A spatiotemporal joint sparse deconvolution method and system for enhancing geological body recognition accuracy

Abstract

The present invention discloses a spatiotemporal joint sparse deconvolution method and system for enhancing the accuracy of geological body identification, the method comprising the following steps: extracting seismic wavelets of input seismic data along the time direction; preliminarily estimating the boundary smoothing window function of the target geological body along the space direction, and then obtaining the boundary smoothing function that meets the preset requirements by the knife edge method; obtaining a synthetic point spread function based on the seismic wavelet and the boundary smoothing function, and performing spatiotemporal joint deconvolution on the seismic data based on the synthetic point spread function, updating the boundary smoothing function until the boundary identification accuracy of the target geological body meets the preset requirements; performing spatial smoothing on the deconvolution output result to obtain the output result of spatiotemporal joint sparse deconvolution. The present invention has good algorithm integration, can be processed modularly, is convenient for use by seismic data processing personnel, and has high promotion value.

Description

Technical Field

[0001] The present invention belongs to the technical field of clear identification of geological bodies, and in particular relates to a spatiotemporal joint sparse deconvolution method and system for enhancing the accuracy of geological body identification. Background Art

[0002] Post-stack seismic data used for studies such as horizon tracing, fault interpretation, and sedimentary system identification require high temporal and spatial resolution. Temporal resolution can be improved through deconvolution, and spatial resolution is usually improved based on migration methods. Although there are super-resolution algorithms based on deep learning algorithms, this requires a large amount of training sets and time.

[0003] Deconvolution methods for post-stack seismic data include sparse deconvolution, time-frequency domain deconvolution, spectral inversion, etc. Sparse deconvolution assumes that the reflection coefficient is sparse, uses the compressed sensing algorithm to invert the reflection coefficient, and then convolves the high-resolution wavelet. At the same time, by combining the continuity constraints in the spatial direction, a multi-channel deconvolution algorithm that maintains lateral continuity can be obtained; time-frequency domain deconvolution adds window segmentation to the time direction and considers the window to be stable, which makes time-frequency domain deconvolution suitable for high-resolution processing of non-stationary signals; spectral inversion introduces parity decomposition to reduce tuning phenomena, and can independently select the seismic frequency bands involved in deconvolution; deep learning algorithms use the powerful nonlinear fitting ability of deep learning to broaden the frequency band and improve the resolution of data by learning the mapping relationship from low-resolution to high-resolution seismic data.

[0004] In terms of improving spatial resolution, the migration method can improve the structural ambiguity in the spatial direction caused by noise suppression, spatial smoothing and other processing methods. The migration algorithm based on wave theory can improve the spatial resolution to a certain extent; and considering the seismic attenuation of wave theory, the Q migration algorithm can compensate for both temporal resolution and spatial resolution. When the data contains geological bodies with spatial mutations such as faults, incised valleys or river channels, improving the lateral resolution through these methods is particularly important for identifying these geological bodies.

[0005] Existing methods of improving the spatial resolution of seismic data through migration technology usually combine spatial smoothing functions to spatially smooth seismic profiles. Spatial smoothing functions have similar characteristics to seismic wavelets, which can be summarized as follows: 1. Spatial smoothing functions have a greater impact on large angles, and the greater the inclination angle, the more serious the impact (such as faults); 2. Strata with smaller inclination angles only have an impact at the boundaries of the spatial direction; 3. Spatial smoothing functions are usually short in duration to avoid excessive loss of effective signals. Based on the above characteristics, smoothing in the spatial direction will blur geological bodies with smaller widths (branch rivers, caves, etc.) and geological phenomena with too large inclination angles (faults), and it is still impossible to accurately identify those specific geological bodies with sudden spatial changes. Summary of the invention

[0006] In view of the shortcomings of the prior art, the present invention proposes a spatiotemporal joint sparse deconvolution method and system for enhancing the accuracy of geological body identification. The defuzzification theory of image processing is introduced into post-stack data processing, seismic wavelets are extracted in the time direction, and the spatial smoothing function is extracted in the spatial direction by the "knife edge method", and the two are combined to construct a point spread function. At the same time, sparse constraints are added to the high-resolution seismic data, and the reflection coefficient is obtained by solving the optimization equation, and then the high-resolution wavelet is convolved to obtain data with improved time and space resolution, so as to accurately identify those specific geological bodies with sudden changes in space.

[0007] To achieve the above object, the present invention provides the following solutions:

[0008] A spatiotemporal joint sparse deconvolution method for enhancing the accuracy of geological body recognition comprises the following steps:

[0009] Extracting seismic wavelets from input seismic data along the time direction;

[0010] A preliminary estimate of the boundary smoothing window function of the target geological body is made along the spatial direction, and then a boundary smoothing function that meets the preset requirements is obtained by the knife edge method;

[0011] Based on the seismic wavelet and the boundary smoothing function, a synthetic point spread function is obtained, and based on the synthetic point spread function, a spatiotemporal joint deconvolution is performed on the seismic data, and the boundary smoothing function is updated until the target geological body boundary recognition accuracy meets the preset requirements;

[0012] The deconvolution output is spatially smoothed to obtain the output of spatiotemporal joint sparse deconvolution.

[0013] Preferably, the method for extracting seismic wavelets from input seismic data along the time direction comprises:

[0014]

[0015] in, is the sparse reflection coefficient in the seismic frequency band, is the reflection coefficient, For low spatial resolution spatiotemporal data, is the Fourier transform matrix in the time direction, is the Fourier transform matrix in the spatial direction, is the dot product, is the point spread function, is the regularization operator, represents discrete Fourier transform.

[0016] Preferably, the method of preliminarily estimating the boundary smoothing window function of the target geological body along the spatial direction and then obtaining the boundary smoothing function that meets the preset requirements by the knife edge method includes:

[0017]

[0018] in, for The inverse transform of the modulus of the Fourier transform, is the differential of a segment of data taken from the seismic data.

[0019] Preferably, the method of performing spatial smoothing on the deconvolution output result to obtain the spatiotemporal joint sparse deconvolution output result includes:

[0020] Test the high-resolution processing effects of boundary smoothing functions of different lengths, use geological knowledge to select the deconvolution processing results that meet the preset requirements, and use spatial 3-point smoothing denoising as needed;

[0021] The pseudo-Wheeler domain data is reverse flattened, and the wavelet is convolved as needed to output high-resolution seismic records or directly output reflection coefficients.

[0022] The present invention also provides a spatiotemporal joint sparse deconvolution system for enhancing the accuracy of geological body identification, comprising: a seismic wavelet extraction module, a boundary smoothing window function calculation module, a spatiotemporal joint deconvolution module and a spatial smoothing module;

[0023] The seismic wavelet extraction module is used to extract the seismic wavelet of the input seismic data along the time direction;

[0024] The boundary smoothing window function calculation module is used to preliminarily estimate the boundary smoothing window function of the target geological body along the spatial direction, and then obtain the boundary smoothing function that meets the preset requirements through the knife edge method;

[0025] The spatiotemporal joint deconvolution module is used to obtain a synthetic point spread function based on the seismic wavelet and the boundary smoothing function, and to perform spatiotemporal joint deconvolution on the seismic data based on the synthetic point spread function, and to update the boundary smoothing function until the target geological body boundary recognition accuracy meets the preset requirements;

[0026] The spatial smoothing module is used to perform spatial smoothing on the deconvolution output result to obtain the spatiotemporal joint sparse deconvolution output result.

[0027] Preferably, the process of extracting seismic wavelets from input seismic data along the time direction includes:

[0028]

[0029] in, is the sparse reflection coefficient in the seismic frequency band, is the reflection coefficient, For low spatial resolution spatiotemporal data, is the Fourier transform matrix in the time direction, is the Fourier transform matrix in the spatial direction, is the dot product, is the point spread function, is the regularization operator, represents discrete Fourier transform.

[0030] Preferably, the process of preliminarily estimating the boundary smoothing window function of the target geological body along the spatial direction and then obtaining the boundary smoothing function that meets the preset requirements by the knife edge method includes:

[0031]

[0032] in, for The inverse transform of the modulus of the Fourier transform, is the differential of a segment of data taken from the seismic data.

[0033] Preferably, the process of performing spatial smoothing on the deconvolution output result to obtain the spatiotemporal joint sparse deconvolution output result includes:

[0034] Test the high-resolution processing effects of boundary smoothing functions of different lengths, use geological knowledge to select the deconvolution processing results that meet the preset requirements, and use spatial 3-point smoothing denoising as needed;

[0035] The pseudo-Wheeler domain data is reverse flattened, and the wavelet is convolved as needed to output high-resolution seismic records or directly output reflection coefficients.

[0036] Compared with the prior art, the present invention has the following beneficial effects:

[0037] 1. The present invention is a technology for enhancing the accuracy of geological body recognition based on spatiotemporal joint sparse deconvolution. It is first applied to seismic data imaging to improve spatial detail information and has achieved good results in two-dimensional seismic data testing.

[0038] 2. The present invention introduces the deblurring theory of image processing into post-stack data processing, and uses the "knife edge method" to extract the smoothing function of the edge of the target geological body in the pseudo-Wheeler domain, thereby avoiding the blurring phenomenon caused by the geological body with a large dip angle in the seismic data;

[0039] 3. The present invention firstly uses point spread function and seismic records for time-space joint deconvolution to determine whether the boundary of the target geological body is clear. The obtained results greatly enrich the detailed information of the target geological body.

[0040] 4. The present invention utilizes a spatial smoothing module to remove artifacts from the output deconvolution data, thereby greatly improving the accuracy of identifying geological bodies on seismic profiles;

[0041] 5. The present invention can realize modular processing and has high promotion value. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] In order to more clearly illustrate the technical solution of the present invention, the following briefly introduces the drawings required for use in the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.

[0043] Figure 1 This is a flow chart of a spatiotemporal joint sparse deconvolution technique for enhancing the accuracy of geological body recognition according to an embodiment of the present invention;

[0044] Figure 2 A schematic diagram of extracting a spatial smoothing function using the knife edge method according to an embodiment of the present invention;

[0045] Figure 3 The Marmousi model reflection coefficient, synthetic seismic record and their local magnified diagrams of the embodiment of the present invention; wherein, Figure 3 (a) is a schematic diagram of reflection coefficient; Figure 3 (b) is a schematic diagram of synthetic seismic records;

[0046] Figure 4 : are reflection coefficients obtained by different deconvolution methods according to the embodiments of the present invention and their enlarged diagrams; wherein, Figure 4 (a) is a schematic diagram of sparse spike deconvolution; Figure 4 (b) is a schematic diagram of multidimensional sparse deconvolution;

[0047] Figure 5 A schematic diagram of a spatial smoothing function of actual seismic data and different position estimates according to an embodiment of the present invention;

[0048] Figure 6 Schematic diagram of the data before and after deconvolution and the fault extracted by the Maxlikelihood method in the embodiment of the present invention; wherein, Figure 6 (a) is a schematic diagram of the original seismic data; Figure 6 (b) is a schematic diagram of sparse spike deconvolution; Figure 6 (c) is a schematic diagram of multidimensional sparse deconvolution;

[0049] Figure 7 It is a partial enlarged diagram of the data before and after deconvolution in an embodiment of the present invention; wherein, Figure 7 (a) is a schematic diagram of the original seismic data; Figure 7 (b) is a schematic diagram of sparse spike deconvolution; Figure 7 (c) is a schematic diagram of multidimensional sparse deconvolution;

[0050] Figure 8 Schematic diagram of the FK spectrum of data before and after deconvolution according to an embodiment of the present invention; wherein, Figure 8 (a) is a schematic diagram of the original seismic data; Figure 8 (b) is a schematic diagram of sparse spike deconvolution; Figure 8 (c) in the figure is a schematic diagram of multidimensional sparse deconvolution. DETAILED DESCRIPTION

[0051] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0052] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the present invention is further described in detail below with reference to the accompanying drawings and specific embodiments.

[0053] Embodiment 1

[0054] like Figure 1 As shown, the present invention provides a spatiotemporal joint sparse deconvolution method for enhancing the accuracy of geological body recognition, comprising the following steps:

[0055] Extracting seismic wavelets from input seismic data along the time direction;

[0056] Make a preliminary estimate of the boundary smoothing window function of the target geological body along the spatial direction, and then obtain a suitable boundary smoothing function through the knife edge method;

[0057] The point spread function calculated from the results of the first two steps is used to perform spatiotemporal joint deconvolution on the seismic data to obtain geological body model information. In the process of spatiotemporal joint deconvolution, the boundary smoothing function obtained in the previous step is iteratively updated to re-characterize the target geological body until the boundary of the target geological body is clearly identified.

[0058] The spatial smoothing is performed on the spatiotemporal joint deconvolution result output after iteration to eliminate the illusions that may be caused in the iteration process, and finally the output result of the spatiotemporal joint sparse deconvolution is obtained.

[0059] The algorithm of the present invention is well integrated and can be processed in a modular manner, so it is convenient for seismic data processing personnel to use and has high promotion value.

[0060] Sparse pulse deconvolution method considers post-stack seismic data It can be regarded as an earthquake wavelet and reflection coefficient The convolution of

[0061] (1)

[0062] in, is a random noise with Gaussian distribution; when the seismic data has completed good amplitude and frequency compensation and the spatial structure is relatively stable, the seismic wavelet Without time-varying and space-varying characteristics, then, by estimating the seismic wavelet, the reflection coefficient approximation within the seismic frequency band can be obtained based on the deconvolution method, that is,

[0063] (2)

[0064] In the formula, represents the regularization operator. It can also be assumed that the reflection coefficient is sparse, that is,

[0065] (3)

[0066] Formula (2) is the least square deconvolution, and formula (3) is the sparse pulse deconvolution. Under the action of a suitable low-pass filter, the low-pass filtering of formula (3) is equivalent to formula (2). There are mature studies on the solution method of formula (3), including iterative soft threshold, alternating direction multiplier method, convex set projection method, etc.

[0067] Due to the convolution assumption of self-excitation and self-collection, and the one-dimensional characteristics of the wavelet, this deconvolution method only improves the resolution in the time direction; however, the acquisition quality of a considerable portion of data is poor, or algorithms such as spatial smoothing are used to suppress noise during processing, resulting in a reduction in the resolution of the data in the spatial direction, affecting the identification of geological bodies such as faults and reservoir boundaries. It is very necessary to perform spatial deconvolution on data with a high degree of spatial smoothness. Spatial deconvolution focuses more on using a deblurring algorithm to improve lateral resolution, and is a super-resolution algorithm.

[0068] Assume that there is spatial smoothing along the formation strike when processing data, and this smoothing is independent of time, that is,

[0069] (4)

[0070] in, It is spatiotemporal data with low spatial resolution. is a low-pass filter, which suppresses random noise, large-angle coherent noise, etc. by reducing the energy of high wave numbers. Then, substitute equation (1) into equation (4), that is,

[0071] (5)

[0072] Then low-resolution data It can be regarded as a point spread function and reflection coefficient Two-dimensional linear convolution is formed. Therefore, if the point spread function is known , through deconvolution, the temporal and spatial resolutions can be enhanced together. According to the properties of Fourier transform, the convolution in the time and space domain is equivalent to the product in the frequency and wave number domain, so equation (5) can be written as:

[0073] (6)

[0074] in, , and Respectively , and The two-dimensional Fourier transform of represents the dot product. The matrix or vector form is shown in bold, i.e.,

[0075] (7)

[0076] in, represents the discrete Fourier transform, that is,

[0077] (8)

[0078] in, .when hour, is the Fourier transform matrix in the time direction, when hour, is the Fourier transform matrix in the spatial direction. We hope to obtain seismic data with enhanced resolution , with L1 norm constraint (7) , we can get the seismic data with enhanced resolution. The process can be expressed as:

[0079] (9)

[0080] In this method, if the data dimension is high, equation (9) can be generalized in the form of a tensor. For example, for three-dimensional data, it is only necessary to transform the FK in two directions into the FKK transformation in three directions.

[0081] In addition, when the dip angle of the formation data is large, or there is an unconformity caused by formation pinching or missing, we need to use the existing layers or obtain the formation dip angle to approximately flatten the formation. There are many methods for obtaining the relevant formation dip angle, such as the cross-correlation method, the plane wave decomposition method, etc. The flattening of the layers corresponding to multiple phase axes of the same geological age can obtain the seismic data in the Wheeler domain. If the thickness of these layers is very different, direct flattening will destroy the structure of the seismic wavelet. Therefore, only a single layer is flattened here to obtain the seismic data in the pseudo-Wheeler domain, and then the spatial low-pass filter is extracted. The high-resolution seismic data is obtained using formula (9), and finally the data is flattened to obtain the seismic data with enhanced temporal and spatial resolution.

[0082] The spatial smoothing function has similar characteristics to the seismic wavelet, and has the following characteristics:

[0083] 1. The spatial smoothing function has a greater impact on large angles, and the greater the dip angle, the more serious the impact (such as faults);

[0084] 2. Strata with small dip angles only have an impact at the boundaries of the spatial direction;

[0085] 3. The spatial smoothing function is usually short in duration to avoid excessive loss of valid signals.

[0086] Based on the above characteristics, smoothing in the spatial direction will blur the geological bodies with small width (such as branched rivers, caves, etc.) and geological phenomena with large inclination angles (such as faults); then, based on the edge method, it is only necessary to find the boundary of the geological body in the pseudo-Wheeler seismic data body, and then obtain the smoothing function through differentiation, and finally zero-phase and re-normalize it. The schematic diagram of the process is shown in Figure 2 shown.

[0087] Assume that the target geological body position is defined as , because only the spatial boundary of the geological body needs to be selected, so , which means taking a section of data from the seismic data. First, differentiate this part of data, that is:

[0088] (12)

[0089] Generally speaking, spatial filtering and smoothing of data require a zero-phase filter to avoid changing the formation position after filtering. Therefore, we need to use Fourier transform to Zero phase, that is, take The inverse transform of the modulus of the Fourier transform, that is,

[0090] (13)

[0091] in, and Represents the forward and inverse Fourier transform. Finally, Normalization can obtain the extracted spatial smoothing function, that is,

[0092] (14)

[0093] The extracted spatial smoothing function can be approximated using a Gauss window function with a clear expression, which makes it easy to change the length of the window function, that is,

[0094] (15)

[0095] Obviously, the same The length of the closest Gauss window function. However, due to the spatial smoothing function Very short, easy to observe directly The waveform of gives the closest Gauss window. In addition to the Gaussian window, a window function with a clear interpretation can also be used for fitting.

[0096] In summary, the specific process of spatiotemporal joint sparse deconvolution technology for enhancing the accuracy of geological body recognition can be described as follows:

[0097] 1. Flatten the seismic data into the pseudo-Wheeler domain using horizons or calculated dips;

[0098] 2. Use the data amplitude spectrum to extract the wavelet, find the fault or smaller geological body, use the edge method to estimate the spatial smoothing function, and synthesize the seismic wavelet and smoothing function into a point spread function;

[0099] 3. Test the high-resolution processing effects of smoothing functions of different lengths, select the most appropriate deconvolution processing results based on geological knowledge, and use spatial three-point smoothing and denoising as needed;

[0100] 4. Anti-flatten the pseudo-Wheeler domain data, convolve the wavelet as needed to output high-resolution seismic records or directly output reflection coefficients.

[0101] Embodiment 2

[0102] The present invention also provides a spatiotemporal joint sparse deconvolution system for enhancing the accuracy of geological body identification, which mainly includes four modules: the first is a seismic wavelet extraction module for calculating the point spread function; the second is a boundary smoothing window function calculation module, which obtains the boundary smoothing function of the appropriate target geological body by combining the knife edge method with the Gauss window function fitting; the third is a spatiotemporal joint deconvolution module, which performs spatiotemporal joint deconvolution on the point spread function calculated by the results output by the first two modules and the input seismic data, and the obtained result is used to judge whether the boundary of the target geological body is clear; the fourth is a spatial smoothing module, which is used to eliminate the illusion that may appear in module 3. Through the above steps, the output result of the spatiotemporal joint sparse deconvolution is finally obtained.

[0103] Specifically, the seismic wavelet extraction module is used to extract the seismic wavelet of the input seismic data along the time direction;

[0104] The boundary smoothing window function calculation module is used to preliminarily estimate the boundary smoothing window function of the target geological body along the spatial direction, and then obtain the boundary smoothing function that meets the preset requirements through the knife edge method;

[0105] The spatiotemporal joint deconvolution module is used to obtain a synthetic point spread function based on the seismic wavelet and the boundary smoothing function, and to perform spatiotemporal joint deconvolution on the seismic data based on the synthetic point spread function, and to update the boundary smoothing function until the target geological body boundary recognition accuracy meets the preset requirements;

[0106] The spatial smoothing module is used to perform spatial smoothing on the deconvolution output result to obtain the spatiotemporal joint sparse deconvolution output result.

[0107] In this embodiment, the process of extracting the seismic wavelet of the input seismic data along the time direction includes:

[0108]

[0109] in, is the sparse reflection coefficient in the seismic frequency band, is the reflection coefficient, For low spatial resolution spatiotemporal data, is the Fourier transform matrix in the time direction, is the Fourier transform matrix in the spatial direction, is the dot product, is the point spread function, is the regularization operator, represents discrete Fourier transform.

[0110] In this embodiment, the process of preliminarily estimating the boundary smoothing window function of the target geological body along the spatial direction and then obtaining the boundary smoothing function that meets the preset requirements by the knife edge method includes:

[0111]

[0112] in, for The inverse transform of the modulus of the Fourier transform, is the differential of a segment of data taken from the seismic data.

[0113] In this embodiment, the process of performing spatial smoothing on the deconvolution output result to obtain the spatiotemporal joint sparse deconvolution output result includes:

[0114] Test the high-resolution processing effects of boundary smoothing functions of different lengths, use geological knowledge to select the deconvolution processing results that meet the preset requirements, and use spatial 3-point smoothing denoising as needed;

[0115] The pseudo-Wheeler domain data is reverse flattened, and the wavelet is convolved as needed to output high-resolution seismic records or directly output reflection coefficients.

[0116] Embodiment 3

[0117] Test 1:

[0118] In order to test the applicability of the algorithm to complex models, the present invention performs a reflection coefficient inversion test based on the Marmousi model ( Figure 3 (a) in the figure), the data sampling interval is 2ms, based on 30Hz Ricker wavelet convolution synthesis, and a 15-point Gaussian filter is applied in the spatial direction to simulate smoothing ( Figure 3 (b) in the figure); the test results show that for data containing faults, spatial smoothing reduces the clarity of the faults, and for these high-angle strata, direct reflection coefficient inversion may cause false axes in the data ( Figure 4 (a) in the figure), while the data with enhanced lateral resolution can better adapt to the reflection coefficient inversion of high-angle strata, thereby enhancing the clarity of faults ( Figure 4 (b) in the figure.

[0119] Test 2:

[0120] For seismic data with high-angle faults, the spatial smoothing may also reduce the clarity of the faults, affecting the seismic interpretation and the identification of related structural oil and gas reservoirs. We used Wu's actual data to test the two-dimensional sparse deconvolution algorithm. We directly obtained this public data from the article using screenshots. Due to resolution issues, the data is inherently blurred. The spatial smoothing function is estimated based on the data breakpoints. The corresponding positions and extracted spatial smoothing functions are shown in Figure 2. Figure 5 As shown in Figure 2. Since the fault dip is large and the data structure is relatively flat, no pseudo-Wheeler domain conversion was performed. We applied traditional sparse pulse deconvolution for comparison, applied low-pass filtering with the same bandwidth to the deconvolved reflection coefficient, and extracted faults from the data before and after deconvolution based on the widely recognized Maxlikelihood algorithm. After deconvolution, we found that both methods improved the vertical resolution, and the event axis became significantly thinner ( Figure 6 ), but interestingly, the profile after sparse pulse deconvolution is not significantly different from that before deconvolution after tomography extraction ( Figure 6 (a) and Figure 6 (b) in the figure), but after we enhanced the lateral resolution, the inheritance of deep and shallow faults was improved, and some small faults could be identified through the fault extraction algorithm ( Figure 6 (a) and Figure 6 (c) in ). And if Figure 7 As shown in Figure 2, after the spatial resolution is improved, some detailed information is added to the data profile. Figure 8 ) It can also be observed that the width of the wavenumber has been improved. This data shows that our method of enhancing the lateral resolution is effective in improving the clarity of the fault.

[0121] The embodiments described above are only descriptions of the preferred embodiments of the present invention and are not intended to limit the scope of the present invention. Without departing from the design spirit of the present invention, various modifications and improvements made to the technical solutions of the present invention by ordinary technicians in this field should all fall within the protection scope determined by the claims of the present invention.

Claims

1. A spatiotemporal joint sparse deconvolution method for enhancing the accuracy of geological body recognition, characterized in that: The following steps are involved: Extracting seismic wavelets from input seismic data along the time direction; A preliminary estimate of the boundary smoothing window function of the target geological body is made along the spatial direction, and then a boundary smoothing function that meets the preset requirements is obtained by the knife edge method; Based on the seismic wavelet and the boundary smoothing function, a synthetic point spread function is obtained, and based on the synthetic point spread function, a spatiotemporal joint deconvolution is performed on the seismic data, and the boundary smoothing function is updated until the target geological body boundary recognition accuracy meets the preset requirements; Perform spatial smoothing on the deconvolution output to obtain the output of spatiotemporal joint sparse deconvolution; The method of extracting seismic wavelets from input seismic data along the time direction includes: in, is the sparse reflection coefficient in the seismic frequency band, is the reflection coefficient, For low spatial resolution spatiotemporal data, is the Fourier transform matrix in the time direction, is the Fourier transform matrix in the spatial direction, is the dot product, is the point spread function, is the regularization operator, represents discrete Fourier transform; Methods for preliminarily estimating the boundary smoothing window function of the target geological body along the spatial direction and then obtaining the boundary smoothing function that meets the preset requirements by the knife edge method include: in, for The inverse transform of the modulus of the Fourier transform, is the differential of a segment of data taken from the seismic data; The method of performing spatial smoothing on the deconvolution output to obtain the output result of spatiotemporal joint sparse deconvolution includes: Test the high-resolution processing effects of boundary smoothing functions of different lengths, use geological knowledge to select the deconvolution processing results that meet the preset requirements, and use spatial 3-point smoothing denoising as needed; The pseudo-Wheeler domain data is reverse flattened, and the wavelet is convolved as needed to output high-resolution seismic records or directly output reflection coefficients.

2. A spatiotemporal joint sparse deconvolution system for enhancing the accuracy of geological body recognition, characterized in that: include: Seismic wavelet extraction module, boundary smoothing window function calculation module, spatiotemporal joint deconvolution module and spatial smoothing module; The seismic wavelet extraction module is used to extract the seismic wavelet of the input seismic data along the time direction; The boundary smoothing window function calculation module is used to preliminarily estimate the boundary smoothing window function of the target geological body along the spatial direction, and then obtain the boundary smoothing function that meets the preset requirements through the knife edge method; The spatiotemporal joint deconvolution module is used to obtain a synthetic point spread function based on the seismic wavelet and the boundary smoothing function, and to perform spatiotemporal joint deconvolution on the seismic data based on the synthetic point spread function, and to update the boundary smoothing function until the target geological body boundary recognition accuracy meets the preset requirements; The spatial smoothing module is used to perform spatial smoothing on the deconvolution output result to obtain the output result of spatiotemporal joint sparse deconvolution; The process of extracting seismic wavelets from input seismic data along the time direction includes: in, is the sparse reflection coefficient in the seismic frequency band, is the reflection coefficient, For low spatial resolution spatiotemporal data, is the Fourier transform matrix in the time direction, is the Fourier transform matrix in the spatial direction, is the dot product, is the point spread function, is the regularization operator, represents discrete Fourier transform; The process of preliminarily estimating the boundary smoothing window function of the target geological body along the spatial direction and then obtaining the boundary smoothing function that meets the preset requirements through the knife edge method includes: in, for The inverse transform of the modulus of the Fourier transform, is the differential of a segment of data taken from the seismic data; The process of spatially smoothing the deconvolution output to obtain the spatiotemporal joint sparse deconvolution output includes: Test the high-resolution processing effects of boundary smoothing functions of different lengths, use geological knowledge to select the deconvolution processing results that meet the preset requirements, and use spatial 3-point smoothing denoising as needed; The pseudo-Wheeler domain data is reverse flattened, and the wavelet is convolved as needed to output high-resolution seismic records or directly output reflection coefficients.

Images