A method, system and device for improving the imaging accuracy of ultra-deep seismic data

Abstract

The application provides a method, system and device for improving imaging precision of super-deep seismic data, and belongs to the field of super-deep seismic oil and gas exploration. The method for improving imaging precision of super-deep seismic data adopts a time domain processing system for protecting low frequency and expanding frequency band, an approximate true surface full-depth velocity modeling technical process and an angle domain mirror imaging fixed structure and full wave field imaging fixed reservoir imaging technical system, protects the signal of a sensitive and effective frequency band 8-14 Hz of a super-deep target layer, improves the precision of a full-depth velocity model and a full-depth, and outputs final seismic data for the super-deep target layer from two aspects of stratum structure and reservoir, so that the imaging precision of super-deep seismic data is effectively improved.

Description

Technical Field

[0001] This invention belongs to the field of ultra-deep seismic oil and gas exploration, specifically relating to a method, system, and apparatus for improving the imaging accuracy of ultra-deep seismic data. Background Technology

[0002] Ultra-deep seismic imaging (strata with a depth exceeding 6000m are called ultra-deep strata) suffers from significant burial depth, severe ground absorption attenuation, and large influence from shallow and medium-depth layers. Its imaging effect is directly affected by the shallow and medium-depth layers. The accumulated errors at the target layer will gradually amplify, ultimately making it difficult to image ultra-deep target layers. At the same time, for reservoir imaging, ultra-deep target layers have low signal-to-noise ratios, requiring high signal-to-noise ratio seismic data to determine the stratigraphic structure. High-fidelity and amplitude-preserving seismic data are also needed for reservoir prediction.

[0003] Currently, the main technical processes and systems adopted by the industry are still mainly derived from improvements to conventional seismic data processing procedures, attempting to improve data quality through higher precision speed, higher resolution, and gather processing for ultra-deep strata.

[0004] The invention patent with publication number CN105005076B discloses a method for seismic wave full waveform inversion based on a least squares gradient update velocity model, including the following steps: 1) obtaining the acoustic equation of the seismic wave field in the time domain; 2) constructing an initial velocity model, setting the velocity model update iteration number N and the allowable minimum error value ε; 3) constructing the wave field error vector of the observed wave field data and the calculated wave field data; 4) constructing an objective function; 5) calculating the standard equation for seismic wave full waveform inversion from the objective function; 6) introducing the velocity model update gradient direction gk and the update step size α; 7) solving the velocity model update gradient direction gk using the least squares method; 8) interpolating the update step size α; 9) updating the velocity model to obtain: mk = mk-1 + αgk; when |αgk| < ε or the velocity model update number reaches the velocity model update iteration number N, the velocity model update ends; otherwise, proceed to step 3). This invention can quickly update the velocity model and is widely used in seismic wave full waveform inversion methods, but it does not specifically solve the problem of imaging difficulties in ultra-deep seismic data.

[0005] Chinese patent application CN112198547A discloses a method and apparatus for processing deep or ultra-deep seismic data. The method includes: acquiring deep or ultra-deep seismic data from the field; performing frequency- and domain-based denoising processing on the deep or ultra-deep seismic data based on actual coordinates to obtain a denoising result; the actual coordinates being the actual coordinates of the shot point and receiver point; sequentially performing amplitude compensation processing and deconvolution processing on the denoising result to obtain a deconvolution result; and generating a deep or ultra-deep seismic data processing result based on the deconvolution result. This method sequentially applies denoising processing, amplitude compensation processing, and deconvolution processing to the seismic data, which can protect low-frequency signals, improve low-frequency signal imaging capabilities, and thus improve the imaging quality of deep or ultra-deep seismic data. In order to improve the signal-to-noise ratio of seismic data, considering that deep data imaging requires more low-frequency information, more careful parameter suppression is adopted for noise at specified frequencies, such as noise below 10Hz, so that low-frequency signals are not damaged.

[0006] It is known that existing technologies have not proposed a complete imaging solution and process for ultra-deep seismic data imaging, resulting in various problems such as poor accuracy, low signal-to-noise ratio, low bandwidth, poor amplitude fidelity, and unclear wave group characteristics in current ultra-deep seismic data imaging.

[0007] To address the key challenges and difficulties of ultra-deep seismic data, it is necessary to innovatively establish a complete technical process and system for improving the accuracy of seismic data. This system should comprehensively consider the velocity variations at shallow, medium, and deep depths, protect and expand the complete and effective frequency bands, improve the signal-to-noise ratio while maintaining amplitude and fidelity in imaging. In this way, a technical system and process for improving the quality of ultra-deep seismic data can be truly established from a process perspective. Summary of the Invention

[0008] This invention addresses the problems existing in the prior art by providing a process and technical system suitable for improving ultra-deep seismic data. It provides a method, system, and device for improving the imaging accuracy of ultra-deep seismic data. This method has targeted effects on signal-to-noise ratio, amplitude fidelity, and completeness from shallow to deep, thereby achieving the goal of systematically improving the accuracy of seismic data in ultra-deep exploration target layers.

[0009] To achieve the above objectives, the technical solution adopted by the present invention is as follows:

[0010] In a first aspect, the present invention provides a method for improving the imaging accuracy of ultra-deep seismic data, comprising the following steps:

[0011] Step S1: Perform approximate true surface migration processing on the seismic data of the ultra-deep target layer to obtain the processed data;

[0012] Step S2: For the processed data, a two-step well control deconvolution process with extended frequency band is used to protect the effective low-frequency signal of 8-14Hz, and the data after deconvolution is obtained.

[0013] Step S3: Using the data processed by deconvolution, perform approximate true surface full-depth and all-round velocity modeling to obtain the approximate true surface full-depth velocity modeling results for the ultra-deep target layer.

[0014] Step S4: Based on the approximate true surface full-depth velocity modeling results of the ultra-deep target layer, perform dual imaging of the ultra-deep target layer.

[0015] In step S1, the approximate true surface offset surface processing includes:

[0016] Step S11: Select the smooth surface of the ground elevation through the observation system to determine the approximate true ground surface offset surface;

[0017] Step S12: Using micro-logging constrained inversion of the near-surface model, and based on the selected surface elevation smooth surface, calculate the static correction amount based on the approximate true ground surface and the static correction surface based on the reference surface.

[0018] Step S13: Based on the static correction of the reference plane, apply the pre-stack denoising technique of classifying and step-by-step preservation of effective low frequencies to suppress noise;

[0019] Step S14: Conduct surface uniform amplitude processing.

[0020] In step S13, the effective low frequency includes an effective signal of 8-14 Hz that is sensitive to ultra-deep target layers.

[0021] In step S13, the pre-stack denoising technique employs radial domain subtraction for selective denoising.

[0022] Step S1 performs an approximate true surface offset process on the ultra-deep target layer. This process corrects the phase and amplitude errors caused by complex terrain, ensuring the accuracy of subsequent processing and interpretation.

[0023] In step S2, the extended frequency band is the extended high-frequency frequency band.

[0024] Step S2 enhances the quality of the reflected signal by reducing the influence of multiple waves and random noise, providing a clearer data foundation for subsequent velocity modeling.

[0025] The approximate true surface full-depth and omnidirectional velocity modeling process in step S3 includes:

[0026] Step S31: Perform stacking velocity analysis, dynamic correction, and residual static correction processing;

[0027] Step S32: Remove the static correction based on the reference surface from the gather, apply the static correction based on the approximate true ground surface, and apply the static correction to the first arrival.

[0028] Step S33: Based on the statically corrected first arrival at an approximate true surface, conduct first arrival tomography based on the approximate true surface to invert the shallow and middle layer velocity model;

[0029] Step S34: Perform full-depth joint tomography and simultaneously invert the velocity model across the entire depth range;

[0030] Step S35: Evaluate the velocity model using a quantitative quality control system.

[0031] Step S3 establishes a detailed underground velocity structure model, taking into account velocity variations throughout the exploration area, thus providing accurate velocity field information for the final imaging.

[0032] In step S4, the dual imaging is the formation structure and reservoir imaging of the ultra-deep target layer.

[0033] Step S4 specifically includes the following steps:

[0034] Step S41: Based on the approximate true surface full-depth velocity modeling results of the ultra-deep target layer, perform angle domain migration to obtain a dip domain three-dimensional gather;

[0035] Step S42: Based on the dip domain 3D gather, perform dip domain mirror imaging on the formation structure of the ultra-deep target layer to obtain mirror results for the formation structure of the ultra-deep target layer; for the ultra-deep target layer reservoir, perform angle domain illumination-compensated migration imaging and full-wavelength stacking to obtain full-wavelength reservoir imaging results for the ultra-deep target layer.

[0036] Step S4 provides high-resolution geological structure images and more accurately depicts reservoir characteristics, which is of vital importance for the assessment and development of oil and gas resources.

[0037] The method for improving the accuracy of seismic imaging of ultra-deep target layers described in this invention employs a time-domain processing system that preserves low frequencies and expands the frequency band; (2) a near-true surface full-depth velocity modeling technology process and angle-domain mirror imaging to determine the structure; and (3) a full-wavefield imaging technology system to determine the reservoir. Among these, the time-domain processing system that preserves low frequencies and expands the frequency band protects the signals in the sensitive effective frequency band of 8-14Hz for ultra-deep target layers. The near-true surface full-depth velocity modeling technology process improves the accuracy of the full-depth velocity model. Finally, the final velocity and gathers are used to perform full-wavefield angle-domain illumination-compensated migration imaging to determine the reservoir and dip-domain mirror imaging to determine the structure. The final seismic data for ultra-deep target layers are output from both the stratigraphic structure and reservoir aspects, which can effectively improve the quality and interpretation accuracy of ultra-deep target layer seismic data.

[0038] Secondly, the present invention provides a system for improving the imaging accuracy of ultra-deep seismic data, comprising the following modules:

[0039] The data processing module is used to execute the methods described in the above technical solution;

[0040] The observation module is used to select the smooth surface of the ground elevation and determine the approximate true ground surface offset surface;

[0041] The deconvolution module is used to implement well-controlled two-step deconvolution processing with extended frequency bands;

[0042] The velocity modeling module is used to perform velocity modeling processing that approximates the true Earth surface at all depths and in all directions.

[0043] The imaging module is used for dual imaging of ultra-deep target layers based on velocity modeling results.

[0044] Thirdly, the present invention also provides an apparatus for improving the imaging accuracy of ultra-deep seismic data, comprising the following units:

[0045] The processing unit is used to execute the method described in the above technical solution;

[0046] Storage units are used to store seismic data and intermediate results generated during processing;

[0047] The input / output unit is used to receive input data and output processing results;

[0048] The display unit is used to display the processing results and the image.

[0049] Compared to existing technologies, this invention has the following advantages: It can be widely applied to ultra-deep seismic oil and gas exploration, enabling accurate identification of facies-controlled dolomite reservoirs in multiple exploration strata, including the Penglaiba Formation-Xiaqiulitag Group, platform margin zone, and Xiaoerbulake Formation + Qigebulake Formation in the North Tarim Basin. The key is deep seismic reflection structure and amplitude-preserving, high-fidelity imaging. The imaging technology method for improving the accuracy of ultra-deep seismic data described in this invention proposes a noise suppression technique that preserves low frequencies and expands high frequencies through frequency division and classification. It integrates near-true surface full-depth velocity modeling and proposes an angle-domain mirror imaging technique for determining structure and a full-wavefield imaging technique for determining reservoirs. This provides technical support for effectively improving the imaging quality of ultra-deep dolomite reservoirs. This technology has been applied in the processing of the Tashen 1 well area, and the imaging of the platform margin structure and internal anomaly reflection characteristics further confirms the effectiveness of this technology system. Attached Figure Description

[0050] Figure 1 A flowchart of a method for improving the accuracy of ultra-deep seismic data;

[0051] Figure 2 The images show the effects of radial transformation on a single shot before and after; the left image is a schematic diagram of a single shot before radial transformation, and the right image is a schematic diagram of a single shot after radial transformation; the horizontal axis represents the track number (unit: track), and the vertical axis represents the recording time (unit: seconds);

[0052] Figure 3 To preserve low frequencies, a comparison of the ultra-deep target layer spectrum before and after pre-stack denoising is presented. A represents the stacked profile after denoising, with the horizontal axis representing the tie line number (none) and the vertical axis representing the recording time (milliseconds). The green box indicates the region for spectrum and signal-to-noise ratio (SNR) analysis. B represents the stacked profile before denoising, with the horizontal axis representing the tie line number (none) and the vertical axis representing the recording time (milliseconds). The red box indicates the region for spectrum and SNR analysis. C represents a schematic diagram of the SNR analysis results for the target layer, with the horizontal axis representing the channel number and the vertical axis representing the SNR value. The green curve represents the SNR value distribution curve of the stacked profile after denoising, and the red curve represents the SNR value distribution curve before denoising. D represents a schematic diagram of the spectrum analysis results for the target layer, with the horizontal axis representing the frequency (Hertz) and the vertical axis representing the amplitude (decibels). The green curve represents the spectrum distribution curve after denoising, and the red curve represents the spectrum distribution curve before denoising.

[0053] Figure 4 This is a comparison chart of well-controlled deconvolution correlation coefficients; where 0.3141 represents the cross-correlation coefficient between synthetic records and well-side seismic traces corresponding to a predicted deconvolution step size of 20ms, 0.3158 represents the cross-correlation coefficient between synthetic records and well-side seismic traces corresponding to a predicted deconvolution step size of 24ms, 0.3295 represents the cross-correlation coefficient between synthetic records and well-side seismic traces corresponding to a predicted deconvolution step size of 28ms, and 0.3265 represents the cross-correlation coefficient between synthetic records and well-side seismic traces corresponding to a predicted deconvolution step size of 32ms.

[0054] Figure 5 This is a comparison chart of the spectrum analysis before and after well-controlled deconvolution; the red curve is the spectrum curve before deconvolution, and the green curve is the spectrum curve after deconvolution. The horizontal axis represents frequency (in Hz), and the vertical axis represents amplitude (in decibels).

[0055] Figure 6 The diagram shows the velocity model before and after the initial tomography; the left diagram is the initial velocity profile in the depth domain, and the right diagram is the depth profile in the depth domain after three rounds of tomography; the horizontal axis represents the connecting line number, and the vertical axis represents the depth (unit: meters);

[0056] Figure 7 A planar plot of travel error calculated from initial arrival information; where the legend represents the percentage of relative travel error.

[0057] Figure 8The figures show the velocity profiles before and after combined tomography, as well as the remaining velocity profiles. The legend represents the percentage of relative travel time error. A is the depth-domain velocity profile before combined tomography, with the horizontal axis representing the connecting line number and the vertical axis representing depth (in meters); B is the depth-domain velocity profile after combined tomography, with the horizontal axis representing the connecting line number and the vertical axis representing depth (in meters); C is the percentage of relative travel time error, with the horizontal axis representing the connecting line number and the vertical axis representing depth (in meters).

[0058] Figure 9 This is a three-dimensional plot showing the relative error of the first arrival information, the relative error of the reflection information, and the well depth error; where the X-axis represents the main survey line direction number, the Y-axis represents the connecting line direction number, the Z-axis represents the depth (unit: meters), and the legend represents the percentage of relative travel time error;

[0059] Figure 10 This is a comparison diagram of the mirror imaging results and the full-wavelength interpretation results of the ultra-deep target layer stratigraphic structure in an embodiment of the present invention; wherein, the upper figure is a schematic diagram of the depth domain seismic migration profile, and the lower figure is a schematic diagram of the mirror energy superposition profile. The red arrow indicates the location of the improved continuity of the seismic phase axis in the mirror energy superposition profile. Different colored lines in the depth domain seismic migration profile represent geological strata, and different colored lines in the mirror energy superposition profile represent corresponding geological strata.

[0060] Figure 11 This is a comparison diagram of the reservoir imaging effect of the present invention and conventional results; wherein, the blue arrows represent plate-like reflections, the blue frame lines represent ultra-deep target layers, A is a profile diagram of the result obtained by the conventional method, the horizontal axis is the main survey line number, and the vertical axis is time (unit: milliseconds); B is a profile diagram of the result obtained by the present method, the horizontal axis is the main survey line number, and the vertical axis is time (unit: milliseconds). Detailed Implementation

[0061] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0062] It should be noted that, unless otherwise specified, the embodiments and features described in this application can be combined with each other. The specific implementation methods of the present invention will now be described in detail with reference to the accompanying drawings and embodiments.

[0063] The flowchart of the method for improving the accuracy of ultra-deep seismic data provided in this embodiment of the invention is as follows: Figure 1 As shown, the method includes steps S1-S4:

[0064] Step S1: Approximate true surface migration surface processing. Specifically, the seismic data of the ultra-deep target layer is processed using an approximate true surface migration surface to obtain the processed data.

[0065] In step S1, the approximate true surface offset surface processing includes steps S11-S14.

[0066] Step S11: Select the smooth surface of the ground elevation through the observation system to determine the approximate true ground surface offset surface;

[0067] Specifically, the choice of the approximate true surface determines the starting and ending points of the velocity modeling ray tracing, and also the starting point for updating the velocity. The shot point of the seismic data acquisition is located at the excitation well depth, while the receiver point is located on the surface. Therefore, there are two possible choices for the approximate true surface: the first is a combination of two surface types—the excitation surface corresponding to the shot point and the surface corresponding to the receiver point; the second is a small, smooth surface. The smoothing radius of the approximate true surface is determined by the shot distance and receiver distance, and must be greater than both.

[0068] Both offset surfaces can meet the requirements of approximating the true surface. The difference lies in the different static correction calculations. The first type of static correction calculation requires calculating the high-frequency static correction amount at the corresponding position of the shot receiver. The second type, in addition to calculating the high-frequency static correction amount, also requires calculating the static correction amount from the well depth to the smooth surface. This time can be calculated through the wellhead time or through tomographic static correction.

[0069] Step S12: Using micro-logging constraint inversion of the near-surface model, and based on the selected surface elevation smooth surface, calculate the static correction amount based on the approximate true ground surface and the static correction surface based on the reference surface.

[0070] Specifically, static correction for approximate true surfaces differs from conventional static correction. The main static correction calculated for approximate true surfaces involves high-frequency static corrections between the actual and approximate surfaces. The calculation process for approximate true surface static correction utilizes a micro-logging-constrained inversion of the surface velocity model. Then, given the approximate true surface as the final surface for static correction calculation, the shot point static correction and receiver static correction are directly calculated. These static corrections are then applied to the CMP gathers and first arrivals of the actual surface.

[0071] Step S13: Based on the static correction of the reference plane, a pre-stack denoising technique that classifies and steps to preserve effective low frequencies is applied to suppress noise. The effective low frequencies include the 8-14Hz effective signal, which is sensitive to ultra-deep target layers, and this 8-14Hz effective signal is given priority protection. The pre-stack denoising technique employs a radial domain subtraction method for selective denoising.

[0072] Specifically, in the pre-stack denoising stage, it is necessary to suppress noise while preserving amplitude and fidelity. Therefore, the overall design steps are based on a simulation subtraction method. First, the noise is simulated, and then the noise is subtracted from the original data. For pre-stack denoising of ultra-deep exploration target layers, the key lies in subtracting noise in the critical frequency band of 8-14Hz. For this part of the noise, a radial domain subtraction method is used for targeted denoising. This method is a process of extracting amplitude along a zero-offset straight line that is a function of time. The process is as follows: Figure 2 As shown, Figure 2 The image shows the effect before and after radial transformation of a single shot, where straight lines are superimposed on the shot gather. The amplitude along these lines is extracted, and these lines form radial paths. This process is a simple mapping of amplitude.

[0073] After determining the key steps of the radial domain subtraction method, a pre-stack denoising technique system for classifying and step-by-step preserving effective low frequencies for ultra-deep target layers was developed. As shown in Table 1, applying this technique system achieved the goal of protecting the low-frequency signals of ultra-deep target layers. The comparison of the ultra-deep target layer spectrum before and after pre-stack denoising to preserve low frequencies is as follows: Figure 3 As shown, from Figure 3 The spectrum analysis shows that the 8-14Hz low-frequency signal is well protected.

[0074] Table 1. Classification and Step-by-Step Effective Low-Frequency Technology System for Ultra-Deep Exploration Target Layers

[0075]

[0076] Step S14: Conduct surface consistency amplitude processing. This addresses consistency issues caused by varying surface shot detection conditions.

[0077] Next, proceed to step S2. Step S2: well-controlled two-step deconvolution, including surface-consistent deconvolution and predictive deconvolution selection. Specifically, for the processed data, well-controlled two-step deconvolution with extended frequency band is used to protect the effective low-frequency signal of 8-14Hz, resulting in deconvolution-processed data.

[0078] Specifically, the well-controlled deconvolution correlation coefficient is for example... Figure 4 As shown, the spectral analysis before and after well-controlled deconvolution is compared to... Figure 5 As shown in the figure. Through comparative analysis, it can be seen that the target layer in ultra-deep exploration is very sensitive to low-frequency signals, especially the effective low-frequency signal of 8-14Hz, which is key to ultra-deep imaging. However, 8-14Hz is also the main frequency band for noise development. Therefore, pre-stack denoising and frequency band extension for the target layer in ultra-deep exploration to protect the effective signal in this frequency band are key to improving the imaging of the target layer in ultra-deep exploration.

[0079] As mentioned above, the effective low-frequency signal of 8-14Hz is the key frequency band for imaging ultra-deep target layers. Therefore, protecting the low frequency and expanding the high frequency during the deconvolution process is a key strategy for imaging ultra-deep target layers.

[0080] Next, proceed to step S3. Step S3: Approximate true surface full-depth and omnidirectional velocity modeling. Specifically, using the data processed by deconvolution, approximate true surface full-depth and omnidirectional velocity modeling is performed to obtain the approximate true surface full-depth velocity modeling results for the ultra-deep target layer.

[0081] Ultra-deep target layers are deeply buried and greatly affected by shallow and intermediate layers. The accuracy of shallow and intermediate layer seismic data processing directly determines the imaging effect of ultra-deep target layers. Approximate true surface full-depth velocity modeling fundamentally considers the continuous changes from the surface to the deep layers, completely solving the problem of singular processing or only considering reflection wave processing, and minimizing the accuracy changes caused by the surface, thereby truly improving the imaging accuracy of ultra-deep target layers.

[0082] The approximate true surface full-depth and all-round velocity modeling process in step S3 includes steps S31-S35.

[0083] Step S31: Perform stacking velocity analysis, dynamic correction and residual static correction processing.

[0084] Using deconvolution-processed data, stacking velocity analysis, dynamic correction, and residual static correction are performed on ultra-deep target layer seismic data based on effective low-frequency preservation and bandwidth extension processing. The velocity obtained from the common reflection point time-distance curve is called the stacking velocity. Because stacking is performed after dynamic correction, if the velocity used for dynamic correction is appropriate, the effective wave energy after stacking will be the strongest; this velocity is called the optimal stacking velocity. Conversely, if the velocity used for dynamic correction is inappropriate, the effective wave energy after stacking will be weakened. The amplitude of the stacked record varies with the stacking velocity; this is the stacking velocity spectrum.

[0085] The stacking velocity analysis includes: (1) performing dynamic correction processing (NMO processing) on ​​ultra-deep target layer seismic data that has been processed to preserve effective low frequency and extended bandwidth; (2) stacking the dynamically corrected seismic data with a certain time delay to establish the relationship between the observation data; (3) velocity model assumptions: assuming different velocity models, and determining the most suitable velocity model by comparing the best fit between the stacking results under different velocity models and the observation data.

[0086] The residual static correction process aims to eliminate the residual error of the reference plane static correction and adjust the superposition phase of the common center point gathers in order to achieve in-phase superposition.

[0087] The remaining static corrections include: (1) Picking the time of each layer: In seismic data processing, it is necessary to first determine the time of each layer, which is the basis for subsequent processing. (2) Decomposing the static corrections of the source and receiver: By analyzing the seismic data, the static corrections of the source and receiver, as well as the tectonic time difference and dynamic correction time difference are decomposed. (3) Calculating the time difference caused by elevation differences: In the common center point gather, the data is separated according to the offset range, the average elevation of the near and far offset data is calculated, and the time difference caused by the elevation difference is obtained. (4) Obtaining the static corrections by surface consistency iteration: The surface consistency iteration is performed using the difference in the elevation static corrections of the near and far offsets to obtain the static corrections of the shot point and receiver point, and applied to the common center point gather.

[0088] S32: Remove the static correction based on the reference surface from the gather, apply the static correction based on the approximate true ground surface, and apply the static correction to the first arrival.

[0089] S33: Based on the static correction of the first arrival at an approximate true surface, conduct first arrival tomography based on the approximate true surface to invert the shallow and middle layer velocity model.

[0090] Based on the statically corrected first arrival at an approximate true ground surface, tomographic imaging is performed. This method mainly addresses the velocity inversion problem in shallow layers. Conventional reflection tomography is affected by the migration gather and cannot effectively invert the velocity in very shallow layers. As a result, the overall velocity model can only converge to flatten the gather. Since the target layer is very deep, the error will gradually amplify with the depth, ultimately preventing the imaging accuracy of the target layer from being effectively improved. Therefore, tomographic imaging for shallow and medium layers is an important step in improving accuracy.

[0091] The technical process of this method is as follows: by using two-point ray tracing, the error between the first arrival travel time and the travel time of the existing velocity model is calculated. The error is used to establish a tomography matrix, and the least squares method is used to solve the matrix update rate. This method does not limit the reflection depth, so the larger the first arrival offset, the greater the reflection depth, the greater the first arrival density, and the relatively high reflection accuracy.

[0092] Initial tomography does not require migration. The tomography process is iterated around the approximate migration plane. Quality control is mainly based on quantitative quality control, checking the relative travel time error. When the relative travel time error is less than 80% of a single sample point, the tomography can be considered complete and the speed converges. Figure 6 The velocity model before and after the initial arrival chromatography is shown. Figure 7 The diagram showing the travel time error for calculating the initial arrival information is presented.

[0093] S34: Conduct full-depth joint tomography imaging and simultaneously invert the velocity model across the entire depth range.

[0094] A near-true surface joint tomography was performed to update the full-depth velocity, and a near-true surface full-depth velocity model was constructed.

[0095] For velocity model updates, first-arrival tomography has high vertical resolution but relatively low horizontal resolution because the ray tracing process mainly involves lateral propagation. Reflection tomography, on the other hand, has high horizontal resolution but relatively low vertical resolution because the final ray propagation process is longitudinal. Combining the advantages of both methods can maximize the accuracy of the velocity model.

[0096] This method is based on approximate true ground surface migration. First-arrival tomography and reflection tomography use different methods for ray tracing, but calculate the same travel time error. A matrix equation is established for each pair of rays, and finally, the velocity is inverted using the least squares method. Because this method simultaneously utilizes first-arrival information and residual delay information from reflection gathers, establishes different methods for calculating travel time errors on the same matrix, and performs inversion simultaneously, it is one of the best methods currently available for improving the accuracy of velocity models. Figure 8 The velocity profiles before and after combined chromatography and their remaining velocity profiles are shown.

[0097] S35: Evaluate the velocity model through a quantitative quality control system.

[0098] The quality control of the approximate true surface full-depth velocity model is performed using the travel time error calculated from the first arrival information and the travel time error calculated from the reflection gather.

[0099] Quality control of the velocity model is a key aspect of velocity modeling. Conventional quality control mainly relies on qualitative methods such as velocity spectrum and gather flattening. Well error is the only quality control factor in conventional quality control.

[0100] In the full-depth modeling process, since the first arrival information is introduced, the error calculated from the first arrival information and the error calculated from the reflection gather can be used as the basis for quality control, thus realizing a quality control system from shallow to deep. This system mainly consists of three dimensions: the relative error with the first arrival information, the relative error with the reflection information, and the well depth error. The main indicators are: the error of the first arrival information is less than 0.02 sample points in 99% of cases, the relative error of the reflection information is less than 0.02 in 99% of cases, and the well depth error is less than one-thousandth. Figure 9 A three-dimensional diagram showing the relative error of the initial arrival information, the relative error of the reflection information, and the well depth error is presented.

[0101] Next, proceed to step S4. Step S4: Perform dual imaging of the ultra-deep target layer. Specifically, based on the approximate true surface full-depth velocity modeling results of the ultra-deep target layer, perform dual imaging of the ultra-deep target layer.

[0102] The dual imaging refers to the formation structure and reservoir imaging of the ultra-deep target layer.

[0103] Ultra-deep target layers suffer from low signal-to-noise ratios (SNR). Current methods to improve SNR primarily involve post-stack enhancement or noise attenuation targeting ultra-deep formations. Analysis shows that these methods can improve SNR to some extent, but their impact on fidelity is difficult to assess. Furthermore, the resulting data differs significantly from bare, biased data, failing to truly meet the needs of ultra-deep target layer exploration. Therefore, this invention proposes a two-step imaging strategy focusing on the stratigraphic structure and reservoir of ultra-deep target layers to simultaneously improve the SNR of ultra-deep target layers and ensure seismic data fidelity.

[0104] Step S4 specifically includes steps S41-S42.

[0105] Step S41: Based on the approximate true surface full-depth velocity modeling results of the ultra-deep target layer, image the stratigraphic structure of the ultra-deep target layer.

[0106] To improve the signal-to-noise ratio (SNR) of ultra-deep target layers, a highly reliable SNR enhancement technique is needed. Image processing based on angle-domain dip gathers is one of the most reliable and effective techniques currently available. This method first requires angle-domain migration imaging to obtain a three-dimensional dip-domain gather. Then, image energy separation is performed in the dip domain, and the gather is correlated with the original gather within a given time window to output the imaged dip-domain gather. This gather is then stacked to obtain the image. The results show that dip-domain image imaging can significantly improve the SNR of seismic data. Furthermore, through comprehensive interpretation, the image results are completely consistent with the full-wavefield imaging in terms of stratigraphic structure. Figure 10 The results show a comparison between mirror imaging of the stratigraphic structure of the ultra-deep target layer and full-field interpretation.

[0107] Step S42: Based on the approximate true surface full-depth velocity modeling results of the ultra-deep layers, image the reservoir of the ultra-deep target layer.

[0108] Ultra-deep target reservoirs are buried at great depths, resulting in severe absorption and attenuation by the earth. Meanwhile, reservoir prediction requires high-fidelity and amplitude-preserving imaging. Therefore, angle-domain illumination-compensated migration technology is selected as a key technology to improve the imaging of ultra-deep target reservoirs. This technology performs imaging in the local angle domain, and during the migration process, illumination compensation is used to solve the amplitude changes caused by lateral changes and longitudinal attenuation of the observation system during the acquisition process. At the same time, it combines the characteristics of X-ray beam imaging and tilt stacking technology to ensure the signal-to-noise ratio and fidelity and amplitude preservation characteristics, thus ensuring the correct imaging of ultra-deep target reservoirs. Figure 11The invention illustrates a comparison between reservoir imaging results from embodiments of the present invention and conventional results. The conventional imaging method is as follows: In the static correction stage, the conventional processing flow generally adopts a high-low frequency separation processing technology and does not perform processing based on full-depth true surface imaging. In the deconvolution stage, it generally focuses on wavelet consistency and high-resolution application effects, and pays less attention to low-frequency extended parameters. In the velocity modeling stage, it generally adopts a two-step depth domain modeling imaging technology and does not perform full-depth velocity modeling. In the imaging stage, it only uses velocity for depth migration imaging and does not perform mirror imaging for ultra-deep target layers.

[0109] Finally, it should be noted that the above content is only used to illustrate the technical solution of the present invention, and is not intended to limit the scope of protection of the present invention. Simple modifications or equivalent substitutions made by those skilled in the art to the technical solution of the present invention do not depart from the essence and scope of the technical solution of the present invention.

Claims

1. A method for improving the imaging accuracy of ultra-deep seismic data, characterized in that, Includes the following steps: Step S1: Perform approximate true surface migration surface processing on the seismic data of the ultra-deep target layer to obtain the processed data; Step S2: For the processed data, a two-step well control deconvolution process with extended frequency band is used to protect the effective low-frequency signal of 8-14Hz, and the data after deconvolution is obtained. Step S3: Using the data processed by deconvolution, perform approximate true surface full-depth and all-round velocity modeling to obtain the approximate true surface full-depth velocity modeling results for the ultra-deep target layer. Step S4: Based on the approximate true surface full-depth velocity modeling results of the ultra-deep target layer, perform dual imaging of the ultra-deep target layer.

2. The method according to claim 1, characterized in that, In step S2, the extended frequency band is the extended high-frequency frequency band.

3. The method according to claim 1, characterized in that, In step S1, the approximate true surface offset surface processing includes: Step S11: Select the smooth surface of the ground elevation through the observation system to determine the approximate true ground surface offset surface; Step S12: Using micro-logging constrained inversion of the near-surface model, and based on the selected surface elevation smooth surface, calculate the static correction amount based on the approximate true ground surface and the static correction surface based on the reference surface. Step S13: Based on the static correction of the reference plane, apply the pre-stack denoising technique of classifying and step-by-step preservation of effective low frequencies to suppress noise; Step S14: Conduct surface uniform amplitude processing.

4. The method according to claim 3, characterized in that, In step S13, the effective low frequency includes an effective signal of 8-14 Hz that is sensitive to ultra-deep target layers.

5. The method according to claim 3, characterized in that, In step S13, the pre-stack denoising technique employs radial domain subtraction for selective denoising.

6. The method according to claim 1, characterized in that, Step S3, which involves approximating the true surface velocity modeling across all depths and directions, includes: Step S31: Perform stacking velocity analysis, dynamic correction, and residual static correction processing; Step S32: Remove the static correction based on the reference surface from the gather, apply the static correction based on the approximate true ground surface, and apply the static correction to the first arrival. Step S33: Based on the statically corrected first arrival at an approximate true surface, conduct first arrival tomography based on the approximate true surface to invert the shallow and middle layer velocity model; Step S34: Perform full-depth joint tomography and simultaneously invert the velocity model across the entire depth range; Step S35: Evaluate the velocity model using a quantitative quality control system.

7. The method according to claim 1, characterized in that, In step S4, the dual imaging is the formation structure and reservoir imaging of the ultra-deep target layer.

8. The method according to claim 7, characterized in that, Step S4 specifically includes the following steps: Step S41: Based on the approximate true surface full-depth velocity modeling results of the ultra-deep target layer, perform angle domain migration to obtain a dip domain three-dimensional gather; Step S42: Based on the dip domain 3D gather, perform dip domain mirror imaging on the formation structure of the ultra-deep target layer to obtain mirror results for the formation structure of the ultra-deep target layer; for the ultra-deep target layer reservoir, perform angle domain illumination-compensated migration imaging and full-wavelength stacking to obtain full-wavelength reservoir imaging results for the ultra-deep target layer.

9. A system for improving the imaging accuracy of ultra-deep seismic data, characterized in that, Includes the following modules: The data processing module is configured to perform the method as described in any one of claims 1-8; The observation module is used to select the smooth surface of the ground elevation and determine the approximate true ground surface offset surface; The deconvolution module is used to implement well-controlled two-step deconvolution processing with extended frequency bands; The velocity modeling module is used to perform velocity modeling processing that approximates the true Earth surface at all depths and in all directions. The imaging module is used for dual imaging of ultra-deep target layers based on velocity modeling results.

10. A device for improving the imaging accuracy of ultra-deep seismic data, characterized in that, Includes the following units: Processing unit, configured to perform the method as described in any one of claims 1 to 7; Storage units are used to store seismic data and intermediate results generated during processing; The input / output unit is used to receive input data and output processing results; The display unit is used to display the processing results and the image.

Images