Micro-structure seismic exploration detection method and system based on pre-stack depth migration

Abstract

The present application relates to oil and gas exploration technology field, especially to a microstructure seismic exploration detection method and system based on prestack depth migration, including generating equivalent velocity model set, calculating velocity uncertainty data, probabilistic detection of microstructure, and result fusion and display; the present application is relative to the prior art blindly pursuing an absolute correct velocity model, but ultimately still cannot overcome the influence of model error on the identification of small structure, the present application changes the velocity model error from a kind of interference needing minimization to a kind of valuable detection signal; by automatically generating multiple sets of globally equivalent velocity models and analyzing their uncertainty, those areas most unstable in the modeling process become direct evidence indicating the existence of microstructure, realizing the paradigm shift from eliminating error to using error, significantly improving the detection ability of microstructure.

Description

Technical Field

[0001] This invention relates to the field of oil and gas exploration technology, and in particular to a method and system for micro-scale structural seismic exploration based on pre-stack depth migration. Background Technology

[0002] In the field of geophysical exploration, pre-stack depth migration is a key technology for imaging complex structures, and its accuracy heavily depends on the accuracy of the subsurface velocity model. As oil and gas exploration targets increasingly concealed reservoirs, effectively identifying micro-structures such as small faults, micro-blocks, and lithological boundaries has become a core issue directly affecting drilling success rates and reservoir evaluation accuracy. These micro-structures are typically small in scale and have weak seismic responses, making them difficult to identify directly on conventional seismic profiles, but they play a crucial role in controlling fluid blockage and migration.

[0003] For a long time, the industry has generally adopted a strategy of pursuing the construction of an "optimal" or "perfect" velocity model, aiming to achieve optimal flattening of seismic gathers through iterative inversion and to identify structures based on this single model through migration imaging. However, geophysical inversion itself has strong ambiguity, and any velocity model inevitably contains errors. This pursuit of a "unique solution" has led to the deliberate neglect or minimization of the inherent uncertainties in the velocity modeling process, which contain rich geological information. The direct consequence is that imaging results based on a single velocity model cannot distinguish between artifacts caused by velocity errors and real micro-tectonic responses, resulting in either missed detections of small faults or low reliability in their detection, failing to provide a reliable basis for pre-drilling risk assessment.

[0004] To address the aforementioned issues, this invention proposes a micro-scale structural seismic exploration and detection method and system based on pre-stack depth migration. Summary of the Invention

[0005] To overcome the problems mentioned in the background art, the present invention proposes a micro-scale structural seismic exploration and detection method and system based on pre-stack depth migration.

[0006] The technical solution of this invention is: a micro-scale structural seismic exploration and detection method based on pre-stack depth migration, comprising the following steps:

[0007] S11: Generate a set of equivalent velocity models. Based on seismic data, generate multiple sets of equivalent velocity models that can flatten seismic gathers globally and are within the allowable error range.

[0008] S12: Calculate the velocity uncertainty data volume by performing point-by-point statistics on the equivalent velocity model set, calculating the standard deviation of the velocity value at each spatial location point, and generating a three-dimensional velocity uncertainty data volume.

[0009] S13: Probabilistic detection of micro-structures, interpreting high-value regions in velocity uncertainty data as probability regions where micro-structures exist, where high standard deviations are associated with a high probability of micro-structures.

[0010] S14: Results fusion and display, which overlays the velocity uncertainty data volume with the seismic migration profile and seismic attribute volume to comprehensively verify the micro-structure detection results.

[0011] Preferably, the generation of the equivalent velocity model set specifically includes:

[0012] S21: Obtain initial modeling data, including pre-stack seismic gather data, initial velocity model, and stratigraphic data for seismic interpretation;

[0013] S22: Define the perturbation parameters and range, determine the parameter types, perturbation space, and allowable range of perturbation amount for perturbing the initial velocity model, and ensure that the perturbation amount can still flatten the seismic gathers globally after the perturbation.

[0014] S23: Perform cyclic perturbation and model verification;

[0015] S24: Set generation determination, repeatedly execute loop perturbation and model verification until the number of models in the equivalent speed model set reaches the preset value.

[0016] Preferably, the cyclic perturbation and model validation process includes the following:

[0017] S231: Within the perturbation allowable range, generate a random perturbation field for the current cycle;

[0018] S232: Apply a random perturbation field to the initial velocity model to obtain a candidate velocity model;

[0019] S233: Seismic migration imaging using candidate velocity models and calculation of gather flattening;

[0020] S234: Determine whether the flatness of the track gather meets the preset standard. If it does, store the candidate velocity model in the equivalent velocity model set; if it does not, discard the model.

[0021] When defining the disturbance parameters and range, the disturbance parameters include the velocity value itself, the lateral gradient of the velocity along the layer, and the depth of the formation reflection interface.

[0022] In this process, when a random perturbation field is applied to the initial velocity model to obtain a candidate velocity model, the velocity field and the reflection interface are coupled and perturbed. That is, when the depth of the local layer interface is perturbed, the corresponding layer velocity is also adjusted accordingly.

[0023] Preferably, the generation of the random perturbation field for the current cycle specifically includes:

[0024] S31: Define the basic perturbation field and generate a standard Gaussian random field with a mean of 0 and a variance of 1;

[0025] S32: Apply the spatial correlation function, define the covariance function that characterizes spatial correlation, and use this function to perform convolution filtering on the basic perturbation field to generate a correlated random field with a specific spatial structure;

[0026] S33: Apply perturbation amplitude control, multiply the relevant random field by a perturbation amplitude control function based on spatial variation, and generate the final random perturbation field.

[0027] Specifically, when applying the spatial correlation function, the covariance function adopts an anisotropic exponential function, with the following form:

[0028] ;

[0029] in, and For point and points In the horizontal coordinates, and For point and points In the vertical direction coordinates, For horizontal correlation length, This is the vertical correlation length.

[0030] Preferably, the horizontal correlation length Directionality is achieved through coordinate rotation, specifically:

[0031] Let u be the direction along the strike of the geological structure and v be the direction along the dip. Then the covariance function in the horizontal direction becomes:

[0032] ;

[0033] in, , and It is a point and points The coordinate difference in the rotated coordinate system The relevant length along the direction, For the relevant length along the dip, The vertical correlation length.

[0034] Preferably, when applying the spatial correlation function, the process of applying the spatial correlation function adopts a multi-scale strategy, specifically as follows:

[0035] A correlated random field is a linear superposition of multiple sub-random fields with different correlation scales:

[0036] ;

[0037] in, For the relevant random field, For the i-th correlated random field generated by a covariance function with a specific correlation length, These are the corresponding weighting coefficients.

[0038] Furthermore, the application of spatial correlation functions is constrained by the geological framework, specifically:

[0039] The shape and parameters of the covariance function are controlled by the geological framework model. Within continuous sedimentary strata, a larger correlation length is used to ensure the smoothness of the disturbance. At preset faults or lithological boundaries, the correlation length of the covariance function is set to a smaller value or directly set to zero to allow the disturbance field to produce discontinuities on both sides of the boundary.

[0040] Specifically, the quality framework constraint is achieved by constructing a guiding operator, which is defined by the formation normal field, making the covariance function more correlated along the formation plane direction and less correlated in the direction perpendicular to the formation plane.

[0041] Preferably, when defining the fundamental perturbation field, the generation of the fundamental perturbation field adopts the Fast Fourier Transform method, and the specific process is as follows:

[0042] Generate a complex random field in the frequency domain, with a uniform phase distribution on [0, 2π] and an amplitude of a random variable satisfying a specific energy spectrum distribution;

[0043] By performing an inverse Fourier transform on the complex random field, the fundamental perturbation field in the spatial domain is obtained.

[0044] Preferably, when calculating data volumes with uncertain computational speed, the specific methods include:

[0045] S41: Obtain the equivalent velocity model set. Obtain a model set consisting of N sets of equivalent velocity models, where each model has a defined velocity value at a grid point in three-dimensional space.

[0046] S42: Calculate statistics point by point. For each grid point in 3D space, perform the following:

[0047] Data extraction: Extract the M velocity values ​​corresponding to the grid point from the model set to form a velocity value sample set;

[0048] Calculate the statistic: Based on the velocity value sample set, calculate the statistic that characterizes the degree of dispersion of velocity values ​​at that point;

[0049] S43: Generate an uncertain data volume by assigning the calculated statistics at each grid point to that point, thus generating a three-dimensional data volume with the same dimensions as the original velocity model, which is the velocity uncertainty data volume.

[0050] Specifically, when calculating statistics, the standard deviation is included, and the formula is as follows:

[0051] ;

[0052] in, Standard deviation For the velocity value samples of the i-th model, Let be the average velocity value of N models at point 𝑥.

[0053] Specifically, when calculating statistics, the variance, i.e., the square of the standard deviation, is also included.

[0054] Specifically, when calculating statistics, the coefficient of variation is also included, and the formula is as follows:

[0055] ;

[0056] in, The coefficient of variation is 1. Standard deviation Let be the average velocity value of N models at point 𝑥.

[0057] Specifically, when calculating statistics, the calculated statistics also include the range, which is the difference between the maximum and minimum values ​​among the N velocity values ​​at that point.

[0058] Specifically, in addition to calculating the central tendency statistic and the dispersion statistic, the skewness and kurtosis of the velocity value sample set at that point are also calculated to describe the asymmetry and sharpness of the velocity value distribution.

[0059] Preferably, after generating the uncertain data volume, the following post-processing steps are also included:

[0060] S44: Data volume smoothing, anisotropic smoothing filtering is performed on the generated velocity uncertainty data volume. The smoothing operator has a smoothing intensity along the formation plane and a smoothing intensity perpendicular to the formation plane, which can suppress random noise without blurring the micro-structural boundaries.

[0061] S45: Data volume normalization, normalizes the values ​​in the velocity uncertainty data volume to the [0,1] interval, and generates a dimensionless relative uncertainty data volume.

[0062] Preferably, when obtaining the equivalent velocity model set, the generation process of the equivalent velocity model set is constrained by the geological framework and adopts a multi-scale anisotropic perturbation strategy.

[0063] Preferably, the final product generated is the velocity uncertainty data volume, which is stored in a computer-readable medium in the form of an electronic file and can be used for three-dimensional visualization display, wherein high value areas are marked with a prominent color to indicate the probability zone where microstructures exist.

[0064] Preferably, when performing probabilistic detection of micro-amplitude construction, the specific steps include:

[0065] S51: Acquire input data, acquire velocity uncertainty data volume, and simultaneously acquire seismic migration data volume of the study area;

[0066] S52: Determine the probability threshold based on the numerical distribution of the velocity uncertainty data volume to distinguish between high uncertainty and low uncertainty regions;

[0067] S53: Generate a probability volume and a binary detection volume. Mark the positions in the velocity uncertainty data volume with values ​​higher than the selected threshold as high-probability areas of micro-construction, and generate a three-dimensional micro-construction probability detection data volume.

[0068] S54: Geological interpretation and verification. The micro-tectonic probability detection data volume and the seismic migration data volume are comprehensively compared and analyzed. Based on the termination, discontinuity, bending and amplitude variation characteristics of the seismic reflection phase axis, the high-probability area is geologically interpreted and identified as a specific type of micro-tectonic structure.

[0069] Preferably, determining the probability threshold specifically includes:

[0070] S521: Data extraction and statistical calculation: Read all values ​​of the entire three-dimensional velocity uncertainty data volume, form a sample set, and calculate the global statistical characteristics of the sample set.

[0071] S522: Single-level threshold calculation, calculates a single probability threshold based on a preset confidence level;

[0072] S523: Multi-level threshold calculation, calculates a set of incremental probability thresholds based on multiple preset incremental confidence levels;

[0073] S524: Threshold output, which uses the calculated threshold as the judgment criterion for output.

[0074] Specifically, the formula for calculating a single probability threshold is as follows:

[0075]

[0076] in, It is the arithmetic mean. For threshold coefficient, The standard deviation is denoted as .

[0077] Specifically, when calculating a set of increasing probability thresholds, the formula is as follows:

[0078] ;

[0079] in, For the j-th threshold, , The total number of thresholds, It is a set of increasing threshold coefficients.

[0080] Specifically, the generation of the probability volume and the binary detection volume also includes mathematical morphological processing of the generated binary detection volume, specifically:

[0081] Opening operation: Erosion followed by dilation, used to eliminate isolated noise points and smooth the boundaries of high-probability regions;

[0082] Closing operation: first dilate, then erode, used to fill small holes inside high-probability regions and connect adjacent nearest points.

[0083] Specifically, in geological interpretation and verification, a three-dimensional overlay display technology is used for comprehensive comparative analysis. Specifically, the micro-tectonic probability detection data volume is displayed as a semi-transparent color cloud map, while the layer slices or along-layer slices of the seismic migration data volume are overlaid on it in an opaque manner, so that the high uncertainty area and the seismic reflection characteristics are directly correlated in space.

[0084] Specifically, geological interpretation and verification also include the generation of structural index bodies:

[0085] The system automatically extracts the geometric attributes of each connected high-probability region in the microstructure detection data volume, including but not limited to: the center position of the region, the length along the direction, the width along the dip, the area, the volume, and the main direction. These attributes are then associated with the region labels to generate a list of microstructure targets with an attribute table.

[0086] Specifically, after geological interpretation and verification, it also includes:

[0087] S54: Results Quantification and Risk Assessment. Based on the micro-structure probability detection data, calculate the area percentage and volume percentage of high-probability zones within the target layer as quantitative indicators of the micro-structure development intensity in the zone; and for specific identified micro-structure targets, calculate the minimum distance between their spatial location and the predetermined well trajectory as the basis for geological risk assessment of drilling projects.

[0088] Preferred, specific types of microstructures include: small faults, microblocks, fracture zones, lithological pinch-out lines, channel sand body boundaries, small reef-shoal body boundaries, and heterogeneous boundaries of porous reservoirs.

[0089] Preferably, the generation of the probability volume and the binary detection volume specifically includes:

[0090] S531: Data input, read the three-dimensional velocity uncertainty data volume and probability threshold;

[0091] S532: Binarized detection volume generation, which converts the velocity uncertainty data volume into a binary micro-amplitude constructed detection data volume through a hard threshold function;

[0092] S533: Continuous probability volume generation, which converts velocity uncertainty data volume into continuous micro-amplitude construction probability detection data volume through a continuous probability mapping function;

[0093] S534: Data Volume Output: The generated binarized micro-amplitude construction detection data volume and the continuous micro-amplitude construction probability detection data volume are output as the final detection results.

[0094] Specifically, the mathematical definition of the binarized micro-amplitude construction detection data volume is:

[0095] ;

[0096] in, To construct the detection data volume using a binarized micro-amplitude model, For a three-dimensional velocity uncertainty data volume, The value is the probability threshold. A voxel with a value of 1 represents a region with a high probability of micro-construction, while a voxel with a value of 0 represents a region with a low probability.

[0097] Specifically, the mathematical definition of a continuous, micro-amplitude construct probability detection data volume is:

[0098] ;

[0099] in, To construct a continuous, micro-amplitude probability detection data volume, The slope parameter controls the steepness of the function. This is the probability inflection point threshold.

[0100] In this embodiment, the slope parameter The method for determining it is as follows:

[0101] ;

[0102] in, This represents the uncertainty threshold corresponding to a probability close to 0. This represents the uncertainty threshold corresponding to a probability close to 1. It is a decimal number close to 0.

[0103] Preferably, after generating the binary detection volume, mathematical morphology operations are further applied to optimize the detection results:

[0104] The structuring element S is used to perform closing and opening operations on the binarized micro-amplitude constructed detection data volume, where the closing operation is defined as: Used to fill small holes inside the target and smooth the boundaries; the opening operation is defined as: It is used to eliminate isolated noise points.

[0105] A micro-structure seismic exploration and detection system based on pre-stack depth migration includes:

[0106] The data input module is used to receive pre-stack seismic gather data, initial velocity models, seismic interpretation horizon data, and regional geological framework data.

[0107] The model generation module is used to generate multiple equivalent velocity models based on geological constraints and perturbation strategies.

[0108] The uncertainty calculation module is used to perform statistical analysis on the equivalent velocity model set and generate velocity uncertainty data.

[0109] A construction detection module is used to convert velocity uncertainty data into micro-amplitude construction probability detection data.

[0110] The visualization output module is used to display the test results in three dimensions and quantify the risks.

[0111] The beneficial effects of this invention are:

[0112] 1. Compared with existing technologies that blindly pursue an absolutely correct velocity model but ultimately cannot overcome the impact of model errors on the identification of microstructures, this invention transforms velocity model errors from interference that needs to be minimized into a valuable detection signal. By actively generating multiple globally equivalent velocity models and analyzing their uncertainties, the most unstable regions in the modeling process become direct evidence indicating the existence of microstructures. This realizes a paradigm shift from eliminating errors to utilizing errors, significantly improving the detection capability of microstructures.

[0113] 2. Compared to existing technologies where velocity model perturbations often lack geological constraints and easily generate a large number of geologically unrealistic models, thus introducing irrelevant noise, this scheme introduces strict geological framework constraints and a structurally oriented strategy when generating the equivalent velocity model set. This ensures that the model perturbations strictly follow the stratigraphic distribution, allowing discontinuous changes at boundaries such as faults, ensuring that the explored velocity uncertainty space matches the real geological conditions. The resulting uncertainty data volume has a higher signal-to-noise ratio, and the indicated micro-tectonic locations are more accurate and reliable.

[0114] 3. Compared with existing technologies that typically employ single-scale perturbation strategies, which struggle to effectively separate velocity uncertainty signals generated by geological bodies of different scales, resulting in weak detection specificity, this invention employs a multi-scale, anisotropic intelligent perturbation strategy. This strategy can specifically reveal velocity changes caused by geological bodies of different scales, from large-scale tectonic backgrounds to micro-faults. In particular, small-scale perturbations can effectively "capture" signals from micro-structures, thereby significantly improving the method's detection resolution and specificity for micro-structures.

[0115] 4. Compared with existing technologies that rely heavily on the experience of interpreters to make subjective and qualitative inferences about seismic profiles, this invention provides a complete automated and quantitative process. From uncertainty quantification and automatic calculation of probability thresholds to mathematical morphology optimization of detection results, it ultimately generates a directly quantifiable probability data body and a list of structural targets, which greatly reduces the reliance on human experience, makes the micro-structural detection results more objective and repeatable, and can be directly used for subsequent quantitative evaluation and risk assessment.

[0116] 5. Compared to existing technologies that typically only provide a single interpretation result, making it difficult to meet the needs of different levels in exploration decision-making, this invention simultaneously outputs a binarized detection volume and a continuous probability volume. The binarized detection volume can meet the engineering requirements for quickly locating clear targets, while the continuous probability volume retains complete uncertainty information, supporting refined risk classification assessment. This dual data volume output strategy greatly enhances the practicality and flexibility of the method, enabling it to effectively support the entire process from preliminary screening to detailed pre-drilling evaluation.

[0117] 6. Compared to existing technologies that may be limited to specific algorithms or process segments and lack systematic integration, this invention constructs a full-chain technical solution from core methods to a complete system. It organically integrates geological modeling, velocity analysis, uncertainty quantification, visualization interpretation, and risk assessment, and encapsulates them into a set of collaborative system modules. This not only ensures the efficient and stable operation of the technical process, but also promotes the transformation of this innovative method from theoretical research results to mature tools that can be practically applied to the production line, generating significant technical integration benefits. Attached Figure Description

[0118] Figure 1 The diagram shows a flowchart of the micro-structural seismic exploration and detection method based on pre-stack depth migration of the present invention.

[0119] Figure 2 The diagram shown is a schematic representation of the micro-structural seismic exploration and detection system based on pre-stack depth migration of the present invention. Detailed Implementation

[0120] The present invention will be further described below with reference to the accompanying drawings and embodiments.

[0121] Please see Figure 1 The present invention provides an embodiment of a micro-structural seismic exploration and detection method based on pre-stack depth migration, comprising the following steps:

[0122] S11: Generate a set of equivalent velocity models. Based on seismic data, generate multiple sets of equivalent velocity models that can flatten seismic gathers globally and are within the allowable error range.

[0123] S12: Calculate the velocity uncertainty data volume by performing point-by-point statistics on the equivalent velocity model set, calculating the standard deviation of the velocity value at each spatial location point, and generating a three-dimensional velocity uncertainty data volume.

[0124] S13: Probabilistic detection of micro-structures, interpreting high-value regions in velocity uncertainty data as probability regions where micro-structures exist, where high standard deviations are associated with a high probability of micro-structures.

[0125] S14: Results fusion and display, which overlays the velocity uncertainty data volume with the seismic migration profile and seismic attribute volume to comprehensively verify the micro-structure detection results.

[0126] In this embodiment, the present invention abandons the traditional approach of pursuing a "perfect" velocity model and instead innovatively utilizes the inherent uncertainty of velocity models to reveal micro-structures. This method first generates multiple equivalent velocity models that macroscopically "flatten" the seismic gathers but exhibit subtle local differences, and calculates the standard deviation of their velocity values ​​to form a "velocity uncertainty data volume." Then, regions with high standard deviations in the data volume are interpreted as probability zones for the existence of micro-structures. Finally, comprehensive verification is achieved by overlaying the uncertainty data volume with results such as seismic migration profiles. Ultimately, the analysis of velocity model errors is transformed from a passively accepted "defect" to an actively detected "signal," enabling the high-probability detection of micro-structures such as small faults and lithological boundaries that are difficult to identify with conventional seismic profiles. This significantly improves the resolution and reliability of seismic exploration in identifying complex geological bodies, providing strong technical support for fine-grained description of oil and gas reservoirs and drilling geological risk assessment.

[0127] Preferably, the generation of the equivalent velocity model set specifically includes:

[0128] S21: Obtain initial modeling data, including pre-stack seismic gather data, initial velocity model, and stratigraphic data for seismic interpretation;

[0129] S22: Define the perturbation parameters and range, determine the parameter types, perturbation space, and allowable range of perturbation amount for perturbing the initial velocity model, and ensure that the perturbation amount can still flatten the seismic gathers globally after the perturbation.

[0130] S23: Perform cyclic perturbation and model verification;

[0131] S24: Set generation determination, repeatedly execute loop perturbation and model verification until the number of models in the equivalent speed model set reaches the preset value.

[0132] In this embodiment, the present invention first acquires basic data such as seismic gathers, initial velocity models, and interpretation horizons; then defines perturbation parameters and their reasonable ranges for the velocity field and reflection interfaces; subsequently, it intelligently perturbs and filters the initial model by iteratively executing the core steps of "generating a random perturbation field - constructing candidate models - verifying gather flattening"; finally, after generating a sufficient number of qualified models, an equivalent model set is formed. This process, through an automated iterative filtering mechanism, ensures that each generated velocity model maintains "equivalence" in terms of seismic data fitting, and fully explores the uncertainty space of the velocity field by introducing controlled local variations, laying a solid foundation for subsequent quantitative uncertainty analysis and significantly improving the systematic nature of the detection process and the repeatability of the results.

[0133] Preferably, the cyclic perturbation and model validation process includes the following:

[0134] S231: Within the perturbation allowable range, generate a random perturbation field for the current cycle;

[0135] S232: Apply a random perturbation field to the initial velocity model to obtain a candidate velocity model;

[0136] S233: Seismic migration imaging using candidate velocity models and calculation of gather flattening;

[0137] S234: Determine whether the flatness of the track gather meets the preset standard. If it does, store the candidate velocity model in the equivalent velocity model set; if it does not, discard the model.

[0138] When defining the disturbance parameters and range, the disturbance parameters include the velocity value itself, the lateral gradient of the velocity along the layer, and the depth of the formation reflection interface.

[0139] In this process, when a random perturbation field is applied to the initial velocity model to obtain a candidate velocity model, the velocity field and the reflection interface are coupled and perturbed. That is, when the depth of the local layer interface is perturbed, the corresponding layer velocity is also adjusted accordingly.

[0140] In this embodiment, the scheme details the specific steps of cyclic perturbation and model validation: a random perturbation field is generated within a preset range and applied to the initial model to obtain candidate models. The quality is then verified by calculating the gather flattening using migration imaging; only models that meet the standards are included. Specifically, the perturbation parameters are expanded to include multiple factors such as velocity values, lateral gradients, and interface depths, and a coupled perturbation strategy is adopted—when the interface depth is adjusted, the corresponding layer velocity changes synchronously. This design makes the perturbation process more consistent with the geophysical laws of the subsurface medium, effectively avoiding the generation of a large number of geologically unreasonable models, significantly improving the geological reliability and convergence efficiency of the equivalent model set, and providing geologically significant model samples that meet the requirements of seismic data fitting for subsequent uncertainty analysis.

[0141] Preferably, the generation of the random perturbation field for the current cycle specifically includes:

[0142] S31: Define the basic perturbation field and generate a standard Gaussian random field with a mean of 0 and a variance of 1;

[0143] S32: Apply the spatial correlation function, define the covariance function that characterizes spatial correlation, and use this function to perform convolution filtering on the basic perturbation field to generate a correlated random field with a specific spatial structure;

[0144] S33: Apply perturbation amplitude control, multiply the relevant random field by a perturbation amplitude control function based on spatial variation, and generate the final random perturbation field.

[0145] Specifically, when applying the spatial correlation function, the covariance function adopts an anisotropic exponential function, with the following form:

[0146] ;

[0147] in, and For point and points In the horizontal coordinates, and For point and points In the vertical direction coordinates, For horizontal correlation length, This is the vertical correlation length.

[0148] In this embodiment, the present invention first generates a standard Gaussian random field with zero mean and one variance as a basis; then, it uses an anisotropic exponential covariance function to perform convolution filtering on it. This function, by setting different horizontal and vertical correlation lengths, enables the generated perturbation field to have a spatially consistent correlation structure with the anisotropic characteristics of sedimentary strata; finally, the perturbation intensity is finely adjusted through a spatially varying amplitude control function. This method can generate geologically more reasonable structured perturbations, rather than pure random noise, thereby significantly improving the geological representativeness and diversity of the equivalent velocity model set, and enabling the generated velocity uncertainty data volume to more clearly and reliably reveal sensitive areas related to microstructures.

[0149] Preferably, the horizontal correlation length Directionality is achieved through coordinate rotation, specifically:

[0150] Let u be the direction along the strike of the geological structure and v be the direction along the dip. Then the covariance function in the horizontal direction becomes:

[0151] ;

[0152] in, , and It is a point and points The coordinate difference in the rotated coordinate system The relevant length along the direction, For the relevant length along the dip, The vertical correlation length.

[0153] Preferably, when applying the spatial correlation function, the process of applying the spatial correlation function adopts a multi-scale strategy, specifically as follows:

[0154] A correlated random field is a linear superposition of multiple sub-random fields with different correlation scales:

[0155] ;

[0156] in, For the relevant random field, For the i-th correlated random field generated by a covariance function with a specific correlation length, These are the corresponding weighting coefficients.

[0157] Furthermore, the application of spatial correlation functions is constrained by the geological framework, specifically:

[0158] The shape and parameters of the covariance function are controlled by the geological framework model. Within continuous sedimentary strata, a larger correlation length is used to ensure the smoothness of the disturbance. At preset faults or lithological boundaries, the correlation length of the covariance function is set to a smaller value or directly set to zero to allow the disturbance field to produce discontinuities on both sides of the boundary.

[0159] In this embodiment, the present invention significantly improves the geological rationality of the random perturbation field through three key technologies: First, by rotating the coordinates, the horizontal correlation length is oriented to the strike and dip of the geological structure, achieving anisotropic perturbation consistent with the underground structural framework; second, a multi-scale strategy is adopted to linearly superimpose sub-random fields with different correlation lengths, enabling the perturbation field to simultaneously cover geological features of various scales (large, medium, and small); finally, a geological framework constraint is introduced to maintain perturbation smoothness within continuous strata, while pre-setting discontinuities at faults or lithological boundaries. This effectively ensures that the generated perturbation field is not mathematical random noise, but rather a structured variation conforming to geological laws. Consequently, the resulting velocity uncertainty data volume can more clearly and accurately indicate the location of small faults, lithological boundaries, and other micro-structures, significantly improving the reliability and resolution of the detection results.

[0160] Specifically, the quality framework constraint is achieved by constructing a guiding operator, which is defined by the formation normal field, making the covariance function more correlated along the formation plane direction and less correlated in the direction perpendicular to the formation plane.

[0161] Preferably, when defining the fundamental perturbation field, the generation of the fundamental perturbation field adopts the Fast Fourier Transform method, and the specific process is as follows:

[0162] Generate a complex random field in the frequency domain, with a uniform phase distribution on [0, 2π] and an amplitude of a random variable satisfying a specific energy spectrum distribution;

[0163] By performing an inverse Fourier transform on the complex random field, the fundamental perturbation field in the spatial domain is obtained.

[0164] In this embodiment, the present invention optimizes the generation efficiency and geological consistency of random perturbation fields through two key technologies: First, a guiding operator is constructed using the stratigraphic normal field definition, ensuring that random perturbations are strictly distributed along stratigraphic planes, maintaining high continuity within the stratigraphic plane while rapidly decaying in the direction perpendicular to the stratigraphic plane, thus ensuring a high degree of consistency between the perturbation pattern and the actual geological structure; Second, a complex random field satisfying a specific energy spectrum is generated in the frequency domain using fast Fourier transform technology, and then the fundamental perturbation field in the spatial domain is efficiently obtained through inverse transform, transforming complex spatial convolution into simple frequency domain multiplication operations. This combined strategy, while ensuring that the perturbation field possesses geologically reasonable anisotropic characteristics, significantly improves the computational efficiency of generating large-scale three-dimensional random fields, providing key technical support for the rapid generation of a large number of geologically friendly equivalent velocity models in practical production applications.

[0165] Preferably, when calculating data volumes with uncertain computational speed, the specific methods include:

[0166] S41: Obtain the equivalent velocity model set. Obtain a model set consisting of N sets of equivalent velocity models, where each model has a defined velocity value at a grid point in three-dimensional space.

[0167] S42: Calculate statistics point by point. For each grid point in 3D space, perform the following:

[0168] Data extraction: Extract the M velocity values ​​corresponding to the grid point from the model set to form a velocity value sample set;

[0169] Calculate the statistic: Based on the velocity value sample set, calculate the statistic that characterizes the degree of dispersion of velocity values ​​at that point;

[0170] S43: Generate an uncertain data volume by assigning the calculated statistics at each grid point to that point, thus generating a three-dimensional data volume with the same dimensions as the original velocity model, which is the velocity uncertainty data volume.

[0171] Specifically, when calculating statistics, the standard deviation is included, and the formula is as follows:

[0172] ;

[0173] in, Standard deviation For the velocity value samples of the i-th model, Let be the average velocity value of N models at point 𝑥.

[0174] Specifically, when calculating statistics, the variance, i.e., the square of the standard deviation, is also included.

[0175] Specifically, when calculating statistics, the coefficient of variation is also included, and the formula is as follows:

[0176] ;

[0177] in, The coefficient of variation is 1. Standard deviation Let be the average velocity value of N models at point 𝑥.

[0178] Specifically, when calculating statistics, the calculated statistics also include the range, which is the difference between the maximum and minimum values ​​among the N velocity values ​​at that point.

[0179] Specifically, in addition to calculating the central tendency statistic and the dispersion statistic, the skewness and kurtosis of the velocity value sample set at that point are also calculated to describe the asymmetry and sharpness of the velocity value distribution.

[0180] In this embodiment, the present invention first obtains an equivalent velocity model set. Then, for each grid point in three-dimensional space, the velocity values ​​of all models at that point are extracted from the set to form a sample set, and their statistics are calculated. Finally, the statistics of each point are summarized into a three-dimensional data volume. This scheme adopts a multi-index collaborative evaluation strategy. In addition to the core standard deviation, it can also select statistics such as variance, coefficient of variation, range, and even skewness and kurtosis. This not only quantitatively characterizes the dispersion of velocity values, but also reveals more complex statistical characteristics such as the asymmetry and sharpness of their error distribution. This multi-dimensional and in-depth uncertainty quantification method can more comprehensively and sensitively capture the stability differences of velocity models at different spatial locations, thereby providing richer and more reliable probabilistic basis for the identification of micro-structures, significantly enhancing the information content and interpretive depth of the detection results.

[0181] Preferably, after generating the uncertain data volume, the following post-processing steps are also included:

[0182] S44: Data volume smoothing, anisotropic smoothing filtering is performed on the generated velocity uncertainty data volume. The smoothing operator has a smoothing intensity along the formation plane and a smoothing intensity perpendicular to the formation plane, which can suppress random noise without blurring the micro-structural boundaries.

[0183] S45: Data volume normalization, normalizes the values ​​in the velocity uncertainty data volume to the [0,1] interval, and generates a dimensionless relative uncertainty data volume.

[0184] Preferably, when obtaining the equivalent velocity model set, the generation process of the equivalent velocity model set is constrained by the geological framework and adopts a multi-scale anisotropic perturbation strategy.

[0185] Preferably, the final product generated is the velocity uncertainty data volume, which is stored in a computer-readable medium in the form of an electronic file and can be used for three-dimensional visualization display, wherein high value areas are marked with a prominent color to indicate the probability zone where microstructures exist.

[0186] In this embodiment, after generating the velocity uncertainty data volume, the present invention innovatively introduces two key post-processing steps: First, anisotropic smoothing filtering is employed to enhance the smoothing along the stratigraphic planes and suppress the smoothing vertically, effectively suppressing random noise while precisely protecting the boundary morphology of micro-structures; subsequently, data normalization is performed to uniformly transform the uncertainty values ​​into the dimensionless interval [0,1], forming a relative uncertainty data volume. Combined with the model set generated by the previously established geological framework constraints and multi-scale anisotropic perturbation strategy, the final velocity uncertainty data volume can be stored and visualized as an independent product, with high-value areas highlighted in a striking color. This series of processing steps significantly improves the signal-to-noise ratio, interpretability, and comparability of the data volume, making the identification of micro-structure probability zones more intuitive and reliable, and providing a strong decision-making basis for the fine description of oil and gas reservoirs.

[0187] Preferably, when performing probabilistic detection of micro-amplitude construction, the specific steps include:

[0188] S51: Acquire input data, acquire velocity uncertainty data volume, and simultaneously acquire seismic migration data volume of the study area;

[0189] S52: Determine the probability threshold based on the numerical distribution of the velocity uncertainty data volume to distinguish between high uncertainty and low uncertainty regions;

[0190] S53: Generate a probability volume and a binary detection volume. Mark the positions in the velocity uncertainty data volume with values ​​higher than the selected threshold as high-probability areas of micro-construction, and generate a three-dimensional micro-construction probability detection data volume.

[0191] S54: Geological interpretation and verification. The micro-tectonic probability detection data volume and the seismic migration data volume are comprehensively compared and analyzed. Based on the termination, discontinuity, bending and amplitude variation characteristics of the seismic reflection phase axis, the high-probability area is geologically interpreted and identified as a specific type of micro-tectonic structure.

[0192] In this embodiment, the present invention clarifies the transformation process from velocity uncertainty to geological conclusions: First, the velocity uncertainty data volume and the seismic migration data volume are loaded together; then, an objective probability threshold is determined based on the statistical distribution characteristics of the uncertainty values; subsequently, the uncertainty data volume is transformed into a three-dimensional data volume (continuous probability volume or binary detection volume) that quantitatively represents the probability of the existence of microstructures through threshold segmentation; finally, the probability detection results are comprehensively compared and analyzed with the seismic migration profile, and the high-probability areas are geologically interpreted based on typical seismic response characteristics such as the faulting and termination of the phase axis, identifying specific structures such as small faults. This process successfully transforms the abstract concept of velocity uncertainty into a concrete and visualized probability distribution map of geological targets, and through mutual verification with direct seismic evidence, significantly improves the objectivity, intuitiveness, and geological reliability of the microstructure detection results, achieving a reliable leap from geophysical inversion to geological understanding.

[0193] Preferably, determining the probability threshold specifically includes:

[0194] S521: Data extraction and statistical calculation: Read all values ​​of the entire three-dimensional velocity uncertainty data volume, form a sample set, and calculate the global statistical characteristics of the sample set.

[0195] S522: Single-level threshold calculation, calculates a single probability threshold based on a preset confidence level;

[0196] S523: Multi-level threshold calculation, calculates a set of incremental probability thresholds based on multiple preset incremental confidence levels;

[0197] S524: Threshold output, which uses the calculated threshold as the judgment criterion for output.

[0198] Specifically, the formula for calculating a single probability threshold is as follows:

[0199]

[0200] in, It is the arithmetic mean. For threshold coefficient, The standard deviation is denoted as .

[0201] Specifically, when calculating a set of increasing probability thresholds, the formula is as follows:

[0202] ;

[0203] in, For the j-th threshold, , The total number of thresholds, It is a set of increasing threshold coefficients.

[0204] Specifically, the generation of the probability volume and the binary detection volume also includes mathematical morphological processing of the generated binary detection volume, specifically:

[0205] Opening operation: Erosion followed by dilation, used to eliminate isolated noise points and smooth the boundaries of high-probability regions;

[0206] Closing operation: first dilate, then erode, used to fill small holes inside high-probability regions and connect adjacent nearest points.

[0207] In this embodiment, the present invention significantly improves the objectivity and accuracy of micro-structure detection by establishing a systematic probability threshold determination and optimization process. It automatically calculates single-level or multi-level probability thresholds using quantitative formulas based on global statistical characteristics, effectively avoiding the subjective arbitrariness of manually setting thresholds. In particular, the introduction of multi-level thresholds enables risk grading assessment, providing a gradient basis for engineering decisions. After generating the detection results, innovative mathematical morphology processing is applied, effectively eliminating isolated anomalies caused by noise, filling internal voids in valid signals, and smoothing structural boundaries. These measures ensure that the final micro-structure detection results not only have a solid statistical basis but also are more complete, continuous, and geologically sound in spatial morphology, greatly enhancing the interpretability and practical value of the results.

[0208] Specifically, in geological interpretation and verification, a three-dimensional overlay display technology is used for comprehensive comparative analysis. Specifically, the micro-tectonic probability detection data volume is displayed as a semi-transparent color cloud map, while the layer slices or along-layer slices of the seismic migration data volume are overlaid on it in an opaque manner, so that the high uncertainty area and the seismic reflection characteristics are directly correlated in space.

[0209] Specifically, geological interpretation and verification also include the generation of structural index bodies:

[0210] The system automatically extracts the geometric attributes of each connected high-probability region in the microstructure detection data volume, including but not limited to: the center position of the region, the length along the direction, the width along the dip, the area, the volume, and the main direction. These attributes are then associated with the region labels to generate a list of microstructure targets with an attribute table.

[0211] In this embodiment, the present invention innovatively employs three-dimensional overlay display technology in the geological interpretation stage. This technology precisely overlays semi-transparent micro-structural probability data volumes with stratigraphic slices of seismic migration data volumes, directly coupling high-uncertainty areas with seismic reflection characteristics (such as faults and lithological boundaries) in space. This significantly improves the intuitiveness of the interpretation and the accuracy of location. Furthermore, by automatically extracting the geometric attributes (such as location, scale, and orientation) of each high-probability connected region and generating a structured target list, the invention achieves a leap from qualitative interpretation to quantitative assessment. This significantly improves the automation level, interpretation efficiency, and engineering application value of micro-structural identification, providing precise data support for well location deployment and geological modeling.

[0212] Specifically, after geological interpretation and verification, it also includes:

[0213] S54: Results Quantification and Risk Assessment. Based on the micro-structure probability detection data, calculate the area percentage and volume percentage of high-probability zones within the target layer as quantitative indicators of the micro-structure development intensity in the zone; and for specific identified micro-structure targets, calculate the minimum distance between their spatial location and the predetermined well trajectory as the basis for geological risk assessment of drilling projects.

[0214] Preferred, specific types of microstructures include: small faults, microblocks, fracture zones, lithological pinch-out lines, channel sand body boundaries, small reef-shoal body boundaries, and heterogeneous boundaries of porous reservoirs.

[0215] In this embodiment, after completing geological interpretation, the present invention adds a result quantification and risk assessment step. By calculating the area or volume percentage of high-probability micro-structure zones within the target stratigraphic interval, a quantitative evaluation of the intensity of zonal structural development is achieved. Simultaneously, by calculating the minimum distance between a specific micro-structure target and the designed well trajectory, direct geological risk information is provided for drilling engineering. This scheme detects a wide range of target types, covering key geological bodies such as small faults, fracture zones, and lithological boundaries. This design elevates the detection method from qualitative identification to quantitative evaluation and risk warning, enabling the results not only to guide the deepening of geological understanding but also to directly serve engineering decisions such as well location optimization and drilling safety, significantly enhancing the practical value and economic benefits of the technology.

[0216] Preferably, the generation of the probability volume and the binary detection volume specifically includes:

[0217] S531: Data input, read the three-dimensional velocity uncertainty data volume and probability threshold;

[0218] S532: Binarized detection volume generation, which converts the velocity uncertainty data volume into a binary micro-amplitude constructed detection data volume through a hard threshold function;

[0219] S533: Continuous probability volume generation, which converts velocity uncertainty data volume into continuous micro-amplitude construction probability detection data volume through a continuous probability mapping function;

[0220] S534: Data Volume Output: The generated binarized micro-amplitude construction detection data volume and the continuous micro-amplitude construction probability detection data volume are output as the final detection results.

[0221] In this embodiment, the present invention directly converts the uncertain data volume into a binary detection volume with a clear black-and-white distinction through a hard threshold function, clearly delineating the structure boundary. Simultaneously, it utilizes a continuous probability mapping function to generate a continuous probability volume, using gradient colors to reflect the likelihood of the structure's existence. This strategy of simultaneously outputting both "deterministic" and "probabilistic" dual data volumes satisfies the need for rapid location of clear targets in exploration decision-making while retaining the uncertainty information necessary for risk assessment in a continuous probability form. This significantly improves the practicality and flexibility of the detection results, enabling it to support different application scenarios from rapid screening to refined risk assessment, greatly enhancing the method's applicability in actual production.

[0222] Specifically, the mathematical definition of the binarized micro-amplitude construction detection data volume is:

[0223] ;

[0224] in, To construct the detection data volume using a binarized micro-amplitude model, For a three-dimensional velocity uncertainty data volume, The value is the probability threshold. A voxel with a value of 1 represents a region with a high probability of micro-construction, while a voxel with a value of 0 represents a region with a low probability.

[0225] Specifically, the mathematical definition of a continuous, micro-amplitude construct probability detection data volume is:

[0226] ;

[0227] in, To construct a continuous, micro-amplitude probability detection data volume, The slope parameter controls the steepness of the function. This is the probability inflection point threshold.

[0228] In this embodiment, the slope parameter The method for determining it is as follows:

[0229] ;

[0230] in, This represents the uncertainty threshold corresponding to a probability close to 0. This represents the uncertainty threshold corresponding to a probability close to 1. It is a decimal number close to 0.

[0231] Preferably, after generating the binary detection volume, mathematical morphology operations are further applied to optimize the detection results:

[0232] The structuring element S is used to perform closing and opening operations on the binarized micro-amplitude constructed detection data volume, where the closing operation is defined as: Used to fill small holes inside the target and smooth the boundaries; the opening operation is defined as: It is used to eliminate isolated noise points.

[0233] In this embodiment, the present invention utilizes a hard thresholding function to generate a clearly defined binary detection volume, and constructs a continuous probability volume reflecting gradual probability changes using a sigmoid function. The slope parameter q of the sigmoid function is precisely controlled by preset high and low thresholds to ensure the physical meaning of the probability values ​​is clear. In particular, the scheme innovatively introduces mathematical morphological operations (opening and closing operations) to optimize the binary results, effectively eliminating noise interference and improving the structural morphology. This series of mathematical processing significantly improves the signal-to-noise ratio and geological rationality of the detection results, enabling the identification of micro-structures to meet both the engineering requirements of rapid localization and support precise risk quantification assessment.

[0234] like Figure 2 As shown, the micro-structural seismic exploration and detection system based on pre-stack depth migration includes:

[0235] The data input module is used to receive pre-stack seismic gather data, initial velocity models, seismic interpretation horizon data, and regional geological framework data.

[0236] The model generation module is used to generate multiple equivalent velocity models based on geological constraints and perturbation strategies.

[0237] The uncertainty calculation module is used to perform statistical analysis on the equivalent velocity model set and generate velocity uncertainty data.

[0238] A construction detection module is used to convert velocity uncertainty data into micro-amplitude construction probability detection data.

[0239] The visualization output module is used to display the test results in three dimensions and quantify the risks.

[0240] In this embodiment, the system achieves accurate identification of micro-structures through the collaborative operation of five core modules: data input, model generation, uncertainty calculation, structural detection, and visualization output. The system innovatively employs multiple equivalent velocity model generation strategies, generating velocity uncertainty data volumes through statistical analysis, further converting them into probabilistic detection data volumes, and finally quantifying risk through 3D visualization. This technical solution transforms velocity model errors into effective detection signals, significantly improving the detection accuracy and reliability of micro-structures such as small faults and lithological boundaries, providing an effective technical means for fine-grained description of oil and gas reservoirs and drilling risk assessment.

[0241] Example 1: Detection of small faults within a complex fault zone in an onshore basin

[0242] Application Background: The target strata in this basin are deeply buried, with a complex tectonic setting and numerous low-amplitude structures and small faults. These micro-faults play a crucial role in controlling the migration and accumulation of oil and gas, but they are difficult to identify on conventional seismic profiles, resulting in many design wells encountering unpredictable small faults and affecting development effectiveness.

[0243] Specific implementation:

[0244] Data preparation and geological modeling: 3D pre-stack seismic gathers, existing well-seismic calibration velocity models, and interpretations of major stratigraphic levels and large fault frameworks were collected for the work area. Based on regional geological understanding, a geological framework model with a clear NE-E sedimentary trend and major faults trending NNE was established.

[0245] Generation of equivalent velocity model set under geological framework constraints:

[0246] Starting with the initial velocity model, multiple equivalent models are generated using the Monte Carlo simulation method.

[0247] Intelligent perturbation: Key technologies were implemented when generating the random perturbation field for each model: First, the initial random field was efficiently generated using Fast Fourier Transform (FFT). Next, a multi-scale, anisotropic covariance function was applied for filtering—that is, perturbations of three scales (large, medium, and small) were designed, with the perturbations extending further along the stratigraphic strike (NE-NE) and smaller in the vertical direction, making the perturbation shape more consistent with geological body characteristics. Most importantly, a geological framework constraint was introduced. Within known large continuous sedimentary areas, the perturbation was forced to be smooth and continuous, while near pre-set large fault lines, discontinuous jumps in velocity and stratigraphy were allowed. Simultaneously, stratigraphic depth and stratigraphic velocity were coupled with perturbation; that is, when a stratigraphic level is deepened by perturbation, its corresponding stratigraphic velocity also increases accordingly, to conform to the compaction trend.

[0248] Model validation: Each candidate model generated is used for pre-stack depth migration and its gather flattening is calculated. Only those models that can sufficiently flatten the entire work area gathers macroscopically are accepted and stored in the equivalent model set. This process is repeated hundreds of times, ultimately resulting in a set containing hundreds of equivalent velocity models.

[0249] Calculation and post-processing of velocity uncertainty data:

[0250] For each 3D grid point in the model set, the standard deviation of its hundreds of velocity values ​​was calculated, generating an initial velocity uncertainty data volume.

[0251] Anisotropic smoothing was performed on the data volume, primarily along stratigraphic planes, to suppress random noise while protecting structural boundaries. Finally, the data volume values ​​were normalized to the 0-1 range for easier interpretation.

[0252] Micro-construction probability detection and verification:

[0253] Determine the threshold: Analyze the normalized uncertain data volume, define the region with values ​​greater than 0.7 as the micro-construction high probability region, and define the value between 0.5 and 0.7 as the medium probability region.

[0254] Generate detection objects: Based on the threshold, generate binary detection objects (1 for high probability regions and 0 for others) and continuous probability objects.

[0255] Mathematical morphology optimization: Perform opening and closing operations on the binary detection volume to remove sporadic noise points and fill tiny holes, making the fault traces more continuous and complete.

[0256] Comprehensive Interpretation and Verification: High-probability areas were overlaid on a conventional seismic migration profile as semi-transparent red cloud images. Interpreters discovered that these high-standard-deviation areas clearly delineated several previously unidentified small fault trajectories, which highly coincided with the locations of minor fault breaks and amplitude variations in the seismic phase axis. The system automatically extracted the geometric properties of these high-probability areas, generating a target list containing 24 small faults, including their length, strike, and location.

[0257] Risk quantification: The proportion of high-probability zones within the target stratigraphic segment was calculated as an indicator of the intensity of fracture development in that area. Furthermore, the distance between the newly deployed well trajectory and the nearest minor fault was calculated, providing pre-drilling risk warnings.

[0258] Implementation Results: After applying this method, several small faults with vertical displacement of less than 10 meters were successfully predicted. Subsequent drilling confirmed that the accuracy rate was significantly higher than that of traditional methods, providing key basis for well location optimization and geological model correction.

[0259] Example 2: Identification of Lithological Boundaries within Deepwater Fan Reservoirs at Sea

[0260] Application Background: The reservoir in this deep-water area is a large turbidite fan with strong internal lithological heterogeneity. The boundaries of high-quality sand bodies (lithological pinch-out lines) control hydrocarbon accumulation. However, conventional impedance spectroscopy has limited ability to distinguish between sandstone and mudstone, making it difficult to accurately characterize sand body boundaries.

[0261] Specific implementation:

[0262] Data preparation and geological modeling: Using high-quality broadband 3D seismic data and combined with data from multiple wells, a geological model including the top and bottom interfaces of the main fan bodies was established. The geological framework emphasizes the orientation of sedimentary units within the reservoir.

[0263] Generate a set of targeted equivalent velocity models:

[0264] This implementation focuses more on small- and medium-scale disturbances, aiming to capture subtle velocity changes within the reservoir.

[0265] When generating the perturbation field, the anisotropic direction is set according to the paleocurrent direction of the sedimentary unit. The geological constraints are reflected in the fact that the perturbation is continuous and smooth within the interpreted single lobes; while greater velocity variations are allowed at the edge zones where different lobes may come into contact.

[0266] Uncertainty Quantification and Lithological Boundary Detection:

[0267] The standard deviations of hundreds of equivalent models were calculated to generate an uncertain data volume. It was found that regions with high standard deviations were not linear (such as faults), but rather exhibited discontinuous banded or curved linear features consistent with sedimentary patterns.

[0268] By comparing these high-uncertainty stripes with seismic attribute volumes (such as root mean square amplitude) through three-dimensional overlay display, it was found that they are located precisely at the edge of the transition from strong amplitude (interpreted as high-quality sandstone) to weak amplitude (interpreted as mudstone or low-quality sandstone).

[0269] Geological interpretation and risk assessment:

[0270] Interpreters probabilistically interpret these high-uncertainty bands as lithological pinch-out lines or effective reservoir boundaries. This is because velocity modeling is most sensitive to whether sandstone or mudstone velocities are used at these boundary locations, resulting in the greatest variability (high standard deviation) in the model ensemble at these points.

[0271] The system-generated list of micro-structural targets now includes information on the spatial distribution of these lithological boundaries. Using this information, developers optimized horizontal well trajectory design to maximize travel distances within high-quality reservoirs and successfully mitigate potential lithological pinch-out risks.

[0272] Implementation Results: This method successfully revealed the heterogeneous boundaries within the reservoir that are difficult to characterize using traditional methods. The predicted sand body boundaries were verified by subsequent horizontal well drilling, showing a high degree of agreement. This effectively guided the well location deployment for the development of this deepwater oil and gas field and reduced geological risks.

[0273] The embodiments of the present invention have been described in detail above with reference to the accompanying drawings. However, the present invention is not limited to the above embodiments. Within the scope of knowledge possessed by those skilled in the art, various changes can be made without departing from the spirit of the present invention.

Claims

1. A micro-structural seismic exploration and detection method based on pre-stack depth migration, characterized in that: Includes the following steps: S11: Generate a set of equivalent velocity models. Based on seismic data, generate multiple sets of equivalent velocity models that can flatten seismic gathers globally and are within the allowable error range. S12: Calculate the velocity uncertainty data volume by performing point-by-point statistics on the equivalent velocity model set, calculating the standard deviation of the velocity value at each spatial location point, and generating a three-dimensional velocity uncertainty data volume. S13: Probabilistic detection of micro-structures, interpreting high-value regions in velocity uncertainty data as probability regions where micro-structures exist, where high standard deviations are associated with a high probability of micro-structures. S14: Results fusion and display, which overlays the velocity uncertainty data volume with the seismic migration profile and seismic attribute volume to comprehensively verify the micro-structure detection results; The generation of the equivalent velocity model set specifically includes: S21: Obtain initial modeling data, including pre-stack seismic gather data, initial velocity model, and stratigraphic data for seismic interpretation; S22: Define the perturbation parameters and range, determine the parameter types, perturbation space, and allowable range of perturbation amount for perturbing the initial velocity model, and ensure that the perturbation amount can still flatten the seismic gathers globally after the perturbation. S23: Perform cyclic perturbation and model verification; S24: Set generation determination, repeatedly execute loop perturbation and model verification until the number of models in the equivalent speed model set reaches the preset value.

2. The micro-scale structural seismic exploration and detection method based on pre-stack depth migration according to claim 1, characterized in that: The specific steps involved in performing cyclic perturbations and model validation include: S231: Within the perturbation allowable range, generate a random perturbation field for the current cycle; S232: Apply a random perturbation field to the initial velocity model to obtain a candidate velocity model; S233: Seismic migration imaging using candidate velocity models and calculation of gather flattening; S234: Determine whether the flatness of the track gather meets the preset standard. If it does, store the candidate velocity model in the equivalent velocity model set; if it does not, discard the candidate velocity model.

3. The micro-structural seismic exploration and detection method based on pre-stack depth migration according to claim 2, characterized in that: When generating a random perturbation field for the current cycle, the specific steps include: S31: Define the basic perturbation field and generate a standard Gaussian random field with a mean of 0 and a variance of 1; S32: Apply the spatial correlation function, define the covariance function that characterizes spatial correlation, and use the covariance function to perform convolution filtering on the basic perturbation field to generate a correlated random field with a specific spatial structure. S33: Apply perturbation amplitude control, multiply the relevant random field by a perturbation amplitude control function based on spatial variation, and generate the final random perturbation field.

4. The micro-scale structural seismic exploration and detection method based on pre-stack depth migration according to claim 3, characterized in that: When calculating the volume of data with speed uncertainty, the specific components include: S41: Obtain the equivalent velocity model set. Obtain a model set consisting of N sets of equivalent velocity models, where each model has a defined velocity value at a grid point in three-dimensional space. S42: Calculate statistics point by point. For each grid point in 3D space, perform the following: Data extraction: Extract the M velocity values ​​corresponding to the grid point from the model set to form a velocity value sample set; Calculate the statistics: Based on the velocity value sample set, calculate the statistics that characterize the dispersion of velocity values ​​at the grid point; S43: Generate an uncertain data volume by assigning the calculated statistics at each grid point to that grid point, generating a three-dimensional data volume with the same dimensions as the original velocity model, which is the velocity uncertainty data volume.

5. The micro-scale structural seismic exploration and detection method based on pre-stack depth migration according to claim 4, characterized in that: After generating the uncertain data volume, the following post-processing steps are also included: S44: Data volume smoothing, anisotropic smoothing filtering is performed on the generated velocity uncertainty data volume. The smoothing operator has a smoothing intensity along the formation plane and a smoothing intensity perpendicular to the formation plane, which can suppress random noise without blurring the micro-structural boundaries. S45: Data volume normalization, normalizes the values ​​in the velocity uncertainty data volume to the [0,1] interval, and generates a dimensionless relative uncertainty data volume.

6. The micro-scale structural seismic exploration and detection method based on pre-stack depth migration according to claim 5, characterized in that: The probabilistic detection micro-amplitude construction specifically includes: S51: Acquire input data, acquire velocity uncertainty data volume, and simultaneously acquire seismic migration data volume of the study area; S52: Determine the probability threshold based on the numerical distribution of the velocity uncertainty data volume to distinguish between high uncertainty and low uncertainty regions; S53: Generate a probability volume and a binary detection volume. Mark the positions in the velocity uncertainty data volume with values ​​higher than the selected threshold as high-probability areas of micro-construction, and generate a three-dimensional micro-construction probability detection data volume. S54: Geological interpretation and verification. The micro-tectonic probability detection data volume and the seismic migration data volume are comprehensively compared and analyzed. Based on the termination, discontinuity, bending and amplitude variation characteristics of the seismic reflection phase axis, the high-probability area is geologically interpreted and identified as a specific type of micro-tectonic structure.

7. The micro-scale structural seismic exploration and detection method based on pre-stack depth migration according to claim 6, characterized in that: Determining the probability threshold specifically includes: S521: Data extraction and statistical calculation: Read all values ​​of the entire three-dimensional velocity uncertainty data volume, form a sample set, and calculate the global statistical characteristics of the sample set. S522: Single-level threshold calculation, calculates a single probability threshold based on a preset confidence level; S523: Multi-level threshold calculation, calculates a set of incremental probability thresholds based on multiple preset incremental confidence levels; S524: Threshold output, which uses the calculated threshold as the judgment criterion for output.

8. The micro-scale structural seismic exploration and detection method based on pre-stack depth migration according to claim 7, characterized in that: The generation of probability volumes and binary detection volumes specifically includes: S531: Data input, read the three-dimensional velocity uncertainty data volume and probability threshold; S532: Binarized detection volume generation, which converts the velocity uncertainty data volume into a binary micro-amplitude constructed detection data volume through a hard threshold function; S533: Continuous probability volume generation, which converts velocity uncertainty data volume into continuous micro-amplitude construction probability detection data volume through a continuous probability mapping function; S534: Data Volume Output: The generated binarized micro-amplitude construction detection data volume and the continuous micro-amplitude construction probability detection data volume are output as the final detection results.

9. A micro-structural seismic exploration and detection system based on pre-stack depth migration, applied to the micro-structural seismic exploration and detection method based on pre-stack depth migration as described in any one of claims 1-8, characterized in that: include: The data input module is used to receive pre-stack seismic gather data, initial velocity models, seismic interpretation horizon data, and regional geological framework data. The model generation module is used to generate multiple equivalent velocity models based on geological constraints and perturbation strategies. The uncertainty calculation module is used to perform statistical analysis on the equivalent velocity model set and generate velocity uncertainty data. A construction detection module is used to convert velocity uncertainty data into micro-amplitude construction probability detection data. The visualization output module is used to display the test results in three dimensions and quantify the risks.

Images