Seismic data reconstruction method and system
By randomly reordering the missing seismic data and reconstructing it using deep learning network, the problem of insufficient reconstruction accuracy of large-area missing seismic data in the prior art is solved, and higher reconstruction accuracy and more suitable model adaptability are achieved.
Patent Information
- Application Number
- CN202311260827.1
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2023-09-27
- Publication Date
- 2025-03-28
- Estimated Expiration
- 2043-09-27
AI Technical Summary
The prior art has insufficient reconstruction accuracy when processing large-area missing seismic data, especially in the interpolation of low-amplitude missing components, which poses great challenges.
By randomly reordering the missing seismic data, the missing tracks become scattered and the reconstruction distance is reduced. Deep learning network is used as the reconstruction model, and the training model is fine-tuned through transfer learning methods to improve reconstruction accuracy.
Through the combination of random reordering and deep learning networks, the reconstruction accuracy of seismic data is significantly improved, especially in the absence of large areas and low amplitudes, seismic data can be reconstructed more accurately.
Smart Images
Figure CN119716973A_ABST
Abstract
Description
Technical Field
[0001] The present invention belongs to the technical field of seismic data processing, and in particular relates to a seismic data reconstruction method and system. Background Art
[0002] In recent years, the increasingly complex exploration environment, the promotion of the "two wide and one high" seismic exploration technology, and the impact of large-scale obstacles often cause the collected seismic data to be missing in large areas, seriously affecting the fidelity, resolution and signal-to-noise ratio of subsequent data processing, thereby reducing the reliability of the final data interpretation. Therefore, it is necessary to reconstruct incomplete or irregular seismic data to improve the integrity of seismic data.
[0003] With the development of deep learning, many neural networks with different structures have been applied to interpolation of missing seismic data. Among them, CNN can effectively interpolate small gap seismic traces with regular and irregular missing data. However, for seismic traces with large gaps and continuous missing data, the interpolation results of low-amplitude missing components need to be further improved.
[0004] In the single-stage training process of traditional CNN on the interpolation of continuously missing seismic traces, the model tends to interpolate the missing traces that are closer to the non-empty traces first. The farther the missing traces are from the non-empty traces, the worse the reconstruction accuracy. Therefore, for seismic traces with small intervals missing, better reconstruction accuracy can be obtained, but it is not enough to handle large-area missing seismic data well. Summary of the invention
[0005] The purpose of the present invention is to solve the above-mentioned problems existing in the prior art and to provide a method for reconstructing seismic data.
[0006] The present invention is implemented by the following technical scheme: a seismic data reconstruction method, comprising the following steps:
[0007] Obtain complete earthquake data and corresponding missing earthquake data;
[0008] Cutting the missing seismic data into a number of missing data slices, and correspondingly cutting the complete seismic data into an equal number of intact data slices;
[0009] Randomly reorder the missing data slices to obtain randomly ordered data slices;
[0010] Perform the same random reordering of the intact data slices as that of the missing data slices to obtain the label data;
[0011] A sample training reconstruction model is adopted based on the randomly ordered data slices and the label data;
[0012] Generate a randomly ordered data piece to be tested of missing seismic data to be tested, wherein the randomly ordered data piece to be tested has the same random reordering as the sample;
[0013] The randomly ordered data slices to be tested are input into the trained reconstruction model, and the reconstructed data slices are output.
[0014] Furthermore, missing earthquake data are produced based on the complete earthquake data: complete earthquake data are obtained through numerical simulation, and then missing traces are marked on the complete earthquake data according to the missing historical actual data or randomly, and the data of the missing traces are filled with 0 values, thereby obtaining missing earthquake data.
[0015] Furthermore, the same random reordering is performed on the preferred track sets of the missing seismic data and the complete seismic data. The preferred track set refers to the track set with the minimum reconstruction distance. For a single missing track, the reconstruction distance refers to the number of tracks between the missing track and the nearest non-empty track. For a track set, the reconstruction distance refers to the maximum reconstruction distance of all the missing tracks in the track set.
[0016] Furthermore, the steps to obtain the optimal gathers are as follows:
[0017] The missing seismic data are converted into different types of gathers, the reconstruction distances of each gather are compared, and the gather with the smallest reconstruction distance is selected as the preferred gather for the missing seismic data; the type of the preferred gather for the complete seismic data is the same as that for the missing seismic data.
[0018] Furthermore, the missing seismic data are converted into a gather including a common shot point gather, a common receiver point gather, a common center point gather, a common offset gather and a common channel number gather.
[0019] Furthermore, the random reordering step includes: assigning a random number to each piece of data, and then reordering the data in the order of the random numbers.
[0020] Furthermore, a deep learning network is adopted as the reconstruction model.
[0021] Furthermore, after the reconstructed data slices of the missing seismic data to be tested are output, they are reversely randomly reordered to obtain complete reconstructed data slices arranged in the original order, and then the complete reconstructed data slices are spliced into complete reconstructed seismic data.
[0022] Furthermore, after the reconstruction model is trained using samples obtained through numerical simulation, the reconstruction model is fine-tuned and trained using samples constructed based on actual data through a transfer learning method.
[0023] Furthermore, the step of constructing samples based on actual data includes: collecting actual complete seismic data, marking missing traces on the actual complete seismic data according to the missing conditions of historical actual data or randomly, and filling the data of the missing traces with 0 values as actual missing seismic data.
[0024] The present invention also provides a seismic data reconstruction system for implementing the seismic data reconstruction method of the present invention, comprising a random coding network, a reconstruction model and a splicing module, wherein the random coding network is used to randomly reorder the input seismic data and output randomly ordered data slices to the reconstruction model; the reconstruction model is used to output reconstructed data slices based on the randomly ordered data slices; the splicing module is used to reversely randomly reorder the reconstructed data slices to obtain complete reconstructed data slices arranged in the original order, and then splice the complete reconstructed data slices into complete reconstructed seismic data.
[0025] Compared with the prior art, the beneficial effects of the present invention include:
[0026] 1. The present invention randomly reorders the missing seismic data so that the missing traces become dispersed and no longer aggregate to form large-area missing traces, thereby reducing the reconstruction distance. For the reconstruction model, the smaller the reconstruction distance, the closer the reconstructed data is to the label data, thereby improving the reconstruction accuracy.
[0027] 2. The present invention produces missing seismic data based on complete seismic data, obtains complete seismic data through numerical simulation, ensures the quality of complete seismic data, and ensures the matching of complete seismic data with missing seismic data.
[0028] 3. Marking missing traces on complete earthquake data according to the missing historical data to produce missing earthquake data can reduce the training samples; randomly marking missing traces on complete earthquake data to produce missing earthquake data makes the training samples richer and can cover various missing situations.
[0029] 4. The present invention can further reduce the reconstruction distance by optimizing the gathers, thereby further improving the reconstruction accuracy.
[0030] 5. The present invention uses samples constructed according to actual data to fine-tune the training and reconstruction model through the transfer learning method, so as to obtain a model that is more suitable for the current work area. BRIEF DESCRIPTION OF THE DRAWINGS
[0031] Figure 1 Data flow diagram for seismic data reconstruction method;
[0032] Figure 2 Flowchart of refactoring the model for training;
[0033] Figure 3 The flowchart of data reconstruction using the trained reconstruction model. DETAILED DESCRIPTION
[0034] In the single-stage training process of traditional CNN on the interpolation of continuously missing seismic traces, the model tends to interpolate the missing traces that are closer to the non-empty traces first. The farther the missing traces are from the non-empty traces, the worse the reconstruction accuracy. Therefore, for seismic traces with small intervals missing, better reconstruction accuracy can be obtained, but it is not enough to handle large-area missing seismic data well.
[0035] The present invention randomly reorders the missing seismic data so that the missing traces become dispersed and no longer aggregate to form large-area missing traces, thereby reducing the reconstruction distance. For the reconstruction model, the smaller the reconstruction distance, the closer the reconstructed data is to the label data, thereby improving the reconstruction accuracy.
[0036] The present invention is further described in detail below in conjunction with the accompanying drawings:
[0037] refer to Figure 1 As shown, a seismic data reconstruction system is used to implement the seismic data reconstruction method of the present invention, including a random coding network, a reconstruction model and a splicing module. The random coding network is used to randomly reorder the input seismic data and output randomly ordered data slices to the reconstruction model; the reconstruction model is used to output reconstructed data slices according to the randomly ordered data slices; the splicing module is used to reversely randomly reorder the reconstructed data slices to obtain complete reconstructed data slices arranged in the original order, and then splice the complete reconstructed data slices into complete reconstructed seismic data.
[0038] The seismic data reconstruction method of the present invention comprises the following steps:
[0039] Obtain complete earthquake data and corresponding missing earthquake data;
[0040] Cutting the missing seismic data into a number of missing data slices, and correspondingly cutting the complete seismic data into an equal number of intact data slices;
[0041] Randomly reorder the missing data slices to obtain randomly ordered data slices;
[0042] Perform the same random reordering of the intact data slices as that of the missing data slices to obtain the label data;
[0043] A sample training reconstruction model is adopted based on the randomly ordered data slices and the label data;
[0044] Generate a randomly ordered data piece to be tested of missing seismic data to be tested, wherein the randomly ordered data piece to be tested has the same random reordering as the sample;
[0045] The randomly ordered data slices to be tested are input into the trained reconstruction model, and the reconstructed data slices are output.
[0046] A more detailed description is given below.
[0047] 1) Create a training data set
[0048] Produce missing earthquake data based on complete earthquake data: obtain complete earthquake data through numerical simulation, and then mark missing traces on the complete earthquake data according to the missing historical actual data or randomly, and fill the missing traces with 0 values to obtain missing earthquake data.
[0049] The missing seismic data are randomly reordered to obtain randomly ordered data pieces; the steps of random reordering include: assigning a random number to each piece of data, and then reordering according to the order of the random numbers.
[0050] Preferably, the missing seismic data and the complete seismic data are randomly reordered according to their respective preferred track sets, where the preferred track set refers to the track set with the minimum reconstruction distance; for a single missing track, the reconstruction distance refers to the number of tracks between the missing track and the nearest non-empty track; for a track set, the reconstruction distance refers to the maximum value of the reconstruction distances of all the missing tracks in the track set.
[0051] The steps to obtain the optimal gather are as follows:
[0052] The missing seismic data are converted into different types of gathers, the reconstruction distances of each gather are compared, and the gather with the smallest reconstruction distance is selected as the preferred gather for the missing seismic data; the type of the preferred gather for the complete seismic data is the same as that for the missing seismic data.
[0053] The missing seismic data are converted into gathers including common shot point gathers, common receiver point gathers, common center point gathers, common offset gathers and common channel number gathers.
[0054] The complete earthquake data are randomly reordered in the same way as the missing earthquake data to obtain the labeled data;
[0055] Samples are constructed based on randomly sorted data pieces and labeled data. That is, randomly sorted data pieces are used as input data for the reconstruction model, and the model parameters are continuously updated through the reconstruction error between the reconstructed data output by the reconstruction model and the labeled data.
[0056] 2) Training and reconstruction model
[0057] The sample set composed of the above samples is divided into training set, validation set and test set.
[0058] Build appropriate reconstruction models according to needs, using deep learning networks, including convolutional neural networks, long and short-term recurrent neural networks, etc.;
[0059] Use the prepared training set to debug and determine the hyperparameters of network training;
[0060] refer to Figure 2As shown, the network training is performed according to the prepared training set and hyperparameters;
[0061] When the set number of iterations is met or the reconstruction error drops to the specified value, the training is stopped;
[0062] Save the trained reconstruction model.
[0063] Preferably, after the reconstruction model is trained using samples obtained through numerical simulation, the reconstruction model is fine-tuned and trained using samples constructed based on actual data through a transfer learning method.
[0064] The step of constructing samples based on actual data includes: collecting actual complete seismic data, marking missing traces on the actual complete seismic data according to the missing conditions of historical actual data or randomly, and filling the data of the missing traces with 0 values as actual missing seismic data.
[0065] 3) Data reconstruction
[0066] refer to Figure 3 As shown, randomly ordered data slices of missing seismic data to be tested are generated and input into the trained reconstruction model, and the reconstructed data slices are output. The same method as that of making training samples is used to generate randomly ordered data slices of missing seismic data to be tested, that is, the missing seismic data to be tested are sliced, each data is assigned a random number, and then reordered in the order of the random numbers.
[0067] The number of the randomly ordered data slices to be tested is equal to the number of the randomly ordered data slices in the sample, and the sizes are equal. The random numbers of the corresponding tracks of the randomly ordered data slices to be tested and the randomly ordered data slices in the sample are the same, which means that the randomly ordered data slices to be tested have the same random reordering as the sample.
[0068] After the reconstructed data slices of the missing seismic data to be tested are output, the random reordering is reversed to obtain the complete reconstructed data slices arranged in the original order, and then the complete reconstructed data slices are spliced into the complete reconstructed seismic data. The random reordering is the reverse process of the random reordering, which first reverses the random numbering to obtain the original numbering, and then sorts in the order of the original numbering.
[0069] Example 1
[0070] In this example, simulated data is first used for pre-training. After training, a small amount of actual data is used for fine-tuning through transfer learning methods to finally obtain a model suitable for the current work area.
[0071] 1. Create a training data set:
[0072] 1.1. Design the observation system according to the geological model of the work area and the geological tasks, and then perform numerical simulation to obtain a complete seismic data set.
[0073] 1.2. According to the distribution of obstacles in the work area, mark the corresponding positions of the complete seismic data and fill the missing traces with 0 values.
[0074] 1.3. In this example, only the case of missing detection points is considered. For large-area missing detection points, the missing seismic data are converted to common detection point gathers, common shot point gathers, common center point gathers, common offset gathers, and common channel number gathers, and the reconstruction distance of each gather is calculated separately. In this example, the reconstruction distance of the common shot point gather is the smallest, so the missing seismic data are converted to the common shot point gathers.
[0075] 1.4. Perform the same random coding, rearrangement and recording on each set of common shot point seismic data.
[0076] 1.5. Divide the above earthquake data into training set, validation set and test set.
[0077] 1.6. The process of making a training set of actual data is similar to steps 1.2-1.5 and will not be described in detail.
[0078] 2. Deep learning network training:
[0079] 2.1. In this example, the commonly used U-net network structure with attention mechanism is selected for training;
[0080] 2.2. Use the prepared training set to debug and determine the hyperparameters of network training;
[0081] 2.3. Perform network training according to the prepared training set and hyperparameters;
[0082] 2.4. Save the trained network model.
[0083] 3. Deep learning reconstruction part
[0084] 3.1. Use the trained random coding network to reconstruct data. Preprocess the missing seismic data and make random coding data slices and corresponding masks according to the process in step 1;
[0085] 3.2. Input the data slice and the corresponding mask into the trained network model to output the reconstructed seismic data;
[0086] 3.3. Anti-random reordering of the reconstructed data;
[0087] 3.4. Splice the output seismic data into the original size.
[0088] Example 2
[0089] In this example, simulated data is first used for pre-training. After training, a small amount of actual data is used for fine-tuning through transfer learning methods to finally obtain a model suitable for the current work area.
[0090] 1. Create a training data set:
[0091] 1.1. Design the observation system according to the geological model of the work area and the geological tasks, and then perform numerical simulation to obtain a complete seismic data set.
[0092] 1.2. According to the distribution of obstacles in the work area, mark the corresponding positions of the complete seismic data and fill the missing traces with 0 values.
[0093] 1.3. In this example, only the case of missing shot points is considered. For large-area missing shot points, the missing seismic data is converted into common detection point gathers, common shot point gathers, common center point gathers, common offset gathers, and common channel number gathers, and the reconstruction distance of each gather is calculated separately. In this example, the reconstruction distance of the common detection point gather is the smallest, so the missing seismic data is converted into the common detection point gathers.
[0094] 1.4. Perform the same random coding, rearrangement and recording on each set of common shot point seismic data.
[0095] 1.5. Divide the above earthquake data into training set, validation set and test set.
[0096] 1.6. The process of making a training set of actual data is similar to steps 1.2-1.5 and will not be described in detail.
[0097] 2. Deep learning network training:
[0098] 2.1. In this example, convolutional neural network CNN is selected for training;
[0099] 2.2. Use the prepared training set to debug and determine the hyperparameters of network training;
[0100] 2.3. Perform network training according to the prepared training set and hyperparameters;
[0101] 2.4. Save the trained network model.
[0102] 3. Deep learning reconstruction part
[0103] 3.1. Use the trained random encoding network to reconstruct data. Figure 3 As shown. Preprocess the missing seismic data and make random coded data slices and corresponding masks according to the process in step 1;
[0104] 3.2. Input the data slice and the corresponding mask into the trained network model to output the reconstructed seismic data;
[0105] 3.3. Anti-random reordering of the reconstructed data;
[0106] 3.4. Splice the output seismic data into the original size.
[0107] Example 3
[0108] In this example, simulated data is first used for pre-training. After training, a small amount of actual data is used for fine-tuning through transfer learning methods to finally obtain a model suitable for the current work area.
[0109] 1. Create a training data set:
[0110] 1.1. Design the observation system according to the geological model of the work area and the geological tasks, and then perform numerical simulation to obtain a complete seismic data set.
[0111] 1.2. According to the distribution of obstacles in the work area, mark the corresponding positions of the complete seismic data and fill the missing traces with 0 values.
[0112] 1.3. In this example, only the case of missing offset is considered. For large-area missing offset, the missing seismic data is converted to common detection point gathers, common shot point gathers, common center point gathers, common offset gathers, and common channel number gathers, and the reconstruction distance of each gather is calculated separately. In this example, the reconstruction distance of the common center point gather is the smallest, so the missing seismic data is converted to the common center point gathers.
[0113] 1.4. Perform the same random coding, rearrangement and recording on each set of common shot point seismic data.
[0114] 1.5. Divide the above earthquake data into training set, validation set and test set.
[0115] 1.6. The process of making a training set of actual data is similar to steps 1.2-1.5 and will not be described in detail.
[0116] 2. Deep learning network training:
[0117] 2.1. In this example, recurrent neural network RNN is selected for training;
[0118] 2.2. Use the prepared training set to debug and determine the hyperparameters of network training;
[0119] 2.3. Perform network training according to the prepared training set and hyperparameters;
[0120] 2.4. Save the trained network model.
[0121] 3. Deep learning reconstruction part
[0122] 3.1. Use the trained random encoding network to reconstruct data. Figure 3 As shown. Preprocess the missing seismic data and make random coded data slices and corresponding masks according to the process in step 1;
[0123] 3.2. Input the data slice and the corresponding mask into the trained network model to output the reconstructed seismic data;
[0124] 3.3. Anti-random reordering of the reconstructed data;
[0125] 3.4. Stitch the output seismic data into the original size.
[0126] Example 4
[0127] In this example, simulated data is first used for pre-training. After training, a small amount of actual data is used for fine-tuning through transfer learning methods to finally obtain a model suitable for the current work area.
[0128] 1. Create a training data set:
[0129] 1.1. Design the observation system according to the geological model of the work area and the geological tasks, and then perform numerical simulation to obtain a complete seismic data set.
[0130] 1.2. According to the distribution of obstacles in the work area, mark the corresponding positions of the complete seismic data and fill the missing traces with 0 values.
[0131] 1.3. In this example, only the case of missing center points is considered. For large-area missing center points, the missing seismic data are converted to common detection point gathers, common shot point gathers, common center point gathers, common offset gathers, and common channel number gathers, and the reconstruction distance of each gather is calculated separately. In this example, the reconstruction distance of the common offset gather is the smallest, so the missing seismic data are converted to the common offset gathers.
[0132] 1.4. Perform the same random coding, rearrangement and recording on each set of common shot point seismic data.
[0133] 1.5. Divide the above earthquake data into training set, validation set and test set.
[0134] 1.6. The process of making a training set of actual data is similar to steps 1.2-1.5 and will not be described in detail.
[0135] 2. Deep learning network training:
[0136] 2.1. In this example, the long short-term memory network LSTM is used for training;
[0137] 2.2. Use the prepared training set to debug and determine the hyperparameters of network training;
[0138] 2.3. Perform network training according to the prepared training set and hyperparameters;
[0139] 2.4. Save the trained network model.
[0140] 3. Deep learning reconstruction part
[0141] 3.1. Use the trained random encoding network to reconstruct data. Figure 3 As shown. Preprocess the missing seismic data and make random coded data slices and corresponding masks according to the process in step 1;
[0142] 3.2. Input the data slice and the corresponding mask into the trained network model to output the reconstructed seismic data;
[0143] 3.3. Anti-random reordering of the reconstructed data;
[0144] 3.4. Stitch the output seismic data into the original size.
[0145] A person skilled in the art will appreciate that all or part of the steps in the above-described embodiment method may be implemented by instructing related hardware through a program, and the program may be stored in a computer-readable storage medium, such as ROM / RAM, a magnetic disk, an optical disk, etc.
[0146] The above technical scheme is only a specific implementation method of the present invention. For those skilled in the art, it is easy to make various types of improvements or modifications based on the principles disclosed in the present invention, and it is not limited to the technical scheme described in the above specific embodiments of the present invention. Therefore, the above description is only preferred and does not have a restrictive meaning.
Claims
1. A method for reconstructing seismic data, characterized in that: The following steps are involved: Obtain complete earthquake data and corresponding missing earthquake data; Cutting the missing seismic data into a number of missing data slices, and correspondingly cutting the complete seismic data into an equal number of intact data slices; Randomly reorder the missing data slices to obtain randomly ordered data slices; Perform the same random reordering of the intact data slices as that of the missing data slices to obtain the label data; A sample training reconstruction model constructed according to the randomly ordered data slices and the label data is used; Generate a randomly ordered data piece to be tested of missing seismic data to be tested, wherein the randomly ordered data piece to be tested has the same random reordering as the sample; The randomly ordered data slices to be tested are input into the trained reconstruction model, and the reconstructed data slices are output.
2. The seismic data reconstruction method according to claim 1, characterized in that: Produce missing earthquake data based on complete earthquake data: obtain complete earthquake data through numerical simulation, and then mark missing traces on the complete earthquake data according to the missing historical actual data or randomly, and fill the missing traces with 0 values to obtain missing earthquake data.
3. The seismic data reconstruction method according to claim 1, characterized in that: The same random reordering is performed on the preferred track sets of missing seismic data and complete seismic data. The preferred track set refers to the track set with the smallest reconstruction distance. For a single missing track, the reconstruction distance refers to the number of tracks between the missing track and the nearest non-empty track. For a track set, the reconstruction distance refers to the maximum reconstruction distance of all missing tracks in the track set.
4. The seismic data reconstruction method according to claim 3, characterized in that: The steps to obtain the optimal gather are as follows: The missing seismic data are converted into different types of gathers, the reconstruction distances of each gather are compared, and the gather with the smallest reconstruction distance is selected as the preferred gather for the missing seismic data; the type of the preferred gather for the complete seismic data is the same as that for the missing seismic data.
5. The seismic data reconstruction method according to claim 4, characterized in that: The missing seismic data are converted into gathers including common shot point gathers, common receiver point gathers, common center point gathers, common offset gathers and common channel number gathers.
6. The seismic data reconstruction method according to claim 1, characterized in that: The random reordering step includes: assigning a random number to each piece of data, and then reordering the data in the order of the random numbers.
7. The seismic data reconstruction method according to claim 1, characterized in that: A deep learning network is used as the reconstruction model.
8. The seismic data reconstruction method according to claim 1, characterized in that: After the reconstructed data slices of the missing seismic data to be tested are output, they are reversely randomly reordered to obtain complete reconstructed data slices arranged in the original order, and then the complete reconstructed data slices are spliced into complete reconstructed seismic data.
9. The seismic data reconstruction method according to claim 1, characterized in that: After the reconstruction model is trained using samples obtained through numerical simulation, the reconstruction model is fine-tuned and trained using transfer learning methods using samples constructed based on actual data.
10. The seismic data reconstruction method according to claim 9, characterized in that: The step of constructing samples based on actual data includes: collecting actual complete seismic data, marking missing traces on the actual complete seismic data according to the missing conditions of historical actual data or randomly, and filling the data of the missing traces with 0 values as actual missing seismic data.
11. A seismic data reconstruction system, characterized in that: Used to implement the seismic data reconstruction method as described in any one of claims 1 to 10, comprising a random coding network, a reconstruction model and a splicing module, the random coding network is used to randomly reorder the input seismic data and can output randomly ordered data slices to the reconstruction model; the reconstruction model is used to output reconstructed data slices based on the randomly ordered data slices; the splicing module is used to reverse the random reordering of the reconstructed data slices to obtain complete reconstructed data slices arranged in the original order, and then splice the complete reconstructed data slices into complete reconstructed seismic data.
Citation Information
Patent Citations
Method and device for reconstructing irregular seismic data
CN106199701A
Vertical seismic profiling seismic data regularization reconstruction method and device and storage medium
CN111123361A
Three-dimensional seismic data reconstruction method and device
CN113093273A