Anomaly first arrival removal method based on automatic region matching

Abstract

The present invention proposes a method for removing abnormal first arrivals based on automatic regional matching, including: dividing the shot gathering data in a work area into multiple sub-areas; obtaining the abnormal first arrival characteristics of each sub-area; and identifying and removing abnormal first arrivals based on the abnormal first arrival characteristics. By testing the seismic data of different controllable vibroseis work areas, the results show that the method of the present invention can quickly identify and remove abnormal first arrivals, effectively reduce the time of manual interaction operation, and improve the accuracy of the first arrival picking results.

Description

Technical Field

[0001] The invention belongs to the field of earth science, relates to petroleum geophysical exploration technology, and specifically relates to an abnormal first arrival removal method based on automatic regional matching in the process of seismic data processing. Background Art

[0002] First-arrival picking is the basis and key business link of seismic data processing, and plays an important role in applications such as refraction wave static correction, shallow refraction wave exploration and tomography exploration. First-arrival picking accounts for a large proportion of the seismic processing workload and is also the most cumbersome. Especially in areas with complex near-surface conditions and low signal-to-noise ratio, the difficulty and time cost of manual picking increase exponentially, and its accuracy and efficiency are difficult to guarantee. Therefore, automated first-arrival picking has also become an important topic for seismic exploration work under complex near-surface conditions. So far, there are many technical methods for automatic picking in the industry, such as correlation method, fractal dimension method, neural network method, etc., but the application effect is often not stable, especially in complex areas, the accuracy and stability of data picking still have a lot of room for improvement. At the same time, the explosive growth of exploration data brought about by the advancement of acquisition technology has also made the processing efficiency of massive earthquakes a new challenge.

[0003] First-arrival picking is the most time-consuming and labor-intensive part of seismic data processing. With the surge in the amount of collected data and the promotion and application of controllable vibrators, the processing time occupied by the first-arrival picking part is getting longer and longer, which seriously affects the cycle and processing quality of this processing project. Most of the existing first-arrival picking methods focus on automatic picking of first-arrivals, and rarely mention the automatic quality control of first-arrival data. At the same time, although existing commercial software provides a variety of quality control methods, manual interaction is required for the quality control of first-arrivals, and abnormal first-arrivals are identified by human eyes and removed or corrected.

[0004] In recent years, with the surge in seismic data, how to quickly perform quality control on first-arrival picking results has become a major concern for processors. In many commercial software, quality control is performed using manual interaction. Although a variety of different tools are provided for quality control, quality control issues are actually fed back to users, which greatly increases the workload of user interactive quality control while providing help to users.

[0005] When the amount of data is relatively large, manual interactive processing greatly increases the workload of the processing personnel, and due to different personal experiences, it is difficult to guarantee the validity of the final result. Based on the above facts, it is not conducive to the initial picking work. Summary of the invention

[0006] Based on the existing quality control status of first-arrival picking, the present invention proposes an abnormal first-arrival removal method of automatic regional matching, which automatically removes abnormal first-arrival results of automatic picking, greatly reducing the workload of user interactive quality control and improving the efficiency and effect of seismic data processing.

[0007] According to one aspect of the present invention, a method for removing abnormal first arrivals based on automatic region matching is provided, comprising:

[0008] Divide the shot gathering data in a work area into multiple sub-areas;

[0009] Obtain the first arrival characteristics of anomalies in each sub-region;

[0010] Abnormal first arrivals are identified and removed based on the abnormal first arrival features.

[0011] Furthermore, the said dividing the shot gather data in a work area into a plurality of sub-areas is to divide and balance the first arrival data based on the geodetic coordinates of the shot points in units of shot gathers.

[0012] Furthermore, the first arrival data is divided and balanced, including:

[0013] Count the coordinates of all the shot points in the work area, calculate the minimum circumscribed rectangle, and obtain the coordinates of the four corners of the rectangle;

[0014] Divide the grid according to a given scale based on the coordinates of the four corners of the rectangle;

[0015] Calculate the total number of grids and count the number of initial shot gatherings in each grid;

[0016] Calculate the average number of shot gathers and the variance of the number of shot gathers on the grid;

[0017] Based on the relationship between the difference between the initial shot gather number and the average shot gather number and the standard deviation of the shot gather number, the shot gather data in the grid are divided and balanced according to the geographical location according to the neighbor principle.

[0018] Furthermore, when the difference between the average shot gather number and the initial shot gather number is greater than the shot gather number variance, all shot gather data in the current grid need to be split into adjacent grids, and the shot gather data are divided into the nearest adjacent grids according to the adjacent principle.

[0019] Furthermore, the coordinates of the four corners of the rectangle are (x1, y1), (x1, y2), (x2, y1), (x2, y2), |x2-x1| is divided according to a given scale, |y2-y1| is divided, and the number of grid points GNX in the x direction and the number of grid points GNY in the y direction are recorded;

[0020] For any grid G ij, where i∈[1, GNX], j∈[1, GNY], calculate the number of guns SN in the grid ij , where i∈[1,GNX], j∈[1,GNY];

[0021] Calculate the average number of shots for all grids MeanSN = ∑SN ij and variance RmsSN;

[0022] For any grid G ij , the number of guns is SN ij , when MeanSN-SN ij >RmsSN, split all guns in the current grid into adjacent grids.

[0023] Further splitting all guns in the current grid into adjacent grids includes:

[0024] Find the current grid as G ij The set S of all adjacent grids, where S∈{G (i-1)j , G (i-1)(j+1) , G i(j+1) , G (i+1)(j+1) , G (i+1)j , G (i+1)(j-1) , G i(j-1) , G (i-1)(j-1)};

[0025] Calculate the center point set C of all grids in the set S, where

[0026] C∈{C (i-1)j , C (i-1)(j+1) , C i(j+1) , C (i+1)(j+1) , C (i+1)j , C (i+1)(j-1) , C i(j-1) , C (i-1)(j-1)};

[0027] For the current grid G ij For any gun S in the range, calculate the distance set D from S to different points in the center point set C, where

[0028] D∈{D (i-1)j , D (i-1)(j+1) , D i(j+1) , D (i+1)(j+1) , D (i+1)j , D (i+1)(j-1) , D i(j-1) , D (i-1)(j-1)};

[0029] According to the nearest neighbor principle, the grid with the smallest distance is selected to store the current gun S.

[0030] Furthermore, obtaining the abnormal first arrival characteristics of each sub-region includes:

[0031] For the shot data in each sub-area, a first arrival time offset intersection diagram is formed. Based on the human eye recognition rule, points that deviate from the majority are abnormal points. The "human eye recognition" process is automatically implemented to extract abnormal first arrival features.

[0032] Furthermore, the extraction of abnormal first arrival features includes:

[0033] Divide the grid in the first arrival time offset intersection diagram, divide it in the first arrival time direction and the offset direction, and calculate the corresponding grid position for any first arrival;

[0034] Count the number of first arrivals, first arrival density, and first arrival speed in each grid;

[0035] Calculate the average velocity in different grids in different offset intervals;

[0036] For any grid, the abnormal first arrival characteristics are obtained based on the first arrival density, the difference between the first arrival velocity and the average velocity.

[0037] Further, identifying and removing abnormal first arrivals based on the abnormal first arrival features includes:

[0038] The abnormal first arrival is identified according to the abnormal first arrival feature, and the shot collection and channel number where the abnormal first arrival is located are found according to the corresponding relationship, and the first arrival corresponding to the shot collection and channel number is set to an invalid value.

[0039] By testing the seismic data of different vibrator work areas, the results show that the method of the present invention can quickly identify and remove abnormal first arrivals, effectively reduce the time of manual interaction operation, and improve the accuracy of first arrival picking results. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] The above and other objects, features and advantages of the present disclosure will become more apparent through a more detailed description of exemplary embodiments of the present disclosure in conjunction with the accompanying drawings, wherein like reference numerals generally represent like components in the exemplary embodiments of the present disclosure.

[0041] Figure 1 The present invention is a flowchart of the method for removing abnormal first arrivals based on automatic region matching.

[0042] Figure 2 4 is a flowchart of abnormal first arrival removal according to an embodiment of the present invention.

[0043] Figure 3 The effect of automatically picking the first arrival result according to the embodiment of the present invention is displayed in the time offset intersection diagram.

[0044] Figure 4 The automatic picking of the first arrival result according to the embodiment of the present invention is used to display the effect in the shot gather record.

[0045] Figure 5 The effect of removing abnormal first arrival results according to an embodiment of the present invention is displayed in the time offset intersection diagram.

[0046] Figure 6 The effect of removing abnormal first arrival results according to an embodiment of the present invention is displayed in the shot gather record. DETAILED DESCRIPTION

[0047] The preferred embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings. Although the preferred embodiments of the present disclosure are shown in the accompanying drawings, it should be understood that the present disclosure can be implemented in various forms and should not be limited by the embodiments described herein. On the contrary, these embodiments are provided to make the present disclosure more thorough and complete, and to fully convey the scope of the present disclosure to those skilled in the art.

[0048] First arrival picking is an important part of seismic data processing, but it is very time-consuming and labor-intensive. The reason is that the processor needs to perform a lot of manual interaction and modification to meet production requirements. The present invention proposes a method for removing abnormal first arrivals based on automatic regional matching.

[0049] like Figure 1 As shown, the present invention discloses a method for removing abnormal first arrivals based on automatic region matching, comprising:

[0050] Divide the shot gathering data in a work area into multiple sub-areas;

[0051] Obtain the first arrival characteristics of anomalies in each sub-region;

[0052] Abnormal first arrivals are identified and removed based on the abnormal first arrival features.

[0053] Specifically, first, for the shot gathering data in a work area, several sub-areas are automatically divided according to the geographical location, and the number of shot gatherings in each sub-area is automatically adjusted so that the data volume in the shot gathering partition presents a stable trend.

[0054] Based on the basic premise that adjacent locations have similar underground structures, the present invention uses shot gathers as units and automatically divides and balances the first arrival data according to the geodetic coordinates of the shot points, so that the amount of first arrival data in each partition is basically the same. This is the most important step in the present invention, which directly affects the effect of subsequent abnormal first arrival removal. Usually, the amount of first arrival data in a work area is very large. In the first arrival time offset intersection diagram, when there are many abnormal first arrivals and they show similar patterns, it is difficult to identify the abnormal first arrivals, and the effect of removing the abnormal first arrivals cannot be achieved. Therefore, the abnormal first arrivals can only be removed within a certain data domain.

[0055] Preferably, automatic division and automatic balancing are accomplished through the following steps:

[0056] 1) Count the coordinates of all the shot points in the work area, calculate the minimum circumscribed rectangle, and obtain the coordinates of the four corners of the rectangle (x1, y1), (x1, y2), (x2, y1), and (x2, y2).

[0057] 2) Divide the range into grids according to a certain range, for example, divide |x2-x1| in the x direction into units of 1000 meters, and divide |y2-y1| similarly, and record the number of grid points GNX in the x direction; the number of grid points GNY in the y direction.

[0058] 3) For any grid G ij , where i∈[1, GNX], j∈[1, GNY], calculate the number of guns SN in the grid ij , where i∈[1,GNX], j∈[1,GNY].

[0059] 4) Balance the number of guns in each grid:

[0060] 4.1) Calculate the average number of guns in all grids MeanSN = ∑SN ij And the variance RmsSN can be calculated according to the conventional method;

[0061] 4.2) For any grid G ij , the number of its guns is SN ij , when MeanSN-SN ij >RmsSN, all guns in the current grid need to be split into adjacent grids. The specific steps are as follows:

[0062] ①Assume that the current grid is G ij , find all its adjacent grid sets S, where S∈{G (i-1)j , G (i-1)(j+1) , G i(j+1) , G (i+1)(j+1) , G (i+1)j , G (i+1)(j-1) , G i(j-1) , G (i-1)(j-1)};

[0063] ② Calculate the center point set C of all grids in the set S, where

[0064] C∈{C (i-1)j , C (i-1)(j+1) , C i(j+1) , C (i+1)(j+1) , C (i+1)j , C (i+1)(j-1) , C i(j-1), C (i-1)(j-1)};

[0065] ③For the current grid G ij For any gun S in the range, calculate the distance set D from S to different points in the center point set C, where

[0066] D∈{D (i-1)j , D (i-1)(j+1) , D i(j+1) , D (i+1)(j+1) , D (i+1)j , D (i+1)(j-1) , D i(j-1) , D (i-1)(j-1)};

[0067] ④ According to the nearest neighbor principle, select the grid with the smallest distance to store the current gun.

[0068] By grouping according to the above rules, the artillery collection can be divided as equally as possible according to geographical location.

[0069] Next, obtain the abnormal first arrival characteristics of each sub-region:

[0070] The first arrival wave is the first signal that returns to the ground after the excitation source signal passes through the stratum. In a shot, the first arrival at the near-offset position is mostly a direct wave, and the first arrival at the far-offset position is mostly a refracted wave, a reflected wave and other waveforms. In a sub-area, the underground structure is similar, and the speed of the first arrival wave is also similar, which is reflected in the first arrival time-offset intersection diagram. The near offset has a similar speed, and the far offset has a similar speed. When the first arrival result automatically picked up on a seismic trace is inaccurate, it will be presented as an outlier in the first arrival time-offset intersection diagram, that is, a point that deviates from the normal. When manual interaction is performed, it is identified by observing whether it deviates from the normal area. The following will describe the "human eye recognition" process to realize the automatic recognition of abnormal first arrivals.

[0071] 1) In the first arrival time offset intersection diagram, divide the grid, for example, in the offset direction, divide it every 50 meters, and in the first arrival time direction, divide it every 20ms. For any first arrival (of f i ,fa i ), of f i Indicates the offset distance of the ith first arrival, in meters, fa i It represents the first arrival time of the ith first arrival, in seconds.

[0072] Calculate its corresponding grid position Where m = off i / offInterval, offInterval represents the offset interval, the unit is m, n = fa i / faInterval, faInterval represents the first arrival time interval in ms; all grid positions with the same subscript are in one grid.

[0073] 2) Count the number of first arrivals in each grid. For the mnth grid, the number of first arrivals is recorded as FaN mn .

[0074] 3) Count the initial arrival density in each grid. For the mnth grid, its initial arrival density is recorded as where k∈[m-1,m+1], t∈[n-1,n+1], s kt Represents the weight coefficient.

[0075] 4) Count the initial arrival velocity in each grid. For the mnth grid, its initial arrival velocity is recorded as Among them (off i ,fa i ) represents all first arrivals in the grid mn.

[0076] 5) Calculate the average velocity in different grids at different offset intervals. For example, with an interval of 2000 meters, the average velocity in the grid on the offset interval [-1000, 1000] is Where m = off i / offInterval,off i ∈[-1000, 1000]; Similarly, calculate The average speed in different areas.

[0077] 6) For any grid, its eigenvector is recorded as

[0078] In a certain work area, the abnormal feature vector generally has a low first arrival density in the grid, for example, the first arrival density is less than 3; i ∈[-1000, 1000] as an example, when in It indicates the standard deviation of the first arrival velocity in all grids with offsets in [-1000, 1000]. When it deviates significantly from the average velocity, it is considered abnormal. The characteristics of the abnormal first arrival are recorded as (3, 3), where the first 3 indicates the first arrival density in a grid, and the second 3 indicates that the difference between the velocity in a grid and the average velocity is greater than 3 times the standard deviation. Different values ​​can be set for different work areas.

[0079] Finally, based on the above abnormal first arrival features, abnormal first arrivals are identified and removed.

[0080] In the mnth grid, if and Set all the first arrivals in the grid to invalid values;

[0081] By testing the seismic data of different vibrator work areas, the results show that this method can quickly identify and remove abnormal first arrivals, effectively reduce the time of manual interaction, and improve the accuracy of first arrival picking results.

[0082] In the present invention, the input data is the first arrival result automatically picked up in a work area, and the output data is the result of removing abnormal first arrivals. The intermediate processing process is automatically completed without the need for manual interactive operation, which is very convenient.

[0083] To facilitate understanding of the solution and effects of the embodiments of the present invention, a specific application example is given below. Those skilled in the art should understand that this example is only for facilitating understanding of the present invention, and any specific details thereof are not intended to limit the present invention in any way.

[0084] Example 1

[0085] The key content of the method of this embodiment is that for the first arrival data in a work area, based on the common sense that adjacent geographical locations have similar underground structures, the work area is automatically divided into multiple sub-areas according to the geographical location and the shot data volume in each sub-area is automatically maintained balanced; then, for the first arrival data in each sub-area, a first arrival time offset intersection diagram is formed, and based on the human eye recognition law, the abnormal first arrival pattern is automatically extracted; finally, the abnormal first arrival pattern is applied to automatically identify the abnormal first arrival and remove the abnormal first arrival in the first arrival data.

[0086] The specific process of this embodiment is as follows Figure 2 As shown:

[0087] First, input the first arrival result of automatic picking within a work area.

[0088] Next, multiple sub-areas are automatically divided based on the geographic location. Specifically, the first arrival data is automatically divided into multiple sub-areas based on the geodetic coordinates of the shot points, taking the shot gather as a unit.

[0089] Then, the amount of first arrival data in each sub-region is automatically balanced.

[0090] Next, the anomalous first arrival patterns are extracted in each sub-region.

[0091] Then, the abnormal first arrival elimination mode is applied to identify the abnormal first arrival.

[0092] Finally, the abnormal first arrivals in the automatically picked first arrival results are removed.

[0093] Example 2

[0094] The content of the present invention has been integrated into the self-developed software AFAPA of the Software Institute of Sinopec Petroleum Geophysical Exploration Technology Research Institute for application. This embodiment is the actual process of using the software AFAPA, which is as follows:

[0095] 1. Input the first arrival automatic picking results into the software;

[0096] 2. Specify the initial distance for automatic division of shot gather data in the software, in meters;

[0097] 3. Run the software to automatically remove abnormal arrivals.

[0098] In order to verify the effect of this embodiment, we selected loess plateau data with low signal-to-noise ratio and large elevation change for testing. The seismic data of 3824 shots and 17217691 first arrivals in this work area were divided into 56 groups, and the abnormal first arrivals in different groups were removed. A total of 22739 abnormal first arrivals were removed, and 0.13% of the abnormal first arrivals were removed. According to the first arrival results picked up automatically, their distribution in the first arrival time offset intersection diagram is first displayed. Figure 3 It can be seen that there are very few first arrivals that deviate from the majority of the population, especially at positive offsets. Obviously, there are very few first arrival picking results that are smaller than the actual first arrival results, which is not in line with reality. The following is displayed on the seismic record as follows Figure 4 As shown in the figure, the first arrival picking of the seismic trace with channel number 4692 is inaccurate.

[0099] By using the method of this embodiment, the above abnormal first arrivals can be removed. First, check their distribution display in the first arrival time offset intersection diagram, such as Figure 5 As shown in the figure, no anomalous first arrivals can be seen. Figure 6 It can be seen that the abnormal first arrival on the seismic trace with channel number 4692 has been removed.

[0100] Those skilled in the art should understand that the purpose of the above description of the embodiments of the present invention is only to exemplarily illustrate the beneficial effects of the embodiments of the present invention, and is not intended to limit the embodiments of the present invention to any given examples.

[0101] The embodiments of the present invention have been described above, and the above description is exemplary, not exhaustive, and is not limited to the disclosed embodiments. Many modifications and changes will be apparent to those skilled in the art without departing from the scope and spirit of the described embodiments.

Claims

1. A method for removing abnormal first arrivals based on automatic region matching, characterized in that: include: Divide the shot gathering data in a work area into multiple sub-areas; Obtain the first arrival characteristics of anomalies in each sub-region; Identify and remove abnormal first arrivals based on the abnormal first arrival features; The shot gather data in a work area is divided into a plurality of sub-areas, and the first arrival data is divided and balanced according to the geodetic coordinates of the shot points in units of shot gathers; Partition and balance the first arrival data, including: Count the coordinates of all the shot points in the work area, calculate the minimum circumscribed rectangle, and obtain the coordinates of the four corners of the rectangle; Divide the grid according to a given scale based on the coordinates of the four corners of the rectangle; Calculate the total number of grids and count the number of initial shot gatherings in each grid; Calculate the average number of shot gathers and the variance of the number of shot gathers on the grid; Based on the relationship between the difference between the initial shot gathering number and the average shot gathering number and the shot gathering number variance, the shot gathering data in the grid are divided and balanced according to the geographical location according to the neighboring principle; When the difference between the average number of shot collections and the initial number of shot collections is greater than the variance of the number of shot collections, all shot collection data in the current grid need to be split into adjacent grids, and the shot collection data are divided into the nearest adjacent grids according to the adjacent principle; The coordinates of the four corners of the rectangle are (x1, y1), (x1, y2), (x2, y1), (x2, y2), |x2-x1| is divided according to a given scale, and |y2-y1| is divided, and the number of grid points GNX in the x direction and the number of grid points GNY in the y direction are recorded; For any grid G ij , where i∈[1,GNX],j∈[1,GNY], calculate the number of guns SN in the grid ij , where i∈[1,GNX],j∈[1,GNY]; Calculate the average number of shots for all grids MeanSN = ∑SN ij and standard deviation RmsSN; For any grid G ij , the number of guns is SN ij , when MeanSN-SN ij >RmsSN, split all guns in the current grid into adjacent grids.

2. The method for removing abnormal first arrivals based on automatic region matching according to claim 1, characterized in that: Splitting all guns in the current grid into adjacent grids includes: Find the current grid as G ij The set S of all adjacent grids, where S∈{G (i-1)j , G (i-1)(j+1) ,G i(j+1) ,G (i+1)(j+1) ,G (i+1)j ,G (i+1)(j-1) ,G i(j-1) ,G (i-1)(j-1) }; Calculate the center point set C of all grids in the set S, where C∈{C (i-1)j ,C (i-1)(j+1) ,C i(j+1) ,C (i+1)(j+1) ,C (i+1)j ,C (i+1)(j-1) ,C i(j-1) ,C (i-1)(j-1) }; For the current grid G ij For any gun S in the range, calculate the distance set D from S to different points in the center point set C, where D∈{D (i-1)j ,D (i-1)(j+1) ,D i(j+1) ,D (i+1)(j+1) ,D (i+1)j ,D (i+1)(j-1) ,D i(j-1) ,D (i-1)(j-1) }; According to the nearest neighbor principle, the grid with the smallest distance is selected to store the current gun S.

3. The method for removing abnormal first arrivals based on automatic region matching according to claim 1, characterized in that: The obtaining of the abnormal first arrival characteristics of each sub-region comprises: For the shot data in each sub-area, a first arrival time offset intersection diagram is formed. Based on the human eye recognition rule, points that deviate from the majority are abnormal points. The "human eye recognition" process is automatically realized to extract abnormal first arrival features.

4. The method for removing abnormal first arrivals based on automatic region matching according to claim 3 is characterized in that: The extracted abnormal first arrival features include: Divide the grid in the first arrival time offset intersection diagram, divide it in the first arrival time direction and the offset direction, and calculate the corresponding grid position for any first arrival; Count the number of first arrivals, first arrival density, and first arrival speed in each grid; Calculate the average velocity in different grids in different offset intervals; For any grid, the abnormal first arrival characteristics are obtained based on the first arrival density, the difference between the first arrival velocity and the average velocity.

5. The method for removing abnormal first arrivals based on automatic region matching according to claim 1, characterized in that: Identifying and removing abnormal first arrivals based on the abnormal first arrival features includes: The abnormal first arrival is identified according to the abnormal first arrival feature, and the shot collection and channel number where the abnormal first arrival is located are found according to the corresponding relationship, and the first arrival corresponding to the shot collection and channel number is set to an invalid value.

Images