Broadband elastic parameter inversion method based on seismic frequency division body

Abstract

The present application provides a kind of wideband elastic parameter inversion method based on seismic frequency body, which includes: step 1, seismic effective band analysis and time-frequency decomposition are carried out;Step 2, extract frequency seismic matching seismic wavelet;Step 3, carry out joint reflection coefficient inversion of frequency seismic data;Step 4, integrate reflection coefficient integral impedance body with low frequency model body.Compared with the traditional seismic inversion method, the wideband elastic parameter inversion method based on seismic frequency body considers the energy difference characteristics of seismic high, medium and low frequency band signals, uses time-frequency decomposition body, constructs joint inversion objective function, increases balance constraint parameters and optimizes solution, weakens the suppression and influence of strong energy signal band on weak signal band in medium frequency dominant band, fully plays the role of seismic full frequency information, and finally obtains wideband inversion data body, improves the ability of geological reservoir prediction and description.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of seismic data processing, in particular to a wideband elastic parameter inversion method based on seismic frequency volume. BACKGROUND

[0002] Seismic inversion is an important geophysical reservoir prediction technology to remove seismic wavelet tuning effect and convert seismic interface reflection information into interlayer information, which plays an important role in improving the prediction and description ability of geological body. Seismic is a multi-frequency finite signal. Due to the difference in excitation energy and propagation energy attenuation, there is a difference in energy of different frequency bands of seismic signal, and the energy of middle frequency band is often strong, which is the dominant frequency band of seismic. The conventional seismic inversion objective function construction and solving do not consider the difference in energy of different frequency bands, and the multi-frequency information is not distinguished and applied at the same time. The strong signal in the main frequency band plays a leading role in seismic inversion, and suppresses the weak energy frequency band signal, which cannot fully play the role of all frequency band information of seismic in inversion, especially the application of high frequency weak signal, which affects the improvement of the identification and description ability of thin sand body by seismic and inversion.

[0003] In the Chinese patent application with the application number CN201811140059.5, a frequency division iteration seismic inversion method and system are involved, which includes: obtaining low, medium and high frequency seismic data based on original seismic data, and corresponding low, medium and high frequency seismic wavelets; establishing an inversion initial model; performing low frequency band inversion based on the inversion initial model, low frequency seismic data and low frequency seismic wavelet to obtain low frequency seismic wave impedance inversion result; updating the inversion initial model based on the low frequency seismic wave impedance inversion result and performing medium frequency band inversion to obtain medium frequency seismic wave impedance inversion result; updating the inversion model using the inversion result and performing high frequency band inversion to obtain final seismic wave impedance inversion result. The patent method needs to use the same method to perform multi-step iteration inversion on the frequency division data, which has many inversion steps, complex process, error accumulation and influence on the final inversion accuracy.

[0004] In the Chinese patent application with the application number CN201510559753.0, a seismic inversion method and device based on frequency division data are involved, which relates to the field of geophysical exploration technology, and it includes the following steps: obtaining seismic inversion initial model and different frequency seismic data volume based on collected seismic and well logging data in the region; obtaining inversion results of different frequency seismic data volume based on different frequency seismic data volume and seismic inversion initial model; obtaining quantitative reservoir prediction results of the region based on the inversion results of different frequency seismic data volume. The patent method also needs to perform multiple inversions, which also has excessive error accumulation, and uses formula fitting method to combine multiple frequency division inversion data into one inversion result. This fitting belongs to a simple mathematical algorithm, which lacks geophysical theoretical basis.

[0005] In the Chinese patent application with the application number: CN201811045066.7, a frequency division fusion wave impedance inversion method is involved. The method is to use initial seismic data and logging data of multiple wells to carry out conventional logging constrained model inversion to obtain logging constrained model inversion results; using the logging constrained model inversion results as the initial model, high-resolution inversion is carried out under the constraint of tuning frequency to obtain high-resolution inversion results; the logging constrained model inversion results and the high-resolution inversion results are combined in frequency segments to obtain frequency division fusion inversion results. The patent method is also to carry out multiple inversions, and the cumulative error affects the inversion accuracy, and the application of low frequency band of the earthquake is lacking, which is not conducive to the inversion of thick reservoirs.

[0006] Therefore, we have invented a new wideband elastic parameter inversion method based on seismic frequency division body, which solves the above technical problems. SUMMARY

[0007] The purpose of the present application is a wideband elastic parameter inversion method based on seismic frequency division body, which can balance the constraint of weak seismic frequency band signal, enhance the effect of weak energy band seismic signal, fully utilize the effective full frequency information of the earthquake, and improve the seismic inversion accuracy and reservoir description ability.

[0008] The purpose of the present application can be realized by the following technical measures: a wideband elastic parameter inversion method based on seismic frequency division body, which comprises the following steps: step 1, seismic effective frequency band analysis and time-frequency decomposition; step 2, extracting frequency division seismic matching seismic wavelet; step 3, carrying out joint reflection coefficient inversion of frequency division seismic data; step 4, fusing reflection coefficient integral impedance body and low frequency model body.

[0009] The purpose of the present application can also be realized by the following technical measures:

[0010] In step 1, by analyzing the seismic frequency band and signal-to-noise ratio, the seismic effective frequency band range is determined, then a high-precision time-frequency decomposition method is selected, and seismic time-frequency decomposition is carried out, and the seismic data body is divided and decomposed into low, medium and high frequency bands in order of frequency from high to low.

[0011] In step 1, wavelet transform, S transform, matching pursuit and other high-precision seismic time-frequency decomposition methods are selected to decompose the seismic data body into low, medium and high frequency bands.

[0012] In step 2, based on the frequency division seismic body, the statistical wavelet extraction method is used to obtain the frequency division seismic wavelet consistent with the frequency division seismic amplitude energy, frequency and phase.

[0013] In step 3, the frequency division seismic body and the matching wavelet are used to construct the inversion objective function:

[0014]

[0015] wherein, r is the formation reflectivity, which is the variable to be inverted; D low ,D mid ,D hig respectively represent low, medium and high frequency band seismic data obtained by seismic time-frequency decomposition; W low ,W mid ,W hig is a convolution matrix composed of matched wavelets extracted from low, medium and high frequency band seismic data respectively; β is a balanced constraint weight, which constrains the weight of different frequency band information in the objective function.

[0016] In step 3, let:

[0017]

[0018] wherein: G1=αW low , G2=W mid , G3=βW hig ;

[0019] D1=αD low ,D2=D mid , D3=βD hig

[0020] By matrix substitution, while increasing the L-P norm regularization constraint term to ensure the stability and accuracy of inversion, the above inversion objective function is reconstructed and simplified as:

[0021]

[0022] wherein, λ is a regularization parameter, p is 1 or 2, and the formation reflectivity can be obtained by using least square, gradient descent and other objective function optimization solving methods.

[0023] Step 4 includes:

[0024] Step 41, integral operation is performed on the obtained reflectivity to obtain a formation relative wave impedance body;

[0025] Step 42, the obtained relative impedance body is fused with a low frequency model by using a time-frequency fusion method, and finally a wide frequency wave impedance body is obtained.

[0026] In step 41, the integral formula is as follows:

[0027]

[0028] Where AI(0) and AI(t) represent the relative wave impedance of the first point and the point of time t respectively, and r is the formation reflection coefficient.

[0029] In step 42, the low-frequency model volume is constructed by using the logging interpolation under the tectonic constraint, or by using the seismic velocity in combination with the logging petrophysical relationship.

[0030] The wide-frequency elastic parameter inversion method based on the seismic frequency division volume has the following advantages: the method is directed to the characteristics of the amplitude energy strength of different frequency band information of the earthquake, the construction of the joint inversion objective function is used to balance the strong and weak signals, the constraint ability of the weak energy band signal on the inversion is improved, the role of all frequency band information of the earthquake is fully played, the reflection coefficient inversion precision is improved, and finally the integration volume and the low-frequency model volume are fused and supplemented, the frequency band of the inverted elastic parameter volume is further widened, and this is beneficial to the seismic and geological reservoir prediction and description. BRIEF DESCRIPTION OF DRAWINGS

[0031] Figure 1 A flowchart of a wide-frequency elastic parameter inversion method based on a seismic frequency division volume according to the present application;

[0032] Figure 2 A schematic diagram of an original seismic profile according to an embodiment of the present application;

[0033] Figure 3 A seismic spectrum analysis diagram according to an embodiment of the present application;

[0034] Figure 4 A schematic diagram of a seismic low-frequency band decomposition profile according to an embodiment of the present application;

[0035] Figure 5 A schematic diagram of a seismic medium-frequency band decomposition profile according to an embodiment of the present application;

[0036] Figure 6 A schematic diagram of a seismic high-frequency band decomposition profile according to an embodiment of the present application;

[0037] Figure 7 A schematic diagram of an inverted reflection coefficient profile according to an embodiment of the present application;

[0038] Figure 8 A schematic diagram of an inverted wide-frequency wave impedance profile according to an embodiment of the present application;

[0039] Figure 9 A schematic diagram of an inverted wave impedance profile according to an embodiment of the present application. DETAILED DESCRIPTION

[0040] It should be noted that the following detailed description is exemplary in nature and is intended to provide further description of the application. Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs.

[0041] It is to be understood that the singular forms "a," "an," and "the" include plural referents unless the context clearly dictates otherwise. It is also to be understood that the term "comprising" as used herein is intended to mean "including but not limited to."

[0042] As shown in Figure 1 , Figure 1 A flow chart of a wideband elastic parameter inversion method based on seismic frequency decomposition body of the application, the wideband elastic parameter inversion method based on seismic frequency decomposition body of the application includes the following steps:

[0043] Step 1, seismic effective frequency band analysis and time-frequency decomposition; the seismic effective frequency band analysis and time-frequency decomposition determine the effective seismic frequency band through seismic spectrum and signal-to-noise ratio analysis, and select wavelet transform, S transform, matching pursuit and other high-precision seismic time-frequency decomposition methods to decompose the seismic data into low, medium and high frequency band data.

[0044] Step 2, frequency-decomposed seismic matching seismic wavelet extraction; the frequency-decomposed seismic matching seismic wavelet extraction respectively extracts frequency-decomposed seismic wavelets consistent with the frequency-decomposed seismic amplitude energy, frequency and phase based on the frequency-decomposed seismic body using statistical wavelet extraction method.

[0045] Step 3, joint reflection coefficient inversion of frequency-decomposed seismic data;

[0046] The joint reflection coefficient inversion of frequency-decomposed seismic data constructs an inversion objective function:

[0047]

[0048] Wherein, r is the formation reflection coefficient, which is the variable to be inverted; D low ,D mid ,D hig Respectively represent the low, medium and high frequency band seismic data obtained by seismic time-frequency decomposition; W low ,W mid ,W hig Is the convolution matrix composed of the matching wavelets extracted by low, medium and high frequency band seismic data respectively; Beta is the balance constraint weight, which constrains the weight of different frequency band information in the objective function, and its size is obtained by comparing the energy of the frequency-decomposed body.

[0049] make:

[0050]

[0051] Where: G1=αW low G2 = W mid G3 = βW hig ;

[0052] D1=αD low D2=D mid D3 = βD hig

[0053] By using matrix substitution and adding LP norm regularization constraints to ensure inversion stability and accuracy, the above inversion objective function can be reconstructed and simplified as follows:

[0054]

[0055] Where λ is the regularization parameter, and p is usually 1 or 2. The formation reflection coefficient can be obtained by using objective function optimization methods such as least squares and gradient descent.

[0056] Step 4: Fusion of the reflection coefficient integral impedance volume with the low-frequency model volume. This fusion of the reflection coefficient integral impedance volume with the low-frequency model volume includes two steps.

[0057] Sub-step 41 involves integrating the reflection coefficients obtained from the inversion to obtain the relative wave impedance volume of the formation. The integration formula is as follows:

[0058]

[0059] Where AI(0) and AI(t) represent the relative wave impedance at the first point and at time t, respectively.

[0060] Sub-step 42 involves fusing the obtained relative impedance volume with the low-frequency model using a time-frequency fusion method. The low-frequency model can be constructed using well logging interpolation under structural constraints or by combining seismic velocity with well logging rock physics relationships.

[0061] In a specific embodiment 1 of the present invention, the broadband elastic parameter inversion method based on seismic frequency division bodies of the present invention includes the following steps:

[0062] Step 1: Effective seismic frequency band analysis and time-frequency decomposition; The effective seismic frequency band analysis and time-frequency decomposition are carried out by analyzing the seismic spectrum and signal-to-noise ratio to determine the effective seismic frequency band, and the wavelet scale decomposition method is used to decompose the seismic data into three frequency bands: low, medium and high.

[0063] Step 2, frequency division seismic matching seismic wavelet extraction; the frequency division seismic matching seismic wavelet extraction respectively utilizes low, middle and high three frequency band seismic bodies D low ,D mid ,D hig , respectively obtains seismic wavelets W corresponding to three frequency division seismic data by using a seismic statistical wavelet extraction method low ,W mid ,W hig .

[0064] Step 3, joint reflection coefficient inversion of frequency division seismic data

[0065] An inversion objective function is used:

[0066] Wherein, r is the stratum reflection coefficient, which is a variable to be inverted; D low ,D mid ,D hig respectively represent low, middle and high frequency band seismic data obtained by seismic time-frequency decomposition; matched wavelets W low ,W mid ,W hig are respectively extracted by using low, middle and high frequency band seismic data; convolution matrices are respectively constructed; the energy ratio of three frequency division bodies is used to determine the balanced constraint weight β, which constrains the weight of different frequency band information in the objective function.

[0067] Let:

[0068]

[0069] Wherein: G1=αW low , G2=W mid , G3=βW hig ;

[0070] D1=αD low ,D2=D mid , D3=βD hig

[0071] By matrix substitution, while increasing the L-P norm regularization constraint term to ensure the stability and accuracy of the inversion, the above inversion objective function is reconstructed, which can be simply represented as:

[0072]

[0073] Wherein, λ is a regularization parameter, in order to ensure the sparsity of the reflection coefficient inversion, p is set to 1, the objective function becomes L1 regularization least square problem, and the formation reflection coefficient can be obtained by using the Newton interior point method for objective function optimization solving method.

[0074] Step 4, the reflection coefficient integral impedance body is fused with the low-frequency model body. The reflection coefficient integral impedance body is fused with the low-frequency model body, including two steps.

[0075] Substep 41, the reflection coefficient obtained by inversion is subjected to integral operation to obtain a seismic wideband relative wave impedance body, and the integral formula is as follows:

[0076]

[0077] Wherein, AI(0) and AI(t) represent the first point and the relative wave impedance at the point t, respectively.

[0078] Substep 42, the seismic imaging velocity is converted into layer velocity, then the linear relationship between velocity and impedance is constructed by combining the logging petrophysical relationship, the low-frequency impedance model is constructed by using the relationship, and the relative impedance body is fused with the low-frequency impedance model body by using the time-frequency fusion method, so that the wideband elastic parameter body can be obtained.

[0079] In a specific embodiment 2 of the application, the wideband elastic parameter inversion method based on the seismic frequency division body of the application comprises the following steps:

[0080] Step 1, seismic effective band analysis and time-frequency decomposition; the seismic effective band analysis and time-frequency decomposition are used to determine the effective seismic frequency band by analyzing the seismic spectrum and signal-to-noise ratio, and the S transform time-frequency decomposition method is selected to decompose the seismic data into three frequency band data of 3-35 Hz, 35-110 and 110-150 Hz.

[0081] Step 2, frequency division seismic matching seismic wavelet extraction; the frequency division seismic matching seismic wavelet extraction is used to respectively extract the low, medium and high frequency band seismic data D low ,D mid ,D hig , and the seismic statistical wavelet extraction method is used to obtain the seismic wavelets W low ,W mid ,W hig corresponding to the three frequency division seismic data.

[0082] Step 3, joint reflection coefficient inversion of frequency division seismic data;

[0083] The inversion objective function is as follows:

[0084] wherein, r is formation reflectivity, which is a variable to be inverted; D low ,D mid ,D hig respectively represent low, medium and high frequency band seismic data obtained by seismic time-frequency decomposition; matched wavelets W low ,W mid ,W hig are respectively constructed.

[0085] Let:

[0086]

[0087] wherein: G1=αW low , G2=W mid , G3=βW hig ;

[0088] D1=αD low ,D2=D mid , D3=βD hig

[0089] By matrix substitution, while increasing L-P norm regularization constraint term to ensure the stability and accuracy of inversion, the above inversion objective function is re-constructed, which can be simplified as:

[0090]

[0091] wherein, λ is a regularization parameter, p is set to 2, the objective function becomes L2 regularization, and the layer reflectivity can be obtained by using the least square method.

[0092] Step 4, the reflectivity integral impedance body is fused with the low frequency model body. The reflectivity integral impedance body is fused with the low frequency model body, including two steps.

[0093] Sub-step 41, the reflectivity obtained by inversion is integrated to obtain a seismic wideband relative wave impedance body, and the integral formula is as follows:

[0094]

[0095] wherein AI(0) and AI(t) respectively represent the relative wave impedance of the first point and the point with time t.

[0096] Sub-step 42, the impedance low frequency information constructed by using logging velocity and density is used to establish a low frequency impedance model under the logging structure constraint, the obtained relative impedance body is fused with the low frequency impedance model body, and a wideband elastic parameter body can be obtained.

[0097] Figure 2The high frequency information of the original seismic profile in an embodiment of the present application is responsive to the thin layer drilled by the well, but the seismic axis energy is weak, the lateral is discontinuous, and it is difficult to track and describe.

[0098] Figure 3 The original seismic spectrum analysis diagram in an embodiment of the present application uses a wide seismic frequency band, the energy of the low frequency and high frequency bands is weak compared with the energy of the medium frequency band, and the low, medium and high frequency bands are divided according to the frequency band analysis and the energy intensity.

[0099] Figure 4 、 Figure 5 and Figure 6 are respectively the low frequency band, medium frequency band and high frequency band seismic profiles obtained by the seismic time-frequency decomposition in an embodiment of the present application, it can be seen that the energy of the low frequency and high frequency bands is weak and the signal-to-noise ratio is poor compared with the medium frequency band seismic; in the medium and low frequency band seismic, the thin layer drilled by the well has no response due to the lack of high frequency band signals, and the thick layer is reflected more clearly; in the high frequency band seismic, the thin layer has a response, but the thick layer is reflected poorly.

[0100] Figure 7 The inversion reflection coefficient profile in an embodiment of the present application is shown, the thin geological body is more clear, but the information of the stratigraphic interface is still not used for the description of the reservoir thickness.

[0101] Figure 8 The inversion wide frequency wave impedance profile in an embodiment of the present application is obtained by integrating the reflection coefficient and fusing the low frequency body, the wide frequency wave impedance body is obtained, which not only reflects the spatial variation characteristics of the thin layer but also reflects the spatial variation characteristics of the thick layer, and is more suitable for reservoir prediction and description.

[0102] Figure 9 The inversion wave impedance profile in an embodiment of the present application is not used, the inversion reflects more characteristics of the medium frequency band seismic information, and the description ability for the thin layer and the thick layer is poor compared with the wide frequency wave impedance body obtained by the present application.

[0103] Compared with the traditional seismic inversion method, the wide frequency elastic parameter inversion method based on the seismic frequency division body considers the energy difference characteristics of the high, medium and low frequency bands of the seismic signal, uses the time-frequency decomposition body, balances the constraint parameters by constructing a joint inversion objective function and optimizing the solution, weakens the suppression and influence of the strong energy signal of the medium frequency dominant band on the weak signal band, fully plays the role of the full frequency information of the seismic, and finally obtains the wide frequency inversion data body, and improves the geological reservoir prediction and description ability.

[0104] Finally, it should be noted that the above only describes the preferred embodiments of the present application and is not intended to limit the present application. Although the present application has been described in detail with reference to the foregoing embodiments, those skilled in the art will appreciate that modifications can be made to the technical solutions described in the foregoing embodiments, or some of the technical features thereof can be replaced equivalently, without departing from the spirit and principle of the present application. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principle of the present application should be included in the protection scope of the present application.

[0105] All that is not described in the specification is known to those skilled in the art.

Claims

1. A method for broadband elastic parameter inversion based on seismic frequency-volume, characterized in that, The wideband elastic parameter inversion method based on the seismic frequency division body comprises: Step 1, performing seismic effective frequency band analysis and time-frequency decomposition; Step 2, extracting frequency division seismic matching seismic wavelet; Step 3, performing joint reflection coefficient inversion of frequency division seismic data; Step 4, fusing the reflection coefficient integral impedance body with a low-frequency model body; In step 1, by analyzing the seismic frequency band and signal-to-noise ratio, the effective frequency band range of the seismic data is determined, then a high-precision time-frequency decomposition method is selected, and the seismic time-frequency decomposition is carried out, and the seismic data is sequentially divided and decomposed into low, medium and high frequency band seismic data bodies according to the high, medium and low frequency order in the effective frequency band range; In step 2, based on the frequency division seismic body, the statistical wavelet extraction method is used to obtain the frequency division seismic wavelet consistent with the frequency division seismic amplitude energy, frequency and phase, and an air-variable wavelet field is constructed; In step 3, the frequency division seismic body and the matching wavelet are used to construct an inversion objective function: wherein r is the formation reflection coefficient, and is the variable to be inverted; D low ,D mid ,D hig respectively represent low, medium and high frequency band seismic data obtained by seismic time-frequency decomposition; W low ,W mid ,W hig is a convolution matrix composed of matched wavelet fields extracted from low, medium and high frequency band seismic data respectively; β is a balanced constraint weight, which constrains the weight of different frequency band information in the objective function; In step 3, let: wherein: G2= W mid , G3= βW hig ; D2= D mid , D3= βD hig ; By matrix substitution and simultaneously adding an L-P norm regularization constraint term to ensure the stability and accuracy of the inversion, the above inversion objective function is reconstructed and simplified as: Wherein, λ is a regularization parameter, p is 1 or 2, and the least square gradient descent objective function optimization solving method is used to obtain the formation reflection coefficient; Step 4 comprises: Step 41, performing integral operation on the obtained reflection coefficient to obtain a formation relative wave impedance body; Step 42, using the frequency fusion and time-frequency conversion method, fusing the obtained relative impedance body with a low-frequency model and a high-frequency model in the frequency domain, and finally converting to the time domain to obtain a wideband wave impedance body.

2. The method of claim 1, wherein, In step 1, the high-precision seismic time-frequency decomposition methods such as wavelet transform, S transform and matching pursuit are selected to decompose the seismic data into low, medium and high frequency band seismic data bodies.

3. The method of claim 1, wherein, In step 41, the integral formula is as follows: Wherein, AI(0) and AI(t) represent the relative wave impedance of the first point and the point with time t respectively, and r is the formation reflection coefficient.

4. The method of claim 1, wherein, In step 42, the low-frequency model body is constructed by using the constrained logging interpolation or constructed by using the seismic velocity and the logging petrophysical relationship.

Images