CN106772587A - Seismic elastic parameter Facies Control Modeling method based on same position multiphase collocating kriging - Google Patents

Abstract

The invention discloses a phase-controlled modeling method of seismic elastic parameters based on co-located multi-phase collaborative kriging, which comprises the steps of: using an equal-scale grid subdivision method to subdivide the modeling grid in the target layer to form several planes , as the target position of the elastic parameter modeling; respectively extract the logging elastic parameters, seismic attribute parameters and sedimentary facies information as the modeling data, select the main variable, select the first and second covariate variables; variables and the second covariate are used as calculation parameters, interpolation calculation is carried out by using the co-location multiphase co-kriging method, and the parameter values of all points to be estimated on the plane are obtained as the multi-source data fusion modeling results; the anisotropic diffusion method is used for generalization After generalized processing, the seismic elastic parameter modeling results after generalized processing are obtained. The invention can eliminate calculation abnormal points and boundary noise points, improve the precision of multi-information parameter comprehensive modeling and incorporate more real geological information, and has good applicability.

Description

Phased Modeling Method of Seismic Elastic Parameters Based on Co-located Multiphase Co-Kriging

technical field

The invention relates to a phase-controlled modeling method of seismic elastic parameters based on co-located multi-phase collaborative kriging, which belongs to the technical field of parameter modeling in geophysics.

Background technique

Seismic elastic parameter modeling is a multi-information comprehensive analysis process. In complex geological environments, it is very difficult to maintain the stability and high precision of wave velocity modeling. Because the change of the entire geological system is affected by the natural environment, the buried depth of layers, and the thickness of soil and rock formations, and the complexity of geological structures will lead to great heterogeneity and uncertainty in the velocity model. Therefore, a modeling method that can integrate multiple geological conditions is needed to represent this heterogeneity and reduce uncertainty. To obtain fine seismic inversion or depth migration results, it also depends on the initial model constructed by seismic wave velocity modeling. Although the linear inversion method can use the logging information at the well location to construct appropriate boundary conditions and combine it with seismic information, the accuracy of the initial model will greatly affect the accuracy of the inversion. The seismic wave travel time velocity analysis method converges through continuous modification of the initial model, but it is not adaptable to strong lateral changes and is prone to large cumulative errors in the calculation process.

The geological conditions in offshore carbonate rock areas are complex, with strong heterogeneity, and the logging data as the main variable is relatively scarce. In order to improve the modeling accuracy of elastic parameters and reduce uncertainty, it is necessary to add effective Restrictions. The constraints that can be obtained in the study area are often seismic attribute parameters and sedimentary facies information, and these information can reflect the variation characteristics of the main variables to a certain extent, so it is necessary to select appropriate attribute parameters to constrain the modeling process. In addition to the primary variable, the elastic parameter modeling method based on co-kriging also uses a secondary variable as a constraint condition, which effectively expresses the prior information of geological conditions and reduces the uncertainty of modeling. Since the estimated value obtained by using the co-kriging method comprehensively expresses the parameter values of the surrounding samples, the weight coefficient of the sample points is not only related to the main variable space covariance function, but also controlled by a covariate constraint, so to a certain extent It improves the stability of modeling. Obviously, when the constraints continue to be added, the co-kriging method with multiple secondary variables can further improve the accuracy of modeling, but there are still many difficulties in the extraction and use of constraints.

Using conventional seismic attribute parameters to constrain different main variables, the modeling results obtained based on the co-kriging method often have no clear boundaries and a lot of noise, which cannot truly reflect the distribution of elastic parameters of the underground medium, especially in the medium In areas with poor lateral continuity of attributes, it is difficult to distinguish between real local disturbances and small errors caused by calculations, which will cause difficulties for subsequent inversion and related work.

Contents of the invention

The technical problem to be solved by the present invention is to overcome the deficiencies of the prior art, provide a phase-controlled modeling method of seismic elastic parameters based on co-kriging, and solve the problem that the modeling results obtained based on the collaborative kriging method are often It does not have a clear boundary and there are a lot of noise points, which cannot truly reflect the distribution of the elastic parameters of the underground medium. It can eliminate calculation abnormal points and boundary noise points, improve the accuracy of multi-information parameter comprehensive modeling and incorporate more real geology. The information has good applicability.

The present invention specifically adopts the following technical solutions to solve the above technical problems:

The phase-controlled modeling method of seismic elastic parameters based on co-located multi-phase co-kriging includes the following steps:

Step S1, subdividing the modeling grid by using an equal-scale grid subdivision method in the target layer, forming several planes at positions with the same ratio thickness, as target positions for elastic parameter modeling;

Step S2, respectively extract the logging elastic parameters of known points on the plane and the seismic attribute parameters and sedimentary facies information distributed in the whole area as the modeling data, that is, select the low-frequency parameter background value of the logging parameters as the main variable, and select the seismic Attribute parameters are used as the first covariate, and the sedimentary facies digital model is obtained by generalizing the sedimentary facies information by using the anisotropic diffusion method as the second covariate;

Step S3, using the main variable, the first covariate and the second covariate obtained in step S2 as calculation parameters, using the co-location multiphase co-kriging method to perform interpolation calculations, and obtaining the parameter values of all points to be estimated on the plane as elastic parameters Multi-source data fusion modeling results;

Step S4, using the anisotropic diffusion method to generalize the multi-source data fusion modeling result obtained in step S3, and obtain the generalized seismic elastic parameter modeling result.

Further, as a preferred technical solution of the present invention: in the step S2, the low-frequency parameter background value of the logging parameter is selected as the main variable, specifically including:

Use the polynomial fitting method to smooth the gas-permeable logging elastic parameters, and obtain several groups of smoothed low-frequency parameter background values according to different polynomial fitting times;

A group of background values of low-frequency parameters larger than the variation range of the preset low-frequency parameter background values are selected as the main variables.

Further, as a preferred technical solution of the present invention: in the step S2, the seismic attribute parameter is selected as the first collaborative variable according to the correlation coefficient between the main variable and the seismic attribute parameter.

Further, as a preferred technical solution of the present invention: in the step S2, the anisotropic diffusion method is used to generalize the sedimentary facies information, using the formula:

Among them, Z is the parameter value of the point to be processed, is the gradient value of adjacent points in each direction, and A, B, and C are weight coefficients in each direction.

Further, as a preferred technical solution of the present invention: in the step S3, the calculation is performed by interpolation using the co-located polyphase co-kriging method, which specifically includes:

Step S31, using the co-location collaborative multiphase kriging method to establish a parameter value model of the point to be estimated on the calculation plane;

Step S32, obtaining the kriging equations according to the unbiased optimal condition, and converting the kriging equations into a matrix;

Step S33, calculating the covariance and cross-covariance between the parameters of the points to be estimated at different positions, so as to obtain all variable values in the matrix obtained in step S32;

Step S34, obtain the weight coefficient of each variable by solving the matrix obtained in step S32, and substitute it into the parameter value model of the point to be estimated established in step S31 to calculate the parameter value of the point to be estimated.

Further, as a preferred technical solution of the present invention: the multi-source data fusion modeling result of elastic parameters obtained in the step S3 is:

Among them, the estimated parameter value of the position of the point to be estimated is Z ^* (u ₀ ), the logging elastic parameter of the main variable is Z(u _i ), u is the position of the parameter point, and the seismic attribute parameter as the first covariate is Y ₁ , the numerical model of sedimentary facies as the second covariate is Y ₂ , and the weight coefficients are α _i , β ₁ and β ₂ .

Further, as a preferred technical solution of the present invention: in the step S4, the generalization process using the anisotropic diffusion method specifically includes:

Step S41, setting the parameter change threshold of the point to be estimated, and removing the parameter value of the point to be estimated that is larger or smaller than the set parameter change threshold as an abnormal point;

Step S42 , using the anisotropic diffusion method to generalize the parameter values of the remaining points to be estimated, and obtaining the generalized seismic elastic parameter modeling results through repeated iterations.

The present invention adopts above-mentioned technical scheme, can produce following technical effect:

The present invention obtains a method for rationally optimizing parameters in the modeling process through the extraction of multiple variable parameter values and the elastic parameter fusion modeling of multiple parameters based on the co-kriging method, and realizes the use of sedimentary facies digital models as The phase-controlled modeling of collaborative variables improves the modeling accuracy of elastic parameters in the geophysical field, and uses the anisotropic diffusion method to remove calculation noise while retaining the local information contained in geological bodies. Finally, the phase-controlled modeling of multi-parameter fusion is realized, which improves the accuracy and completeness of elastic parameter modeling based on conventional data. It can effectively eliminate calculation abnormal points and boundary noise, improve the accuracy of multi-information parameter comprehensive modeling and incorporate more real geological information. modeling method.

Description of drawings

Fig. 1 is a flow chart of the phase-controlled modeling method of elastic parameters based on co-located multi-phase co-kriging of the present invention.

Fig. 2 is a schematic diagram of equal-scale grid division in the present invention.

Fig. 3 is a schematic diagram of original logging parameters and smoothed low-frequency parameter background values used in the example of the present invention.

Fig. 4 is a schematic diagram of a digital model of sedimentary facies used in examples of the present invention.

Fig. 5 is a schematic diagram of the results obtained by modeling the well logging shear wave velocity parameters using the co-located multiphase co-kriging-based elastic parameter phase-controlled modeling method used in the example of the present invention.

detailed description

Embodiments of the present invention will be described below in conjunction with the accompanying drawings.

As shown in Figure 1, the elastic parameter phase-controlled modeling method based on co-located multi-phase collaborative Kriging proposed by the present invention mainly includes modeling mesh division, multi-source information extraction, and interpolation calculation based on co-located multi-phase Kriging and generalization processing based on the anisotropic diffusion method, the details are as follows:

Step S1, use the equal-proportion grid subdivision method to subdivide the modeling grid in the target layer, as shown in Figure 2, form multiple planes at the position with the same ratio thickness, as the target position for elastic parameter modeling ; If there is no data point at the target position, use the nearest neighbor point in the vertical direction as the value of the target position. Wherein, there are several known points and points to be estimated on the plane, and the known points can directly obtain the logging elastic parameters.

Step S2, multi-source information extraction, respectively extract the logging elastic parameters of all known points on the plane and the seismic attribute parameters and sedimentary facies information distributed in the whole area as modeling data, that is, select the low-frequency parameter background value of the logging parameters as As the main variable, choose the seismic attribute parameter as the first covariate, and use the anisotropic diffusion method to generalize the sedimentary facies information to obtain the sedimentary facies digital model as the second covariate; analyze the correlation of the data volume and analyze the extracted data Preliminary processing is carried out to realize the multi-source data fusion modeling process and improve the modeling accuracy of geophysical elastic parameters. The process includes the following steps:

Step S21, extract the low-frequency background values of all logging elastic parameters at known points, try different fitting times and compare the modeling requirements with the original logging data to obtain the data frequency that can meet the modeling accuracy requirements, as shown in the figure 3.

It is specifically:

Step S211 , using a polynomial fitting method to perform smoothing on all logging elastic parameter data at known points, and obtaining multiple sets of smoothed low-frequency parameter background values according to different polynomial fitting times.

Step S212, draw the low-frequency parameter background value curve, and set the preset low-frequency parameter background value range to 40m. Since the frequency required by the modeling result is actually lower than the original data collected by logging, it is selected according to the characteristics of elastic parameter changes. A group of low-frequency background values that can reflect changes in the range of more than 40m are used as the main variables, so that the low-frequency background values can reflect large sections of rock formations and reflect a certain change trend in thin-bed areas in the depth domain data volume.

Step S22, extracting seismic attribute parameters, obtaining parameter types that can reflect changes in main variables laterally through statistical methods, calculating the correlation coefficient between different seismic attribute data volumes in the entire region and low-frequency background values of well logging, and selecting generally applicable seismic attributes according to the calculation results As the first covariate, the optimization of the modeling parameters is achieved. It includes the following steps:

Step S221, drawing curves at different well locations for each seismic attribute, counting the variation trends of seismic attributes at different well locations, macroscopically understanding the variation characteristics of seismic attribute parameters in different sedimentary facies belts, and comparing the seismic attribute characteristics at the well locations with the Comparative analysis of sedimentary facies information.

Step S222, count the correlation coefficients between the main variables obtained in step S21 and the seismic attributes at all known well locations, and combine the degree of agreement between the seismic attributes and sedimentary facies information in step S221, and select the seismic attribute parameters with larger correlation coefficients as first covariate.

Step S23, transforming the sedimentary facies information into a digital model of sedimentary facies by means of classification and labeling, so that the sedimentary facies is added as a control condition into the modeling calculation process, and real facies-controlled modeling is realized. The specific process is as follows:

Step S231, according to the description of sedimentary facies in the geological survey results, use numbers to represent changes in sedimentary facies, and the size of the numbers is related to the distance away from the land, specifically: 1, 2 represent intra-platform reef-shoal facies; 3 represent platform tidal flat facies; 4, 5 represents the platform margin slope facies, and the conventional text description is replaced by quantitative digital representation, so that the digital sedimentary facies information obtained in this way can directly constrain the modeling process; so that the sedimentary facies digital representation of each point on the plane can integrate the sedimentary facies information Included in the modeling calculation process.

Step S232, using the anisotropic diffusion method to generalize the obtained digitized sedimentary facies information to obtain a digital sedimentary facies model as a second collaborative variable;

The calculation formula of the anisotropic diffusion method is as follows:

Among them, Z is the parameter value of the point to be processed, is the gradient value of the adjacent points in each direction, A, B, and C are the weight coefficients in each direction, i, j, and k are the relative positions of the processed points, and m is the number of iterations of the points. The generalized sedimentary facies model can be obtained by solving formula (1).

Among them, by constantly modifying the position of the points to be processed, the entire digital sedimentary facies information data body is generalized, and finally the digital model of the sedimentary facies on the processed plane is obtained, as shown in Figure 4.

Step S3, using the main variable, the first covariate and the second covariate obtained in step S2 as calculation parameters, using the co-location multiphase co-kriging method to perform interpolation calculations, and obtaining the parameter values of all points to be estimated on the plane as elastic parameters Multi-source data fusion modeling results.

Among them, the process of obtaining parameter values of all points to be estimated on the plane is as follows:

Step S31, using the main variable, the first covariate and the second covariate described in step S2 as calculation parameters, perform interpolation calculation according to the co-location collaborative multiphase Kriging method, and establish a parameter value model of the point to be estimated on the calculation plane:

Among them, the estimated parameter value of the position to be estimated is Z ^* (u ₀ ), the main logging variable parameter is Z(u _i ), u is the parameter point position, and the seismic attribute parameter extracted as the first covariate is Y ₁ , the numerical model of sedimentary facies as the second covariate is Y ₂ , and the weight coefficients are α _i , β ₁ and β ₂ .

Step S32, according to the unbiased optimal condition, by constructing the Lagrangian equation to minimize the estimated Kriging variance, the Kriging equation system can be obtained:

Where C represents the covariance or cross-covariance between parameters. In order to express the relationship between parameters more clearly, the equation system is converted into a matrix form:

Among them, M represents the covariance matrix, Λ represents the weight coefficient vector of the main variable, ε represents the error parameter, and the expression of M is:

Step S33, calculating the covariance and cross-covariance among the logging parameters, seismic attributes and sedimentary facies of the points to be estimated at different positions, and all variable values in the matrix M can be obtained;

Step S34 , by solving the formula (4), the weight coefficient of each variable can be obtained, and substituting it into the formula (2) to calculate the parameter value of the point to be estimated.

The above steps S31 to S34 can calculate a parameter value of a point to be estimated on the plane. In order to obtain the parameter values of all points to be estimated on the whole plane, the position of the point to be estimated is modified, and the parameter value calculated in step S34 is regarded as already Know the point, repeat step S3, calculate the parameter value of the next point to be estimated, and end when all the points on the plane are calculated.

Step S4, using the anisotropic diffusion method to generalize the multi-source data fusion modeling result obtained in step 3, and obtain the generalized seismic elastic parameter modeling result. Specifically include:

Step S41, set the parameter change threshold of the point to be estimated, and regard the parameter value of the point to be estimated that is greater or smaller than the set parameter change threshold as an abnormal point and remove it, that is, the statistical parameter change range will be within the negative value and the order of magnitude beyond the maximum value The parameter values of are regarded as outliers, and then the outliers in the multi-source data fusion modeling results in step S3 are removed.

Step S42, use the anisotropic diffusion method to generalize the multivariate data fusion modeling results to remove abnormal points, that is, to generalize the parameter values of the remaining points to be estimated, and to obtain generalized processing through repeated iterations of the formula (1) The final seismic elastic parameter modeling results are shown in Fig. 5. The obtained results retain the complete parameter change boundaries and local changes, and largely eliminate the noise that interferes with the modeling clarity. It can be seen from the figure that By adopting the method of the invention, more reasonable elastic parameter modeling results can be obtained.

Therefore, the method of the present invention realizes phase-controlled modeling using the digital model of sedimentary facies as a collaborative variable, improves the modeling accuracy of elastic parameters in the field of geophysics, and uses anisotropic diffusion method to remove calculation noise while retaining the properties of geological bodies. The local information included can eliminate calculation abnormal points and boundary noise points, improve the accuracy of multi-information parameter comprehensive modeling and incorporate more real geological information, which has good applicability.

The embodiments of the present invention have been described in detail above in conjunction with the accompanying drawings, but the present invention is not limited to the above embodiments, and can also be made without departing from the gist of the present invention within the scope of knowledge possessed by those of ordinary skill in the art. Variations.

Claims (7)

1. based on the seismic elastic parameter Facies Control Modeling method with position multiphase collocating kriging, it is characterised in that including following step Suddenly：

Step S1, in destination layer using equal proportion grid cutting algorithm to modeling grid carry out subdivision, in same ratio thickness Position on form several planes, as elastic parameter model target location；

Step S2, respectively extract plane on known point well logging elastic parameter and be distributed in whole region Seismic Attribute Parameters and Deposition phase information selects the low-frequency parameter background value of log parameter as master variable as modeling data, selects earthquake Property parameters are used as the first collaboration variable, and deposition phase information is carried out generalizing processing and deposited using anisotropic diffusion Phase mathematical model is used as the second collaboration variable；

Step S3, using step S2 gained master variable, first collaboration variable and second collaboration variable as calculating parameter, using same position Multiphase collocating kriging method carries out interpolation calculation, obtains the parameter value of all points to be estimated in plane as many of elastic parameter Source data integration modeling result；

Step S4, using anisotropic diffusion to step S3 gained multisource data fusion modeling result carry out generalizing processing, obtain Seismic elastic parameter modeling result after to generalizing processing.

2. the seismic elastic parameter Facies Control Modeling method based on same position multiphase collocating kriging according to claim 1, its It is characterised by, selects the low-frequency parameter background value of log parameter as master variable in the step S2, specifically includes：

The well logging elastic parameter extracted is smoothed using polynomial fitting method, according to different fitting of a polynomials Number obtain some groups it is smooth after low-frequency parameter background value；

Selection is more than one group of low-frequency parameter background value of default low-frequency parameter background value excursion as master variable.

3. the seismic elastic parameter Facies Control Modeling method based on same position multiphase collocating kriging according to claim 1, its It is characterised by, in the step S2, is made according to the coefficient correlation selection Seismic Attribute Parameters between master variable and Seismic Attribute Parameters It is the first collaboration variable.

4. the seismic elastic parameter Facies Control Modeling method based on same position multiphase collocating kriging according to claim 1, its It is characterised by, generalizing processing is carried out to deposition phase information using anisotropic diffusion in the step S2, using formula：

$Z_{i,j}^{(m+1)}={\mathrm{AZ}}_{i,j}^{\left(m\right)}+\frac{B}{8}\underset{k,l=-1}{\overset{1}{\Σ}}{C}_{i+k,j+l}\▿Z_{i+k,j+l}$

Wherein, Z be pending parameter value, ▽ Z be all directions point of proximity Grad, A, B, C for all directions weight coefficient, i, J, k are the relative position of handled point, and m is the iterations of point.

5. the seismic elastic parameter Facies Control Modeling method based on same position multiphase collocating kriging according to claim 1, its It is characterised by, is calculated using same position multiphase collocating kriging method interpolation in the step S3, specifically includes：

Step S31, using radioactive tritium multiphase kriging method, set up the parameter value model of point to be estimated on Calculation Plane；

Step S32, obtained gram according to unbiased optimal conditions in golden equation group, and will gram in golden equation group be converted to matrix；

Covariance and cross covariance between step S33, each parameter of calculating diverse location point to be estimated, to obtain step S32 institutes Obtain all variate-values in matrix；

Step S34, the matrix as obtained by solution procedure S32 obtain the weight coefficient of each variable, are substituted into step S31 and are set up The parameter value model of point to be estimated is calculated the parameter value of point to be estimated.

6. the seismic elastic parameter Facies Control Modeling method based on same position multiphase collocating kriging according to claim 1, its It is characterised by, the multisource data fusion modeling result that elastic parameter is obtained in the step S3 is：

$Z^{*}\left(u_0\right)=\underset{i=1}{\overset{n}{\Σ}}{\mathrm{\α}}_{i}Z\left(u_i\right)+{\mathrm{\β}}_1{Y}_1\left(u_0\right)+{\mathrm{\β}}_2{Y}_2\left(u_0\right)$

Wherein, the to be estimated estimates of parameters of position is Z^*(u₀), the well logging elastic parameter of master variable is Z (u_i), u is parameter Point position, is Y as the Seismic Attribute Parameters of the first collaboration variable₁, as second collaboration variable sedimentary facies mathematical model be Y₂, weight coefficient is respectively α_i、β₁And β₂。

7. the seismic elastic parameter Facies Control Modeling method based on same position multiphase collocating kriging according to claim 1, its It is characterised by, in the step S4, is specifically included using anisotropic diffusion generalizing processing：

Step S41, the Parameters variation threshold value for setting point to be estimated, will be greater than or less than the point to be estimated of setup parameter change threshold Parameter value is considered as abnormity point removal；

Step S42, the parameter value using the remaining point to be estimated of anisotropic diffusion generalizing processing, obtain general by iterating Seismic elastic parameter modeling result after change treatment.