Abnormal delay detection method for first arrival automatic picking based on azimuthal anisotropy control

Abstract

The present invention provides a method for detecting abnormal delay of first arrival automatic picking based on azimuth anisotropy control, comprising: automatically picking first arrivals for all seismic data; iteratively calculating the delay of the automatically picked first arrivals to obtain the shot delay time and formation velocity; eliminating the influence of the shot delay time and offset on the first arrival time; spatially arranging the processed first arrival picking points in a polar coordinate system according to the azimuth and offset of each detection point relative to the shot point; dividing the first arrival picking points into azimuth intervals according to the complexity of the terrain; performing distance-based outlier detection on the first arrival time in each azimuth interval under the control of a sliding time window, detecting and deleting the first arrival corresponding to the outlier; and outputting a final first arrival picking result. The present invention has good noise interference resistance, can automatically detect and eliminate abnormal values ​​in the automatically picked first arrivals, and effectively improve the efficiency and accuracy of the first arrival automatic picking.

Description

Technical Field

[0001] The present invention relates to the technical field of oil and gas exploration seismic data processing, and in particular to a first arrival automatic picking anomaly delay detection method based on azimuthal anisotropy control. Background Art

[0002] With the continuous advancement of oil and gas exploration, single-point high-density seismic exploration is increasingly being used. With the acquisition of this type of data, conventional seismic data first-break picking methods face numerous challenges in practical application. These single-point high-density seismic data often have low signal-to-noise ratios, making automatically picked first-breaks often infeasible for direct application. First-breaks are crucial in static correction processing, particularly first-break tomographic static correction, which has become a common technique in seismic data processing. Conventional first-break picking methods rely on software for automatic picking, followed by manual modification based on the automatically picked data to eliminate outliers. This strategy is difficult to implement due to the enormous workload. Automatic picking methods are even more difficult to achieve ideal first-break picking results in data with low signal-to-noise ratios. While some missing first-breaks in automatic first-break picking may not significantly impact the static correction results, the presence of erroneous first-breaks can lead to errors in the static correction calculations. Therefore, developing algorithms for automatically removing anomalous first-break picking is of great significance. Current research in this area primarily relies on least-squares fitting methods to eliminate anomalous first-breaks.

[0003] A Chinese patent application with application number CN201711009894.0 describes a method for rapidly removing earthquake first-arrival outliers. The method includes: A. inputting raw first-arrival data; B. plotting a central tendency curve for effective first-arrival data to obtain first-arrival data; C. classifying and sorting the first-arrival data to obtain a first-arrival sequence; D. obtaining a second-arrival sequence; E. obtaining a first-arrival difference sequence based on the differences between consecutive pairwise first-arrivals in the second first-arrival sequence, and obtaining a first singular value editing threshold range based on the mean and standard deviation of the first-arrival difference sequence and its multiple coefficient values; F. determining a second editing threshold range and the number of editing iterations; G. screening the first-arrival data based on the second editing threshold range and the number of editing iterations to obtain second-arrival data. This method can increase the speed of removing first-arrival outliers, has low effective first-arrival loss, and can improve the efficiency of removing first-arrival outliers while ensuring first-arrival redundancy.

[0004] In the Chinese patent application with application number: CN201811388862.0, a two-step method and system for eliminating abnormal earthquake first arrivals is involved. The method includes: obtaining an SPS file of a project to be processed; obtaining an earthquake first arrival data file based on the earthquake first arrival time of each shot in the SPS file; collecting and generating earthquake first arrival data and corresponding offset data for each receiving arrangement of each shot according to positive and negative offset distances; obtaining a fitting formula for earthquake first arrival based on the offset data and the earthquake first arrival data, calculating the earthquake first arrival fitting time difference of each effective seismic trace, comparing the earthquake first arrival fitting time difference with a predetermined threshold, and eliminating abnormal earthquake first arrivals in the earthquake first arrival data file based on the comparison result to obtain preliminary earthquake first arrival data; collecting and generating earthquake first arrival data and corresponding offset data for all receiving arrangements of each shot according to positive and negative offset distances; repeating the above process to obtain corrected earthquake first arrival data; performing first arrival inversion modeling and calculating residual static correction based on the corrected earthquake first arrival data.

[0005] Chinese patent application number CN201910664529.6 discloses a method and apparatus for identifying abnormal first arrivals. The method includes determining the energy ratio of each sampling point in each seismic trace in acquired single-shot seismic data, determining a first arrival position curve based on the energy ratio, and then averaging the time differences of first arrivals of adjacent traces of the single-shot seismic data based on the first arrival position curve. Using a permutation and combination method, the method detects anomalies in the first arrivals of the trace to be identified, identifies and removes discontinuous abnormal first arrivals, and selects the segment with the largest number of arrivals from the continuous first arrival position curve as the reference first arrival. Using a sliding linear fit of the arrival times, the method calculates the average velocity corresponding to the first arrival at one end of the reference first arrival, the average time difference of the first arrival before and after correction, and the average time difference of the adjacent segments to be identified. Continuous abnormal first arrivals are identified based on the absolute value of the difference between the two average values. This solution improves the quality of automatically picking and identifying abnormal first arrivals.

[0006] The Chinese patent application with application number CN201410743510.8 discloses a method and system for removing abnormal first arrivals. The method includes: acquiring seismic data; obtaining shot offsets and first arrival data from the seismic data; gridding the shot offsets and first arrival data from the seismic data; setting a threshold for the number of first arrivals; and removing abnormal first arrivals from the gridded shot offsets and first arrival data based on the threshold for the number of first arrivals. This embodiment of the application provides a method and system for removing abnormal first arrivals. This method utilizes statistical data on the shot offsets and first arrivals of all shots to grid the shot offsets and first arrivals, deleting all first arrivals in grids where the number of first arrivals is less than a preset threshold, thereby improving the accuracy of removing abnormal first arrivals.

[0007] The above existing technologies are all significantly different from the present invention and fail to solve the technical problem we want to solve. Therefore, we have invented a first arrival automatic picking abnormal delay detection method based on azimuthal anisotropy control. Summary of the Invention

[0008] The present invention provides a new method that can automatically detect and eliminate outliers in the first-arrival picking of massive seismic data. The method does not require human intervention and automatically completes the detection and elimination of outlier picking values ​​by statistically analyzing the spatial anisotropy of the first-arrival picking time points, so that the first-arrival automatic picking values ​​meet the requirements of static correction calculations.

[0009] The object of the present invention can be achieved by the following technical measures: a first arrival automatic picking abnormal delay detection method based on azimuthal anisotropy control, the first arrival automatic picking abnormal delay detection method based on azimuthal anisotropy control comprising:

[0010] Step 1: Automatically pick the first arrival of all seismic data;

[0011] Step 2: Perform delay time iteration calculation on the automatically picked first arrival to obtain the shot point delay time and formation velocity;

[0012] Step 3: Eliminate the effects of shot point delay time and offset distance in the automatically picked first arrival time, and only retain the effect of receiver point delay time on the first arrival time;

[0013] Step 4: Arrange the first arrival points processed in step 3 in a polar coordinate system according to the azimuth and offset of each detection point relative to the shot point.

[0014] Step 5: Divide the first arrival point picked in step 4 into azimuth intervals according to the terrain complexity;

[0015] Step 6: Using sliding window control, perform distance-based outlier detection on the first arrival time in each azimuth interval, and detect and delete the first arrival corresponding to the outlier;

[0016] Step 7: Output the final first arrival picking result.

[0017] The purpose of the present invention can also be achieved by the following technical measures:

[0018] In step 1, before the first arrival automatic picking, the seismic data is subjected to conventional denoising processing to improve the first arrival signal-to-noise ratio and enhance the automatic picking accuracy.

[0019] In step 2, the delayed time is iterated using the first arrivals picked automatically. Before the iterative calculation, the initial formation velocity corresponding to the shot point is assigned an initial value when the shot point is delayed. The formation velocity is obtained by fitting the existing first arrivals, and its value is the inverse of the slope of the fitting line. The shot point delay is taken as half of the intercept of the fitting line.

[0020] In step 2, when performing delay iteration, it is assumed that the first arrival wave is a refracted wave and there is a single refraction layer underground. Let S be the shot point, R be the receiver point, and the surface thickness of point S and point R be Z respectively. s 、Z r , the critical angle is α, v1 and v2 are the velocities of the upper and lower layers of the refraction surface respectively, then the first arrival time of the refracted wave received by point R when point S is excited is:

[0021]

[0022] The delays for defining shot points and receiver points are:

[0023]

[0024]

[0025] set up Then formula (1) can be written as:

[0026] T SR =t S +t AB +t R (4)

[0027] An equation such as formula (4) can be established for each first arrival time. Assuming that the number of shot points in a work area is N, the number of receiver points is M, and each shot has W channels, then the number of equations that can be established in the work area is N*W, and the number of unknowns to be calculated is approximately 2*N+M. Therefore, in the case of high shot channel density, the number of equations is much larger than the number of unknowns. The equation can be solved using the Gauss-Seidel iterative algorithm to calculate the delay time and formation velocity corresponding to each receiver point and shot point.

[0028] In step 3, the delay time of the corresponding shot point is subtracted from each picked first arrival time, and the effect of the offset distance on the delay time is eliminated according to the formation velocity.

[0029] In step 3, for each picked first arrival time, the influence of the shot point delay time and the formation velocity is eliminated to obtain the temporary detection point delay time corresponding to each first arrival time:

[0030] t′ SR =T SR -t S -t AB (5)

[0031] After the delay time decomposition in step 2, t S , t ABRelatively accurate values ​​can be obtained. According to formula (5), the first arrival automatically picked up in step 1 is transformed to the detection point delay time t′ corresponding to the current channel. SR .

[0032] In step 4, in order to make better use of the characteristic that the delay time of the nearby physical point is closer, and at the same time take into account the spatial anisotropy of the actual data, with the shot point as the center and the offset as the radius, the first arrival time that has eliminated the influence of the shot point delay time and offset distance is calculated according to the azimuth and offset distance. SR Arrange the space.

[0033] In step 5, the first arrival points in the polar coordinate system are divided into several main azimuth intervals according to the surface complexity such as surface velocity and undulation characteristics.

[0034] In step 5, the detection point in step 4 is delayed by time t′ SR According to the complexity of the terrain, it is divided into several large directions; the number of divisions is related to the complexity of the anisotropy of the actual stratum, and is usually divided into 2-6 large ranges; general data can be processed well by dividing it into 4 ranges; data with smooth terrain and weak stratum anisotropy are divided into 2 ranges; in the case of particularly complex terrain, the direction division can adopt a non-uniform division method that conforms to the law of terrain distribution.

[0035] In step 6, a distance-based outlier detection method with sliding window control is used for the time within each azimuth. In this distance-based outlier detection algorithm, the definition of distance includes the offset difference and the time difference, which is a value calculated by the combined effect of the two.

[0036] In step 6, a distance-based outlier detection method is used on the data, using a sliding window with a window size of 200-500 meters.

[0037] The present invention provides a method for detecting abnormal delays in automatic first-arrival picking based on azimuthal anisotropy control, which is mainly used for automatic first-arrival picking of massive seismic data. In view of the difficulty in monitoring automatic first-arrival picking in the case of massive seismic data, the present invention proposes an abnormal delay detection algorithm using azimuthal anisotropy control to achieve automatic elimination of abnormal first-arrival picking values. Outlier detection algorithms have been widely used in intrusion detection, financial fraud, stock analysis and other fields, and have good practical application effects. The algorithm automatically detects abnormal points based on statistical laws. Converting the automatically picked first arrival into the delay time of the detection point greatly reduces the difficulty of automatic identification of abnormal first arrivals. The present invention has good applicability. It can be seen from the analysis of the actual seismic data processing results that after processing using the method of the present invention, the problem of automatic detection of first-arrival picking anomalies in massive seismic data is well solved, and abnormal values ​​in the first-arrival picking results are effectively eliminated. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] Figure 1 It is a flowchart of a specific embodiment of the method for detecting abnormal delay of first arrival automatic picking based on azimuthal anisotropy control of the present invention;

[0039] Figure 2 A schematic diagram of a hypothetical refractive layer model used in the present invention;

[0040] Figure 3 This is a diagram showing anisotropic distribution of temporary detection point delay in a specific embodiment of the present invention;

[0041] Figure 4 This is a diagram showing the effect of temporary detection points delayed according to offset distances after division in each direction in a specific embodiment of the present invention;

[0042] Figure 5 This is a diagram showing the effect of distance-based outlier detection when temporary detection points arranged according to offset distances are delayed after each direction is divided in a specific embodiment of the present invention;

[0043] Figure 6 is a diagram showing anisotropy distribution of the temporary detection point delay after processing in a specific embodiment of the present invention;

[0044] Figure 7 An actual earthquake profile diagram in a specific embodiment of the present invention;

[0045] Figure 8 A commercial software picking up the first arrival effect diagram in a specific embodiment of the present invention;

[0046] Figure 9 This is a diagram showing the initial arrival effect after processing in a specific embodiment of the present invention;

[0047] Figure 10This is a diagram showing the effect of distance-based outlier detection when temporary detection points arranged according to offset distances are delayed after each direction is divided in a specific embodiment of the present invention;

[0048] Figure 11 Schematic diagram of the first arrival picking effect at different stages in a specific embodiment of the present invention. DETAILED DESCRIPTION

[0049] It should be noted that the following detailed descriptions are exemplary and intended to provide further explanation of the present invention. Unless otherwise specified, all technical and scientific terms used herein have the same meaning as commonly understood by those skilled in the art to which the present invention belongs.

[0050] It should be noted that the terms used herein are only for describing specific embodiments and are not intended to limit the exemplary embodiments according to the present invention. As used herein, unless the context clearly indicates otherwise, the singular form is intended to include the plural form. In addition, it should be understood that when the terms "comprise" and / or "include" are used in this specification, they indicate the presence of features, steps, operations and / or combinations thereof.

[0051] Aiming at the problem of automatic elimination of first-arrival picking anomalies in the case of massive low-signal-to-noise ratio seismic data, the present invention proposes a first-arrival automatic picking anomaly delay detection method based on azimuthal anisotropy control, which realizes the automatic detection and elimination of abnormal first arrivals. The core contents of this method are twofold. One is to arrange the first-arrival picking results in an azimuthal anisotropic manner, and the other is to automatically detect outliers in a local space-time window. The anisotropic distribution of the first-arrival data mainly refers to the azimuth of the corresponding first-arrival detection point and shot point. In order to better highlight the anisotropic differences in the first-arrival data, it is necessary to eliminate the influence of the shot point delay time and offset distance before performing outlier detection.

[0052] like Figure 1 As shown, Figure 1 The flowchart of the first arrival automatic picking abnormal delay detection method based on azimuthal anisotropy control of the present invention is as follows:

[0053] Step 101, using an automatic picking method to pick the first arrival of all seismic data;

[0054] Step 102, performing delay time iterative calculation on the automatically picked first arrival to obtain the shot point delay time and formation velocity;

[0055] Step 103, eliminating the influence of the shot point delay time and the offset distance in the automatically picked first arrival time, and only retaining the influence of the detection point delay time on the first arrival time;

[0056] Step 104, spatially arranging the first arrival points processed in step 3 in a polar coordinate system according to the azimuth and offset of each detection point relative to the shot point;

[0057] Step 105, dividing the first arrival point picked in step 104 into azimuth intervals according to the terrain complexity;

[0058] Step 106 , using sliding window control, performs distance-based outlier detection on the first arrival time in each azimuth interval, and detects and deletes the first arrival corresponding to the outlier;

[0059] Step 107: Output the final first arrival picking result.

[0060] Through the above specific steps, the difficult problem of automatic identification of outliers in the automatic picking of first arrival of seismic data is solved.

[0061] The following are several specific embodiments of the present invention:

[0062] Example 1

[0063] In a specific embodiment 1 of the present invention, the first arrival automatic picking abnormal delay detection method based on azimuthal anisotropy control includes the following steps:

[0064] Step 1: Use the automatic picking method to pick the first arrival of all seismic data.

[0065] Before automatic first-arrival picking, conventional denoising processing is performed on the seismic data to improve the first-arrival signal-to-noise ratio of the seismic data and enhance the accuracy of automatic picking.

[0066] Step 2: Perform a time delay iteration on the automatically picked first arrivals to obtain the shot delay time and formation velocity. A time delay iteration algorithm is used to calculate the delay time and formation velocity corresponding to each shot point. The time delay iteration algorithm uses the Gauss-Seidel iteration method for solution.

[0067] Use the automatically picked first arrivals for iterative calculations of delayed time. Before iterative calculations, assign an initial value to the formation velocity corresponding to the shot point's delayed time. The formation velocity can be obtained by fitting the existing first arrivals, and its value is the inverse of the slope of the fitted line. The shot point delayed time is taken as half the intercept of the fitted line.

[0068] When performing delay time iteration, it is assumed that the first arrival wave is a refracted wave and there is a single refractive layer underground, such as Figure 2 As shown in the figure: Let S be the shot point, R be the receiver point, and the surface thickness of point S and point R are Z respectively. s 、Z r , the critical angle is α, v1 and v2 are the velocities of the upper and lower layers of the refraction surface respectively, then the first arrival time of the refracted wave received by point R when point S is excited is:

[0069]

[0070] The delays for defining shot points and receiver points are:

[0071]

[0072]

[0073] set up Then formula (1) can be written as:

[0074] T SR =t S +t AB +t R (4)

[0075] An equation such as Equation (4) can be established for each first arrival time. Assuming that the number of shot points in a work area is N, the number of receiver points is M, and each shot has W channels, then the number of equations that can be established in the work area is N*W, and the number of unknowns to be calculated is approximately 2*N+M. Therefore, in the case of high channel density, the number of equations is much greater than the number of unknowns. The equation can be solved using the Gauss-Seidel iterative algorithm to calculate the delay time and formation velocity corresponding to each receiver point and shot point.

[0076] Step 3: Subtract the delay time of the corresponding shot point from each picked first arrival time, and eliminate the effect of the offset distance on the delay time according to the formation velocity.

[0077] After obtaining the shot point delay time and the corresponding formation velocity of the shot point through the iterative calculation of the first arrival delay time, the influence of the shot point delay time and the offset distance is eliminated. The offset distance effect can be calculated as a time correction value by iteratively calculating the velocity and the known offset distance. In the iterative calculation of the delay time, the detection point delay time can be obtained. However, in actual use, the detection point delay time obtained by the iterative calculation of the delay time cannot be used in subsequent steps. Instead, it is necessary to use formula (5) to eliminate the influence of the shot point delay time and the offset distance from the first arrival time obtained in step 1, and the remaining time is used as the temporary delay time of the detection point corresponding to the current first arrival.

[0078] For each picked first arrival time, eliminate the influence of shot point delay time and formation velocity, and obtain the temporary detection point delay time corresponding to each first arrival time:

[0079] t′ SR =T SR -t S -t AB (5)

[0080] After the delay time decomposition in step 2, t S , tAB Relatively accurate values ​​can be obtained. According to formula (5), the first arrival automatically picked up in step 1 can be transformed to the detection point delay time t′ corresponding to the current channel. SR .

[0081] Step 4: Arrange the first arrival points processed in step 3 in a polar coordinate system according to the azimuth and offset of each detection point relative to the shot point. That is, arrange the first arrival points in a spatial manner according to the azimuth and offset.

[0082] In order to make better use of the characteristic that the delay time of the nearby physical point is closer, and take into account the spatial anisotropy of the actual data, with the shot point as the center and the offset as the radius, the first arrival time that has eliminated the influence of the shot point delay time and offset distance is calculated according to the azimuth and offset distance. SR Arrange the space, such as Figure 4 shown.

[0083] Step 5: Divide the first arrival points picked in step 4 into azimuth intervals based on terrain complexity. Based on surface complexity such as surface velocity and undulation characteristics, the first arrival points in the polar coordinate system are divided into several major azimuth intervals.

[0084] The time in step 4 is divided into several broad ranges based on terrain complexity. The number of divisions depends on the anisotropic complexity of the actual strata, typically 2-6. Dividing data into four ranges generally yields good processing results. For data with smooth terrain and weak anisotropy, divide the data into two ranges. In particularly complex terrain, the range division can be non-uniform, consistent with the terrain distribution pattern.

[0085] With the shot point as the center, the azimuth distribution of the surrounding receiver points should be 0 to 360 degrees. In the case of relatively flat terrain, the azimuths can be divided into 2 directions. In the case of complex terrain, the azimuths can be divided into 4 or more directions. In the case of particularly complex terrain, the azimuths can be divided in an uneven manner that conforms to the terrain distribution law. Figure 5 shown.

[0086] In step 6, a sliding window control is used to perform distance-based outlier detection on the first arrival time in each azimuth interval, and the first arrivals corresponding to the outliers are detected and deleted.

[0087] A distance-based outlier detection method with a sliding window control is used for the time within each azimuth. In this distance-based outlier detection algorithm, the definition of distance includes the offset difference and the time difference, which are the values ​​calculated by the combined effect of the two.

[0088] A distance-based outlier detection method is used on the data, and a sliding window is used for calculation with a window size of 200-500 meters.

[0089] The first arrival time of a seismic wave is directly related to the shot delay, the detection point delay, the shot detection offset, and the formation velocity. Therefore, the actual first arrival time distribution is relatively complex. The present invention can greatly improve the accuracy of automatic first arrival picking by eliminating the influence of factors such as the shot delay, the shot detection offset, and the formation velocity, and only retaining the influence of the detection point delay on the first arrival time. In combination with the fact that the detection point delay is directly related to the underground medium, the corresponding relationship between the first arrival time and the detection point position is used to divide the detection point delay obtained after processing into regions. Using a distance-based outlier detection algorithm in each divided area, abnormal first arrivals can be eliminated.

[0090] Step 7: Output the final first arrival picking result.

[0091] Example 2

[0092] In a specific embodiment 2 of the present invention, Figure 3 The data is displayed in polar coordinates, with the temporary detection point delay time represented by different grayscale values. It is clear from the figure that there are some grayscale anomalies in the data. Figure 4 After dividing the range of 0 to 360 degrees into four larger ranges, the temporary detection points are delayed according to the offset distance. Figure 5 The results of the distance-based outlier detection method are processed in each orientation interval. Figure 5 Points with larger grayscale values ​​are monitored as outliers. Figure 6 The figure shows the processed anisotropic distribution temporary detection point delay. It can be seen from the figure that the abnormal points are well processed. Figure 7 This is the original seismic data profile. Figure 8 As can be seen from the figure, there are many abnormal points in the automatic picking first arrival result due to the influence of noise. Figure 9 This is the final result after processing by the method of the present invention. The point with a smaller gray value at time 0 and marked with a triangular arrow is the seismic trace mark where the abnormal first arrival is detected. Figure 9 It can be seen from the figure that the method of the present invention retains as many valid first arrivals as possible while eliminating abnormal first arrivals.

[0093] Example 3

[0094] In a specific embodiment 3 of the present invention, Figure 10The results of outlier detection for the temporary detection point delay time after dividing the directional anisotropy of the data into two large ranges are shown in the figure. As can be seen from the figure, although the data distribution pattern is relatively complex, outliers in the data are still well identified. Figure 11 (a) is the original data section, Figure 11 (b) is the data profile after the first arrival is automatically picked, and there are many abnormal first arrivals in the data. Figure 11 (c) shows the final result after processing using the method of the present invention. The data processing results demonstrate the excellent application effect of the method of the present invention. As can be seen from the processing results, the method of the present invention can eliminate outliers in the first arrival picking to the greatest extent possible, while retaining the correct first arrivals.

[0095] Finally, it should be noted that the above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art may modify the technical solutions described in the aforementioned embodiments or substitute equivalents for some of the technical features therein. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention shall be included within the scope of protection of the present invention.

[0096] Except for the technical features described in the specification, all other technical features are known technologies to those skilled in the art.

Claims

1. A method for detecting abnormal delay of first arrival automatic picking based on azimuthal anisotropy control, characterized in that: The first arrival automatic picking abnormal delay detection method based on azimuthal anisotropy control includes: Step 1: Automatically pick the first arrival of all seismic data; Step 2: Perform delay time iteration calculation on the automatically picked first arrival to obtain the shot point delay time and formation velocity; Step 3: Eliminate the effects of shot point delay time and offset distance in the automatically picked first arrival time, and only retain the effect of receiver point delay time on the first arrival time; Step 4: Arrange the first arrival points processed in step 3 in a polar coordinate system according to the azimuth and offset of each detection point relative to the shot point. Step 5: Divide the first arrival point picked in step 4 into azimuth intervals according to the terrain complexity; Step 6: Using sliding window control, perform distance-based outlier detection on the first arrival time in each azimuth interval, and detect and delete the first arrival corresponding to the outlier; Step 7: Output the final first arrival picking result.

2. The method for detecting abnormal delay of first arrival automatic picking based on azimuthal anisotropy control according to claim 1, characterized in that: In step 1, before the first arrival automatic picking, the seismic data is subjected to conventional denoising processing to improve the first arrival signal-to-noise ratio and enhance the automatic picking accuracy.

3. The method for detecting abnormal delay of first arrival automatic picking based on azimuthal anisotropy control according to claim 1, characterized in that: In step 2, the delayed time is iterated using the first arrivals picked automatically. Before the iterative calculation, the initial formation velocity corresponding to the shot point is assigned an initial value when the shot point is delayed. The formation velocity is obtained by fitting the existing first arrivals, and its value is the inverse of the slope of the fitting line. The shot point delay is taken as half of the intercept of the fitting line.

4. The method for detecting abnormal delay of first arrival automatic picking based on azimuthal anisotropy control according to claim 3, characterized in that: In step 2, when performing delay iteration, it is assumed that the first arrival wave is a refracted wave and there is a single refraction layer underground. Let S be the shot point, R be the receiver point, and the surface thickness of point S and point R be Z respectively. S 、Z R , the critical angle is α, v1 and v2 are the velocities of the upper and lower layers of the refraction surface respectively, then the first arrival time of the refracted wave received by point R when point S is excited is: The delays for defining shot points and receiver points are: set up Then formula (1) can be written as: T SR =t S +t AB +t R (4) An equation such as formula (4) is established for each first arrival time. Assuming that the number of shot points in a work area is N, the number of receiver points is M, and each shot has W channels, the number of equations established in the work area is N*W, and the number of unknowns to be calculated is 2*N+M. Therefore, in the case of high shot channel density, the number of equations is much larger than the number of unknowns. The equation is solved using the Gauss-Seidel iterative algorithm to calculate the delay time and formation velocity corresponding to each receiver point and shot point.

5. The method for detecting abnormal delay of first arrival automatic picking based on azimuthal anisotropy control according to claim 4, characterized in that: In step 3, the delay time of the corresponding shot point is subtracted from each picked first arrival time, and the effect of the offset distance on the delay time is eliminated according to the formation velocity.

6. The method for detecting abnormal delay of first arrival automatic picking based on azimuthal anisotropy control according to claim 5, characterized in that: In step 3, for each picked first arrival time, the influence of the shot point delay time and the formation velocity is eliminated to obtain the temporary detection point delay time corresponding to each first arrival time: t′ SR =T SR -t S -t AB (5) After the delay time decomposition in step 2, t S , t AB Relatively accurate values ​​are obtained. According to formula (5), the first arrival automatically picked up in step 1 is transformed to the detection point delay time t′ corresponding to the current channel. SR .

7. The method for detecting abnormal delay of first arrival automatic picking based on azimuthal anisotropy control according to claim 6, characterized in that: In step 4, in order to make better use of the characteristic that the delay time of the nearby physical point is closer, and at the same time take into account the spatial anisotropy of the actual data, with the shot point as the center and the offset as the radius, the first arrival time that has eliminated the influence of the shot point delay time and offset distance is calculated according to the azimuth and offset distance. SR Arrange the space.

8. The method for detecting abnormal delay of first arrival automatic picking based on azimuthal anisotropy control according to claim 1, characterized in that: In step 5, the first arrival points in the polar coordinate system are divided into several main azimuth intervals according to the surface complexity such as surface velocity and undulation characteristics.

9. The method for detecting abnormal delay of first arrival automatic picking based on azimuthal anisotropy control according to claim 8, characterized in that: In step 5, the detection point in step 4 is delayed by time t′ SR According to the complexity of the terrain, the data is divided into several large directions; the number of divisions is related to the complexity of the anisotropy of the actual stratum, and is divided into 2-6 large ranges; dividing the data into 4 ranges can achieve good processing results; data with smooth terrain and weak stratum anisotropy is divided into 2 ranges; in the case of particularly complex terrain, the direction division adopts a non-uniform division method that conforms to the law of terrain distribution.

10. The method for detecting abnormal delay of first arrival automatic picking based on azimuthal anisotropy control according to claim 1, characterized in that: In step 6, a distance-based outlier detection method with sliding window control is used for the time within each azimuth. In this distance-based outlier detection algorithm, the definition of distance includes the offset difference and the time difference, which is a value calculated by the combined effect of the two.

11. The method for detecting abnormal delay of first arrival automatic picking based on azimuthal anisotropy control according to claim 10, characterized in that: In step 6, a distance-based outlier detection method is used on the data, using a sliding window with a window size of 200-500 meters.

Images