A method for prestack noise suppression of seismic data

Abstract

This invention relates to a pre-stack noise suppression method for seismic data, belonging to the field of seismic exploration technology. The method includes the following steps: 1) After decompiling and defining the observation system of the acquired seismic data, static correction, spherical diffusion compensation, and surface wave suppression are performed sequentially; 2) The first arrival time of the surface wave suppressed seismic data is corrected to a relatively uniform time, and then random noise and residual noise from the surface wave suppression process are suppressed; 3) The first arrival time of the noise-suppressed seismic data is reverse-corrected to restore the first arrival time to its state before correction. This method can effectively suppress shallow noise interference in near-offset data while better preserving the detailed structure of the seismic data, improving the accuracy of seismic processing; and avoids other noise interference introduced by general processing methods that apply strong noise suppression to achieve the purpose of suppressing shallow noise in near-offset data, thus improving the amplitude fidelity of the processing.

Description

Technical Field

[0001] This invention relates to a method for suppressing pre-stack noise in seismic data, belonging to the field of seismic exploration technology. Background Technology

[0002] Pre-stack noise suppression is one of the three main battlegrounds in seismic data processing. It plays a crucial role in the process, primarily by suppressing various types of noise in the seismic data, preserving valuable seismic information, and improving the signal-to-noise ratio.

[0003] Currently, pre-stack noise suppression for seismic data can be mainly divided into two categories: pre-stack random noise suppression and pre-stack regular noise suppression. Pre-stack random noise suppression primarily targets irregular, high-energy noise in seismic data. Typical random noise includes high-energy interference such as spike noise, highway noise, and railway noise. This type of noise generally exhibits no clear pattern in frequency or velocity, but possesses high energy. Pre-stack regular noise suppression primarily targets regular noise in seismic data. Typical regular noise includes energy interference such as surface wave interference, linearity interference, and multiple waves. This type of noise is characterized by clear patterns in velocity or frequency.

[0004] In seismic data processing, methods for suppressing these two main types of noise interference can be broadly categorized into random noise suppression and regular noise suppression. Firstly, for random noise suppression, we often utilize the coherence of the effective signal to perform statistical analysis on the seismic data, thereby identifying random interference and then suppressing the noise interference portion. Secondly, for regular noise suppression, it can be essentially summarized as utilizing the differences in frequency and wavenumber between regular noise and the effective signal to suppress the regular noise. In both types of noise suppression, current processing software requires sampling within a certain time and space. That is, we need to perform statistical analysis on the L and N adjacent seismic channels of the target data within a certain time period to achieve the noise suppression effect. If the number of seismic data channels within the sampling window is too small or the noise within the window is too excessive, it is difficult to effectively suppress the noise. Typically, a single seismic gather contains a large number of seismic traces. However, during acquisition, it's inevitable that the first arrival time of seismic data increases with the offset. This results in a very small number of effective traces (N) for shallow layers in near-offset data, and strong noise interference at near-offset points makes it difficult to distinguish between effective information and noise interference, thus failing to effectively suppress shallow noise in near-offset seismic data. Although we can increase the number of samples by increasing the time window length (L), the accuracy of identifying effective information in the seismic data will decrease significantly with the increase in the sampling time window length. Fine structural information in the seismic data cannot be effectively protected, and the accuracy of seismic processing cannot be effectively guaranteed. Furthermore, due to the excessively large time window length (L) and strong suppression, some low-frequency square waves may be generated in the seismic data. Summary of the Invention

[0005] The purpose of this invention is to provide a pre-stack noise suppression method for seismic data, which can solve the problems that existing pre-stack noise suppression methods for seismic data cannot effectively suppress shallow noise in near-offset data, and that increasing the noise suppression intensity to suppress near-offset noise leads to loss of effective information and introduction of noise.

[0006] To achieve the above objectives, the technical solution adopted by this invention is as follows:

[0007] A method for pre-stack noise suppression of seismic data includes the following steps:

[0008] 1) After decompiling and defining the observation system of the acquired seismic data, static correction, spherical diffusion compensation, and surface wave suppression are performed in sequence.

[0009] 2) Correct the first arrival time of the seismic data obtained by surface wave suppression to a relatively uniform time, and then perform noise suppression processing on random noise and residual noise from surface wave suppression.

[0010] 3) Perform first arrival time inverse correction on the seismic data obtained from noise suppression in step 2) so that the first arrival time of the seismic data obtained from noise suppression is restored to the state before the first arrival time correction in step 2).

[0011] The pre-stack noise suppression method for seismic data of the present invention can effectively suppress shallow noise interference in near-offset data, especially strong noise interference, while better preserving the detailed structure of seismic data, improving the accuracy of seismic processing, and avoiding other noise interference introduced by general processing methods that apply strong noise suppression to achieve the purpose of suppressing shallow noise in near-offset data, thereby improving the amplitude fidelity of processing.

[0012] The pre-stack noise suppression method for seismic data of this invention corrects the first arrival time of seismic data to a relatively uniform time before performing noise suppression. This effectively solves the problem of near-offset shallow data not being sampled and analyzed when there is no adjacent seismic data. After the first arrival time is corrected, the effective information sampling of near-offset shallow data increases, which can effectively identify noise and effective information, thereby effectively suppressing noise. It effectively maintains the continuity of effective information in seismic data and does not cause temporal misalignment of seismic data. It effectively shortens the length of the sampling window, thereby increasing the recognition of small structures and increasing processing accuracy. It effectively reduces the noise suppression intensity and avoids square wave and other noise introduced by excessive denoising intensity in near-offset data.

[0013] The pre-stack random noise suppression of this invention is performed in the single-shot domain. Further, in step 2), the correction includes the following steps: first-arrival time picking is performed on the seismic data obtained from surface wave suppression, and the first-arrival time of the seismic data obtained from surface wave suppression is corrected to a relatively uniform time using the picked-up first-arrival time. When using first-arrival picking software to pick up the first-arrival time, a static correction library file with the first-arrival time as the static correction value can be created (there are many commercial static correction processing software programs available, and most can be used). Applying the library file to the seismic data obtained from surface wave suppression will correct the first-arrival time of the seismic data obtained from surface wave suppression to a relatively uniform time. Alternatively, when the first-arrival time fluctuation of a single shot is not significant, a simple first-arrival time cut-out library file can be created (commercial first-arrival time cut-out library files can all be implemented) to replace the static correction file with the first-arrival time as the static correction value, thus correcting the first-arrival time of the seismic data obtained from surface wave suppression to a relatively uniform time. In other words, there are many methods to correct the first-arrival time of a single shot to a relatively uniform time; regardless of the method used, as long as the first-arrival time of a single shot is corrected to a relatively uniform time, it is acceptable.

[0014] Furthermore, in step 3), the reverse correction is performed using the initial arrival time picked up in step 2).

[0015] Many existing methods for suppressing surface waves can be used for pre-stack noise suppression of the seismic data in this invention. Furthermore, to achieve better surface wave suppression, the surface wave suppression process employs a cross-shaped arrangement domain adaptive surface wave suppression method.

[0016] In step 2), random noise mainly includes irregular high-energy noise such as railway noise, pulse interference, and mechanical noise. Residual noise from surface wave suppression is mainly distributed in shallow layers near the offset. By effectively suppressing the strong random noise and residual noise from surface wave suppression in the seismic data, strong energy within the same shot can be eliminated, making the energy of the same shot relatively consistent. There are many methods for suppressing random noise and residual noise from surface wave suppression, and many mature methods are available in commercial processing software. Attached Figure Description

[0017] Figure 1 This is a flowchart of the pre-stack noise suppression processing method for seismic data in the embodiment.

[0018] Figure 2 This is a schematic diagram of single-shot seismic data after the acquired seismic data has been decoded and the observation system defined in the embodiment.

[0019] Figure 3 This is a schematic diagram of single-shot seismic data after static correction processing in the embodiment.

[0020] Figure 4 This is a schematic diagram of single-shot seismic data after surface wave suppression processing in the embodiment.

[0021] Figure 5 This is a schematic diagram illustrating the initial arrival time acquisition in the embodiment.

[0022] Figure 6 This is a schematic diagram of single-shot seismic data after first arrival time correction in the embodiment.

[0023] Figure 7 This is a schematic diagram of single-shot seismic data after residual noise suppression following random noise and surface wave suppression processing in the embodiment.

[0024] Figure 8 This is a schematic diagram of single-shot seismic data after first arrival time inverse correction in the embodiment.

[0025] Figure 9 The following diagrams illustrate the methods for noise suppression in the embodiment (9a), the method for noise suppression in the embodiment (9b), the method for noise suppression in the comparative example (9c), and the method for noise suppression in the comparative example (9d).

[0026] Figure 10This is a schematic diagram comparing the spectrum analysis of the method before noise suppression in the embodiment, the method after noise suppression in the embodiment, and the method after noise suppression in the comparative example.

[0027] Figure 11 The following diagrams illustrate the low-frequency portion of a single gun before noise suppression in the embodiment (11a), the low-frequency portion of a single gun after noise suppression in the embodiment (11b), and the low-frequency portion of a single gun after noise suppression in the comparative example (11c).

[0028] Figure 12 The following diagrams illustrate the following: (12a) a schematic diagram of a single gun arranged at a distance before noise suppression in the method of the embodiment; (12b) a schematic diagram of a single gun arranged at a distance after noise suppression in the method of the embodiment; (12c) a schematic diagram of a single gun arranged at a distance after noise suppression in the method of the comparative example; (12d) a schematic diagram of a noise-suppressed far-distance arrangement in the method of the embodiment; and (12e) a schematic diagram of a noise-suppressed far-distance arrangement in the method of the comparative example. Detailed Implementation

[0029] This invention applies an first-arrival time correction method to seismic data. While maintaining the continuity of effective information in the seismic data, it increases the sampling number of shallow data at near offsets, improving the identification of effective information and thus effectively distinguishing noise from effective information, thereby effectively suppressing noise. The following detailed implementation of the pre-stack noise suppression method for seismic data using a specific 3D seismic dataset is provided as an example.

[0030] Example 1

[0031] The pre-stack noise suppression method for seismic data in this embodiment is as follows: Figure 1 As shown, it includes the following steps:

[0032] 1) The acquired single-shot seismic data A is decompiled and the observation system is defined to obtain seismic data B, such as... Figure 2 As shown.

[0033] 2) Apply static correction processing technology to the processed data B from step 1) to eliminate the static correction time lag in seismic data caused by differences in surface elevation, enhance the continuity of effective information, and obtain seismic data C. For example... Figure 3 As shown, the continuity of the effective information reflection waves in the seismic data is significantly improved, the characteristics are obvious, the coherence of the ground data is enhanced, and it is beneficial to highlight the effective information in the process of noise suppression.

[0034] 3) Apply spherical diffusion compensation technology to the seismic data C to eliminate energy loss during the propagation of seismic waves, so that the energy of the effective information of the seismic data in the same single shot is relatively consistent, and obtain the seismic data D.

[0035] 4) Noise suppression follows the principle of prioritizing strong noises over weak ones. Adaptive surface wave suppression technology using a cross-shaped arrangement domain is applied to suppress surface waves in seismic data D. Seismic data E is obtained through this processing, such as... Figure 4 The surface wave interference in the deep layers of the single-shot data was well suppressed, but the noise in the shallow layers near the offset was not effectively suppressed due to the excessive noise energy and the reduction of available sampling statistics within the sampling window. Figure 4 It is easy to see that as the strata become shallower, the more noise remains from surface waves; on the other hand, there is strong noise at the near offset of a single shot. We need to suppress these noises along with other random high-energy noise in the seismic data.

[0036] 5) Current noise suppression methods all require sampling within a certain time window width (generally represented by the N adjacent channels of the target data in processing) and time window length L for coherent enhancement. Most current high-energy noise suppression methods and their basic principles are similar. To avoid the situation in step 4) where near-offset shallow noise is not effectively suppressed due to insufficient effective sampling within the time window width N, we can increase the sampling time window length L to increase the number of effective sampling points. However, as L increases, the denoising accuracy will be significantly reduced, and the shallow noise will not be significantly improved. Therefore, we need to increase the number of samples that can be taken in the near-offset shallow space without changing the continuity (i.e., coherence) of the seismic data.

[0037] Because the seismic data acquisition and observation system dictates that the first arrival time of the data increases with the offset, the upper part of the single-shot data exhibits a triangular-like structure. This means that there are no adjacent data points in the shallow horizontal direction (spatially) near the offset, which is detrimental to the identification of effective information. If we correct the first arrival times of all seismic traces in a single shot to a relatively uniform time, the upper part of the single-shot seismic data will be approximately horizontal, while maintaining the continuity of effective information in the lower part.

[0038] First arrival time picking is performed on the seismic data E (e.g.) Figure 5 As shown), when using first-arrival picking software to pick first-arrival times, a static correction library file can be generated with the first-arrival time as the static correction value. Applying this first-arrival time library file to seismic data E corrects the first-arrival times of the seismic data to a relatively uniform time (e.g., ...). Figure 6 As shown in the figure, the seismic data F is obtained.

[0039] from Figure 6 It can be seen that the first arrival times of the earthquake data are approximately on a horizontal line, and the continuity of effective information in the lower part is well maintained. In this way, the problem of lack of adjacent effective data in shallow layers near the earthquake offset is well solved.

[0040] 6) After addressing the issue of insufficient adjacent effective data for shallow seismic data at near offsets, random noise suppression was applied to seismic data F to suppress high-energy random noise interference, thus obtaining seismic data G (e.g., ...). Figure 7 (As shown). From Figure 7 We can see that the high-energy noise on the left side is well suppressed, and the interference from the near-offset shallow layer, which is considered high-energy random noise, is also effectively suppressed. This eliminates the strong energy within the same shot and achieves relative consistency in the energy of individual shots within the same shot.

[0041] 7) Apply the first arrival time database file generated in step 5) to perform first arrival time inverse correction on the seismic data G, restoring the first arrival time of the seismic data G to the state before processing in step 4), obtaining the result data H (e.g., ...). Figure 8 (As shown).

[0042] Comparative Example

[0043] The pre-stack noise suppression method for seismic data in this comparative example differs from the pre-stack noise processing method for seismic data in the embodiment only in that: this comparative example omits steps 5) and 7) of the embodiment and directly performs high-energy random noise suppression on the seismic data E obtained in step 4); the window length for noise suppression effect is increased to twice that of random noise suppression in step 6) of the embodiment.

[0044] If the comparative example omits only steps 5) and 7) and directly applies high-energy random noise suppression to the seismic data E obtained in step 4), the processing effect is much worse than that of the example and fails to meet the requirements for noise suppression of seismic data.

[0045] Experimental Example 1

[0046] Comparing the results of the method in the embodiment (processing object: data from another single shot in the area where the single shot in the embodiment is located) with the results of the method in the comparative example, it can be seen that the method in the embodiment, after a series of processing steps, effectively suppresses noise in the seismic data. At the same time, the pre-stack noise suppression method of the seismic data in the embodiment can effectively preserve the effective information.

[0047] The schematic diagrams of a single gun before noise suppression in the embodiments and comparative examples, the schematic diagram of a single gun after noise suppression in the embodiments, the schematic diagram of a single gun after noise suppression in the comparative examples, the schematic diagram of noise suppression in the embodiments, and the schematic diagram of noise suppression in the comparative examples are respectively shown in [reference 1]. Figure 9 a-9e. Passed. Figure 9As can be seen from a-9e, the noise in the data processed by the method in the embodiment is well suppressed; while the method in the comparative example, by increasing the time window length L and increasing the noise suppression intensity, does not show obvious effect in suppressing shallow noise at near offset; at the same time, it can be seen from the noise profile of the suppressed noise that the present invention suppresses has no effective information, while the general method, due to the increase of the time window length L and the suppression intensity, suppresses a large amount of effective information in the noise.

[0048] The noise suppression spectra before noise suppression, the noise suppression spectra of the example, and the noise suppression spectra of the comparative example are shown below. Figure 10 ,pass Figure 10 The corresponding spectrum analysis shows that the spectrum processing in the embodiment effectively preserved the spectral characteristics of the original data.

[0049] Schematic diagrams of the low-frequency range of a single gun before noise suppression in the embodiment method, the low-frequency range of a single gun after noise suppression in the embodiment method, and the low-frequency range of a single gun after noise suppression in the comparative example are shown below. Figure 11 a-11c. Passed. Figure 11 As can be seen from a-11c, the noise suppression of the comparative method resulted in low-frequency square waves in the near-offset data due to the increase in the time window length L and the suppression intensity. However, the pre-stack noise suppression method of the embodiment method effectively avoided this phenomenon.

[0050] Schematic diagrams of single-gun element arrangement before noise suppression, single-gun element arrangement after noise suppression, and far-field arrangement of single-gun after noise suppression in the embodiment and comparative methods are shown below. Figure 12 a-12e. From Figure 12 As can be seen in a-12e, due to the improper handling of shallow data by the noise suppression method in the comparative method, high-frequency noise was introduced into the first arrival portion of the seismic data. This can be further confirmed by suppressing the noise profile. In contrast, the pre-stack noise suppression method in the embodiment effectively addresses the problem of shallow data and effectively suppresses noise, while preserving effective information well, thus improving the processing accuracy and amplitude fidelity of seismic data.

Claims

1. A method for pre-stack noise suppression of seismic data, characterized in that: Includes the following steps: 1) After decompiling and defining the observation system of the acquired seismic data, static correction, spherical diffusion compensation, and surface wave suppression are performed in sequence. 2) Correct the first arrival time of the seismic data obtained by surface wave suppression to a relatively uniform time, and then suppress the random noise and residual noise from surface wave suppression. 3) Perform first arrival time inverse correction on the seismic data obtained from noise suppression in step 2) so that the first arrival time of the seismic data obtained from noise suppression is restored to the state before the first arrival time correction in step 2).

2. The pre-stack noise suppression method for seismic data according to claim 1, characterized in that: In step 2), the correction includes the following steps: picking the first arrival time of the seismic data obtained by surface wave suppression processing, and using the picked first arrival time to correct the first arrival time of the seismic data obtained by surface wave suppression to a relatively uniform time.

3. The pre-stack noise suppression method for seismic data according to claim 1 or 2, characterized in that: In step 3), the inverse correction is performed using the initial arrival time picked up in step 2).

4. The pre-stack noise suppression method for seismic data according to claim 1 or 2, characterized in that: The surface wave suppression process employs a cross-shaped arrangement domain adaptive surface wave suppression method.

Images