Earthquake data fidelity denoising method and device based on effective signal extraction

Abstract

The invention provides a seismic data fidelity denoising method and device based on effective signal extraction. By analyzing seismic conditions of a complex earth surface work area, researching noise types and distribution, utilizing the principle that leveling conditions after effective fluctuation correction are obviously different from noise, and extracting effective signals by adopting an inter-channel subtraction method, the problem that the effective signals are seriously damaged is solved, and the effect of fidelity denoising is finally achieved. According to the method, the defects of a single method in de-noising can be overcome, the problem that a large number of effective signals exist in noise after de-noising of a conventional single method but cannot be solved is solved, and the resolution and fidelity of seismic data are effectively improved. Therefore, a clear seismic section is obtained, the signal-to-noise ratio is improved, and high resolution, high fidelity and high signal-to-noise ratio are realized.

Description

Technical Field

[0001] The present invention relates to the field of seismic data processing in the field of earth sciences, and more specifically, to a seismic data fidelity denoising method and apparatus based on effective signal extraction. Background Art

[0002] With the continuous innovation and in-depth development of oil exploration technologies, exploration targets are becoming increasingly complex, exploration difficulties are increasing, and the requirements for seismic data fidelity are getting higher and higher. How to effectively suppress seismic data noise in complex areas is an urgent problem to be solved. Affected by complex surface conditions and acquisition factors, it is extremely easy to cause a phenomenon where a large amount of effective signals exist in the noise after denoising with ordinary processes but cannot be solved, and the fidelity processing is difficult. Therefore, how to suppress noise while effectively improving data fidelity has become the top priority in the research of denoising methods for complex surface work areas.

[0003] Currently, for seismic data with weak seismic emission signals, methods for extracting weak signals therefrom include: tomography method, high-order statistic method, wavelet transform method, singular value decomposition (SVD) method, pre-stack adaptive F-X domain coherent noise attenuation technology, random noise suppression method based on F-X domain prediction denoising technology, etc.

[0004] The tomography method uses the surface correlation of reflected waves and parameterizes the formation using migration and ray parameters, concentrating the signal energy on a few pixels to achieve separation and suppression of ground interference waves. This method is mainly used to reduce the intensity of surface scattered waves and protect formation reflected waves, but has poor suppression effect on shallow layers.

[0005] The high-order statistic method uses the difference between the high-order statistics of signal waves and noise waves, and uses a metric standard and threshold judgment to distinguish signal waves and noise waves. Commonly used high-order statistics include skewness, peak factor, average amplitude, etc. This type of method has good effect in extracting seismic random noise, but has poor effect in extracting directional noise.

[0006] The wavelet transform method uses wavelet transform to perform multi-resolution representation of signals, which can effectively highlight useful information and suppress noise. By using different mother wavelets and determining appropriate decomposition levels, noise can be removed while well maintaining signal characteristics. However, the degree of denoising of this method is closely related to the selection of basis functions.

[0007] The main idea of applying the singular value decomposition (SVD) method to signal processing is that, in a certain sense, several largest singular values of a matrix mainly reflect the main characteristics of the matrix and correspond to the main components of the signal; while smaller singular values and corresponding singular vectors mainly correspond to the noise part. Therefore, by intercepting the main singular values, the purpose of denoising can be achieved. However, when the signal-to-noise ratio is low, the effect of this method will be significantly reduced.

[0008] The pre-stack adaptive F-X domain coherent noise attenuation technology realizes noise reduction by predicting the correlation between data samples in the migration domain, constructing an autoregressive model, and estimating the signal under certain continuity constraints. This method has a good suppression effect on periodic noise, but it requires sacrificing a certain signal bandwidth.

[0009] The F-X domain prediction denoising technology is an efficient denoising algorithm. This technology uses the local coherence of data in the migration domain to construct an autoregressive model, estimates the model coefficients by the least squares method, predicts the signal, and thus eliminates noise. However, the selection of the prediction length will affect the authenticity of the signal waveform.

[0010] In addition, the T-p domain denoising method, the K-L transform denoising method, and the wavelet domain Wiener filtering method are also often used to extract the effective signal in seismic data. These methods can all suppress noise to a certain extent and extract the effective wave components. However, using only one method often cannot achieve ideal results, and many researchers have extracted weak seismic signals based on the combination of different methods and achieved certain results.

[0011] For example, Research One uses the method of combining wavelet transform and singular spectrum analysis. First, denoise the seismic record through wavelet decomposition; then, use the singular spectrum analysis method to extract the effective signal from the denoised record. This method makes full use of the protection of signal details by wavelet transform and the effective signal extraction ability of the singular spectrum analysis method and has achieved good results.

[0012] Research Two proposes an improved singular value decomposition method. This method introduces the concept of threshold, makes the selection of the noise subspace more reasonable, and thus improves the adaptability and anti-noise interference ability of the singular value decomposition method. The test results show that this method can effectively improve the signal-to-noise ratio and enhance the effective signal.

[0013] Research Three uses a denoising method based on the combination of prediction migration and wavelet transform. This method introduces the prediction-based denoising technology into the wavelet domain, fully combines the advantages of wavelet analysis in maintaining signal detail features and the prediction method in suppressing noise, and has achieved relatively ideal results.

[0014] Research Four proposes a wavelet domain non-local denoising method. This method uses wavelet transform to obtain a multi-resolution representation, uses the non-local mean algorithm to eliminate noise, and then performs reconstruction. This method does not rely on any noise model and can effectively eliminate noise while retaining complex waveforms by aggregating similar sample information.

[0015] The above research has improved the signal-to-noise ratio to a certain extent and extracted the effective signals. However, in complex surface areas, the difference between deep effective weak signals and noise interference is small and difficult to identify, so these methods are subject to certain limitations. How to further effectively identify and protect these weak signals in the data of complex areas is a technical problem worthy of exploration. Summary of the Invention

[0016] In view of this, the present invention discloses a seismic signal denoising scheme, which can solve the problem of serious damage to effective signals and strive to finally achieve the effect of signal fidelity and noise reduction.

[0017] According to one aspect of the present invention, a method for seismic data fidelity denoising based on effective signal extraction is proposed. The method includes:

[0018] Step 1, respectively record the shot gather data before and after seismic signal denoising as A1 and A2, and output the subtraction result A3 according to the inter-trace subtraction operation;

[0019] Step 2, perform NMO flattening processing on the subtraction result A3 and output the NMO-processed data A4;

[0020] Step 3, perform multi-trace stacking processing on the NMO-processed data A4, output the result A5 after multiple stackings, and use A5 as the extracted effective signal;

[0021] Step 4, add the effective signal A5 back to the shot gather data A2 after seismic data denoising through inter-trace operation to obtain the signal A6 after fidelity denoising.

[0022] In some embodiments, step 2 specifically includes:

[0023] Set the length of the NMO window;

[0024] Within this NMO window, select the best translation amount using the criterion of energy maximization to achieve data fitting of the subtraction result A3 on the time axis;

[0025] Slide the NMO window and repeat the above operations until the entire time axis is scanned;

[0026] Finally, output the NMO-processed data A4.

[0027] In some embodiments, step 3 specifically includes:

[0028] Step 31, sort the NMO-processed data A4 in the order of trace numbers, and group every n traces as a group. Adjacent n traces are divided into a trace group, where n is a set value;

[0029] Step 32, within each trace group, perform stacking summation on the data of each trace;

[0030] Step 33: Connect the results superimposed within different trace groups to obtain the result of this superimposition.

[0031] Step 34: For the result of the superimposition, re-divide the trace groups, with each group including n adjacent traces. Repeat Step 32 and Step 33 until the number of superimposition times reaches the preset number of superimposition times, and output the result after multiple superimpositions, which is the effective signal A5.

[0032] In some embodiments, the method further includes:

[0033] Within the limited apparent velocity range and limited frequency range, perform surface wave suppression on the shot gather data A1 before denoising to obtain the denoised shot gather data A2.

[0034] According to one aspect of the present invention, there is also provided a seismic data fidelity denoising device based on effective signal extraction. The device includes:

[0035] An inter-trace subtraction unit, configured to respectively denote the shot gather data before and after seismic signal denoising as A1 and A2, and output the subtraction result A3 according to the inter-trace subtraction operation.

[0036] A NMO flattening unit, configured to perform NMO flattening processing on the subtraction result A3 and output the data A4 after NMO.

[0037] A superimposition unit, configured to perform multi-trace superimposition processing on the data A4 after NMO, output the result A5 after multiple superimpositions, and use A5 as the extracted effective signal.

[0038] An inter-trace addition unit, configured to add the effective signal A5 back to the shot gather data A2 after seismic data denoising through inter-trace operation to obtain the signal A6 after fidelity denoising.

[0039] In some embodiments, the NMO flattening unit is specifically configured to:

[0040] Set the length of the NMO window.

[0041] Within this NMO window, select the optimal translation amount according to the criterion of energy maximization to achieve data fitting of the subtraction result A3 on the time axis.

[0042] Slide the NMO window and repeat the above operations until the entire time axis is scanned.

[0043] Finally, output the data A4 after NMO.

[0044] In some embodiments, the superimposition unit is specifically configured to:

[0045] Step 31: Sort the data A4 after NMO in the order of trace numbers, group every n traces as a group, and adjacent n traces are divided into a trace group, where n is a set value.

[0046] Step 32: Within each trace group, sum up the data on each trace.

[0047] Step 33: Connect the results of stacking within different trace groups to obtain the result of this stacking.

[0048] Step 34: For the stacked result, re-divide the trace groups, with each group including n adjacent traces, and repeat Step 32 and Step 33 until the stacking times reach the preset stacking times, and output the result after multiple stackings, which is the effective signal A5.

[0049] In some embodiments, the device further includes:

[0050] A surface wave suppression unit, configured to perform surface wave suppression on the pre-denoised shot gather data A1 within a limited apparent velocity range and a limited frequency range to obtain the denoised shot gather data A2.

[0051] According to another aspect of the present invention, an electronic device is further provided, and the electronic device includes:

[0052] A memory storing executable instructions;

[0053] A processor that runs the executable instructions in the memory to implement the seismic data fidelity denoising method based on effective signal extraction described above.

[0054] According to another aspect of the present invention, a computer-readable storage medium is further provided, and the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, it implements the seismic data fidelity denoising method based on effective signal extraction described above.

[0055] The present invention is a fidelity denoising method based on effective signal extraction. By analyzing the seismic conditions in complex surface areas and studying the noise types and distributions, using the principle that the situation of the effective wave after calibration and flattening is significantly different from that of the noise, and then adopting the method of subtraction between traces to extract the effective signal, the problem of serious damage to the effective signal is solved, and the effect of fidelity denoising is ultimately achieved. The present invention can solve the defects of a single method in denoising, overcome the problem that a large amount of effective signals still exist in the noise after denoising by a conventional single method and cannot be solved, and effectively improve the resolution and fidelity of seismic data. Thus, a clear seismic profile is obtained and the signal-to-noise ratio is improved, realizing high resolution, high fidelity, and high signal-to-noise ratio.

[0056] The beneficial effects of the present invention are further analyzed in detail below.

[0057] 1) Effectively extract and protect weak signals

[0058] Through dynamic calibration leveling and multi-channel stacking processing, the present invention can effectively extract and protect weak signals directly from a complex noise environment. This avoids the additional operations in other methods for protecting weak signals, simplifies the process, and reduces the risk of weak signal loss.

[0059] 2) Improve the signal-to-noise ratio

[0060] In the present invention, multi-channel stacking processing can significantly increase the amplitude of the effective signal, thereby increasing the ratio of the signal to the noise, and effectively improving the signal-to-noise ratio.

[0061] 3) Ensure the fidelity of the signal waveform

[0062] After the effective signal extracted by the present invention undergoes dynamic calibration leveling and stacking gain, it can fully retain the detailed information of the original signal, with high waveform fidelity. This provides a reliable basis for subsequent precise data analysis and interpretation.

[0063] 4) Simple operation and high computational efficiency

[0064] The core steps in the present invention, such as dynamic calibration leveling and multi-channel stacking, only rely on simple data rearrangement and stacking operations, without complex waveform analysis and modeling, with a small amount of calculation and high computational efficiency. This makes it easy to implement the technology in engineering applications.

[0065] 5) Independent of the noise model and strong adaptability

[0066] The extraction and discrimination of the effective signal in the present invention do not rely on modeling the noise characteristics, so there are no specific restrictions on the types and distributions of the noise present in the data, and it has strong adaptability. This is especially effective for dealing with complex and variable on-site noise.

[0067] 6) Novel technical route

[0068] There is currently no report on using the combination of dynamic calibration leveling and multi-channel stacking for fidelity denoising. This research provides a new technical route idea, which has certain innovative significance for promoting the development of technologies in this field.

[0069] The methods and devices of the present invention have other characteristics and advantages, which will be obvious in the accompanying drawings and subsequent specific embodiments incorporated herein, or will be described in detail in the accompanying drawings and subsequent specific embodiments incorporated herein. These accompanying drawings and specific embodiments are used together to explain the specific principles of the present invention. Brief Description of the Drawings

[0070] By describing the exemplary embodiments of the present invention in more detail in conjunction with the accompanying drawings, the above and other objects, features, and advantages of the present invention will become more obvious. Among them, in the exemplary embodiments of the present invention, the same reference numerals generally represent the same components.

[0071] Figure 1 Shows a flowchart of a seismic data fidelity denoising method based on effective signal extraction according to an embodiment of the present invention.

[0072] Figure 2 Shows the single-shot octave sweep map of a certain actual data work area.

[0073] Figure 3 (a), (b) and (c) respectively show the single-shot data before denoising, the single-shot data after denoising and the single-shot noise data according to an embodiment of the present invention.

[0074] Figure 4 (a), (b) and (c) respectively show the single-shot noise data, the effective signal data extracted from the noise and the pure noise data according to an embodiment of the present invention.

[0075] Figures 5(a), (b) and (c) respectively show the stacked section before denoising of the actual seismic data, the stacked section after denoising and the residual section according to an embodiment of the present invention.

[0076] Figures 6(a), (b) and (c) respectively show the stacked section before denoising of the actual seismic data, the stacked section after the effective signal extraction and restoration and the residual section according to an embodiment of the present invention. Detailed implementation manners

[0077] The preferred embodiments of the present invention will be described in more detail below with reference to the accompanying drawings. Although the preferred embodiments of the present invention are shown in the drawings, it should be understood that the present invention can be implemented in various forms and should not be limited by the embodiments set forth herein. On the contrary, these embodiments are provided so that the present invention will be more thorough and complete, and will fully convey the scope of the present invention to those skilled in the art.

[0078] Example 1

[0079] Figure 1 Shows a flowchart of a seismic data fidelity denoising method based on effective signal extraction according to an embodiment of the present invention. As shown in the figure, the method includes steps 1 to 4.

[0080] Step 1: Denote the gather data before and after denoising the seismic signal as A1 and A2 respectively, and output the subtraction result A3 according to the inter-trace subtraction operation.

[0081] The following are some common methods for obtaining the gather data A1 before denoising:

[0082] 1) Surface acquisition

[0083] Using a seismic acquisition vehicle group or a portable seismograph group, receiving points and source points are arranged in the field according to a preset seismic profile, and artificial excitation is carried out to obtain the original record. The data obtained by this method has a high degree of authenticity, but the efficiency is relatively low.

[0084] 2) Box Burial Acquisition

[0085] In the area to be measured, a small seismic data acquisition device is used, and wireless technology is adopted for detection. This detection instrument is so small that it can be buried underground, thereby reducing the influence of environmental noise and obtaining high-quality original records. However, the cost is relatively high.

[0086] 3) Downhole Acquisition

[0087] High-quality reflection wave records of the underlying formation and the target formation are obtained by using geophones or seismic while-drilling tools in vertical or inclined shafts. This method can obtain data with a high SNR, but the sampling range is limited.

[0088] 4) Seabed Acquisition

[0089] Mobile seismic recorders are arranged on the seabed in the sea area, and the seabed formation is excited by a ship to obtain its reflection record. This method can efficiently obtain data, but it is greatly affected by sea conditions.

[0090] 5) Reacquisition

[0091] The archived historical seismic data is reprocessed and quality-improved by applying new technologies to obtain the quality-improved seismic profile as the data before denoising. This reacquisition method has a low cost and a fast speed, but it is difficult to greatly improve the quality.

[0092] In practical applications, according to the cost, the position of the target formation, and the expected quality requirements, etc., one or several means are selected to obtain the original pre-denoising shot gather data.

[0093] In some embodiments, within a limited apparent velocity range and a limited frequency range, surface wave suppression is performed on the pre-denoising shot gather data A1 to obtain the post-denoising shot gather data A2.

[0094] In some examples, the specific implementation steps for performing surface wave suppression on the pre-denoising shot gather data A1 are as follows:

[0095] 1) Perform time-frequency analysis on the pre-denoising shot gather data A1 to determine the main surface wave frequency range and the main distribution range of surface waves in the time-distance domain;

[0096] 2) According to the time-frequency analysis results, set the apparent velocity range limit and the frequency range limit;

[0097] 3) Under the above-mentioned apparent velocity and frequency range limitation conditions, use band-stop filtering technology to filter out surface wave noise in the shot gather data A1. Common filtering methods include FK filtering, time-distance filtering, etc.;

[0098] 4) Further eliminate the residual interface waves introduced by the filtering operation through the discontinuous correlation technology;

[0099] 5) Use the signal coherent superposition technology to enhance the effective signal amplitude of the signal after surface wave suppression;

[0100] Thus, output the shot gather data A2 with the main surface wave noise components suppressed.

[0101] Through surface wave suppression, the surface wave noise in the complex environment can be initially eliminated.

[0102] In some examples, the specific steps of inter-trace subtraction are as follows:

[0103] 1) Perform profile geometric coordination on the shot gather data A1 and A2 before and after denoising to ensure that the two sets of data are strictly aligned in time and space coordinates;

[0104] 2) Read the data of A1 and A2 channel by channel respectively, and match and arrange the two data sets in the order of channel numbers;

[0105] 3) For the two channels of data with the same serial number, perform simple numerical subtraction on each time sample point. That is, for the time sample point t, calculate A3(t) = A1(t) - A2(t);

[0106] 4) Connect the subtraction results of all channels to obtain the final overall subtraction profile A3.

[0107] It is necessary to ensure that the two sets of data participating in the subtraction are strictly aligned in the geographic coordinate system, that is, the channels with the same serial number correspond to the same location in the actual acquisition. If the coordinates of the two sets of input data are inconsistent, the above-mentioned coordination adjustment is also required to meet the basic requirements of inter-trace subtraction, so as to ensure that the difference result A3 can truly reflect the distribution of noise.

[0108] Step 2, perform dynamic correction and flattening processing on the subtraction result A3, and output the data A4 after dynamic correction.

[0109] Dynamic correction is a means in seismic data processing, mainly used to improve the coherence of data between adjacent channels and provide a basis for subsequent multi-channel stacking.

[0110] Adjacent seismic wave channels theoretically have highly similar signals because they depict the continuous variation of the same formation interface. However, due to surface conditions, instrument errors, etc., the energy maximum points of adjacent channels do not completely align, and there are small time deviations. Dynamic correction flattening can adjust the time shift slightly to align the signal waveforms of adjacent channels as much as possible, thereby improving coherence and providing a basis for subsequent noise suppression. Noise, due to its lack of consistency, is difficult to produce enhanced superposition after dynamic correction flattening and may instead produce an amplitude reduction and cancellation effect due to decoherence.

[0111] In short, dynamic correction flattening can achieve better coherent superposition of effective wave signals on adjacent channels through slight time offset alignment, while noise does not have this effect.

[0112] The inventor utilizes the principle that the situation after flattening the effective wave by dynamic correction is significantly different from that of noise, and performs dynamic correction flattening on the result of channel subtraction. Through dynamic correction flattening, the quality of coherent superposition between adjacent channels can be significantly improved, effectively highlighting the effective signal and providing a basis for subsequent extraction of the effective signal.

[0113] In some embodiments, dynamic correction flattening can be achieved in the following ways:

[0114] Set the length of the dynamic correction window;

[0115] Within this dynamic correction window, select the best translation amount using the criterion of energy maximization to achieve data fitting of the subtraction result A3 on the time axis;

[0116] Slide the dynamic correction window and repeat the above operations until the entire time axis is scanned;

[0117] Finally, output the data A4 after dynamic correction.

[0118] Compared with random noise, the overall waveform of the effective seismic signal is more consistent after channel subtraction. Therefore, using dynamic correction flattening can make the effective waves superpose well subsequently, while it is difficult for noise to achieve such consistency.

[0119] Through dynamic correction flattening processing, the amplitude of the effective signal relative to noise can be significantly increased, so as to better extract the effective signal from the mixed signal. This not only ensures the signal fidelity but also lays a foundation for subsequent steps.

[0120] Step 3, perform multi-channel superposition processing on the data A4 after dynamic correction, output the result A5 after multiple superpositions, and use A5 as the extracted effective signal.

[0121] In some embodiments, the multi-channel superposition processing of the data A4 after dynamic correction specifically includes:

[0122] Step 31: Sort the data A4 after dynamic correction in the order of trace numbers, and group every n traces as a group. Adjacent n traces are divided into a trace group, where n is a set value;

[0123] Step 32: Within each trace group, sum up the data on each trace;

[0124] Step 33: Connect the results of stacking within different trace groups to obtain the result of this stacking;

[0125] Step 34: For the result of stacking, re-divide the trace groups, with each group including adjacent n traces. Repeat Step 32 and Step 33 until the stacking times reach the preset stacking times, and output the result after multiple stackings, which is the effective signal A5.

[0126] Compared with the usual stacking, the advantage of multi-trace stacking is that signals on different traces can be mutually stacked to enhance the effective components and weaken the random noise. Both n and the preset stacking times are adjustable parameters, and in practice, they can be specifically set according to the noise situation of the data to achieve the best denoising gain. For example, n can be taken as 5 - 10, such as 9; the preset stacking times can be 3 - 8 times, such as 5 times.

[0127] Step 4: Add the effective signal A5 back to the shot gather data A2 after seismic data denoising through inter-trace operations to obtain the signal A6 after fidelity denoising.

[0128] In some embodiments, the effective signal A5 can be added back to A2 through the following method:

[0129] Perform strict geometric coordination on the denoised shot gather data A2 and the effective signal A5 to ensure that the two sets of data are accurately corresponding in the time and space coordinate systems;

[0130] Extract the traces in A2 and A5 for matching one by one according to the trace numbers;

[0131] For each pair of matching traces, directly stack the data in A5 onto the data of the A2 trace sample by sample, that is:

[0132] A6(x, y, t) = A2(x, y, t) + A5(x, y, t)

[0133] where x and y respectively represent the spatial dimensions in the coordinate system, and t is the time axis sample point;

[0134] Connect all the stacked trace data A6 in the order of trace numbers to generate the final summation result data volume A6.

[0135] Through simple direct addition between traces, signal duplication and possible losses are effectively avoided, while the computational load is small and the computational efficiency is high. In the finally output data A6, the effective signal is enhanced, achieving the purpose of signal fidelity and noise reduction.

[0136] Similar to the trace subtraction in Step 1, special attention should be paid to the coordination synchronization of the two sets of input data during trace addition, otherwise accurate and reliable addition output cannot be obtained.

[0137] This embodiment is a seismic data fidelity and noise reduction method based on effective signal extraction. By analyzing the seismic conditions in complex surface areas and studying the noise types and distributions, using the principle that the flattened situation of the effective wave after correction is significantly different from the noise, and then adopting the trace subtraction method to extract the effective signal, the problem of serious damage to the effective signal is solved, and the effect of signal fidelity and noise reduction is finally achieved. The present invention can solve the defects of a single method in noise reduction, overcome the problem that a large amount of effective signals still exist in the noise after denoising by conventional single methods and cannot be solved, and effectively improve the resolution and fidelity of seismic data. Thus, a clear seismic profile is obtained and the signal-to-noise ratio is improved, achieving high resolution, high fidelity, and high signal-to-noise ratio.

[0138] Example 2

[0139] According to an embodiment of the present invention, a seismic data fidelity and noise reduction device based on effective signal extraction is provided. The device includes:

[0140] A trace subtraction unit, configured to record the seismic signal gather data before and after noise reduction as A1 and A2 respectively, and output a subtraction result A3 according to the trace subtraction operation;

[0141] A NMO flattening unit, configured to perform NMO flattening processing on the subtraction result A3 and output the data A4 after NMO;

[0142] An overlay unit, configured to perform multi-trace overlay processing on the data A4 after NMO, output the result A5 after multiple overlays, and use A5 as the extracted effective signal;

[0143] A trace addition unit, configured to add the effective signal A5 back to the seismic signal gather data A2 after noise reduction through trace operation to obtain the signal A6 after fidelity and noise reduction.

[0144] In some embodiments, the NMO flattening unit is specifically configured to:

[0145] Set the length of the NMO window;

[0146] Within this NMO window, select the best translation amount according to the criterion of energy maximization to achieve data fitting of the subtraction result A3 on the time axis;

[0147] Slide the dynamic correction window and repeat the above operations until the entire timeline is scanned;

[0148] Finally, output the data A4 after dynamic correction.

[0149] In some embodiments, the superimposing unit is specifically configured to:

[0150] Step 31: Sort the data A4 after dynamic correction in the order of trace numbers, and group every n traces into a group. Adjacent n traces are divided into a trace group, where n is a set value;

[0151] Step 32: Within each trace group, sum up the data on each trace;

[0152] Step 33: Connect the results of superposition in different trace groups to obtain the result of this superposition;

[0153] Step 34: For the result of superposition, re-divide the trace groups, with each group including adjacent n traces. Repeat Step 32 and Step 33 until the number of superposition times reaches the preset number of superposition times, and output the result after multiple superpositions, which is the effective signal A5.

[0154] In some embodiments, the device further includes:

[0155] A surface wave suppression unit, configured to perform surface wave suppression on the pre-denoising shot gather data A1 within a limited apparent velocity range and a limited frequency range to obtain the post-denoising shot gather data A2.

[0156] This embodiment is a seismic data fidelity denoising device based on effective signal extraction. By analyzing the seismic conditions in complex surface work areas and studying the noise types and distributions, using the principle that the flattened situation of effective wave after dynamic correction is significantly different from that of noise, and then adopting the method of subtraction between traces to extract effective signals, it solves the problem of serious damage to effective signals, and strives to finally achieve the effect of fidelity denoising. The present invention can solve the defects of a single method in denoising, overcome the problem that a large amount of effective signals still exist in the noise after denoising by conventional single methods and cannot be solved, and effectively improve the resolution and fidelity of seismic data. Thus, a clear seismic profile is obtained and the signal-to-noise ratio is improved, achieving high resolution, high fidelity, and high signal-to-noise ratio.

[0157] For other detailed descriptions and advantages of this embodiment, reference can be made to the corresponding descriptions in the foregoing embodiments, and details are not repeated herein.

[0158] Example 3

[0159] According to another aspect of the present invention, an electronic device is further provided. The electronic device includes:

[0160] A memory storing executable instructions:

[0161] A processor that runs the executable instructions in the memory to implement the seismic data denoising method based on mathematical morphology according to the present invention.

[0162] The method includes the following steps:

[0163] Step 1, denote the gather data before and after seismic signal denoising as A1 and A2 respectively, and output the subtraction result A3 according to the inter-trace subtraction operation;

[0164] Step 2, perform NMO flattening processing on the subtraction result A3 and output the NMO-corrected data A4;

[0165] Step 3, perform multi-trace stacking processing on the NMO-corrected data A4, output the result A5 after multiple stackings, and take A5 as the extracted effective signal;

[0166] Step 4, add the effective signal A5 back to the gather data A2 after seismic data denoising through inter-trace operation to obtain the signal A6 after fidelity denoising.

[0167] In some embodiments, step 2 specifically includes:

[0168] Set the length of the NMO window;

[0169] Within this NMO window, select the optimal translation amount using the criterion of energy maximization to achieve data fitting of the subtraction result A3 on the time axis;

[0170] Slide the NMO window and repeat the above operations until the entire time axis is scanned;

[0171] Finally, output the NMO-corrected data A4.

[0172] In some embodiments, step 3 specifically includes:

[0173] Step 31, sort the NMO-corrected data A4 in the order of trace numbers, group every n traces as a group, and adjacent n traces are divided into a trace group, where n is a set value;

[0174] Step 32, within each trace group, perform stacking summation on the data of each trace;

[0175] Step 33, connect the stacking results within different trace groups to obtain the result of this stacking;

[0176] Step 34, for the stacking result, re-divide the trace groups, each group includes adjacent n traces, and repeat steps 32 and 33 until the stacking times reach the preset stacking times, and output the result after multiple stackings, which is the effective signal A5.

[0177] In some embodiments, the method further includes:

[0178] Within a limited apparent velocity range and a limited frequency range, surface wave suppression is performed on the gather data A1 before denoising to obtain the gather data A2 after denoising.

[0179] This embodiment is a seismic data fidelity denoising scheme based on effective signal extraction. By analyzing the seismic conditions in complex surface areas and studying the noise types and distributions, using the principle that the flattened situation of the effective wave after correction is significantly different from that of the noise, and then adopting the method of subtraction between traces to extract the effective signal, the problem of serious damage to the effective signal is solved, and the effect of fidelity denoising is ultimately achieved. The present invention can solve the defects of a single method in denoising, overcome the problem that a large amount of effective signals exist in the noise after denoising by a conventional single method but cannot be solved, and effectively improve the resolution and fidelity of seismic data. Thus, a clear seismic profile is obtained and the signal-to-noise ratio is improved, achieving high resolution, high fidelity, and high signal-to-noise ratio.

[0180] For the detailed description of this embodiment, reference can be made to the corresponding descriptions in the foregoing embodiments, and details will not be repeated here.

[0181] Example 4

[0182] According to another aspect of the present invention, there is also provided a computer-readable storage medium storing a computer program, which when executed by a processor, implements the seismic data denoising method based on mathematical morphology according to the present invention.

[0183] The method includes the following steps:

[0184] Step 1, respectively denote the gather data before and after seismic signal denoising as A1 and A2, and output the subtraction result A3 according to the subtraction operation between traces;

[0185] Step 2, perform NMO flattening processing on the subtraction result A3, and output the data A4 after NMO;

[0186] Step 3, perform multi-trace stacking processing on the data A4 after NMO, output the result A5 after multiple stackings, and use A5 as the extracted effective signal;

[0187] Step 4, add the effective signal A5 back to the gather data A2 after seismic data denoising through the operation between traces to obtain the signal A6 after fidelity denoising.

[0188] In some embodiments, the specific content of step 2 includes:

[0189] Set the length of the NMO window;

[0190] Within this NMO window, select the optimal translation amount using the criterion of energy maximization to achieve data fitting of the subtraction result A3 on the time axis;

[0191] Slide the dynamic calibration window and repeat the above operations until the entire timeline is scanned;

[0192] Finally, output the data A4 after dynamic calibration.

[0193] In some embodiments, step 3 specifically includes:

[0194] Step 31: Sort the data A4 after dynamic calibration in the order of trace numbers, and group every n traces as a group. Adjacent n traces are divided into a trace group, where n is a set value;

[0195] Step 32: Within each trace group, perform superposition summation on the data of each trace;

[0196] Step 33: Connect the superposition results of different trace groups to obtain the result of this superposition;

[0197] Step 34: For the superposition result, re-divide the trace groups, with each group including adjacent n traces. Repeat steps 32 and 33 until the number of superpositions reaches the preset number of superpositions, and output the result after multiple superpositions, which is the effective signal A5.

[0198] In some embodiments, the method further includes:

[0199] Suppress surface waves for the shot gather data A1 before denoising within the limited apparent velocity range and limited frequency range to obtain the denoised shot gather data A2.

[0200] According to the computer-readable storage medium of the embodiments of the present invention, non-temporary computer-readable instructions are stored thereon. When the non-temporary computer-readable instructions are run by a processor, all or part of the steps of the methods of the various embodiments of the present invention described above are executed.

[0201] The above-mentioned computer-readable storage medium includes but is not limited to: optical storage media (such as CD-ROM and DVD), magneto-optical storage media (such as MO), magnetic storage media (such as magnetic tapes or external hard drives), media with built-in rewritable non-volatile memories (such as memory cards), and media with built-in ROM (such as ROM cartridges).

[0202] Those skilled in the art should understand that in order to solve the technical problem of how to obtain a good user experience effect, this embodiment may also include well-known structures such as communication buses and interfaces, and these well-known structures should also be included in the protection scope of the present invention.

[0203] This embodiment is a seismic data fidelity denoising scheme based on effective signal extraction. By analyzing the seismic conditions in complex surface work areas and studying the noise types and distributions, using the principle that the flattened situation of effective fluctuations after correction is significantly different from that of noise, and then adopting the method of subtracting between traces to extract effective signals, the problem of serious damage to effective signals is solved, and the effect of fidelity denoising is finally achieved. The present invention can solve the defects of a single method in denoising, overcome the problem that a large number of effective signals still exist in the noise after denoising by a conventional single method and cannot be solved, and effectively improve the resolution and fidelity of seismic data. Thus, a clear seismic profile is obtained and the signal-to-noise ratio is increased, achieving high resolution, high fidelity, and high signal-to-noise ratio.

[0204] For the detailed description of this embodiment, reference can be made to the corresponding descriptions in the foregoing embodiments, and details will not be repeated here.

[0205] Example 5

[0206] This embodiment verifies and illustrates the effects of the present invention.

[0207] Select a set of actual seismic data with various noises developed and strong surface wave energy. Figure 2 It is the single-shot octave sweep map of the actual data work area. It can be seen from it that the frequencies of the shallow effective waves and the surface waves are similar, which increases the complexity of noise suppression.

[0208] Figure 3 (a) shows the single-shot data before denoising. According to the embodiment of the present invention, the apparent velocity range is limited to 0 - 350 m / s, and the frequency range is limited to 0 - 12 HZ, and surface wave suppression is performed on the data before denoising. Figure 3 (b) The single-shot data after denoising according to the embodiment of the present invention. Figure 3 (c) shows the single-shot data of the noise. It can be seen from Figure 3 (c) that there are still a large number of effective signals in the noise signal.

[0209] When performing effective signal extraction, the specific steps are as follows:

[0210] 1) Denote the gather data before and after seismic signal denoising as A1 (gather data before denoising) and A2 (gather data after denoising), and output the subtraction result A3 through trace-by-trace subtraction operation.

[0211] 2) Using the principle that the flattened situation of effective fluctuations after correction is significantly different from that of noise, perform dynamic correction and flattening processing on the trace-by-trace subtraction result A3, and output the data A4 after dynamic correction.

[0212] 3) Combine the data A4 after dynamic correction and perform stack on the combined traces. After several stacks, the output is denoted as A5. At this time, the effective wave in A5 is significantly stronger in energy than the noise signal after multiple stacks and can be regarded as a pure effective signal with a weak noise signal. In this example, every 9 traces of data are used as a trace group, and the number of stacks is 5.

[0213] Figure 4 (a), (b), and (c) respectively show the single-shot noise data, the effective signal data extracted from the noise, and the pure noise data according to an embodiment of the present invention. From Figure 4 the pure noise data shown in (c), it can be seen that the residual effective signal is significantly restored.

[0214] Figures 5(a), (b), and (c) respectively show the stacked section before denoising of the actual seismic data, the stacked section after denoising, and the residual section according to an embodiment of the present invention. Figures 6(a), (b), and (c) respectively show the stacked section before denoising of the actual seismic data, the stacked section after adding and restoring the extracted effective signal, and the residual section according to an embodiment of the present invention.

[0215] Comparing the stacked sections before and after denoising shown in Figures 5(a) and (b) and Figures 6(a) and (b), it can be clearly seen that after extracting the effective signal according to the present invention, the effective wave mixed in the noise is significantly restored, and the fidelity of the seismic data is significantly improved.

[0216] For other detailed descriptions of this exemplary embodiment, reference can be made to the corresponding descriptions in the foregoing embodiments, which will not be repeated here.

[0217] Combining the above embodiments, the present invention provides a seismic data fidelity denoising scheme based on effective signal extraction. Each aspect of each embodiment of the present invention has the following beneficial effects:

[0218] 1) Effectively extract and protect weak signals

[0219] Through the processing of dynamic correction flattening and multi-trace stacking, the present invention can directly extract and protect weak signals from a complex noise environment, which avoids the additional operations for protecting weak signals in other methods, simplifies the process, and also reduces the risk of weak signal loss;

[0220] 2) Improve the signal-to-noise ratio

[0221] In the present invention, the multi-trace stacking processing can significantly increase the amplitude of the effective signal, thereby increasing the ratio of the signal to the noise and effectively improving the signal-to-noise ratio;

[0222] 3) Ensure the fidelity of the signal waveform

[0223] After the effective signals extracted by the present invention are subjected to dynamic correction flattening and stacking gain, the detailed information of the original signals can be fully retained, and the waveform fidelity is high, which provides a reliable basis for subsequent precise data analysis and interpretation;

[0224] 4) Simple operation and high calculation efficiency

[0225] The core steps in the present invention, such as dynamic correction flattening and multi-channel stacking, only rely on simple data re-coordination and stacking operations, without complex waveform analysis and modeling, with small calculation amount and high operation efficiency, which makes it easy to realize engineering applications of this technology;

[0226] 5) Independent of noise model and strong adaptability

[0227] The extraction and distinction of the effective signals in the present invention do not depend on the modeling of the noise characteristics. Therefore, there are no specific restrictions on the types and distributions of the noises existing in the data, and the adaptability is strong, which is especially effective for dealing with complex and variable on-site noises;

[0228] 6) Novel technical route

[0229] There is currently no report on using the combination of dynamic correction flattening and multi-channel stacking for fidelity denoising. This research provides a new technical route idea, which has certain innovative significance for promoting the technological development in this field.

[0230] The embodiments of the present invention have been described above. The above description is exemplary, not exhaustive, and is not limited to the disclosed embodiments. Many modifications and variations are obvious to those of ordinary skill in the art in the technical field without departing from the scope and spirit of the described embodiments. The selection of the terms used herein is intended to best explain the principles of the embodiments, practical applications, or technical improvements to the technologies in the market, or to enable other ordinary skill in the art in the technical field to understand the embodiments disclosed herein.

Claims

1. A seismic data fidelity denoising method based on effective signal extraction, characterized in that, the method includes: Step 1, respectively record the shot gather data before and after seismic signal denoising as A1 and A2, and output the subtraction result A3 according to the inter-trace subtraction operation; Step 2, perform NMO flattening processing on the subtraction result A3, and output the data A4 after NMO; Step 3, perform multi-trace stacking processing on the data A4 after NMO, output the result A5 after multiple stackings, and use A5 as the extracted effective signal; Step 4, add the effective signal A5 back to the shot gather data A2 after seismic data denoising through inter-trace operation to obtain the fidelity denoised signal A6.

2. The method according to claim 1, wherein, the specific steps of Step 2 include: Set the length of the NMO window; Within this NMO window, select the best translation amount according to the criterion of maximum energy to achieve data fitting of the subtraction result A3 on the time axis; Slide the NMO window and repeat the above operations until the entire time axis is scanned; Finally, output the data A4 after NMO.

3. The method according to claim 1, characterized in that, the specific steps of Step 3 include: Step 31, sort the data A4 after NMO in the order of trace numbers, group every n traces as a group, and adjacent n traces are divided into a trace group, where n is a set value; Step 32, within each trace group, perform stacking summation on the data of each trace; Step 33, connect the results of stacking within different trace groups to obtain the result of this stacking; Step 34, for the result of stacking, re-divide the trace groups, each group includes adjacent n traces, and repeat Step 32 and Step 33 until the stacking times reach the preset stacking times, and output the result after multiple stackings, which is the effective signal A5.

4. The method according to claim 1, characterized in that, the method further includes: Within the limited apparent velocity range and limited frequency range, perform surface wave suppression on the shot gather data A1 before denoising to obtain the shot gather data A2 after denoising.

5. A seismic data fidelity denoising device based on effective signal extraction, characterized in that, the device includes: An inter-trace subtraction unit, which is used to respectively record the shot gather data before and after seismic signal denoising as A1 and A2, and output the subtraction result A3 according to the inter-trace subtraction operation; An NMO flattening unit, which is used to perform NMO flattening processing on the subtraction result A3 and output the data A4 after NMO; A stacking unit, which is used to perform multi-trace stacking processing on the data A4 after NMO, output the result A5 after multiple stackings, and use A5 as the extracted effective signal; An inter-trace addition unit, which is used to add the effective signal A5 back to the shot gather data A2 after seismic data denoising through inter-trace operation to obtain the fidelity denoised signal A6.

6. The device according to claim 5, wherein, the NMO flattening unit is specifically used for: Set the length of the NMO window; Within this NMO window, select the best translation amount according to the criterion of maximum energy to achieve data fitting of the subtraction result A3 on the time axis; Slide the NMO window and repeat the above operations until the entire time axis is scanned; Finally, output the data A4 after NMO.

7. The device according to claim 5, characterized in that, the stacking unit is specifically used for: Step 31: Sort the data A4 after dynamic correction in the order of trace numbers, and group them in sets of n traces. Adjacent n traces are divided into a trace group, where n is a set value; Step 32: Within each trace group, sum up the data on each trace; Step 33: Connect the summation results of different trace groups to obtain the result of this summation; Step 34: For the summation result, re-divide the trace groups, with each group including adjacent n traces, and repeat Step 32 and Step 33 until the number of summation times reaches the preset number of summation times, and output the result after multiple summations, which is the effective signal A5.

8. The method according to claim 5, wherein, the device further comprises: a surface wave suppression unit, configured to perform surface wave suppression on the pre-denoising shot gather data A1 within a limited apparent velocity range and a limited frequency range to obtain the post-denoising shot gather data A2.

9. An electronic device, wherein, the electronic device comprises: a memory storing executable instructions; a processor, the processor running the executable instructions in the memory to implement the method according to any one of claims 1-4.

10. A computer-readable storage medium storing a computer program, which when executed by a processor implements the method according to any one of claims 1-4.

Images