True surface seismic geology integrated speed modeling and pre-stack depth imaging method

Abstract

The invention provides a true surface seismic geology integrated velocity modeling and pre-stack depth imaging method, and belongs to the technical field of seismic imaging, and the method comprises the steps: solving a near-surface velocity model through the tomography inversion of a first-motion wave, and fusing the near-surface velocity model with a mid-deep layer velocity model obtained by a large floating surface of a reflected wave, so as to form an initial depth migration velocity field; the velocity model is optimized through reflection wave tomography velocity inversion, and a true earth surface isotropic velocity model is built; the method comprises the following steps of: extracting azimuth angle and dip angle data of an isotropic migration seismic database on the basis of the isotropic migration seismic database, analyzing through an anisotropic model, calculating various parameters of the anisotropic model, solving an anisotropic speed, and repeatedly iterating and updating for multiple times through a network tomography inversion method to obtain a multi-information constraint anisotropic speed model; according to the method, the problems of low modeling speed, poor accuracy, poor imaging quality and low precision of the existing true surface seismic geological velocity modeling and pre-stack depth imaging are solved.

Description

Technical Field

[0001] The present invention belongs to the technical field of seismic imaging, and in particular relates to a true surface seismic-geological integrated velocity modeling and pre-stack depth imaging method. Background Art

[0002] Multiple phases of tectonic movements can lead to "double complexity" on the surface and underground: the surface fluctuates violently, with large relative height differences, and the speed and thickness of the low-velocity zone change rapidly; the underground structure is complex, resulting in a complex wave field, making it difficult to obtain high-quality seismic data with accurate and reliable imaging.

[0003] For seismic imaging studies in areas with complex geological structures and complex surface structures, the most effective method currently is to approximate the "true" surface pre-stack depth migration. The final imaging effect depends on two factors: the advancement and applicability of the migration algorithm; and the reasonable and accurate pre-stack depth migration velocity modeling that matches it.

[0004] Traditional migration velocity iteration processes all aim to flatten reflection events, using the residual delay (residual RMS velocity) of reflections to invert accurate migration velocities. However, this method is significantly limited in shallow layers due to low reflection coverage and high noise levels. Furthermore, the accuracy of shallow velocity in prestack depth migration directly impacts the imaging quality of the underlying strata. Consequently, traditional migration velocity iterations result in poor imaging quality and low precision for the underlying strata.

[0005] The prior art discloses a method for improving the accuracy of pre-stack depth migration imaging on an undulating surface. The method includes performing a small smoothing on the surface elevation surface and using the smoothed small smoothed surface elevation surface as the pre-stack depth migration surface. The method uses the first arrival time of the cannon to perform tomographic inversion iteration to obtain a shallow velocity model, where the starting position of the shallow velocity model during travel is consistent with the pre-stack depth migration surface. The method obtains a low-frequency static correction corresponding to the surface low-velocity zone in the shallow velocity model. The low-frequency static correction is removed from the migration input data. The migration input data is shifted in the time direction onto the pre-stack depth migration surface. The method obtains a mid-deep velocity model and fuses the shallow velocity model with the mid-deep velocity model to obtain an initial pre-stack depth migration velocity model. However, the invention patent uses a weighted fusion method for fusion and splicing, which greatly reduces the accuracy of the shallow velocity of pre-stack depth migration, resulting in poor imaging quality and very low accuracy. Summary of the Invention

[0006] In response to the above-mentioned deficiencies in the prior art, the present invention provides a true surface seismic geological integrated velocity modeling and prestack depth imaging method, which solves the problems of existing true surface seismic geological velocity modeling and prestack depth imaging, such as slow modeling speed and poor accuracy, as well as poor imaging quality and low precision.

[0007] In order to achieve the above objectives, the present invention adopts a technical solution: a real surface seismic geological integrated velocity modeling and prestack depth imaging method, comprising the following steps: S1. Collect seismic exploration observation data and isotropic migration seismic database in the study area to obtain the mid-deep velocity model. Use first-arrival wave tomographic inversion to obtain the near-surface velocity model. Select the splicing surface and fusion surface. Based on the fusion surface, fuse the splicing surface to the near-surface velocity model and the mid-deep velocity model to form the initial depth migration velocity field. S2. The near-surface velocity model and the mid-deep velocity model fused by the splicing surface are fused to obtain a fused velocity model. The common center point gathers are corrected. Based on the geological facies, structural style, and well information, a constraint framework and correction criteria are selected from multiple angles. The fused velocity model is corrected based on seismic data drive, seismic-geological integration, and multiple rounds of deep fusion to obtain a large-scale floating surface velocity model of the common center point gathers. S3. Based on the isotropic migration seismic database, multi-scale grid tomography is performed. The true surface isotropic velocity model is modeled using the large-scale floating surface velocity model of the common center point gathers, and the anisotropic parameter field is obtained. S4. Based on the seismic exploration observation data and the isotropic migration seismic database, anisotropic parameter fields are used to perform modeling to obtain a depth velocity model. The depth velocity model is constrained and errors are eliminated using well location stratification to model a multi-information constrained anisotropic velocity model. S5. Use the initial depth migration velocity field as the input of prestack depth migration, and use the multi-information constrained anisotropic velocity model to perform the preset prestack depth migration to generate post-migration seismic imaging results.

[0008] The beneficial effects of the present invention are as follows: in the process of true surface seismic geological integrated velocity modeling, the present invention continuously improves the accuracy of the depth domain velocity model through the deep integration and processing interpretation of seismic data drive and geological multi-information guidance, and multiple rounds of seismic geological feedback, thereby improving the depth and occurrence accuracy of geological targets, meeting the requirements of field tracking for imaging accuracy of the target layer, and achieving high accuracy and high efficiency of true surface seismic geological velocity modeling, as well as high-quality and high-precision pre-stack depth imaging.

[0009] Furthermore, the S1 includes the following steps: S101. Collect seismic exploration observation data and isotropic migration seismic database in the study area, perform small-scale smoothing on the surface elevation data to obtain the migration starting surface, build a mid-deep velocity model based on the large floating surface of the reflected wave, and perform first-arrival tomographic inversion to build an initial near-surface velocity model. S102, using a fusion technique of tomographic statics and residual statics, combined with the migration starting plane, to fine-tune the initial near-surface velocity model; in response to the preliminary statics, constraining the initial near-surface velocity model through micro-logging, and performing densely spaced sampling to obtain an optimized near-surface velocity model; S103. Select a splicing surface and a fusion surface based on the seismic exploration observation data, and fuse the splicing surface into the optimized near-surface velocity model and the mid-deep velocity model based on the fusion surface to establish an initial depth migration velocity field.

[0010] Furthermore, the S101 includes the following steps: S1011. Collect seismic exploration observation data and an isotropic migration seismic database in the study area, perform small-scale smoothing on the real surface elevation data in the seismic exploration observation data, and obtain a migration starting surface; S1012. Based on the large floating surface of the reflected wave, a velocity model of the mid-deep layer is built. Using the first arrival of the cannon with the preset offset distance, first arrival tomographic inversion is performed to build an initial near-surface velocity model.

[0011] The beneficial effects of the above further scheme are as follows: the present invention obtains the migration starting surface through small-scale smoothing processing, thereby achieving a more realistic reflection of the actual near-surface wave field characteristics, reducing the distortion of the wave field in the depth migration process caused by time domain static correction, and improving the accuracy of migration imaging; and by adopting the first-arrival wave tomographic inversion method to obtain the near-surface velocity model, it is ensured that the velocity field of the low-deceleration zone with small thickness, low velocity, drastic changes and large time delay is reasonably applied in depth migration.

[0012] Furthermore, the S102 includes the following steps: S1021. Use the tomographic static correction and residual static correction fusion technology, combined with the migration starting surface, to fine-tune the near-surface lateral velocity difference in the initial near-surface velocity model; S1022. In response to the preliminary static correction, constrain the first-arrival wave tomography static correction inversion through micro-logging to obtain a fine near-surface velocity model, and perform encrypted and equally spaced sampling on the fine near-surface velocity model to obtain an optimized near-surface velocity model.

[0013] The beneficial effects of the above further scheme are as follows: the present invention adopts the fusion technology of tomographic static correction and residual static correction to realize the correction of near-surface lateral velocity differences, and makes the near-surface velocity model meet the requirements of establishing the medium-deep velocity field on the common center point gather reference surface; and constrains the first-arrival wave tomographic static correction inversion through micro-logging, so that the obtained fine near-surface velocity model is matched with the static correction, and uses encrypted equal-interval sampling to achieve lossless output of the optimized near-surface velocity model, laying a good foundation for subsequent velocity fusion.

[0014] Furthermore, the S103 includes the following steps: S1031. Based on the seismic exploration observation data, use the replacement velocity of the study area as the near-surface velocity fusion interface, the replacement velocity smooth surface of the study area as the splicing surface, and the ray density interface smooth surface as the fusion surface; S1032. Based on the fusion surface, the splicing surface is fused into the optimized near-surface velocity model and the mid-deep velocity model to establish an initial depth migration velocity field for first-arrival wave and reflection wave joint depth velocity modeling.

[0015] The beneficial effects of the above further scheme are as follows: the present invention uses the replacement velocity of the study area as the basis for selecting the near-surface velocity fusion interface, and uses its replacement velocity smooth surface as the splicing surface; and uses the ray density interface smooth surface as the fusion surface, thereby improving the reliability of the near-surface inversion velocity above the fusion surface, achieving the near-surface velocity above the fusion surface close to the seismic velocity below the fusion surface, and avoiding the artificially caused seismic traveltime anomaly due to the addition of the fusion surface; adopting a method of fusing the near-surface velocity inverted by tomographic static correction with the floating surface velocity model of the common center point gather, thereby reducing the influence of the near-surface low velocity reduction zone on the traveltime calculation of the underlying strata.

[0016] Furthermore, the S2 includes the following steps: S201, fusing the near-surface velocity model and the mid-deep layer velocity model fused by the splicing surface, replacing the velocity between the splicing surface and the surface with the replacement velocity, establishing an equivalent true surface velocity field, and obtaining a fused velocity model; S202, removing the tomographic static correction amount obtained by the near-surface velocity model, regressing the seismic exploration observation data to a value close to the actual surface, and using the elevation or elevation smooth surface as a new floating surface. The common center point gathers are moved as a whole to a small smooth surface corresponding to the new floating surface. Using travel time equivalence as a criterion, the splicing surface as the base, and the final datum surface as the top, the travel times of the refined near-surface velocity model and the fused velocity model are respectively obtained. S203, applying the travel time difference between the refined near-surface velocity model and the fused velocity model as a high-frequency quantity to the seismic exploration observation data to correct the seismic exploration observation data, and using the residual static correction method of the seismic exploration observation data to calibrate the travel time error caused by sampling the refined near-surface velocity model; S204, calibrating the seismic exploration observation data from the common center point gather floating surface to the true surface corresponding to the true surface velocity field of the fused velocity model, performing depth step sampling and smoothing processing on the refined near-surface velocity model, fusing the offset velocity to the surface velocity of the near-true surface velocity field, and performing correction processing on the seismic exploration observation data; S205. Select a marker layer with a strong reflection interface as a constraint framework, pick N horizons from shallow to deep based on geological facies and structural styles, establish an initial structural model, convert the pre-stack time migration root mean square velocity field into interval velocity in the depth domain through a constrained velocity inversion method, and form an initial background velocity for pre-stack depth migration. Under the constraints of a large set of geological horizons, combine well information from multiple angles, normalize the well logging P-wave velocity curve, extract the low-frequency component, and select the low-frequency information of the well logging P-wave as the correction standard. S206. Based on the constraint framework and correction standards, a large-scale floating surface velocity field of the common center point gather is formed, and the fused velocity model is corrected based on seismic data drive, seismic-geological integration, and multi-round deep fusion to obtain a large-scale floating surface velocity model of the common center point gather.

[0017] The beneficial effects of the above-mentioned further solution are as follows: the present invention adopts a method for converting the true surface to a small smooth surface, and the velocity model obtained is based on the true surface. For small smooth surfaces above the surface, replacement velocities are used to fill in the area, and the area below the area is directly removed. The true surface velocity model is transitioned to the small smooth surface velocity model, thereby improving the requirements for layer imaging accuracy. By applying travel time differences as high-frequency quantities to the data, errors caused by near-surface model processing, fusion, and migration surface changes are eliminated. After the migration velocity is fused with the near-surface velocity, the data is subjected to corresponding inverse correction or other time correction processing, thereby improving the improvement effect of migration imaging. The present invention calibrates the seismic exploration observation data from the common center point gather floating surface to the real surface corresponding to the real surface velocity field of the fused velocity model, and adopts depth step sampling and smoothing processing for the near-surface velocity model, so that the near-surface velocity model can effectively express the near-surface wave field and better realize splicing with the depth velocity model, thereby improving the coupling effect between the near-surface model and the offset surface; based on the integration of seismic data drive and seismic geology and multiple rounds of seismic geological corrections, the accuracy and coupling degree of the fused velocity model are continuously improved.

[0018] Furthermore, the S3 includes the following steps: S301. Extracting a dip volume, an azimuth volume, and a continuity factor volume from the offset volume based on an isotropic offset seismic database, defining a reflection surface and its direction, storing reflection points on the reflection surface in a preset database, and performing ray tracing toward the surface using the reflection points based on different opening angles and azimuths to obtain an automatic offset volume picking method; S302, automatically picking up the residual delay in the offset gather to obtain an automatic picking method for the residual delay; S303, creating a tomographic imaging matrix using an automatic picking method based on the offset volume and the residual delay and a point-to-feature operation mode based on anisotropic ray tracing; S304, based on the tomography matrix, using small-scale high-resolution geological attributes, constraining the grid tomography along the layer to obtain multi-scale grid tomography; S305. Based on the multi-scale grid tomography, a high-resolution processing technique is used to process the seismic data volume in the isotropic migration seismic database, and the seismic data volume is denoised, and a dip volume, an azimuth volume, and a continuity factor volume of the high-resolution processed and denoised seismic data volume are calculated. S306. Perform ray tracing based on the inclination body, azimuth body, and continuity factor body obtained in S305 to obtain velocity variation. Based on the velocity variation, fuse the seismic exploration observation data with the common center point gather large-scale floating surface velocity model to form an initial true surface isotropic velocity field, and perform modeling to obtain a true surface isotropic velocity model. S307: Using well information constraints, the velocity error is taken as the depth difference between the migration imaging layer and the well layer, thereby obtaining an anisotropic vertical velocity field and an initial anisotropic parameter field. The anisotropic parameters in the initial anisotropic parameter field are iteratively updated using the anisotropic migration value, and the dip angle field and azimuth angle field reflecting the tilted symmetry axis are obtained by calculating the dip angle of the formation structure. S308. In response to establishing the true surface isotropic velocity model, based on the dip field and the azimuth field, according to the geological information and anisotropic parameters in the seismic exploration observation data, combined with the horizon model, interpolation and extrapolation along the layer are performed to obtain the anisotropic parameter field.

[0019] Furthermore, the step S304 includes the following steps: S3041. Based on the tomographic matrix, ray tracing forward modeling is performed using structural horizons and internal sublayers as reflection points. Globally uniformly distributed points are used as the inversion grid. The conjugate gradient method is used to obtain the least squares solution. Interpolation is performed under the constraints of the structural model to update the depth velocity volume. S3042. Solve the tomography matrix using different geological constraints to obtain multi-scale grid tomography; The tomography matrix solving steps are specifically as follows: Using the common reflection point gather generated by prestack depth migration of the target line, the residual delay is defined as the residual delay value of the common reception point gather at the reflection point position. The residual delay of the gather is picked up using the common reflection point gather, and the residual delay is used to perform step-by-step inversion of the layer velocity of the formation.

[0020] The beneficial effects of the above further solution are as follows: the present invention achieves the regression of seismic exploration observation data to a value close to the actual surface by removing the tomographic static correction value obtained by the near-surface velocity model; the present invention applies the travel time difference between the refined near-surface velocity model and the fused velocity model as a high-frequency quantity to the seismic exploration observation data, eliminating errors caused by near-surface model processing, fusion, and offset surface changes; This invention combines a new deep velocity modeling workflow with a grid tomographic inversion technique that constrains the layers from shallow to deep. The velocity field and the anisotropy parameter field are updated iteratively in an interleaved manner. The anisotropy parameter field is introduced at the beginning of the modeling process, and anisotropic modeling is performed layer by layer for each set of well stratification. This allows the model to be constrained by well stratification throughout the entire modeling process, eliminating errors in imaging and well stratification caused by anisotropy layer by layer. As a result, the cumulative depth error of each small loop iteration between layers is small, accelerating the iterative convergence rate and improving the consistency of the obtained continuous anisotropy results with the well stratification. The present invention is based on grid tomography technology, which is a powerful supplement to layer-based tomography technology and solid model-based tomography technology. It is committed to flattening each strong phase axis of each depth-migrated CRP data gather, regardless of whether the phase axis is located at the layer position or between two layers, ensuring the imaging effect of areas with interlayer velocity anomalies or large interlayer velocity variations.

[0021] Furthermore, the S4 includes the following steps: S401, extracting the azimuth and dip data volumes of the isotropic migration seismic data volumes from the seismic data volumes in the isotropic migration seismic database, and modeling them using anisotropy to establish a depth velocity model; S402, obtaining anisotropic velocity by calculating various parameters of the anisotropic parameter field, and performing multiple iterations on the constrained and error-eliminated depth velocity model using a grid tomographic inversion method to obtain anisotropic velocity field for prestack depth migration; S403. Based on the anisotropic velocity field of pre-stack depth migration, multi-scale fracture detection is performed on the seismic data volume to obtain a discontinuity data volume. Based on the true surface isotropic velocity model and the interpreted horizon, combined with well information constraints, the relative coefficients of the velocities of each formation discontinuity volume and the average formation velocity are analyzed. The discontinuity data volume is 3D-carved to obtain a 3D discontinuity data volume. The relative coefficients are used to update the velocities of each formation to obtain a refined velocity model, and the refined velocity model is repeatedly updated and constructed. S404. In response to the updated refined velocity model, the depth velocity model is constrained by using well location stratification and the errors in imaging and well location stratification caused by anisotropy are eliminated layer by layer to obtain a multi-information constrained anisotropic parameter field and to model a multi-information constrained anisotropic velocity model.

[0022] The beneficial effects of the above further scheme are as follows: the present invention updates the velocity model after repeated iterations of the network tomography inversion method to obtain the final pre-stack depth migration velocity anisotropy velocity model. During the modeling process, through the deep fusion and processing interpretation of seismic data drive and geological multi-information guidance, and multiple rounds of seismic and geological feedback, the accuracy of the depth domain velocity model is continuously improved, the depth and occurrence accuracy of the geological target are improved, and the requirements of on-site tracking for the imaging accuracy of the target layer are met. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] Figure 1 Flow chart of the method of the present invention.

[0024] Figure 2 FIG. 4 is a flowchart of the grid tomography of this embodiment.

[0025] Figure 3 This is the first comparison diagram of the new prestack depth migration results and the previous processing results in this embodiment.

[0026] Figure 4 This is the second comparison chart between the new prestack depth migration results and the previous processing results in this embodiment.

[0027] Figure 5 This is the third comparison diagram between the new prestack depth migration results and the previous processing results in this embodiment. DETAILED DESCRIPTION

[0028] The specific embodiments of the present invention are described below to facilitate understanding of the present invention by those skilled in the art. However, it should be clear that the present invention is not limited to the scope of the specific embodiments. For those skilled in the art, as long as various changes are within the spirit and scope of the present invention as defined and determined by the appended claims, these changes are obvious, and all inventions and creations utilizing the concepts of the present invention are protected.

[0029] Before describing this embodiment, the following terms are explained: CMP: common midpoint gathers; CDP: common reflection point gather; CRP: Common Acceptance Point Gather; PPF: point pair feature; RMS speed: root mean square speed; Kirchhoff prestack depth migration: Kirchhoff prestack depth migration; TTI: transverse isotropy about the tilted axis of symmetry; Pencils database: a new library that stores the locations of reflection points; Tomography RMO: Tomography Rayleigh Mode refers to an interference phenomenon caused by the scattering of seismic waves during propagation due to the inhomogeneity of the underground medium. QC: Quality Control.

[0030] Example In this embodiment, the idea and strategy for depth velocity modeling are as follows: The traditional migration velocity iteration process aims to flatten the reflection wave phase axis, that is, to use the residual delay of the reflection wave (residual RMS velocity) to invert the accurate migration velocity. However, in shallow layers, due to the low number of reflection wave coverage and high noise development, this method has significant limitations. For prestack depth migration, the accuracy of shallow layer velocity directly affects the imaging quality of the underlying strata. Therefore, the depth velocity modeling strategy adopted by the present invention is as follows: the first arrival wave tomographic inversion is used to obtain an accurate near-surface velocity field, which is then integrated with the mid-deep velocity model obtained from the large floating surface of the reflected wave to form an initial depth migration velocity field. Considering the influence of the near-surface low velocity reduction zone on the travel time calculation of the underlying strata, a method is adopted to integrate the near-surface velocity obtained by tomographic static correction inversion with the common center point gather floating surface velocity model; on this basis, the velocity model is optimized by using the reflected wave tomographic velocity inversion to build a true surface isotropic pre-stack depth migration velocity field; based on the isotropic migration seismic database, its azimuth and dip data are extracted, and then the various parameters of the anisotropic model are calculated and the anisotropic velocity is obtained after anisotropic model analysis. On this basis, the model is updated after repeated iterations of the network tomographic inversion method to obtain the final pre-stack depth migration velocity anisotropic velocity model.

[0031] During the modeling process, the accuracy of the depth domain velocity model is continuously improved through the deep integration and processing interpretation of seismic data drive and geological multi-information guidance, as well as multiple rounds of seismic and geological feedback. This improves the accuracy of the depth and occurrence of geological targets, and meets the requirements of field tracking for the imaging accuracy of the target layer.

[0032] In this embodiment, all velocity models are three-dimensional velocity bodies that vary with x, y, and z.

[0033] Based on this, Figure 1 As shown, the present invention provides a real surface seismic geological integrated velocity modeling and prestack depth imaging method, which is implemented as follows: S1. Collect seismic exploration observation data and isotropic migration seismic database in the study area to obtain the mid-deep velocity model. Use first-arrival wave tomographic inversion to obtain the near-surface velocity model. Select the splicing surface and fusion surface. Based on the fusion surface, fuse the splicing surface to the near-surface velocity model and the mid-deep velocity model to form the initial depth migration velocity field. The specific steps are as follows: S101. Collect seismic exploration observation data and isotropic migration seismic database in the study area, perform small-scale smoothing on the surface elevation data to obtain the migration starting surface, build a mid-deep velocity model based on the large floating surface of the reflected wave, and perform first-arrival tomographic inversion to build an initial near-surface velocity model. The specific steps are as follows: S1011. Collect seismic exploration observation data and an isotropic migration seismic database in the study area, perform small-scale smoothing on the real surface elevation data in the seismic exploration observation data, and obtain a migration starting surface; S1012. Based on the large floating surface of the reflected wave, a velocity model of the mid-deep layer is built. Using the first arrival of the cannon with the preset offset distance, first arrival tomographic inversion is performed to build an initial near-surface velocity model.

[0034] In this embodiment, the migration surface is selected by performing a small-scale smoothing process on the true surface elevation of the seismic exploration observation data collected in the study area to obtain the migration starting surface, also known as the small smoothing migration surface, or simply the small smoothing surface. Its purpose is to more realistically reflect the true near-surface wavefield characteristics, reduce the distortion of the wavefield during the depth migration process caused by time-domain static correction, and thus improve the accuracy of the migration imaging. Small smooth surface migration is a method for approximating the true surface. Given the differences between conventional floating surfaces and the true surface, it has the following advantages: Small smooth surfaces are still floating surfaces that are extremely close to the true surface, making it easy to transition and correct data from floating surfaces to small smooth surfaces. Furthermore, most common migration algorithms, such as the Kirchhoff algorithm, are based on floating surfaces, so using small smooth surface migration is consistent with the applicability of the algorithm. According to the large floating surface of the reflected wave, the mid-deep velocity model is modeled to obtain the mid-deep velocity model; in order to obtain a near-surface velocity model that is similar to the actual near-surface velocity structure, the first arrival of the cannon with a preset offset distance is selected to perform first arrival tomographic inversion and establish the initial near-surface velocity model.

[0035] S102: Using the tomographic static correction and residual static correction fusion technology, combined with the migration starting surface, the initial near-surface velocity model is fine-tuned. In response to the preliminary static correction, the initial near-surface velocity model is constrained by micro-logging, and densely sampled at equal intervals to obtain an optimized near-surface velocity model. The specific steps are as follows: S1021. Use the tomographic static correction and residual static correction fusion technology, combined with the migration starting surface, to fine-tune the near-surface lateral velocity difference in the initial near-surface velocity model; S1022. In response to the preliminary static correction, constrain the first-arrival wave tomography static correction inversion through micro-logging to obtain a fine near-surface velocity model, and perform encrypted and equally spaced sampling on the fine near-surface velocity model to obtain an optimized near-surface velocity model.

[0036] In this embodiment, the near-surface lateral velocity difference in the initial near-surface velocity model is fine-tuned by using the tomographic static correction and residual static correction fusion technology in combination with the migration starting surface; In response to the preliminary static correction during fine-tuning using the tomographic static correction and residual static correction fusion technology, a fine near-surface velocity model matching the static correction is obtained through micro-logging constrained first-arrival wave tomographic static correction inversion. The fine near-surface velocity model is then densely sampled at equal intervals to output the optimized near-surface velocity model.

[0037] S103. Based on the seismic exploration observation data, a splicing surface and a fusion surface are selected. Based on the fusion surface, the splicing surface is fused into the optimized near-surface velocity model and the mid-deep velocity model to establish an initial depth migration velocity field. The specific steps are as follows: S1031. Based on the seismic exploration observation data, use the replacement velocity of the study area as the near-surface velocity fusion interface, the replacement velocity smooth surface of the study area as the splicing surface, and the ray density interface smooth surface as the fusion surface; S1032. Based on the fusion surface, the splicing surface is fused into the optimized near-surface velocity model and the mid-deep velocity model to establish an initial depth migration velocity field for first-arrival wave and reflection wave joint depth velocity modeling.

[0038] In this embodiment, based on seismic exploration observation data, the replacement velocity of the study area is used as the near-surface velocity fusion interface, the replacement velocity smooth surface of the study area is used as the splicing surface, and the ray density interface smooth surface is selected as the fusion surface. Based on the fusion surface, the splicing surface is fused into the optimized near-surface velocity model and the mid-deep velocity model to complete the initial depth migration velocity field of the first arrival wave and reflection wave joint depth velocity modeling. The requirements for selecting the fusion surface are as follows: since the near-surface inversion velocity above the fusion surface must be more reliable, in practice the ray density of the reference tomographic inversion is used. Therefore, when selecting, the ray density interface smooth surface can be used as the reference position for selecting the fusion surface; the replacement velocity of the study area is used as the basis for selecting the near-surface velocity fusion interface, and its replacement velocity smooth surface is used as the splicing surface; since the near-surface velocity above the fusion surface should be as close as possible to the seismic velocity below the fusion surface, the artificial travel time anomaly of the earthquake caused by the addition of the fusion surface is avoided; after the fusion surface selection is completed, the processed near-surface velocity model and the conventional floating surface velocity modeling velocity are well spliced to complete the initial depth migration velocity field of the first arrival wave and reflection wave joint depth velocity modeling.

[0039] S2. The near-surface velocity model and the mid-deep velocity model fused by the splicing surface are fused to obtain a fused velocity model. The common center point gathers are corrected. Based on the geological facies, structural style, and well information, a constraint framework and correction criteria are selected from multiple angles. The fused velocity model is corrected based on seismic data drive, seismic-geological integration, and multiple rounds of deep fusion to obtain a large-scale floating surface velocity model of the common center point gathers. The specific steps are as follows: S201. Fusing the near-surface velocity model and the mid-deep layer velocity model fused by the splicing surface, replacing the velocity between the splicing surface and the surface with the replacement velocity, establishing an equivalent true surface velocity field, and obtaining a fused velocity model.

[0040] In this embodiment, the optimized near-surface velocity model and the mid-deep velocity model after the splicing surface fusion are fused. The fusion processing method is as follows: the velocity between the splicing surface and the surface is replaced by the replacement velocity to establish an equivalent true surface velocity field. The fusion process has good adaptability in areas with thin and stable low-velocity reduction zones. The fused velocity model is obtained through the fusion process. Near-surface model fusion is the key and most challenging aspect of the entire process, centered around selecting the fusion surface and matching the data. The fusion process also addresses the coupling issue between the refined near-surface model and migration. Because depth migration requires discrete sampling (extend step) of the velocity field in the x, y, and z directions, such as the commonly used 5m step, the accuracy of the velocity field after discrete sampling is significantly reduced for the near-surface model, impacting the quality of the migration imaging. This is primarily manifested in the following ways: Because the depth migration sampling step is much larger than the spatial sampling scale of the near-surface model obtained through inversion, the depth-migrated velocity model cannot accurately represent near-surface characteristics. The near-surface low-velocity zone is often thin and has the most significant impact on the true wavefield. Furthermore, the wavefield under the refined near-surface velocity model is complex, making it difficult for existing migration methods to effectively represent it, resulting in distorted migration imaging. During the processing process, depth-step sampling and smoothing are applied to the near-surface velocity model, enabling it to effectively represent the near-surface wavefield and better integrate with the depth velocity model.

[0041] S202. Remove the tomographic static correction obtained by the near-surface velocity model, regress the seismic exploration observation data to a value close to the actual surface, and use the elevation or elevation smooth surface as the new floating surface. Move the common center point gather as a whole to the small smooth surface at the corresponding position of the new floating surface. Use travel time equivalence as the criterion, the splicing surface as the bottom, and the final datum surface as the top to respectively obtain the travel times of the refined near-surface velocity model and the fused velocity model.

[0042] In this embodiment, in order to replace the conventional floating surface, the seismic exploration observation data is regressed to a surface close to the real surface by removing the tomographic static correction amount obtained by the near-surface velocity model. However, since the seismic exploration observation data contains residual static correction amount, it is not completely equivalent to the real surface. The elevation (real surface) or elevation smooth surface (small smooth surface) is used as the new floating surface, that is, the low-frequency amount, and the common center point gather is moved as a whole to the small smooth surface at the corresponding position of the new floating surface, that is, (the common center point gather is corrected to the real surface corresponding to the real surface velocity field in the fused velocity model); and the travel time equivalence is used as the criterion, with the splicing surface as the bottom and the final reference surface as the top, to respectively calculate the travel time of the refined near-surface model and the fused velocity model. The present invention adopts a conversion method between the true surface and the small smooth surface. The velocity model obtained is based on the true surface. Due to the small-scale difference between the surface elevation and the elevation of the small smooth surface, the most reasonable processing method at present is to fill up and cut down, that is, for the small smooth surface above the surface, the replacement velocity is used to fill it; the part below it is directly cut off; the true surface velocity model can be transitioned to the small smooth surface velocity model, thereby greatly improving the requirements for layer imaging accuracy.

[0043] S203. Apply the travel time difference between the refined near-surface velocity model and the fused velocity model as a high-frequency quantity to the seismic exploration observation data to correct the seismic exploration observation data, and use the residual static correction method of the seismic exploration observation data to calibrate the travel time error caused by sampling the refined near-surface velocity model.

[0044] In this embodiment, the travel time difference between the refined near-surface velocity model and the fused velocity model is used as a high-frequency quantity and applied to the seismic exploration observation data to eliminate errors caused by near-surface model processing, fusion, and offset surface changes, and to perform corresponding corrections on the seismic exploration observation data. Since the sampled refined near-surface velocity model will cause travel time errors, the residual static correction method of the seismic exploration observation data is used to calibrate the travel time errors caused by the sampling of the refined near-surface velocity model. Since the residual static correction is used as a supplement to the near-surface model error and is independent of the near-surface model fusion process, it is applied and retained in the present invention.

[0045] S204. Calibrate the seismic exploration observation data from the common center point gather floating surface to the true surface corresponding to the true surface velocity field of the fused velocity model, perform depth step sampling and smoothing on the refined near-surface velocity model, fuse the offset velocity to the surface velocity of the near-true surface velocity field, and perform correction processing on the seismic exploration observation data.

[0046] In this embodiment, when calibrating the seismic exploration observation data from the floating surface of the common center point gather to the true surface corresponding to the true surface velocity field of the fused velocity model, data matching is required to fuse the offset velocity to the surface velocity of the near-true surface velocity field, and perform corresponding inverse correction or other time correction processing on the seismic exploration observation data; The present invention calibrates the seismic exploration observation data from the CMP surface to the real surface corresponding to the corresponding real surface velocity field, and adopts depth step sampling and smoothing processing for the near-surface velocity model, so that the near-surface velocity model can effectively express the near-surface wave field and better achieve splicing with the depth velocity model, thereby achieving a better fusion effect between the near-surface model and the offset surface.

[0047] S205. Select a marker layer with a strong reflection interface as a constraint framework, pick N horizons from shallow to deep based on geological facies and structural styles, establish an initial structural model, convert the pre-stack time migration root mean square velocity field into interval velocity in the depth domain through a constrained velocity inversion method, and form an initial background velocity for pre-stack depth migration. Under the constraints of a large set of geological horizons, combine well information from multiple angles, normalize the well logging P-wave velocity curve, extract the low-frequency component, and select the low-frequency information of the well logging P-wave as the correction standard. S206. Based on the constraint framework and correction standards, a large-scale floating surface velocity field of the common center point gather is formed, and the fused velocity model is corrected based on seismic data drive, seismic-geological integration, and multi-round deep fusion to obtain a large-scale floating surface velocity model of the common center point gather.

[0048] In this embodiment, the process of constructing a CMP large-scale smooth floating surface velocity model is as follows: a marker layer of a geologically significant strong reflection interface is selected as a constraint framework, N horizons are picked from shallow to deep, an initial structural model is established, and a stratigraphic structure framework is constructed based on seismic data and geological facies belt division. The initial structural model is a constraint framework for velocity field modeling. The pre-stack time migration root mean square velocity field is converted into layer velocity in the depth domain through a constrained velocity inversion method to form the initial background velocity for pre-stack depth migration. Under the constraints of a large set of geological layers, the velocity field is corrected by using the logging curve. Combined with the multi-angle well information, the logging P-wave velocity curve is normalized, and the low-frequency components in the normalized logging P-wave velocity curve are extracted. The low-frequency information of the logging P wave is selected as the correction standard to constrain the seismic velocity and form a CMP large-scale floating surface velocity field. The fusion velocity model is corrected based on seismic data drive, seismic-geological integration and multi-round deep fusion to obtain a large-scale floating surface velocity model of the common center point gather. Framework constraints and corrections are not completed in one go. Well data will increase with the implementation of the project. Therefore, multi-round fusion is based on multiple updates of well data, and the selection and use of constraint frameworks and correction standards are also carried out multiple times.

[0049] S3. Based on the isotropic migration seismic database, multi-scale grid tomography is performed. The common center point gather large-scale floating surface velocity model is used to build a true surface isotropic velocity model and obtain the anisotropic parameter field. The specific steps are as follows: S301. Extracting a dip volume, an azimuth volume, and a continuity factor volume from the offset volume based on an isotropic offset seismic database, defining a reflection surface and its direction, storing reflection points on the reflection surface in a preset database, and performing ray tracing toward the surface using the reflection points based on different opening angles and azimuths to obtain an automatic offset volume picking method; S302, automatically picking up the residual delay in the offset gather to obtain an automatic picking method for the residual delay; S303: Based on the automatic picking method of the offset volume and the residual delay, a point-to-feature operation mode based on anisotropic ray tracing is adopted to create a tomographic imaging matrix.

[0050] In this example, to establish a detailed structural model, the Xiaoping velocity model, after near-surface fusion, is used as the initial model in the depth domain. Under the constraints of the geological structural model, multi-scale grid tomography is used to gradually characterize the velocity field. Large-scale grid tomography is first used throughout the entire region, with multiple rounds of iterations gradually updating velocity information at structural levels and between layers. Once relatively accurate velocities are obtained, small-scale grid tomography and multi-attribute constraints are used to enhance the accuracy of tomographic imaging of large velocity body models and complex geological volumes. The present invention is challenging to perform tomography for updating subsurface velocities using CRP gathers obtained after a preset Kirchhoff prestack depth migration. The tomography equation linearly links the travel-time error along the reflected ray to the updated subsurface velocity model parameters. The ray is traced from dense subsurface reflection points to the surface according to different reflection angles and azimuths. In practice, the present invention determines the travel-time error from the unflattened common imaging gathers in the offset range or angle domain. The goal of tomography is to find an optimal update parameter to globally minimize the travel-time error using a constrained least squares method. This results in a model that is infinitely close to the subsurface structure, which can be used for a new round of migration, ultimately resulting in an ideal migration output gather.

[0051] In this embodiment, to enhance the accuracy of velocity tomography for large velocity models and complex geological volumes, grid tomography supports multi-azimuth updates, anisotropic parameter updates, and automatic picking of individual reflectors and residual delays. All operations in this method (including data allocation, automatic picking, matrix formation and matrix solution during grid tomography operations) utilize the PPF model to ensure efficient processing of large computational workloads. The layer-based tomography and solid model-based tomography technologies mainly consider the average layer velocity of a large set of layers, have an averaging effect on the layer velocity between layers, and are suitable for most phase axes within a layer. However, it is difficult to guarantee the accuracy of those areas with abnormal inter-layer velocity or large inter-layer velocity variations, and the velocity is either too high or too low. The grid-based tomography technology is a powerful supplement to the layer-based tomography and solid model-based tomography technologies, and is committed to flattening each strong phase axis of each depth-migrated CRP data gather, regardless of whether the phase axis is at the layer position or between two layers.

[0052] In this embodiment, Figure 2 As shown, based on the isotropic migration seismic database, automatic extraction is performed in the migration volume, the dip volume, azimuth volume and continuity factor volume are extracted, the reflection surface and direction are defined, the reflection surface is automatically picked up, the reflection points on the reflection surface are stored in a preset database, and according to different opening angles and azimuths, rays are traced toward the surface using the reflection points to achieve automatic picking and imaging; Figure 2 FASTVEL / VEVLNAN automatic / manual residual speed analysis, respectively, represents three speed data extraction methods, automatic, semi-automatic and manual; Automatically pick up residual delay in offset gathers; residual delay is automatically picked up in previously defined new points. Residual delay data supports two modes: vector and per-trace residual delay value support, and multi-directional residual delay is fully supported by the system. Based on automatic picking imaging and automatic picking residual delay, the point-to-feature operation mode based on anisotropic ray tracing is adopted to create a tomographic imaging matrix. The PPF operation mode adopted improves the computing efficiency of each node on the cluster.

[0053] S304: Based on the tomography matrix, small-scale high-resolution geological attributes are used to constrain the grid tomography along the layer to obtain multi-scale grid tomography. The specific steps are as follows: S3041. Based on the tomographic matrix, ray tracing forward modeling is performed using structural horizons and internal sublayers as reflection points. Globally uniformly distributed points are used as the inversion grid. The conjugate gradient method is used to obtain the least squares solution. Interpolation is performed under the constraints of the structural model to update the depth velocity volume. S3042. Solve the tomography matrix using different geological constraints to obtain multi-scale grid tomography; The tomography matrix solving steps are specifically as follows: Using the common reflection point gather generated by prestack depth migration of the target line, the residual delay is defined as the residual delay value of the common reception point gather at the reflection point position. The residual delay of the gather is picked up using the common reflection point gather, and the residual delay is used to perform step-by-step inversion of the layer velocity of the formation.

[0054] In this embodiment, small-scale, high-resolution grid tomographic modeling is used to control the changes in velocity anomalies caused by faults, etc. According to the tomographic imaging matrix, ray tracing forward modeling is performed using structural horizons and internal small layers as reflection points. Globally uniformly distributed points are used as the inversion grid, and the conjugate gradient method is used to obtain the least squares solution. Interpolation is performed under the constraints of the structural model to update the depth velocity volume. Different geological constraints are used to solve the tomography matrix and obtain multi-scale grid tomography. The specific solution process is as follows: using the common reflection point gather generated by pre-stack depth migration of the target line, the residual delay is defined as the residual delay value of the common reception point gather at the reflection point position, the residual delay of the gather is picked up using the common reflection point gather, and the residual delay is used to perform step-by-step inversion of the layer velocity of the formation.

[0055] S305. Based on the multi-scale grid tomography, a high-resolution processing technique is used to process the seismic data volume in the isotropic migration seismic database, and the seismic data volume is denoised, and a dip volume, an azimuth volume, and a continuity factor volume of the high-resolution processed and denoised seismic data volume are calculated. S306. According to the inclination body, azimuth body and continuity factor body of S305, ray tracing is performed to obtain velocity variation. Based on the velocity variation, the seismic exploration observation data is integrated with the large-scale floating surface velocity model of the common center point gather to form an initial true surface isotropic velocity field, and modeling is performed to obtain a true surface isotropic velocity model.

[0056] In this embodiment, based on a multi-scale grid tomography method, a high-resolution processing technique is used to process the seismic data volume in the isotropic migration seismic database, and the seismic data volume is denoised. Then, the dip volume, azimuth volume, and continuity factor volume of the high-resolution processed and denoised seismic data volume are calculated. Ray tracing is performed based on this, and the detailed changes in velocity are recorded. Based on the detailed changes in velocity, the seismic exploration observation data are fused with a large-scale floating surface velocity model of a common center point gather to form an initial true surface isotropic velocity field, and modeling is performed to obtain a true surface isotropic velocity model.

[0057] S307: Using well information constraints, the velocity error is taken as the depth difference between the migration imaging layer and the well layer, thereby obtaining an anisotropic vertical velocity field and an initial anisotropic parameter field. The anisotropic parameters in the initial anisotropic parameter field are iteratively updated using the anisotropic migration value, and the dip angle field and azimuth angle field reflecting the tilted symmetry axis are obtained by calculating the dip angle of the formation structure. S308. In response to establishing the true surface isotropic velocity model, based on the dip field and the azimuth field, according to the geological information and anisotropic parameters in the seismic exploration observation data, combined with the horizon model, interpolation and extrapolation along the layer are performed to obtain the anisotropic parameter field.

[0058] In this embodiment, after establishing the isotropic velocity model, the present invention incorporates well information constraints, converts the depth difference between the offset imaged layer and the well layer into a velocity error, and obtains an anisotropic vertical velocity field and an anisotropic parameter field. The anisotropic parameters in the anisotropic parameter field are then iteratively updated using the anisotropic offset value. Finally, by calculating the dip angle of the formation structure, the dip angle field and azimuth angle field reflecting the tilted symmetry axis are obtained. When establishing the isotropic velocity model, the geological information and anisotropic parameters obtained from the drilled wells are combined with the layer model to perform interpolation and extrapolation along the layer to finally obtain the anisotropic parameter field.

[0059] S4. Based on the seismic exploration observation data and the isotropic migration seismic database, anisotropic parameter fields are used to build a model to obtain a depth velocity model. The depth velocity model is constrained and errors are eliminated using well location stratification, and a multi-information constrained anisotropic velocity model is built. The specific steps are as follows: S401, extracting the azimuth and dip data volumes of the isotropic migration seismic data volumes from the seismic data volumes in the isotropic migration seismic database, and modeling them using anisotropy to establish a depth velocity model; S402. Anisotropic velocity is obtained by calculating various parameters of the anisotropic parameter field. The constrained and error-eliminated depth velocity model is iterated multiple times using a grid tomographic inversion method to obtain anisotropic velocity field for prestack depth migration.

[0060] In this embodiment, although the layer velocity constraint is performed with reference to the well velocity during the isotropic velocity modeling stage, when the well-seismic velocity consistency survey is conducted on the isotropic results, there will still be local unsatisfactory consistency. This is because the well velocity constraint in the isotropic migration stage is only based on the principle of common imaging point flattening, which reduces the multi-solution range of travel time error. However, using a simple isotropic algorithm for migration imaging cannot really solve the anisotropy problem. However, anisotropy is widely present in nature. To obtain pre-stack depth migration results that are closer to the actual underground structure, it is necessary to carry out anisotropic velocity modeling to meet the current accuracy requirements for exploration and mining. In particular, the present invention requires that the depth domain seismic results should be used to predict the horizontal well guidance of the main target layers such as the multi-peak group, improve the box drilling rate, and meet the requirements of drilling tracking. Therefore, higher requirements are placed on pre-stack depth imaging. Based on the seismic data volume in the isotropic migration seismic database, the azimuth and dip data volumes of the isotropic migration seismic data volume are extracted, and modeling and analysis are carried out through anisotropy. A depth velocity model is established, and the various parameters of the anisotropic parameter field are calculated and the anisotropic velocity is obtained. Then, the model is updated after repeated iterations through the grid tomography inversion method to obtain the anisotropic velocity field for pre-stack depth migration.

[0061] In this embodiment, the currently common depth-velocity modeling process is as follows: first, an isotropic velocity model is established. In this modeling stage, migration is performed according to the isotropic assumption, and only event focusing is considered. The coincidence of the imaged layer and the well layer is not considered. Even if well information constraints are added, they are generally limited to the low-frequency component of the acoustic logging velocity to constrain the large velocity trend of each layer. Second, after the isotropic velocity field is established, the depth difference between the migrated imaged layer and the well layer is converted into a velocity error to obtain the anisotropic vertical velocity field (Vp0) and the anisotropic parameter field. The anisotropic migration is then used to calculate the anisotropic parameter delta ( ) and epsilon ( ) for iterative update; finally, by obtaining the tectonic dip angle of the stratum, the dip angle field and azimuth angle field reflecting the TTI symmetry axis are obtained.

[0062] However, this process often suffers from the following issues: During the isotropic velocity modeling phase, there are no well location constraints. When anisotropy is significant in the region, the imaging depth will differ significantly from the well location stratification, resulting in significant errors when converting to Vp0. To address these issues, a new depth-velocity modeling workflow and a grid-based tomographic inversion technique with stratum constraints, working from shallow to deep, have been adopted. Specifically, the velocity field and the anisotropy parameter field are updated iteratively, with the anisotropy parameter field introduced at the outset of modeling. Anisotropy modeling is performed layer by layer for each well location stratification. Throughout the modeling process, the well location stratification is consistently used to constrain the model, gradually eliminating errors in imaging and well location stratification caused by anisotropy. As a result, the cumulative depth error of each small iteration between layers is small, the iterative convergence is rapid, and the resulting continuous anisotropy results are generally highly consistent with the well stratification.

[0063] S403. Based on the anisotropic velocity field of pre-stack depth migration, multi-scale fracture detection is performed on the seismic data volume to obtain a discontinuity data volume. Based on the true surface isotropic velocity model and the interpreted horizon, combined with well information constraints, the relative coefficients of the velocities of each formation discontinuity volume and the average formation velocity are analyzed. The discontinuity data volume is 3D-carved to obtain a 3D discontinuity data volume. The relative coefficients are used to update the velocities of each formation to obtain a refined velocity model, and the refined velocity model is repeatedly updated and constructed. S404. In response to the updated refined velocity model, the depth velocity model is constrained by using well location stratification and the errors in imaging and well location stratification caused by anisotropy are eliminated layer by layer to obtain a multi-information constrained anisotropic parameter field and to model a multi-information constrained anisotropic velocity model.

[0064] In this embodiment, multi-information constrained anisotropic modeling is performed. Based on the anisotropic velocity field of pre-stack depth migration, multi-scale fracture detection is performed on the seismic data volume to obtain a discontinuity data volume. According to the true surface isotropic velocity model and the interpreted horizon, combined with well information constraints, the relative coefficients of the velocities of the discontinuity bodies of each formation and the average velocity of the formation are analyzed. The discontinuity data volume is three-dimensionally carved to obtain a three-dimensional discontinuity data volume. Combined with the relative coefficients, the velocity of each formation is updated to obtain a refined velocity model, and the refined velocity model is repeatedly updated and constructed. In response to the updated and constructed refined velocity model, the depth velocity model is constrained by using well location stratification, and the errors between imaging and well location stratification caused by anisotropy are eliminated layer by layer to obtain a multi-information constrained anisotropic parameter field. The multi-information constrained anisotropic velocity model is then modeled to obtain a multi-information constrained anisotropic velocity model. The expression of the multi-information constrained anisotropic velocity model is as follows: ; in, represents the multi-information constrained anisotropic velocity model, represents the isotropic velocity model, and are anisotropy parameters, represents the angle of incidence, Indicates the observation range, Indicates the orientation of the crack symmetry axis.

[0065] S5. Use the initial depth migration velocity field as the input of prestack depth migration, and use the multi-information constrained anisotropic velocity model to perform the preset prestack depth migration to generate post-migration seismic imaging results.

[0066] In this embodiment, the preset pre-stack depth migration adopts Kirchhoff pre-stack depth migration. The initial depth migration velocity field is used as the input of the pre-stack depth migration. The multi-information constrained anisotropic velocity model is used to perform Kirchhoff pre-stack depth migration, and finally the post-migration seismic imaging results are generated. like Figure 3 、 Figure 4 ,as well as Figure 5 As shown in the figure, a set of large-scale pre-stack depth migration data with clear formation and fault imaging, relatively reasonable structural strike, and good well-seismic matching effect has been formed; Figure 3 、 Figure 4 ,as well as Figure 5 In the figure, the left side is a picture of existing seismic imaging results, and the right side is a picture of post-migration seismic imaging results formed by the present invention. The present invention solves the problems of existing true surface seismic geological velocity modeling and prestack depth imaging, such as slow modeling speed and poor accuracy, as well as poor imaging quality and low precision.

[0067] In this embodiment, the advantages of the present invention are as follows: First, the present invention establishes a detailed geological structure model, using the near-surface fused small smooth surface velocity field as the initial depth domain model. Under the constraints of the geological structure model, the velocity field characteristics are gradually characterized through multi-scale grid tomography technology. First, a large-scale grid tomography is used throughout the entire area, and multiple rounds of iterations are performed to gradually update the velocity information of the structural layers and between layers. After obtaining relatively accurate velocities, small-scale grid tomography and multi-attribute constraints can be used to enhance the accuracy of tomographic imaging of large-scale velocity body models and complex geological body velocities. Conventional methods are based on tomography of interpreted horizons and tomography technology based on solid models. They mainly consider the average interval velocity of a large set of horizons, have an averaging effect on the interval velocity between layers, and are suitable for most events within a layer. However, they are difficult to guarantee for areas with abnormal or large inter-layer velocity variations, and the velocity is either too high or too low. The grid-based tomography technology of the present invention is a powerful supplement to the layer-based tomography technology and the solid model-based tomography technology, and is dedicated to flattening each strong seismic reflection event of each depth-migrated CRP gather, regardless of whether the event is located at the layer position or between two layers. To enhance the accuracy of tomographic imaging of large-scale velocity models and complex geological bodies, a small-grid tomography technique with multi-attribute constraints is required. Before extracting structural attributes, a combined denoising technique is used to improve the signal-to-noise ratio and enhance the stability of the velocity field.

[0068] Although the isotropic velocity modeling phase employed inter-layer velocity constraints based on well velocities, a well-seismic velocity fit investigation of the isotropic results revealed that the migration results still exhibited localized poor fit. This is because the well velocity constraints during the isotropic migration phase were based solely on the principle of common imaging point flattening, which reduced the range of multi-solutions for traveltime errors. However, using a purely isotropic algorithm for migration imaging does not truly address anisotropy. However, anisotropy is widespread in nature, and to obtain pre-stack depth migration results that are closer to the true subsurface structure, anisotropic velocity modeling is necessary to meet current exploration and production accuracy requirements. Pre-stack depth imaging must meet even higher standards, especially for applications requiring deep-domain seismic data to predict horizontal well guidance for key target layers and improve the box drilling rate.

[0069] Second, the current depth velocity modeling process commonly used in the seismic processing industry often suffers from the following problems: During the isotropic velocity modeling phase, there is a lack of well location constraints. When the anisotropy in the region is significant, the imaging depth will have a large error with the well location layer, and the error will also be large when converting the anisotropic vertical velocity field (Vp0). To address the above-mentioned issues, the present invention utilizes seismic exploration observation data to guide the process of multi-information constrained anisotropic modeling, adjusting the conventional process to a new deep velocity modeling workflow and a grid tomography inversion technique with layer constraints from shallow to deep; namely, the velocity field and the anisotropic parameter field are updated in an alternating and iterative manner, the anisotropic parameter field is introduced at the beginning of the modeling process, and anisotropic modeling is performed layer by layer for each set of well locations. During modeling, the geological information and anisotropic parameters obtained from the wells that have been drilled are fully utilized, and interpolation and extrapolation along the layers are performed in combination with the layer model to obtain a complete anisotropic parameter field. In this way, throughout the entire modeling process, the model is always constrained by the well location layer, and the errors in imaging and well location layering caused by anisotropy are eliminated layer by layer. Therefore, the depth cumulative error of the small cycle iteration between each set of strata is small, the iterative convergence is also faster, and the final continuous anisotropic results are usually more consistent with the well location layering. Practical applications have shown that this technical invention has achieved very good results. The new depth migration profile has been significantly improved compared with the previous old results. Actual drilling has revealed that the drilling rate of horizontal well trajectories has also been greatly improved.

Claims

1. A real surface seismic geological integrated velocity modeling and prestack depth imaging method, characterized in that: The following steps are involved: S1. Collect seismic exploration observation data and isotropic migration seismic database in the study area to obtain the mid-deep velocity model. Use first-arrival wave tomographic inversion to obtain the near-surface velocity model. Select the splicing surface and fusion surface. Based on the fusion surface, fuse the splicing surface to the near-surface velocity model and the mid-deep velocity model to form the initial depth migration velocity field. S2. The near-surface velocity model and the mid-deep velocity model fused by the splicing surface are fused to obtain a fused velocity model. The common center point gathers are corrected. Based on the geological facies, structural style, and well information, a constraint framework and correction criteria are selected from multiple angles. The fused velocity model is corrected based on seismic data drive, seismic-geological integration, and multiple rounds of deep fusion to obtain a large-scale floating surface velocity model of the common center point gathers. S3. Based on the isotropic migration seismic database, multi-scale grid tomography is performed. The true surface isotropic velocity model is modeled using the large-scale floating surface velocity model of the common center point gathers, and the anisotropic parameter field is obtained. S4. Based on the seismic exploration observation data and the isotropic migration seismic database, anisotropic parameter fields are used to perform modeling to obtain a depth velocity model. The depth velocity model is constrained and errors are eliminated using well location stratification to model a multi-information constrained anisotropic velocity model. S5. Use the initial depth migration velocity field as the input of prestack depth migration, and use the multi-information constrained anisotropic velocity model to perform the preset prestack depth migration to generate post-migration seismic imaging results.

2. The real surface seismic-geological integrated velocity modeling and prestack depth imaging method according to claim 1, characterized in that: Said S1 comprises the following steps: S101. Collect seismic exploration observation data and isotropic migration seismic database in the study area, perform small-scale smoothing on the surface elevation data to obtain the migration starting surface, build a mid-deep velocity model based on the large floating surface of the reflected wave, and perform first-arrival tomographic inversion to build an initial near-surface velocity model. S102, using a fusion technique of tomographic statics and residual statics, combined with the migration starting plane, to fine-tune the initial near-surface velocity model; in response to the preliminary statics, constraining the initial near-surface velocity model through micro-logging, and performing densely spaced sampling to obtain an optimized near-surface velocity model; S103. Select a splicing surface and a fusion surface based on the seismic exploration observation data, and fuse the splicing surface into the optimized near-surface velocity model and the mid-deep velocity model based on the fusion surface to establish an initial depth migration velocity field.

3. The real surface seismic-geological integrated velocity modeling and prestack depth imaging method according to claim 2, characterized in that: The S101 includes the following steps: S1011. Collect seismic exploration observation data and an isotropic migration seismic database in the study area, perform small-scale smoothing on the real surface elevation data in the seismic exploration observation data, and obtain a migration starting surface; S1012. Based on the large floating surface of the reflected wave, a velocity model of the mid-deep layer is built. Using the first arrival of the cannon with the preset offset distance, first arrival tomographic inversion is performed to build an initial near-surface velocity model.

4. The real surface seismic-geological integrated velocity modeling and prestack depth imaging method according to claim 2, characterized in that: The S102 includes the following steps: S1021. Use the tomographic static correction and residual static correction fusion technology, combined with the migration starting surface, to fine-tune the near-surface lateral velocity difference in the initial near-surface velocity model; S1022. In response to the preliminary static correction, constrain the first-arrival wave tomography static correction inversion through micro-logging to obtain a fine near-surface velocity model, and perform encrypted and equally spaced sampling on the fine near-surface velocity model to obtain an optimized near-surface velocity model.

5. The real surface seismic-geological integrated velocity modeling and prestack depth imaging method according to claim 2, characterized in that: The S103 includes the following steps: S1031. Based on the seismic exploration observation data, use the replacement velocity of the study area as the near-surface velocity fusion interface, the replacement velocity smooth surface of the study area as the splicing surface, and the ray density interface smooth surface as the fusion surface; S1032. Based on the fusion surface, the splicing surface is fused into the optimized near-surface velocity model and the mid-deep velocity model to establish an initial depth migration velocity field for first-arrival wave and reflection wave joint depth velocity modeling.

6. The real surface seismic-geological integrated velocity modeling and prestack depth imaging method according to claim 4, characterized in that: The S2 comprises the following steps: S201, fusing the near-surface velocity model and the mid-deep layer velocity model fused by the splicing surface, replacing the velocity between the splicing surface and the surface with the replacement velocity, establishing an equivalent true surface velocity field, and obtaining a fused velocity model; S202, removing the tomographic static correction amount obtained by the near-surface velocity model, regressing the seismic exploration observation data to a value close to the actual surface, and using the elevation or elevation smooth surface as a new floating surface. The common center point gathers are moved as a whole to a small smooth surface corresponding to the new floating surface. Using travel time equivalence as a criterion, the splicing surface as the base, and the final datum surface as the top, the travel times of the refined near-surface velocity model and the fused velocity model are respectively obtained. S203, applying the travel time difference between the refined near-surface velocity model and the fused velocity model as a high-frequency quantity to the seismic exploration observation data to correct the seismic exploration observation data, and using the residual static correction method of the seismic exploration observation data to calibrate the travel time error caused by sampling the refined near-surface velocity model; S204, calibrating the seismic exploration observation data from the common center point gather floating surface to the true surface corresponding to the true surface velocity field of the fused velocity model, performing depth step sampling and smoothing processing on the refined near-surface velocity model, fusing the offset velocity to the surface velocity of the near-true surface velocity field, and performing correction processing on the seismic exploration observation data; S205. Select a marker layer with a strong reflection interface as a constraint framework, pick N horizons from shallow to deep based on geological facies and structural styles, establish an initial structural model, convert the pre-stack time migration root mean square velocity field into interval velocity in the depth domain through a constrained velocity inversion method, and form an initial background velocity for pre-stack depth migration. Under the constraints of a large set of geological horizons, combine well information from multiple angles, normalize the well logging P-wave velocity curve, extract the low-frequency component, and select the low-frequency information of the well logging P-wave as the correction standard. S206. Based on the constraint framework and correction standards, a large-scale floating surface velocity field of the common center point gather is formed, and the fused velocity model is corrected based on seismic data drive, seismic-geological integration, and multi-round deep fusion to obtain a large-scale floating surface velocity model of the common center point gather.

7. The method for real surface seismic-geological integrated velocity modeling and prestack depth imaging according to claim 1, characterized in that: The S3 includes the following steps: S301. Extracting a dip volume, an azimuth volume, and a continuity factor volume from the offset volume based on an isotropic offset seismic database, defining a reflection surface and its direction, storing reflection points on the reflection surface in a preset database, and performing ray tracing toward the surface using the reflection points based on different opening angles and azimuths to obtain an automatic offset volume picking method; S302, automatically picking up the residual delay in the offset gather to obtain an automatic picking method for the residual delay; S303, creating a tomographic imaging matrix using an automatic picking method based on the offset volume and the residual delay and a point-to-feature operation mode based on anisotropic ray tracing; S304, based on the tomography matrix, using small-scale high-resolution geological attributes, constraining the grid tomography along the layer to obtain multi-scale grid tomography; S305. Based on the multi-scale grid tomography, a high-resolution processing technique is used to process the seismic data volume in the isotropic migration seismic database, and the seismic data volume is denoised, and a dip volume, an azimuth volume, and a continuity factor volume of the high-resolution processed and denoised seismic data volume are calculated. S306. Perform ray tracing based on the inclination body, azimuth body, and continuity factor body obtained in S305 to obtain velocity variation. Based on the velocity variation, fuse the seismic exploration observation data with the common center point gather large-scale floating surface velocity model to form an initial true surface isotropic velocity field, and perform modeling to obtain a true surface isotropic velocity model. S307: Using well information constraints, the velocity error is taken as the depth difference between the migration imaging layer and the well layer, thereby obtaining an anisotropic vertical velocity field and an initial anisotropic parameter field. The anisotropic parameters in the initial anisotropic parameter field are iteratively updated using the anisotropic migration value, and the dip angle field and azimuth angle field reflecting the tilted symmetry axis are obtained by calculating the dip angle of the formation structure. S308. In response to establishing the true surface isotropic velocity model, based on the dip field and the azimuth field, according to the geological information and anisotropic parameters in the seismic exploration observation data, combined with the horizon model, interpolation and extrapolation along the layer are performed to obtain the anisotropic parameter field.

8. The real surface seismic-geological integrated velocity modeling and prestack depth imaging method according to claim 7, characterized in that: The S304 includes the following steps: S3041. Based on the tomographic matrix, ray tracing forward modeling is performed using structural horizons and internal sublayers as reflection points. Globally uniformly distributed points are used as the inversion grid. The conjugate gradient method is used to obtain the least squares solution. Interpolation is performed under the constraints of the structural model to update the depth velocity volume. S3042. Solve the tomography matrix using different geological constraints to obtain multi-scale grid tomography; The tomography matrix solving steps are specifically as follows: Using the common reflection point gather generated by prestack depth migration of the target line, the residual delay is defined as the residual delay value of the common reception point gather at the reflection point position. The residual delay of the gather is picked up using the common reflection point gather, and the residual delay is used to perform step-by-step inversion of the layer velocity of the formation.

9. The method for real surface seismic-geological integrated velocity modeling and prestack depth imaging according to claim 1, characterized in that: The S4 comprises the following steps: S401, extracting the azimuth and dip data volumes of the isotropic migration seismic data volumes from the seismic data volumes in the isotropic migration seismic database, and modeling them using anisotropy to establish a depth velocity model; S402, obtaining anisotropic velocity by calculating various parameters of the anisotropic parameter field, and performing multiple iterations on the constrained and error-eliminated depth velocity model using a grid tomographic inversion method to obtain anisotropic velocity field for prestack depth migration; S403. Based on the anisotropic velocity field of pre-stack depth migration, multi-scale fracture detection is performed on the seismic data volume to obtain a discontinuity data volume. Based on the true surface isotropic velocity model and the interpreted horizon, combined with well information constraints, the relative coefficients of the velocities of each formation discontinuity volume and the average formation velocity are analyzed. The discontinuity data volume is 3D-carved to obtain a 3D discontinuity data volume. The relative coefficients are used to update the velocities of each formation to obtain a refined velocity model, and the refined velocity model is repeatedly updated and constructed. S404. In response to the updated refined velocity model, the depth velocity model is constrained by using well location stratification and the errors in imaging and well location stratification caused by anisotropy are eliminated layer by layer to obtain a multi-information constrained anisotropic parameter field and to model a multi-information constrained anisotropic velocity model.

Images