Multi-scale variable frequency modeling method
Through the multi-scale variable frequency modeling method, combined with the wave equation and ray theory, velocity models of different scales are gradually established, which solves the problems of sparseness of rays and low calculation efficiency of waveform tomography, and achieves higher inversion accuracy and resolution , an accurate near-surface velocity model was obtained, and the static correction amount and pre-stack depth offset effect were improved.
Patent Information
- Application Number
- CN202110508994.8
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2021-05-11
- Publication Date
- 2024-11-01
- Estimated Expiration
- 2041-05-11
AI Technical Summary
When establishing a near-surface velocity model in the prior art, there is a problem of ray sparseness in travel time tomography, waveform tomography calculation efficiency is low and errors are large, making it difficult to obtain an accurate shallow velocity model in areas such as the front zone of the mountain.
Using multi-scale variable frequency modeling method, through multiple rounds of inversion at different scales, the ray path is determined using wavefront propagation time and the annular wave band. Combining the wave equation and ray theory, a large, medium and small scale velocity model is gradually established to improve the inversion. Performance accuracy and resolution.
The ray sparse problem is effectively solved, the inversion accuracy and resolution are improved, a more accurate near-surface velocity model is obtained, and the effect of static correction amount calculation and pre-stack depth offset is improved.
Smart Images

Figure CN115327635B_ABST
Abstract
Description
Technical Field
[0001] The invention relates to the technical field of oil and gas exploration seismic data processing, and in particular to a multi-scale variable frequency modeling method. Background Art
[0002] The accuracy of the near-surface velocity model is the most important factor affecting the quality of seismic data processing. Currently, there are two main methods for establishing near-surface velocity models, namely travel-time tomography and waveform tomography. These two methods have their own advantages and disadvantages.
[0003] Travel time tomography based on ray theory inverts underground velocity distribution by interpreting the travel time of seismic waves, which is the line integral of the slowness along the ray path between the excitation point and the receiving point. The assumption of travel time tomography is that the frequency of seismic waves is high enough and the seismic energy propagates along a ray of zero volume, which is a purely idealized mathematical ray. This method is easy to calculate and has a fast calculation speed. However, in practical applications, when encountering a low-velocity body, the ray will deviate, and when encountering a high-speed gathering area, there will be serious instability. Therefore, for areas with large lateral velocity changes, such as the piedmont belt, many grid cells do not even have a single ray passing through, which greatly reduces the accuracy of the inversion.
[0004] Waveform tomography based on the dynamic characteristics of seismic waves takes into account the frequency factor of seismic waves and the effects of all media in the entire model space. Theoretically, this method has a higher resolution than travel-time tomography. However, in actual calculations, it is necessary to use the initial model to calculate the forward propagation of seismic waves and the reverse propagation of the residual wave field, and at the same time calculate the integral of the correlation between the two along time. The calculation efficiency of this method is very low, and it has high requirements on the initial velocity model. The seismic wavelets required for calculation can only be given artificially, which brings large errors to the final results.
[0005] The quality of seismic data in piedmont zones, desert areas, gravel areas, etc. is generally poor, characterized by strong noise interference, large velocity changes, low shallow data coverage times, and low signal-to-noise ratio. It is difficult to obtain an accurate shallow velocity model using traditional reflection wave tomography inversion.
[0006] In the Chinese patent application with application number: CN201010520346.6, a method for establishing a near-surface velocity model in high-density seismic static correction processing is involved. The method is to assume that the near-surface velocity increases linearly with depth, calculate the refracted wave velocities at different offsets from the travel time difference of the first-arrival waves of adjacent detection points of the high-density seismic based on the physical properties of the near-surface velocity field, and then establish the near-surface gradual velocity field parameters based on the relationship between the offset and the refracted wave, and then establish the near-surface velocity model. The modified velocity model can be directly used for static correction calculation in seismic processing, and can also be used as an initial model for tomographic inversion of near-surface velocity.
[0007] In the Chinese patent application with application number: CN201310520865.6, a multi-azimuth tomographic velocity modeling method in the common offset and common azimuth domain is involved, which belongs to the technical field of oil exploration seismic data processing. The seismic data in the work area are processed conventionally to establish an initial velocity model; the above initial velocity model is used to perform pre-stack migration in the common offset and common azimuth domain, and the common reflection point (CRP) gather has azimuth information after migration; the azimuth division scheme is determined according to the development direction of the main faults in the work area, and the CRP gather is divided into several groups of sub-CRP gathers according to the azimuth information in the CRP gather; the residual time difference (RMO) is picked for each sub-CRP gather; all azimuth RMO picking values are uniformly subjected to ray tracing and tomographic inversion to obtain the update amount of the velocity model; the above update amount is used to update the original velocity model to obtain an updated velocity model; a new round of pre-stack depth migration velocity model iteration is performed until the velocity model meets the requirements.
[0008] In the Chinese patent application with application number: CN202010058741.0, a deep learning velocity modeling method based on seismic reflection waveform and velocity spectrum is involved, which belongs to the field of petroleum geophysical exploration technology. The present invention includes the following steps: generation of training set seismic velocity model, generation and processing of training set data, construction of deep neural network architecture, training of deep neural network model, preparation of input data, and prediction of seismic velocity model.
[0009] The above existing technologies are significantly different from the present invention and fail to solve the technical problem we want to solve. Therefore, we have invented a new multi-scale variable frequency modeling method. Summary of the invention
[0010] The purpose of the present invention is to provide a multi-scale variable frequency modeling method which not only solves the ray sparseness problem of travel time tomography, but also utilizes the dynamic characteristics of waveform theory.
[0011] The purpose of the present invention can be achieved by the following technical measures: a multi-scale variable frequency modeling method, the multi-scale variable frequency modeling method comprising:
[0012] Step 1: Data preparation, first arrival picking;
[0013] Step 2: Establish an initial large-scale discrete grid velocity model;
[0014] Step 3: Determine the wavefront propagation time in the large-scale discrete grid velocity model;
[0015] Step 4: Determine the large-scale ray path;
[0016] Step 5: Determine the large-scale annular wave band in the low-frequency state;
[0017] Step 6: Tomographic inversion to obtain a large-scale velocity model;
[0018] Step 7: Replace the initial velocity model with the large-scale velocity model, and repeat steps 2 to 6 with the mesoscale discrete grid to obtain the mesoscale velocity model;
[0019] Step 8: Replace the large-scale velocity model with the mesoscale velocity model, and repeat steps 2 to 6 with a small-scale discrete grid to obtain the final velocity model.
[0020] The purpose of the present invention can also be achieved by the following technical measures:
[0021] In step 1, the prepared data include the coordinates and elevation of the single-shot loading shot point and receiving point collected in the field, the shot point well depth, and the wellhead time; the first arrival picking is to pick the early arrival wave time of each trace.
[0022] In step 2, an initial large-scale discrete grid velocity model is established, the surface model is discretized into a number of cubic units using a regular grid, and the velocity model of the grid is represented by the velocity values at the grid nodes.
[0023] In step 3, when forward calculating the first-arrival wave travel in the large-scale discrete grid velocity model, the wavefront extension method based on the level set is used to gradually calculate the wavefront propagation time on the discrete medium grid nodes starting from the earthquake source to obtain the wavefront propagation time at a large scale.
[0024] In step 4, the ray path is reversely determined in the large-scale discrete grid velocity model, and the wavefront propagation time on the grid nodes calculated in step 3 is used to trace the ray path unit by unit starting from the receiving point toward the source direction.
[0025] In step 5, the low-frequency frequency is obtained according to the effective frequency band of the first arrival wave, the radius of the annular wave band is first calculated, and then the spatial distribution of the annular wave body is determined by the radius of the annular wave band with the ray as the central axis.
[0026] In step 6, the distribution of the annular wave body space and its energy flux density is used to establish a linear relationship between the first arrival travel time residual and the grid node slowness correction, and the large-scale velocity model is obtained by tomographic inversion.
[0027] In step 7, the initial velocity model of step 2 is replaced by the large-scale velocity model of step 6, the frequency in step 5 is changed to obtain the main frequency according to the spectrum of the first arrival wave, and steps 2 to 6 are repeated with the mesoscale to obtain the mesoscale velocity model.
[0028] In step 7, the large-scale discrete grid is a grid that is 2 to 4 times larger than the original data acquisition grid, and the medium-scale discrete grid is a grid that is the same size as the original data acquisition grid.
[0029] In step 8, the velocity model of step 2 is replaced with the mesoscale velocity model of step 7, and in step 5, the high-frequency frequency is obtained according to the effective frequency band of the first arrival wave, and steps 2 to 6 are repeated at a small scale to obtain the final velocity model.
[0030] In step 8, the mesoscale discrete grid is a grid with the same scale as the original data acquisition grid, and the small-scale discrete grid is a half or a quarter of the original data acquisition grid.
[0031] According to the characteristics of actual seismic data, such as large velocity variation in low-deceleration zones, low coverage of shallow data and low signal-to-noise ratio, the present invention has developed a multi-scale variable frequency modeling method that is between the wave equation and the ray theory and fully considers the frequency characteristics of seismic waves. This method not only solves the ray sparsity problem of travel time tomography, but also utilizes the dynamic characteristics of waveform theory. It is a method with high calculation accuracy and efficiency.
[0032] The multi-scale variable frequency modeling method of the present invention, which is between the wave equation and the ray theory, fully considers that the actual earthquake frequency is limited, and the earthquake energy mainly propagates in the annular belt including the ideal ray; the travel time is related to all media in the annular belt, which is more in line with the actual propagation law of seismic waves; the inversion stability and resolution are improved, and the accuracy of the velocity model of complex near-surface seismic data can be improved, laying a foundation for the subsequent static correction or pre-stack depth migration. The beneficial effects of the present invention are as follows:
[0033] First, multiple rounds of inversion at different scales are used in the modeling process to avoid the influence of empty channels and empty shots caused by various reasons, which can effectively improve the accuracy of inversion and make the inverted velocity model more refined.
[0034] Second, in the process of determining the annular wave body, the order from low frequency to high frequency is adopted, which solves the problem of sparse ray density in conventional tomography, reduces the multi-solution problem in inversion, and makes the inverted velocity model more accurate. BRIEF DESCRIPTION OF THE DRAWINGS
[0035] Figure 1 A flowchart of a specific embodiment of the multi-scale variable frequency modeling method of the present invention;
[0036] Figure 2 It is an elevation map after loading the observation system in a specific embodiment of the present invention;
[0037] Figure 3 A schematic diagram of a velocity model inverted by a multi-scale variable frequency modeling method in a specific embodiment of the present invention;
[0038] Figure 4 It is a schematic diagram of a cross section of a residual static correction of a certain commercial software in a specific embodiment of the present invention;
[0039] Figure 5 It is a schematic diagram of a cross section of an application of the velocity model of the present invention for calculating static correction in a specific embodiment of the present invention. DETAILED DESCRIPTION
[0040] It should be noted that the following detailed descriptions are exemplary and are intended to provide further explanation of the present invention. Unless otherwise specified, all technical and scientific terms used herein have the same meanings as those commonly understood by those skilled in the art to which the present invention belongs.
[0041] It should be noted that the terms used herein are only for describing specific embodiments and are not intended to limit exemplary embodiments according to the present invention. As used herein, unless the context clearly indicates otherwise, the singular form is also intended to include the plural form. In addition, it should be understood that when the terms "comprising" and / or "including" are used in this specification, it indicates the presence of features, steps, operations and / or combinations thereof.
[0042] In the current seismic data processing, the velocity modeling technology corresponding to prestack depth migration has developed slowly, which has affected the application effect of prestack depth migration. An important part of velocity modeling technology is to establish an accurate near-surface velocity model, because the accuracy of the near-surface model directly affects the depth domain imaging effect. The present invention is aimed at the problems existing in the prior art, and provides a multi-scale variable frequency modeling method to accurately calculate the near-surface velocity, which provides a possible way for the practical application of prestack depth migration and provides better processing solutions and results for subsequent seismic interpretation.
[0043] The multi-scale variable frequency modeling method of the present invention comprises the following steps:
[0044] Step 1: Data preparation, first arrival picking;
[0045] The data preparation and first arrival picking include: loading the shot point and receiving point coordinates, elevation, shot point well depth, wellhead time and other information for the single shot collected in the field; first arrival picking is picking the early arrival wave time of each trace.
[0046] Step 2: Establish an initial large-scale discrete grid velocity model;
[0047] An initial large-scale discrete grid velocity model is established. Since the gun check point variable observation causes missing tracks and missing shots, the purpose of using a large scale is to remove the impact of missing tracks and missing shots. The surface model is discretized into several cubic units using a regular grid, and the velocity value at the grid node is used to represent the velocity model of the grid.
[0048] Step 3: Determine the wavefront propagation time in the large-scale discrete grid velocity model;
[0049] When forward calculating the travel of the first arrival wave in the large-scale discrete grid velocity model, the wavefront expansion method based on the level set is used to gradually calculate the wavefront propagation time on the discrete medium grid nodes starting from the earthquake source, and then the wavefront propagation time at a large scale is obtained.
[0050] Step 4: Determine the large-scale ray path;
[0051] The ray path is determined in reverse in the large-scale discrete grid velocity model. The wavefront propagation time on the grid nodes calculated in the previous step is used to trace the ray path unit by unit from the receiving point toward the source.
[0052] Step 5: Determine the large-scale annular wave band in the low-frequency state;
[0053] The low frequency is obtained according to the effective frequency band of the first arrival wave. The low frequency is used in the first cycle to avoid the ray vacancy problem to the greatest extent. The radius of the annular wave band is first calculated, and then the ray is used as the central axis to determine the spatial distribution of the annular wave body with the radius of the annular wave band.
[0054] Step 6: Tomographic inversion to obtain a large-scale velocity model;
[0055] By using the distribution of the annular wave space and its energy flux density, a linear relationship between the first arrival travel time residual and the grid node slowness correction is established, and the large-scale velocity model is obtained by tomographic inversion.
[0056] Step 7: Replace the initial velocity model with the large-scale velocity model, and repeat steps 2 to 6 at the mesoscale, where in step 5, the main frequency is used to obtain the mesoscale velocity model;
[0057] Replace the initial velocity model in step 2 with the large-scale velocity model in step 6, change the frequency in step 5 to obtain the main frequency according to the spectrum of the first arrival wave, and repeat steps 2 to 6 with the mesoscale to obtain the mesoscale velocity model.
[0058] Step 8: Replace the large-scale velocity model with the mesoscale velocity model, and repeat steps 2 to 6 at a small scale. In step 5, use a high frequency to obtain the final velocity model.
[0059] Replace the velocity model in step 2 with the mesoscale velocity model in step 7, obtain the high-frequency frequency according to the effective frequency band of the first arrival wave in step 5, repeat steps 2 to 6 at a small scale, and the final velocity model can be obtained.
[0060] As a preferred technical solution, the large-scale discrete grid is a grid that is 2 to 4 times larger than the original data acquisition grid, the medium-scale discrete grid is a grid with the same scale as the original data acquisition grid, and the small-scale discrete grid is half or one-quarter of the original data acquisition grid.
[0061] In a specific embodiment 1 of the present invention, a multi-scale variable frequency modeling method includes the following steps:
[0062] (1) Preliminary data preparation and first arrival picking include: first loading the shot point and receiving point coordinates, elevation, shot point depth, wellhead time and other information for the single shot collected in the field, and then picking the early arrival wave time of each trace.
[0063] (2) Establish an initial large-scale discrete grid velocity model. Since the gun check point variation observation causes missing tracks and missing shots, the purpose of using a large scale is to remove the impact of missing tracks and missing shots. Use a regular grid to discretize the surface model into several cubic units, and use the velocity values at the grid nodes to represent the velocity model of the grid.
[0064] (3) When forward calculating the travel of the first arrival wave in the large-scale discrete grid velocity model, the wavefront extension method based on the level set is used to gradually calculate the wavefront propagation time on the discrete medium grid nodes starting from the earthquake source. At this time, the wavefront propagation time at a large scale is obtained.
[0065] (4) Determine the ray path in reverse in the large-scale discrete grid velocity model. Using the wavefront propagation time on the grid nodes calculated in the previous step, trace the ray path unit by unit from the receiving point toward the source.
[0066] (5) The low frequency is obtained based on the effective frequency band of the first arrival wave. The low frequency is used in the first cycle to avoid the ray vacancy problem to the greatest extent. The radius of the annular wave band is first calculated, and then the ray is used as the central axis to determine the spatial distribution of the annular wave body using the radius of the annular wave band.
[0067] (6) Using the distribution of the annular wave space and its energy flux density, a linear equation is established between the first arrival travel time residual and the grid node slowness correction, and a large-scale velocity model is obtained by tomographic inversion.
[0068] (7) The initial velocity model of step (2) is replaced by the large-scale velocity model of step (6), the frequency in step (5) is changed to the main frequency obtained according to the spectrum of the first arrival wave, and steps (2) to (6) are repeated at the mesoscale to obtain the mesoscale velocity model.
[0069] (8) Replace the velocity model of step (2) with the mesoscale velocity model of step (7), obtain the high-frequency frequency according to the effective frequency band of the first arrival wave in step (5), repeat steps (2) to (6) at a small scale, and the final velocity model can be obtained.
[0070] In a specific embodiment 2 of the present invention, Figure 1 The flowchart of a specific embodiment of the multi-scale variable frequency modeling method of the present invention comprises the following steps:
[0071] Step 101, data preparation and first arrival picking;
[0072] Step 102, establishing an initial large-scale discrete grid velocity model;
[0073] Step 103, forward calculating the travel time of the first arrival wave in the large-scale discrete grid velocity model to obtain the wavefront propagation time at a large scale;
[0074] Step 104, reversely determine the ray path in the large-scale discrete grid velocity model;
[0075] Step 105, determining the spatial distribution of the ring wave body using the ring wave zone radius at low frequency;
[0076] Step 106, obtaining a large-scale velocity model by tomographic inversion;
[0077] Step 107, replace the initial velocity model with a large-scale velocity model, use the main frequency, and repeat steps 102 to 106 with the mesoscale to obtain a mesoscale velocity model.
[0078] Step 108, replace the velocity model of step 102 with the mesoscale velocity model of step 107, obtain the high frequency according to the effective frequency band of the first arrival wave, repeat steps 102 to 106 at a small scale, and obtain the final velocity model.
[0079] In a specific embodiment 3 of the present invention, for the convenience of explanation, actual seismic data of a certain exploration area is used to apply the multi-scale frequency conversion modeling method according to the operating steps of the present invention to illustrate the application effect of the invention.
[0080] The present invention takes the mountain three-dimensional seismic data of the YD area of XX oil field as the target area, and applies this method to process the data to verify the effect of this method. The specific flow chart is shown in Figure 1 The actual data covers an area of 17km×32km, with 47 shot lines, 18,377 shots, 400m between shot lines, and 50m between shots; each shot is received by 12 receiving lines with 2,400 detection points, 400m between receiving lines, 50m between traces, and a grid of 25mX25m. The above method is used to process the data.
[0081] 1) First, enter step 101, use the SPS file provided in the field to establish an observation system, load the observation system into the original single shot, so that the single shot has the trace information required for subsequent calculations, and then perform first arrival picking. Figure 2This is a topographic map of the work area extracted from the prepared data. The terrain in the work area is undulating, with peaks, canyons, rivers, etc. The elevation changes from 290m to 1020m. There are peaks in the northern part of the work area, and the surface elevation difference is large. Modeling the near-surface velocity of such complex seismic data has always been a difficult problem in the industry.
[0082] 2) Then, according to step 102, an initial large-scale discrete grid velocity model is established. According to the changes in the observation system of this work area and the interference of the first arrival wave, the scale is selected to be 3 times the grid of the original data acquisition, that is, a cube unit of 75mX75mX75m grid. The initial velocity is estimated based on the time-offset distance of the first arrival wave to give a given gradual velocity.
[0083] 3) Then, according to step 103, when forward calculating the travel time of the first arrival wave in the 75mX75mX75m grid model, the wavefront expansion method based on the level set is used to gradually calculate the wavefront propagation time on the discrete medium grid nodes starting from the earthquake source, and the wavefront propagation time in the 75mX75mX75m grid is obtained.
[0084] 4) Then, according to step 104, the ray path is reversely determined in the 75mX75mX75m grid model, and the wavefront propagation time on the grid node calculated in the previous step is used to trace the ray path unit by unit from the receiving point toward the source direction.
[0085] 5) Then, according to step 105, the spectrum analysis of the first arrival wave in this work area is carried out to determine the low frequency of the effective frequency band, and the radius of the annular wave band is calculated using this frequency. Then, with the ray as the central axis, the spatial distribution of the annular wave body is determined using the radius of the annular wave band.
[0086] 6) According to step 106, the distribution of the annular wave body space and its energy flux density is used to establish a linear equation satisfied by the first arrival travel time residual and the grid node slowness correction, and a coarser velocity model of 75mX75mX75m grid is obtained by tomographic inversion.
[0087] 7) According to step 107, the initial velocity model of step 102 is replaced with the 75mX75mX75m grid velocity model of step 106, the frequency in step 105 is changed to the main frequency obtained according to the spectrum of the first arrival wave, and steps 102 to 106 are repeated with a 25mX25mX25m grid to obtain a more refined mesoscale velocity model.
[0088] 8) Finally, according to step 108, the velocity model of step 102 is replaced with the 25mX25mX25m grid velocity model of step 107, and the frequency in step 105 is set to the high frequency according to the effective frequency band of the first arrival wave. The small scale of this work area selects a 12.5mX12.5mX12.5m grid, and repeats steps 102-106 to obtain an accurate final velocity model. Figure 3 This is the final velocity model. It can be seen that the overall surface velocity is about 4 km / s. There are low-speed areas in the mountains and river areas, with the minimum speed being around 3 km / s.
[0089] In order to verify the accuracy of the final velocity model and compare it with the existing commonly used software, the static correction process is performed using the technology of the present invention and a commercial software. The stacked sections after static correction are shown in the figure below. Figure 4 and Figure 5 As shown, Figure 4 This is the result of static correction using a commercial software. Figure 5 This is the static correction result processed by the method of the present invention. Figure 4 and Figure 5 It can be clearly seen that Figure 4 There are obvious distortions and misalignments in the event axis. Figure 5 The event axis in the model is more continuous and the event axis energy is more focused. This shows that the velocity model established by the multi-scale variable frequency modeling method of the present invention is more accurate and has a better static correction effect.
[0090] Finally, it should be noted that the above is only a preferred embodiment of the present invention and is not intended to limit the present invention. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art can still modify the technical solutions recorded in the aforementioned embodiments or replace some of the technical features therein with equivalents. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.
[0091] Except for the technical features described in the specification, all other technical features are known technologies to those skilled in the art.
Claims
1. A multi-scale variable frequency modeling method, characterized in that: The multi-scale variable frequency modeling method includes: Step 1: Data preparation, first arrival picking; Step 2: Establish an initial large-scale discrete grid velocity model; Step 3: Determine the wavefront propagation time in the large-scale discrete grid velocity model; Step 4: Determine the large-scale ray path; Step 5: Determine the large-scale annular wave band in the low-frequency state; Step 6: Tomographic inversion to obtain a large-scale velocity model; Step 7: Replace the initial velocity model with the large-scale velocity model, and repeat steps 2 to 6 with the mesoscale discrete grid to obtain the mesoscale velocity model; Step 8: Use the mesoscale velocity model to replace the large-scale velocity model, change to a small-scale discrete grid and repeat steps 2 to 6 to obtain the final velocity model; In step 7, the initial velocity model of step 2 is replaced by the large-scale velocity model of step 6, the frequency in step 5 is changed to obtain the main frequency according to the spectrum of the first arrival wave, and steps 2 to 6 are repeated with the mesoscale to obtain the mesoscale velocity model; In step 7, the large-scale discrete grid is a grid that is 2 to 4 times larger than the original data acquisition grid, and the medium-scale discrete grid is a grid that is the same as the original data acquisition grid; In step 8, the velocity model in step 2 is replaced with the mesoscale velocity model in step 7, and in step 5, the high frequency is obtained according to the effective frequency band of the first arrival wave, and steps 2 to 6 are repeated at a small scale to obtain the final velocity model; In step 8, the mesoscale discrete grid is a grid with the same scale as the original data acquisition grid, and the small-scale discrete grid is a half or a quarter of the original data acquisition grid.
2. The multi-scale variable frequency modeling method according to claim 1, characterized in that: In step 1, the prepared data include the coordinates and elevation of the single-shot loading shot point and receiving point collected in the field, the shot point well depth, and the wellhead time; the first arrival picking is to pick the early arrival wave time of each trace.
3. The multi-scale variable frequency modeling method according to claim 1, characterized in that: In step 2, an initial large-scale discrete grid velocity model is established, the surface model is discretized into a number of cubic units using a regular grid, and the velocity model of the grid is represented by the velocity values at the grid nodes.
4. The multi-scale variable frequency modeling method according to claim 1, characterized in that: In step 3, when forward calculating the first-arrival wave travel in the large-scale discrete grid velocity model, the wavefront extension method based on the level set is used to gradually calculate the wavefront propagation time on the discrete medium grid nodes starting from the earthquake source to obtain the wavefront propagation time at a large scale.
5. The multi-scale variable frequency modeling method according to claim 1, characterized in that: In step 4, the ray path is reversely determined in the large-scale discrete grid velocity model, and the wavefront propagation time on the grid nodes calculated in step 3 is used to trace the ray path unit by unit starting from the receiving point toward the source direction.
6. The multi-scale variable frequency modeling method according to claim 1, characterized in that: In step 5, the low-frequency frequency is obtained according to the effective frequency band of the first arrival wave, the radius of the annular wave band is first calculated, and then the spatial distribution of the annular wave body is determined by the radius of the annular wave band with the ray as the central axis.
7. The multi-scale variable frequency modeling method according to claim 1, characterized in that: In step 6, the distribution of the annular wave body space and its energy flux density is used to establish a linear relationship between the first arrival travel time residual and the grid node slowness correction, and the large-scale velocity model is obtained by tomographic inversion.
Citation Information
Patent Citations
Method for establishing near-surface velocity model in high-density seismic static correction processing
CN101980054A
Multidirectional tomographic velocity modeling method within common offset and common azimuth domains
CN104570088A
Deep learning speed modeling method based on seismic reflection waveform and speed spectrum
CN111239802A
Multi-scale crosshole SIRT tomography method
CN101561512A
Multi-scale regular grid tomography inversion statics correction method
CN103217715A
