A High-Dimensional Deconvolution Processing Method for Seismic Images Based on Gradual Stabilization Factor

Abstract

This invention provides a high-dimensional deconvolution processing method for seismic images based on a gradient stabilization factor, relating to the field of petroleum seismic exploration. Within the classical least-squares inversion imaging framework, it utilizes a locally supported point spread function (PSF) to approximate the Hessian matrix, and then employs a high-dimensional deconvolution method based on the gradient stabilization factor and PSF for seismic image processing. The constructed gradient stabilization factor prevents computational instability, making the calculation process more robust and enabling better illumination compensation and correction for depth-dependent uneven illumination, effectively improving the resolution of seismic images. The beneficial effects of this invention are: it can more flexibly adapt to changes in subsurface structure, improve the interpretation accuracy of seismic data, provide more accurate and clear subsurface structure information for seismic exploration, and overcome some limitations of traditional methods, thereby improving image quality and interpretability.

Description

Technical Field

[0001] This invention relates to the field of petroleum seismic exploration, and in particular to a high-dimensional deconvolution processing method for seismic images based on a gradual stability factor. Background Technology

[0002] High-resolution and high-fidelity seismic inversion imaging is a primary goal of oil and gas exploration, aiming to reveal subsurface structures and geological features in the most accurate way. High-precision imaging of subsurface structures characterized by reflection coefficients and quantitative estimation of subsurface parameters are crucial aspects of inversion imaging; however, the resolution of the imaging results is typically affected by multiple factors. Traditional single-channel deconvolution methods based on one-dimensional convolution models cannot simultaneously achieve both lateral and longitudinal resolution and are insufficient for fitting complex geological structures. Furthermore, uneven illumination can lead to amplitude imbalances, making traditional deconvolution methods unable to meet the requirements of high-fidelity and high-resolution processing. Therefore, to overcome these problems and comprehensively consider multiple factors affecting imaging resolution, it is essential to extend the one-dimensional deconvolution model to a high-dimensional deconvolution model based on migration imaging for further high-resolution processing.

[0003] In the framework of inversion imaging, the Hessian matrix of a single-reflection wave imaging system is crucial in affecting the resolution of the imaging results, causing blurred profiles and highly non-uniform amplitudes. Least-squares migration (LSM) is an extension of classical migration imaging within the inversion imaging framework. Its purpose is to remove this blurring effect of the Hessian matrix, thereby achieving amplitude uniformity and obtaining higher resolution and more accurate subsurface imaging results, thus improving image resolution. In 2021, Liu et al., in their paper "An Effective Acoustic Impedance Imaging Based on a Broadband Gaussian Beam Migration" published in *Energies*, approximate the Hessian matrix using a locally supported point spread function (PSF) and obtain the broadband reflectance coefficient by performing a high-dimensional deconvolution on the image and PSF. In the spatial domain, the broadband reflectance coefficient needs to be obtained by deconvolving the image and PSF, which is equivalent to directly dividing the image and PSF in the wavenumber domain. Therefore, the broadband reflectance coefficient can be obtained by directly dividing the image by the PSF in the wavenumber domain. This correction method can achieve relatively stable illumination equalization and improve resolution to some extent. However, this method of direct division in the wavenumber domain also has its drawbacks, namely, the value of the denominator PSF cannot be zero or close to zero.

[0004] However, in real-world data, due to factors such as system noise and sampling effects, the PSF value may be close to or even zero in certain regions. Therefore, a stabilization factor needs to be added to the denominator so that the broadband reflectance coefficient can be obtained by directly dividing the image by the PSF in the wavenumber domain for any region. However, considering that the illumination effects at different depths are not uniform, using a fixed stabilization factor cannot effectively compensate for uneven illumination that varies with depth. Summary of the Invention

[0005] To address the aforementioned problems, this invention provides a high-dimensional deconvolution processing method for seismic images based on a gradual stability factor, mainly comprising:

[0006] S1: Data preparation: Based on the background velocity and reflection coefficient of the horizontal thin-layer model, a conventional migration imaging profile is obtained; a scattering point model is generated by inserting scattering points into the background velocity, and the scattering point model is subjected to Born forward modeling to generate pre-stack scattering data. The pre-stack scattering data is then processed using a conventional reverse time migration method to generate the PSF corresponding to all underground scattering points for correction.

[0007] S2: Input data: Store the conventional offset imaging profile and PSF blocks as I and P files;

[0008] S3: Local Fourier Transform: Performs a Fourier transform on each local block data in the I and P files, converting it from the spatial domain to the wavenumber domain;

[0009] S4: Constructing the gradual stabilization factor: Perform two-dimensional spectral analysis on the PSF data after Fourier transform to construct the gradual stabilization factor ε(z);

[0010] S5: Image Correction: In the wavenumber domain, a block correction method is used. A high-dimensional deconvolution method based on the gradient stabilization factor and PSF is employed to perform wavenumber domain correction on each image block. The deconvolution calculation formula is as follows: in, This represents the profile corrected using the gradual stability factor in the wavenumber domain. This represents a conventional migration imaging profile in the wavenumber domain. Let ε(z) represent the PSF in the wavenumber domain, ε(z) be the gradual stabilization factor, x be the position in the imaging space, k be the local wavenumber, and z be the depth.

[0011] S6: Local Inverse Fourier Transform: The corrected result... Perform an inverse Fourier transform to convert the wavenumber domain data into spatial domain data and store it in an Ic file;

[0012] S7: Output results: Output the Ic file image in the spatial domain after gradual stabilization factor correction and local inverse Fourier transform.

[0013] Further, in step S1, the horizontal thin-layer model has a size of 6000m×3800m, and the spatial sampling interval in both the horizontal and vertical directions is 10m; the scattering point model has a sampling interval of 270m in both the horizontal and vertical directions.

[0014] Furthermore, in step S2, in order to obtain better correction results, a Gaussian window function is used to perform windowing operations on each local block.

[0015] Furthermore, in step S2, the conventional offset imaging profile and PSF are stored in blocks using a column-first storage method.

[0016] Furthermore, in step S4, the gradient stabilization factor ε(z) is 5% of the maximum PSF value at the same depth z.

[0017] Furthermore, in step S6, for After performing the inverse Fourier transform, the data is stored column-wise in the Ic file.

[0018] A storage device that stores instructions and data for implementing a high-dimensional deconvolution processing method for seismic images based on a gradual stability factor.

[0019] A high-dimensional deconvolution processing device for seismic images based on a gradual stabilization factor includes: a processor and a storage device; the processor loads and executes instructions and data in the storage device to implement a high-dimensional deconvolution processing method for seismic images based on a gradual stabilization factor.

[0020] The beneficial effects of the technical solution provided by this invention are as follows: Within the classical least-squares inversion imaging framework, this invention utilizes a locally supported point spread function (PSF) to approximate the Hessian matrix, and performs seismic image processing based on a gradient stabilization factor and a high-dimensional deconvolution method using PSF. The constructed gradient stabilization factor not only prevents computational instability, making the computation process more robust, but also better compensates for depth-dependent uneven illumination, effectively improving the resolution of seismic images. This method allows for more flexible adaptation to changes in subsurface structures, thereby improving the interpretation accuracy of seismic data. Consequently, seismic exploration provides more accurate and clearer subsurface structure information, overcomes some limitations of traditional methods, and improves image quality and interpretability. Attached Figure Description

[0021] The present invention will be further described below with reference to the accompanying drawings and embodiments. In the accompanying drawings:

[0022] Figure 1 This is a flowchart of a high-dimensional deconvolution processing method for seismic images based on a gradual stabilization factor, as described in an embodiment of the present invention.

[0023] Figure 2 This is a schematic diagram of the background velocity of the model used in this embodiment of the invention.

[0024] Figure 3 This is a schematic diagram of the reflection coefficient of the model used in this embodiment of the invention.

[0025] Figure 4 This is a schematic diagram of a conventional offset imaging profile in an embodiment of the present invention.

[0026] Figure 5 This is a schematic diagram of the PSF in the spatial domain in an embodiment of the present invention.

[0027] Figure 6 This is a schematic diagram of the wavenumber spectrum of the PSF in an embodiment of the present invention.

[0028] Figure 7 This is a schematic diagram illustrating the spectrum variation of PSF under the same horizontal coordinate in an embodiment of the present invention.

[0029] Figure 8 This is a schematic diagram of the change in the stability factor in an embodiment of the present invention.

[0030] Figure 9 This is a schematic diagram of the blurring effect of PSF on images under different illumination angles in an embodiment of the present invention.

[0031] Figure 10 This is a schematic diagram of the block correction approach in an embodiment of the present invention.

[0032] Figure 11 This is a schematic diagram of the broadband reflection coefficient obtained after the conventional reverse time offset profile is corrected by the gradual stabilization factor in an embodiment of the present invention.

[0033] Figure 12 This is a schematic diagram of comparative analysis in an embodiment of the present invention.

[0034] Figure 13 This is a schematic diagram of the hardware device working in an embodiment of the present invention. Detailed Implementation

[0035] To provide a clearer understanding of the technical features, objectives, and effects of the present invention, specific embodiments of the present invention will now be described in detail with reference to the accompanying drawings.

[0036] Please refer to Figure 1 , Figure 1This is a flowchart of a high-dimensional deconvolution processing method for seismic images based on a gradual stability factor, as described in an embodiment of the present invention, specifically including:

[0037] S1. Data Preparation

[0038] This embodiment uses a horizontal thin-layer model WX, with a model size of 6000m × 3800m, and a spatial sampling interval of 10m in both the horizontal and vertical directions. Background velocity is simulated by Gaussian smoothing of the true velocity. For example... Figure 2 The background velocity of the model shown is Figure 3 The reflection coefficients shown were obtained using the Born forward modeling method from the observed data, i.e., as shown below. Figure 4 The image shows a conventional offset imaging profile. A total of 141 shots are used, with shot points evenly distributed across the uppermost layer of the model from 0 to 6000m, and the interval between adjacent shots is 50m. For each shot, 151 fixed detectors are evenly distributed across the uppermost layer of the model from 0 to 6000m.

[0039] The scattering point model in this embodiment is obtained by inserting 308 scattering points into the background velocity, with a sampling interval of 270m in both the horizontal and vertical directions. Pre-stack scattering data is obtained by using Born forward modeling on the scattering point model. The obtained pre-stack scattering data is then processed using a conventional inverse time migration method to obtain the point spread function (PSF) corresponding to all subsurface scattering points. The PSF in the spatial domain is as follows: Figure 5 .

[0040] S2, Input Data

[0041] The obtained conventional offset imaging profiles and PSFs are divided into I and P files and stored in a column-major order. To achieve better processing results, a Gaussian window function is used to perform windowing operations on each local block in this embodiment.

[0042] S3, Local Fourier Transform

[0043] Performing a Fourier transform on each local block transforms it from the spatial domain to the wavenumber domain. The wavenumber spectrum of the PSF is as follows: Figure 6 As shown;

[0044] S4. Constructing the gradual stabilization factor:

[0045] Two-dimensional spectral analysis of the PSF after Fourier transform reveals that illumination inhomogeneity is more significantly affected by depth, with shallow and mid-depth layers exhibiting better seismic illumination and coverage. The two dimensions of the seismic image are depth z and distance x. Therefore, the independent variable of the gradient stabilization factor is depth z, and the gradient stabilization factor is ε(z), with a magnitude of 5% of the maximum PSF value at the same depth z. That is, the gradient stabilization factor ε(z) is 5% of the maximum PSF value at different distances within the same depth of the seismic image. The spectral variation of PSF under the same horizontal coordinate is as follows: Figure 7 The gradual change in the stabilizing factor is as follows Figure 8 , where the horizontal axis z is the depth, and the vertical axis ε(z) is the percentage of the value among all values ​​of ε(z) to the maximum value.

[0046] S5, Image Correction

[0047] In the wavenumber domain, conventional imaging results can be represented by the dot product of the reflectance coefficient and the PSF:

[0048] I(x,k)=PSF(x,k)R(x,k)

[0049] Where I(x,k) represents the imaging result, PSF(x,k) represents the point spread function (PSF), and R(x,k) represents the reflectance coefficient.

[0050] PSF is a comprehensive characterization of all factors affecting imaging resolution. Figure 9 This demonstrates the blurring effect of a PSF on the same image under different lighting angles. A well-lit PSF produces relatively little blurring.

[0051] Since the correction process needs to be performed in blocks, the corrected result in the wavenumber domain is obtained by iterating through each image block and performing a division operation with its corresponding PSF. The formula is then used for each image and its corresponding PSF. Perform corrections.

[0052] Figure 9 The image shows the blurring effect of the PSF (Power Seed) on the same image under different lighting angles: (a) blurring effect of the PSF with right-side lighting, (b) blurring effect of the PSF with left-side lighting, and (c) blurring effect of the PSF with good lighting; the block correction approach is as follows: Figure 10 ;

[0053] S6, Local Inverse Fourier Transform

[0054] For the obtained Perform an inverse Fourier transform to convert the wavenumber domain data to the spatial domain, and store it in the Ic file in column-major order.

[0055] S7, Output Results

[0056] The output is the Ic file image in the spatial domain after correction and inverse Fourier transform.

[0057] The broadband reflection coefficients obtained after correction by the gradual stabilization factor for a conventional reverse time migration profile are as follows: Figure 11 ;

[0058] like Figure 12 As shown, after outputting the results, a comparative analysis is performed using the conventional reverse-time migration imaging results, the conventional results without graded stabilization factor correction, and the results with graded stabilization factor high-dimensional deconvolution correction, respectively. (a), (b), and (c) represent the conventional reverse-time migration imaging results, the conventional results without graded stabilization factor PSF correction, and the results with graded stabilization factor high-dimensional deconvolution correction, respectively, with local magnification at the same location; (d), (e), and (f) represent the two-dimensional spectra corresponding to the above figures. Comparative Analysis Figure 12 Among the three images (a), (b), and (c), it was found that image (c) has a higher resolution, and the deep amplitude of the image can also be well recovered after high-dimensional deconvolution correction by the gradient stabilization factor.

[0059] Therefore, it can be seen that by using the high-dimensional deconvolution processing method of seismic images with gradual stability factors, it is possible to adapt more flexibly to changes in underground structures, effectively improve the resolution of seismic images, and thus improve the interpretation accuracy of seismic data.

[0060] Please see Figure 13 , Figure 13 This is a schematic diagram of the hardware device in operation according to an embodiment of the present invention. The hardware device specifically includes: a seismic image high-dimensional deconvolution processing device 401 based on a gradual stability factor, a processor 402, and a storage device 403.

[0061] A seismic image high-dimensional deconvolution processing device 401 based on a gradual stabilization factor: The seismic image high-dimensional deconvolution processing device 401 based on a gradual stabilization factor implements the seismic image high-dimensional deconvolution processing method based on a gradual stabilization factor.

[0062] Processor 402: The processor 402 loads and executes the instructions and data in the storage device 403 to implement the high-dimensional deconvolution processing method for seismic images based on the gradual stability factor.

[0063] Storage device 403: The storage device 403 stores instructions and data; the storage device 403 is used to implement the high-dimensional deconvolution processing method for seismic images based on a gradual stability factor.

[0064] The beneficial effects of this invention are as follows: Within the classical least-squares inversion imaging framework, this invention obtains conventional migration imaging profiles and scattering point models. The scattering point models are then subjected to Born forward modeling to generate pre-stack scattering data. The pre-stack scattering data is then processed using a conventional inverse time migration method to generate PSFs corresponding to all subsurface scattering points for correction. Two-dimensional spectral analysis is performed on the Fourier-transformed PSFs to construct a gradual stabilization factor for seismic image processing. This constructed gradual stabilization factor not only prevents computational instability, making the calculation process more robust, but also better compensates for depth-dependent uneven illumination, effectively improving the resolution of seismic images. This method allows for more flexible adaptation to changes in subsurface structures, thereby improving the interpretation accuracy of seismic data. Consequently, seismic exploration provides more accurate and clear subsurface structural information, overcomes some limitations of traditional methods, and improves image quality and interpretability.

[0065] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. 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.

Claims

1. A high-dimensional deconvolution processing method for seismic images based on a gradual stability factor, characterized in that: include: S1: Data preparation: Obtain conventional migration imaging profiles based on background velocity and reflection coefficient of the horizontal thin-layer model; A scattering point model is generated by inserting scattering points into the background velocity, and pre-stack scattering data is generated by Born forward modeling of the scattering point model. The pre-stack scattering data is then used to generate PSF corresponding to all underground scattering points for correction using a conventional inverse time migration method. S2: Input data: Store the conventional offset imaging profile and PSF blocks as I and P files; S3: Local Fourier Transform: Performs a Fourier transform on each local block data in the I and P files, converting it from the spatial domain to the wavenumber domain; S4: Constructing the gradual stabilization factor: Perform two-dimensional spectral analysis on the PSF data after Fourier transform to construct the gradual stabilization factor ε(z); S5: Image Correction: In the wavenumber domain, a block correction method is used. A high-dimensional deconvolution method based on the gradient stabilization factor and PSF is employed to perform wavenumber domain correction on each image block. The deconvolution calculation formula is as follows: in, This represents the profile corrected using the gradual stability factor in the wavenumber domain. This represents a conventional migration imaging profile in the wavenumber domain. Let ε(z) represent the PSF in the wavenumber domain, ε(z) be the gradual stabilization factor, x be the position in the imaging space, k be the local wavenumber, and z be the depth. S6: Local Inverse Fourier Transform: The corrected result... Perform an inverse Fourier transform to convert the wavenumber domain data into spatial domain data and store it in an Ic file; S7: Output results: Output the Ic file image in the spatial domain after gradual stabilization factor correction and local inverse Fourier transform.

2. The high-dimensional deconvolution processing method for seismic images based on a gradual stability factor as described in claim 1, characterized in that: In step S1, the size of the horizontal thin-layer model is 6000m×3800m, and the spatial sampling interval in both the horizontal and vertical directions is 10m. The scattering point model has a sampling interval of 270m in both the horizontal and vertical directions.

3. The high-dimensional deconvolution processing method for seismic images based on a gradual stability factor as described in claim 1, characterized in that: In step S2, in order to obtain better correction results, a Gaussian window function is used to perform windowing operations on each local block.

4. The high-dimensional deconvolution processing method for seismic images based on a gradual stability factor as described in claim 1, characterized in that: In step S2, the conventional offset imaging profile and PSF are stored in blocks using a column-first storage method.

5. The high-dimensional deconvolution processing method for seismic images based on a gradual stability factor as described in claim 1, characterized in that: In step S4, the gradient stabilization factor ε(z) is 5% of the maximum PSF value at the same depth z.

6. The high-dimensional deconvolution processing method for seismic images based on a gradual stability factor as described in claim 1, characterized in that: In step S6, for After performing the inverse Fourier transform, the data is stored column-wise in the Ic file.

7. A storage device, characterized in that: The storage device stores instructions and data for implementing the high-dimensional deconvolution processing method for seismic images based on the gradual stability factor as described in any one of claims 1 to 6.

8. A high-dimensional deconvolution processing device for seismic images based on a gradual stability factor, characterized in that: include: A processor and a storage device; the processor loads and executes instructions and data in the storage device to implement the high-dimensional deconvolution processing method for seismic images based on a gradual stability factor as described in any one of claims 1 to 6.

Images