Method and device for improving signal-to-noise ratio of ultra-shallow layer development data

Abstract

The invention relates to the technical field of signal processing, in particular to a method and device for improving the signal-to-noise ratio of ultra-shallow layer development data. Aiming at the requirement of ultra-shallow development for high signal-to-noise ratio of a signal processing result, data regularization and noise suppression are innovatively formed as a basic technical means, a quality control method based on a geological mode is adopted, the noise suppression effect is improved through multi-domain multi-round whole-process quality control, effective signals are protected at the same time, and the method is suitable for large-scale popularization and application. And finally, the signal-to-noise ratio is improved, so that support and development of seismic geology research basic data for removing counterfeit and preserving true are realized, high-quality signal processing results are used for supporting development of normal, non-normal and new energy resources, and the target of rapidly forming productivity by new energy resources is realized.

Description

Technical Field

[0001] The present invention relates to the technical field of signal processing, and particularly to a method and device for improving the signal-to-noise ratio of ultra-shallow development data. Background Art

[0002] In the development practices for conventional, unconventional, and new energy sources, considerable potential has been seen in many basins in China. With the overall upgrade of development seismic technology, especially the significant acceleration of new energy development, fine seismic exploration for targets and reservoirs has replaced the development fine geological research that supports the determination of structures, and higher requirements are put forward for the fidelity and signal-to-noise ratio of seismic data that supports development research.

[0003] Data regularization and noise suppression are important techniques for improving the signal-to-noise ratio of data. They are not only the basic links of the whole process but also determine the signal-to-noise ratio of the final processing results. The demand for the fidelity of processing results in development seismic research is unprecedented. Data regularization and denoising directly determine the upper limit of the final fidelity of a set of data processing by improving the signal-to-noise ratio. Traditional noise suppression and quality control methods have mature processes. Especially in the past decade, high-density three-dimensional acquisition, "two-wide and one-high" or even "two-wide and two-high" acquisition have gradually become the mainstream types of seismic acquisition observation systems. The description of noise is clearer and more developed. There are two possible problems with traditional noise suppression and quality control methods: inconsistent denoising scales, incomplete denoising, or neglecting fidelity and signal protection. The target objects of new energy exploration and development are different from those of traditional conventional and unconventional resources. Taking uranium ore as an example, oil and gas resources usually do not take 0 - 300m as the imaging target, but uranium ore exploration requires precise imaging of the ultra-shallow layer. In traditional processing, the denoising link will suppress both the noise and signals in this part to a great extent. There is an urgent need for a new fidelity noise suppression method to improve the fidelity of the basic data processing results of new energy development. Summary of the Invention

[0004] In view of the above problems, the present invention provides a method and device for improving the signal-to-noise ratio of ultra-shallow development data, which can improve the signal-to-noise ratio of development basic data, thereby enhancing the success rate and timeliness of conventional energy and new energy development.

[0005] In the first aspect, the present invention provides a method for improving the signal-to-noise ratio of ultra-shallow development data, the method comprising:

[0006] Performing surface wave analysis and surface wave dispersion prediction on seismic data to obtain surface wave noise, and performing coherent noise prediction to obtain coherent noise;

[0007] After respectively reconstructing the coherent noise and the surface wave noise, performing stacking at the same velocity to obtain a noise stacking profile, and performing spectral analysis to obtain the dominant frequency band range of the noise;

[0008] Filter the seismic data within the noise dominant frequency band range; and perform gather extraction and stacking processing on the filtered data to obtain a processed data volume.

[0009] In the processed data volume after filtering within the noise frequency band range, determine the seismic response characteristics of the effective reflection signals through analysis, and quality control whether the same seismic response characteristics exist in the noise stacked section. If there are no identical response characteristics, use the method of adaptive subtraction in the full-band seismic processing results to remove coherent noise and surface wave noise, and obtain a denoised data volume.

[0010] Furthermore, for the seismic data, perform automatic first arrival picking on the original seismic data based on a machine learning method. After picking, perform quality control and select the offset range with excellent picking effect for static correction calculation. If the picking accuracy rate is less than the threshold, readjust the learning samples and repeat the automatic picking until the accuracy rate is greater than the threshold, and perform zero-padding on the upper part of the first arrival.

[0011] Furthermore, the seismic data also includes:

[0012] Calculate the tomographic static correction amount for the first arrival information, and perform quality control after applying the static correction amount to obtain the seismic data after static correction application.

[0013] Furthermore, perform surface wave analysis and surface wave dispersion prediction on the seismic data to obtain surface wave noise, including:

[0014] Perform surface wave analysis and surface wave dispersion prediction on the seismic data after static correction application to obtain surface wave noise.

[0015] Furthermore, perform coherent noise prediction to obtain coherent noise, including:

[0016] Perform gather extraction on the seismic data after static correction application, and complete the regularization processing of the radial five-dimensional data using the matching pursuit method to obtain a seismic data volume with uniform distribution in each azimuth sector;

[0017] Perform Fourier transform on the seismic data volume and perform coherent noise prediction in the fk domain to obtain coherent noise.

[0018] Furthermore, after separately reconstructing the coherent noise and surface wave noise, stack them using the same velocity to obtain a noise stacked section, and perform spectral analysis to obtain the noise dominant frequency band range, including:

[0019] After separately reconstructing the coherent noise and surface wave noise, stack them using the same velocity to obtain a noise stacked section;

[0020] Perform spectral analysis on the noise stacked section to obtain the noise dominant frequency band range with the target layer as the window.

[0021] Furthermore, it also includes:

[0022] When there are the same seismic response characteristics in the noise superposition profile, reduce the noise prediction degree control parameter, and re-predict the surface wave noise and the coherent noise.

[0023] In a second aspect, the present invention provides an apparatus for improving the signal-to-noise ratio of ultra-shallow development data, including: a noise prediction unit, a noise dominant frequency band calculation unit, and a denoising unit;

[0024] The noise prediction unit is configured to perform surface wave analysis and surface wave dispersion prediction on seismic data to obtain surface wave noise, and perform coherent noise prediction to obtain coherent noise;

[0025] The noise dominant frequency band calculation unit is configured to respectively reconstruct the coherent noise and the surface wave noise and then perform superposition at the same velocity to obtain a noise superposition profile, and perform spectral analysis to obtain the noise dominant frequency band range;

[0026] The denoising unit is configured to perform filtering processing on the seismic data within the noise dominant frequency band range; and perform trace gathering and superposition processing on the filtered data to obtain a processed data volume;

[0027] The denoising unit is further configured to, in the processed data volume filtered within the noise frequency band range, determine the seismic response characteristics of the effective reflection signal by analysis, and quality control whether there are the same seismic response characteristics in the noise superposition profile. If there are no same response characteristics, in the full-frequency band seismic processing results, remove the coherent noise and the surface wave noise by an adaptive subtraction method to obtain a denoised data volume.

[0028] In a third aspect, the present invention provides an electronic device, including a processor, a communication interface, a memory, and a communication bus. Among them, the processor, the communication interface, and the memory complete communication with each other through the communication bus;

[0029] The memory stores a computer program;

[0030] The processor is configured to, when executing the computer program stored on the memory, implement the method for improving the signal-to-noise ratio of ultra-shallow development data as described above.

[0031] In a fourth aspect, the present invention provides a computer-readable storage medium storing a computer program, and when the computer program is executed by a processor, the method for improving the signal-to-noise ratio of ultra-shallow development data as described above is implemented.

[0032] The present invention at least has the following beneficial effects:

[0033] In response to the demand for high signal-to-noise ratio in signal processing results for ultra-shallow development, through the innovation of the present invention, data regularization and noise suppression are used as basic technical means, and a quality control method based on geological models is adopted. Through multi-domain and multi-round full-process quality control, the effect of noise suppression is improved while protecting the effective signals. The parameters are iteratively updated to ultimately improve the signal-to-noise ratio, thereby realizing the "removing false and retaining true" of the basic data for development seismic geology research, and supporting the development of conventional, unconventional, and new energy with high-quality signal processing results to achieve the goal of quickly forming production capacity for new energy.

[0034] Other features and advantages of the present invention will be described in the following specification, and, in part, will be obvious from the specification, or will be understood by implementing the present invention. The objectives and other advantages of the present invention can be achieved and obtained by the structures pointed out in the specification and the drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for the description of the embodiments or the prior art. Obviously, the drawings in the following description are some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings.

[0036] Figure 1 It is a flowchart of the method for improving the signal-to-noise ratio in the embodiments of the present invention;

[0037] Figure 2 It is a schematic structural diagram of the device for improving the signal-to-noise ratio in the embodiments of the present invention;

[0038] Figure 3 It is a schematic structural diagram of an electronic device;

[0039] Figure 4 It is a specific flowchart of the method for improving the signal-to-noise ratio of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0040] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the following will clearly and completely describe the technical solutions in the embodiments of the present invention with reference to the drawings in the embodiments of the present invention. Obviously, the described embodiments are some, but not all, of the embodiments of the present invention. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts fall within the protection scope of the present invention.

[0041] The present invention improves the signal-to-noise ratio of ultra-shallow development data based on machine learning with geological model quality control, which can achieve the improvement of the signal-to-noise ratio of development basic data, thereby enhancing the success rate and timeliness of conventional energy and new energy development.

[0042] Such asFigure 1 As shown in the figure, the present invention provides a method for improving the signal-to-noise ratio of ultra-shallow development data, and the method includes:

[0043] S101, performing surface wave analysis and surface wave dispersion prediction on seismic data to obtain surface wave noise, and performing coherent noise prediction to obtain coherent noise;

[0044] S102, respectively reconstructing the coherent noise and the surface wave noise and then stacking them at the same velocity to obtain a noise stacking profile, and performing spectral analysis to obtain the dominant frequency band range of the noise;

[0045] S103, performing filtering processing on the seismic data within the dominant frequency band range of the noise; and performing trace gathering and stacking processing on the filtered data to obtain a processed data volume;

[0046] S104, in the processed data volume filtered within the noise frequency band range, determining the seismic response characteristics of the effective reflection signal through analysis, and quality controlling whether there are the same seismic response characteristics in the noise stacking profile. If there are no same response characteristics, in the full-frequency band seismic processing results, the coherent noise and the surface wave noise are removed through an adaptive subtraction method to obtain a denoised data volume.

[0047] In specific implementation, the so-called five dimensions in the five-dimensional anti-aliasing data regularization technology refer to performing search and interpolation processing in five dimensions: the main survey line direction (x direction of the spatial coordinate system), the connecting survey line direction (y direction of the spatial coordinate system), the time axis direction (negative z direction underground of the spatial coordinate system), offset, and azimuth, which is called five dimensions. When the spatial sampling is insufficient, it is very easy to appear spatial aliasing without considering the interpolation processing of the high cut-off frequency. In order to prevent aliasing, based on time-frequency analysis, the start and end frequencies in the frequency domain are given before interpolation, and the possible spatial aliasing can be eliminated through prior parameter control.

[0048] The algorithm that is relatively successful in the Fourier transform-based seismic data regularization method is the interpolation algorithm based on the orthogonal matching pursuit algorithm. This method uses an extended function set to represent seismic data. The coefficients of this extended function are calculated by successively extracting the minimum data error function. This method usually produces a sparse expression of seismic data and can overcome spatial aliasing in some cases. The proposed Fourier transform method with anti-aliasing ability is to minimize the leakage of the energy of Fourier coefficients as much as possible. From the perspective of the signal, the sampling theorem is repeatedly used to re-estimate the Fourier coefficients, and a simple subtraction is applied to realize the re-orthogonalization of the basis functions in the Fourier transform. The non-uniform Fourier transform is used to obtain the spectrum of irregular data, as shown in the following formula:

[0049]

[0050] In formula (1), k is the frequency domain component; x s is the spatial non-uniform sampling point position; f(x s ) corresponds to x s The signal value at ; f(k) is the spectrum corresponding to the signal; N is the number of non-uniform samples; ΔD is the maximum space; Δx s is the spatial sampling interval.

[0051] In the data reconstruction process, the sparse component with the largest energy in the spectrum is first extracted, and the seismic data is converted back to the time domain through IFFT. Then the time domain data minus the maximum energy component in the frequency domain is transformed by FFT, and the maximum energy component in the existing data is extracted. And so on, each sparse frequency component is extracted one by one according to the energy strength through iterative extraction, and then the a priori anti-aliasing high cutoff frequency parameter control is used to prevent aliasing from being generated when the inverse Fourier transform is used to reconstruct the data. The specific method is as follows:

[0052]

[0053]

[0054] In formula (2) and formula (3), f k (x s ) is the time domain seismic data after IFFT, It is the seismic data returned to the time domain after the strongest energy component is subtracted from the input FFT data in the nth round through IFFT. After multiple rounds of iterations, the seismic data at the target location can be reconstructed. In the model seismic data trial calculation example based on the Fourier transform prestack data regularization method, it can also be seen that the missing trace data is well recovered and no spatial aliasing is generated.

[0055] The specific implementation is described as follows:

[0056] S1. The original seismic data is automatically picked for the first arrival based on the machine learning method. After picking, quality control is performed to select the offset distance range with the best picking effect for the static correction value calculation in the next step. If the picking accuracy is less than 98%, the learning sample is readjusted and the automatic picking is repeated until the accuracy is greater than 98%, and the upper part of the first arrival is filled with zero.

[0057] S2. Apply the first arrival information obtained in S1 to calculate the tomographic static correction. Perform quality control after applying the static correction. If the continuity of the main event axis is significantly improved, the quality control passes. If the continuity of the main event axis deteriorates, check whether the static correction amount needs to be ×(-1) before applying it until it passes the quality control.

[0058] S3. Process the seismic data after applying the static correction generated in S2 into gather data, and complete the regularization processing of the radial five-dimensional data by using the matching pursuit method to obtain a seismic data volume with uniform distribution in each azimuth sector.

[0059] S4. Perform Fourier transform on the seismic data after regularizing the five-dimensional data obtained in S3.

[0060] S5. Predict coherent noise on the data generated in S4 in the fk domain.

[0061] S6. Perform surface wave analysis and surface wave dispersion prediction on the data generated in S2 in the geophone line domain.

[0062] S7. Reconstruct and extract the coherent noise obtained in S5, and reconstruct the surface wave and surface wave dispersion predicted in S6.

[0063] S8. Superimpose the two types of reconstructed noise obtained in S7. The stacking velocity comes from velocity analysis, and the same velocity must be applied to all stackings and quality controls from S8 to S15.

[0064] S9. Perform spectral analysis on the noise stacking profile generated in S8 to determine the dominant frequency band range of the noise with the target layer as the window.

[0065] S10. Perform filtering on the original seismic data after static correction in S2 in the same frequency band range as the dominant frequency band range of the noise stacking in S9.

[0066] S11. Process the filtered data obtained in S10 into gather data and perform stacking.

[0067] S12. In the quality control of the noise stacking profile obtained in S8, check whether there is an effective reflection signal with a similar structural feature to the original signal in the same frequency band as that in S11. If there is an effective signal, perform step S13; if there is no effective signal, perform step S14.

[0068] The frequency band width of the onshore seismic data processing results is usually between 3 - 80 Hz, including possible noise components, single-frequency interferences such as industrial electricity, effective reflection components, and other types of wave field components. The noise frequency band range is generally 3 - 15 Hz, and some high-frequency noises can reach 30 Hz. The data within the same frequency band is objectively comparable. Therefore, it is necessary to first perform spectral analysis on the noise components with a narrower frequency band range, determine the noise frequency band distribution, and then filter the overall processing results with this frequency band range. It is more scientific, rigorous, and objective to perform quality control and comparison on the data within the same frequency band range.

[0069] S13. Repeat steps S5 and S6 to reduce the noise prediction degree control parameters, and repeat the steps until S12.

[0070] S14. Subtract the noise data that has passed the S12 quality control from the original data through adaptive subtraction. The adaptive subtraction in this step is not the subtraction of the stacked profiles, but the subtraction of the pre-stack original data after static correction and the noise predicted by S5 and S6.

[0071] S15. After performing dynamic correction and stacking on the noise suppression result of the adaptive subtraction obtained in S14, check whether the signal-to-noise ratio of the target layer for quality control meets the geological requirements and exploration design. If the signal-to-noise ratio meets the geological requirements, output the data of S14 for subsequent processing; if the quality control fails, repeat steps S5 - S14, appropriately increase the degree of noise suppression, and achieve the maximum noise suppression on the basis of ensuring data fidelity.

[0072] As Figure 2 shown, the present invention provides a device for improving the signal-to-noise ratio of ultra-shallow development data, including: a noise prediction unit 201, a noise dominant frequency band calculation unit 202, and a denoising unit 203;

[0073] The noise prediction unit 201 is used to perform surface wave analysis and surface wave dispersion prediction on seismic data to obtain surface wave noise, and perform coherent noise prediction to obtain coherent noise;

[0074] The noise dominant frequency band calculation unit 202 is used to reconstruct the coherent noise and surface wave noise respectively and then stack them at the same velocity to obtain a noise stacked profile, and perform spectral analysis to obtain the noise dominant frequency band range;

[0075] The denoising unit 203 is used to perform filtering processing on seismic data within the noise dominant frequency band range; and perform trace gather extraction and stacking processing on the filtered data to obtain a processed data volume;

[0076] The denoising unit 203 is also used to analyze and determine the seismic response characteristics of effective reflection signals in the processed data volume after filtering within the noise frequency band range, and check whether there are the same seismic response characteristics in the noise stacked profile. If there are no same response characteristics, use the method of adaptive subtraction in the full-band seismic processing results to remove the coherent noise and surface wave noise to obtain a denoised data volume.

[0077] As Figure 3 shown, the present invention provides an electronic device, including a processor 301, a communication interface 302, a memory 303, and a communication bus 304. Among them, the processor 301, the communication interface 302, and the memory 303 complete mutual communication through the communication bus 304;

[0078] The memory 303 stores a computer program;

[0079] The processor 301 is used to implement the above method when executing the computer program stored on the memory 303.

[0080] The present invention provides a computer-readable storage medium storing a computer program, which when executed by a processor implements the above-described method.

[0081] The computer-readable storage medium may be included in the device / apparatus described in the above embodiments; or may exist separately without being assembled into the device / apparatus. The above computer-readable storage medium carries one or more programs, which when executed, implement the method according to the embodiments of the present disclosure.

[0082] According to an embodiment of the present disclosure, the computer-readable storage medium may be a non-volatile computer-readable storage medium, which may include, for example, but not limited to: portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the above. In the present disclosure, the computer-readable storage medium may be any tangible medium that contains or stores a program, which can be used by or in combination with an instruction execution system, apparatus, or device.

[0083] To enable those skilled in the art to better understand the present invention, the principle of the present invention is described below with reference to the accompanying drawings:

[0084] The method and device for improving the signal-to-noise ratio of ultra-shallow development data based on machine learning for geological model quality control according to the present invention change the traditional concept and process of quality control after subtraction. The invention can maximize the fidelity of seismic data after noise suppression, provide high-quality basic data for the development of conventional, unconventional, and new energy, and has good application and promotion prospects.

[0085] As Figure 4 shown, taking the 3D seismic data of a certain oilfield development area as an example, the introduction is as follows:

[0086] S1. In this work area, a dynamite source with a charge of 4-8 kg is used for excitation, a geophone with a natural frequency of 10 Hz is used for single-point reception, the bin size is 10 m × 20 m, the maximum offset is 3300 m. The original seismic data is subjected to automatic first arrival picking based on a machine learning method. After picking, quality control is carried out to select the offset range with excellent picking effect for the next step of static correction amount calculation. If the picking accuracy rate is less than 98%, the learning samples are readjusted and the automatic picking is repeated until the machine learning picking accuracy rate is greater than 98%. Zero filling is performed on the upper part of the first arrival.

[0087] S2. Calculate the tomographic static correction amount using the first arrival information obtained in S1. After applying the static correction amount, perform quality control. If the continuity of the main in-phase axis is significantly improved, the quality control passes. If the continuity of the main in-phase axis deteriorates, check whether the static correction amount needs to be multiplied by (-1) and then applied until the quality control passes. When the static correction amount is applied for the first time, if the in-phase axis with a high signal-to-noise ratio shows a deterioration in continuity, the static correction amount can be reversed in sign and then applied.

[0088] S3. Perform gather extraction on the seismic data after applying the static correction generated in S2, and complete the regularization processing of the radial five-dimensional data using the matching pursuit method. In this step, first construct the theoretical acquisition system. The original acquisition bin in the Bayanchagan work area is 10m×20m. In the processing of this step, the bin is constructed as 10m×10m, which makes important preparations for the subsequent sorting of the ovt gathers after radial regularization, and a seismic data volume with uniform distribution in each azimuth sector can be obtained.

[0089] S4. Perform Fourier transform on the seismic data after regularizing the five-dimensional data obtained in S3.

[0090] S5. Predict the coherent noise in the fk domain for the data generated in S4. On the fk spectrum, the plane that does not contain effective signals can be distinguished. Through dip domain measurement and frequency division, the identification and stripping of coherent noise in different frequency ranges are realized.

[0091] S6. Perform surface wave analysis and surface wave dispersion prediction on the data generated in S2 in the geophone line domain. Through surface wave fk spectrum analysis, the characteristics of surface waves and dispersion are identified, and effective prediction of surface waves and surface wave dispersion is realized.

[0092] S7. Reconstruct and extract the coherent noise obtained in S5, and reconstruct the surface waves and surface wave dispersion predicted in S6 to obtain the reconstructed coherent noise and surface wave domain dispersion.

[0093] S8. Perform stacking processing on the two types of reconstructed noise obtained in S7. The stacking velocity comes from velocity analysis. The same velocity must be applied for all stackings and quality controls from S8 to S15.

[0094] S9. Perform spectral analysis on the noise stacking profile generated in S8 to determine the dominant frequency band range of the noise with the target layer as the window.

[0095] S10. Perform filtering processing on the original seismic data after S2 static correction in the same frequency band range as the dominant frequency band range of the noise stacking in S9.

[0096] S11. Perform gather extraction and stacking processing on the filtered data obtained in S10.

[0097] In the noise superposition profile obtained by quality control in S12, check whether there is an effective reflection signal with a similar structural feature to the original signal in the same frequency band as in S11. If there is an effective signal, proceed to step S13; if there is no effective signal, proceed to step S14.

[0098] S13. Repeat steps S5 and S6 to reduce the noise prediction degree control parameter, and repeat the steps until S12.

[0099] S14. Subtract the noise data that has passed the quality control in S12 from the original data through adaptive subtraction. The adaptive subtraction in this step is not the subtraction of the superposition profile, but the pre-stack original data after static correction and the noise predicted in S5 and S6.

[0100] S15. After performing dynamic correction and stacking on the noise suppression result of the adaptive subtraction obtained in S14, check whether the signal-to-noise ratio of the target layer meets the geological requirements and exploration design. If the signal-to-noise ratio meets the geological requirements, output the data in S14 for subsequent processing; if the quality control fails, repeat steps S5 - S14, appropriately increase the noise suppression degree, and achieve the maximum noise suppression on the basis of ensuring data fidelity, thereby improving the signal-to-noise ratio of the basic data for development under machine learning assistance based on geological model quality control in this work area.

[0101] Although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that: they can still modify the technical solutions recorded in the foregoing embodiments, or perform equivalent replacements on some of the technical features; and these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for improving the signal-to-noise ratio of ultra-shallow development data, characterized in that, The method includes: Performing surface wave analysis and surface wave dispersion prediction on seismic data to obtain surface wave noise, and performing coherent noise prediction to obtain coherent noise; After separately reconstructing the coherent noise and the surface wave noise, stacking them at the same velocity to obtain a noise stacked section, and performing spectral analysis to obtain the noise dominant frequency band range; Filtering the seismic data within the noise dominant frequency band range; and performing gather extraction and stacking processing on the filtered data to obtain a processed data volume; In the processed data volume filtered within the noise frequency band range, by analyzing to determine the seismic response characteristics of the effective reflection signal, and quality controlling whether the same seismic response characteristics exist in the noise stacked section. If the same response characteristics do not exist, in the full-band seismic processing results, removing the coherent noise and the surface wave noise through the method of adaptive subtraction to obtain a denoised data volume.

2. A method for improving the signal-to-noise ratio of ultra-shallow development data according to claim 1, characterized in that For seismic data, by performing automatic first arrival picking on the original seismic data based on a machine learning method, after picking, quality controlling and optimizing the offset range with high-quality picking effect for static correction amount calculation. If the picking accuracy rate is less than the threshold, readjusting the learning samples and repeating the automatic picking until the accuracy rate is greater than the threshold, and filling zeros above the first arrival.

3. A method for improving the signal-to-noise ratio of ultra-shallow development data according to claim 2, characterized in that The seismic data further includes: Calculating the tomographic static correction amount for the first arrival information, and performing quality control after applying the static correction amount to obtain the seismic data after static correction application.

4. A method for improving the signal-to-noise ratio of ultra-shallow development data according to claim 3, characterized in that Performing surface wave analysis and surface wave dispersion prediction on seismic data to obtain surface wave noise, including: Performing surface wave analysis and surface wave dispersion prediction on the seismic data after static correction application to obtain surface wave noise.

5. A method for improving the signal-to-noise ratio of ultra-shallow development data according to claim 3, characterized in that Performing coherent noise prediction to obtain coherent noise, including: Performing gather extraction on the seismic data after static correction application, and applying the matching pursuit method to complete the regularization processing of the radial five-dimensional data to obtain a seismic data volume with uniform distribution in each azimuth sector; Performing Fourier transform on the seismic data volume and performing coherent noise prediction in the fk domain to obtain coherent noise.

6. A method for improving the signal-to-noise ratio of ultra-shallow development data according to claim 1, characterized in that After separately reconstructing the coherent noise and the surface wave noise, stacking them at the same velocity to obtain a noise stacked section, and performing spectral analysis to obtain the noise dominant frequency band range, including: After separately reconstructing the coherent noise and the surface wave noise, stacking them at the same velocity to obtain a noise stacked section; Performing spectral analysis on the noise stacked section to obtain the noise dominant frequency band range with the target layer as the window.

7. A method for improving the signal-to-noise ratio of ultra-shallow development data according to claim 1, characterized in that It further includes: When the same seismic response characteristics exist in the noise stacked section, reducing the noise prediction degree control parameter and re-predicting the surface wave noise and the coherent noise.

8. A device for improving the signal-to-noise ratio of ultra-shallow development data, characterized in that Including: A noise prediction unit, a noise dominant frequency band calculation unit, and a denoising unit; The noise prediction unit is configured to perform surface wave analysis and surface wave dispersion prediction on seismic data to obtain surface wave noise, and perform coherent noise prediction to obtain coherent noise; The noise dominant frequency band calculation unit is configured to reconstruct the coherent noise and the surface wave noise respectively and then stack them at the same velocity to obtain a noise stacked section, and perform spectral analysis to obtain the noise dominant frequency band range; The denoising unit is configured to perform filtering processing on the seismic data within the noise dominant frequency band range; and perform trace gather and stacking processing on the filtered data to obtain a processed data volume; The denoising unit is further configured to, in the processed data volume filtered within the noise frequency band range, determine the seismic response characteristics of the effective reflection signals by analysis, and quality control whether there are the same seismic response characteristics in the noise stacked section. If there are no same response characteristics, in the full-frequency band seismic processing results, the coherent noise and the surface wave noise are removed by an adaptive subtraction method to obtain a denoised data volume.

9. An electronic device, characterized in that, It includes a processor, a communication interface, a memory, and a communication bus. Among them, the processor, the communication interface, and the memory complete mutual communication through the communication bus; The memory stores a computer program; The processor is configured to, when executing the computer program stored on the memory, implement the method for improving the signal-to-noise ratio of ultra-shallow development data described in any one of claims 1-7.

10. A computer-readable storage medium storing a computer program, characterized in that, The computer program, when executed by the processor, implements the method for improving the signal-to-noise ratio of ultra-shallow development data described in any one of claims 1-7.

Images