RU2664503C1 - Method for forming cube or section of sites, method of automatic horizons/hodographs tracking and method for automatic detection of tectonic deformation zones and fracture zones - Google Patents

Abstract

FIELD: data processing; geophysics.SUBSTANCE: invention relates to the processing of seismic data in the field of geophysics and can be used in seismic operations. Three methods are proposed, connected by a single inventive concept. Method of forming a cube or a section of sites on the basis of an array of initial seismic data includes the estimation of the parameters of seismic waves along each path of both kinematic and dynamic parameters of the waves, wherein it includes the following operations for each seismic recording trace: directional summation of at least two nearest traces in a predetermined range of horizon/hodograph slope angles to obtain a three-dimensional or two-dimensional scan result (cube or scan section), search for local extremes in scan results, clarifying the position of each extremum, the definition of "side" results and the formation of a cube or a section of sites according to the location of the extremums found. Method for automatically tracing horizons/hodographs along a section or cube of sites, including those obtained by the claimed method of forming a cube or a section of sites, in which the horizon is formed from a set of areas that are piecewise linear approximation of a surface that is smooth or has a finite number of discontinuities along which a continuous smooth curve can be drawn, and as a criterion for combining at least two sites in the horizon, the conditions of mutual consistency of the sites are used. Method of automatic detection of zones of tectonic deformations and zones of fracturing along a cube or a section of sites, including those obtained by the claimed method of forming a cube or a section of sites, consists in detecting portions of an array of initial seismic data in which the regularity condition of the waves is violated, in this case, as a criterion for violating the regularity condition of waves, a measure of the misalignment of a plurality of sites is used.EFFECT: increased accuracy, reliability and automation of the processes of forming a cube or section of sites, tracing horizons/hodographs and identifying zones of tectonic deformations and fracture zones.19 cl, 18 dwg

Description

The invention relates to the processing of seismic data in the field of geophysics and can be used in seismic surveys. The methods described in the invention have much in common with data processing methods in other fields of science and technology, which analyze and convert two or three-dimensional data arrays (bitmap images, video, scalar fields, etc.).

The methods of processing and interpretation of seismic data are based on mathematical models of wave fields recorded by land, sea or borehole seismic methods (common depth point method - MOGT, vertical seismic profiling - VSP, method of refracted waves - MPV, continuous seismic profiling method - LSP and t .P.).

In the terminology of seismic wave fields, there is a fundamental concept of a regular wave - a wave that "is characterized by a fairly high stability of kinematic and dynamic parameters within the observation area" [1]. In parallel with the concept of wave regularity, the concept of spatial coherence of oscillations is used - “if the phase difference, as well as the ratio of the amplitudes of the oscillations, remain unchanged (or almost unchanged), then the oscillations will be coherent (or almost coherent)” [2].

One of the most famous and integral stages of processing and interpretation of seismic data, based on the concepts of regularity and coherence, are the tasks of tracking seismic waves and identifying zones of tectonic deformations and fracture zones. Well-known literary sources consider these problems separately, and existing algorithms solve them independently. The present invention shows methods combining the solution of these two problems in a single approach.

The term "wave tracking", "wave correlation" or "horizon piking" is often understood differently, we will mean by this concept the task of reconstructing the wave equation T (ρ) or another equation that describes the time of arrival of any wave characteristic (envelope maximum , the signal maximum, the moment of entry, etc.) at the observation point ρ [3]. As applied to the processing of seismograms, the arrival time of a wave is called the wave travel time equation (or simply a travel time curve). When applied to the interpretation of temporary seismic cubes or sections, the term horizon is used because connect the target wave with the interface of specific geological strata. The main difference between hodographs and horizons is the assumption that there are intersections of hodographs of various types of waves (longitudinal, transverse, incident, reflected, etc.), while the horizons, although they may have gaps, should be ordered in time / depth (t .e. one horizon always lies strictly above or below the other) and, in some cases, may have a finite number of points of contact. In the general case, horizons / hodographs are described by nonlinear and discontinuous functions, but in most cases they can be approximated by piecewise linear functions. In the case of 3D data, a linear approximating equation in the region [x ± Z ₀ , y ± Z ₀ ], can be represented as

и

- производные в этой точке (тангенсы углов наклона касательных) по соответствующим направлениям (Фиг. 1). В предлагаемом изобретении, участок горизонта/годографа, который может быть линейно аппроксимирован, назван площадкой горизонта/годографа.where t ₀ = T (x, y) is the time horizon / hodograph at a point with coordinates (x ₀ , y ₀ ),

and

- derivatives at this point (tangents of the angles of inclination of the tangents) in the corresponding directions (Fig. 1). In the present invention, a plot of the horizon / hodograph, which can be linearly approximated, is called the site of the horizon / hodograph.

The term “site” is sometimes used to denote a portion of a flat inclined boundary when deriving the analytical equations of the travel time curves of reflected waves (reflecting area), refracted waves (refractive area), etc. [one]. The reflecting or diffracting area is mentioned in the interpretation of the data of controlled directional reception (RNP), for the construction of which the ray and time diagrams are used [4].

Under the general task of tracking waves, it is necessary to understand the solution of a complex of problems: detection of a wave, its identification and estimation of parameters [3]. A fairly large number of domestic works has been devoted to solving these problems for processing 2D data [5, 6]. Tracking tasks in the interpretation of 3D data have appeared relatively recently and are mainly covered in foreign sources [7, 8].

In most practical cases, the analyzed wave field is a complex interference picture. As a rule, in this situation, simple methods for tracking the extremes of the wave pattern (including visual ones) give results that are far from true, because wave interference can form false maxima and minima up to the complete mutual cancellation of signals. Therefore, when solving such problems, the question arises of the correctness of their solution, i.e. existence, uniqueness and stability. The question of existence is usually not posed, because the analyzed wave fields are associated with real geological objects. Uniqueness of the solution can be achieved only when the methods used have obviously sufficient resolution, and the results of the tracking can be confirmed by a priori or a posteriori data (according to well logging data). The presence of random noise that interferes with regular waves of various origins, zones of vertical deformations or wedging out leads to instability of the solution.

SeisWorks, IHS Kingdom, GridPoint Dynamics Geoplat Pro-S) прослеживание волн осуществляется либо визуально, т.е. вручную, либо с применением автоматических корреляционных алгоритмов, либо эти подходы используются совместно. Во всех этих случаях надежность прослеживания волн зависит от наличия ярко выраженных регулярных волн, в противном случае автоматические алгоритмы не работают, а качество ручного прослеживания определяется опытом и интуицией интерпретатора, его способностью к критическому анализу разнообразных ситуаций, его представлению о геологической структуре исследуемого объекта.In modern software systems for the interpretation of seismic data (Shiumberger Petrel,

SeisWorks, IHS Kingdom, GridPoint Dynamics Geoplat Pro-S) wave tracking is carried out either visually, i.e. manually, or using automatic correlation algorithms, or these approaches are used together. In all these cases, the reliability of wave tracking depends on the presence of pronounced regular waves, otherwise automatic algorithms do not work, and the quality of manual tracking is determined by the experience and intuition of the interpreter, his ability to critically analyze various situations, his idea of the geological structure of the object under study.

A known method of constructing continuous seismic stratigraphic models of sections / cubes, patent RU 2516590, publ. 05/20/2014 [9].

и

а затем вычисляют функции перехода для временной или глубинной области всех соседних сейсмических трасс. Данные функции перехода позволяют создать единую непрерывную двумерную для разреза и трехмерную для куба сейсмостратиграфическую модель. Технический результат - повышение точности и достоверности данных картирования горизонтов и восстановления параметров геологической среды.According to the known method, pairwise continuous matching of multiple tracks of a seismic section or cube is carried out. For each pair of seismic traces of the section or each pair of traces of all in-lines and all cross-lines, a two-dimensional function of the mutual difference of fragments of seismic trace records is calculated. The values of this function are calculated for all possible combinations of time (depth) of a given pair of traces, which may belong to one seismostratigraphic level. The functions of the optimal correspondence of time / depth from the i-th to i + k-th seismic track are calculated

and

and then calculate the transition functions for the time or depth region of all adjacent seismic traces. These transition functions allow you to create a single continuous two-dimensional for the section and three-dimensional for the cube seismic stratigraphic model. The technical result is to increase the accuracy and reliability of the data of the mapping of horizons and restoration of the parameters of the geological environment.

The disadvantages of the known invention include the lack of stages for resolving seismic waves in it, which under the conditions of interference casts doubt on the reliability of the results of tracking and suggests the impossibility of using CDP and VSP methods in processing seismograms.

A known method for automatically separating horizons from a three-dimensional array of seismic data signals, a method for automatically selecting horizons of subsurface formations from a three-dimensional array of seismic data signals and a computerized method for automatically extracting horizons of layers from a three-dimensional array of seismic data signals, patent RU 2107931, publ. 03/27/1998 [10], which is the closest analogue (prototype) to the method for automatic tracking of horizons / hodographs presented in the present invention. It describes a method consisting of two phases: batch and interactive. In the first batch phase, a bitmap is obtained from the original cube, “units” of which indicate the possible presence of the horizon, and “zeros” - the absence. At the same time, the presence of the horizon on each route is determined by a specific characteristic of this route, confirmed by the presence of the same characteristic on the next four routes within a narrow time window. In the second interactive phase, starting from the user-specified starting point of the three-dimensional array, the nearest “units” are searched in the bitmap.

The disadvantages of this algorithm include:

- Tracking is based on criteria for assessing the position of the horizon from the initial interference wave pattern, which significantly reduces the efficiency of the algorithm in difficult interference conditions.

- Horizons are traced separately and independently, i.e. excluding possible mutual location options.

- Significant incorrectness of the tracking results in zones of tectonic deformations, pinch-outs and a low signal-to-noise ratio.

- Designed only for the interpretation of three-dimensional seismic cubes, and cannot be adapted for processing field seismograms.

A known method of processing a seismic signal and field exploration, patent RU 2144683, publ. 01/20/2000 [11], which is the closest analogue (prototype) to the method of automatic detection of tectonic deformation zones and fracture zones presented in the present invention. It describes a method based on calculating the cross-correlation function of nearby seismic cube traces and determining the “fall” vector from them as an estimate of coherence. In the technical solution subsequent to this patent, described in patent RU 2169931, publ. 06/27/2001 [12] of the same applicant, without changing the principle of invention of patent RU 2144683, similarity coefficients were used instead of the cross-correlation function. In literature, methods are given that use the eigenvalues of the correlation matrix, local structural entropy, and amplitude prediction error [13]. These algorithms are widely used and are used in the prediction of tectonic deformation zones. It is noted that these methods are most effective when using the results of tracking (taking into account the inclination of the horizons). However, in this case, the quality of wave tracking has a great influence on the assessment of coherence: interpreter errors make a significant contribution to the coherence picture.

The present invention describes the principle of calculating coherence, in which the angle of inclination of the horizons is not used as an additional, but the main parameter. As a result, the process of calculating coherence does not require the involvement of tracking results, thus, does not depend on the “subjective” analysis of the interpreter, and can be carried out over the entire volume of the source data.

The technical problem of the invention is the development of three independent technical solutions related by a single inventive concept, namely the development of a method for forming a cube or section of sites based on an array of source seismic data. And then, based on the data of the cube or section of the sites, including those obtained by the proposed method of forming the cube or section of the sites, the following methods were developed: a method for automatically tracking horizons / hodographs and a method for automatically detecting zones of tectonic deformations and fracture zones.

The technical result of the invention is to increase the accuracy, reliability and automation of the processes (or methods) of forming a cube or section of sites, tracking horizons / hodographs and identifying zones of tectonic deformations and zones of fracture.

An additional technical result for the method of forming a cube or section of sites and the method for automatically selecting horizons / hodographs are the implementation (implementation) of these methods in the conditions of interference of waves.

An additional technical result for the method for automatically detecting zones of tectonic deformations and fracturing zones is the independence of the calculation process from the “subjective” analysis of the interpreter.

The specified technical result is achieved by the fact that the method of forming a cube or section of the sites based on an array of source seismic data includes estimating the parameters of seismic waves along each path, both kinematic, such as the parameters of the sites (x0, y0, t0, α, β), and dynamic parameters waves (for example, form S (t), amplitude A, energy E, phase ϕ), and it includes the following operations for each trace of a seismic record: directional summation of at least two nearest tracks in a given range of horizontal angles ont / hodograph to obtain a three-dimensional or two-dimensional scan result (cube or scan section), search for local extrema in the scan results, clarify the position of each extremum, determine the side effect of summation and form a cube or section of sites according to the position of the found extrema.

With this method, directional summation of at least two traces located near the estimated point (x ₀ , y ₀ ) is carried out:

where Y _in (t, x, y) is the initial seismic data, w (n, m) are the weighting coefficients of summation, Δx and Δy are the distances between the nearest traces along the coordinates, R (t, p, q) is the scan result (cube or scan section).

, и обратное преобразование Фурье спектра суммы.With this method, directional summation of at least two traces located near the estimated point (x ₀ , y ₀ ) is carried out through the one-dimensional Fourier transform of the traces, summing the spectra of traces with weights

, and the inverse Fourier transform of the spectrum of the sum.

With this method, the refinement of the position of each extremum of the scan result is carried out as

where R _{i, j, k} = R (t _i , p _j , q _k ).

With this method, the determination of sites formed due to the side effect of summation is carried out under the condition:

where (p _s , q _s ) are the parameters of the wave sites, ƒ ₀ is the carrier frequency of the wave, (p, q) is the parameters of the tested sites, N × M is the number of traces involved in the summation, Δx and Δy are the distances between the nearest paths along coordinates.

In this method, depth cubes or sections are used as the source data array and directional summation is carried out in a given range of slope angles of the reflecting boundaries.

The specified technical result is also achieved by the fact that by the method of automatically tracking horizons / hodographs by section or cube of sites, including those obtained by the claimed method of forming a cube or section of sites, the horizon is formed from a plurality of sites that are piecewise linear approximation of the surface, smooth or having a finite number of discontinuities along which a continuous smooth curve can be drawn, and mutually conditions are used as a criterion for combining at least two sites into the horizon th site consistency.

With this method, subject to the mutual agreement of the sites, the sum of the absolute errors of the “prediction” is used as a measure of the mismatch between the two sites:

In this method, under the condition of mutual agreement between the sites, the average deviation is used as a measure of the mismatch between the two sites: the average height of the figure bounded by the top and bottom of the sites in the local area Ω

With this method, under the condition of mutual agreement between the sites, the equivalent of the "intensity" of errors in the local area Ω is used as a measure of the mismatch between the two sites

In this method, under the condition of mutual agreement between the sites, the mean square error in the local region Ω is used as a measure of the mismatch between the two sites

The specified technical result is also achieved by the fact that the method for automatically detecting zones of tectonic deformations and fracture zones by cube or section of sites, including those obtained by the claimed method of forming a cube or section of sites, consists in detecting sections of the source seismic data array in which the regularity condition is violated waves, while as a criterion for violating the conditions of regularity of waves, a measure of the mismatch of many sites is used.

With this method, the measure of mismatch of multiple sites is defined as the variance of the angles in a sliding window, limited in space and time:

where N is the number of sites that fell into the window [x ± Z ₀ , y ± Z ₀ , t ± T ₀ ], (p _i , q _i ) are the parameters of these sites, w _i are the summation weights,

With this method, as a measure of the mismatch of multiple sites using the measure of "maximum minimum mismatch" (MMR).

With this method, as a measure of the mismatch of multiple sites using the measure of "the sum of the minimum mismatches" (SMR).

With this method, the sum of the absolute errors of the “prediction” according to the formula (5) is used as a measure of the mismatch between the two sites.

In this method, the average deviation is used as a measure of the mismatch between the two sites: the average height of the figure bounded above and below the sites in the local area Ω according to formula (6).

In this method, the equivalent of the "intensity" of errors in the local domain Ω is used as a measure of the mismatch between the two sites according to formula (7).

In this method, the mean square error in the local area Ω is used as a measure of the mismatch between the two sites by the formula (8).

Disclosure of invention

In accordance with the present invention, at least three methods are proposed that allow solving the above problems in two stages (Fig. 3). At the first stage, the parameters of seismic waves are estimated from the initial data array. Assessment is made on each path of both kinematic, such as the parameters of the sites (x ₀ , y ₀ , t ₀ , α, β), and the dynamic parameters of the waves (for example, shape S (t), amplitude A, energy E, phase ϕ and etc.). The set of all obtained estimates (called a cube of sites for 3D data, a section of sites for 2D data) forms a space of signs of horizon / hodograph detection and identification of tectonic deformation zones and fracture zones (zones of reduced coherence) (Fig. 2). Obtained by the proposed method of forming a cube or section of the sites, the parameters of the sites can be saved and used in the future.

At the second stage, using only the data of the cube or section of the sites, including those obtained by the proposed method, the tasks of tracking horizons / hodographs and identifying zones of tectonic deformations and zones of fracture (zones of reduced coherence) are solved.

To estimate the parameters of the reflecting sites for each seismic field path, it is proposed to apply the scheme of FIG. 4, based on scanning (tangents) of the tilt angles (p, q) and directional summation of traces located near the estimated point (x ₀ , y ₀ ) according to formula (2). The position of local extremes R (t, p, q) is taken as an estimate of the time and tangents of the slope angles of the sites:

Directional summation is one of the most common multichannel methods used in interference to solve problems of estimating wave parameters [3, 14]. Summation was developed in the method of controlled directional reception in [4] and others. The basic principle of this method is laid in the idea of the best reception of a plane wave - when the summation line coincides with the travel time curve of the wave. Considering that the wave tilts are often not a priori known, the summation is performed with different time shifts in the range of possible tilt values of the travel time curves of useful waves. This summation is called adjustable. At the moment, a large number of algorithms for the analysis and interpretation of multichannel seismic data have been built, based on the results of the RNP (summograms, summands). In combination with high-pass filtering, RNP has become an effective tool for processing complex wave fields. By analyzing the so-called amplitude growth regions on the RN summot, algorithms and programs for solving the problems of detection, extraction and estimation of plane wave parameters are developed.

Since the method was developed between 1937 and 1954, its implementation was possible only in an analog way. Why special directional summation devices based on optical trace summation were developed. Significant disadvantages of this technique include the hardware dependence of the resolution of the method, the discreteness of the step of the trace shift in time, and, like any specialized technique, it is not universal. The subsequent modifications of the RNP method that appeared mainly had the character of adapting the method to the specifics of a specific task. So in the works of A.B. Kogan, V.N. Deca, L.D. Korning and V.N. Strakhova [15] presented a method of directional summation with simultaneous frequency filtering (NSCF) to distinguish "anomaly-forming objects elongated linearly along some fixed coordinate" (data of magnetic prospecting, gravity exploration, etc.) and automatic construction of correlation schemes according to logging data. Like RNP, the NSCF was initially implemented in an analogue way, but in their recent works these authors cite the results of digital processing by the NSCF method, without describing the methods for implementing digital directional summation themselves, without analyzing the effectiveness of this approach.

Most algorithms of interference systems impose the condition of an integer value of the directivity coefficient of the system (ie, the quantity pΔx / Δt must be an integer) [14]. Otherwise, it is necessary to apply interpolation calculations, leading to a significant increase in the complexity of the algorithms and requiring a reasonable choice of interpolation parameters. As noted in [16], the implementation of the RNP method in the frequency domain faces similar problems: "... it seems that it is easy to get Y _in (ƒ _t, -pƒ _t ) from Y _in ( _{x ,} ƒ _x ), but in practice this the most difficult part of the process. The straight line ƒ _x = qƒ _t does not pass immediately through all the grid nodes (except for the case p = Δt / Δх). Therefore, it is necessary to perform some amount of interpolation calculations. "

If we use direct trace summation to obtain the scan result, taking into account time discreteness and, in general, not the offset factor pnΔx + qmΔy of the time step Δt, then each trace will need to be interpolated for each direction of summation (Fig. 5). However, the above problems can be solved using the approach described in [17, 18]. According to this approach, the directional summation procedure can be organized through the one-dimensional Fourier transform of traces, summing the spectra of traces with weights

and the inverse Fourier transform of the spectrum of the sum (Fig. 6). Obvious advantages of such an implementation scheme:

- for all summation results, the first stage (Fourier transform) is general and independent of the summation parameters, and, therefore, can be performed once for all scan cubes in the formation of which this trace is involved;

- Fourier transform implements the best interpolation of seismic signals - Kotelnikov interpolation;

- when summing the spectra of traces, it is possible to summarize not the entire frequency range (up to the Nyquist frequency), but only the frequency range of the signal (usually half the Nyquist frequency), which will reduce the operating time of the algorithm and increase the resolution of the algorithm;

- to accelerate instead of PF, you can use fast PF algorithms (Cooley-Tukey, Vinograd, etc.) algorithms.

- In case it is required to trace such a feature of the waveform as the transition through zero, it’s enough to add the multiplication of the sum spectrum by the imaginary unit and frequency to the spectral summation procedure. Now the extremes of the scan results will correspond to these features

The dynamic parameters of the waves can be estimated by known methods from the initial wave field, however, to increase the resolution, it is preferable to do this according to the summation results. These parameters include shape, amplitude, predominant frequency, energy, phase, etc. Mandatory parameters for the tracking task include the indicator parameter defining this feature (maximum, minimum, transition through zero from minus to plus or transition through zero from plus to minus, etc.).

Clarification of the position of each extremum can be carried out with an arbitrarily small error, using methods of multidimensional unconditional optimization: the gradient method, the steepest descent method, the conjugate direction method, etc. However, such methods require additional directional summation calculations, which will adversely affect the speed of the algorithm. A rough estimate of the position of the extrema, but with an error much less discrete than the scan result, is proposed to be obtained by linear approximation of the scan results, for which the refined estimate (10) will take the form according to formula (3).

The local extrema of the results of directional summation, as shown by studies, do not always correspond to the coefficients of the equation of the horizon / hodograph of a regular wave. False extremes are associated with a violation of the regularity condition of the wave or with the side effect of summation. To distinguish the second situation, it is necessary to know the parameters of the wave sites (p _c , q _c ) and the frequency composition of the waves (carrier frequency - ƒ ₀ ). The parameters (p, q) of the sites formed due to the side effect of summation will satisfy the condition by formula (4). As (p _c , q _c ) it is possible to choose the parameters of the sites with the amplitude of the scan result prevailing in absolute value in a moving time window. Another approach to reducing the side effect of summation is, as is known in the theory of digital signal processing, to use the so-called w (n, m) weighting coefficients of summation. smoothing windows: a triangular window, a Hann window, a Hamming window, etc.

Based on the obtained estimates of the wave parameters (cube or section of the sites), the method for automatically tracking horizons / hodographs is to detect many sites that are piecewise-linear approximations of the surface, smooth or having a finite number of discontinuities, but along which a smooth curve can be drawn. To solve in such a formulation, a criterion for combining at least two sites with the equations τ ₁ (x, y) and τ ₂ (x, y) into one horizon is necessary.

As such a criterion, the condition for limiting some measure of mismatch between two sites, defined by formulas (5-8), can be used:

- The sum of the absolute errors of the "prediction" (Fig. 7).

- Average deviation: the average height of the figure (Fig. 8), bounded above and below the sites in the local area Ω.

- The equivalent of the "intensity" of errors in the local domain Ω.

- The root mean square error in the local area Ω.

The latter measure is preferable for 3D data, as takes into account the discrepancy in all directions and is defined in the dimension of time. For the analysis of 2D data, the sum of the absolute errors of the “prediction” is sufficient.

Based on these measures, we introduce the condition of mutual coordination of the two sites (Fig. 9):

- have the same indicator wave features,

- site number 1 has the smallest measure of discrepancy with the site number 2 of all sites of the route number 2, not exceeding a given threshold,

- Site No. 2 has the smallest measure of discrepancy with Site No. 1 from all sites of route No. 1.

The condition of mutual coordination of the three sites:

- have the same indicator wave features,

- site number 1 is mutually agreed with site number 2,

- site number 1 is mutually agreed with site number 3,

- Site No. 2 is mutually agreed with Site No. 3.

Further, by analogy, these conditions can be extended to the case of mutual coordination of four, five, etc. sites.

A basic algorithm is proposed for automatic tracking of one horizon over the initial set of sites selected by the user, which includes the following steps:

1. The user defines the initial set of sites {(x _i , y _i , t _i , α _i , β _i , A _i , S _i , ...)} that have the same indicator of the wave feature. The inclusion of these sites in the horizon.

2. Identification of routes adjacent to routes where the horizon is defined, and search for sites on them that meet the requirement of mutual consistency with at least one horizon site.

3. Selecting a site from the sites found in Section 2, which has a minimum discrepancy measure (Fig. 10). If such a site is located - inclusion of this site in the horizon, otherwise - the algorithm stops.

4. Repeat p. 2-3 until the algorithm stops or there are no unverified traces.

This algorithm can be improved by limiting the search area of the sites relative to the initial set or within the specified coordinates, track numbers and time, limiting the number of approvals included in the site horizon, introducing additional measures of site mismatch according to dynamic characteristics, eliminating repeated calculations and excluding from analysis sites formed due to the side effect of summation (Fig. 11).

A basic algorithm is proposed for automatic tracking of all possible horizons (Fig. 12):

1. Optionally or at the direction of the user, an initial tracking path is selected. Each site of the initial route, having mutual coordination with at least one site of the nearest route, forms a new horizon.

2. At random or according to the rule specified by the user, the track is selected that is closest to the already analyzed. All sites of this route are checked for mutual consistency with the nearest sites of horizons. If the track area is mutually agreed with at least one horizon site, it is included in this horizon. If the track site has mutual consistency with at least one site of the nearest track, but is not included in any horizon, it forms a new horizon.

3. Point 2 is repeated until an analysis of all traces has been carried out.

Improvement of this algorithm can be carried out by introducing additional measures for the mismatch of the two sites according to dynamic characteristics, adding an analysis of the traced horizons for possible combination or deletion, eliminating repeated calculations and excluding sites from the analysis due to the side effect of summation.

In contrast to the method for automatically tracking horizons / hodographs, the method for automatically detecting tectonic deformation zones and fracture zones is based on determining sections of the seismic field in which the regularity of the wave is violated (i.e., reduced coherence of oscillations is observed): a sharp lateral change in the general case of the waveform . Accordingly, in the cube or section of the sites, such zones will be characterized by the presence of sites that are “weakly” consistent with nearby ones and “false” sites that do not characterize the reflective properties of the medium, but are the result of a side effect of the summation of irregular waves. To identify the locations of such sites, we introduce the following measures for the mismatch of multiple sites as an estimate of the degree of coherence:

- Dispersion of angles (statistical measure)

Statistical characteristic calculated in a sliding window, limited in space and time by the formula (9).

- Measure "maximum minimum discrepancy"

The measure of the maximum minimum mismatch (MMD) is determined according to the following algorithm:

1) for a particular track with coordinates (x ₀ , y ₀ ), all track areas are sequentially selected,

2) for each site of p. 1, sites from neighboring tracks that have a minimum measure of mismatch between the two sites are searched (Fig. 9a),

3) among the found sites in p. 2, a site with a maximum measure of the mismatch of the two sites is selected (Fig. 13b),

4) the value of this measure determines the value of the cube / section of coherence at the point (x ₀ , y ₀ , t ₀ ) and its temporal neighborhood [t ₀ -W, t ₀ + W], if a larger value is not defined in this neighborhood (Fig. 13a, c).

- Measure "the sum of the minimum discrepancies"

The measure of the amount of minimum mismatch (SMR) is determined according to the following algorithm:

1) for a particular track with coordinates (x ₀ , y ₀ ), all track areas are sequentially selected,

2) for each site of p. 1, sites from neighboring tracks that have a minimum measure of mismatch between the two sites are searched (Fig. 9a),

3) among the found sites in paragraph 2, a site with a maximum measure of the mismatch of the two sites is selected (Fig. 14b),

4) the value of this measure is added to the value of the cube / section of coherence at the point (x ₀ , y ₀ , t ₀ ) and its temporal neighborhood [t ₀ -W, t ₀ + W] (Fig. 14a, c).

Unlike the known ones, the proposed measures have low values in areas of high coherence (regularity) of waves.

Despite the fact that the present invention can be implemented in various ways, it should be borne in mind that the above description should be considered only as an example of the implementation of the principles of the invention, which in no way limits the invention to the described specific algorithms and examples of their implementation.

The invention is illustrated by the following figures.

In FIG. 1 is a graphical representation of a site in three-dimensional space.

In FIG. 2 is a graphical representation of a cube of sites.

In FIG. Figure 3 presents the general scheme for solving the problems of tracking horizons / hodographs and assessing coherence using a cube of sites.

In FIG. 4 presents a general scheme for assessing the kinematic parameters of sites on a single track.

In FIG. 5 shows a diagram for obtaining one trace of a scan result using directional summation of traces.

In FIG. 6 is a diagram for computing scan results using the Fourier transform.

In FIG. 7 is a graphical explanation of the concept of absolute errors of “predicting” the position of sites.

In FIG. 8 is a figure, bounded above and below by equations of two sites.

In FIG. Figure 9 shows the image of the tracks of the sites and highlighted the sites corresponding (a, b) and not corresponding (a, c) to the requirement of mutual coordination of the two sites.

In FIG. 10 is a graphical explanation of the decision to include the site in the horizon / hodograph in the task of tracking the horizon / hodograph (projection on the horizontal plane).

In FIG. 11 shows a general scheme of an algorithm for automatic tracking of one horizon / hodograph.

In FIG. 12 is a general diagram of an algorithm for automatically tracking all possible horizons / hodographs.

In FIG. 13 shows the process of forming the path of the coherence cube according to the MMR measure. The path of the coherence cube (a), the analysis of the measures of inconsistency with the nearest paths (b), the change in the path of the coherence cube (c).

In FIG. Figure 14 shows the process of forming a coherence cube trace according to the CMR measure. The path of the coherence cube (a), the analysis of the measures of inconsistency with the nearest paths (b), the change in the path of the coherence cube (c).

Exemplary embodiments of the invention using real materials are shown in FIG. 15-18.

All the results presented were calculated using a single computer based on the Intel Core i7 processor. When processing 2D data, the time for calculating the sections of the sites did not exceed several minutes, and the processes of tracking all possible horizons / hodographs and calculating the sections of coherence lasted no more than a minute. Calculation of the site cube with the initial seismic cube size of 352 inline for 640 crosslines and 3000 trace samples took 2 hours, tracking of each horizon lasted several minutes.

In FIG. Figure 15 shows an example of interpretation of a temporal section of 2D MOGT. The initial seismic section (a), the section of the sites (b), the result of the algorithm for automatically selecting all horizons (c) and the section of coherence according to the “angle dispersion” measure (d).

In FIG. Figure 16 shows an example of tracking waves using VSP materials obtained in one of the wells in Western Siberia. VSP seismograms: Z-component of the near explosion point (a), R-component of the remote explosion point (g); the result of the automatic selection of all hodographs in the range of angles of incident waves (b), in the range of angles of reflected longitudinal waves (c, f) and in the range of angles of incident converted waves (e).

In FIG. Figure 17 shows an example of horizon tracking using a temporary 3D seismic cube 3D MOGT. The 3D CDP seismic cube plan and the position of the cube slice (a), the site cube slice (b), the user defining the set of initial sites by the seismic cube and site cube slice (c), the seismic cube slice and horizons obtained by the automatic tracking algorithm of one horizon (g ), the results of the automatic tracking algorithm for horizon No. 1 (d), horizon No. 2 (e), horizon No. 3 (g), horizon No. 4 (h), horizon No. 5 (i), the image of the traced horizons in three-dimensional space (to ) In automatic horizon tracking algorithms, a small group (up to 10) of initial sites lying on one slice was specified.

In FIG. Figure 18 shows an example of the calculation of coherence cubes from a temporary 3D seismic cube 3D MOGT of one of the hydrocarbon fields in Western Siberia. 3D MOGT seismic cube (a), measures of coherence cubes by measures: angular dispersion (b), ММР (в), СМР (d).

Information sources

1. Boganik G.N., Gurvich I.I. Seismic exploration. Tver: AIS Publishing House, 2006, p. 744.

2. Sivukhin D.V. General physics course. Moscow: Fizmatlit, 2005.T.IV. Optics.

3. Gurvich I.I., Nomokonova V.P., [ed.]. Seismic exploration. Handbook of geophysics. Moscow: Nedra, 1981, p. 464.

4. Ryabinkin L.A. Theory of elastic waves. Textbook for universities. Moscow: Nedra, 1987, p. 182.

5. Goltsman F.M., Limbach Yu.I., Pakhomov A.A. A statistical algorithm for tracking waves by their amplitudes and arrival times. Questions of the dynamic theory of seismic wave propagation. Leningrad: Science, 1975, pp. 195-201.

6. Puzyrev N.N. General issues of correlation of seismic waves. Discrete correlation of seismic waves. Novosibirsk: Nauka, 1971, p. 5-13.

7. International Conference on Acoustics, Speech, and Signal Processing, IEEE, 3. Harrigan E., et al. 1992. Seismic horizon picking using an artificial neural network, pp. 105-108.

8. Seismic horizon detection using image processing algorithms. Bondar I. 1992, Geophys Prospect, T. 40, pp. 785-800.

9. Patent RU 2516590, publ. 05/20/2014

10. Patent RU 2107931, publ. 03/27/1998

11. Patent RU 2144683, publ. 01/20/2000

12. Patent RU 2169931, publ. 06/27/2001

13. Calculation of cubes of coherence and singularity. Shlenkin S.I. et al., 2012, Seismic exploration technologies, vol. 2, p. 5-11.

14. Goldin S.V. 1974. Linear transformations of seismic signals. Moscow: Nedra, 1974, p. 352.

15. Dec V.N., Kogan A.B. Spatial filtering is a universal method for correlating geophysical data. 1993, Physics of the Earth, vol. 2, pp. 29-39.

16. Clareboat J.F. Theoretical foundations of geophysical information. With applications for oil exploration. Moscow: Science, 1981, p. 304.

17. Stepanov D.Yu. Algorithmic and software for processing seismic information based on filtration methods in the direction: dissertation for the degree of candidate of technical sciences. Tomsk, 2000, p. 201.

18. Stepanov D.Yu., Yapparova E.A. A new approach to the implementation of directional filters in the analysis of complex wave fields // Technologies of seismic exploration, 2005, No. 1, p. 32-37.

Claims (30)

1. A method of forming a cube or section of sites based on an array of source seismic data, including estimating the parameters of seismic waves along each path, both kinematic, such as parameters of the sites (x0, y0, t0, α, β), and dynamic wave parameters (for example, form S (t), amplitude A, energy E, phase ϕ), characterized in that it includes the following operations for each trace of the seismic record: directional summation of at least two nearest tracks in a given range of tilt / hodograph angles to obtain three Nogo or two-dimensional scanning result (cc incision or scanning), the search of local extrema in the scanning results, specification of the position of each extremum, certain side effects of the summation and forming a cube or incision sites according to the position of the extrema found. 2. The method according to p. 1, characterized in that the directional summation of at least two tracks located near the estimated point (x ₀ , y ₀ ) is carried out:

where Y _вx (t, x, y) is the initial seismic data, w (n, m) are the weighting coefficients of summation, Δx and Δy are the distances between the nearest traces along the coordinates, R (t, p, q) is the scan result (cube or scan section). 3. The method according to p. 1, characterized in that the directional summation of at least two traces located near the estimated point (x ₀ , y ₀ ) through the one-dimensional Fourier transform of the traces, summation of the spectra of traces with weights

, and the inverse Fourier transform of the spectrum of the sum. 4. The method according to p. 1, characterized in that the refinement of the position of each extremum of the scan result is carried out as

where R _{i, j, k} = R (t _i , p _j , q _k ). 5. The method according to p. 1, characterized in that the determination of sites formed due to the side effect of summation is carried out under the condition:

where (p _s , q _s ) are the parameters of the wave sites, ƒ ₀ is the carrier frequency of the wave, ( ^~ p, ^~ q) are the parameters of the tested sites, N × M is the number of traces involved in the summation, Δx and Δy are the distances between the nearest tracks along the coordinates. 6. The method according to p. 1, characterized in that as the array of source data using deep cubes or sections and directional summation is carried out in a given range of angles of inclination of the reflecting boundaries. 7. A method for automatically tracking horizons / hodographs by section or cube of sites, including those obtained by method 1, characterized in that the horizon is formed from a plurality of sites that are piecewise linear approximation of the surface, smooth or having a finite number of gaps, which can be carried out continuous smooth curve, and as a criterion for combining at least two sites in the horizon using the conditions of mutual consistency of sites. 8. The method according to p. 7, characterized in that in the condition of mutual agreement of the sites as a measure of the mismatch of the two sites using the sum of the absolute errors of the "prediction": ε ₁ = | t ₁ -τ ₂ (x ₁ , y ₁ ) | + | t ₂ -τ ₁ (x ₂ , y ₂ ) |. 9. The method according to p. 7, characterized in that in the condition of mutual agreement of the sites, the average deviation is used as a measure of the mismatch between the two sites: the average height of the figure bounded by the top and bottom of the sites in the local area Ω ε ₂ = 1 / | Ώ | ∫∫ _Ώ | τ ₁ (x ₁ , y ₁ ) - τ ₂ (x ₂ , y ₂ ) | dydx. 10. The method according to p. 7, characterized in that in the condition of mutual agreement of the sites, the equivalent of the "intensity" of errors in the local area Ω ε ₃ = ∫ _x ∫ _y [τ ₁ (x ₁ , y ₁ ) is used as a measure of the mismatch between the two sites -τ ₂ (x ₂ , y ₂ )] ² dydx. 11. The method according to p. 7, characterized in that in the condition of mutual agreement of the sites, the mean square error in the local area Ω is used as a measure of the mismatch between the two sites

12. A method for automatically detecting zones of tectonic deformations and zones of fracture by cube or section of sites, including those obtained by method 1, which consists in detecting sections of the source seismic data array in which the regularity of waves is violated, characterized in that as a criterion for violation of the conditions wave regularities use a measure of the mismatch of multiple sites. 13. The method according to p. 12, characterized in that the measure of the mismatch of multiple sites is defined as the variance of the angles in a sliding window, limited in space and time:

where N is the number of sites that fell into the window [x ± Z ₀ , y ± Z ₀ , t ± T ₀ ], (p _i , q _i ) are the parameters of these sites, w _i are the summation weights,

14. The method according to p. 12, characterized in that as a measure of the mismatch of multiple sites using the measure of "maximum minimum mismatch" (MMR). 15. The method according to p. 12, characterized in that as a measure of the mismatch of multiple sites using the measure of "amount of minimum mismatch" 16. The method according to any one of paragraphs. 14 or 15, characterized in that the sum of the absolute errors of the “prediction” is used as a measure of the mismatch between the two sites:

17. The method according to any one of paragraphs. 14 or 15, characterized in that the mean deviation is used as a measure of the mismatch between the two sites: the average height of the figure bounded above and below the sites in the local area Ω ε ₂ = 1 / | Ώ | ∫∫ _Ώ | τ ₁ (x ₁ , y ₁ ) - τ ₂ (x ₂ , y ₂ ) | dydx. 18. The method according to any one of paragraphs. 14 or 15, characterized in that the equivalent of the "intensity" of errors in the local area Ω is used as a measure of the mismatch between the two sites ε ₃ = ∫ _x ∫ _y [τ ₁ (x ₁ , y ₁ ) -τ ₂ (x ₂ , y ₂ )] ² dydx .. 19. The method according to any one of paragraphs. 14 or 15, characterized in that the mean square error in the local area Ω is used as a measure of the mismatch between the two sites

Images