Depth domain seismic inversion method with adaptive depth matching

Abstract

This invention provides an adaptive depth-matching depth-domain seismic inversion method, comprising: Step 1, preprocessing well logging data; Step 2, obtaining the time vibration function to establish the relationship between depth-domain wave impedance and seismic records; Step 3, performing depth-domain well-seismic matching correction to achieve optimal alignment of two depth sequences; Step 4, constructing an initial depth-domain model; and Step 5, performing model-driven depth-domain inversion to obtain the depth-domain seismic inversion results. This adaptive depth-matching depth-domain seismic inversion method updates the velocity model in each iteration of the inversion process, and the depth-time relationship is corrected in real time during the iteration, making the inversion results more accurate and reliable. It avoids the accuracy loss caused by well logging data during depth-time conversion and overcomes the error accumulation that occurs during the conversion process.

Description

Technical Field

[0001] This invention relates to the field of oilfield development technology, and in particular to an adaptive depth-matching depth-domain seismic inversion method. Background Technology

[0002] As oil and gas exploration and development deepens, seismic exploration is increasingly focused on more complex structures and reservoirs. Depth-domain seismic imaging results provide more reliable information on structural and lithological variations, and high-precision structural interpretation and reservoir prediction in the depth domain are key to seismic exploration.

[0003] For time-domain earthquakes, seismic convolution forward and inversion theories are typically used, where seismic records are the convolution of reflection coefficient sequences and seismic wavelets, to calibrate well-seismic synthetic seismic records and predict reservoirs through seismic inversion. However, in depth-domain earthquakes, the seismic wavefield does not satisfy the condition of "linear depth invariance," making it impossible to directly synthesize seismic records using convolution model theory. Without seismic forward modeling theory, it is impossible to perform well-seismic deep-domain synthetic record calibration and depth error correction, let alone achieve direct inversion of depth-domain earthquakes.

[0004] Existing technologies cannot achieve direct inversion of deep-domain seismic data. Furthermore, due to the difficulty in completely eliminating errors in deep-domain imaging velocity, the reservoir exhibits excessive errors at both logging and seismic depths, making it impossible to perform geological calibration between logging and seismic geophysical data. Most current depth-domain seismic inversion techniques are based on a "depth-time-depth" transformation, which involves first converting depth-domain seismic data to the time domain, performing inversion using mature techniques in the time domain, and then converting the inversion results back to the depth domain. This increases workload and error accumulation. Hu Zhongping and Lin Boxiang (2006) proposed a pseudo-depth-domain seismic inversion method based on the analysis of the "wavelet" in the depth domain. However, this method does not fundamentally change the underlying principles and still fails to demonstrate the advantages of depth-domain seismic inversion. Zhang Jing et al. (2010) attempted to use seismic multi-attribute transformation for depth-domain inversion. This method is not based on a convolution model and avoids the difficulty of extracting the wavelet in the depth domain. It relies on dense well network data, lacks clear geophysical significance and a unified standard, and the inversion effect cannot be evaluated. Fletcher (2012) et al. described a technique that introduces the point spread function (PSF). The basic idea of ​​this method takes into account the seismic imaging illumination problem in complex structures and salt body distribution areas, and adopts the point spread function (PSF, i.e., Hess) to address this issue. The ian operator replaces the seismic wavelet in traditional seismic inversion. The point spread function (PSF) can be used in the inversion process to remove virtual reflections, illumination compensation and attenuation compensation, but its computational efficiency is low.

[0005] Chinese patent application CN202311183468.4 discloses a method for deep-domain reservoir seismic inversion, belonging to the field of seismic inversion technology. This method includes acquiring deep-domain seismic data containing different seismic wave reflection characteristics; converting the deep-domain seismic data into time-domain seismic data using a low-pass filtering interpolation method; establishing an objective function based on the prior probability distribution of reflection coefficients, establishing constraint conditions based on prior wave impedance constraints and reflection coefficient sparsity constraints, solving the objective function based on the time-domain seismic data to obtain the reflection coefficients corresponding to the time-domain seismic data; and inverting the deep-domain seismic data based on the reflection coefficients to obtain the final seismic inversion result. This invention, by fusing deep-domain seismic data containing different seismic wave reflection characteristics, can effectively improve the resolution of deep-domain seismic inversion.

[0006] Chinese patent application CN201710421920.4 discloses a method and apparatus for deep-domain reservoir seismic inversion, comprising: obtaining deep-domain post-stack seismic data based on time-domain post-stack seismic data; calibrating well logging curves and deep-domain post-stack seismic data to obtain a deep-domain well logging curve model; interpreting the stratigraphic horizons of the deep-domain post-stack seismic data and gridding the interpretation results to obtain a deep-domain geological framework model; calculating the fractal parameters of the deep-domain well logging curve model and the deep-domain post-stack seismic data for each well, and converting the fractal parameters of each seismic trace into the fractal parameter volume of the well logging curve; and performing fractal random interpolation on the fractal parameter volume of the well logging curve based on the deep-domain well logging curve model and the deep-domain geological framework model to obtain the deep-domain reservoir seismic inversion result. This invention, based on a fractal algorithm, effectively ensures high resolution of the seismic inversion result while avoiding the acquisition of deep-domain wavelets.

[0007] Chinese patent application CN201610537549.3 discloses a post-stack seismic inversion method and apparatus. The method includes: obtaining synthetic seismic data based on an initial wave impedance model and actual seismic data; optimizing the initial wave impedance model based on the actual seismic data and synthetic seismic data to obtain a first optimal wave impedance; performing high-pass filtering on the first optimal wave impedance to obtain a high-frequency inversion result of the wave impedance; obtaining an actual seismic data envelopment (DEA) based on the actual seismic data and a synthetic seismic data envelopment (SGA) based on the synthetic seismic data; optimizing the initial wave impedance model based on the DEA and SGA to obtain a second optimal wave impedance; performing low-pass filtering on the second optimal wave impedance to obtain a low-frequency inversion result of the wave impedance; and obtaining a final wave impedance inversion result based on the high-frequency and low-frequency inversion results of the wave impedance. By implementing this invention, inversion accuracy is improved, enabling quantitative description in reservoir prediction.

[0008] Chinese patent application CN202110296191.0 discloses a method for creating depth-domain synthetic seismic records. This method addresses the problems of low wavelet extraction accuracy and low well-seismic matching in existing depth-domain synthetic records. The method includes the following steps: First, the depth-domain seismic imaging data traces passing through the well are converted to a pseudo-depth domain using the depth-domain imaging velocity at the corresponding location. Seismic wavelets are extracted using the linear time-invariant pseudo-depth. Next, well logging data is used to calculate the formation reflection coefficient, considering formation transmission attenuation effects. This coefficient is then converted to the pseudo-depth domain using a stable and smooth velocity, and convolution is performed with the pseudo-depth domain seismic wavelet. Finally, the coefficient is converted back to the depth domain using a stable and smooth velocity, thus completing the creation of the depth-domain synthetic seismic record. This method effectively solves the difference between well logging data velocity and surface seismic data velocity, achieves linear time-invariant extraction of deep-varying seismic wavelets, improves the correlation coefficient of depth-domain well-seismic calibration and the accuracy of depth-domain synthetic record creation, and reduces errors.

[0009] Currently, there is no complete and mature method or technology for directly using depth-domain seismic data for seismic reservoir inversion and prediction. Existing time-domain seismic inversion methods cannot be directly applied to depth-domain seismic data, and methods based on depth-domain seismic data also suffer from errors in well-seismic depth and the inability to compare and calibrate well-seismic geology. All of these existing technologies differ significantly from our present invention and fail to solve the technical problems we aim to address. Therefore, we have invented a novel adaptive depth-matching depth-domain seismic inversion method. Summary of the Invention

[0010] The purpose of this invention is to provide an adaptive depth-matching depth-domain seismic inversion method that updates the time-depth relationship in real time with each velocity update iteration.

[0011] The objective of this invention can be achieved through the following technical measures: an adaptive depth-matching depth-domain seismic inversion method, which includes:

[0012] Step 1: Preprocess the well logging data;

[0013] Step 2: Calculate the time vibration function to establish the relationship between depth domain wave impedance and seismic records;

[0014] Step 3: Perform depth-domain well-seismic matching correction to achieve optimal alignment of the two depth sequences;

[0015] Step 4: Construct the initial model for the depth domain;

[0016] Step 5: Perform model-driven depth domain inversion to obtain depth domain seismic inversion results.

[0017] The objective of this invention can also be achieved through the following technical measures:

[0018] In step 1, scale matching, or resampling, is performed on the well-seismic data to bring the frequency band of the well logging data closer to the seismic record. Then, the resampled well logging curves are selected, and the well logging parameters are appropriately depth-corrected in conjunction with the interpretation of the stratigraphic horizons in this block, so that the depth of parameter changes in the well logging data corresponds to the depth of the stratigraphic horizon interface.

[0019] In step 1, the well log is resampled according to the depth sampling rate of the seismic data. The seismic sampling rate is n meters. The average value of the layer velocity of all sampling points from 0 to n is taken as the first sample point output. Then, the average value of the layer velocity of all sampling points from n to 2n is taken as the second sample point output, and so on, until all sampling points are converted. A well log parameter curve with the same depth sampling as the seismic sampling is obtained.

[0020] In step 2, the time-depth conversion and depth domain superposition are integrated, and the relationship between depth domain wave impedance and seismic records is directly established using the vibration function w(t) as a medium:

[0021]

[0022] Where s is the seismic record in the depth domain, imp is the wave impedance in the depth domain, v is the velocity value in the depth domain, w is the time vibration function, Δh is the seismic sampling interval in the depth domain, and Δt is the sampling interval of the time vibration function.

[0023] In step 2, in order to better extract the time-vibration function, it needs to be transformed into a time-depth relationship, converting it to a depth-based transformation, as shown in the following formula:

[0024]

[0025] Establish the objective function Q for obtaining the vibration function w:

[0026]

[0027] Solving in the least squares sense yields the expression for the time-vibration function:

[0028]

[0029] Where r is the reflection coefficient:

[0030]

[0031] In step 3, the section required for inversion from the well logging data is interpreted in conjunction with the stratigraphic horizon of the corresponding region and compared with the seismic section of the same length. The similarity is calculated using the dynamic depth adjustment method, and the slip depth at the best similarity is selected as the depth error, which is also used as the final depth calibration result of the well and seismic data, thus completing the optimal alignment of the two depth sequences.

[0032] In step 3, the similarity calculation process is as follows: Each data point in the two recording segments is compared using a distance metric; a distance matrix is ​​constructed to represent the distance between each data point in the two depth sequences; a shortest path matrix is ​​constructed using the DP algorithm, and the result is obtained by adding the elements at corresponding positions on the distance matrix; the final similarity can be represented by dividing the shortest path length by the total length of the two depth sequences; after obtaining the deviation correction amount of each small layer of the well-side seismic trace, a depth error volume is established corresponding to the three-dimensional depth data volume to correct the depth domain seismic data and seismic imaging velocity, ensuring that the errors of logging, seismic, and seismic imaging velocity are minimized.

[0033] In step 4, the first step is to interpret the stratigraphic positions in detail based on the seismic data and establish a geological framework. The target stratigraphic positions are then traced and interpreted according to the inversion task to ensure that they are close to the target reservoir and can reflect the spatial variation relationship of the target reservoir. Then, under its control, an appropriate interpolation method is selected to perform interpolation and extrapolation of well logging data along the layers to establish a detailed initial model.

[0034] In step 4, the interpolation methods used to establish the initial model include: Gaussian interpolation algorithm, spline curve interpolation algorithm, Lagrange interpolation algorithm, which has gradually evolved into inverse distance weighted interpolation method, natural neighborhood interpolation method, Kriging interpolation method, and Shebeder interpolation method.

[0035] In step 4, the main stratigraphic layers in the target area are selected as control layers. Between each layer, interpolation is performed using well logging data based on the actual thickness of the formation to obtain the initial velocity model and the initial wave impedance model.

[0036] In step 5, the forward modeling results of the inversion constraint model parameters are added to the objective function using the inversion initial impedance model:

[0037]

[0038] Where F is the root mean square error between the observed data d and the predicted data G(m), the above equation is a typical least squares problem; since the observed data d and the model parameters m to be inverted are often nonlinear in actual data, when solving the above equation linearly, it is necessary to modify the model parameters by adding prior information, i.e. well logging constraints, to constrain the size of its solution space, and obtain the optimal solution by calculating the maximum probability distribution, so as to obtain a result that conforms to the actual situation and is relatively stable and accurate; since the above inversion depth domain model parameters have a prior probability distribution, it can be derived from equation (4) according to Bayes' theorem:

[0039]

[0040] Where A is the Jacobian matrix, I is the identity matrix, and the iterative formula is m. k+1 =m k +Δm, d represents the actual earthquake data, G(m) 0 () represents the depth-domain synthetic seismic record formed by the parametric model; in the above equations The term is equivalent to the damping factor in the traditional damped least squares inversion method, but the difference is that the traditional damped least squares inversion method often selects a white noise factor empirically, while here... It has a clear physical meaning, respects objective facts, and increases the accuracy of the inversion; among them, σ m The value σ represents the size of the solution space with the initial model as the expectation. m The larger σ is, the larger the solution space. m The smaller the value, the smaller the solution space.

[0041] In step 5, after each model update, the oscillation function is reconstructed based on the rock physics relationship and then used in the next model optimization solution. Therefore, the oscillation function will not only change with the spatial velocity, but also with the update of the iteration speed.

[0042] The objective of this invention can also be achieved through the following technical measures: an adaptive depth-matching depth domain seismic inversion system, which uses an adaptive depth-matching depth domain seismic inversion method to perform depth domain seismic inversion.

[0043] The adaptive depth-matching depth-domain seismic inversion method in this invention performs depth-domain synthetic record creation and calibration, automatically picks up well points and corrects seismic depth errors, and establishes a connection between the time vibration function and the depth-domain wave impedance to achieve direct inversion of depth-domain seismic data. This avoids error accumulation caused by repeated "depth-time-depth" conversion and loss of thin interlayer information caused by well data resampling. With each velocity update iteration, the new time-depth relationship is updated in real time, and the amplitude and waveform of the depth-domain seismic data are automatically corrected to achieve optimal matching between the depth-domain seismic data and the synthetic record. Attached Figure Description

[0044] Figure 1 A flowchart of a specific embodiment of the adaptive depth matching depth domain seismic inversion method of the present invention;

[0045] Figure 2 This is a schematic diagram comparing well-seismic depth matching before and after in a specific embodiment of the present invention;

[0046] Figure 3 This is a schematic diagram of the depth domain seismic depth error before correction in a specific embodiment of the present invention;

[0047] Figure 4 This is a schematic diagram of the depth domain seismic depth error correction in a specific embodiment of the present invention;

[0048] Figure 5 This is a schematic diagram of the depth domain model for logging impedance interpolation in a specific embodiment of the present invention;

[0049] Figure 6 This is a schematic diagram of the depth domain wave impedance inversion results in a specific embodiment of the present invention;

[0050] Figure 7 This is a schematic diagram illustrating the deep-domain seismic tectonic interpretation in a specific embodiment of the present invention;

[0051] Figure 8 This is a schematic diagram of the wave impedance depth domain model constructed based on the imaging velocity-rock physics relationship in a specific embodiment of the present invention;

[0052] Figure 9 This is a schematic diagram of the depth domain impedance inversion results in a specific embodiment of the present invention. Detailed Implementation

[0053] It should be noted that the following detailed descriptions are exemplary and intended to provide further illustration of the invention. Unless otherwise specified, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0054] It should be noted that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the exemplary embodiments of the present invention. As used herein, the singular form is intended to include the plural form as well, unless the context clearly indicates otherwise. Furthermore, it should be understood that when the terms "comprising" and / or "including" are used in this specification, they indicate the presence of features, steps, operations, and / or combinations thereof.

[0055] This invention integrates the depth-time conversion relationship into the forward and inverse processes, directly realizing synthetic recording and velocity or wave impedance inversion in the depth domain; during the inversion process, after each velocity iteration, the depth-time relationship is updated synchronously, and the amplitude and waveform of the reflected wave at each point are automatically and double-corrected.

[0056] like Figure 1 As shown, Figure 1 This is a flowchart of the adaptive depth-matching depth-domain seismic inversion method of the present invention. The adaptive depth-matching depth-domain seismic inversion method includes:

[0057] Step 1: Preprocess the well logging data. Scale matching, or resampling, is performed on the well-seismic data to bring the frequency band of the well logging data closer to the seismic record. Then, the resampled well logging curves are selected, and combined with the interpretation of the stratigraphic horizons in this block, appropriate depth correction is performed on the well logging parameters so that the depth location of parameter changes in the well logging data corresponds to the interface depth location of the stratigraphic horizons.

[0058] Step 2: Determining the time-vibration function. The time-depth conversion and depth-domain superposition steps are integrated, using the vibration function w(t) as a medium to directly establish the relationship between depth-domain wave impedance and seismic records.

[0059] Step 3, Depth Domain Well-Seismic Matching Correction. The section required for inversion from the well logging data is interpreted in conjunction with the corresponding stratigraphic horizon and compared with seismic segments of the same length. A dynamic depth straightening method is used to calculate the similarity, and the slip depth at the optimal similarity is selected as the depth error, which is also used as the final well-seismic depth calibration result. The optimal alignment of the two depth sequences is automatically completed.

[0060] Step 4: Construct the initial depth domain model. First, based on the detailed interpretation of seismic data, establish a geological framework. Then, according to the inversion task, trace and interpret the target layers to ensure they approximate the target reservoir and reflect its spatial variation. Next, under this control, select an appropriate interpolation method and perform interpolation and extrapolation of well logging data along the layers to establish a detailed initial model.

[0061] Step 5, Model-driven depth-domain inversion. Using the inversion initial impedance model, the forward modeling results of the inversion constraint model parameters are added to the objective function.

[0062] Compared with existing deep domain inversion techniques, this invention has two main advantages:

[0063] (1) In each iteration of the inversion process, the velocity model is updated, which also means that the deep time relation is updated accordingly. The deep time relation is corrected in real time during the iteration process, making the inversion results more accurate and reliable. This is the core point that distinguishes this invention from other methods.

[0064] (2) It avoids the loss of accuracy caused by the deep-time conversion of logging data; compared with the traditional "deep-time-deep" conversion in the industry, it overcomes the accumulation of errors that occur during the conversion process.

[0065] The following are specific embodiments of the application of the present invention.

[0066] Example 1

[0067] like Figure 1 As shown, Figure 1 This is a flowchart of the adaptive depth-matching depth-domain seismic inversion method of the present invention. The adaptive depth-matching depth-domain seismic inversion method includes:

[0068] Step 1: Preprocessing of well logging data

[0069] In current seismic exploration, the subsurface formation parameters obtained from well logging sampling are usually expressed in depth units. However, the depth sampling rate is much lower than the seismic depth sampling rate. Therefore, it is necessary to first perform scale matching on the well-seismic data. This typically involves sampling and processing the well logs to bring their frequency bands closer to the seismic records. The specific method is as follows:

[0070] Based on the depth sampling rate of the seismic data, the well logs are resampled. Assuming the seismic sampling rate is n meters, the average layer velocity of all sampling points from 0 to n is taken as the first sample point output. Then, the average layer velocity of all sampling points from n to 2n is taken as the second sample point output, and so on, until all sampling points are converted. This yields a well log parameter curve with the same depth sampling as the seismic sampling.

[0071] Subsequently, the resampled logging curves were selected, and combined with the interpretation of the stratigraphic positions in this block, the logging parameters were appropriately depth-corrected so that the depth of parameter changes in the logging data corresponded to the interface depth of the stratigraphic positions.

[0072] Step 2: Determining the time-vibration function

[0073] The industry-standard "depth-time-depth" method is based on a depth-domain model and a time-domain vibration function. First, the reflection coefficient in the depth domain is converted to the time domain using a depth-time transformation. This time-domain reflection coefficient is then convolved with the time-domain vibration function to obtain the seismic record. Finally, the velocity field is transformed back to the depth domain. By integrating the above process—that is, combining the time-depth transformation and depth-domain superposition—the relationship between depth-domain wave impedance and the seismic record is directly established using the vibration function w(t) as a medium.

[0074]

[0075] Where s is the seismic record in the depth domain, imp is the wave impedance in the depth domain, v is the velocity value in the depth domain, and w is the time vibration function.

[0076] To better extract the time-vibration function, it needs to be transformed into a time-depth relationship, converting it to a time-based depth conversion, as shown in the following formula:

[0077]

[0078] Establish the objective function for obtaining the vibration function w:

[0079]

[0080] Solving in the least squares sense yields the expression for the time-vibration function:

[0081]

[0082] Where r is the reflection coefficient:

[0083]

[0084] Step 3: Depth-domain well-seismic matching correction

[0085] In the subsequent establishment of the initial inversion model and the depth domain inversion process, well logging and seismic data with good matching relationships are required. The quality of the matching results directly affects the quality of subsequent work.

[0086] The sections required for inversion from well logging data are interpreted in conjunction with the corresponding stratigraphic horizons of the region and compared with seismic segments of the same length. A dynamic depth rounding method is used to calculate the similarity, and the slip depth at the optimal similarity is selected as the depth error, which is also used as the final well-seismic depth calibration result. The optimal alignment of the two depth sequences is automatically completed. The similarity calculation process is as follows:

[0087] Compare each data point in the two recording segments using any distance metric, such as Euclidean distance or Manhattan distance; construct a distance matrix to represent the distance between each data point in the two depth sequences; use the dynamic programming (DP) algorithm to construct a shortest path matrix, which is obtained by adding the elements at corresponding positions in the distance matrix; the final similarity can be represented by dividing the shortest path length by the total length of the two depth sequences.

[0088] After obtaining the deviation correction values ​​for each sub-layer of the well-side seismic trace, a depth error volume is established corresponding to the three-dimensional depth data volume. This volume is used to correct the depth-domain seismic data and seismic imaging velocity, ensuring that the errors in logging, seismic data, and seismic imaging velocity are minimized. Theoretically, if the depth migration is accurate enough, the resulting depth-domain seismic data should be consistent with the actual subsurface structure in terms of tectonic morphology. If the well logging data is also accurate enough, the two can be perfectly matched.

[0089] Step 4: Construct the initial model in the depth domain

[0090] Due to bandwidth limitations, seismic data lacks high- and low-frequency information. To obtain broadband inversion results during impedance retrieval, an initial model is often constructed to incorporate this missing high- and low-frequency information. The following is the calculation method for constructing the initial model:

[0091] First, the stratigraphic levels must be meticulously interpreted based on seismic data to establish a geological framework. Then, the target stratigraphic levels are traced and interpreted according to the inversion task to ensure proximity to the target reservoir and to reflect its spatial variation. Under this control, an appropriate interpolation method is selected, and well logging data is interpolated and extrapolated along the stratigraphic layers to establish a refined initial model. Interpolation methods for establishing the initial model include Gaussian interpolation, spline curve interpolation, and Lagrange interpolation, which have evolved to inverse distance weighted interpolation, natural neighborhood interpolation, Kriging interpolation, and Shebeder interpolation.

[0092] In the target area, the main stratigraphic layers are selected as control layers. Between each layer, interpolation is performed using well logging data based on the actual thickness of the strata to obtain the initial velocity model and the initial wave impedance model.

[0093] Step 5: Model-driven deep domain inversion

[0094] Using the inverted initial impedance model, the forward modeling results of the inverted constraint model parameters are added to the objective function:

[0095]

[0096] Where F is the root mean square error between the observed data d and the predicted data G(m), the above equation is a typical least squares problem. Since the relationship between the observed data d and the model parameters m to be inverted is often nonlinear in actual data, when solving the above equation linearly, it is necessary to modify the model parameters by adding prior information, i.e., well logging constraints, to constrain the size of its solution space, and obtain the optimal solution by calculating the maximum probability distribution, so as to obtain a result that conforms to the actual situation and is relatively stable and accurate. Since the above inversion depth domain model parameters have a prior probability distribution, it can be derived from equation (4) according to Bayes' theorem:

[0097]

[0098] Where A is the Jacobian matrix, I is the identity matrix, and the iterative formula is m. k+1 =m k +Δm, d represents the actual earthquake data, G(m) 0 () represents the depth-domain synthetic seismic record formed by the parametric model. The equations above... The term is equivalent to the damping factor in the traditional damped least squares inversion method, but the difference is that the traditional damped least squares inversion method often selects a white noise factor empirically, while here... It has a clear physical meaning, respects objective facts, and increases the accuracy of the inversion. Among them, σ m The value σ represents the size of the solution space with the initial model as the expectation. m The larger σ is, the larger the solution space. m The smaller the value, the smaller the solution space.

[0099] After each model update, the oscillation function is reconstructed based on the rock physics relationship and then used in the next model optimization solution. Therefore, the oscillation function changes not only with the spatial velocity but also with the update of the iteration speed, resulting in higher inversion accuracy.

[0100] Example 2

[0101] In a specific embodiment of the present invention, the adaptive depth-matching depth-domain seismic inversion method includes the following steps:

[0102] (1) After preprocessing the well logging data, the time vibration function is obtained by combining the well-side depth domain seismic record and the well logging depth domain data. This is then convolved to obtain the synthetic well logging seismic record, which is compared with a seismic segment of the same length from the well-side seismic trace and automatically matched and corrected. For example... Figure 2As shown, after well-seismic depth matching correction, the position of the seismic wave peak corresponds to the interface depth position of the well logging. With the well logging data as a constraint, it can be adapted to the well logging synthetic seismic record to the greatest extent, which greatly improves the similarity coefficient between the well-side seismic record and the well logging synthetic record, and helps the subsequent inversion steps.

[0103] After obtaining the deviation correction values ​​for each sub-layer of the seismic trace near the well, a depth error volume is established corresponding to the three-dimensional depth data volume. This volume is then used to correct the depth-domain seismic data and seismic imaging velocity. Figure 3 The earthquake before correction. Figure 4 The corrected earthquake profile shows that the structural morphology before and after correction is basically the same, and the depth error of the earthquake geological reflection interface has been corrected and removed.

[0104] (2) The distance-weighted interpolation method is used to perform interpolation calculations on the initial wave impedance model. With geological structural interpretation as a constraint, inter-layer interpolation is performed according to the specific layer thickness of different seismic traces, which serves as the initial model for subsequent depth-domain wave impedance inversion. Figure 5 The initial wave impedance model obtained after the corresponding steps is divided into multiple fault blocks by the fault.

[0105] (3) Combining the depth domain seismic profile and its structural interpretation, and using well logging data parameters as constraints, the parameters of the subsurface medium are directly inverted in the depth domain. Figure 6 This is a schematic diagram of the depth domain impedance inversion results. The inversion results are consistent with the well logging data, have high resolution, and reveal the high impedance anomaly zone in the area.

[0106] Example 3

[0107] Figure 7 This is another example of a three-dimensional depth domain seismic inversion in the work area. Due to the large area and complex structural changes in the work area, only the top and bottom control layers of the target layer were explained.

[0108] Figure 8 To construct the inversion constraint model using imaging velocity, due to the imprecise structural interpretation, the structural model establishment and constraint accuracy were low. Therefore, in step four, well logging interpolation modeling was not used. Instead, the inversion wave impedance constraint model was constructed based on the statistical relationship between velocity and impedance using depth domain imaging velocity. The model was then subjected to depth error correction in step three to ensure that the depth of the seismic and constraint models were consistent.

[0109] Figure 9 To utilize the corrected depth-domain seismic and wave impedance constrained model, step five is used to invert the depth-domain inverted wave impedance, which shows that the high-impedance sandstone and conglomerate in the steep slope zone is clearly distributed.

[0110] Finally, it should be noted that the above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

[0111] Except for the technical features described in the specification, all other technologies are known to those skilled in the art.

Claims

1. An adaptive depth-matching depth-domain seismic inversion method, characterized in that, The adaptive depth-matching depth-domain seismic inversion method includes: Step 1: Preprocess the well logging data; Step 2: Calculate the time vibration function to establish the relationship between depth domain wave impedance and seismic records; Step 3: Perform depth-domain well-seismic matching correction to achieve optimal alignment of the two depth sequences; Step 4: Construct the initial model for the depth domain; Step 5: Perform model-driven depth domain inversion to obtain depth domain seismic inversion results.

2. The depth-domain seismic inversion method with adaptive depth matching according to claim 1, characterized in that, In step 1, scale matching, or resampling, is performed on the well-seismic data to bring the frequency band of the well logging data closer to the seismic record. Then, the resampled well logging curves are selected, and the well logging parameters are appropriately depth-corrected in conjunction with the interpretation of the stratigraphic horizons in this block, so that the depth of parameter changes in the well logging data corresponds to the depth of the stratigraphic horizon interface.

3. The depth-domain seismic inversion method with adaptive depth matching according to claim 2, characterized in that, In step 1, the well log is resampled according to the depth sampling rate of the seismic data. The seismic sampling rate is n meters. The average value of the layer velocity of all sampling points from 0 to n is taken as the first sample point output. Then, the average value of the layer velocity of all sampling points from n to 2n is taken as the second sample point output, and so on, until all sampling points are converted. A well log parameter curve with the same depth sampling as the seismic sampling is obtained.

4. The depth-domain seismic inversion method with adaptive depth matching according to claim 1, characterized in that, In step 2, the time-depth conversion and depth domain superposition are integrated, and the relationship between depth domain wave impedance and seismic records is directly established using the vibration function w(t) as a medium: Where s is the seismic record in the depth domain, imp is the wave impedance in the depth domain, v is the velocity value in the depth domain, w is the time vibration function, Δh is the seismic sampling interval in the depth domain, and Δt is the sampling interval of the time vibration function.

5. The depth-domain seismic inversion method with adaptive depth matching according to claim 4, characterized in that, In step 2, in order to better extract the time-vibration function, it needs to be transformed into a time-depth relationship, converting it to a depth-based transformation, as shown in the following formula: Establish the objective function Q for obtaining the vibration function w: Solving in the least squares sense yields the expression for the time-vibration function: Where r is the reflection coefficient:

6. The depth-domain seismic inversion method with adaptive depth matching according to claim 1, characterized in that, In step 3, the section required for inversion from the well logging data is interpreted in conjunction with the stratigraphic horizon of the corresponding region and compared with the seismic section of the same length. The similarity is calculated using the dynamic depth adjustment method, and the slip depth at the best similarity is selected as the depth error, which is also used as the final depth calibration result of the well and seismic data, thus completing the optimal alignment of the two depth sequences.

7. The depth-domain seismic inversion method with adaptive depth matching according to claim 6, characterized in that, In step 3, the similarity calculation process is as follows: Each data point in the two recording segments is compared using a distance metric; a distance matrix is ​​constructed to represent the distance between each data point in the two depth sequences; a shortest path matrix is ​​constructed using the DP algorithm, and the result is obtained by adding the elements at corresponding positions on the distance matrix; the final similarity can be represented by dividing the shortest path length by the total length of the two depth sequences; after obtaining the deviation correction amount of each small layer of the well-side seismic trace, a depth error volume is established corresponding to the three-dimensional depth data volume to correct the depth domain seismic data and seismic imaging velocity, ensuring that the errors of logging, seismic, and seismic imaging velocity are minimized.

8. The depth-domain seismic inversion method with adaptive depth matching according to claim 1, characterized in that, In step 4, the first step is to interpret the stratigraphic positions in detail based on the seismic data and establish a geological framework. The target stratigraphic positions are then traced and interpreted according to the inversion task to ensure that they are close to the target reservoir and can reflect the spatial variation relationship of the target reservoir. Then, under its control, an appropriate interpolation method is selected to perform interpolation and extrapolation of well logging data along the layers to establish a detailed initial model.

9. The depth-domain seismic inversion method with adaptive depth matching according to claim 8, characterized in that, In step 4, the interpolation methods used to establish the initial model include: Gaussian interpolation algorithm, spline curve interpolation algorithm, Lagrange interpolation algorithm, which has gradually evolved into inverse distance weighted interpolation method, natural neighborhood interpolation method, Kriging interpolation method, and Shebeder interpolation method.

10. The depth-domain seismic inversion method with adaptive depth matching according to claim 8, characterized in that, In step 4, the main stratigraphic layers in the target area are selected as control layers. Between each layer, interpolation is performed using well logging data based on the actual thickness of the formation to obtain the initial velocity model and the initial wave impedance model.

11. The depth-domain seismic inversion method with adaptive depth matching according to claim 1, characterized in that, In step 5, the forward modeling results of the inversion constraint model parameters are added to the objective function using the inversion initial impedance model: Where F is the root mean square error between the observed data d and the predicted data G(m), the above equation is a typical least squares problem; since the observed data d and the model parameters m to be inverted are often nonlinear in actual data, when solving the above equation linearly, it is necessary to modify the model parameters by adding prior information, i.e. well logging constraints, to constrain the size of its solution space, and obtain the optimal solution by calculating the maximum probability distribution, so as to obtain a result that conforms to the actual situation and is relatively stable and accurate; since the above inversion depth domain model parameters have a prior probability distribution, it can be derived from equation (4) according to Bayes' theorem: Where A is the Jacobian matrix, I is the identity matrix, and the iterative formula is m. k+1 =m k +Δm, d represents the actual earthquake data, G(m) 0 () represents the depth-domain synthetic seismic record formed by the parametric model; in the above equations The term is equivalent to the damping factor in the traditional damped least squares inversion method, but the difference is that the traditional damped least squares inversion method often selects a white noise factor empirically, while here... It has a clear physical meaning, respects objective facts, and increases the accuracy of the inversion; among them, σ m The value σ represents the size of the solution space with the initial model as the expectation. m The larger σ is, the larger the solution space. m The smaller the value, the smaller the solution space.

12. The depth-domain seismic inversion method with adaptive depth matching according to claim 11, characterized in that, In step 5, after each model update, the oscillation function is reconstructed based on the rock physics relationship and then used in the next model optimization solution. Therefore, the oscillation function will not only change with the spatial velocity, but also with the update of the iteration speed.

13. An adaptive depth-matching depth-domain seismic inversion system, characterized in that, The adaptive depth-matching depth-domain seismic inversion system uses the adaptive depth-matching depth-domain seismic inversion method described in any one of claims 1-12 to perform depth-domain seismic inversion.

Images