An efficient first-arrival picking method for massive 3D seismic data
By employing data thinning experiments to obtain the optimal solution in the processing of massive 3D seismic data, reducing the density of receiver points and the arrangement of receiver lines, the problem of first arrival picking quality relying on manual correction was solved, achieving efficient and accurate first arrival picking and shortening the processing cycle.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- CHINA NAT PETROLEUM CORP
- Filing Date
- 2022-08-23
- Publication Date
- 2026-06-30
AI Technical Summary
In the processing of massive 3D seismic data, existing technologies rely on manual correction for initial picking quality, resulting in a huge workload and extended processing cycles.
The optimal data thinning scheme was obtained through data thinning experiments. The detector point density and receiver line arrangement were halved to ensure that the thinned data could replace all data for first arrival picking. A tomographic static correction method independent of the initial model was used to compare static correction values, and the scheme that meets the ±1ms difference was selected as the optimal scheme.
It greatly reduces the workload of manual picking and correction, improves the efficiency of initial arrival picking, shortens the processing cycle, and ensures that the initial arrival picking quality of thinned data is similar to that of all data, making it suitable for processing massive 3D seismic data.
Smart Images
Figure CN117665921B_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of petroleum geophysical exploration technology, and in particular to an efficient first arrival picking method for massive 3D seismic data. Background Technology
[0002] In the processing of onshore seismic data, it is usually necessary to correct the seismic data to a unified reference plane, which is generally a horizontal plane. Seismic exploration and interpretation theories assume that the excitation point and receiver point are on the same horizontal plane and that the formation velocity is uniform. However, in reality, the ground is often uneven, the depth of each excitation point may vary, and the wave velocity in the low-velocity zone differs significantly from the wave velocity in the formation, which inevitably affects the shape of the measured time-distance curve. To eliminate these effects, the raw seismic data undergoes topographic correction, excitation depth correction, and low-velocity zone correction. These corrections are invariant across different seismic interfaces at the same observation point and are therefore collectively referred to as static correction. In a broader sense, static correction also includes phase correction and correction for instrument-related factors. With the development of digital processing technology, various automatic static correction methods and procedures have emerged.
[0003] Currently, seismic exploration data processing suffers from numerous serious static correction problems. The effectiveness of methods such as refractive static correction and tomographic static correction heavily relies on the quality of first arrival picking. Existing first arrival picking methods, such as Chinese patent document CN114660656A, disclose a seismic data first arrival picking method and system, including: acquiring the seismic data to be measured; inputting the seismic data to be measured into a seismic first arrival picking model to obtain the seismic first arrival; wherein, the seismic first arrival picking model is trained based on a training set and a calibrated deep convolutional neural network; the training set includes sample seismic data and corresponding label information; the label information is a 0-1 map of the sample seismic data; the calibrated deep convolutional neural network includes a first attention subnetwork, a second attention subnetwork, a third attention subnetwork, an encoding subnetwork, and a decoding subnetwork. This invention introduces an attention mechanism into the Unet network, allowing the model parameters in the shallow feature layer to be updated according to the given task-related spatial region, making the network more focused on the required first arrival information and achieving accurate extraction. However, due to the complex surface conditions and the influence of various factors on first arrival waves, the quality of automatic first arrival picking by software is often not ideal, requiring a large amount of manual picking and correction, which is a huge workload and prolongs the processing cycle. Therefore, it is necessary to develop a first arrival picking and processing method that can meet the requirements of seismic wave first arrival picking quality and efficiency for static correction calculation. Summary of the Invention
[0004] The massive amount of 3D seismic data still requires manual picking and correction, resulting in a huge workload and extended processing cycle. Based on the actual situation of seismic data in this area, this invention uses 3D seismic data as a foundation and sets up a data thinning scheme by halving the detector point density and simultaneously halving the receiver arrangement. Qualitative and quantitative tests are conducted on the first arrival picking values corresponding to each scheme to calculate the static correction comparison model and the difference between the static correction values and all picked values. When the difference between the static correction values calculated from the first arrivals obtained from all picked values is very small, this scheme is a usable and efficient first arrival picking scheme that can replace all first arrival picking.
[0005] This invention is achieved by adopting the following technical solution:
[0006] An efficient first-land picking method for massive 3D seismic data includes the following steps:
[0007] S1, acquire earthquake data;
[0008] S2, preprocessing seismic data;
[0009] S3, Use data thinning experiments to obtain the optimal data thinning scheme;
[0010] S4: Acquire target seismic data and preprocess the target seismic data;
[0011] S5, Based on the optimal data thinning scheme obtained in step S3, perform thinning processing on the preprocessed target seismic data.
[0012] S6, based on the thinned data, performs first arrival picking to obtain the first arrival of the earthquake.
[0013] Preferably, in step S1, earthquake data is obtained by retrieving historical data.
[0014] Preferably, in step S2, the preprocessing of the seismic data includes loading the seismic data based on a three-dimensional observation system.
[0015] Preferably, in step S3, the data thinning experiment includes the following steps:
[0016] S31. For seismic data after loading by a three-dimensional observation system, based on the principle of halving the density of receiver points and simultaneously halving the arrangement of receiver lines, several data thinning schemes are exhaustively proposed.
[0017] S32, implement all data thinning schemes in sequence, observe and analyze the characteristics of all data thinning schemes, and retain the data thinning schemes without missing shot points and receiver points.
[0018] S33, obtain all the first arrivals of the earthquake data, and obtain the first arrivals after implementing all the data thinning schemes that were retained, and set the first arrivals as the thinned first arrivals;
[0019] S34. Compare the initial arrival values of all retained data thinning schemes with all initial arrival values to select the best data thinning scheme from all retained data thinning schemes.
[0020] Preferably, in step S31, there are seven exhaustive data thinning schemes, including:
[0021] Option 1: Dilute every other firing line by one line.
[0022] Option 2: Thin out one detector line evenly every other detector line;
[0023] Option 3: When the FFID is odd, reduce the number of shots.
[0024] Option 4: Select the right half of the shot point and arrange the thinning detector lines;
[0025] Option 5: Arrange half of the shots closest to the firing point for each shot.
[0026] Option 6: For each shot, starting from the furthest position, samples are taken and arranged sequentially every other detector line.
[0027] Option 7: First, select the odd-numbered detector line for the odd-numbered FFID number and then thin out the detector line. Then, select the even-numbered detector line for the even-numbered FFID number and thin out the detector line.
[0028] Preferably, in step S32, both Scheme 6 and Scheme 7 involve picking up each shot, with no missing shot points or receiver points, so Scheme 6 and Scheme 7 are ultimately retained.
[0029] Preferably, step S33, obtaining the initial result after implementing all retained data thinning schemes, includes the following steps:
[0030] S33-1, Implement Scheme 6 and Scheme 7 respectively to obtain the diluted data corresponding to Scheme 6 and Scheme 7 respectively;
[0031] S33-2, based on the sparsed data of Scheme 6 and Scheme 7, load the data volume headers of Scheme 6 and Scheme 7 respectively according to all the first arrivals, and then output the first arrival texts corresponding to the data volume headers of Scheme 6 and Scheme 7 respectively.
[0032] Preferably, in step S34, a tomographic static correction method that does not depend on the initial model is used to compare the static correction values of the model, and then the data thinning scheme that satisfies the static correction difference of ±1ms is selected as the optimal data thinning scheme.
[0033] The beneficial technical effects of this invention are as follows:
[0034] 1) This technical solution obtains the optimal data thinning scheme through thinning experiments, ensuring that the data obtained by implementing the data thinning scheme can replace all seismic data for first arrival picking. Based on the optimal data thinning scheme, the seismic data is thinned, which greatly reduces the amount of data used for first arrival picking. As a result, the workload of manual picking and correction can be greatly reduced, which plays an important role in improving the efficiency of first arrival picking and shortening the processing cycle of massive 3D data.
[0035] 2) In the thinning test, this technical solution is based on halving the density of the receiver points and halving the arrangement of the receiver lines to thin the seismic data by halving it. This achieves the maximum thinning of the seismic data within a reasonable range, so that the effects of reducing the workload of manual picking and correction, improving the efficiency of first arrival picking, and shortening the processing cycle through data thinning are maximized within a reasonable range.
[0036] 3) This technical solution, through thinning experiments, determined an optimal data thinning scheme using a cross-arrangement. The near-surface model obtained by this scheme corresponding to the first arrival is highly similar to the standard model using all first arrivals, with the difference in static correction between the two models within ±1 ms. Therefore, implementing cross-arrangement data thinning followed by first arrival picking is an efficient picking method that eliminates the need for replacement with all arrangements. This method can be directly used in subsequent seismic data first arrival picking and has withstood historical testing. Attached Figure Description
[0037] Figure 1 A schematic diagram of the scheme for extracting all data from each shot and each receiving line;
[0038] Figure 2 This is a schematic diagram of data thinning corresponding to Scheme 1;
[0039] Figure 3 This is a schematic diagram of data thinning corresponding to Scheme 2;
[0040] Figure 4 This is a schematic diagram of data thinning corresponding to Scheme 3;
[0041] Figure 5 This is a schematic diagram of data thinning corresponding to Scheme 4;
[0042] Figure 6 This is a schematic diagram of data thinning corresponding to Scheme 5;
[0043] Figure 7 This is a schematic diagram of data thinning corresponding to Scheme 6;
[0044] Figure 8 This is a schematic diagram of data thinning corresponding to Scheme 7;
[0045] Figure 9 This is a schematic diagram illustrating an example of data thinning for Scheme 4;
[0046] Figure 10 This is a schematic diagram illustrating an example of data thinning for Scheme 5;
[0047] Figure 11 This is a schematic diagram illustrating an example of data thinning for Scheme Six;
[0048] Figure 12 This is a schematic diagram of an example of thinning the detector line for the corresponding FFID single-number selection line in Scheme 7;
[0049] Figure 13 This is a schematic diagram of an example of the FFID double-number selection double-number line thinning detector line in Scheme 7;
[0050] Figure 14 This is a diagram showing the difference between the values calculated by each dilution scheme and the total extraction (standard) chromatography method; where (B1) corresponds to scheme seven, which is a diagram showing the difference between the values calculated by scheme seven and the total extraction chromatography method, and so on, (B2) corresponds to scheme six, (B3) corresponds to scheme five, (B4) corresponds to scheme three, and (B5) corresponds to scheme four.
[0051] Figure 15 The diagram shows the comparison of the tomographic inversion models corresponding to each thinning scheme; where (C1) corresponds to scheme seven, which is the tomographic inversion model diagram of scheme seven, and so on, (C3) corresponds to scheme three, (C4) corresponds to scheme six, (C5) corresponds to scheme five, (C6) corresponds to scheme four, and (C7) is the tomographic inversion model diagram of all first arrivals.
[0052] Figure 16 The diagram shows a comparison between the superposition effect of the first arrival inversion static correction for Scheme 7 and the superposition effect of all first arrival inversion static corrections; where (A1) is a diagram showing the superposition effect of all first arrival inversion static corrections and (A2) is a diagram showing the superposition effect of the first arrival inversion static correction for Scheme 7. Detailed Implementation
[0053] To make the purpose, technical solution and advantages of the invention clearer, the technical solution of the invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are some embodiments of the invention, but not all embodiments.
[0054] Therefore, the following detailed description of the invention provided in the accompanying drawings is not intended to limit the scope of the claimed invention, but merely to illustrate selected embodiments of the invention. All other embodiments obtained by those skilled in the art based on the embodiments of the invention without inventive effort are within the scope of protection of the invention.
[0055] Example 1
[0056] This embodiment discloses an efficient first-arrival picking method for massive 3D seismic data. As a preferred embodiment of the present invention, it includes the following steps:
[0057] S1, Obtain seismic data. This seismic data is mainly used for data thinning experiments. Therefore, if historical data is available, seismic data can be obtained by retrieving historical data; if no historical data is available, seismic data can be obtained through on-site exploration.
[0058] S2, Preprocessing of seismic data. This mainly involves loading data using electronic devices to achieve data visualization. In this technical solution, the preprocessing of seismic data primarily includes loading (3D) seismic data based on a 3D observation system to prepare for data thinning experiments.
[0059] S3. A data thinning experiment was conducted to obtain the optimal data thinning scheme. This ensures that the data obtained after implementing the data thinning scheme can replace all seismic data for first-arrival picking.
[0060] S4: Acquire the target seismic data (i.e., the seismic data to be picked up after initial arrival) and preprocess the target seismic data. To meet the requirements of the optimal data thinning scheme, the target seismic data is preprocessed by loading the target seismic data based on the 3D observation system.
[0061] S5. Based on the optimal data thinning scheme obtained in step S3, the preprocessed target seismic data is thinned to obtain the thinned data.
[0062] S6. First arrival picking is performed based on the thinned data to obtain the earthquake first arrival. The first arrival picking can be done manually or using an existing first arrival picking model (such as a first arrival picking model trained on a training set and a calibrated deep convolutional neural network).
[0063] Example 2
[0064] This embodiment discloses an efficient first-arrival picking method for massive 3D seismic data. As a preferred embodiment of the present invention, it includes the following steps:
[0065] S1 obtains earthquake data by retrieving historical data.
[0066] S2 loads seismic data based on a three-dimensional observation system.
[0067] S3. A data thinning experiment is conducted to obtain the optimal data thinning scheme. The data thinning experiment includes the following steps:
[0068] S31. For seismic data after loading by a three-dimensional observation system, based on the principle of halving the density of receiver points and simultaneously halving the arrangement of receiver lines, several data thinning schemes are exhaustively proposed.
[0069] S32, implement all data thinning schemes in sequence, observe and analyze the characteristics of all data thinning schemes, and retain the data thinning schemes without missing shot points and receiver points.
[0070] S33, Obtain all first arrivals of the seismic data, and separately obtain the first arrivals after implementing all the data thinning schemes to be retained; wherein, all first arrivals are obtained by using the existing first arrival picking method to obtain the first arrivals from all the seismic data;
[0071] S34, perform static correction calculations on all first arrivals of the seismic data and all first arrivals obtained after implementing the data thinning scheme, so as to obtain the static correction values of all first arrivals and the static correction values corresponding to all retained data thinning schemes.
[0072] S35, compare the static correction values corresponding to all retained data thinning schemes with all initial static correction values in turn, and select the data thinning scheme that satisfies the static correction difference of ±1ms as the best data thinning scheme.
[0073] S4: Acquire target seismic data and preprocess the target seismic data.
[0074] S5. Based on the optimal data thinning scheme obtained in step S3, the preprocessed target seismic data is thinned to obtain the thinned data.
[0075] S6, based on the thinned data, performs first arrival picking to obtain the first arrival of the earthquake.
[0076] Example 3
[0077] This embodiment discloses an efficient first-arrival picking method for massive 3D seismic data. As a preferred embodiment of the present invention, it includes the following steps:
[0078] S1, acquire earthquake data.
[0079] S2 is used for preprocessing seismic data.
[0080] S3. A data thinning experiment is conducted to obtain the optimal data thinning scheme. The data thinning experiment includes the following steps:
[0081] S31, for seismic data loaded by a three-dimensional observation system, based on the principle of halving the receiver point density while simultaneously halving the receiver line arrangement, exhaustively lists seven data thinning schemes, including two overall observation system thinning schemes, namely: Scheme 1, such as... Figure 2 As shown, the firing line is thinned out evenly by one line every other firing line (spreading the firing line); Option 2, as shown... Figure 3 As shown, every other detector line is uniformly thinned by one (spinning arrangement); and there are five shot-by-shot differentiated thinning schemes, namely: Scheme 3, as shown Figure 4 As shown, this corresponds to the number of sparse cannon shots (spinning shots) when the FFID is odd; Scheme 4, as... Figure 5 and Figure 9 As shown, select the right half of the shot point and arrange the thinned detector lines (spinning arrangement); Scheme 5, as... Figure 6 and Figure 10 As shown, half of the shots are taken from the nearest firing point and arranged in a scattering pattern; Scheme 6, as shown... Figure 7 and Figure 11 Each shot starts from the furthest arrangement and sequentially extracts samples from every other detector line (spinning arrangement); Scheme 7, such as Figure 8 and Figure 12 As shown, first, select the odd-numbered FFID (odd number) detector line and thin out the detector line, then... Figure 8 and Figure 13 As shown, the even-numbered FFID numbers (even numbers) are selected, and the even-numbered (even numbers) detector lines are thinned out (cross-arranged).
[0082] S32. Implement all data thinning schemes sequentially. After observing and analyzing the characteristics of each scheme, retain the scheme that does not lack shot points or receiver points. Further, to facilitate the implementation of the thinning scheme, combine the line number (four digits) and station number (four digits) of the shot / receiver point into an eight-digit line station number. All line station numbers are arranged in a sequence of an+1 = an + (n-1)*d, where an represents the line station number corresponding to the nth shot / receiver point, and d equals the difference between the line station numbers corresponding to the 1st and 2nd shot / receiver points. Based on the sequence of all line station numbers, the corresponding data trace head converts the line station numbers into a sequence index of the corresponding natural number n (1, 2, 3, ...) to receive the line. Based on this, implement the corresponding data thinning scheme. After implementing all data thinning schemes, Schemes 1 and 3 lack shot points in their qualitative analysis; Scheme 2 lacks receiver points, resulting in missing corresponding detector point calibrations in the calculation; Schemes 4 and 5 have missing 1 / 2 and 1 / 4 arrangements at the boundaries in their qualitative analysis, which also leads to missing corresponding detector point calibrations in the calculation. For example... Figure 14The diagrams (B4), (B5), and (B3) illustrating the differences between Schemes 3, 4, and 5 and the total extraction chromatography method show that the absence of shot points and their arrangement can lead to significant differences in the calculated values compared to the total extraction chromatography method. Figure 15 It can be seen that the tomographic inversion models (C3), (C5), and (C6) of Schemes 3, 4, and 5 differ significantly from the tomographic inversion model (C7) for all first arrivals. For Schemes 6 and 7, each shot participates in the pickup process, with no missing shot or receiver points. Scheme 6 thins the receiver arrangement every other receiver line, compensating for this with three-dimensional beam rolling. Scheme 7 uses a randomized alternating arrangement of odd and even field numbers (FFID numbers), not entirely relying on beam rolling. Based on this, Schemes 6 and 7 are the retained data thinning schemes.
[0083] S33, This technical solution prepares for subsequent first-arrival picking of the target seismic data. Therefore, it is necessary to verify the feasibility of the solution, and thus, a correlation comparison (or difference comparison) needs to be made between the first-arrival obtained after data thinning and the first-arrival obtained from all seismic data. Based on this, it is necessary to obtain all the first-arrivals of the seismic data using existing methods, and then separately obtain the first-arrivals after implementing all retained data thinning schemes. Obtaining the first-arrivals after implementing all retained data thinning schemes includes the following steps:
[0084] S33-1, Implement Scheme 6 and Scheme 7 respectively to obtain the diluted data corresponding to Scheme 6 and Scheme 7 respectively;
[0085] S33-2, Obtaining initial arrivals based on the thinned data from Scheme 6 and Scheme 7 respectively. The initial arrivals can be obtained from the thinned data using the same existing methods as for obtaining all initial arrivals. For simplicity and speed, it is preferred to load the data volume headers of Scheme 6 and Scheme 7 respectively based on the thinned data from Scheme 6 and Scheme 7, and then output the corresponding initial arrival text for each. The obtained initial arrivals are then used for massive 3D initial arrival picking using these two schemes (Scheme 6 and Scheme 7). Specifically, all initial arrivals are loaded into the data volume headers of Scheme 6 and Scheme 7 respectively based on the line station number.
[0086] S34, compare the static correction values of the thinned first arrivals corresponding to all retained data thinning schemes with all first arrivals to select the optimal data thinning scheme from all retained data thinning schemes. Specifically, a tomographic static correction method independent of the initial model is used to compare the model static correction values (excluding the remaining first arrivals), and then the data thinning scheme that satisfies the static correction difference of ±1ms is selected as the optimal data thinning scheme. In this technical solution, static correction calculations are performed on the first arrivals corresponding to Scheme 6 and Scheme 7, as well as all first arrivals. The static correction values of the near-surface model and reference surface obtained by the tomographic static correction inversion of the first arrivals corresponding to Scheme 6 and Scheme 7 independent of the initial calculation model are compared with all first arrivals participating in the tomographic static correction calculation. Figure 14 Compare (B1) and (B2), and use... Figure 15 Under the standard condition of (C2), comparing (C1) and (C4), Scheme 7 is superior to Scheme 6. Specifically, the near-surface model obtained based on the first arrival of Scheme 7 is highly similar to the standard model involving all first arrivals, and the difference in static correction of the datum surface calculated with all first arrivals is within ±1ms. It can be seen that the correction calculated using the first arrivals obtained after data thinning is comparable to the correction calculated with all first arrivals, and there is essentially no difference from the cross-section. Figure 16 As shown, the stacking effect of the first-arrival inversion static correction values corresponding to Scheme 7 is basically consistent with the stacking effect of all first-arrival inversion static correction values. Through qualitative and quantitative analysis, first-arrival picking after implementing Scheme 7 can be a highly efficient picking method that does not require all values to be included in the picking replacement. It has stood the test of history. That is, if at any time, facing any new set of seismic data, if you want to verify the reliability of this technical scheme, after completing the first-arrival picking using this technical scheme, you can perform all first-arrival picking based on the same set of seismic data and compare the corresponding model static correction values. You can obtain the same conclusion as the data thinning experiment of this technical scheme (that is, the near-surface model obtained based on the first-arrival values corresponding to Scheme 7 is highly similar to the standard model with all first-arrivals included, and the difference between the reference surface static correction values calculated with all first-arrivals included is within ±1ms).
[0087] S4: Acquire target seismic data and preprocess the target seismic data.
[0088] S5. Based on the optimal data thinning scheme obtained in step S3, the preprocessed target seismic data is thinned to obtain the thinned data. That is, according to the requirements of scheme seven, the odd-numbered receiver line data corresponding to odd-numbered FFIDs and the even-numbered receiver line data corresponding to even-numbered FFIDs are extracted respectively. The two sets of data extracted above are used to replace the entire massive three-dimensional data for first arrival picking and static correction calculation.
[0089] S6, based on the thinned data, performs first arrival picking to obtain the first arrival of the earthquake.
Claims
1. An efficient first-arrival picking method for massive 3D seismic data, characterized in that, Includes the following steps: S1, acquire earthquake data; S2, preprocessing the seismic data, which includes loading the seismic data based on the three-dimensional observation system; S3, Use data thinning experiments to obtain the optimal data thinning scheme; the data thinning experiment includes the following steps: S31. For seismic data after loading by a three-dimensional observation system, based on the principle of halving the density of receiver points and simultaneously halving the arrangement of receiver lines, several data thinning schemes are exhaustively proposed. S32, implement all data thinning schemes in sequence, observe and analyze the characteristics of all data thinning schemes, and retain the data thinning schemes without missing shot points and receiver points. S33, obtain all the first arrivals of the earthquake data, and obtain the first arrivals after implementing all the data thinning schemes that were retained, and set the first arrivals as the thinned first arrivals; S34, compare the initial arrival values of all data thinning schemes with all initial arrival values to select the best data thinning scheme from all the retained data thinning schemes. S4: Acquire target seismic data and preprocess the target seismic data; S5, Based on the optimal data thinning scheme obtained in step S3, perform thinning processing on the preprocessed target seismic data. S6, based on the thinned data, performs first arrival picking to obtain the first arrival of the earthquake.
2. The efficient first arrival picking method for massive 3D seismic data as described in claim 1, characterized in that: In step S1, earthquake data is obtained by retrieving historical data.
3. The efficient first arrival picking method for massive 3D seismic data as described in claim 1, characterized in that, In step S31, there are seven exhaustive data thinning schemes, including: Option 1: Dilute every other firing line by one line. Option 2: Thin out one detector line evenly every other detector line; Option 3: When the FFID is odd, reduce the number of shots. Option 4: Select the right half of the shot point and arrange the thinning detector lines; Option 5: Arrange half of the shots closest to the firing point for each shot. Option 6: For each shot, starting from the furthest position, samples are taken and arranged sequentially every other detector line. Option 7: First, select the odd-numbered detector line for the odd-numbered FFID number and then thin out the detector line. Then, select the even-numbered detector line for the even-numbered FFID number and thin out the detector line.
4. The efficient first arrival picking method for massive 3D seismic data as described in claim 3, characterized in that: In step S32, both Scheme 6 and Scheme 7 involved each shot in the picking process, with no missing shot points or receiver points. Therefore, Scheme 6 and Scheme 7 were ultimately retained.
5. The efficient first arrival picking method for massive 3D seismic data as described in claim 4, characterized in that: In step S33, obtaining the initial result after implementing all retained data thinning schemes includes the following steps: S33-1, Implement Scheme 6 and Scheme 7 respectively to obtain the diluted data corresponding to Scheme 6 and Scheme 7 respectively; S33-2, based on the sparsed data of Scheme 6 and Scheme 7, load the data volume headers of Scheme 6 and Scheme 7 respectively according to all the first arrivals, and then output the first arrival texts corresponding to the data volume headers of Scheme 6 and Scheme 7 respectively.
6. The efficient first arrival picking method for massive 3D seismic data as described in claim 3, characterized in that: In step S34, a tomographic static correction method that does not depend on the initial model is used to compare the static correction values of the model, and then the data thinning scheme that satisfies the static correction difference of ±1ms is selected as the optimal data thinning scheme.