Attribute-joint method for compensating near-surface attenuation of seismic signals

Abstract

The present invention relates to an attribute-joint seismic signal near-surface attenuation compensation method, comprising the following steps: S1 collecting seismic signals to obtain a data set of common excitation points and a data set of common receiving points; S2 obtaining a relative amplitude attenuation coefficient and a relative main frequency value based on a surface consistency principle; S3 combining the relative amplitude attenuation coefficient with the relative main frequency to perform near-surface attenuation compensation: calculating a Q value by averaging the main frequency value, calculating a compensation curve using the Q value, and adopting a standard deviation normalization algorithm to unify the spatial variation of the amplitude of the compensation curve to the same level, that is, when applying the relative main frequency value for compensation, only the frequency loss is compensated, and the signal amplitude level is not changed; then, applying the relative amplitude attenuation coefficient of the excitation point or the receiving point to the compensation curve, making an overall adjustment to the amplitude level of the compensation curve, thereby compensating for the frequency and amplitude loss caused by the near-surface stratum on the signal attenuation, and restoring the amplitude and frequency consistency to the greatest extent.

Description

Technical Field

[0001] The present invention belongs to the field of seismic signal processing for oil and gas exploration, and mainly aims to compensate and restore attenuated seismic exploration acquisition signals. Background Art

[0002] The near-surface strata have loose sediments, which strongly attenuate seismic signals. Spatial variations in lithology and sediment thickness near the surface lead to significant lateral variations in signals collected at different ground points. This invention aims to address the attenuation effects of the near-surface on seismic signals and improve the lateral consistency of signal waveforms.

[0003] Seismic signals received at ground points carry a wealth of stratigraphic information. Processing and interpreting these signals can reveal the distribution of underground rock formations, guiding oil and gas exploration and development. However, seismic signal propagation is often subject to interference, impacting the accuracy of stratigraphic information. Signal attenuation is a major interfering factor. Unlike deep diagenetic strata, near-surface strata are primarily composed of loose sand and clay. Studies have shown that near-surface strata attenuate over 70% of signal energy. Furthermore, near-surface stratigraphic structure, including lithology and sediment thickness, varies rapidly in space. Consequently, the attenuation process causes significant differences in frequency, energy, and phase between signals received at different ground points (Figure 1). Without restoring the signals to the closest possible consistency to the original signal, valid subsurface rock formation information cannot be readily analyzed and interpreted from the signals.

[0004] To address this issue, the field of seismic signal processing for oil and gas exploration has developed corresponding technical means, such as surface consistency deconvolution, surface consistency amplitude compensation, etc. [1] These two methods have been integrated into commercial software and are widely used as a standard technique in oil and gas exploration. These two methods are often used together to address waveform inconsistencies caused by near-surface attenuation. Surface-consistent deconvolution improves frequency consistency, while surface-consistent amplitude compensation addresses amplitude consistency. Surface-consistent deconvolution attempts to separate the attenuated wavelet from the acquired signal using certain assumptions and then restores the signal frequency using a deconvolution algorithm. However, because these assumptions differ significantly from reality, the results are often suboptimal. Surface-consistent amplitude compensation, on the other hand, simply calculates amplitude differences and balances amplitude consistency by adjusting an amplitude coefficient. Neither algorithm is based on a mathematical model of attenuation.

[0005] In recent years, there has been an increasing number of studies on near-surface attenuation compensation. One of the difficulties in attenuation compensation is the calculation of the earth's quality factor Q value. In theory, the Q value requires the amplitude and frequency information of the wavelet (the signal obtained by exciting the ground point) before and after attenuation. However, since the received signal not only contains the wavelet information but also carries the formation information, the attenuated wavelet cannot be accurately obtained. At present, there are studies that use actual measurements of well logging data to obtain the Q value. [2][3] However, it is impossible to conduct a large-scale well drilling survey and calculation of the Q value in the entire work area from a cost perspective, and only a small-scale verification can be performed. Another study based on near-surface structure modeling to obtain velocity information, and through the empirical formula between velocity and Q value, to approximate the size and spatial variation of Q. [4][5] However, due to the low accuracy or poor matching with the actual received signal, the compensation result cannot meet the exploration needs. Some scholars have considered the importance of making full use of the actual received signal, and based on the frequency shift theory, the relationship between Q value and frequency is calculated by statistically analyzing the relative relationship of the actual signal frequency. [6-9] , calculate the near-surface Q value, or statistical amplitude attributes

[10]

[11] , using the spectral ratio method to calculate the Q value, thereby compensating and restoring waveform consistency. This method often achieves a higher degree of match with the data, making it more reliable. However, focusing solely on a single attribute, such as frequency or amplitude, cannot guarantee the complete consistency of the signal waveform.

[0006] References

[0007] [1] Chen Chaoqun, Dai Haitao, et al. Research and application of consistency processing methods for seismic data under complex surface conditions [J]. Geophysical and Geochemical Exploration. 2023, 47(4), 954-964.

[0008] [2] Luo Yong, Mao Haibo, et al. Estimation of near-surface Q value using polynomial fitting centroid frequency shift method [J]. Petroleum Geophysical Exploration. 2016, 51(3), 589-595.

[0009] [3]Shi Zhanjie, Tian Gang, et al. Near-surface absorption compensation technology and its application in the Daqing Oilfields[J].AppliedGeophysics.2009,6(2),184-191.

[0010] [4] Cui Qinghui. Research on near-surface absorption compensation method for seismic data in desert areas[J]. Science, Technology and Engineering. 2016, 16(8), 50-53.

[0011] [5] Chen Li, Zhou Qiang, et al. Research and application of near-surface Q compensation technology in western China [J]. Proceedings of the 2022 International Conference on Oil and Gas Field Exploration and Development II.

[0012] [6] Wang Jing, Liu Weiming, et al. Near-surface Q compensation technology and its application in continuous data processing [J]. Proceedings of the 2019 Annual Conference on Oil and Gas Geophysics.

[0013] [7] Feng Xinyuan, Zeng Ming, et al. Data-driven near-surface Q compensation method and its application [J]. Inner Mongolia Petrochemical, 2023(11), 27-30.

[0014] [8] Wang Yihui, Wang Xiaowei, et al. Application of near-surface Q compensation to deep-seated seismic data in western China [J]. Proceedings of the 2019 Annual Conference on Oil and Gas Geophysics.

[0015] [9] Liu Huan, Su Qin, et al. Application of near-surface Q compensation technology in tight gas exploration in central Sichuan [J]. Lithologic Reservoirs. 2021, 33(3), 104-112.

[0016]

[10] Zeng Huahui, Wang Xiao, et al. Application of near-surface absorption compensation technology in tight oil reservoir treatment in Wuxia area [J]. 2015 Geophysical Exploration Technology Seminar.

[0017]

[11] Jiang Li, Luo Yong, et al. Research and application of surface consistency surface relative Q calculation and compensation method [J]. Xinjiang Geology. 2015, 33(3), 415-420. Summary of the Invention

[0018] The present invention aims to overcome the aforementioned shortcomings of the prior art by providing a method for compensating for near-surface attenuation of seismic signals using a combined attribute approach. Based on a mathematical model of attenuation, the present invention calculates the relative amplitude attenuation coefficient and relative dominant frequency by actually receiving the signal and simultaneously counting the amplitude and frequency attributes. The relative dominant frequency is used to derive the Q value associated with the near-surface, and both the Q value and the relative amplitude attenuation coefficient are simultaneously applied to the compensation process. The technical solution is as follows:

[0019] A method for compensating near-surface attenuation of seismic signals using a combined attribute approach comprises the following steps:

[0020] S1 collects seismic signals and obtains data sets of common excitation points and common receiving points;

[0021] S2 obtains the relative amplitude attenuation coefficient and relative main frequency value based on the surface consistency principle

[0022] Assume that the data received by the excitation point i and the receiving point j is A ij , i, j are the excitation point and receiving point index numbers respectively, i = 1…n, n is the number of excitation points, j = 1…l, l is the number of receiving points;

[0023] The process of S21 calculating the relative amplitude attenuation coefficient between the excitation point and the receiving point is as follows:

[0024] S211 will A ij Sorted into n co-excitation point data sets S according to the co-excitation points i , i=1…n;

[0025] S212 for each co-excitation point data set S i , i = 1…n, find the average amplitude mean_a of each channel data it contains i ;

[0026] S213 For all co-excitation point data sets S i , i = 1…n, calculate the average amplitude mean_a i , the maximum value max_a among i=1…n;

[0027] S214 performs maximum normalization to obtain the data set S of each co-excitation point i The maximum normalized value of the average amplitude norm_a i , i=1…n;

[0028] S215 inverts the normalized data to obtain the temporary relative amplitude attenuation coefficient ini0_α of each excitation point i =1 / norm_a i , i=1…n;

[0029] S216 sets the temporary relative amplitude attenuation coefficient ini0_α of each excitation point i , i = 1…n and the corresponding co-excitation point data set S i Multiply all the channel data in to remove the amplitude attenuation effect of the corresponding excitation point and obtain the adjusted A ij ;

[0030] S217 adjusts the A of S216 ij Sorted by index number j into a common reception point dataset R j , according to the same calculation method as S211 to S216, the temporary relative amplitude attenuation coefficient ini0_α of each receiving point is obtained j , j=1…l and the adjusted A ij ;

[0031] S218 specifies the number of repetitions m, repeats S211 to S217, and obtains the temporary relative amplitude attenuation coefficient ini1_α of each excitation point in turn. i , ini2_α i ...and the temporary relative amplitude attenuation coefficient ini1_α at each receiving point j, ini2_α j …, multiply the temporary relative amplitude attenuation coefficients of the m groups of excitation points and receiving points obtained m times respectively, and obtain the relative amplitude attenuation coefficients α of the excitation point and the receiving point respectively i and α j ;

[0032] S22 calculates the relative main frequency value γ of the excitation point i and the receiving point j i and γ j , the calculation method is as follows:

[0033] S221 specifies the number of repetitions p, and calculates the temporary relative main frequency coefficients of the excitation point and the temporary relative main frequency coefficients of the receiving point p times according to the calculation method of S21. When calculating, the average amplitude of each data channel is replaced by the main frequency of the spectrum of each data channel;

[0034] S222 multiplies the temporary relative main frequency coefficients of the p excitation points obtained by p calculations by the average main frequency value of all co-excitation point data sets to obtain the relative main frequency value γ of the excitation point i. i ;

[0035] S223 multiplies the temporary relative main frequency coefficients of the p receiving points obtained by p calculations by the average main frequency value of all the common receiving point data sets to obtain the relative main frequency value γ of the receiving point j. j ;

[0036] The S3 relative amplitude attenuation coefficient is combined with the relative dominant frequency to compensate for near-surface attenuation: the Q value is calculated by averaging the dominant frequency value, and the compensation curve is calculated using the Q value. A standard deviation normalization algorithm is used to unify the spatial variations in the compensation curve amplitude to the same level. That is, when the relative dominant frequency value is used for compensation, only the frequency loss is compensated, and the signal amplitude level is not changed. The relative amplitude attenuation coefficient of the excitation point or receiving point is then applied to the compensation curve, making an overall adjustment to the compensation curve's amplitude level, thereby compensating for the frequency and amplitude losses caused by signal attenuation due to near-surface strata and restoring amplitude and frequency consistency to the greatest extent possible.

[0037] Furthermore, S1 includes: after the signal excited by an excitation point propagates to a certain depth underground, its reflected signal is recorded by a certain number of receiving points, and one receiving point records the signals propagated from different excitation points; by sorting the data recorded by the receiving points, data with the same excitation point but different receiving point positions are obtained to form a data set of common excitation points; data with the same receiving point positions but different excitation points are obtained to form a data set of common receiving points.

[0038] Furthermore, in S222, the relative main frequency value γ of the excitation point i is obtained. i The formula is:

[0039]

[0040] Among them, mean_γ i is the excitation point dataset S i The average main frequency, ini0_γ i , ini1_γ i …are the temporary relative main frequency coefficients of the excitation points obtained during the p-times repetition process.

[0041] Furthermore, in S222, the relative main frequency value γ of the receiving point j is obtained. j The formula is:

[0042]

[0043] Among them, mean_γ j is the receiving point dataset R j The average main frequency, ini0_γ j , ini1_γ j …are the temporary relative main frequency coefficients of the receiving points obtained during the p-times repetition process.

[0044] Furthermore, the S3 method is as follows:

[0045] S31 according to γ i and γ j The Q value of the excitation point is calculated by the frequency shift method, that is, Q i , and the Q value of the receiving point, that is, Q j :

[0046]

[0047] When calculating the Q value of the excitation point, it is assumed that the receiving point is located at the bottom of the ground near the excitation point. The f in formula (1) is r is the main frequency of the excitation signal;

[0048] Assuming that the excitation point is located at the bottom of the near-surface formation opposite the receiving point, calculate the receiving point Q using the same calculation method. j ;

[0049] S32 according to Q i and Q j Construct the initial compensation curve A corresponding to the excitation point and the receiving point i (f) and A j (f), called the initial compensation curve constructed by frequency attributes;

[0050] S33 retains the steepness relationship of the initial compensation curve constructed by frequency attributes, and adopts the standard deviation normalization algorithm to realize the combination of frequency attributes and amplitude attributes in the compensation curve.

[0051] Furthermore, the method of step S33 is as follows:

[0052] S331 Calculate the expected value μ of the compensation amount of the compensation curve of the excitation point and the receiving point i and μ j :

[0053]

[0054] Where Δf is the discrete signal frequency sampling interval, μ i and μ j are the expected values ​​of the compensation at the excitation point and the receiving point respectively;

[0055] S332 calculates the standard deviation of the compensation amount of the excitation point and the receiving point compensation curve:

[0056]

[0057] Among them, σ i and σ j are the standard deviations of the compensation corresponding to the excitation point and the receiving point, respectively, x is a certain frequency value, x is the index value of a certain frequency in the effective frequency band of the seismic signal, and X is the index value corresponding to the upper limit value of the effective frequency band of the seismic signal;

[0058] S333 by σ i and σ j Normalize the compensation curves of the excitation point and the receiving point respectively:

[0059] AR i (f) = A i (f) / σ i , AR j (f) = A j (f) / σ j (4) Among them, AR i (f)With AR j (f) The compensation curves of the excitation point and the receiving point after normalization;

[0060] S334 applies the relative amplitude attenuation coefficient to the normalized compensation curve to obtain the final near-surface attenuation compensation curve:

[0061] AF ij (f) = α i α j AR i (f)AR j (f) (5)

[0062] Compared with the prior art, the technical solution of the present invention has the following beneficial effects:

[0063] 1) It is convenient for industrial application. The frequency and main frequency extraction adopt the same algorithm framework, which can simultaneously extract and calculate two sets of key attribute data, and simultaneously calculate the relative amplitude attenuation coefficient and the relative main frequency. The algorithm has high execution efficiency and can realize efficient calculation of data volumes above T level in the entire work area.

[0064] 2) The obtained Q value has a high degree of match with the data and is completely dependent on the actual data. By extracting the effective information in the data, the attenuation-related properties are obtained, thereby obtaining the Q value used for attenuation compensation.

[0065] 3) The surface consistency principle is combined with attenuation compensation technology. The calculation of the relative amplitude attenuation coefficient and relative main frequency is actually based on the surface consistency principle, which is widely used in conventional processing. Based on this principle, the Q value is calculated and the compensation curve is further calculated and corrected.

[0066] 4) Strong anti-interference ability. Since the frequency and amplitude attributes of the data are extracted simultaneously, it avoids the situation where the attribute extraction is inaccurate due to interference from noise when using only the frequency or amplitude attribute, thereby affecting the compensation effect.

[0067] 5) Taking into account both theoretical and actual data conditions, the waveform signal consistency can be fully improved from both the amplitude and frequency aspects. Starting from the mathematical model of attenuation, the Q value is calculated and the core compensation algorithm is based on the theoretical model. In improving the consistency of the signal waveform, the frequency and amplitude attributes are integrated into the compensation curve to fully take into account the actual data conditions. In the theoretical model, after attenuation, the frequency and amplitude should have the same change trend. However, in actual data, due to various external factors (instrument reasons and environmental noise, etc.), the theory is sometimes not met, that is, the frequency and amplitude have different change trends. Therefore, when the frequency attribute is used alone, the amplitude consistency requirement cannot be fully met.

[0068] In summary, the method proposed in this paper accounts for the discrepancies between theoretical models and actual data, while simultaneously utilizing frequency and amplitude properties to compensate for near-surface formation attenuation. This effectively improves signal waveform consistency, restores signal frequency and amplitude, and enables data to effectively guide oil and gas exploration. This invention exhibits strong anti-interference capabilities and high execution efficiency, enabling large-scale industrial data processing applications. BRIEF DESCRIPTION OF THE DRAWINGS

[0069] Figure 1 Schematic diagram of near-surface attenuation seismic signals

[0070] Figure 2 Schematic diagram of seismic signal excitation and reception

[0071] Figure 3 Frequency and attribute plane distribution diagram obtained by analyzing the same set of data (different positions represent different receiving points)

[0072] Figure 4 Implementation Flowchart

[0073] Figure 5 Schematic diagram of obtaining the Q value of the excitation point

[0074] Figure 6 Schematic diagram of obtaining the Q value of the receiving point

[0075] Figure 7 Initial compensation curve constructed based on frequency attributes

[0076] Figure 8 Compensation curve after normalization

[0077] Figure 9 Actual data application effect DETAILED DESCRIPTION

[0078] When actual data is inconsistent with the theoretical model, there are certain problems in calculating Q and compensating based on a single attribute. In theory, the signal amplitude and frequency attributes caused by attenuation should have the same change trend, but through the analysis of actual data, it is confirmed that the spatial distribution of amplitude and frequency is often quite different ( Figure 3 ). Therefore, when attenuation compensation is performed based solely on frequency attributes, not only will the amplitude consistency not be improved, but it may even deteriorate. The same principle applies to amplitude. The present invention utilizes both attributes simultaneously, aiming to balance the amplitude and frequency characteristics of signals at different excitation points and different receiving points, thereby eliminating the attenuation effect of near-surface formations. The specific steps are as follows:

[0079] 1) Seismic signal acquisition

[0080] Seismic signal acquisition work Figure 2 As shown, after the signal excited by an excitation point propagates to a certain depth underground, its reflected signal is recorded by a certain number of receiving points. A receiving point also needs to record the signals propagated from different excitation points. By sorting all the data recorded by the receiving points, a data set of common excitation points (the excitation point is the same, but the receiving point positions are different) or a data set of common receiving points (the receiving point positions are the same, but the excitation points are different) can be obtained. In this way, the overall amplitude or frequency difference between the data sets of different excitation points is caused by the attenuation of the near-surface stratum directly below the excitation point. For data sets of different common receiving points, the difference is caused by the attenuation of the near-surface stratum directly below the receiving point. By statistically analyzing the sorted data separately, as shown in Figure 2 The three sets of data from the receiving point are used to calculate and compare the amplitude and main frequency, respectively, to obtain the relative amplitude attenuation coefficient and relative main frequency value. This can be further analyzed to determine the relative attenuation of the near-surface formation corresponding to the receiving point. Low frequencies and weak amplitudes indicate greater attenuation, and vice versa. The same process is repeated for the excitation point.

[0081] 2) Obtaining the relative amplitude attenuation coefficient and relative main frequency value based on the surface consistency principle

[0082] Because the spatial variation of near-surface strata is much greater than that of mid-depth strata, amplitude and frequency differences between datasets at different excitation points are primarily due to near-surface absorption. This also applies to the receiving points. Therefore, based on the principle of surface consistency, the amplitude and frequency differences in the data can be decomposed. After decomposition, the amplitude and frequency differences at the excitation point—that is, the relative amplitude attenuation coefficient and relative dominant frequency—are attributed to the near-surface strata directly below the excitation point. The same principle applies to the receiving point.

[0083] First calculate the relative amplitude attenuation coefficient. Assume that the data received by the excitation point i and the receiving point j is A ij , i, j are the excitation point and receiving point index numbers respectively, i = 1...n, n is the number of excitation points, j = 1...l, l is the number of receiving points, the calculation process is as follows:

[0084] 1) A ij Sorted into n data sets S according to the co-excitation points i , i=1…n.

[0085] 2) For each data set S i Find the average amplitude mean_a i , as follows:

[0086] For the excitation point i, let the total number of signal waveform data received by each receiving point be T,

[0087]

[0088] where a t For the dataset S i The average amplitude of each data in .

[0089] 3) Find the maximum value max_a,

[0090]

[0091] 4) Perform maximum value normalization,

[0092] norm_a i =mean_a i / max_a i =1…n

[0093] 5) Invert the normalized data to obtain the temporary relative amplitude attenuation coefficient of the excitation point,

[0094] ini0_α i =1 / norm_a i

[0095] 6) Set the temporary relative amplitude attenuation coefficient ini0_α i With the corresponding dataset S i Multiply all the channel data in to remove the amplitude attenuation effect of the excitation point and get the adjusted A ij ;

[0096] 7) The adjusted A ij Sorted by index number j into a common reception point dataset R j According to the same principle as steps 1)-6), the temporary relative amplitude attenuation coefficient ini0_α of the receiving point is obtained j and the adjusted A ij .

[0097] Doing steps 1-7 only once can only obtain partial attenuation effects. It is necessary to repeat a certain number of times. The number of repetitions m is specified based on experience, and the temporary relative amplitude attenuation coefficient ini1_α of each excitation point is obtained in turn. i , ini2_α i ...and the temporary relative amplitude attenuation coefficient ini1_α at each receiving point j , ini2_α j …, multiply the temporary relative amplitude attenuation coefficients of the m groups of excitation points and receiving points obtained m times respectively to obtain the final relative amplitude attenuation coefficient α i and α j

[0098] α i =ini0_α i *ini1_α i *ini2_α i *…

[0099] α j =ini0_α j *ini1_α j *ini2_α j *…

[0100] Calculate the relative main frequency value γ of the excitation point i and the receiving point j i and γ j The calculation method is as follows: specify the number of repetitions p based on experience, and use the same principle to calculate the temporary relative main frequency coefficients of the p excitation points and the p receiving points. When calculating, the average amplitude needs to be replaced by the main frequency calculated from the spectrum of the calculated data channel.

[0101] Since the Q value needs to be calculated by the relative dominant frequency, it is also necessary to multiply the temporary relative dominant frequency coefficients of the p excitation points obtained by p calculations by the average dominant frequency value of all co-excitation point data sets, that is,

[0102]

[0103] Among them, mean_γ i is the excitation point dataset S i The average main frequency, ini0_γ i , ini1_γ i …are the temporary relative main frequency coefficients of the excitation points obtained during the p-times repetition process.

[0104]

[0105] mean_γ j is the receiving point dataset R j The average main frequency, ini0_γ j , ini1_γ j …are the temporary relative main frequency coefficients of the receiving points obtained during the p-times repetition process.

[0106] 3) Combining the relative amplitude attenuation coefficient and the relative main frequency to compensate for near-surface attenuation

[0107] This section primarily applies the relative amplitude attenuation coefficient and relative dominant frequency values ​​from the above results to construct the compensation curve. Since the relative amplitude attenuation coefficient, relative dominant frequency, and Q value obtained vary spatially, the constructed compensation curve also varies spatially. The key is how to simultaneously apply the relative amplitude attenuation coefficient and relative dominant frequency value to the compensation curve. According to this approach, applying the relative dominant frequency value only changes the frequency properties of the signal, but the amplitude of the actual compensation curve varies with the Q value, which also changes the amplitude properties. To address this issue, we use a standard deviation normalization algorithm to normalize the spatial variation of the compensation curve's amplitude to the same level. This allows only frequency loss to be compensated when applying the dominant frequency value. After this, the relative amplitude attenuation coefficient is applied to the compensation curve to adjust the overall amplitude level. This process effectively compensates for the frequency and amplitude losses caused by near-surface strata, restoring the maximum possible consistency between amplitude and frequency.

[0108] First, according to γ i and γ j The Q value of the excitation point is calculated by the frequency shift method, that is, Q i And the Q value of the receiving point is Q j .

[0109]

[0110] When calculating the Q value of the excitation point, it is assumed that the receiving point is located at the bottom of the ground near the excitation point ( Figure 5 ), f in formula (1) ris the main frequency of the excited signal. Here we only consider the attenuation effect, so it is assumed that the main frequency of the excited signal at different points is the same. Calculate the receiving point Q j The process is the same when , and it is assumed that the excitation point is located at the bottom of the near-surface formation ( Figure 6 ).

[0111] Next, according to Q i and Q j Construct the initial compensation curve A corresponding to the excitation point and the receiving point i (f) and A j (f), called the initial compensation curve constructed by frequency attributes, such as Figure 7 As shown in the figure, this is a set of curves with different compensation amounts corresponding to different Q values. The smaller the Q value, the steeper the curve is at the mid- and low-frequency ends, indicating that the frequency is recovered more. The smaller the Q value, the larger the amplitude of the compensation curve, indicating that the amplitude is recovered more. However, if the relative amplitude attenuation coefficient obtained at this point is a value with a small attenuation, that is, from the amplitude perspective, a smaller compensation amount is required, which will be counterproductive. Therefore, it is only necessary to retain the steepness relationship of the initial compensation curve constructed by the frequency attribute, while the overall energy level of the curve is unified. It is necessary to find an effective energy normalization algorithm to realize the combination of the two attributes in the compensation curve. Due to the nonlinearity of the compensation curve, standard deviation normalization is used. First, the expected value of the compensation amount of the compensation curve is calculated (the seismic signal is a broadband signal, and the effective frequency band of the seismic signal is approximately 0-100Hz):

[0112]

[0113] Where Δf is the discrete signal frequency sampling interval, μ i and μ j are the expected values ​​of the compensation at the excitation point and the receiving point respectively. The standard deviation of the compensation at different frequencies is further calculated:

[0114]

[0115] Among them, σ i and σ j are the standard deviations of the compensation corresponding to the excitation point and the receiving point, respectively, x is a certain frequency value, x is an arbitrary frequency index value, and X is the index value corresponding to the upper limit of the effective frequency band of the seismic signal. i and σ j Perform the final normalization processing on the compensation curves of the excitation point and the receiving point respectively:

[0116] AR i (f) = A i (f) / σ i , AR j (f) = A j(f) / σ j (4)

[0117] Among them, AR i (f)With AR j (f) are the compensation curves of the excitation point and the receiving point after normalization, such as Figure 8 shown.

[0118] Finally, the relative amplitude attenuation coefficient is applied to the normalized compensation curve to obtain the final near-surface attenuation compensation curve.

[0119] AF ij (f) = α i α j AR i (f)AR j (f) (5)

[0120] Figure 9 This is a demonstration of the effects of applying this invention to actual data. The vertical coordinate represents time, indicating the signal continuously propagating upward from the underground, received by a specific excitation point. The horizontal coordinate represents the coordinate position, corresponding to different ground points or receiving points. Due to the confidentiality of exploration data, the coordinate position and time information are not displayed here. It can be seen that the raw data has large amplitude and frequency differences in the horizontal direction. Figure 9 b is the result of attenuation compensation using only frequency attributes. Due to the inconsistency between the frequency and amplitude attributes of the original data, although the frequency is increased during the compensation process, the horizontal difference in amplitude is actually greater than that of the original data, which also confirms the problem mentioned in the implementation plan. Figure 9 Figure c is the result of attenuation compensation using the method of the present invention. It can be seen that the lateral frequency and amplitude consistency are greatly improved.

Claims

1. A method for compensating near-surface attenuation of seismic signals using a combined attribute method, comprising the following steps: S1 collects seismic signals and obtains data sets of common excitation points and common receiving points; S2 obtains the relative amplitude attenuation coefficient and relative main frequency value based on the surface consistency principle Assume that the data received by the excitation point i and the receiving point j is A ij , i, j are the excitation point and receiving point index numbers respectively, i = 1…n, n is the number of excitation points, j = 1…l, l is the number of receiving points; The process of S21 calculating the relative amplitude attenuation coefficient between the excitation point and the receiving point is as follows: S211 will A ij Sorted into n co-excitation point data sets S according to the co-excitation points i ; S212 for each co-excitation point data set S i , find the average amplitude mean_a of each channel data it contains i ; S213 For all co-excitation point data sets S i , calculate the average amplitude mean_a i The maximum value max_a; S214 performs maximum normalization to obtain the data set S of each co-excitation point i The maximum normalized value of the average amplitude norm_a i ; S215 inverts the normalized data to obtain the temporary relative amplitude attenuation coefficient ini0_α of each excitation point i =1 / norm_a i ; S216 sets the temporary relative amplitude attenuation coefficient ini0_α of each excitation point i and the corresponding co-excitation point dataset S i Multiply all the channel data in to remove the amplitude attenuation effect of the corresponding excitation point and get the adjusted A ij ; S217 adjusts the A of S216 ij According to the j index number, the common reception point dataset R is sorted j , according to the same calculation method as S211 to S216, the temporary relative amplitude attenuation coefficient ini0_α of each receiving point is obtained j and the adjusted A ij ; S218 specifies the number of repetitions m, repeats S211 to S217, and obtains the temporary relative amplitude attenuation coefficient ini1_α of each excitation point in turn. i , ini2_α i ...and the temporary relative amplitude attenuation coefficient ini1_α at each receiving point j , ini2_α j …, multiply the temporary relative amplitude attenuation coefficients of the m groups of excitation points and receiving points obtained m times respectively, and obtain the relative amplitude attenuation coefficients α of the excitation point and the receiving point respectively i and α j ; S22 calculates the relative main frequency value γ of the excitation point i and the receiving point j i and γ j , the calculation method is as follows: S221 specifies the number of repetitions p, and calculates the temporary relative main frequency coefficients of the excitation point and the temporary relative main frequency coefficients of the receiving point p times according to the calculation method of S21. When calculating, the average amplitude of each data channel is replaced by the main frequency of the spectrum of each data channel; S222 multiplies the temporary relative main frequency coefficients of the p excitation points obtained by p calculations by the average main frequency value of all co-excitation point data sets to obtain the relative main frequency value γ of the excitation point i. i ; S223 multiplies the temporary relative main frequency coefficients of the p receiving points obtained by p calculations by the average main frequency value of all the common receiving point data sets to obtain the relative main frequency value γ of the receiving point j. j ; The S3 relative amplitude attenuation coefficient is combined with the relative main frequency to compensate for near-surface attenuation: the Q value is calculated using the relative main frequency value, and the compensation curve is calculated using the Q value. A standard deviation normalization algorithm is used to unify the spatial variations in the compensation curve amplitude to the same level. That is, when the relative main frequency value is applied for compensation, only the frequency loss is compensated, and the signal amplitude level is not changed. Then, the relative amplitude attenuation coefficient of the excitation point or receiving point is applied to the compensation curve, and the amplitude level of the compensation curve is adjusted as a whole, thereby compensating for the frequency and amplitude losses caused by signal attenuation due to near-surface strata and restoring the amplitude and frequency consistency to the greatest extent possible.

2. The attribute-combined near-surface attenuation compensation method for seismic signals according to claim 1, characterized in that: S1 includes: After a signal emitted by an excitation point propagates to a certain depth underground, its reflected signal is recorded by a certain number of receiving points. Each receiving point records signals transmitted from different excitation points. By sorting the data recorded by the receiving points, data with the same excitation point but different receiving point locations are obtained to form a data set of common excitation points. The data with the same receiving point position but different excitation points are obtained to form a data set of common receiving points.

3. The attribute-combined near-surface attenuation compensation method for seismic signals according to claim 1, characterized in that: In S222, the relative main frequency value γ of the excitation point i is obtained i The formula is: Among them, mean_γ i is the excitation point dataset S i The average main frequency, ini0_γ i , ini1_γ i …are the temporary relative main frequency coefficients of the excitation points obtained during the p-times repetition process.

4. The attribute-combined near-surface attenuation compensation method for seismic signals according to claim 1, characterized in that: In S222, the relative main frequency value γ of the receiving point j is obtained j The formula is: Among them, mean_γ j is the receiving point dataset R j The average main frequency, ini0_γ j , ini1_γ j …are the temporary relative main frequency coefficients of the receiving points obtained during the p-times repetition process.

5. The attribute-combined near-surface attenuation compensation method for seismic signals according to claim 1, characterized in that: The S3 method is as follows: S31 according to γ i and γ j The Q value of the excitation point is calculated by the frequency shift method, that is, Q i , and the Q value of the receiving point, that is, Q j : When calculating the Q value of the excitation point, it is assumed that the receiving point is located at the bottom of the ground near the excitation point. The f in formula (1) is r is the main frequency of the excitation signal; Assuming that the excitation point is located at the bottom of the near-surface formation opposite the receiving point, calculate the receiving point Q using the same calculation method. j ; S32 according to Q i and Q j Construct the initial compensation curve A corresponding to the excitation point and the receiving point i (f) and A j (f), called the initial compensation curve constructed by frequency attributes; S33 retains the steepness relationship of the initial compensation curve constructed by frequency attributes, and adopts the standard deviation normalization algorithm to realize the combination of frequency attributes and amplitude attributes in the compensation curve.

6. The attribute-combined near-surface attenuation compensation method for seismic signals according to claim 5, characterized in that: The method of step S33 is as follows: S331 Calculate the expected value μ of the compensation amount of the compensation curve of the excitation point and the receiving point i and μ j : Where Δf is the discrete signal frequency sampling interval, μ i and μ j are the expected values ​​of the compensation at the excitation point and the receiving point respectively; S332 calculates the standard deviation of the compensation amount of the excitation point and the receiving point compensation curve: Among them, σ i and σ j are the standard deviations of the compensation corresponding to the excitation point and the receiving point, respectively, x is a certain frequency value, x is the index value of a certain frequency in the effective frequency band of the seismic signal, and X is the index value corresponding to the upper limit value of the effective frequency band of the seismic signal; S333 by σ i and σ j Normalize the compensation curves of the excitation point and the receiving point respectively: ON i (f)=A i (f) / σ i ,AR j (f)=A j (f) / σ j (4) Among them, AR i (f)With AR f (f) The compensation curves of the excitation point and the receiving point after normalization; S334 applies the relative amplitude attenuation coefficient to the normalized compensation curve to obtain the final near-surface attenuation compensation curve: AF ij (f)=α i α j AR i (f)AR j (f) (5)。

Images