The Seismic Identification of Small Strike-Slip Faults in the Deep Sichuan Basin (SW China)

Abstract

Recently, the “sweet spot” of a fractured reservoir, controlled by a strike-slip fault, has been found and become the favorable target for economic exploitation of deep (>4500 m) tight gas reservoirs in the Sichuan Basin, Southwestern China. However, hidden faults of small vertical displacements (<20 m) are generally difficult to identify using low signal–noise rate seismic data for deep subsurfaces. In this study, we propose a seismic processing method to improve imaging of the hidden strike-slip fault in the central Sichuan Basin. On the basis of the multidirectional and multiscale decomposition and reconstruction processes, seismic information on the strike-slip fault can be automatically enhanced to improve images of it. Through seismic processing, the seismic resolution increased to a large extent enhancing the fault information and presenting a distinct fault plane rather than an ambiguous deflection of the seismic wave, as well as a clearer image of the sectional seismic attributes. Subsequently, many more small strike-slip faults, III–IV order faults with a vertical displacement, in the range of 5–20 m, were identified with the reprocessing data for the central Sichuan Basin. The pre-Mesozoic intracratonic strike-slip fault system was also characterized using segmentation and paralleled dispersive distribution in the Sichuan Basin, suggesting that this seismic process method is applicable for the identification of deep, small strike-slip faults, and there is great potential for the fractured reservoirs along small strike-slip fault zones in deep tight matrix reservoirs.

1. Introduction

Faults and their fracture networks occur widely in the Earth’s crust and have significant roles in the mechanical, petrophysical and fluid effects of rocks [1,2,3,4]. The complicated architecture and mechanism of a fault zone are important issues for scientific study and engineering applications [3,4,5,6,7]. Geological, geophysical and drilling technologies have been used to detect fault networks in subsurfaces. Geophysical responses to fault displacements/relief are universals signs used in the detection and mapping of subsurface faults [8,9,10]. New seismic processing methods have been proposed to enhance the seismic resolution of small-scale faults [11,12,13]. Furthermore, artificial intelligent (AI)-based methods can also be used to identify small faults [14,15]. However, hidden faults under subsurfaces are generally too small in scale to be identified by conventional technologies.

Recently, a large amount of petroleum resources were found along the pre-Mesozoic strike-slip fault zones in deep (>4500 m) intracratonic basins [16,17]. Because most deep pre-Mesozoic matrix reservoirs were tight during their long burial histories, the “sweet spots” of fractured carbonate reservoirs along fault zones are becoming favorable drilling targets to enhance economic oil/gas exploitation and production in deep subsurfaces [16,17,18]. In tight matrix reservoirs, fracture networks and their related dissolution porosities along hidden faults play important roles in hydrocarbon migration and accumulation [19], and they have significant impacts on enhancing permeability, by more than one order of magnitude, and porosity, by more than 50%, forming the “sweet spots” of fractured reservoirs [20,21]. Generally, fractured reservoirs are closely correlated with hidden faults in fault zones [10,16,17,20]. High-resolution 3D seismic methods, such as seismic coherence, curve, amplitude attribute, ant body and maximum likelihood, have been used to identify hidden faults in deep petroliferous basins. These hidden strike-slip faults generally present as low-amplitude, small-sized structural anomalies with lateral dimensions of 2–5 km and an amplitude of less than 20 m [16,17]. Owing to the limitations of the seismic resolution, small hidden faults do not present as visually distinct during seismic imaging of deep subsurfaces, and the presence of many pseudomorphic faults can decrease the accuracy of the identification of small hidden faults [16,17,18]. These issues have hampered industry mappings of small fault-related reservoirs by seismic data and, subsequently, fractured reservoir descriptions for well optimization in deep subsurfaces.

Recently, a large strike-slip fault system was found in the central Sichuan Basin of Southwestern China [22,23]. The strike-slip faults could connect the source-reservoir assemblage and gas enrichment [22,24]. Wells that penetrate the “sweet spots” of fracture–vug reservoirs in strike-slip fault zones can achieve a much higher level of gas production than matrix reservoirs [21,24,25,26]. The deep strike-slip fault-related gas reservoir has been a major gas exploitation frontier in the Sichuan Basin. The seismic coherence and amplitude attributes have been used to describe the strike-slip faults [17,22]; however, it is still difficult to describe the small (vertical displacements of less than 20 m) hidden strike-slip faults in the deep subsurface [17]. Considering that the “sweet spots” of fractured reservoirs are distributed in the small hidden fault damage zones [21], the seismic identification of small strike-slip faults is important for the reservoir’s description and the targeted optimization in the basin.

In this study, we propose a seismic processing method that improves seismic resolution in the identification of small hidden strike-slip faults, as well as provide a discussion of its application in deep pre-Mesozoic carbonate reservoirs.

2. Geological Background

The Sichuan Basin is a typical intracratonic basin that has undergone more than 10 stages of tectonic–sedimentary evolution with relatively complete Ediacaran–Quaternary strata (Figure 1) [27,28,29]. A carbonate platform of the late Ediacaran–Ordovician and late Permian–middle Triassic has developed across the basin. The Lower Cambrian shale is a regional source, and the carbonate reservoirs in the Upper Ediacaran, Lower Cambrian and Upper Permian are the major gas-producing strata in the central basin [30,31].

The largest carbonate gas field in China, with a more than 1 × 10¹² m³ geological reserve, was found in the central uplift [31]. The Upper Ediacaran is one of the major gas-bearing strata. The ancient dolomite reservoirs of the Upper Ediacaran are generally attributed to be high-energic facies-controlled reservoirs in the carbonate platform [30,31]. However, the ancient carbonates underwent a significant reduction in porosity during the long diagenetic process, resulting in the low porosity–permeability and low levels of production [21]. Recently, the NWW-trending strike-slip faults were dispersively distributed in the Ediacaran–Permian [22,23,32] (Figure 1a). They are generally composed of several oblique fault segments with multiple inherited strike-slip fault activities (Figure 1). In low-permeability (<0.5 mD) matrix reservoirs, strike-slip faults can increase the permeability by more than one order of magnitude [21,22,25]. Furthermore, the dissolution porosity along a strike-slip fault damage zone can improve the porosity by more than 50% in low-porosity (<3%) matrix reservoirs [21,25]. More than 80% of high-production gas wells are located along strike-slip fault zones. However, it is still a significant challenge to identify small, hidden strike-slip faults in deep, tight carbonate reservoirs [17].

3. Methods

The coherence, curve, likelihood and fusion attributes have been used to identify strike-slip faults in the Sichuan Basin [17,22], but they are not favorable for the identification of small (vertical displacement <20 m) strike-slip faults in deep subsurfaces (Figure 2b,c). According to a seismic data analysis, multiple factors influence the imaging of the strike-slip faults. Firstly, seismic data cannot show the distinct offsets of the small strike-slip faults by conventional seismic processing [17,22]. This is correlated with the small vertical displacement that is beyond the seismic resolution, with a low main frequency of 20–30 Hz in the deep strata. Thus, the small strike-slip fault generally presents weak flexure without a distinct offset between the two fault walls. Owing to most of the vertical displacements along the strike-slip fault zones being less than 30 m in the study area [23], the low-amplitude fold in the faulted block is observed instead of the offset of the seismic event. Particularly, the image of the vertical fault surface is ambiguous in the deep subsurface. Secondly, the thrust belt, evaporate, and fold can mask the imaging of strike-slip faults and strata images (Figure 2a). Low-resolution seismic data generally present ambiguous images of the strike-slip faults and more pseudomorphic fault images in the seismic sections. This is closely related to the low signal–noise ratio of the seismic data in the deep subsurface. Subsequently, there are a multiplicity of seismic interpretations of the strike-slip faults in the central Sichuan Basin [22,23]. In addition, the ambiguous fault image in the seismic data is unfavorable for the mapping of fault combination. In this context, enhancing the hidden fault image by seismic data processing is significant for fault identification and interpretation.

Considering the weak seismic responses and low resolutions of small strike-slip faults, we propose a multiscale and multi-angle intelligent processing method to improve the seismic resolution for identifying small hidden strike-slip faults. With the directional controllable filtering using a linear combination of directional filters and multiscale image decomposition, the seismic signal is decomposed into sub-band information that comprises different geological information at different scales and directions (Figure 3).

The optimal direction of the decomposed data is enhanced to investigate and highlight the geological information from the seismic data, as well as more clearly characterize and identify deep, small faults. Through a linear multiscale and multidirectional decomposition, the fault image can be automatically changed from the small scale to the large scale [33]. Small-scale information can reflect a detailed image of a hidden fault, while large-scale information can better reflect the macro information of a large fault. The decomposition processing is carried out by recursively calling the low-pass radial filter in the frequency domain to realize a pyramid-like structure [32]. Then, the bandpass filtering image from a different direction can be obtained by the directional controllable filter processing of each layer. Therefore, the image can be decomposed into sub-band information at different scales, and each layer can be decomposed into different directional data to obtain sub-band images that contain different directional information.

The directional controllable filter processing uses a set of base filters and has directional rotation. The processing of the directional controllable filter has controllability, and the processing result is characterized by fewer calculations but much higher precision. With the seismic data, directional controllable filter processing is filtered using the direction of the in-phase axis, and its function expression is as follows:

$$f^{\theta }\left(x,y\right)={\sum }_{j=1}^N{k}_j\left(\theta \right)f^{{\theta }_j}\left(x,y\right)$$

where $f^{\theta }\left(x,y\right)$ is the function of the controllable filter in the θ direction, which can be obtained by the linear combination of the interpolation function $k_j\left(\theta \right)$ and the basis function $f^{{\theta }_j}\left(x,y\right)$ in the θ direction; j is the number of layers of decomposition; and N is the logarithm of the basis function and the interpolation function.

To solve the problem of how to choose the $f^{{\theta }_j}\left(x,y\right)$ and $k_j\left(\theta \right)$ functions and the number of filters, the rectangular coordinate system is converted to a polar coordinate system, in which $r=\sqrt{x^2+y^2}$, $\varphi =\mathit{arg}\left(x,y\right)$, and then $f^{\theta }\left(x,y\right)$ is expressed in polar coordinates as follows:

$$f^{\theta }\left(r,\varphi \right)={\sum }_{N=-M}^N{k}_n\left(\theta \right)g_i\left(r,\varphi \right)$$

where $g_i\left(r,\varphi \right)$ is the basis of the function $f^{\theta }\left(r,\varphi \right)$.

In the seismic processing, the first process of the directional controllable filter is convolution with a set of base directional filters (three different directions) of the input images. Then, the pyramid image processing is performed by assigning the weight coefficient of the interpolation function corresponding to the directional filtered image by multiscale analysis. The process of image reconstruction is mainly used to determine the interpolation function [33]. Then, the bandpass-filtered image and its corresponding interpolation function are multiplied and added to obtain the filtered image. This procedure mainly includes the estimation of the direction of spatial change and the analysis of different dimensional attributes. The fault information is highlighted by stacking to reconstruct the fault images.

To improve the seismic resolution of the strike-slip fault, an artificial intelligence method and anisotropic diffusion compensation algorithm were used in the processing to carry out special de-noising of the 3D seismic data. The deep learning workflow can be used for the optimal detection and imaging of the deep fault [14,15]. The sampling of the seismic response of the strike-slip fault is from the large-scale fault that has been confirmed by the well data and distinct seismic data. Although there are multiple and complicated geological and seismic samples of the strike-slip faults, they are generally similar for the large-scale and small-scale strike-slip faults in the deep subsurface [22,23]. The absolute amplitude of the preferential directional-filtered data is favorable for strike-slip fault imaging.

4. Results

Through seismic data processing, the seismic imaging of the strike-slip fault improved significantly, making it favorable for the identification and interpretation of deep faults.

4.1. Improving the Seismic Resolution of the Strike-Slip Faults

The seismic reprocessing improved the seismic resolution of the strike-slip fault (Figure 4). In the study area, the formation wave groups have large differences from the surface to the deep strata, and the main frequency decreased with depth, by greater than 30 Hz to less than 20 Hz. In the deep strata, the low frequency and strong continuity of the seismic wave groups generally covered the image of the strike-slip faults (Figure 4a). Hence, it is not easy to identify the strike-slip fault in the deep strata. By extracting large-scale information from the original data with strong background information and enhancing the small-scale information, the seismic resolution in both the deep and shallow strata can be improved (Figure 4b). The main frequency and bandwidth of the seismic data are enhanced, particularly for the deep strata. This is very helpful in strengthening the fault discontinuity information conducive to seismic imaging of the vertical strike-slip fault.

Owing to the low-resolution seismic data, the small strike-slip fault plane generally presents deflection and fold in the original seismic section (Figure 5a). These features could be correlate to the strata folding that is hard to distinguish with the conventional seismic section. Through seismic reprocessing (Figure 5b), the strike-slip fault plane has become clearer, showing a distinct offset of the seismic reflection. With the improved imaging of the strike-slip fault plane, the offset between the fault walls is obvious, showing the continuous fault plane. These reprocessed seismic data can be used to identify the strike-slip fault and analyze the fault style in the seismic section.

The seismic wave is more continuous in the deep strata, which can be attributed to the low-resolution seismic data that have shed the seismic response of the strike-slip fault (Figure 6a). Through seismic reprocessing (Figure 6b), the imaging of the deep strike-slip fault has significantly improved. The seismic resolution of the strike-slip fault increased with the increase in depth. With this enhanced seismic resolution, the strike-slip fault in the deep strata also presents a distinct seismic response. At the same time, it can easily be identified that the strike-slip faults developed along the vertical deformation belts.

Furthermore, this processing method can highlight the discontinuity in the fault damage zone. Owing to the strike-slip fault zone being characterized with broad fracture zones, it usually shows discontinuous seismic reflection waves (Figure 6a). Although it is difficult to identify the smaller strike-slip faults, the reprocessed seismic data allow for highlighting the discontinuity in the damage zone and facilitate the identification of the fracture zone. The results show that the small strike-slip fault zone generally has a wide damage zone of more than 1 km (Figure 6b). The width of the fault damage zone is more than those in other areas with the same displacement [34,35,36]. Although the major fault plane is unclear and microfractures are difficult to identify, the boundary of the fault damage zone is distinct in the seismic images of the seismic discontinuity. In addition, the architecture and damage strength of the fault zones have a distinct seismic response too. Through seismic processing, the fault damage zone presents a distinct seismic response in the reprocessed seismic section (Figure 6b). The seismic discontinuity is favorable for the identification and description of the widths and strengths of strike-slip fault damage zones.

Furthermore, seismic processing can improve the image of the strike-slip fault in seismic planar attribute. Generally, the planar coherence, curvature, and amplitude attributes are of the most importance in the identification of strike-slip faults [22]. However, the planar seismic attributes generally show weak and ambiguous responses by the strike-slip fault (Figure 2b,c and Figure 7a). This has led to difficulty in identifying the strike-slip fault and subsequently different interpretation results in the Sichuan Basin [23,24,25,26]. Through the seismic processing, the extracted coherence attribute (Figure 7b) shows a clear fault line response, which can be used to identify the planar distributions of strike-slip faults. Combined with the distribution characteristics of en échelon/oblique faults, the seismic planar attribute is more advantageous than the seismic section for identifying hidden strike-slip faults.

4.2. Discovery and Mapping of Small Strike-Slip Faults

Through the seismic processing, small strike-slip faults can be identified by the planar seismic attributes and seismic section. The seismic data can be used to identify and interpret small strike-slip faults of class III (length > 10 km and maximum vertical throw > 20 m) and class IV (length < 10 km and maximum vertical throw < 20 m) in the central Sichuan Basin (Figure 8). The more reliable image of the strike-slip fault in the planar and sectional attributes provides data upon which to base the identification and interpretation of the strike-slip faults.

Furthermore, many more class III–IV strike-slip faults can be identified that were not distinguished by the unprocessed data. Seismic interpretation and industrial mapping of the strike-slip faults were carried out to identify the distribution of the strike-slip faults in the 3D seismic area (Figure 8). The major strike-slip fault zones were mapped with a total length of 720 km in the Anyue Gasfield [22], while a total length of 1860 km of strike-slip faults have been discovered with the seismically processed data. The NE-strike faults and class III–IV strike-slip faults are newly discovered faults. The results show that there are 8 class I strike-slip fault zones (length > 50 km and maximum vertical displacement > 60 m), 14 class II strike-slip fault zones (length > 20 km and maximum vertical displacement > 40 m), and 688 III–IV strike-slip faults with a total length of 1146 km. These suggest that there are many smaller strike-slip faults with dispersive distribution in the Sichuan Basin, which have been confirmed by recent borehole data across the strike-slip fault zone (Figure 9) [21]. The results indicate that the strike-slip faults are pervasively distributed throughout the basin [22,23].

5. Discussions

5.1. Seismic Methods for Improving the Resolution of Strike-Slip Faults

Geological, geophysical and engineering methods are helpful in hidden fault identification in subsurfaces [7,8,9,10,11,12,13,14,15]. The seismic method has been widely used in fault and fractured reservoir description in the petroleum basin [10,11,12,13]. The low seismic resolution and the interference of seismic noise have hampered the identification of hidden strike-slip faults with small-scale displacements.

In the central Sichuan Basin, the major class I–II strike-slip fault zones have been found using multiple seismic attributes [22,23,24]. Owing to the complicated seismic responses of the weak faults, there are many different mappings of the strike-slip fault network. Particularly, the small hidden strike-slip faults present different fault distributions and assemblages. Furthermore, the coherence, curve and amplitude attributes are favorable for the identification of strike-slip faults in the Tarim Basin, but this is not the case in the Sichuan Basin [10,16]. Using the seismic reprocessing method in this study, seismic noise was suppressed to a large extent, and imaging of the fault was improved. This further processing benefitted from the calibration of well data, which allowed for small strike-slip faults to be imaged clearly. The seismic imaging of strike-slip faults improved significantly, making them favorable for the identification and mapping of deep faults in the Sichuan Basin (Figure 5, Figure 6, Figure 7, Figure 8 and Figure 9). Owing to the small-scale and vertical fault plane, this is generally difficult with small strike-slip fault imaging using the seismic process, and the class III-IV strike-slip faults need further study to increase the seismic resolution (Figure 7b, Figure 8b,c and Figure 9b). For example, well GS018-1-H2 has penetrated three small fracture zones (Figure 9b), which have ambiguous responses in the planar and sectional attributes. Hidden strike-slip faults with vertical displacements as low as 5–10 m can be identified with this seismic reprocessing method, whereas the seismic resolution and precision for identifying hidden faults are not easily evaluated and corrected using the available data. The prestack/poststack seismic processing methods need further study to improve the deep smaller fault imaging [17].

Furthermore, the seismic attributes used to identify strike-slip faults are generally from within a large window of time to enhance the fault responses [10,16,17]. These attributes are helpful for confirming the distribution of the strike-slip fault zone, whereas the fault segmentation and exact fault location are ambiguous according to these data. On the other hand, the horizontal attribute is generally too weak to show the smaller fault trace in a planar map. In this context, AI-based multiple directional and scaled processing could be much helpful for the next seismic imaging of the smaller strike-slip faults. Owing to multiple factors and the changeable seismic responses of the strike-slip faults, the sample learning and AI-based process method could be the optimal solution for deep hidden strike-slip fault identification and characterization.

5.2. Characterization of Strike-Slip Faults and Fractured Reservoirs

It has been shown that there are weak strike-slip fault activities in the intracratonic basin [16,17]. The strike-slip faults generally present multiple styles and activities in the seismic section, as well as multiple segmentations and types in the planar attributes. However, these depend on the seismic resolution used in the identification of hidden strike-slip faults.

Through seismic processing, the seismic attribute shows a distinct fault segmentation (Figure 7b and Figure 8a). The strike-slip fault zone is characterized by several en échelon or oblique segments. Segmentation may be affected by the fault’s maturity, with fault growth and interaction [23]. It has been shown that most strike-slip fault zones, particularly of small fault zones, are in the isolated or soft-linkage stages. This suggests that the intracratonic strike-slip faults have obvious segmentation along the fault strike. This is consistent with the fault segments obtained by fault mapping [22,23]. Small fault zones generally present isolated and short segments. Large fault zones can present an overlapping zone with wider and longer segments. In this way, most small displacement but long fault segments need to be reinterpreted as having several segments. Furthermore, the overlapping fault zone could have developed from fault tip interaction. This is consistent with the power law of fault elements [25,34]. Although it is generally difficult to distinguish fault segmentation, the vertical throw and width measurement along the strike-slip fault zone is very helpful in fault segment discrimination and mapping (Figure 8a). The vertical throw of the single fault segment decreases from the center toward both ends, and it exhibits a sudden change between different fault segments, which can be attributed to the fault linkage [37]. However, the strike-slip fault is characterized with the much lower immaturity of the en échelon/oblique segments in the Sichuan intracratonic basin.

The detailed seismic interpretation results show that strike-slip faults have developed widely in the Anyue Gasfield. On the basis of a fine interpretation of the small strike-slip faults, the 3D visualization mapping presents a detailed fracture network and damage zone along the strike-slip fault zone (Figure 10) [33]. With the correction of the seismic section and planar attributes, it has been shown that the fracture network is widely distributed in the Ediacaran dolomites. The discontinuous fracture network suggests an isolated, soft-linked fracture system, consistent with the immature strike-slip fault system [37]. Although the 3D visualization needs improvement in mapping the detailed spatial distribution of small strike-slip faults, it has achieved the 3D modeling of the small fracture network in the deep subsurface [16,17]. The 3D visualization of the fracture network suggests a dispersive fracture system in the central Sichuan Basin.

The results show that the major strike-slip fault zone is segmented with several soft–hard-linked segments (Figure 11). Soft linkage refers to noninteraction between two adjacent fault segments, but interactions occur among the hard-linked fault segments [37]. The displacement and deformation generally decrease from the segment center to the tip of the soft-linked segment but could be transferred and localized to the overlapping zone between two hard-linked segments. The strong interaction between two hard-linked segments can lead to overlapping, connection, deformation and, subsequently, a complicated fracture network [37]. The vertical displacement in the hard-linked zone typically presents a sudden increase that could be correlated with the secondary hidden fault. In addition, the secondary strike-slip faults are widespread to the east, showing isolated fault segments at the tip of the major fault zone. This can be used to describe the spatial network along the fault damage zone. The segments’ lengths and widths are correlated with the fault maturity [37]. The variable intensity of the fault zone’s vertical and planar attributes suggests a complicated fracture connectivity in the carbonate fault zones.

It has been shown that strike-slip faults significantly contributed to the tight carbonate reservoirs and high gas production more than once in the deep subsurface, and fractured reservoirs with “sweet spots” are favorable targets for the economic development of deep tight reservoirs [17,21,33]. In a 3D area of 8870 km² in the Anyue Gasfield, the total length of the identified strike-slip faults reached up to 1860 km (Figure 8a). Generally, the width of the fault damage zone reached 400–1200 m, and the total area was about 1440 km² (Figure 9 and Figure 11), which reveal the great resource potential of the deep strike-slip fault-controlled “sweet spot” gas reservoir in the Sichuan Basin.

6. Conclusions

1. A seismic process method based on multidirectional and multiscale data decomposition and reconstruction is proposed and applicable for improving the seismic imaging of hidden (vertical displacement < 20 m) strike-slip faults in deep subsurfaces.

2. Many additional small strike-slip faults (vertical displacement less than 10 m) were identified in the central Sichuan Basin, and it was first found a pervasive distribution of strike-slip faults throughout the intracratonic basin.

3. The intracratonic strike-slip fault zone is characterized by segmentation and low maturity but presents a wide fault damage zone.

4. There is great development potential for the strike-slip faults related to the “sweet spots” of fractured reservoirs in the deep Sichuan Basin.