Seismic wavelet extraction method based on statistical parameter linear fitting

Abstract

The present application belongs to the technical field of oil and gas exploration seismic data processing, and particularly relates to a seismic wavelet extraction method based on statistical parameter linear fitting, which comprises inputting single-channel post-stack seismic data and performing time-frequency analysis processing; obtaining point spectrum of the seismic record based on the time-frequency analysis result; calculating statistical parameters of the energy spectrum; linearly fitting the statistical parameters of each time of the whole seismic record; constructing time-varying wavelet frequency domain analytical expression based on the statistical parameters and combining weighted exponential function; obtaining weighted exponential function spectrum based on the time-varying wavelet frequency domain analytical expression, and performing inverse Fourier transform on the weighted exponential function spectrum to obtain the corresponding time domain waveform. According to the statistical parameters of the seismic record point spectrum and the change trend thereof, the present application proposes a high-precision time-varying wavelet extraction technology, aiming to overcome the shortcomings of conventional wavelet extraction methods which are sensitive to noise and reflection coefficient, and providing guarantee for subsequent high-resolution processing of seismic data.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of oil and gas exploration seismic data processing, and particularly relates to a seismic wavelet extraction method based on statistical parameter linear fitting. BACKGROUND

[0002] High-resolution processing of seismic data has been a hot and difficult point in the research and production of the field of oil and gas exploration. In recent years, the targets of seismic exploration are gradually dominated by thin layers and small structures, which requires seismic data to have higher resolution, aiming at accurate positioning and fine description of thin layer information. Seismic wavelet is the basis of seismic data processing and interpretation, and if an error wavelet is used for seismic data inversion, the resolution of the inversion result will be directly reduced. Therefore, the research on time-varying wavelet extraction technology has important theoretical significance and practical application value.

[0003] Time-varying wavelet extraction methods are mainly divided into the following two categories: one is to divide the seismic record into several segments under the assumption of segmental stability, and to assume that the seismic record in each segment is approximately stable, and then to process by using conventional methods; the other is to directly extract the seismic wavelet from the non-stationary seismic record, and to realize point-by-point extraction of the seismic wavelet through spectrum simulation. Since the extraction accuracy of the former wavelet is easily affected by the length of the time window, the point-by-point extraction of the seismic wavelet in the time-frequency domain has become one of the research hotspots in recent years.

[0004] Spectral simulation technique is one of the commonly used methods to extract wavelet amplitude spectrum, mostly based on the empirical formula of seismic wavelet amplitude spectrum summarized by Rosa et al. (1991). In 1996, Zhao Bo et al. determined the minimum phase seismic wavelet from amplitude spectrum by using the Kolsky method. In 2011, Wang Deying et al. combined spectral simulation technique with autocorrelation method to estimate wavelet amplitude, and realized the adaptive selection of spectral simulation parameters. In 2014, Diao Rui et al. and Shang Xinmin et al. introduced the generalized S transform and improved S transform into spectral simulation method respectively, and proposed the time-frequency domain spectral simulation method, aiming to improve the ability of thin reservoir resolution of seismic data. Considering the time-varying characteristics of the phase of seismic wavelet, Dai et al. improved this kind of method. From 2014 to 2019, the research team of Dai Yongshou fitted the amplitude spectrum of seismic wavelet by using time-frequency domain spectral simulation technique, and then extracted the phase of time-varying wavelet by using the principle of local similarity, and proposed an extraction method which could better reflect the time-varying characteristics of seismic wavelet, and realized the estimation of time-varying mixed phase wavelet in frequency domain. Zhang and Fomel (2017) estimated time-varying wavelet by using local spectrum and autocorrelation theory and applied it to seismic data inversion, and obtained higher precision wave impedance inversion results. In 2019, Yao Zhenan et al. obtained high-resolution time-frequency analysis results by using basis pursuit spectral decomposition method, and combined with generalized seismic wavelet function to carry out point spectrum simulation processing. In 2020, Li Jing et al. extracted time-varying wavelet by using polynomial fitting smoothing method and constructed time-varying wavelet convolution model, and realized higher precision inversion results.

[0005] Most of these methods directly fit the spectrum of each moment of seismic record. Therefore, when the adjacent reflection coefficients have tuning effect, there is a certain error between the extraction result of wavelet and the theoretical value, which leads to that the thin layer information is not easy to be distinguished. SUMMARY

[0006] The purpose of the present application is to overcome the shortcomings of the conventional wavelet extraction method which is sensitive to noise and reflection coefficient, and to provide guarantee for subsequent high-resolution processing of seismic data.

[0007] The present application is realized by adopting the following technical solutions:

[0008] The wavelet extraction method based on linear fitting of statistical parameters comprises the following steps:

[0009] S1, input single-channel post-stack seismic data, and perform time-frequency analysis processing on the single-channel post-stack seismic data;

[0010] S2, obtain the point spectrum of the seismic record based on the time-frequency analysis result;

[0011] S3, obtain the energy spectrum based on the point spectrum of the seismic record;

[0012] S4, calculating statistical parameters of the energy spectrum, including the centroid frequency and the half-bandwidth;

[0013] S5, linearly fitting the centroid frequency and the half-bandwidth of each time of the whole seismic record;

[0014] S6, constructing a time-varying wavelet frequency domain analytical expression of each time based on the fitted centroid frequency and the half-bandwidth, and combining a weighted exponential function;

[0015] S7, obtaining a weighted exponential function spectrum based on the time-varying wavelet frequency domain analytical expression, and performing inverse Fourier transform on the weighted exponential function spectrum to obtain a corresponding time domain waveform.

[0016] Preferably, in the step S1, the time-frequency analysis method for the single-channel post-stack seismic record is short-time Fourier transform, wavelet transform or S transform.

[0017] Preferably, in the step S2, the method for obtaining the seismic record point is: based on the time-frequency analysis result, extracting a spectrum of each time along the time direction, and the spectrum is the point spectrum of the seismic record.

[0018] Preferably, in the step S3, the method for obtaining the energy spectrum is: taking each time as a sampling point, calculating the amplitude square of the point spectrum of the seismic record of each time, and the calculation result of the amplitude square is the energy spectrum of the corresponding seismic record sampling point.

[0019] Preferably, in the step S4, the centroid frequency is calculated by the following formula:

[0020]

[0021] wherein, f c(p) represents the centroid frequency, f represents the frequency, A(f) 2 represents the energy spectrum of the seismic record.

[0022] Preferably, in the step S6, the half-bandwidth is calculated by the following formula:

[0023]

[0024] wherein, represents the half-bandwidth; f c(p) represents the centroid frequency, f represents the frequency, A(f) 2 represents the energy spectrum of the seismic record.

[0025] Preferably, in the step S6, the time-varying wavelet frequency domain analytical expression is represented by:

[0026]

[0027] wherein A(f) FWE denotes a weighted exponential function spectrum, a denotes an amplitude parameter, f0 denotes a control band width factor, n denotes a control symmetry form factor, and exp denotes an exponential function with base of natural constant e.

[0028] Preferably, in the step S6, the control symmetry form factor n is obtained by the following formula:

[0029]

[0030] wherein f c(p) denotes a centroid frequency, denotes a half band width.

[0031] Preferably, in the step S6, the control band width factor f0 is obtained by the following formula:

[0032]

[0033] wherein f c(p) denotes a centroid frequency, denotes a half band width.

[0034] The present application has the beneficial technical effects:

[0035] The present application provides a seismic wavelet extraction method based on statistical parameter linear fitting, which overcomes the false appearance of up and down jitter of statistical parameter curve of seismic record due to tuning effect, accurately extracts the real centroid frequency and half band width at each moment by linear fitting the trend of statistical parameter, and ensures the accurate extraction of time-varying wavelet; further, the statistical parameter is extracted by energy spectrum, which has higher calculation accuracy and better noise resistance than amplitude spectrum. BRIEF DESCRIPTION OF DRAWINGS

[0036] Figure 1 is a basic implementation flowchart of the present technical solution;

[0037] Figure 2 is an amplitude and energy spectrum noise resistance ability number model test comparison chart;

[0038] Figure 3 is a centroid frequency error distribution comparison chart of amplitude spectrum and energy spectrum;

[0039] Figure 4 is a non-steady state seismic data model chart;

[0040] Figure 5 is a statistical parameter chart of each moment of seismic record;

[0041] Figure 6 is a fitted statistical parameter chart;

[0042] Figure 7 Time-varying wavelet matrix diagram of time-varying wavelet extraction result;

[0043] Figure 8 Actual result and theoretical result comparison diagram of time-varying wavelet extraction;

[0044] Figure 9 Wavelet extraction effect diagram for actual post-stack data;

[0045] Figure 10 Actual seismic data deconvolution effect;

[0046] Figure 11 Spectrum comparison diagram before and after actual seismic data processing; DETAILED DESCRIPTION

[0047] In order to make the objects, technical solutions and advantages of the present application clearer, the technical solutions in the present application will be described clearly and completely below in conjunction with the drawings in the present application. Obviously, the described embodiments are only some of the embodiments of the present application, rather than all the embodiments.

[0048] Therefore, the following detailed description of the present application provided in the drawings is not intended to limit the scope of the claimed present application, but only represents selected embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the present application.

[0049] Embodiment 1

[0050] This embodiment discloses a seismic wavelet extraction method based on statistical parameter linear fitting, which is a basic embodiment of the present application, as shown in Figure 1 The method comprises the following steps:

[0051] S1, input single-channel post-stack seismic data, and perform time-frequency analysis processing on the single-channel post-stack seismic data;

[0052] S2, obtain a point spectrum of the seismic record based on the time-frequency analysis result;

[0053] S3, obtain an energy spectrum based on the point spectrum of the seismic record;

[0054] S4, calculate statistical parameters of the energy spectrum, including a centroid frequency and a half-bandwidth;

[0055] S5, linearly fit the centroid frequency and the half-bandwidth of each time of the whole seismic record;

[0056] S6, based on the fitted centroid frequency and half-bandwidth, a time-varying wavelet frequency domain analytical expression is constructed by combining a weighted exponential function, that is, a wavelet frequency domain analytical expression at each time;

[0057] S7, based on the time-varying wavelet frequency domain analytical expression, a weighted exponential function spectrum (time-varying wavelet) is obtained, and inverse Fourier transform is performed on the weighted exponential function spectrum to obtain a corresponding time-domain waveform.

[0058] The anti-noise property in the technical solution refers to adding random noises of different degrees in the energy spectrum to observe the influence on the extraction of statistical parameters. Figure 2 As shown in the figure, the figure is a comparison diagram of amplitude and energy spectrum anti-noise ability number model test, the figure is tested by using the Ricker wavelet commonly used in seismic data processing (main frequency is 30 Hz), (2a) and (2b) in the figure are test diagrams of the energy spectrum, the amplitude spectrum and the energy spectrum of the Ricker wavelet, wherein the solid line is under the condition of no noise, and the dashed line is under the condition of containing medium intensity (signal-to-noise ratio SNR=6dB); under the condition of containing noise, the error distribution of the centroid frequency is calculated by selecting the effective frequency band range (0-80Hz) of the amplitude spectrum and the energy spectrum respectively. As shown in Figure 2 It can be known that under the condition of containing noise, the low frequency and high frequency parts of the amplitude spectrum (2a) are disturbed, while the curve of the energy spectrum (2b) is smoother and closer to the frequency spectrum distribution under the condition of no noise, because the energy spectrum obtained by calculating the square of the amplitude has a certain ability to suppress noise. Figure 3 As shown in (3a) and (3b) in the figure, (3a) and (3b) are respectively the centroid frequency error distribution diagram of the amplitude spectrum and the centroid frequency error distribution diagram of the energy spectrum, and by comparing Figure 3 (3a) and (3b) in the figure, it can be known that because the amplitude spectrum is easily affected by noise, even if the centroid frequency (3a) extracted by selecting the advantage frequency band of the amplitude spectrum under the condition of containing noise still has a large error. Under the same condition, the result based on the energy spectrum (3b) is more accurate and objective. In summary, the application proposes to calculate the statistical parameters of the time-varying wavelet by using the energy spectrum, so as to ensure the accuracy and stability of the calculation result.

[0059] Further, the technical solution obtains a weighted exponential function spectrum (time-varying wavelet) by establishing a time-varying wavelet frequency domain analytical expression, as shown in Figure 4 The figure is a non-stationary seismic data model diagram, (4a) in the figure is a non-stationary seismic record model constructed by using a time-varying wavelet and a random reflection coefficient, wherein the sampling interval is 1ms, the propagation time is 1s, the main frequency of the time-varying wavelet is linearly reduced from 38Hz to 32Hz, and the half-bandwidth (half-bandwidth) is linearly reduced from 13.6 to 11.4; (4b) in the figure is a time-frequency analysis result obtained by using synchronous squeezing transformation.

[0060] Embodiment 2

[0061] The embodiment discloses a seismic wavelet extraction method based on statistical parameter linear fitting, as a preferred embodiment of the application, comprising the following steps:

[0062] S1, input single-channel post-stack seismic data, and performing time-frequency analysis processing on the single-channel post-stack seismic data; wherein the method for performing time-frequency analysis processing on the single-channel post-stack seismic data is short-time Fourier transform, wavelet transform or S transform.

[0063] S2, acquiring a point spectrum of a seismic record based on a time-frequency analysis result; wherein the method for acquiring the point of the seismic record is as follows: based on the time-frequency analysis result, a frequency spectrum of each time is extracted along a time direction, and the frequency spectrum is the point spectrum of the seismic record.

[0064] S3, acquiring an energy spectrum based on the point spectrum of the seismic record; wherein the method for acquiring the energy spectrum is as follows: each time is taken as a sampling point, and the amplitude square of the point spectrum of the seismic record at each time is calculated, and the calculation result of the amplitude square is the energy spectrum of the corresponding seismic record sampling point.

[0065] S4, calculating statistical parameters of the energy spectrum, including a centroid frequency and a half-bandwidth.

[0066] S5, linearly fitting the centroid frequency and the half-bandwidth of each time of the whole seismic record.

[0067] S6, constructing a time-varying wavelet frequency domain analytical expression based on the fitted centroid frequency and the half-bandwidth, and combining a weighted exponential function, that is, a wavelet frequency domain analytical expression of each time.

[0068] S7, acquiring a weighted exponential function spectrum based on the time-varying wavelet frequency domain analytical expression, and performing inverse Fourier transform on the weighted exponential function spectrum to obtain a corresponding time domain waveform.

[0069] Embodiment 3

[0070] The embodiment discloses a seismic wavelet extraction method based on statistical parameter linear fitting, as a basic embodiment of the application, comprising the following steps:

[0071] S1, input single-channel post-stack seismic data, and performing time-frequency analysis processing on the single-channel post-stack seismic data; wherein the method for performing time-frequency analysis processing on the single-channel post-stack seismic data is short-time Fourier transform, wavelet transform or S transform.

[0072] S2, acquiring a point spectrum of a seismic record based on a time-frequency analysis result; wherein the method for acquiring the point of the seismic record is as follows: based on the time-frequency analysis result, a frequency spectrum of each time is extracted along a time direction, and the frequency spectrum is the point spectrum of the seismic record.

[0073] S3, obtaining the energy spectrum based on the point spectrum of the seismic record; wherein, the method for obtaining the energy spectrum is: taking each moment as a sampling point, calculating the square of the amplitude of the point spectrum of the seismic record at each moment, and the result of the square of the amplitude calculation is the energy spectrum of the corresponding seismic record sampling point.

[0074] S4 calculates the statistical parameters of the energy spectrum, including the centroid frequency and half-bandwidth. Specifically:

[0075] The centroid frequency is calculated using the following formula:

[0076]

[0077] The half-bandwidth is calculated using the following formula:

[0078]

[0079] Among them, f c(p) Let f represent the frequency of the centroid, and A(f) represent the frequency. 2 This represents the energy spectrum of the earthquake record. This indicates the half-bandwidth.

[0080] like Figure 5 The statistical parameters of the energy spectrum at each moment of the seismic record are shown, where (5a) is the unfit centroid frequency and (5b) is the unfit half-bandwidth. Along Figure 4 In section (4b), the time-frequency analysis results are used to extract the spectrum of each moment in the time direction and obtain its energy spectrum. Then, the centroid frequency and half-bandwidth at that moment are calculated using the above formula. In summary, the statistical parameters at each moment are calculated, and the results are as follows: Figure 5 The centroid frequency (5a) and half-bandwidth (5b) are shown in the figure. Figure 5 As shown in the figure, the unfitted statistical parameters exhibit fluctuations, but theoretically, the dominant frequency and half-bandwidth of the wavelet should decrease linearly. This is due to the abnormal increase in statistical parameters in some regions caused by the tuning effect of adjacent reflection coefficients. Therefore, this invention proposes linearly fitting the trend of statistical parameters to determine the true centroid frequency and half-bandwidth at each moment of the seismic record.

[0081] S5, linear fitting of the centroid frequency and half-bandwidth of each moment of the whole seismic record. According to the absorption attenuation theory, the energy of high frequency component attenuates faster during the propagation of the wavelet, so the dominant frequency of the wavelet moves to the low frequency direction, and the frequency bandwidth becomes narrower. In theory, both the centroid frequency and the half-bandwidth will decrease. However, due to the tuning effect of adjacent reflection coefficients, that is, the influence of the complex wave, individual statistical parameters may increase. The complex wave refers to the wavelet waveform that appears aliasing when the interval between adjacent reflection coefficients is small, forming a complex wave. Therefore, the trend of the centroid frequency and the half-bandwidth needs to be linearly fitted to determine the true statistical parameters at this moment rather than the values affected by the complex wave.

[0082] As shown in the fitted statistical parameter diagram shown in Figure 6 . In (6a), the solid line is the theoretical centroid frequency, and the dashed line is the equivalent centroid frequency after linear fitting; in (6b), the solid line is the theoretical half-bandwidth, and the dashed line is the equivalent half-bandwidth after linear fitting. Figure 6 It can be seen that the results of the linear fitting of the present application are basically consistent with the theoretical parameters, which illustrates the effectiveness and accuracy of the present application.

[0083] S6, based on the fitted centroid frequency and half-bandwidth, a time-varying wavelet frequency domain analytical expression is constructed by combining a weighted exponential function (FWE), that is, a wavelet frequency domain analytical expression at each moment, which is specifically expressed as:

[0084]

[0085] Wherein, A(f) FWE represents the spectrum of the weighted exponential function; a represents the amplitude parameter; f0 represents the control band width factor, which is obtained by formula ; n represents the control symmetry factor, which is obtained by formula ; exp represents the exponential function with natural constant e as the base. The calculation formulas of the control symmetry factor n and the control band width factor f0 are obtained by the following reasoning, and the time-varying wavelet frequency domain analytical expression is respectively brought into the centroid frequency calculation formula and the half-bandwidth calculation formula, so that:

[0086] Therefore, we have:

[0087] Based on the above formula, the seismic wavelet is extracted, Figure 10 , which is the actual seismic data deconvolution effect of the present application. Since the actual seismic data cannot determine whether the wavelet extraction is accurate or not, the seismic wavelet extracted by the present application is used in the reflection coefficient inversion. Figure 10 a is the reflection coefficient obtained by using the time-varying wavelet extracted by the present application, Figure 10b is the reflection coefficient estimated using well logging data. For example... Figure 10 As shown, the deconvolution results obtained based on the time-varying wavelet extracted by this invention are basically consistent with the trend of the reflection coefficient extracted from the well data, and the similarity is high. The correlation coefficient (c = 0.68) between the deconvolution results (10a) and the reflection coefficient of the well data (10b) was calculated, and the qualitative and quantitative analysis demonstrated the accuracy and effectiveness of the wavelet extraction by this invention.

[0088] S7. Based on the frequency domain analytical expression of the time-varying wavelet, the spectrum of the weighted exponential function is obtained, and the inverse Fourier transform is performed on the spectrum of the weighted exponential function to obtain the corresponding time-domain waveform. That is, the time-domain waveform of the time-varying wavelet is extracted using the inverse Fourier transform, such as... Figure 7 The image shows the results of time-varying wavelet extraction. Figure 6 Substituting the statistical parameters at each time step after fitting into the above formula yields the corresponding wavelet frequency domain analytical expression. Then, the inverse Fourier transform is used to extract the time-domain waveform at that moment, as shown in the figure. Figure 7 and Figure 8 As shown. It is worth noting that, as Figure 7 Each column in the time-varying wavelet matrix shown represents the seismic wavelet at that moment, with the dominant frequency shifting towards lower frequencies and the bandwidth gradually narrowing, consistent with the dynamic attenuation characteristics of wavelets during propagation. Furthermore, as... Figure 8 As shown, wavelets at four time points in the time-varying wavelet matrix are extracted for specific analysis. In 8a, 8b, 8c, and 8d, the dashed lines represent the wavelet spectra (top) and time-domain waveforms (bottom) extracted from the seismic record at 2.189, 2.192, 2.450, and 2.800 s, respectively, while the solid lines represent the theoretical wavelet spectrum and waveform. Figure 8 It can be seen that the time-varying wavelet proposed in this invention is in high agreement with the theoretical value and has a small error. Figure 8 b and Figure 8 The wavelet extraction results show that even when there is a tuning effect between adjacent reflection coefficients, the present invention can still accurately extract the wavelet at that moment, thereby improving the ability of seismic data to distinguish thin layers.

[0089] Furthermore, Figure 9 This is a diagram showing the actual wavelet extraction effect of the post-stack data in this invention. Figure 9 a represents the well-side data of the post-stack seismic data along the road; Figure 9 b represents the time-varying wavelet matrix extracted using this invention, where each column represents the seismic wavelet at that moment. For example... Figure 9As shown in b, with the increase of propagation time, the main frequency of the wavelet moves to low frequency direction, the frequency band becomes narrow, and the time domain waveform becomes wide, which reflects the non-stationary characteristics of the actual seismic data. Meanwhile, the time-varying wavelet extracted by the present application has the characteristics of weak sidelobe energy and high resolution. In the seismic data processing, the narrower the main lobe width of the seismic wavelet is, the weaker the sidelobe amplitude is, and the higher the time resolution is, so that the error caused by the sidelobe energy of the wavelet can be avoided in the seismic data inversion.

[0090] Further, Figure 11 The figure shown is the spectrum comparison chart of the actual seismic data before and after the processing of the present application. Figure 11 The middle solid line is the spectrum of the seismic record before the processing, and the dotted line is the spectrum after the processing. Figure 11 It can be known that the result after the processing of the present application better protects the original frequency components, and the low frequency and high frequency energy are both lifted. According to the existing research, compared with the high frequency information, the low frequency information has a greater contribution to the relative frequency width, and is more difficult to recover. The present application improves the accuracy of the wavelet extraction, and lifts the low frequency energy of the seismic record when used in the inversion, so as to improve the resolution of the inversion result. The actual seismic data application result is consistent with the theoretical model test, which again proves the accuracy and practical application value of the time-varying wavelet extracted by the present application.

Claims

1. A seismic wavelet extraction method based on linear fitting of statistical parameters, characterized in that, The method comprises the following steps: S1, inputting single-channel stacked seismic data, and performing time-frequency analysis on the single-channel stacked seismic data; S2, obtaining a point spectrum of a seismic record based on a time-frequency analysis result; S3, obtaining an energy spectrum based on the point spectrum of the seismic record, that is, taking each time as a sampling point, calculating an amplitude square of the point spectrum of the seismic record at each time, and the calculation result of the amplitude square is an energy spectrum of a corresponding sampling point of the seismic record; S4, calculating statistical parameters of the energy spectrum, including a centroid frequency and a half-bandwidth; A formula for calculating the centroid frequency is represented as: ； A formula for calculating the half-bandwidth is represented as: ； wherein, denotes the centroid frequency, denotes the frequency, denotes the energy spectrum of the seismic record, denotes the half-bandwidth; S5, linearly fitting the centroid frequency and the half-bandwidth of each time of the whole seismic record; S6, constructing a time-varying wavelet frequency domain analytical expression based on the fitted centroid frequency and the half-bandwidth, and combining a weighted exponential function, that is, a wavelet frequency domain analytical expression at each time; wherein, the time-varying wavelet frequency domain analytical expression is represented as: ; wherein, represents a weighted exponential function spectrum; represents an amplitude parameter; represents a control band width factor; represents a control symmetry form factor; represents an exponential function with a natural constant as a base; S7, obtaining a weighted exponential function spectrum based on the time-varying wavelet frequency domain analytical expression, and performing inverse Fourier transform on the weighted exponential function spectrum to obtain a corresponding time domain waveform.

2. The method for extracting seismic wavelet based on statistical parameter linear fitting according to claim 1, wherein, In the step S1, a method for performing time-frequency analysis on the single-channel stacked seismic data is short-time Fourier transform, wavelet transform or S transform.

3. The method of claim 1, wherein the statistical parameter is a correlation coefficient. In the step S2, a method for obtaining a point of a seismic record is as follows: based on a time-frequency analysis result, a frequency spectrum at each time is extracted along a time direction, and the frequency spectrum is the point spectrum of the seismic record.

4. The method of claim 1, wherein the statistical parameter is a correlation coefficient. In the step S6, the control symmetry pattern factor is obtained by the following equation : ； wherein, denotes the center frequency, denotes the half-bandwidth.

5. The method of claim 1, wherein the statistical parameter is a correlation coefficient. In the step S6, the control band width factor is obtained by the following equation : ； wherein, denotes the center frequency, denotes the half-bandwidth.

Images