A method for predicting volcanic rock fractures based on enhanced dip imaging

Through inclination imaging enhancement and fault reinforcement treatment, combined with tensor direction field constraints, the problem of low prediction accuracy of volcanic rock reservoirs is solved, and more accurate crack prediction and reservoir distribution identification is achieved, providing efficient technical support for volcanic rock seismic exploration .

CN114578421BInactive Publication Date: 2024-11-22CHINA UNIV OF PETROLEUM (EAST CHINA)
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202210172079.0
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2022-02-24
Publication Date
2024-11-22
Estimated Expiration
Not applicable · inactive patent

AI Technical Summary

Technical Problem

It is difficult to predict volcanic rock reservoirs, especially the deep data quality problems and the low quantitative prediction accuracy of reservoirs, which leads to unclear seismic reflection characteristics of volcanic rocks and difficulty in structural characterization and reservoir prediction.

Method used

Inclination imaging enhancement and fault reinforcement treatment methods are adopted, combined with tensor direction field constraint weighted fitting, the signal-to-noise ratio and resolution of seismic data are improved, and the development intensity and reservoir distribution of volcanic rock fractures are accurately predicted.

Benefits of technology

It significantly improves the accuracy and reliability of volcanic reservoir prediction, clarifies the description and channel location of hidden craters, provides reliable data support, and provides an important basis for efficient exploration and development of oil and gas fields.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN114578421B_ABST
    Figure CN114578421B_ABST
Patent Text Reader

Abstract

The present invention relates to the technical field of oil and gas seismic exploration, and discloses a method for predicting volcanic rock fractures based on dip imaging enhancement. The method comprises: S1 dip imaging enhancement: multi-scale and multi-azimuth dip imaging enhancement is performed on the original seismic data, and the similarity of adjacent traces is calculated by using the changes in dip and azimuth to improve the lateral signal-to-noise ratio of the seismic data; S2 fault enhancement processing: fault enhancement processing is performed on the original seismic data to improve the vertical resolution of the seismic data and improve the identifiability of the fault; S3 fracture prediction based on tensor directional field constraints: on the basis of the seismic data after dip imaging enhancement and fault enhancement processing, volcanic rock fracture prediction is performed based on tensor field directional constraint weighted fitting, the fracture development intensity is predicted, and the volcanic rock reservoir distribution area is finally determined in combination with regional geological analysis. The present invention greatly improves the accuracy of fracture prediction, and for the description and identification of hidden craters, the volcanic morphology is clearer and the location of the volcanic channel is more accurate.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The invention relates to the technical field of oil and gas seismic exploration, and in particular to a method for predicting volcanic rock fracture pore type reservoirs, so as to improve the prediction accuracy of volcanic rock reservoirs and optimize favorable well location areas. Background Art

[0002] With the increasing number of volcanic oil and gas reservoirs discovered, volcanic rocks have become a new field for oil and gas exploration. However, due to the deep burial depth of volcanic rocks and the influence of factors such as volcanic eruption periods and structural deformation, the internal strata and sequences of volcanic rocks are very complex, resulting in chaotic reflections in seismic data, unclear boundaries between different volcanic facies belts and lithology combinations, serious overlap of wave impedance characteristics, and the fact that volcanic reservoir types are mostly dual media with diverse reservoir lithologies, making it difficult to predict volcanic reservoirs. The prediction of high-quality volcanic reservoirs has become a difficulty in the efficient development of volcanic gas reservoirs. The accuracy and precision of reservoir prediction determine the benefits of volcanic gas reservoir development. At present, the reservoir prediction methods for volcanic rocks in the industry are relatively single, and the prediction effect of a single method is low in accuracy.

[0003] At present, the macroscopic exploration methods of volcanic rocks are limited to finding volcanic rock bodies. Gravity exploration, magnetic exploration, acoustic magnetic field method, amplitude, phase and frequency analysis of composite channels and other technologies are mainly used to study the thickness distribution, lithofacies and physical properties of underground volcanic rocks; microscopic research on the petrological characteristics of volcanic rocks, diagenesis and its impact on reservoir physical properties is relatively detailed. Post-stack seismic data is the most commonly used data in the oil industry. Most of the previous fracture prediction techniques can only be used to qualitatively describe the degree of fracture development. However, for example, the deep volcanic gas reservoirs in the Junggar Basin have complex underground geological conditions, strong heterogeneity, rapid changes in lithology and lithofacies, and complex seismic reflection characteristics. Conventional methods are difficult to meet the requirements of actual production.

[0004] There are mainly the following problems in reservoir prediction of volcanic rocks:

[0005] (1) Deep data quality issues. Due to the deep burial of volcanic rocks, the formation matrix is ​​variable, the lithology is complex, and the lithofacies vary greatly, resulting in a low overall data signal-to-noise ratio, unclear seismic response characteristics of volcanic rocks, and great difficulty in characterizing volcanic structures and predicting subsequent volcanic reservoirs;

[0006] (2) The quantitative prediction accuracy of volcanic reservoirs is low.

[0007] Conventional reservoir prediction methods are mostly based on phase-controlled inversion for quantitative prediction of volcanic rocks. However, for fracture-pore volcanic reservoirs, the rock composition and pore structure of volcanic rocks in different lithofacies vary greatly, and the relationship between reservoir physical properties and elastic parameters is also different. Conventional inversion cannot effectively predict high-quality reservoirs. Summary of the invention

[0008] The main technical problem to be solved by the present invention is to provide a volcanic rock fracture reservoir prediction method to avoid the influence of the volcanic rock reservoir prediction accuracy existing in the traditional method, to improve the prediction accuracy of the volcanic rock high-quality reservoir, to optimize the favorable drilling target area, to reduce the drilling risk of the research target, and to provide reliable data for the efficient exploration and development of oil and gas fields.

[0009] In order to solve the above technical problems, the technical solution adopted by the present invention is: a volcanic rock fracture prediction method enhanced by dip imaging, comprising the steps of:

[0010] S1 dip imaging enhancement: Perform multi-scale and multi-azimuth dip imaging enhancement on the original seismic data, and use the changes in dip and azimuth to calculate the similarity of adjacent traces to improve the lateral signal-to-noise ratio of the seismic data;

[0011] S2 Fault Enhancement Processing: Perform fault enhancement processing on the original seismic data to improve the vertical resolution of the seismic data and improve the fault identifiability;

[0012] S3 Fracture prediction based on tensor directional field constraints: Based on the seismic data after dip imaging enhancement and fault strengthening, volcanic rock fracture prediction is carried out based on tensor field directional constraint weighted fitting to predict the fracture development intensity, and finally determine the volcanic reservoir distribution area in combination with regional geological analysis.

[0013] Preferably, the step S1 comprises:

[0014] S11: Based on the seismic 3D data volume, define the sub-azimuth and sub-scale characteristics of the seismic data as ψ u,v (x),

[0015]

[0016] In formula (1), i is a complex operator, σ is the filter bandwidth, and k u,v =k v (cosθ,sinθ) T , k v =2 (2-v) / 2 π, θ = u·π / K, v corresponds to the scale of the filter, u corresponds to the direction of the filter, ‖*‖ represents the modulus;

[0017] For a given 3D seismic data I(x), for a given arbitrary spatial point x = (in, cl, t), seismic data of different orientations and scales are defined as D u,v (x):

[0018] D u,v (x) = ψ u,v (x)*I(x) (2),

[0019] In formula (2), D u,v(x) is the seismic data of different orientations and scales obtained by using filters determined by different (u, v);

[0020] The seismic data were jointly analyzed using three scales and eight directions to obtain D u,v (x) has 24 forms of expression:

[0021]

[0022] In formula (3), variables m and n are subscript index variables;

[0023] S12: For D u,v (x) is encoded,

[0024]

[0025]

[0026] In formula (4) and (5), and D u,v (x) Binary encoding of multi-scale and multi-directional filtering data under (u, v) parameters and calculation of synthetic data and

[0027]

[0028]

[0029] In formula (6) and (7), and To process through filters of different scales and orientations;

[0030] S13: According to the correlation of the decomposed data, the correlation matrix is ​​defined as:

[0031]

[0032]

[0033] Will It can be expressed as:

[0034]

[0035] In formula (10), v is the scale subscript, for has three non-negative eigenvalues ​​and satisfies They are The corresponding eigenvector, Construct a local orthogonal coordinate system, is the direction with the largest contrast in the local area, Into a perpendicular The local plane of Find the inclination property:

[0036]

[0037] Will It can be expressed as:

[0038]

[0039] In formula (12), v is the scale subscript, for has three non-negative eigenvalues ​​and satisfies They are The corresponding eigenvector, Construct a local orthogonal coordinate system, is the direction with the largest contrast in the local area, Into a perpendicular The local plane of Find the inclination property:

[0040]

[0041] The optimal dip angle dip(x) is:

[0042]

[0043] Preferably, the step S2 comprises:

[0044] For the original seismic data x, select a 3×3 sub-block Y(x) along the dip(x) direction, and define the block histogram of Y(x) as:

[0045]

[0046] In formula (15), is the multi-scale and multi-directional filtering data obtained in step S1, i=1, 2, ...k, k is a pre-set data level;

[0047] Perform matrix decomposition on Y(x),

[0048]

[0049] Sort all eigenvalues, λ i >λ i+1, all variations are defined by the first eigenvector, such as λ 1 >>λ 2 =λ 3 ≈0, then the messy attribute value c=-1; if the three eigenvectors are equal λ 1 =λ 2 =λ 3 , then c=+1; if there is a boundary λ 1 =λ 2 >>λ 3 =0, then c = 0,

[0050]

[0051] In the filter tilt estimation, a set of overlapping windows is determined to ensure the best plane value is:

[0052]

[0053] In formula (18), r is the optimal plane value of the filter reflector, and the range of r is between r = 1 / 3 and r = 1. Then the plane calculation value of the filter reflector is:

[0054]

[0055] (19) The plane calculation value for the filter reflector is until the plane calculation value for the filter reflector Then the edge eigenvalue f is:

[0056]

[0057] In formula (20), for the boundary f = 1, then λ 1 ≈λ 2 >>λ3≈0, for the boundary f=0, then λ 1 >>λ 2 =λ 3 ≈0, or λ 1 ≈λ 2 ≈λ 3 ,

[0058] Combination And f to define the similarity property:

[0059]

[0060] In formula (21), s is the similarity attribute, f is the edge feature value, Calculates values ​​for the plane of the filtered reflector.

[0061] Preferably, step S3 comprises:

[0062] The similarity attributes calculated in steps S1 and S2 are used as the input data of the crack model, and the directional kernel function of the crack model is K m , the directional kernel function of the similarity attribute is K coh , generate a new directional kernel function K, let K = K coh *K m , by weighting the direction of the cracks in the simulation window and then optimizing the cluster fitting, the tensor-direction-based crack simulation properties are obtained;

[0063] Based on the fracture simulation attributes, the relative fracture density Fden and the main fracture development orientation Fazi are calculated:

[0064]

[0065]

[0066] Formula (22) is used to calculate the relative crack density. Assume that the region is divided into M blocks with a block radius of R. There is a crack number N in each radius. f ,max(N f ) represents the maximum value within the area, and the number of cracks N within the unit radius R is calculated. f Get the relative crack density;

[0067] Formula (23) is used to calculate the main orientation of fracture development, azi is the fracture azimuth attribute, coh is the similarity attribute, L is the fracture length, i represents the number of fractures, fi represents the i-th fracture, and coh i and L i As the weight attribute constraint of the i-th crack.

[0068] Compared with the prior art, the present invention has the following beneficial effects:

[0069] The present invention proposes a method for predicting volcanic rock fractures based on dip imaging enhancement. First, the original seismic data is specially processed, and multi-scale and multi-directional dip control enhancement is performed to improve the lateral signal-to-noise ratio of the seismic data; secondly, the seismic data is subjected to fault enhancement processing to improve the vertical resolution of the seismic data and retain its discontinuous wave group characteristics, thereby smoothing the seismic data and making discontinuities (such as faults) easier to identify; then, volcanic rock fractures are predicted based on tensor field direction constraint weighted fitting, the distribution area of ​​volcanic rock high-quality reservoirs is predicted, and volcanic rock reservoir classification evaluation is performed. The method of the present invention greatly improves the accuracy of fracture prediction. Compared with conventional prediction methods, the description and identification of hidden craters make the volcanic morphology clearer and the location of the volcanic channel more accurate, providing reliable support for subsequent prediction of high-quality volcanic rock reservoirs, etc., and has very important guiding significance for future volcanic rock seismic exploration technology.

[0070] The method of the present invention makes full use of the rich information of drilling, logging data and seismic data in the study area and fully exploits the information of seismic data. The volcanic reservoir prediction method based on dip imaging enhancement of the present invention verifies that the well matching rate reaches 80%. BRIEF DESCRIPTION OF THE DRAWINGS

[0071] Figure 1 The present invention is a flow chart of a method for predicting volcanic rock fractures based on dip angle imaging enhancement.

[0072] Figure 2 This is a comparison diagram of the sections before and after fault enhancement processing.

[0073] Figure 3 Schematic diagram of volcanic rock fracture prediction based on the method of the present invention. DETAILED DESCRIPTION

[0074] In order to facilitate a further understanding of the method of the present invention and the effects achieved, the present invention is described in detail below with reference to the accompanying drawings. However, the accompanying drawings are only provided for reference and illustration and are not intended to limit the present invention.

[0075] The present invention provides a volcanic rock fracture reservoir prediction method based on dip imaging enhancement. The method improves the data signal-to-noise ratio and fault recognizability by performing dip imaging enhancement and fault enhancement processing on the original seismic data. Then, fracture prediction based on tensor directional field constraints is performed on the specially processed data to highlight the volcanic rock geological targets, improve the fracture prediction accuracy, and provide an important basis for the characterization of volcanic rock structures. The present invention is achieved by the following steps:

[0076] S1 dip imaging enhancement: Multi-scale and multi-azimuth dip imaging enhancement is performed on the original seismic data. The similarity of adjacent traces is calculated using the changes in dip and azimuth to improve the lateral signal-to-noise ratio of the seismic data.

[0077] S2 Fault Enhancement Processing: Perform fault enhancement processing on the original seismic data to improve the vertical resolution of the seismic data and improve the identifiability of faults.

[0078] S3 Fracture prediction based on tensor directional field constraints: Based on the seismic data after dip imaging enhancement and fault strengthening, volcanic rock fracture prediction is carried out based on tensor field directional constraint weighted fitting to predict the fracture development intensity, and finally determine the volcanic reservoir distribution area in combination with regional geological analysis.

[0079] The present invention mainly provides an important basis for comprehensive evaluation and classification of high-quality volcanic reservoirs by integrating various information, including well logging information, mud logging information and seismic information. The specific implementation methods of the present invention are as follows:

[0080] S1 tilt imaging enhancement: data enhancement processing based on multi-scale azimuth adaptive transformation

[0081] This step uses the changes in dip and azimuth to calculate the similarity of adjacent traces, improves the seismic lateral signal-to-noise ratio, and because it uses the method of calculating dip and azimuth, the result significantly enhances the ability to characterize faults, which is conducive to subsequent research on faults and fractures. In the calculation process, multi-scale azimuth change decomposition data is used for each sampling point for fusion analysis, and each sampling point involved in the calculation is statistically calculated by block histograms. The dip calculation method is mainly to determine the dip of the formation by finding the direction with the largest change in seismic data through a small analysis window. The methods are mainly divided into:

[0082] S11: Based on the seismic 3D data volume, a multi-scale azimuth transformation is defined. Since the traditional gradient structure tensor (GST) only uses a single point and a single direction calculation method, the result is easily affected by noise and has low accuracy. This time, considering the spatial combination method, a multi-scale azimuth change method is used to improve the resolution and signal-to-noise ratio of the data:

[0083]

[0084] In formula (1), i is a complex operator, σ is the filter bandwidth, and k u,v =k v (cosθ,sinθ) T , k v =2 (2-v) / 2 π, θ = u·π / K, v corresponds to the scale (frequency) of the filter, u corresponds to the direction of the filter, and ‖*‖ represents the modulus. Different filter directions can extract different azimuth and scale characteristics of seismic data.

[0085] For a given 3D seismic data I(x), for a given arbitrary spatial point x = (in, cl, t), the sub-azimuth and sub-scale characteristics of the seismic data are defined as the Gabor transform ψ u,v (x) is convolved with the seismic data I(x) to obtain seismic data of different directions and scales:

[0086] D u,v (x) = ψ u,v (x)*I(x) (2)

[0087] In formula (2), D u,v (x) is the multi-scale azimuth data obtained by the filter determined by different (u, v). In addition, the combination method and number of (u, v) need to be determined in advance by the seismic interpreter. In the actual application process, 3 scales and 8 directions are used to jointly analyze the seismic data to obtain D u,v (x) in 24 forms, namely:

[0088]

[0089] In formula (3), variables m and n are subscript index variables.

[0090] S12: Based on step S11, the optimal fusion data is calculated using the encoding strategy.

[0091] For the obtained multi-scale orientation data set D u,v (x) is encoded as follows:

[0092]

[0093]

[0094] In formula (4) and (5), and The data set D u,v (x) Binary encoding of multi-scale and multi-directional filtering data under (u, v) parameters and calculation of synthetic data and Right now:

[0095]

[0096]

[0097] In formula (6) and (7), the data and Through filtering processing at different scales and orientations, the correlation between adjacent seismic trace data is reduced, and the lateral difference characteristics of different orientations between adjacent traces of seismic data are highlighted. Compared with the original seismic data, the dip of the stratum can be better determined based on the direction in which the seismic data changes the most.

[0098] S13: Find the optimal inclination angle according to the correlation of the decomposed data, where the correlation matrix is ​​defined as:

[0099]

[0100]

[0101] Will Expressed in terms of eigenvalues ​​and eigenvectors, then:

[0102]

[0103] In formula (10), v is the scale subscript, for has three non-negative eigenvalues ​​and satisfies They are The corresponding feature vector. Construct a local orthogonal coordinate system, is the direction with the largest contrast in the local area, Into a perpendicular The local plane of The inclination property can be obtained:

[0104]

[0105] Will Expressed in terms of eigenvalues ​​and eigenvectors, then:

[0106]

[0107] In formula (12), v is the scale subscript, for has three non-negative eigenvalues ​​and satisfies They are The corresponding feature vector. Construct a local orthogonal coordinate system, is the direction with the largest contrast in the local area, Into a perpendicular The local plane of The inclination property can be obtained:

[0108]

[0109] Therefore, the optimal dip angle dip(x) is:

[0110]

[0111] S2 fault enhancement processing.

[0112] First, different methods of improving the signal-to-noise ratio were tested on the inner strata of the volcanic rock to enhance the identifiability of the faults. The main technical idea is to use the geometric attributes guided by the discontinuity of the event axis to highlight the fault identification, and to perform normalization processing on the basis of enhanced dip filtering and increased fractures, which can highlight small-scale fractures while retaining the information of large-scale fractures. Here, the joint attributes of the block histogram and the multi-scale multi-directional filtering dip body are used to predict fractures.

[0113] First, calculate the block histogram and select a 3×3 sub-block Y(x) along the dip(x) direction for the original seismic data x. Define the block histogram of Y(x) as:

[0114]

[0115] In formula (15), is the multi-scale and multi-directional filtering data mentioned above, i=1,2,...k, k is the pre-set data level, that is, the seismic data is divided into k levels.

[0116] Perform matrix decomposition on Y(x),

[0117]

[0118] Sort all eigenvalues, λ i >λ i+1 , where all variations are defined by the first eigenvector, such as λ 1 >>λ 2 =λ 3 ≈0, then the messy attribute value c=-1; if the three eigenvectors are equal λ 1 =λ 2 =λ 3 , then c=+1; if there is a boundary λ 1 =λ 2 >>λ 3 =0, then c = 0. The calculation is as follows:

[0119]

[0120] In the filter tilt estimation, a set of overlapping windows is determined to ensure the best plane value, which is:

[0121]

[0122] In formula (18), r is the optimal plane value of the filter reflector, and the range of r is between r = 1 / 3 (for random reflectors) and r = 1 (for planar reflectors), then:

[0123]

[0124] Until the random reflector is Then the edge feature can be simplified to:

[0125]

[0126] In formula (20), the range is for the boundary f = 1, and its λ 1 ≈λ 2 >>λ3≈0, for the boundary f=0, then λ 1 >>λ 2 =λ 3 ≈0, or λ 1 ≈λ 2 ≈λ 3 .

[0127] In order to better represent the edge geometric properties, combined And f to define the similarity attribute:

[0128]

[0129] In formula (21), f is the edge eigenvalue, Calculates values ​​for the plane of the filtered reflector.

[0130] S3 Crack prediction based on tensor field direction constraint weighted fitting

[0131] Based on the seismic data after special processing (processing in steps S2 and S3), the third-generation intrinsic coherence calculation is performed (a method for quantitatively calculating waveform similarity, which is achieved by defining a "global" aperture in time and space and using the calculation of inclination and azimuth).

[0132] Assuming that the crack direction distribution conforms to the Gaussian function, by establishing crack models with different directions, the direction kernel function is established to simulate the direction of the crack distribution tensor field of each point within the unit time window. The similarity attributes calculated in steps S1 and S2 are used as the input data of the crack model, and the direction kernel function of the crack model is set to K m , the directional kernel function K of the similarity attribute coh , generate a new directional kernel function, let K = K coh *K m By weighting the direction of the cracks in the simulation window and then performing optimal cluster fitting (the direction weighting and optimal cluster fitting can be performed using conventional mathematical methods), the tensor-direction-based crack simulation properties are obtained.

[0133] Based on the fracture simulation properties, the relative fracture density (Fden) and the main orientation of fracture development (Fazi) are calculated, and finally the fracture development intensity is obtained (fracture density can represent the fracture development intensity, and the fracture development intensity is greater where the density is greater). The specific formula is as follows:

[0134]

[0135]

[0136] Formula (22) is used to calculate the relative crack density. Assume that the region is divided into M blocks with a block radius of R. There is a crack number N in each radius. f ,max(N f ) represents the maximum value within the area, and the number of cracks N within the unit radius R is calculated. f Get the relative crack density;

[0137] Formula (23) is used to calculate the main orientation of fracture development, azi is the fracture azimuth attribute, coh is the similarity attribute, L is the fracture length, i represents the number of fractures, fi represents the i-th fracture (each fracture has an orientation attribute), and cohi and Li are the weight attribute constraints of the i-th fracture.

[0138] like Figure 2 and Figure 3 As shown, by comparing the specific embodiments of the method of the present invention with the prior art, the method of the present invention has better effects.

[0139] Figure 2 This is a comparison between the profile after fault enhancement processing (right) and the original seismic profile (left). From the comparison profile, it can be seen that after fault enhancement processing, the signal-to-noise ratio of the seismic data is significantly improved, the fault is confirmed, the fracture is more crisp and neat, and the profile breakpoints are clearer.

[0140] Figure 3 In the study, the prediction of volcanic rock fractures based on tensor directional field constraints was carried out on the data of fault enhancement processing. From the plane map, it can be seen that the characteristics of central craters and central-fissure craters are clear, and the volcanic structure is more accurately described. The central craters are distributed in a ring-shaped concave shape on the plane map of fracture properties, and the central-fissure craters are distributed in clusters with fracture development zones.

[0141] The above embodiments are only preferred embodiments for fully illustrating the present invention, and the protection scope of the present invention is not limited thereto. Any equivalent substitution or change made by a person skilled in the art based on the present invention is within the protection scope of the present invention.

Claims

1. A volcanic rock fracture prediction method based on dip imaging enhancement, characterized in that: Includes steps: S1 dip imaging enhancement: Perform multi-scale and multi-azimuth dip imaging enhancement on the original seismic data, and use the changes in dip and azimuth to calculate the similarity of adjacent traces to improve the lateral signal-to-noise ratio of the seismic data; S2 Fault Enhancement Processing: Perform fault enhancement processing on the original seismic data to improve the vertical resolution of the seismic data and improve the fault identifiability; S3 Fracture prediction based on tensor directional field constraints: Based on the seismic data after dip imaging enhancement and fault strengthening, volcanic rock fracture prediction is carried out based on tensor field directional constraint weighted fitting, the fracture development intensity is predicted, and the volcanic rock reservoir distribution area is finally determined in combination with regional geological analysis; The step S1 comprises: S11: Based on the seismic 3D data volume, define the sub-azimuth and sub-scale characteristics of the seismic data as ψ u,v (x), In formula (1), i is a complex operator, σ is the filter bandwidth, and k u,v =k v (cosθ,sinθ) Τ , k v =2 (2-v) / 2 π, θ = u·π / K, v corresponds to the scale of the filter, u corresponds to the direction of the filter, and ||*|| represents the modulus; For a given 3D seismic data I(x), for a given arbitrary spatial point x = (in, cl, t), seismic data of different orientations and scales are defined as D u,v (x): D u,v (x)=ψ u,v (x)*I(x) (2), In formula (2), D u,v (x) is the seismic data of different orientations and scales obtained by using filters determined by different (u, v); The seismic data were jointly analyzed using three scales and eight directions to obtain D u,v (x) has 24 forms of expression: In formula (3), variables m and n are subscript index variables; S12: For D u,v (x) is encoded, In formula (4) and (5), and D u,v (x) Binary encoding of multi-scale and multi-directional filtering data under (u, v) parameters and calculation of synthetic data and In formula (6) and (7), and To process through filters of different scales and orientations; S13: According to the correlation of the decomposed data, the correlation matrix is ​​defined as: Will It can be expressed as: In formula (10), v is the scale subscript, for has three non-negative eigenvalues ​​and satisfies They are The corresponding eigenvector, Construct a local orthogonal coordinate system, is the direction with the largest contrast in the local area, Into a perpendicular The local plane of Find the inclination property: Will It can be expressed as: In formula (12), v is the scale subscript, for has three non-negative eigenvalues ​​and satisfies They are The corresponding eigenvector, Construct a local orthogonal coordinate system, is the direction with the largest contrast in the local area, Into a perpendicular The local plane of Find the inclination property: The optimal dip angle dip(x) is: The step S2 comprises: For the original seismic data x, select a 3×3 sub-block Y(x) along the dip(x) direction, and define the block histogram of Y(x) as: In formula (15), is the multi-scale and multi-directional filtering data obtained in step S1, i=1, 2, ...k, k is a pre-set data level; Perform matrix decomposition on Y(x), Sort all eigenvalues, λ i >λ i+1 , all changes are defined by the first eigenvector, such as λ1>>λ2=λ3≈0, then the messy attribute value c=-1; if the three eigenvectors are equal λ1=λ2=λ3, then c=+1; if there is a boundary λ1=λ2>>λ3=0, then c=0, In the filter tilt estimation, a set of overlapping windows is determined to ensure the best plane value is: In formula (18), r is the optimal plane value of the filter reflector, and the range of r is between r = 1 / 3 and r = 1. Then the plane calculation value of the filter reflector is: (19) The plane calculation value for the filter reflector is until the plane calculation value for the filter reflector Then the edge eigenvalue f is: In formula (20), for the boundary f = 1, then λ1≈λ2>>λ3≈0, for the boundary f = 0, then λ1>>λ2=λ3≈0, or λ1≈λ2≈λ3, Combination And f to define the similarity attribute: In formula (21), s is the similarity attribute, f is the edge feature value, Calculate the value for the plane of the filter reflector; The step S3 comprises: The similarity attributes calculated in steps S1 and S2 are used as the input data of the crack model, and the directional kernel function of the crack model is K m , the directional kernel function of the similarity attribute is K coh , generate a new directional kernel function K, let K = K coh *K m , by weighting the direction of the cracks in the simulation window and then optimizing the cluster fitting, the tensor direction-based crack simulation properties are obtained; Based on the fracture simulation attributes, the relative fracture density Fden and the main fracture development orientation Fazi are calculated: Formula (22) is used to calculate the relative crack density. Assume that the region is divided into M blocks with a block radius of R. There is a crack number N in each radius. f ,max(N f ) represents the maximum value in the area, and the number of cracks N within the unit radius R is calculated. f Get the relative crack density; Formula (23) is used to calculate the main orientation of fracture development, azi is the fracture azimuth attribute, coh is the similarity attribute, L is the fracture length, i represents the number of fractures, fi represents the i-th fracture, and coh i and L i As the weight attribute constraint of the i-th crack.