A seismic data denoising method and system based on adaptive equalization

Abstract

The present invention relates to a seismic data denoising method and system based on adaptive equalization. The method comprises acquiring seismic data, performing two-dimensional smoothing on the acquired seismic data to obtain a smoothed profile; performing equalization on the acquired seismic data using the smoothed profile; and sequentially performing curvelet transform, curvelet domain denoising, inverse curvelet transform, and inverse equalization on the equalized seismic data to obtain denoised target seismic data. This method solves the existing problem of poor denoising due to poor energy equalization in shallow, medium, and deep depth seismic data.

Description

Technical Field

[0001] The present invention belongs to the technical field of oil and gas geophysical exploration engineering, and in particular relates to a seismic data denoising method and system based on adaptive equalization. Background Art

[0002] In current oil seismic exploration, the quality of seismic data collected has a significant impact on subsequent processing. However, in actual acquisition, due to the influence of many factors such as terrain, vehicles, and weather, the collected seismic data generally contains a lot of seismic noise, which has an adverse effect on subsequent seismic data processing and causes considerable trouble for seismic data processing and reservoir prediction.

[0003] Existing seismic data denoising methods include a curvelet domain seismic data denoising method and a block method. Among them, the curvelet domain seismic data denoising is a better seismic data denoising method, but the denoising effect is not ideal when the shallow, medium and deep energy in the shot set is uneven. To address the problem of energy imbalance, a Chinese invention patent application document with application publication number CN104181600A discloses a seismic data linear noise attenuation method and device. The method includes collecting first seismic data, arranging the shot records to generate left-arranged seismic data and right-arranged seismic data; performing energy balance processing on the left-arranged seismic data and the right-arranged seismic data to generate second Seismic data; scan out samples with noise and perform inverse energy equalization processing on the amplitude values ​​to form third seismic data; perform Fuji transform on the third seismic data to generate frequency-space domain seismic data; perform linear noise prediction processing on the frequency-space domain seismic data to generate fourth seismic data; perform inverse Fuji transform on the fourth seismic data to the time domain to generate fifth seismic data; combine the left-aligned seismic data and the right-aligned seismic data in the fifth seismic data to generate noise-predicted seismic data; subtract the shot record of the noise-predicted seismic data from the shot record of the first seismic data to generate linear noise-attenuated seismic data. Although the energy equalization method is mentioned in the application document, no specific technical means are provided. Regarding the block method, for example, Chinese invention patent application publication number CN109100788B discloses a non-local means denoising method for seismic data. The method obtains a data point to be detected in the seismic data to be detected; calculates the amplitude difference between the data point to be detected and its neighboring data points, and calculates the SUSAN area of ​​the data point to be detected; uses the SUSAN area to calculate the edge response of the data point to be detected; repeats these steps to calculate the edge response of all data points to obtain the edge response of the seismic data to be detected; divides the seismic data to be detected into optimized regions and conventional regions; denoises the conventional region using the non-local means denoising method; and performs optimized denoising on the optimized region. SUSAN edge detection can be used to extract edge information of the data structure, thereby demarcating the optimized region from the data. Within the optimized region, independent data similarity calculations are performed, which can improve the weighting effect between data within the optimized region and compensate for the incomplete denoising of the conventional non-local means denoising method in the structural edge regions of the data, namely the optimized region. This method primarily improves on the threshold method and the division of different regions. However, the block method increases the computational complexity and often leaves artifacts at the boundaries.

[0004] Therefore, the existing technology has the problem that the energy balance effect of shallow, medium and deep seismic data is not good, resulting in poor denoising effect. Summary of the Invention

[0005] The present invention provides a seismic data denoising method and system based on adaptive equalization, which is used to solve the problem in the prior art that the denoising effect is poor due to poor shallow, medium and deep energy equalization effect of seismic data.

[0006] To solve the above technical problems, the present invention provides a seismic data denoising method based on adaptive equalization, which includes:

[0007] 1) Acquire seismic data and perform two-dimensional smoothing on the acquired seismic data to obtain a smooth profile;

[0008] 2) Using the smoothed section to perform equalization processing on the acquired seismic data to obtain equalized seismic data; the equalization processing is: h (x,z)=p obs (x,z) / p abs (x,z), where p h (x,z) represents the equalized seismic data at the (x,z) coordinate, p obs (x,z) represents the seismic data acquired at the (x,z) coordinate, p abs (x,z) represents a smooth profile;

[0009] 3) Curvelet transform, curvelet domain denoising, inverse curvelet transform and inverse equalization are sequentially performed on the equalized seismic data to obtain denoised target seismic data. The inverse equalization process is as follows: in represents the target seismic data at (x,z) coordinates, Represents the seismic data after inverse curve transformation.

[0010] The above technical solution has the following beneficial effects: 2D smoothing is performed on the acquired seismic data to obtain a smoothed profile, and the smoothed profile is used to perform equalization on the acquired seismic data. The equalization is performed by dividing the acquired seismic data by the smoothed profile, and then denoising is performed in the curvelet domain. This improves the energy balance of shallow, medium, and deep seismic shot recordings, thereby improving the denoising accuracy of the seismic data during curvelet domain denoising, thereby achieving effective denoising. Furthermore, the problems of increased computational complexity and artifacts at boundaries are avoided.

[0011] Furthermore, in order to improve the denoising effect, the present invention provides a seismic data denoising method based on adaptive equalization, which further includes the curvelet domain denoising processing as filtering processing, and the filtering processing is in is the filtered seismic data, c m is the seismic data after curvelet transformation, T(c m |c m>α) indicates that c m Perform filtering and retain the part greater than α, where α is the threshold and m represents the label of the image corresponding to the seismic data.

[0012] Furthermore, in order to better obtain a smooth profile, the present invention provides a seismic data denoising method based on adaptive equalization, which also includes a step of performing initial denoising on the acquired seismic data before performing two-dimensional smoothing in step 1).

[0013] Furthermore, in order to better obtain a smooth profile, the present invention provides a seismic data denoising method based on adaptive equalization, which also includes initial denoising processing as Fourier transform denoising.

[0014] Furthermore, in order to better obtain a smooth profile, the present invention provides a seismic data denoising method based on adaptive equalization, which also includes that the two-dimensional smoothing processing in step 1) is a two-dimensional Gaussian smoothing processing.

[0015] Furthermore, in order to better obtain a smooth profile, the present invention provides a seismic data denoising method based on adaptive equalization, which also includes in step 1) two-dimensional smoothing processing including horizontal and vertical smoothing processing, and the degree of vertical smoothing is greater than the degree of horizontal smoothing.

[0016] Furthermore, in order to better obtain a smooth profile, the present invention provides a seismic data denoising method based on adaptive equalization, further comprising a two-dimensional smoothing process being a two-dimensional Gaussian smoothing process, wherein the two-dimensional Gaussian smoothing process is: in, represents the initial denoised seismic data at the (x, z) coordinate, abs represents the absolute value, sx represents the number of horizontal smoothing points, and sz represents the number of vertical smoothing points. Express A two-dimensional Gaussian smoothing function with sx points in the horizontal direction and sz points in the vertical direction, where δ represents a stable number that is not 0.

[0017] Furthermore, in order to better improve the denoising effect, the present invention provides a seismic data denoising method based on adaptive equalization, which further includes performing the inverse curvelet transform in step 3) as follows: in are the basis functions of the curvelet transform.

[0018] Furthermore, in order to better improve the denoising effect, the present invention provides a seismic data denoising method based on adaptive equalization, which also includes a curvelet transform in step 3), where the curvelet transform is: in is the basis function of the curvelet transform, c m is the seismic data after curvelet transformation.

[0019] The present invention also provides a seismic data denoising system based on adaptive equalization, which includes a memory and a processor, wherein the processor is used to execute instructions stored in the memory to implement the above-mentioned seismic data denoising method based on adaptive equalization. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] Figure 1 is a flow chart of a seismic data denoising method based on adaptive equalization of the present invention;

[0021] Figure 2 A complete seismic gun record for an example of the present invention;

[0022] Figure 3 Noisy seismic shot records for examples of the present invention;

[0023] Figure 4 is a smooth cross section of the present invention;

[0024] Figure 5 The denoised seismic shot record of the present invention;

[0025] Figure 6 Seismic shot records denoised by conventional methods;

[0026] Figure 7(a) shows the error between the denoised profile and the true profile using the method of the present invention;

[0027] Figure 7(b) shows the error between the denoised profile and the true profile using the conventional method. DETAILED DESCRIPTION

[0028] The basic concept of the present invention is to perform two-dimensional smoothing on the acquired seismic data to obtain a smoothed profile, use the smoothed profile to perform equalization on the acquired seismic data, and sequentially perform curvelet transform, curvelet domain denoising, inverse curvelet transform, and inverse equalization on the equalized seismic data to obtain denoised target seismic data. The equalization process involves dividing the acquired seismic data by the smoothed profile. Thus, the two-dimensional smoothed profile obtained by the two-dimensional smoothing process is used to equalize the seismic shot records, thereby improving the energy balance of the shallow, medium, and deep seismic shot records. This improves the denoising accuracy of the seismic data during curvelet domain denoising, thereby achieving effective denoising.

[0029] In order to make the purpose, technical solutions and technical effects of the present invention more clear, the present invention is further described in detail below with reference to the accompanying drawings and specific embodiments.

[0030] Embodiment of a seismic data denoising method based on adaptive equalization:

[0031] Figure 1This is a flow chart of the seismic data denoising method based on adaptive equalization of the present invention. The specific process is as follows:

[0032] Step 1: Acquire seismic data and perform initial denoising on the acquired seismic data.

[0033] In step 1, the seismic data may be the original data in the seismic acquisition data. The seismic data may also be referred to as seismic shot records. If the seismic data is presented in the form of a profile, the seismic data obtained in step 1 is the original profile. For this embodiment, in order to facilitate the technical effect of the denoising method of this embodiment in conjunction with the accompanying drawings, this embodiment uses the noise-containing Figure 3 As an example of the acquired seismic data. Figure 2 This is a complete gun record for testing the present invention. Figure 3 This is a noisy record used for testing the present invention. Figure 3 The specific method is as follows: in seismic exploration, according to the needs of seismic acquisition, observe the seismic shot records. If wave equation imaging is performed, it is often necessary to record the seismic shot records of the common shot. The finite difference forward simulation of the layered model is used to obtain the following: Figure 2 Single shot record, yes Figure 2 Add random noise to the single shot record, the noise is 1% of the maximum value, and get Figure 3 The noisy seismic shot record shown in Figure 1 is shown in Figure 2. Since the energy of the direct wave is very strong, the noise is not very significant. Figure 2 The deep signal is heavily polluted. Figure 3 The data is very noisy.

[0034] In step 1, initial denoising is performed to obtain initial denoised seismic data. The initial denoising in this embodiment refers to a conventional denoising method, such as Fourier transform denoising. This removes large outliers from the seismic data, facilitating subsequent acquisition of a smoothed profile.

[0035] Step 2: Smoothing the seismic data after initial denoising.

[0036] Specifically, in step 2, the smoothing process is a two-dimensional smoothing process. The two-dimensional smoothing process includes horizontal and vertical smoothing processes. After the two-dimensional smoothing process, the horizontal and vertical smoothed sections are obtained, and the degree of vertical smoothing is greater than that of horizontal smoothing. In this case, the smoothing process can better stabilize the local relative relationship of the seismic data, thereby better obtaining a smooth section. For this embodiment, Figure 4 is a smooth cross section of the present invention. Figure 3 After initial denoising and smoothing, we get Figure 4 Smoothed cross section shown.

[0037] In step 2, if Figure 1As shown in FIG, before smoothing, the absolute value of the initial denoised seismic data (i.e., the initial denoised seismic data) is taken, and then the seismic data after taking the absolute value is subjected to two-dimensional smoothing. The two-dimensional smoothing process uses a two-dimensional Gaussian smoothing function. The two-dimensional smoothing process is:

[0038]

[0039] Where p in formula (1) abs (x,z) represents the smooth section, (x,z) represents the coordinates, represents the initial denoised seismic shot record at the (x, z) coordinate, abs represents the absolute value, sx represents the number of points for lateral smoothing, and sz represents the number of points for longitudinal smoothing. Express A two-dimensional Gaussian smoothing function with sx points horizontally and sz points vertically, where δ represents a non-zero stability factor. In this case, taking the square root during smoothing better preserves the overall relative relationship of the seismic data, while taking the absolute value ensures that the divisor remains positive during subsequent processing. Setting a stability factor prevents the denominator from being zero in subsequent processing.

[0040] In this embodiment, δ is set to one ten-thousandth of the maximum absolute value, sx=20, sz=100.

[0041] Step 3: Perform equalization processing on the acquired seismic data.

[0042] Specifically, in step three, the seismic data obtained in step one (e.g., the original profile) is equalized using the horizontal and vertical smoothed sections to obtain equalized seismic data. Since the smoothed section is obtained based on the acquired seismic data, and the smoothing process can ensure the relative relationship between the overall and local aspects of the seismic data, in this case, if the acquired seismic data is large, the obtained smoothed section is also large, and if the acquired seismic data is small, the obtained smoothed section is also small. Equalization utilizes the acquired seismic data divided by the smoothed section, that is, the seismic data acquired in different layers (e.g., the front, middle, and deep layers) is divided by the smoothed section of the corresponding layer to obtain the equalized seismic data of the corresponding layer. The energy relationship between the equalized data of different layers is more balanced than the energy relationship between the seismic data of different layers obtained in step one. Therefore, after equalization, the energy balance of the shallow, middle, and deep seismic shot records can be improved, which facilitates the subsequent denoising process in the curvelet domain.

[0043] In step three, the equalization process is:

[0044]

[0045] Where p in formula (2) h(x, z) represents the equalized seismic shot record (i.e., equalized seismic data), p obs (x,z) represents the input seismic shot record (ie, acquired seismic data) at the (x,z) coordinates.

[0046] In step three, if the original profile is equalized, the obtained equalized seismic data is the seismic shot recording profile.

[0047] Step 4: Perform curvelet transform on the equalized seismic data.

[0048] In step 4, after the curvelet transform, the seismic data in the curvelet domain (i.e., the seismic data after the curvelet transform) is obtained. Specifically, since most of the events of the seismic data are curves, the curvelet transform is an ideal transform for sparsely representing the seismic wave field. Based on the fact that the reflection coefficient is more sparse in the curvelet domain, the curvelet transform is used to implement the denoising method. In the curvelet transform, the curvelet transform in the two-dimensional case is:

[0049]

[0050] Wherein formula (3) is the basis function of the curvelet transform, p h (x,z) is the equalized seismic shot record at the (x,z) coordinate, c m is the seismic shot record after curvelet transformation, and m represents the label of the corresponding image (i.e., section) of the seismic shot record. In this way, the denoising effect can be better improved.

[0051] In step 4, the seismic data after curvelet transformation can be presented in the form of a matrix. If the seismic shot record profile is obtained after equalization in step 3, then Figure 1 As shown, step 4 performs a curvelet transform on the seismic shot record section to obtain a curvelet domain expression matrix (i.e., seismic data after curvelet transform). In this case, the expression matrix can represent most of the seismic shot record information.

[0052] Step 5: Perform filtering in the curvelet domain.

[0053] Specifically, in step 5, the curvelet domain seismic data obtained in step 4 is filtered to obtain filtered seismic data. In this case, only the main information is retained after filtering, thereby further improving the denoising effect.

[0054] In step 5, if the filtering process adopts the threshold function method, the filtering process is:

[0055]

[0056] Wherein formula (4) is the filtered seismic shot record (i.e., filtered seismic data), c m is the seismic shot record after curvelet transformation, T(c m |c m >α) indicates that c m Filtering is performed, retaining the portion greater than α, where α is a threshold and m represents the index of the corresponding image recorded by the seismic shot. In this case, filtering only retains the key information of the seismic data (i.e., the portion greater than α), thereby improving the denoising effect.

[0057] In step 5, the filtered seismic data can be presented in the form of a matrix. If the expression matrix in the curvelet domain is obtained after the curvelet transformation in step 4, the expression matrix is ​​filtered using a threshold function in step 5 to obtain a filter matrix.

[0058] Step 6: Perform inverse curve transform on the filtered seismic data.

[0059] Specifically, in step 6, the inverse curvelet transform is referred to as the inverse transform, and the filter matrix obtained in step 5 is inversely transformed to obtain the secondary denoised seismic shot record.

[0060]

[0061] Wherein the formula (5) is the basis function of the curvelet transform, is the seismic shot record after inverse curvelet transform. In this case, the seismic data in the curvelet domain is converted back to the time domain through inverse curvelet transform, thereby better improving the denoising effect. In addition, the formula (5) of the inverse curvelet transform can also be called the sparse denoising formula of the curvelet transform. It is called the denoising result after sparse constraint.

[0062] Step 7: Perform inverse equalization processing on the secondary denoised seismic data to obtain denoised target seismic data.

[0063] In step seven, the de-equalization process is:

[0064]

[0065] Wherein the formula (6) is the anti-equalized seismic shot record at the (x, z) coordinate (i.e., target seismic data), is the seismic shot record after inverse curve transformation. For this embodiment, Figure 5 This is the denoised seismic shot record of the present invention. Figure 4 The smooth cross section of Figure 3 The seismic cannon records are processed from step 3 to step 7 to obtain Figure 5The denoised seismic shot record is shown. Figure 3 It is a single shot record, so Figure 5 Also recorded for seismic single shots.

[0066] For this embodiment, Figure 6 Figure 7(a) shows the error between the denoised seismic shot record obtained using the present invention and the conventional denoising method, and a standard shot record. Figure 7(b) shows the error between the denoised seismic shot record obtained using the present invention and the true seismic shot record, while Figure 7(a) shows the error between the denoised seismic shot record obtained using the conventional denoising method. This shows that the denoised seismic shot record obtained using the denoising method of this embodiment has a smaller error.

[0067] According to the seismic data denoising method based on adaptive equalization of this embodiment, the acquired seismic data is subjected to two-dimensional smoothing to obtain a smoothed profile, and the acquired seismic data is equalized using the smoothed profile, wherein the acquired seismic data is divided by the smoothed profile for equalization, and then denoised in the curvelet domain. This can improve the energy balance of shallow, medium and deep seismic shot records, thereby improving the denoising accuracy of the seismic data when performing curvelet domain seismic data denoising, thereby achieving effective denoising. In addition, the seismic data after equalization is subjected to curvelet transform, curvelet domain denoising, inverse curvelet transform and inverse equalization in sequence to obtain the denoised target seismic data. In this way, seismic data with a high denoising effect can be obtained. In addition, the problems of increased computational complexity and traces at the boundaries are avoided.

[0068] In this embodiment, the seismic data after the initial denoising is smoothed. In other embodiments, the seismic data obtained in step 1 can be directly smoothed.

[0069] Example of a seismic data denoising system based on adaptive equalization:

[0070] This embodiment discloses a seismic data denoising system based on adaptive equalization. Through the seismic data denoising system based on adaptive equalization of this embodiment, a seismic data denoising method based on adaptive equalization introduced in the method embodiment of the present invention can be implemented.

[0071] In this embodiment, a seismic data denoising system based on adaptive equalization includes a processor and a memory. The processor is configured to execute instructions stored in the memory to implement the seismic data denoising method based on adaptive equalization in the method embodiment of the present invention. This seismic data denoising method based on adaptive equalization has been described in detail in the above-mentioned method embodiment. Those skilled in the art can generate corresponding computer instructions based on this seismic data denoising method to obtain a seismic data denoising system based on adaptive equalization, and the details will not be repeated here. The memory is configured to store the computer instructions generated by the seismic data denoising method based on adaptive equalization.

[0072] The seismic data denoising system based on adaptive equalization according to this embodiment can solve the problem of ineffective denoising when the energy difference between shallow, medium and deep parts of seismic data is large, that is, it can solve the problem in the prior art that the denoising effect is poor due to the poor equalization effect of the shallow, medium and deep parts of seismic data.

Claims

1. A seismic data denoising method based on adaptive equalization, characterized in that: include: 1) Acquire seismic data and perform two-dimensional smoothing on the acquired seismic data to obtain a smooth profile; 2) Using the smoothed section to perform equalization processing on the acquired seismic data to obtain equalized seismic data; the equalization processing is: h (x,z)=p obs (x,z) / p abs (x,z), where p h (x,z) represents the equalized seismic data at the (x,z) coordinate, p obs (x,z) represents the seismic data acquired at the (x,z) coordinate, p abs (x,z) represents a smooth profile; 3) Curvelet transform, curvelet domain denoising, inverse curvelet transform and inverse equalization are sequentially performed on the equalized seismic data to obtain denoised target seismic data. The inverse equalization process is as follows: in represents the target seismic data at (x,z) coordinates, Represents the seismic data after inverse curve transformation.

2. The seismic data denoising method based on adaptive equalization according to claim 1, characterized in that: The curvelet domain denoising process is a filtering process, and the filtering process is in is the filtered seismic data, c m is the seismic data after curvelet transformation, T(c m |c m >α) indicates that c m Perform filtering and retain the part greater than α, where α is the threshold and m represents the label of the image corresponding to the seismic data.

3. The seismic data denoising method based on adaptive equalization according to claim 1, characterized in that: In step 1), before performing the two-dimensional smoothing process, the method further includes performing an initial denoising process on the acquired seismic data.

4. The seismic data denoising method based on adaptive equalization according to claim 3, characterized in that: The initial denoising process is Fourier transform denoising.

5. The seismic data denoising method based on adaptive equalization according to claim 1, characterized in that: In step 1), the two-dimensional smoothing process is a two-dimensional Gaussian smoothing process.

6. The seismic data denoising method based on adaptive equalization according to claim 1, characterized in that: In step 1), the two-dimensional smoothing process includes horizontal and vertical smoothing processes, and the degree of vertical smoothing is greater than that of horizontal smoothing.

7. The seismic data denoising method based on adaptive equalization according to claim 3, characterized in that: The two-dimensional smoothing process is a two-dimensional Gaussian smoothing process, and the two-dimensional Gaussian smoothing process is: in, represents the initial denoised seismic data at the (x, z) coordinate, abs represents the absolute value, sx represents the number of horizontal smoothing points, and sz represents the number of vertical smoothing points. Express A two-dimensional Gaussian smoothing function with sx points in the horizontal direction and sz points in the vertical direction, where δ represents a stable number that is not 0.

8. The seismic data denoising method based on adaptive equalization according to claim 2, characterized in that: The inverse curvelet transform in step 3) is: in are the basis functions of the curvelet transform.

9. The seismic data denoising method based on adaptive equalization according to claim 1, characterized in that: In the curvelet transform of step 3), the curvelet transform is: in is the basis function of the curvelet transform, c m is the seismic data after curvelet transformation.

10. A seismic data denoising system based on adaptive equalization, characterized in that: include: A memory and a processor, wherein the processor is configured to execute instructions stored in the memory to implement the seismic data denoising method based on adaptive equalization according to any one of claims 1 to 9.

Images