Multistage enhanced fault interpretation method

Abstract

The invention belongs to the technical field of petroleum geophysical prospecting data processing and interpretation, and particularly relates to a multistage enhanced fault interpretation method. The invention relates to a multistage enhanced fault interpretation method, which comprises the following steps of: on the basis of constructing guiding filtering, performing band-pass filtering on a post-stack data volume, searching a dominant frequency band of fault response, realizing first enhanced display of a fault, and then performing coherent and curvature combined application, considering anisotropy of seismic data, and performing ant tracking calculation to obtain a fault interpretation result. Second-time enhanced display of the fault is achieved; data, capable of reflecting fault information, of the coherence-ant body and the curvature-ant body are mutually fused, fault features are highlighted, and third-time enhanced display of the fault is achieved; and finally, fusing the enhanced data volume and the original seismic data in an equal proportion to realize accurate interpretation of the fault on the profile. According to the method, processing of double filtering, coherence, curvature, ant tracking and the like is carried out on the post-stack data volume, step-by-step enhancement of fault display is achieved, and the accuracy of fault interpretation is improved.

Description

Technical Field

[0001] The present invention belongs to the technical field of petroleum geophysical data processing and interpretation, and particularly relates to a fault interpretation method with multi-level enhancement. Background Art

[0002] Faults in oil and gas bearing basins play an important role in the migration and accumulation of oil and gas. Faults not only control the formation of traps such as fault noses and fault blocks, but also can serve as channels for oil and gas migration, controlling the accumulation and formation of oil and gas. Through the fine interpretation of faults, it is more conducive to the analysis and research of the laws of oil and gas enrichment. In recent years, with the application of 3D seismic data, people have carried out fault interpretation work using methods such as seismic coherence analysis, frequency division processing, and ant tracking, and achieved certain exploration results. However, due to the influence of objective factors such as acquisition and processing, the signal-to-noise ratio and resolution of seismic data cannot meet the requirements of fine structure interpretation, so that the potential of 3D seismic data cannot be fully exploited. Especially in the implementation of structures in complex structural areas and the interpretation of minor faults with a fault throw less than 10m, it is necessary to not only refer to the quality of seismic data, but also conform to the characteristics of tectonic movements in the study area, so that the distribution of faults and tectonic characteristics conform to the laws of regional geological tectonic movements. Due to the single prediction method, faults with a fault throw less than 10m cannot be accurately identified.

[0003] Currently, the existing fault interpretation mainly involves manually applying seismic profiles to observe whether there are discontinuities in seismic event axes, and whether there are changes in phase, amplitude, etc., and using attributes such as coherence and curvature to assist in interpretation. Among them, the coherent C3 algorithm is mainly used to calculate the coherence values of adjacent seismic traces to characterize faults. Its coherence results are greatly affected by seismic data noise. Due to the defects of the algorithm itself, it can only clearly identify faults with discontinuities in seismic event axes, and cannot identify minor faults with flexures in seismic event axes. Curvature is used to characterize the degree of deformation and bending at a certain point. The more severely the layer is deformed and bent, the larger the curvature value. Its seismic event axis is very sensitive to folds, flexures, etc. For those without flexural deformation on both sides of the fault, it is not clear on the curvature volume. The ant tracking technology is a newly emerging fault prediction technology in recent years. To a certain extent, it overcomes the subjectivity of fracture interpretation. It can not only clearly depict large regional faults, but also qualitatively describe small faults in the strata. The results of the ant tracking technology are greatly affected by the input data, and at the same time, it requires that the regional formation dip angle should not be too large, the seismic main frequency should not be too low, and the data signal-to-noise ratio should be high. However, due to the influence of actual seismic data and human subjective input, the results of the ant body are poor.

[0004] Patent Invention CN112068199B discloses a method for rapid fault interpretation of plane sections. First, the difference in interpreted horizons is made into a horizon trend surface, then time windows are opened along the horizon trend surface to extract coherence attributes, and plane fault lines are extracted based on the coherence attributes within the plane range; then a seismic section is selected within the range of the plane fault lines, and the fault lines are interpreted in the section, and the plane fault lines and the section fault lines intersect to form a surface; in the section, with the intersection point of the horizon fault as the rotation point, the projection fault line of the fault surface is scanned for rotation angles, and the fault line is adjusted to the position of the fault scanning angle corresponding to the maximum entropy, and all sections are traversed to obtain a finely interpreted fault surface. The present invention is guided by coherence attributes and can only identify faults with relatively large fault offsets.

[0005] Patent Invention CN1737607A discloses a method for coherence processing of dominant frequency bands for fine fault interpretation. This method uses a forward model to determine the dominant frequency band and correctly selects coherence processing parameters, and can effectively identify small faults of 4 - 8 ms in the seismic data volume. Although this method improves the accuracy of fault interpretation to a certain extent, the established forward model is an idealized model that differs from the actual geological situation. At the same time, due to the limitations of the coherence algorithm principle, its fault identification ability is limited.

[0006] Patent Invention CN112965111A discloses a fault interpretation method. This method uses the ratio of the formation thicknesses at the upper and lower fault planes of a relatively large fault to identify the types of relatively large faults within the target interval, and determines whether each relatively large fault is a single fault with only one fault plane or a multi - stage multi - segment fault with multiple fault planes; uses the maximum change rate of seismic reflection amplitude within the target interval to identify micro - faults within the target interval. The present invention is applicable to solving the phenomena of inconsistent formation attitudes between complex fault blocks and unreasonable thickness relationships in the target interval.

[0007] Patent Invention CN109001801A discloses a variable - scale fault identification method based on a multi - iterative ant colony algorithm. This method first performs noise reduction processing on seismic data, and on this basis, extracts coherence attributes using the C3 algorithm, and determines the threshold values of various parameters and weight coefficients for variable - scale fault identification through multiple experiments using the multi - iterative ant colony algorithm; this method can perform semi - quantitative variable - scale fault identification, but due to the limitations of the coherence algorithm itself, it is difficult to identify faults with in - phase axis flexures on the fault surface. Summary of the Invention

[0008] Aiming at the above problems, the object of the present invention is to provide a multi - level enhanced fault interpretation method. This method realizes the gradual enhancement of fault display through processing such as "double filtering", coherence, curvature, and ant tracking on the post - stack data volume. At the same time, considering the anisotropy of seismic data, various types of attribute data are comprehensively used to reduce the multi - solution nature of interpretation, thereby improving the accuracy of fault interpretation.

[0009] The technical solution of the present invention lies in: a multi-level enhanced fault interpretation method, including the following steps:

[0010] S1: Perform structure-oriented filtering on the post-stack seismic data volume for preliminary denoising;

[0011] S2: Carry out band-pass filtering on the seismic data volume after structure-oriented filtering, find the dominant response frequency band of the fault, and separate to obtain the frequency-divided data volume;

[0012] S3: According to the frequency-divided data volume obtained in step S2, through the C3 eigenvalue coherence algorithm, using an adaptive time window, obtain the initial coherence volume and curvature volume, and extract the plane coherence and curvature attributes along the layer to determine the distribution of the main large faults;

[0013] S4: According to the distribution of the main large faults determined in step S3, select the azimuthal data volume perpendicular to the strike of the main large faults, and repeat the process of steps S1 to S3 to obtain the final coherence volume and curvature volume;

[0014] S5: Perform ant tracking operation on the final coherence volume and curvature volume obtained in step S4, determine the threshold values of the fault identification parameters at different scales through multiple iterations, and obtain the final coherence-ant volume and curvature-ant volume;

[0015] S6: Normalize the coherence-ant volume and curvature-ant volume obtained in step S5 and fuse them into one data volume;

[0016] S7: Slice the data volume fused in step S6 along the layer, and at the same time perform proportional fusion with the original seismic data volume to achieve enhanced display of the fault, and then it can guide the development of fault profile interpretation and plane combination.

[0017] In step S1, when performing structure-oriented filtering on the post-stack seismic data volume, the specific process is as follows: determine the dip attributes of all seismic sampling points on the main survey line and connecting survey line of the post-stack seismic data volume through dip scanning, and finally obtain a data volume with dip and azimuth information at each seismic sampling point. Orient the reflection axis of the dip and azimuth data volume, and smooth the information parallel to the seismic event axis direction of the post-stack seismic data volume. If the seismic emission event axis is discontinuous horizontally, no smoothing is performed. The smoothing process is as follows:

[0018]

[0019] u(x, y, 0) = u 0 (x, y) (1)

[0020] In the formula, u(x, y, t) is the seismic wave amplitude, S is the structure tensor, G σrepresents a Gaussian function with scale σ, and ε is a continuity factor, where 0 ≤ ε ≤ 1.

[0021] In step S3, through the C3 eigenvalue coherence algorithm, an adaptive time window is used to obtain the initial coherence volume and curvature volume, and plane coherence and curvature attributes are extracted along the layers to determine the distribution of the main large faults. The specific process is as follows:

[0022] S31: Calculate the spectral component frequencies and average power spectrum data obtained in step S2 through matching pursuit. The calculation formula is as follows:

[0023]

[0024] In the formula, <R j f,xr j > is the vertical projection component of the signal Y decomposed into the best-matched atom xr j , R k+1 f is the residual value, and k is the number of decomposition times;

[0025] S32: Calculate the size of the adaptive time window at different depths according to the calculated average power spectrum data, and select a combination mode of 9 seismic traces, namely the adjacent 8 seismic traces centered on the target point, with stronger anti-noise ability for coherence to obtain the final coherence data volume. The calculation formula is as follows:

[0026]

[0027] In the formula, C3 represents the coherence attribute based on eigenvalues, λ 1 is the eigenvalue corresponding to the maximum vector, and ∑λ i is the total energy of the seismic traces within the analysis window.

[0028] In step S5, ant tracking operation is performed on the final coherence body and curvature body to obtain ant body data. The specific process is as follows: First, select the initial ant distribution boundary, with values 1, 2, 3, 4, 5, 6, 7, 9, 11, 13, 15, 20, 25, 30, ant tracking deviation, with values 0, 1, 2, 3, ant search step size, with values 2, 3, 4, 5, 6, 7, 8, 9, 10, illegal step size, with values 0, 1, 2, 3, legal step size, with values 0, 1, 2, 3, termination criterion, with values 0, 5, 10, 15, 20, 25, 30, 40, 50. Calculate a result for the median of a total of 6 parameters; then change one parameter while keeping the other parameters unchanged, take the maximum and minimum values respectively, and calculate two data volumes respectively. Compare the influence of different parameter values on the tracking result; through comparative analysis, the larger the initial ant distribution boundary value, the fewer the number of ants and the less detailed information is captured; the larger the ant tracking deviation, ant search step size, illegal step size, and termination annotation values, the denser the ant body traced, and the closer its amplitude correlation is to +1; the larger the legal step size value, the sparser the ant body traced, and the closer the amplitude correlation coefficient is to -1; for the coherence body, take a small value for the initial ant distribution boundary of ant tracking, large values for ant tracking deviation, ant search step size, illegal step size, and termination annotation, and a large value for the legal step size. The obtained coherence-ant body is mainly used to depict fractures with a fault throw greater than 10m; for the curvature body, take a large value for the initial ant distribution boundary of ant tracking, small values for ant tracking deviation, ant search step size, illegal step size, and termination annotation, and a small value for the legal step size to obtain the curvature-ant body, which is used to depict fractures with a fault throw less than 10m.

[0029] In step S6, the coherence-ant body and the curvature-ant body are normalized and fused. The fusion calculation formula is as follows:

[0030] F(A)=K 1 F(Acor) / F max (Acor)+K 2 F(Acur) / F max (Acur)(4)

[0031] In the formula, F(A) is the fusion attribute, F(Acor) is the coherence-ant body attribute, F max (Acor) is the maximum value of the coherence-ant body attribute, F(Acur) is the curvature-ant body attribute, F max (Acur) is the maximum value of the curvature-ant body attribute, K 1 、K 2 are the contribution coefficients of the coherence-ant body and the curvature-ant body, and their sum value is 1. When depicting large fractures with obvious phase axis discontinuity on the section with a fault throw greater than 10m, K 1 takes a value greater than K 2, the fused data volume mainly shows large faults; on the contrary, when depicting small faults with a fault throw less than 10m and no obvious offset of the in-phase axis on the section, mainly flexural small faults, K 2 greater than K 1 , the fused data volume mainly shows small faults; when the values of K1 and K2 are equivalent, the fused data retains the original characteristics of both data volumes.

[0032] In step S7, the data volume fused in step S6 is sliced along the layer, and at the same time, it is proportionally fused with the original seismic data volume, that is, the two data are fused together according to a proportional relationship. Specifically, part of the data in the seismic data volume and the attribute data volume that can reflect fault information are fused proportionally to generate a new data volume. The new data volume not only retains the information of the seismic data, but also contains the fault information of the attribute volume.

[0033] The technical effect of the present invention is as follows: A multi-level enhanced fault interpretation method of the present invention performs band-pass filtering on the post-stack data volume on the basis of structure-oriented filtering to find the dominant frequency band of the fault response, realizing the first enhancement display of the fault. Then, through the combined application of coherence and curvature, considering the anisotropy of seismic data at the same time, and through ant tracking calculation, the second enhancement display of the fault is realized; then, the data in the coherence-ant body and the curvature-ant body that can reflect fault information are fused with each other to highlight the fault characteristics and realize the third enhancement display of the fault; finally, the enhanced data volume is proportionally fused with the original seismic data to realize the accurate interpretation of the fault on the section.

[0034] The following will be further described with reference to the accompanying drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] Figure 1 is a flowchart of a multi-level enhanced fault interpretation method according to an embodiment of the present invention.

[0036] Figure 2 is a "double filtering" cross-section view according to an embodiment of the present invention.

[0037] Figure 3 is a coherence plan view before and after double filtering processing according to an embodiment of the present invention.

[0038] Figure 4 is a curvature plan view before and after double filtering processing according to an embodiment of the present invention.

[0039] Figure 5 is a coherence plan view according to an embodiment of the present invention.

[0040] Figure 6 is a cross-sectional view of the coherence-curvature ant body along the layer according to an embodiment of the present invention.

[0041] Figure 7This is the post - fusion Y8 cross - layer slice map of the coherent - curvature ant body in the embodiment of the present invention.

[0042] Figure 8 This is the profile map of the ratio fusion of attributes and seismic data in the embodiment of the present invention.

[0043] Figure 9 This is the plane map of fault distribution in the embodiment of the present invention. Detailed implementation manners

[0044] Example 1 As Figure 1 shown, a multi - level enhanced fault interpretation method includes the following steps:

[0045] S1: Perform structure - oriented filtering on the post - stack seismic data volume for preliminary denoising.

[0046] S2: Conduct band - pass filtering on the seismic data volume after structure - oriented filtering to find the dominant response frequency band of faults and separate the frequency - divided data volume.

[0047] S3: According to the frequency - divided data volume obtained in step S2, through the C3 eigenvalue coherence algorithm, using an adaptive time window, obtain the initial coherence volume and curvature volume, extract plane coherence and curvature attributes along the layer, and determine the distribution of major faults.

[0048] S4: According to the distribution of major faults determined in step S3, select the azimuth - divided data volume perpendicular to the strike of major faults, and repeat the process of steps S1 to S3 to obtain the final coherence volume and curvature volume.

[0049] S5: Perform ant tracking operations on the final coherence volume and curvature volume obtained in step S4, determine the threshold values of fault identification parameters at different scales through multiple iterations, and obtain the final coherent - ant body and curvature - ant body.

[0050] S6: Normalize and fuse the coherent - ant body and curvature - ant body obtained in step S5 into one data volume.

[0051] S7: Slice the data volume fused in step S6 along the layer, and at the same time perform ratio fusion with the original seismic data volume to achieve enhanced display of faults, which can guide the development of fault profile interpretation and plane combination.

[0052] In the step S1, structural orientation filtering is performed on the stacked seismic data volume. The specific process is as follows: By dip scanning, the dip attributes of all seismic sampling points on the main survey line and tie lines of the stacked seismic data volume are determined. Finally, a data volume with dip and azimuth information at each seismic sampling point is obtained. The reflection axis of the dip and azimuth data volume is oriented, and the information parallel to the direction of the seismic event axis of the stacked seismic data volume is smoothed. If the seismic event axis is discontinuous horizontally, no smoothing is performed. The smoothing process is as follows:

[0053]

[0054] u(x, y, 0) = u 0 (x, y) (1)

[0055] In the formula, u(x, y, t) is the seismic wave amplitude, S is the structure tensor, G σ represents the Gaussian function with scale σ, and ε is the continuity factor, where 0 ≤ ε ≤ 1.

[0056] In the step S3, through the C3 eigenvalue coherence algorithm, an adaptive time window is used to obtain the initial coherence volume and curvature volume, and the planar coherence and curvature attributes are extracted along the layer to determine the distribution of the main large faults. The specific process is as follows:

[0057] S31: Calculate the spectral component frequency and average power spectrum data obtained in the step S2 through matching pursuit. The calculation formula is as follows:

[0058]

[0059] In the formula, <R j f,xr j > is the vertical projection component of the signal Y decomposed into the best matching atom xr j , R k+1 f is the residual, and k is the number of decomposition times;

[0060] S32: Calculate the size of the adaptive time window at different depths according to the calculated average power spectrum data, and select the combination mode of 9 seismic traces, namely the adjacent 8 seismic traces centered on the target point, with stronger anti-noise ability for coherence to obtain the final coherence data volume. The calculation formula is as follows:

[0061]

[0062] In the formula, C3 represents the coherence attribute based on eigenvalues, λ 1 is the eigenvalue corresponding to the maximum vector, and ∑λ i is the total energy of the seismic traces within the analysis window.

[0063] In step S5, ant tracking operations are performed on the final coherence body and curvature body to obtain ant body data. The specific process is as follows: First, select the initial ant distribution boundary, with values 1, 2, 3, 4, 5, 6, 7, 9, 11, 13, 15, 20, 25, 30, ant tracking deviation, with values 0, 1, 2, 3, ant search step size, with values 2, 3, 4, 5, 6, 7, 8, 9, 10, illegal step size, with values 0, 1, 2, 3, legal step size, with values 0, 1, 2, 3, termination criterion, with values 0, 5, 10, 15, 20, 25, 30, 40, 50. Calculate a result for the median of a total of 6 parameters; then change one parameter while keeping the other parameters unchanged, take the maximum and minimum values respectively, and calculate two data volumes respectively to compare the influence of different parameter values on the tracking results; through comparative analysis, the larger the initial ant distribution boundary value, the fewer the number of ants and the less detailed information is captured; the larger the ant tracking deviation, ant search step size, illegal step size, and termination annotation values, the denser the ant body traced, and the closer its amplitude correlation is to +1; the larger the legal step size value, the sparser the ant body traced, and the closer the amplitude correlation coefficient is to -1; for the coherence body, take a small value for the initial ant distribution boundary of ant tracking, large values for the ant tracking deviation, ant search step size, illegal step size, and termination annotation, and a large value for the legal step size. The obtained coherence-ant body is mainly used to depict fractures with a fault throw greater than 10m; for the curvature body, take a large value for the initial ant distribution boundary of ant tracking, small values for the ant tracking deviation, ant search step size, illegal step size, and termination annotation, and a small value for the legal step size to obtain the curvature-ant body, which is used to depict fractures with a fault throw less than 10m.

[0064] In step S6, the coherence-ant body and the curvature-ant body are normalized and fused. The fusion calculation formula is as follows:

[0065] F(A) = K 1 F(Acor) / F max (Acor) + K 2 F(Acur) / F max (Acur)(4)

[0066] In the formula, F(A) is the fused attribute, F(Acor) is the coherence-ant body attribute, F max (Acor) is the maximum value of the coherence-ant body attribute, F(Acur) is the curvature-ant body attribute, F max (Acur) is the maximum value of the curvature-ant body attribute, K 1 、K 2 are the contribution coefficients of the coherence-ant body and the curvature-ant body, and their sum value is 1. When depicting a major fracture with obvious dislocation of the in-phase axis on a section with a fault throw greater than 10m, K 1 takes a value greater than K 2, the fused data volume is mainly characterized by large faults; conversely, when depicting small faults with a fault throw less than 10m where the in-phase axis on the section has no obvious offset and is mainly flexural, K 2 greater than K 1 , the fused data volume is mainly characterized by small faults; when the values of K1 and K2 are equivalent, the fused data retains the original characteristics of both data volumes.

[0067] In step S7, the data volume fused in step S6 is sliced along the layer, and at the same time, it is proportionally fused with the original seismic data volume, that is, the two data are fused together according to a proportional relationship. Specifically, the part of the seismic data volume and the attribute data volume that can reflect fracture information is fused proportionally to generate a new data volume. The new data volume not only retains the information of the seismic data but also contains the fracture information of the attribute volume.

[0068] Taking the 3D fault interpretation of a certain block in the coal-bearing strata as an example in Embodiment 2, a multi-level enhanced fault interpretation method described in Embodiment 1 is adopted. In this embodiment, the specific steps are as follows:

[0069] Step 1: Perform structure-oriented filtering on the 3D post-stack time migration data in the block. By calculating the dip angle and azimuth angle of the seismic reflection interface, the reflection axis is oriented, separating the effective signal from the random noise and the coherent noise in other directions intersecting the reflection axis, thus avoiding smoothing on both sides of faults, fractures or other valuable structures, ensuring its discontinuity, and improving the resolution.

[0070] Step 2: Perform spectral analysis on the data volume obtained in Step 1. The data frequency band range is 6 - 60Hz, and then band-pass filtering is performed to divide the seismic data into data volumes of different frequency bands. For example, Figure 2 as shown, the low-frequency data volume information of 6 - 12Hz is missing, and the high-frequency data seismic profile of 36 - 54Hz has distortion phenomena. The seismic data in the main frequency band of 24 - 48Hz is preferably selected. The signal-to-noise ratio of the data volume in this frequency band is higher, and the section imaging is clearer;

[0071] Step 3: Perform coherence and curvature operations on the data volume obtained in Step 2 respectively to obtain a coherence volume and a curvature volume. As Figure 3 , Figure 4 shown, Figure 3 , Figure 4 By slicing along the layer, it can be seen that after double filtering, the noise is reduced, and the imaging quality of coherence and curvature is significantly improved, determining that there are two main sets of fractures developed in the NE and NW directions in the 3D area;

[0072] Step 4: Combining the results obtained in Step 3, respectively select the split-azimuth data perpendicular to the fault strike at 45° and 135°, and repeat the operations in Steps 1 to 3 to obtain the split-azimuth coherence volume and curvature volume data, asFigure 5 As shown in Figure 5 it can be seen that along the layer coherent slice, the information on the along-layer slice of the azimuth data of the fault strike is richer and the fault is clearer;

[0073] Step Five: Perform ant tracking operations on the coherent body and the curvature body. By adjusting the ant tracking input parameters through multiple iterations, the final coherent-ant body and curvature-ant body are obtained. As shown in Figure 6 it can be clearly shown on its along-layer slice the planar distribution of faults at different levels. The simultaneous application of the two can accurately depict the fault. Through formula operations for data fusion, a fused data body with enhanced fault display is obtained. As shown in Figure 7 the fused data body can better depict the planar distribution form of the fault;

[0074] Step Six: Proportionally fuse the information in the data obtained in Step Five that can reflect the fault with the seismic data to achieve enhanced display of the fault on the fused section. As shown in Figure 8 and achieve accurate interpretation of the fault on the section and rapid planar combination. As shown in Figure 9 it is shown

[0075] The above is only a preferred specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any changes or substitutions that can be easily thought of by those skilled in the art within the technical scope disclosed by the present invention should be covered by the protection scope of the present invention.

Claims

1. A multi - level enhanced fault interpretation method, characterized in that: It includes the following steps: S1: Conduct structure - oriented filtering on the post - stack seismic data volume for preliminary denoising; S2: Conduct band - pass filtering on the seismic data volume after structure - oriented filtering to find the dominant response frequency band of the fault and separate to obtain a frequency - divided data volume; S3: According to the frequency - divided data volume obtained in step S2, through the C3 eigenvalue coherence algorithm, using an adaptive time window, obtain an initial coherence volume and a curvature volume, and extract plane coherence and curvature attributes along the layer to determine the distribution of the main large faults; S4: According to the distribution of the main large faults determined in step S3, select a cross - azimuth data volume perpendicular to the strike of the main large faults, and repeat the process of steps S1 to S3 to obtain the final coherence volume and curvature volume; S5: Conduct ant tracking operations on the final coherence volume and curvature volume obtained in step S4, determine the threshold values of fault recognition parameters at different scales through multiple iterations, and obtain the final coherence - ant volume and curvature - ant volume; S6: Normalize the coherence - ant volume and curvature - ant volume obtained in step S5 respectively and fuse them into one data volume; S7: Slice the data volume fused in step S6 along the layer, and at the same time perform proportional fusion with the original seismic data volume to achieve enhanced display of the fault, and then it can guide the development of fault profile interpretation and plane combination.

2. The multi - level enhanced fault interpretation method according to claim 1, characterized in that: In step S1, when conducting structure - oriented filtering on the post - stack seismic data volume, the specific process is as follows: Determine the dip attributes of all seismic sampling points on the main survey line and connecting survey line of the post - stack seismic data volume through dip scanning, and finally obtain a data volume with dip and azimuth information at each seismic sampling point. Orient the reflection axis of the dip and azimuth data volume, and smooth the information parallel to the seismic event axis direction of the post - stack seismic data volume. If the seismic event axis is not continuous transversely, no smoothing is performed. The smoothing process is as follows: u(x, y, 0) = u 0 (x, y) (1) where \(u(x, y, t)\) is the seismic wave amplitude, \(S\) is the structure tensor, G σ represents the Gaussian function with scale \(\sigma\), \(\varepsilon\) is the continuity factor, and \(0\leqslant\varepsilon\leqslant1\).

3. The multi - level enhanced fault interpretation method according to claim 1, characterized in that: In step S3, through the C3 eigenvalue coherence algorithm, using an adaptive time window, obtain an initial coherence volume and a curvature volume, and extract plane coherence and curvature attributes along the layer to determine the distribution of the main large faults. The specific process is as follows: S31: Calculate the spectral component frequencies and average power spectrum data obtained in step S2 through matching pursuit. The calculation formula is as follows: where <R j f, xr j > is the vertical projection component of the signal Y decomposed into the best matching atom xr j , R k+1 f is the residual value, and k is the number of decomposition times; S32: Calculate the size of the adaptive time window at different depths according to the calculated average power spectrum data, and select a combination mode of 9 seismic traces, which are the adjacent 8 seismic traces centered on the target point with stronger anti - noise ability, for coherence to obtain the final coherence data volume. The calculation formula is as follows: where C3 represents the coherence attribute based on the eigenvalue, λ 1 is the eigenvalue corresponding to the maximum vector, ∑λ i is the total energy of the seismic traces within the analysis window.

4. The multi - level enhanced fault interpretation method according to claim 1, characterized in that: In step S5, ant tracking operations are performed on the final coherence volume and curvature volume to obtain ant volume data. The specific process is as follows: First, select the initial ant distribution boundary, with values 1, 2, 3, 4, 5, 6, 7, 9, 11, 13, 15, 20, 25, 30, ant tracking deviation, with values 0, 1, 2, 3, ant search step size, with values 2, 3, 4, 5, 6, 7, 8, 9, 10, illegal step size, with values 0, 1, 2, 3, legal step size, with values 0, 1, 2, 3, termination criterion, with values 0, 5, 10, 15, 20, 25, 30, 40, 50. Calculate a result by taking the median of a total of 6 parameters; then change one parameter while keeping the other parameters unchanged, take the maximum and minimum values respectively, and calculate two data volumes respectively. Compare the influence of different parameter values on the tracking result; through comparative analysis, the larger the initial ant distribution boundary value, the fewer the number of ants and the less detailed information is captured; the larger the ant tracking deviation, ant search step size, illegal step size, and termination annotation values, the denser the ant volume traced out, and the closer the amplitude correlation is to +1; the larger the legal step size value, the sparser the ant volume traced out, and the closer the amplitude correlation coefficient is to -1; for the coherence volume, take a small value for the initial ant distribution boundary, large values for the ant tracking deviation, ant search step size, illegal step size, and termination annotation, and a large value for the legal step size. The obtained coherence-ant volume is mainly used to depict fractures with a fault throw greater than 10m; for the curvature volume, take a large value for the initial ant distribution boundary, small values for the ant tracking deviation, ant search step size, illegal step size, and termination annotation, and a small value for the legal step size. The obtained curvature-ant volume is used to depict fractures with a fault throw less than 10m.

5. A multi-level enhanced fault interpretation method according to claim 1, characterized in that: In step S6, the coherence-ant volume and the curvature-ant volume are normalized and fused. The fusion calculation formula is as follows: F(A) = K 1 F(Acor) / F max (Acor) + K 2 F(Acur) / F max (Acur) (4) Where F(A) is the fused attribute, F(Acor) is the coherence-antibody attribute, and F max (Acor) is the maximum value of the coherence-antibody attribute, F(Acur) is the curvature-antibody attribute, and F max (Acur) is the maximum value of the curvature-antibody attribute, K 1 、K 2 are the contribution coefficients of the coherence-antibody and the curvature-antibody, and their sum is 1. To depict a major fault with an obvious offset of the in-phase axis on a section with a fault throw greater than 10m, K 1 takes a value greater than K 2 , and the fused data volume mainly shows major faults; conversely, when depicting minor faults mainly characterized by flexures with no obvious offset of the in-phase axis on a section with a fault throw less than 10m, K 2 is greater than K 1 , and the fused data volume mainly shows minor faults; when the values of K1 and K2 are equivalent, the fused data retains the original characteristics of both data volumes at the same time.

6. A multi-level enhanced fault interpretation method according to claim 1, characterized in that: In step S7, the data volume after fusion in step S6 is sliced along the layer, and at the same time, it is proportionally fused with the original seismic data volume, that is, the two data are fused together according to a proportional relationship. Specifically, the part of the seismic data volume and the attribute data volume that can reflect fracture information is fused proportionally to generate a new data volume. The new data volume not only retains the information of the seismic data but also contains the fracture information of the attribute volume.

Images