributary deposits. Multiple channel branches converge near well YL171, generating thick sand bodies (Figure c). Overall, during the transition from the lower submember to the upper submember, the spatial distribution of sand bodies gradually shifts, and the sediment provenance direction evolves from northeast–northwest to northeast–east. Convergence and intersection of fluvial sand bodies are widely observed throughout the interval. These evolutionary variations control the complex distribution of effective reservoirs within the Qian-2 member. Consequently, refined predictions and targeted analyses are required to delineate favorable sand bodies and gas-rich zones.

3.

Distribution of sedimentary microfacies for each submember of the Qian-2 Member. (a) Lower submember. (b) Middle submember. (c) Upper submember. Note: The red rectangular dashed box in the figure represents the seismic work area of this study.

4.1.2. Gas Reservoir-Cap Rock Conditions

4.1.2.1. Reservoir Properties

The Qianfoya Formation reservoirs are generally characterized by low porosity and low permeability. Laboratory measurements on core samples reveal that porosity values in the Qianfoya Formation in the study area range from 0.89 to 7.19%, with most values between 2.0 and 5.0%. Permeability varies between 0.0006 × 10^–3 and 2.65 × 10^–3 μm², with the main range of (0.01–2.1) × 10^–3 μm². A favorable correlation exists between porosity and permeability in this interval, indicating relatively favorable percolation capacity of matrix pores.

4.1.2.2. Gas Layer Types and Characteristics

The gas reservoirs within the Qian-2 Member can be classified into three categories, ranked from high to low in terms of gas-bearing properties: gas layers, oil-gas layers, and gas-bearing layers. Detailed geological parameters for these three gas layer types were obtained via core observations and statistical analysis of logging curves from multiple wells (Table ). Further analysis yields the following three key conclusions: first, gas-bearing intervals of the Qian-2 Member are predominantly distributed within the middle and upper submembers; second, the thickness of individual gas layers varies greatly, ranging from 1 to 7 m. Although several gas layers at similar depths are separated by thin mudstone interbeds (thickness < 1.5 m), they share identical reservoir-base-cap assemblages. Third, distinct differences are observed in mud content (Vsh), porosity (φ), and water saturation (Sw) among gas layers of different types.

1. Statistics of Logging Interpreted Geological Parameters for Various Gas-Bearing Interval Types.

well name	stratum	depth	interpretation result	thickness	Vsh	φ	Sw	lithology combination
		(m)		(m)	(%)	(%)	(%)
YL17	J2q2 3	3585–3586	Oil-gas layer	1	5.6	3.68	35.64	Mudstone/sandstone/sandstone
YL171	J2q2 3	3647–3651	Oil-gas layer	4	5.09	4.16	36.24	Mudstone/sandstone/mudstone
		3651–3656.5	Gas layer	5.5	2.17	8.24	24.7
YL173	J2q2 2	3742.2–3746.8	Gas-bearing layer	4.6	11.8	3.92	97.7	Sandstone/sandstone/mudstone
		3749–3750	Oil-gas layer	1	9.5	2.56	48.82	Sandstone/sandstone/sandstone
		3751–3753.8	Oil-gas layer	2.8	22.19	3.68	53.17
		3755.2–3761.1	Oil-gas layer	5.9	10.85	4.27	45.6
		3765–3769.5	Oil-gas layer	4.5	7.4	5.12	37.1	Mudstone/sandstone/sandstone
YL175	J2q2 3	3740–3744	Gas-bearing layer	4	15.57	5.66	53.74	Mudstone/sandstone/sandstone
		3744–3751	Gas layer	7	18.9	6.16	34.7
	J2q2 1	3820–3824	Gas-bearing layer	4	25.7	2.67	56.72	Mudstone/sandstone/mudstone
		3825–3828	Gas-bearing layer	3	18.75	5.03	86.53

4.1.2.3. Cap Rock Conditions

The cap rock conditions of the gas-bearing intervals within the Qian-2 Member in the study area are generally favorable and exert a critical control on the enrichment and preservation of tight sandstone gas reservoirs. The Qian-3 Member overlies the Qian-2 Member and has an average thickness of 88 m. Its lithology is dominated by dark-gray mudstone and silty mudstone interbedded with multiple thin argillaceous siltstone layers. The mudstone intervals originated from shallow lacustrine deposits and serve as high-quality regional cap rocks. This mudstone cap rock features a stable distribution and good lateral continuity. It provides effective vertical sealing for the underlying tight sandstone reservoirs of the Qian-2 Member and offers reliable barrier conditions to preserve tight gas reservoirs. The floor strata correspond to the Qian-1 member, which is approximately 127 m thick. This unit mainly consists of gray and dark-gray mudstones intercalated with gray siltstone, argillaceous siltstone, and fine sandstone. The strata exhibit an overall stable distribution, accompanied by strong heterogeneity, and demonstrate favorable vertical and lateral sealing performance.

4.1.3. Fracture Distribution

Fractures greatly modify the physical properties of tight sandstone reservoirs. They are key geological factors in improving the percolation capacity of low-porosity, low-permeability reservoirs and promoting gas migration and accumulation. Fractures in the Qianfoya Formation of the study area show complex types and variable occurrence. Statistical analysis indicates that fractures within the Qianfoya Formation predominantly distribute along faults, with three dominant dip directions: NE, NW, and SW, among which the NE direction dominates (Figure a). Fracture dips are primarily concentrated in two intervals of 20–60° and 80–90°. This suggests that both high-angle fractures and abundant medium- to low-angle fractures develop within the formation. Among them, medium-angle fractures have the highest proportion and the best development degree (Figure b).

4.

Statistical diagram of fracture dips in the Qianfoya Formation, Bazhong area. (a) Fracture azimuth rose diagram. (b) Statistical distribution of fracture dip angles.

Furthermore, integrating imaging logging and seismic fault interpretation results, statistical analyses were performed on multiple parameters from six wells in the study area, including well-to-fault distance, fault scale, fault dip angle, structural location, average single-layer thickness, and sand-stratum ratio (Table ). The following distribution rules for fracture development are summarized in the study area: first, medium- and small-scale faults are widely developed within the study area. Fracture density increases with proximity to large, steep faults. Specifically, wells located within 500 m of faults with dip angles exceeding 50° and extension lengths greater than 2 km exhibit abundant fractures. Second, fractures preferentially develop in the hanging wall of reverse faults with high structural curvature. Third, fractures are prone to form in strata with thick and relatively pure single-layer and high sand-stratum ratios. For instance, formations with an average single-layer thickness exceeding 2 m and a sand-stratum ratio greater than 0.3 exhibit more fractures. According to the above rules, fractures are relatively well developed around wells YL171, YL17, YL175, and YL176 and poorly developed around wells YL173 and YL172.

2. Statistics of Fracture Development Conditions for Multiple Wells in the Study Area.

fracture characteristics	YL17	YL171	YL175	YL172	YL173	YL176
Distance to faults	0.6 km	1.3 km	0.83 km	1.4 km	1.12 km	1.1 km
Fault scale	2.02 km	4.95 km	4.13 km	1.46 km	1.42 km	12.22 km
Fault dip angle	57°	50°	64°	23°	64°	58°
Structural location	Hanging wall of reverse fault, high structural curvature	Hanging wall of reverse fault, high structural curvature	Hanging wall of reverse fault, high structural curvature	Central fault block, low structural curvature	Footwall of reverse fault, moderately low structural curvature	Central fault block, high structural curvature
Average single- layer thickness	2.15 m	3.33 m	3.01 m	2.10 m	2.94 m	3.35 m
sand-stratum ratio	0.46	0.34	0.30	0.37	0.27	0.36

4.2. Geophysical Prediction of Gas-Rich Zones

4.2.1. Seismic Prediction of Sand-Body Distribution

4.2.1.1. Well-Logging Curve Characteristics of Sand Bodies

Based on actual well-logging data, this study analyzes differences in logging response characteristics between sandstone reservoirs and mudstone in the Qian-2 Member. It observes variations in γ-ray (GR), φ, density (DEN), and petrophysical parameter curves from six wells in the study area. Taking well YL171 as an example, numerical statistics are performed on its logging curves from the Qian-2 Member (Figure ). The value range features of logging curves for sandstone and mudstone can be identified. Sandstone intervals show low GR, high φ, relatively low DEN, high P-wave velocity (Vp) and S-wave velocity (Vs), high Zp and Zs, and low Vp/Vs ratio. Mudstone intervals show the opposite logging characteristics (Table ).

5.

Logging response characteristics of well YL171 in the study area. Note: The yellow intervals in the figure represent the oil and gas reservoir in the sandstone section, as derived from gas-testing data.

3. Comparison of Geological and Petrophysical Parameters of the Qianfoya Formation.

lithology	GR	φ	DEN	Vp	Vs	Zp	Zs	Vp/Vs ratio
	(%)	(%)	(g/cm3)	(m/s)	(m/s)	((m/s)·(g/cm3))	((m/s)·(g/cm3))
Mudstone	73–94	0.16–1.8	2.6–2.66	3380–4560	1920–2550	9460–11,550	4750–6545	1.89–1.95
Sandstone	42–60	3.6–7.2	2.48–2.6	4780–5032	2760–2970	12,210–14,200	6850–8250	1.76–1.81

4.2.1.2. Cross-Plot Analysis of Elastic Parameters

In this study, the GR log is used as the lithology indicator. A histogram statistical analysis of GR logging curves from six wells in the study area identifies a GR cutoff value of 73 API for distinguishing sandstones and mudstones. Under the background of GR, two-parameter cross-plot analysis is performed on elastic parameter curves of the Qian-2 Member for the six wells to optimize the sensitive parameter pair for lithology identification (Figure ). For the multiwell cross plot, the sampling interval of elastic parameter curves for each well is 1 m, yielding approximately 572 samples, which satisfies the requirements for cross-plot statistics. Comparative analysis shows that sandstone and mudstone samples can be well separated on cross plots of Zp versus Zs, Zp versus Vp/Vs, and Lamé impedance LR versus shear impedance MR (both LR and MR are derived from Zp and Zs). In contrast, sandstone and mudstone samples cannot be distinguished on the DEN versus Vp/Vs cross plot. Among these cross plots, the Zp–Zs cross plot exhibits relatively concentrated value distributions for sandstone and mudstone samples and achieves satisfactory lithological discrimination. Therefore, the Zp–Zs pair is selected as the sensitive elastic parameter pair for sandstone identification.

6.

Cross-plot analysis of elastic parameters for sand-body identification. (a) Cross plot of Zp versus Zs. (b) Cross plot of Zp versus Vp/Vs. (c) Cross plot of LR versus MR. (d) Cross plot of DEN versus Vp/Vs.

Based on the determined sensitive parameter pair of Zp and Zs, the coordinate rotation method is adopted to establish the prediction model for the sand-body indicator factor. This model fuses the two sand-sensitive parameters into a single sand-body indicator factor for sandstone identification. On the Zp–Zs cross plot (Figure a), sandstone samples display high values of both Zp and Zs, whereas mudstone samples show low values. From sandstone to mudstone, both Zp and Zs decrease, and the variation amplitude of Zp is larger than that of Zs, resulting in an obvious separation trend between the two lithologies. Accordingly, the boundary line between sandstones and mudstones is determined (the red straight line in the figure) using the following linear equation

$$\mathrm{Zs}=-0.497\times \mathrm{Zp}+723.6$$ 1

It can be seen from eq () that the slope of the boundary line is −0.497, from which the angle between the boundary line and the horizontal axis is estimated to be 26.41°.

7.

Cross-plot analysis of elastic parameters and coordinate rotation. (a) Cross plot of Zp versus Zs. (b) Cross plot after counterclockwise coordinate rotation. In the figure, X′ = LI+466, Y′ is the vertical coordinate after a counterclockwise rotation of the coordinate system.

Subsequently, by analyzing the slope characteristics of the sandstone–mudstone boundary on the cross plot, the formula for horizontal-axis rotation in the counterclockwise coordinate system is adopted to construct the prediction model for the sand-body indicator factor, given by

$$\mathrm{LI}=\mathrm{Zp}\times \cos \theta +\mathrm{Zs}\times \sin \theta -C$$ 2

In eq (), θ denotes the counterclockwise rotation angle of the coordinate system. This angle is calculated as 116.41° by adding 90° to the aforementioned 26.41° (the angle between the boundary line and the horizontal axis). After rotation by this angle, the boundary line appears as a vertical straight line on the cross plot in the new coordinate system. C represents the horizontal axis value corresponding to the boundary line when it is perpendicular to the horizontal axis on the new cross plot, the sandstone–mudstone boundary value, which is read as 466 (Figure b). Subtracting the constant C in eq () shifts the sandstone–mudstone boundary to zero. This produces positive and negative LI values for convenient discrimination between sandstones and mudstones. Finally, substituting the above parameters into eq () yields the prediction model for the sand-body indicator factor

$$\mathrm{LI}=-0.4447\times \mathrm{Zp}+0.8955\times \mathrm{Zs}-466$$ 3

In practical application, the estimated sand-body indicator factor can be normalized, and its calculation formula is given by

$${\mathrm{LI}}_{\mathrm{N}}=(\mathrm{LI}-{\mathrm{LI}}_{\mathrm{av}})/|\mathrm{LI}{|}_{\max }$$ 4

In eq (), LI_av represents the average value of the sand-body indicator factor. Since LI contains both positive and negative values, a value of 0 is adopted for normalization. |LI|_max denotes the maximum absolute value of LI, which can be statistically derived from the value range of the data volume. Based on statistics for the subsequent predicted data volume, this value is determined to be 1200. For the outputs calculated by eq (), the spatial distribution where LI_N > 0 is regarded as the predicted sandstone distribution.

4.2.1.3. Sand-Body Prediction by Prestack Simultaneous Inversion

Before seismic inversion, synthetic seismograms are constructed using DEN and Vp curves to conduct well-seismic calibration and horizon tying with poststack or prestack seismic traces. The YL171 well is an example. As shown in Figure , the wavelet used for synthetic seismogram generation is a statistical wavelet with a dominant frequency of approximately 42 Hz. The synthetic trace (blue) shows a good correlation with the actual well-side trace (red), with a correlation coefficient of 0.78. The well-side trace is derived from poststack data or the zero-angle trace of prestack angle gathers. We can see from the figure that poststack seismic data have continuous seismic events. Horizon tracking and identification are easy tasks. The prestack angle gathers have good quality. Seismic events at the top and bottom boundaries of the Qian-2 member are continuous. Obvious AVO trends are observed. These data meet the basic requirements for prestack inversion.

8.

Synthetic seismogram and horizon calibration of Well YL171.

Prestack simultaneous inversion is performed on prestack angle-gather data. It mainly includes the following steps: building the low-frequency model, extracting angle-dependent wavelets, analyzing the mudstone background trend, and tuning inversion parameters at well locations. First, after well-seismic calibration, a framework model is constructed from seismic interpreted horizons. Within the constraints of this framework model, multiwell Zp and Zs curves are interpolated to construct the low-frequency model. The quality of the framework model and well curves strongly influences the low-frequency model. Second, statistical wavelet sets for near angles (0–10°), medium angles (10–15°), and far angles (20–30°) are extracted separately from angle gathers. These wavelet sets are used for inversion calculation. Next, a background-trend analysis of the mudstone is performed for the Qian-2 Member. Quantitative regression formulas for Zp versus Zs and Zp versus DEN are established. Finally, inversion accuracy is improved by tuning inversion procedures and parameters. The main procedures and parameters include interpolation methods for the framework model and low-frequency model, wavelet shapes, coefficients of quantitative regression formulas, the background P-to-S-wave velocity ratio, and iteration times.

To verify the performance of prestack inversion, inversion parameter tests and error analysis are conducted at well locations. Five wells were selected in this study. They are used for low-frequency model construction and mudstone background-trend analysis. Well YL173 is taken as a blind well for error verification (Figure ). As shown in the figure, the measured curves (blue) match well with the inverted well-side traces (red). The matching ratios for Zp, Zs, and DEN were 0.88, 0.85, and 0.83, respectively. These results confirm the rationality of the selected inversion parameters and the reliability of the inversion outputs.

9.

Comparison between well location prediction traces of inversion parameters and measured curves.

Prestack inversion yields data volumes of sand-sensitive parameters Zp and Zs. Then, eqs () and (4) are used to calculate the sand-body indicator factor data volume (Figure ). Combined with petrophysical characteristics, the inverted Zp and Zs profiles (Figure a,b) show high Zp and Zs values (red-yellow areas) in the upper and middle submembers of the Qian-2 Member. The values are low in the lower submember. On the predicted profile of the sand-body indicator factor (Figure c), positive values of the sand-body indicator factor indicate sand-body distribution (colored areas). Mudstone intervals appear as blank areas. The predicted sand-body distribution matches the characteristics of the GR curves at well locations. The analysis shows that favorable sand bodies of the Qian-2 Member in the study area are mainly distributed in the lower part of the upper submember, the lower part of the middle submember, and the top of the lower submember. Laterally, sand bodies are well developed around Well YL171.

10.

Prestack inversion elastic parameter cross-well profiles. (a) Zp. (b) Zs. (c) LI_N. Note: In (a) and (b), the well-track logs correspond to Zp and Zs curves, respectively. In (c), the well-track log is the GR curve, and its low-value intervals are interpreted as sandstone.

4.2.1.4. Sand-Body Distribution Prediction

Layer-interval attribute slices are extracted from the sand-body indicator factor volume. A comparative analysis is conducted against the previously described sedimentary facies distribution. Overall, sand-body distributions among the three submembers of the Qian-2 Member show an inheritance from lower to upper intervals. They match the planar distribution characteristics of the underwater distributary channel microfacies in the delta front. Sand bodies in the upper and middle submembers exhibit clear geometries and high matching degrees. The matching degree for the lower submember is relatively low.

For the upper submember (Figure a), the seismic prediction map clearly shows the boundary of branch-channel sand bodies. These sand bodies are located north of Well YL172 and extend from the northeast or east. This branch channel merges with another branch channel from the northwest around Well YL171. Sand bodies are relatively poorly developed around and south of Well YL172. Therefore, the predicted sand-body characteristics in this area differ from the understood sedimentary facies distribution. For the middle submember (Figure b), the seismic prediction map indicates that east-oriented channels are well developed south of Well YL172. These channels merge with north-oriented channels around Well YL171. They extend continuously to the area along the line of Wells YL175–YL173. The boundaries of favorable sand bodies are relatively clear. However, branch-channel sand bodies along the line from Well YL176 to Well YL173 are poorly developed. The predicted sand-body boundaries in this area do not match the sedimentary facies distribution. For the lower submember (Figure c), the seismic prediction map shows scattered seismic attributes.

11.

Predicted planar distribution of sand bodies for each submember of the Qian-2 Member. (a) Upper submember. (b) Middle submember. (c) Lower submember.

Only the general geometry of sand bodies from the north can be identified. The predicted sand-body range deviates significantly from the sedimentary facies distribution. This deviation is caused by the limited resolution of seismic data. The vertical resolution of seismic data in the study area is about 28 m. As a result, the prediction accuracy for thin-bed sand bodies in the lower submember is relatively low.

4.2.2. Seismic Prediction of Gas-Bearing Distribution

4.2.2.1. Analysis of Seismic Reflection Characteristics

According to logging interpretations and gas-testing data from six wells in the study area, dry layers and different types of gas-bearing intervals occur in the Qian-2 Member. The gas-bearing intervals include gas layers, oil-gas layers, and gas-bearing layers. On the basis of multiwell petrophysical and seismic response characteristics, this study compares the spectral and near- and far-offset amplitude reflection features of gas-bearing intervals and dry layers. An identification model for gas-bearing intervals and dry layers was constructed.

Statistical analysis of seismic spectral characteristics shows that the effective frequency band of gas-bearing intervals ranges from 20 to 60 Hz, with a dominant frequency of approximately 40 Hz. For dry layers, the effective frequency band is 30–66 Hz, and the dominant frequency is relatively higher at about 47 Hz (Figure ). This indicates that the dominant frequency of gas-bearing intervals is obviously lower than that of dry layers. The difference between them is small and may be affected by the lithology of the gas-bearing reservoirs. It also reflects that the dominant frequency decreases after the tight sandstone contains gas. Comparative analysis of near- and far-offset seismic reflection features, in addition to wells, shows that the far-offset amplitudes are weaker than the near-offset amplitudes for gas-bearing intervals of different grades (Figure ). The amplitude difference between near- and far-offset traces is small for gas-bearing intervals. For dry layers, the far-offset amplitude is greater than the near-offset amplitude. On this basis, the identification pattern of gas-bearing intervals and dry layers on seismic reflection profiles is summarized (Table ), which provides a foundation for effective prediction of gas-bearing properties.

12.

Comparison of spectral characteristics between gas-bearing intervals and dry layers.

13.

Analysis of seismic reflection characteristics of gas-bearing intervals and dry layer. (a) Near-offset profile. (b) Far-offset profile.

4. Identification Patterns of Seismic Reflection Profiles for Gas-Bearing Intervals and Dry Layers.

type	AVO characteristics of near/far-offset stacked profiles			spectral characteristics
	near-offset stack	far-offset stack	amplitude difference
Gas-bearing intervals	Moderate-strong amplitude, large lateral variation, poor continuity	Weak to moderate-strong amplitude	Far-offset amplitude stronger than near-offset amplitude	Low dominant frequency
Dry layers	Weak amplitude, low frequency, discontinuous waveforms	Moderate-weak reflections, good continuity	Far-offset amplitude weaker than near-offset amplitude	High dominant frequency

4.2.2.2. Application of Dominant Amplitude–Frequency Method

The dominant amplitude–frequency method leverages the contrasting seismic spectral and near/far-offset reflection signatures of various reservoir types. It predicts the planar distribution of gas-bearing reservoirs by extracting dominant amplitude, dominant frequency, and their fused attribute values characterizing the “strong far-offset, weak near-offset” AVO response.

In terms of physical mechanism, gas-bearing tight sandstones produce AVO effects. Amplitude changes can be recognized from angle gathers and near- and far-offset stacked profiles. For the Qian-2 Member, we calculate the amplitude difference between far and near offsets (A _mF (t) – A _mN (t)). For gas-bearing reservoirs, the far-offset amplitude (A _mF (t)) is smaller than the near-offset amplitude (A _mN (t)). This difference is negative. The relative magnitude of the difference increases with rising gas saturation, showing a positive correlation, with a correlation coefficient R ² = 0.67 (Figure a). The amplitude difference between far and near offsets is negative for gas-bearing intervals. We need a new parameter. It makes the fused attribute value positive and keeps its value range relatively convergent. Therefore, the amplitude gradient G(t) is introduced. The figure shows that gas-bearing reservoirs have negative amplitude gradient values. Its relative magnitude increases with rising gas saturation. They present a positive correlation with a correlation coefficient R ² = 0.59 (Figure b). Meanwhile, gas-bearing tight sandstones cause seismic wave attenuation. The dominant frequency (f _m (t)) of gas-bearing reservoirs decreases. It shows a negative correlation with increasing gas saturation, with a correlation coefficient R ² = 0.59 (Figure c). Based on the above analysis, the prediction model for the gas-bearing indicator factor B(t) is established. Its calculation formula is given by

$$B(t)=\frac{A_{\mathit{mF}}(t)-A_{\mathit{mN}}(t)}{f_m(t)}\times G(t)$$ 5

From eq (), positive values of the gas-bearing indicator factor B(t) indicate a gas-bearing distribution. The key steps of this method are as follows:

14.

Correlation analysis between seismic attributes and Sg. (a) Correlation between A _mF (t) – A _mN (t) and Sg. (b) Correlation between G(t) and Sg. (c) Correlation between f _m (t) and Sg.

Step 1: Angle bin stacking is performed on prestack seismic gathers over the angle range 0–30° to generate near-offset (0–10°) and far-offset (20–30°) seismic volumes. The differences in reflection characteristics between gas-bearing intervals and dry layers are then analyzed comparatively using the near- and far-offset stacked profiles.

Step 2: Perform continuous wavelet transform (CWT) on near- and far-offset data volumes. The central frequency of the Morlet wavelet is set to 0.812 Hz. The effective frequency band is set to 5–65 Hz. The dominant-amplitude attributes A _mN (t) and A _mF (t) within the effective frequency band are extracted separately. Meanwhile, the dominant-frequency attribute f _m (t) and amplitude gradient G(t) within the effective frequency band are extracted from conventional data volumes.

Step 3: By substituting the aforementioned attributes into eq (), the dominant amplitude–frequency attribute B(t) can be estimated and subsequently used to predict the gas-bearing distribution.

4.2.2.3. Gas-Bearing Distribution Prediction

The dominant amplitude–frequency method was adopted to predict the gas-bearing distribution of reservoirs within the Qian-2 Member. As shown in the predicted profile of B(t) (Figure ), red-yellow areas correspond to high-value zones of B(t), which indicate favorable gas-bearing areas. Light-blue areas represent low-value zones of B(t) and suggest relatively poor gas-bearing potential.

15.

Cross-well profile of predicted gas-bearing distribution. Note: The well logs displayed are Sw curves, and low saturation values indicate gas-bearing intervals.

The planar gas-bearing distribution map (Figure ) illustrates that favorable gas-bearing distribution ranges in the upper and middle submembers are generally comparable, predominantly occurring around Well YL173, the area north of Well YL175, the Well YL176–YL171–YL17 trend, as well as the eastern and northeastern portions of the study area (Figure a,b). The lower submember of the Qian-2 Member is concentrated around Well YL173, Well YL175, and along the Well YL176–YL171–YL17 trend (Figure c), as outlined by black dashed lines in the figure. Overall, the spatial distribution of favorable gas-bearing zones is jointly controlled by sedimentary sand bodies and fault geometries, presenting consistent distribution patterns across the target intervals.

16.

Predicted planar distribution of gas-bearing zones in each submember of the Qian-2 Member. (a) Upper submember. (b) Middle submember. (c) Lower submember.

4.2.3. Seismic Prediction of Fracture Distribution

4.2.3.1. Well-Logging Curve Characteristics of Fracture

As mentioned above, core, imaging logging, and seismic data are combined to identify fracture development patterns of the Qianfoya Formation in the study area. Fractures are relatively well developed around Wells YL175, YL171, and YL17. This study takes Well YL171 as an example. It analyzes the fracture-logging response characteristics of the Qian-2 Member and defines fracture-identification features for a single well (Figure ). As shown in the figure, the lithology is fine sandstone and fine fluorescent sandstone at the depth interval of 3647–3650 m at the bottom of the upper submember. GR shows low values. Caliper log (CAL) reaches high amplitudes. DEN decreases. Deep laterolog (RD) is higher than shallow laterolog (RS). Multiple fractures can be clearly identified on imaging logging images. Similarly, at the depth interval of 3720–3723 m in the middle of the lower submember, the lithology consists of argillaceous siltstone and mudstone. GR presents low values. CAL shows high amplitudes. DEN decreases locally. Acoustic log (AC) drops and exhibits cycle-skipping phenomena. RD is larger than RS. Fractures can also be clearly recognized on imaging logging images. For these two depth intervals, conventional logging fracture-identification features are summarized using imaging logging data. These features include high CAL values, low DEN values, reduced AC values, and RD greater than RS.

17.

Fracture-logging response and fracture identification of Well YL171.

On the basis of the above single well fracture-identification features, multiple fracture developed intervals are identified within the Qian-2 Member. Fracture development is not controlled by the lithology. Fractures occur in both the sandstone and mudstone. They are mainly distributed in the fine sandstone interval at the bottom of the upper submember and in the siltstone-mudstone interbeds of the middle and lower submembers.

4.2.3.1. Fracture-Distribution Prediction

Combined with the fine fault interpretation results, the eigen-coherence attribute (C3 coherence algorithm) is extracted to predict fracture distribution in the Qian-2 Member. The sampling interval of 3D seismic data in the study area is 2 ms. The time window length and horizontal radius need to be optimized. Sensitivity analysis of these two parameters shows that a time window length close to one seismic cycle yields the optimal performance. A larger horizontal radius provides a higher signal-to-noise ratio but increases computational cost. A tradeoff is therefore required. Parameter perturbation comparisons are carried out within reasonable ranges. The overall distribution of fracture developed zones shows no obvious change. Only minor local differences appear in detailed features. Finally, the analysis time window is set to 50 ms and the horizontal radius is set to 3 × 3 traces. A balance is achieved between fracture-identification capability and computational efficiency. The value range of the eigen-coherence coefficient C is [0, 1]. A smaller value indicates stronger stratigraphic discontinuity, meaning more developed faults and fractures.

From the fracture prediction map (Figure ), it can be seen that low eigen-coherence values correspond to fracture developed zones (outlined by blue dashed lines in the figure). These zones are mainly distributed around small- and medium-sized faults (red lines) and areas with concentrated small faults. The comparison shows that fractures are relatively well developed in the middle submember of the Qian-2 Member. The upper submember takes second place. Fractures occur locally in the lower submember. For the upper submember, fractures are predominantly distributed in the western part of the study area, concentrated around faults along the Well YL176–YL171–YL17 trend, adjacent to faults near Well YL175, and in the vicinity and northern zone of Well YL173 (Figure a). In the middle submember, fractures are widely distributed across the whole study area. They concentrate around faults along the Well YL176–YL171–YL17 trend, around and north of Well YL175, and around faults in the northeastern and southeastern parts of the study area (Figure b). In the lower submember, fractures mainly occur around faults in the eastern study area and around faults along the Well YL176–YL171–YL17 trend (Figure c).

18.

Predicted planar fracture distribution derived from coherence attributes along the top boundary of each submember of the Qian-2 Member. (a) Upper submember. (b) Middle submember. (c) Lower submember.

4.2.4. Comprehensive Prediction of Gas-Rich Zones

Following the principle of facies-controlled reservoir prediction, this study adopts a comprehensive prediction framework integrating sand-body distribution, gas-bearing properties, and fracture development. In combination with the sedimentary distribution pattern of the Qian-2 Member, the Boolean intersection method was used for overlay analysis of the predicted ranges of sand bodies, gas-bearing properties, and fractures. Areas satisfying the threshold conditions of all three elements are delineated as gas-rich zones of the tight sandstone gas reservoirs.

Before the overlay analysis, favorable threshold conditions for the three elements need to be determined: (1) For the LI_N, well trace values and GR curves are cross plotted to establish a fitting baseline. A negative correlation exists between the two parameters (R ² = 0.54). The LI_N value corresponding to the upper GR limit of gas-bearing reservoirs is defined as the threshold for favorable sand bodies. Multiwell statistics show that the GR values of gas-bearing reservoirs in the Qian-2 Member range from 46 to 63 API. The LI_N value corresponding to the upper limit of 63 API is 0.18. Areas with LI_N > 0.18 are selected as the distribution range of favorable sand bodies (Figure a). (2) For the dominant amplitude–frequency fluid attribute B(t), well trace values and the gas-bearing saturation (Sg) curve of gas-bearing reservoirs are cross plotted to establish a fitting baseline. A positive correlation exists between the two parameters (R² = 0.70). The B(t) value corresponding to the lower Sg limit of gas-bearing intervals is set as the threshold for favorable gas-bearing zones. Statistics indicate that the lower Sg limit of gas-bearing reservoirs in the Qian-2 Member is approximately 10%, and the corresponding B(t) value is 15 API. Areas with B(t) > 15 API are selected as favorable gas-bearing zones (Figure b). (3) Based on the eigen-coherence attribute, boxplot statistics of coherence coefficient C are carried out around wells with well-developed fractures and poorly developed fractures, respectively. The dividing line between the two groups is determined and taken as the threshold for concentrated development zones of fractures (including small-scale faults). Based on statistical results, areas with coherence values C < 0.7 are selected as concentrated fracture zones (Figure c).

19.

Threshold analysis of LI_N, B(t), and C attributes. (a) GR and LI_N fitting. (b) Sg and B(t) fitting. (c) Statistical boxplot of C at well locations.

Overlay analysis is performed on the three enrichment factors of the Qian-2 Member. The spatial distribution characteristics of gas-rich zones in the Qian-2 Member are clarified (Figure ). As shown in the figure, favorable gas-bearing zones (yellow areas) are mostly located within the extent of favorable sand bodies (areas enclosed by black dashed lines). Fracture-concentrated zones are mainly distributed along fault strikes. They intersect the strike of favorable sand bodies and favorable gas-bearing zones at large angles. The spatial relationship between existing gas-test production and fault distribution supports an inference. The intersection of favorable gas-bearing zones within favorable sand bodies and fracture-concentrated zones forms the gas-rich zones of tight sandstone gas reservoirs. Gas-test data indicate that the daily gas production from Well YL171 is 1.31 × 10⁴ m³. The daily gas production from Well YL175 is 1.12 × 10⁴ m³. By contrast, Well YL173 only has a daily gas production of 0.11 × 10⁴ m³. Such production differences indicate that Wells YL171 and YL175 fall within gas-rich zones. Spatially, both high-production wells are close to faults and located in fracture-concentrated zones.

20.

Comprehensive predicted planar distribution of gas-rich zones for tight sandstone gas reservoirs in the Qian-2 Member. (a) Upper submember. (b) Middle submember. (c) Lower submember.

Based on the above understanding, the extents of gas-rich zones for three submembers are delineated on the comprehensive gas-rich zone prediction map. The gas-rich zones in the upper submember of the Qian-2 Member are relatively concentrated. They mainly occur around well YL173, around well YL175, in the southern area of well YL176, around faults along the well YL171–YL17 trend, and in overlapping areas of developed sand bodies and faults in the northeastern part of the study area (Figure a). The gas-rich zones of the middle submember also show concentrated distribution, mainly surrounding Well YL175, along the fault trend of Well YL171–YL17, and within the overlapping sand bodies and fault zone in the northeastern part of the research area (Figure b). In contrast, the lower submember contains limited gas-rich zones, which are locally distributed around Well YL173, Well YL175, and along the zone connecting Wells YL176, YL171, and YL17 (Figure c). In general, gas-rich zones of the Qian-2 Member are mainly developed in the middle and upper submembers. Their distributions show continuity. These gas-rich zones are concentrated in the overlapping areas of faults and favorable sand bodies.

5. Discussion

5.1. Method Applicability

The matching degree analysis for gas-rich zone prediction of tight sandstone gas reservoirs mainly depends on the application accuracy of two links: favorable sand-body prediction and gas-bearing prediction. On this basis, the applicability of the favorable sand-body prediction and gas-bearing prediction methods can be discussed.

The favorable sand-body prediction uses the petrophysics-based prestack simultaneous inversion method. Due to the limited number of wells in the study area, sparse well constraints reduce the confidence level of seismic inversion results in remote areas. This work uses leave-one-out cross validation (LOOCV) to evaluate the reliability. Well YL173 is taken as a blind well for error testing, and the remaining five wells are involved in prestack simultaneous inversion. From the inversion comparison of well YL173 (Figure ), the correlation coefficient between the synthetic seismic trace and the actual well-adjacent trace reaches 0.83. In addition, the inversion results of other wells are compared on inversion profiles. The well-seismic matching rate reaches 86%. The sand-body prediction distribution is compared with GR log characteristics at well locations, yielding a well-seismic matching rate of 80% (Table ). These comparative analyses verify the reliability of the inversion results.

5. Statistics of Well-Seismic Matching Rates of Prediction Results for Each Method Application Step.

Application link	YL176 (training well)	YL175 (training well)	YL171 (training well)	YL17 (training well)	YL173 (blind well)	YL172 (training well)	Matching rate
Seismic inversion (5 wells used)	Good match (10 points)	Good match (10 points)	Good match (10 points)	Good match (10 points)	Fair match (6 points)	Fair match (6 points)	86%
Sand-body prediction (5 wells used, compared with the GR log)	Good match (10 points)	Good match (10 points)	Good match (10 points)	Fair match (6 points)	Fair match (6 points)	Fair match (6 points)	80%
Gas-bearing prediction (no wells used, compared with the Sg log)	Fair match (6 points)	Good match (10 points)	Good match (10 points)	Good match (10 points)	Fair match (6 points)	Fair match (6 points)	80%

Gas-bearing prediction adopts the dominant amplitude–frequency method. No well-log curves are required throughout its application process. The predicted gas-bearing distribution can be directly compared to the Sg log at well locations. The well-seismic matching rate reaches 80% (Table ). A well-seismic comparison confirms the effectiveness and reliability of the gas-bearing prediction method. It can provide an effective methodological basis and an application reference for prediction of favorable targets in similar tight sandstone gas reservoirs.

The eigen-coherence algorithm is adopted for fracture prediction. This algorithm has a stable and reliable capability for fault identification. Well-seismic calibration shows that the overall fracture prediction results align well with the fracture development patterns observed in actual wells. Gas-rich zones in the study area are favorable geological units jointly constrained by sand-body distribution, gas-bearing seismic response, and fracture development. They only represent the potential enrichment range of natural gas and do not directly correspond to high-production zones. Analysis of multiwell gas-testing production features indicates that the gas-rich zone results constrained by multiple factors can better reflect the actual enrichment rules of tight sandstone gas reservoirs, compared with the prediction derived merely from a single gas-bearing attribute. Based on the above analysis, the complete set of prediction techniques and workflow established in this study can serve as an effective technical reference for the favorable area and target optimization of similar tight gas reservoirs.

5.2. Method Uncertainties

5.2.1. Uncertainties Caused by Well-Seismic Data Limitations

Only six drilling wells are available for well-log calibration and inversion constraints in the study area. The well density is low. Limited well control brings uncertainties to petrophysical analysis, seismic inversion accuracy control, as well as the prediction and verification of sand-body, gas-bearing, and fracture distribution. For example, the construction of the initial model in seismic inversion is strongly affected by the number of involved wells. The limited number of wells leads to smoothed interpolation between wells. It reduces differences in lateral elastic parameters caused by local lithological variations. It further increases uncertainties in inversion and lithology prediction.

In addition, the dominant frequency of target intervals in local 3D seismic data is about 42 Hz. Using a sandstone interval velocity of 5200 m/s, the vertical interpretive resolution limit (one-eighth wavelength) for inversion is calculated to be approximately 14 m. Therefore, the sand-body indicator and gas-bearing indicator show poor prediction performance for thin sand bodies (1–3 m) and gas-bearing reservoirs. They can only realize general predictions for sand–mud interbeds or sand bodies with mud interlayers thicker than 14 m and their gas-bearing reservoirs. Under current seismic resolution conditions, more wells can strengthen well constraints and greatly reduce the uncertainties of this method.

5.2.2. Uncertainties in the Prediction Process

For sand-body prediction, elastic parameter cross-plot analysis is performed. The sample points for sandstone and mudstone show a trend of separation in tight formations. Although a sensitive parameter pair is selected through comparison, the extraction of the sand–mudstone boundary contains multiple solutions. Quantitative discrimination criteria are required for further determination. Moreover, background-trend analysis of mudstone is required in prestack simultaneous inversion. Statistical regression on parameter trends of mudstone intervals from multiple wells builds constraint relationships between Zp and Zs, as well as between Zp and DEN. However, the regression-fitting model is affected by interwell variations in mudstone composition and pore properties. Systematic biases may occur. Such biases exert relatively limited influence and can be mitigated through multiple iterations to achieve optimal solutions.

In terms of gas-bearing property prediction, the dominant amplitude–frequency attribute method has inherent uncertainties. False-positive misjudgments may occur when the amplitude difference between near- and far-offset traces is small or the variation in the seismic dominant frequency for gas-bearing intervals is weak. There are three main reasons. First, calcareous interbeds and high-rigidity tight sand bodies can raise far-offset amplitude and reduce local seismic spectrum. Lithology changes may be mistakenly identified as fluid anomalies. Second, thin-bed tuning effects alter amplitude and spectral features. Nongas intervals adjacent to gas-bearing reservoirs may show similar attribute responses. Third, strong impedance-contrast interfaces in thick interdistributary bay mudstone can generate large-amplitude gradient values and disturb attribute calculations. Therefore, this study adopts the fused dominant amplitude–frequency attribute to reduce uncertainties from single-attribute predictions. The gas-bearing prediction results should be interpreted under the constraint of sand-body distribution. Well-seismic comparison is required to verify its effectiveness.

In terms of fracture prediction, structural attributes such as eigen-coherence can predict macroscale stratigraphic discontinuities, including medium-to-small faults and concentrated fracture zones. Uncertainties exist in the prediction scale. Cores and imaging logging data can reflect microfracture features on wellbore walls. However, they bring uncertainties in lateral fracture extension. In addition, the Qian-2 Member of the study area is dominated by small- and medium-sized faults with relatively small dimensions. Fractures mostly develop in belts along faults. Wells with good gas-bearing performance are located within fracture belts on the hanging wall of reverse faults. Therefore, concentrated fracture zones generally play a positive role in hydrocarbon accumulation and are favorable for gas enrichment.

In target prediction for gas-rich zones, the boundaries of favorable sand bodies, gas-bearing zones, and fracture distributions are delineated using indicator parameters or attribute thresholds. However, some local regions are far away from well locations. For example, areas with abrupt changes in sedimentary microfacies and fragmented attribute indicators have unclear geological understanding. Delineating boundaries only by threshold values will introduce uncertainties. The Boolean intersection method is adopted for multiparameter superimposed interpretation. Geological knowledge must be integrated to gradually reduce the uncertainties of attribute interpretation.

6. Conclusions

• (1)The enrichment-controlling factors in tight sandstone gas reservoirs within the Qian-2 Member in the Bazhong area are clarified. Specifically, the extent of gas-rich zones is governed by the distribution of sedimentary facies, reservoir physical properties, gas-bearing capacity, cap-base rock conditions, and fracture distribution.
• (2)Petrophysical cross-plot analysis identifies Zp and Zs as sensitive parameter pairs for sandstone discrimination. Prestack simultaneous inversion is performed to acquire these two impedance volumes, which are fused via coordinate rotation to generate a sand-body indicator attribute. The distribution of high positive values of its sample points can reliably indicate the spatial distribution of the favorable sand bodies.
• (3)Based on the comparison of near- and far-offset seismic reflection characteristics beside wells, this study summarizes the identification pattern of far-offset weak amplitude and near-offset strong amplitude for gas-bearing intervals, and the difference in the seismic dominant frequency between gas-bearing intervals and dry layers is determined. On this basis, the dominant amplitude–frequency method is applied to extract the corresponding attribute volumes, in which the high-value distribution of sample points can effectively indicate the distribution of gas-bearing zones.
• (4)Fracture prediction results demonstrate that fractures are well developed within the Qian-2 Member, and their spatial distribution is concentrated along fault strikes and zones densely populated with small-scale faults.
• (5)The gas-rich zones of tight sandstone gas reservoirs in the Qian-2 Member are delineated by the superposition of sedimentary sand bodies, gas-bearing properties, and fracture distribution. These zones are greatly affected by small- and medium-sized faults and fracture distribution. Vertically, gas-rich zones are mainly developed within the middle and upper submembers of the Qian-2 Member. Laterally, these zones exhibit a continuous spatial distribution and are clustered in overlapping intervals of faults and favorable sand bodies.
• (6)A set of prediction methods and workflows for gas-rich zones in tight sandstone gas reservoirs is established based on multifactor analysis. A well-seismic comparison shows favorable application performance for this method. However, certain uncertainties remain in several steps of the application. These steps include an elastic parameter cross plot, seismic inversion, analysis of differences in gas-bearing responses, and threshold determination for multifactor prediction. The uncertainties stem from limited well data, seismic resolution constraints, and strong heterogeneity of tight sandstones. Additional well data and model constraints are required to further optimize the prediction results.
• (7)This target prediction of gas-rich zones in this study considers only three major controlling factors. It does not take into account the distribution of the source rocks. Heterogeneity responses, such as rock brittleness and formation pressure variation, were also neglected. Further work can focus on the sweet spot prediction of tight sandstones. Enrichment pattern research can be supplemented. Quantitative production-grade predictions can also be carried out. These efforts will provide more refined references for the exploration of optimization.

The authors declare no competing financial interest.