RU2187130C2 - Process and gear to process seismic signal and to prospect for mineral resources - Google Patents

Abstract

FIELD: seismic prospecting for mineral resources. SUBSTANCE: given process includes following operations: generation of 3D seismic data, division of data by array of relatively small 3D cells, determination of similarity per each cell, determination of inclination and inclination azimuth of seismic routes included in it and display of image of inclination, inclination azimuth and similarity of each cell in the form of 2D map. In agreement with one variant similarity is function of time, number of seismic routes inside cell and visible inclination of visible inclination azimuth of routes inside cell. Similarity of cells is determined by way of conducting aggregate of measurements of similarity of routes inside cell and by selection of largest measurement. Visible inclination and visible inclination azimuth corresponding to largest of measured similarity in cell are considered evaluation of true inclination and true inclination azimuth inside cell. Color map characterized by hue, saturation and brightness is utilized to describe similarity, true inclination and true inclination azimuth of each cell. True inclination azimuth is laid off on axis of color hue, true inclination is plotted on saturation axis and largest measured similarity is laid off on brightness axis of color map. EFFECT: improved reliability of process and gear. 11 cl, 16 dwg

Description

 The invention generally relates to seismic exploration of minerals, and more specifically relates to the creation of methods and devices for identifying structural and stratigraphic characteristics in three dimensions (three-dimensional characteristics).

When conducting seismic exploration of minerals, seismic data are obtained along lines (see lines 10 and 11 in FIG. 1), which are formed by geophonic arrays in coastal areas to the surf zone and hydrophone streamers that cross the marine section. Geophones and hydrophones work as sensors when receiving energy that was previously directed into the soil (the thickness of the earth) and reflected from the surface of the rock boundary of the lower horizon. Energy is usually generated by Vibroseis ^® vehicles, which transmit impulses by exciting ground vibrations on the surface at specified intervals and frequencies. In offshore areas, air guns are often used for this purpose. The slight changes in energy that are received when it is returned to the surface often reflect variations in the stratigraphic, structural, and liquid contents of the reservoirs.

 In the implementation of three-dimensional (3D) seismic exploration, a similar principle is used, however, the lines and lattices are located more closely to provide a more detailed overlap of the lower horizon. With this overlap with high density, it becomes necessary to register, memorize and process extremely large volumes of information before it is possible to obtain the final results of the interpretation. Significant computer resources and complex software are required to process information, which allows amplifying a signal received from the lower horizon and suppressing noises that mask this signal.

 After processing the data, geophysical personnel compile and interpret 3D seismic information in the form of a 3D data cube (see Fig. 2), which effectively displays the characteristics of the lower horizon. When using a data cube, information can be displayed in various forms. Horizontal maps of time slices at selected depths can be built (see FIG. 3). When using a computer workstation, the interpreter can also cut through the desired field and study the outputs of the reservoirs at various seismic horizons. Vertical sections or cross-sections in any direction can also be made using seismic data or data obtained in the well. The seismic peaks of the reflectors can be outlined, as a result of which a temporary map of the horizon can be obtained. Temporal maps of the horizon can be transformed in depth to obtain a structural interpretation in the true scale at a specific level.

 Seismic data is traditionally obtained and processed in order to obtain images of seismic reflectors for structural and stratigraphic interpretation. However, changes in stratigraphy are often difficult to detect using traditional seismic images as a result of the limited amount of information that stratigraphic characteristics give in cross section. When working simultaneously with time slices and with cross sections, it becomes possible to see significantly larger sections of faults, however, fault surfaces in a 3D volume are difficult to identify when reflections from faults are not recorded.

 Coherence is a measure of the similarity or divergence of seismic traces. The greater the coherence between the two seismic tracks, the more they are similar. If we assign coherence to values in the range from 0 to 1, then “0” means the complete absence of coherence, and “1” means complete similarity (that is, the two tracks are identical, but may have a time shift). Coherence for multiple paths can be determined in a similar way.

 One method for calculating coherence is disclosed in US Patent Application 353934 to Bagorich and Farmer (Applicant Corporation Amoko), filed December 12, 1994. Unlike shaded terrain methods, which provide 3D visualization of faults, channels, landslides and other sedimentary characteristics from selected horizons, the process of determining coherence proposed by Bagorich and Farmer operates directly with seismic data. When there is a sufficient change in acoustic impedance, the 3D seismic coherence cube proposed by Bagorich and Farmer can be extremely effective in contouring seismic faults. It is also quite effective in detecting subtle changes in stratigraphy (for example, 3D images of meander distribution channels, elongated shallows (streamers), canyons, landslides and pictures of tidal catchments).

 Despite the fact that the process proposed by Bagorich and Farmer was very successful, it has some limitations. In this process, an assumption (condition) about the average zero value of seismic signals is used. This is approximately true when the correlation window exceeds the duration of the seismic pulse. For seismic data that contain a 10 Hz energy component, this requires the creation of a sufficiently long (long) 100 ms window in which stratigraphic mixing of information from both deeper and shallower time horizons can occur. Shortening the window (for example, up to 32 ms) leads to a higher vertical resolution (resolution), but often due to an increase in side effects associated with the seismic pulse wave. Unfortunately, a more reliable cross-correlation process using a non-zero average window is an order of magnitude more expensive in terms of calculations. Moreover, if seismic data is contaminated with coherent noise, estimates of the true inclination (dip) angle using only two paths will also be relatively noisy.

 Thus, there is a need to create methods and devices that can eliminate the disadvantages of already known solutions. In addition, it is highly desirable to improve the estimates of the angle of inclination in the presence of coherent noise.

 In accordance with the present invention, there is provided a method and an article for localizing underground characteristics, faults and contours. According to a first preferred embodiment of the present invention, the method includes the following operations: obtaining 3D seismic data covering a predetermined volume of the earth; dividing the volume into a grid of relatively small three-dimensional cells, each of these cells having at least five laterally located and mainly vertical seismic traces located in it; determination in each cell of similarity / similarity of traces with respect to two given directions; and displaying the similarity / similarity of each cell in the form of a two-dimensional map. In accordance with another embodiment of the present invention, the similarity / similarity is a function of time, the number of seismic traces in the cell, as well as the apparent slope and the apparent azimuth of the slope of the traces in the cell; cell similarity / similarity is determined by conducting multiple similarity / similarity measurements in the cell and selecting the largest measurement result. In addition, it is assumed that the apparent inclination and the visible azimuth of the inclination, which correspond to the largest similarity / similarity measurement result in the cell, are estimates of the true inclination and the true inclination azimuth of the paths in the cell. Finally, use a color map, which is characterized by color tone, saturation and brightness, to describe the similarity / similarity, true inclination and true azimuth of the inclination for each cell; in particular, the true inclination azimuth is plotted on the hue scale, true inclination is plotted on the saturation scale, and the largest similarity / similarity measurement result is plotted on the brightness scale of the color map.

 According to another preferred embodiment of the present invention, there is provided an article that includes means that can be read by a computer and which contains instructions for the computer to carry out the seismic survey. In accordance with one of the options, the computer receives 3D seismic data that overlap a given volume of the earth, and the tool instructs the computer to carry out the following operations; dividing the volume into a grid of relatively small three-dimensional cells, each of these cells having at least five laterally located and mainly vertical seismic traces located in it; determination in each cell of similarity / similarity of traces with respect to two given directions; and storing (values) of similarity / similarity of each cell in the form of a two-dimensional map. In accordance with one of the options for the command from the specified means determine the similarity / similarity as a function of time, the number of seismic traces in the cell, as well as the apparent inclination and the visible azimuth of the inclination of the traces in the cell; cell similarity / similarity is determined by conducting multiple similarity / similarity measurements in the cell and selecting the largest measurement result. In addition, it is assumed that the apparent inclination and the visible azimuth of the inclination, which correspond to the largest similarity / similarity measurement result in the cell, are estimates of the true inclination and the true inclination azimuth of the paths in the cell. The computer includes a means for creating a color image, which is characterized by color tone, saturation and brightness; and a tool that gives instructions for displaying the true azimuth of the inclination on the hue scale, true inclination on the saturation scale, and the largest similarity / similarity measurement result on the brightness scale.

 The method in accordance with the present invention is particularly well suited for interpreting fault planes in a 3D seismic volume, as well as for detecting thin stratigraphic characteristics in a 3D volume. This is because seismic tracks that intersect the fault line are usually of a different seismic nature than the tracks on either side of the fault. Measurement of multichannel coherence or similarity of paths along a time slice makes it possible to detect outlines (contours) of low coherence along these fault lines. Such measurements make it possible to identify critical details of the lower horizon that are difficult to obtain using traditional seismic sections. By calculating the similarity of the paths along a series of time slices, it is also possible to identify fault planes or fault surfaces using these fault contours.

The method in accordance with the present invention uses a multi-track similarity method, which in general is more reliable in a noisy environment than the three-trace cross-correlation method that was previously used to evaluate seismic coherence. In addition, the method for determining similarity proposed in this application for the invention provides:
- higher vertical resolution for good quality data than the three-trace cross-correlation method for measuring seismic coherence;
- the ability to map coherent events in 3D solid angle (inclination / azimuth);
- the possibility of developing the concept of complex attributes (defining signs) of the "path" for one of the complex attributes of the "reflector"; and
- by combining these enhanced attributes of complex traces with coherence and solid angle provides the basis for obtaining attributes of quantitative 3D seismic data stratigraphy, which allows the use of geostatistical analysis methods.

Moreover, seismic coherence, in contrast to the inclination maps of selected horizons, allows the analysis of:
- structural and stratigraphic frames earlier than the beginning of a detailed survey (the beginning of exploration);
- structural and stratigraphic characteristics of the full amount of data, including zones that are more shallow, deep or adjacent to the original zone of interest;
- subtle characteristics that are not displayed by the peaks of the seismogram in the form of peaks and troughs; and
- characteristics that are internal between the top and bottom of the formation or internal relative to the sequence of boundary peaks of the seismogram.

 Combined with coherence, the slope data cubes of the solid angle of coherent seismic reflection events allow you to quickly determine both structural and stratigraphic relationships (such as an inconsistent transgressive fit or an inconsistent regressive fit) between seismic data and an interpreted boundary sequence.

 Numerous other advantages and characteristics of the present invention will be apparent from the following detailed description of the invention, the examples described therein, and also from the claims and the attached drawings.

 In FIG. 1 is a schematic diagram showing the location of geophones for obtaining 3D seismic data from the lower horizon of the earth, for processing in accordance with the present invention.

 In FIG. 2, information extracted from data in turn obtained using the diagram of FIG. 1 is displayed.

 Figure 3 shows a horizontal time slice (t = 1200 ms) of 3D seismic data processed in accordance with the prior art.

 On figa-4H shows the various analysis windows (calculation stars) that can be used in the analysis of seismic coherence, inclination and azimuth of inclination.

 In FIG. 5 depicts a process in accordance with the present invention using an elliptical window centered on the analysis point.

In FIG. 6A and 6B are examples of a rectangular inclination / azimuth mosaic, useful in analyzing surveys with strike and tilt parallel to the information acquisition axis, as well as when the identified faults are perpendicular to the dominant strike and tilt of the reflector (p ₀ , q ₀ ).

 In FIG. 7A-7C show examples of three mosaics of the solid angle of inclination / azimuth space.

 In FIG. 8A-8D illustrate the construction of a 3D seismic attribute map (φ, s, d) for a 3D color space (H, L, S).

 Fig. 9 shows four surfaces through the color hemisphere of Fig. 8A for four coherence values.

 In FIG. 10A-10C show conventional vertical slices of the seismic data of FIG. 3.

 On figa-11C shows the seismic attributes, the inclination, the azimuth of the inclination and coherence obtained by applying the process in accordance with the present invention, for the data corresponding to figa-10C.

 12A and 12B show time slices (t = 1200 ms and t = 1600 ms) through the inclination azimuth cube in accordance with FIGS. 11A and 11B.

 13A and 13B show coherence images on a gray scale.

 On figa-14C shows sections of coherence, which correspond to the data of figa-10C.

 In FIG. 15A and 15B show the results of applying the similarity algorithm and the inclination / azimuth algorithm in accordance with the present invention.

 16A and 16B are structural diagrams showing information processing operations in accordance with a preferred embodiment of the present invention.

 Despite the fact that the present invention can be implemented in a variety of forms, we will further consider in detail the specific options for its implementation, shown in the drawings. However, it should be borne in mind that the further description is given only as an example of implementation of the present invention and in no way limits the scope of its application to the described specific options for its implementation or specific algorithms.

 The first operation of the process (see figa) is to obtain a set of seismic data in the form of seismic signal traces distributed in a three-dimensional volume of the earth. The methods by which such data are obtained and digitized for processing in the form of 3D data are well known per se.

Similarity process
The next operation is to generate (receive) a “coherence cube”. This operation is carried out by applying a multi-track similarity algorithm to 3D seismic data. This algorithm can take various forms. Regardless of the selected shape, its function is to compare the similarity of nearby (surrounding) areas of seismic data within a 3D seismic volume. The obtained value (or attribute) serves as a fairly reliable estimate of the signal discontinuity within the geological formation, as well as the signal discontinuities through a fault and erosion mismatches.

 The analysis lattice (or computational star) can be either elliptical or rectangular, with the “J” paths centered relative to a given output path (see FIGS. 4A-4H).

 In the drawings, “X” indicates the center of the analysis window, while “O” indicates the additional traces used in the similarity calculation. The round and rectangular windows of the minimum size, which are used to analyze data with an equal gap between the tracks (Δx = Δy), are shown in figa and 4D. Round and rectangular windows of minimum size, which are used to analyze data with a gap between the tracks in the transverse direction (crocc-line / strike) (y) are two times larger than in the longitudinal direction (in-line / dip) (x), and namely Δy = 2Δx shown in figv and 4E. Such unequal gaps are usually used to analyze slow changes in geology in the direction of strike. Larger analysis windows, which are used to obtain greater resolution when evaluating reflector tilt and azimuth, or to increase the signal-to-noise ratio in areas with a weak signal, are shown in FIGS. 4C and 4F.

The elliptical and rectangular analysis windows are centered relative to the analysis point, which is defined by the major axis, minor axis and azimuth of the major axis, as shown in FIGS. 4G and 4H. The data acquisition axes (x, y) are rotated by φ ₀ degrees relative to the North-East axes (x ', y'). Such asymmetric windows are useful in detecting a fault.

В этом выражении тройка (триплет) (τ, p, q) определяет локальное планарное (плоское) событие в момент времени τ, а p и q представляют собой видимые наклонения в направлениях x и y, измеренные в мс/м. Так как p=dsinφ и q=dcosφ, где d представляет собой истинное наклонение, а φ представляет собой азимут наклонения, то из этого следует, что
u_j(t, p, q, x, y)=u_j[t-d(xsinφ+ycosφ), x, y].If we center the axis (x, y) relative to the center of the analysis window containing J seismic traces u _j (t, x _j , y _j ), then the similarity σ (τ, p, q) can be defined as follows:

In this expression, a triple (triplet) (τ, p, q) defines a local planar (flat) event at a time instant τ, and p and q are visible inclinations in the x and y directions, measured in ms / m. Since p = dsinφ and q = dcosφ, where d represents the true inclination, and φ represents the azimuth of the inclination, it follows that
u _j (t, p, q, x, y) = u _j [td (xsinφ + ycosφ), x, y].

 Specialists in this field will easily understand that in the denominator of expression (1), the value J is the normalization coefficient. The numerator displays the average energy, and the terms under the sum sign in the denominator display the total energy of the traces. Thus, expression (1) reflects the ratio of coherent and incoherent energies.

В общем виде нам неизвестно, но мы хотим оценить значение (p, q), объединенное с локальным наклонением и азимутом гипотетического 3D события отражения.The task is to conduct a simultaneous 2D search (see Fig. 5) of the apparent inclination (fall) (p, q) in the longitudinal and transverse directions. However, the similarity estimate given by expression (1) is unstable for small but coherent values of seismic events that can happen if we summarize along the zero intersection of the plane of the coherent pulse. To avoid this, we assume that the coherence with (φ, p, q) at time τ and the visible inclinations (p, q) have an averaged similarity in the time window (or in the vertical analysis window with a height of 2w ms and half length K = w / Δt samples):

In general, we do not know, but we want to estimate the value (p, q) combined with the local inclination and azimuth of a hypothetical 3D reflection event.

Если x_max и y_max составляют половину ширины и половину длины прямоугольного окна анализа и если f_max является наивысшей текущей частотой, которая содержится в сейсмических данных, то тогда критерий Найквиста для выборки данных в двух точках на период ограничивает приращения видимого наклонения Δp и Δq в соответствии с выражениями
x_maxΔр≤1/(2f_max) и
y_mахΔq≤l/(2f_max).In accordance with one embodiment of the present invention, we evaluate (p, q) by searching for brute force over all possible inclinations (see FIGS. 6A and 6B). We assume that the interpreter is able to estimate the maximum true inclination d _max (measured in ms / m) using conventional seismic data displays (e.g., vertical data slices), therefore, the inclination limit is:

If x _max and y _max are half the width and half the length of the rectangular analysis window and if f _max is the highest current frequency that is contained in the seismic data, then the Nyquist criterion for sampling data at two points per period limits the increments of the visible inclination Δp and Δq in matching expressions
x _max Δp≤1 / (2f _max ) and
y _max Δq≤l / (2f _max ).

 It should be noted that the Nyquist criterion is valid for linear operations on seismic data; and expression (2) is non-linear. In practice, we found the need to limit Δp and Δq to half what is required by the Nyquist criterion, which allows us to obtain an exact similarity for a coherent inclination event.

Thus, our search for estimating the apparent inclination (p, q) of a seismic reflector reduces to calculating the similarity with (p _i , q _m ) for n _p * n _q discrete visible inclination pairs (p _i , q _m ), and
n _p = (2d _max / Δp) +1 and
n _q = (2d _max / Δq) +1.

It can be assumed that the visible inclination pairs (p _i , q _m ) serve as an estimate of the apparent inclination of the reflector when:
s (p, q) ≥ s (p _i , q _m ) (3)
for all - n _p ≤ 1 ≤ + n _p , - n _q ≤ 1 ≤ + n _q .

в которых d измеряется в мс/м, а угол φ измеряется в направлении по часовой стрелке от положительной (или северной) оси xⁱ. Простой поворот координат на угол φ₀ необходим в том случае, когда продольное направление получения данных х не совмещено с осью N-S (x') (см. фиг.4G).The estimation of the apparent inclinations (p, q) is related to the estimation of the true inclination d and the true azimuth of the inclination φ by simple geometric relationships:

in which d is measured in ms / m and the angle φ is measured in a clockwise direction from the positive (or north) axis x ⁱ . A simple rotation of the coordinates by an angle φ _{0 is} necessary in the case when the longitudinal direction of obtaining data x is not aligned with the axis NS (x ') (see fig. 4G).

Discretization and indication of a solid angle
Optimal angular discretization is important for two reasons: to minimize the cost of calculations and to limit the number of colors that can be displayed using commercial interpretation workstation software (for example, usually 64 colors for the Landmark Seisworks system and 32 for Geocvest IESX systems).

Индикация
Несмотря на то что имеется возможность отображать независимо подобие, наклонение и азимут, ясно, что последние два атрибута связаны друг с другом. Более того, уверенность, которую мы имеем относительно этих оценок, пропорциональна когерентности/подобию. Другие авторы (см. патент США 4970699 на имя Бушера и др. "Способ цветового отображения геофизических данных", заявитель - корпорация Амоко) показали, что цветовая модель HLS (цветовой тон, яркость, насыщенность) может быть достаточно эффективной при индикации мультикомпонентных сейсмических атрибутов (см. также Foley, J.D. & Van Dam, A., 1981 "Основы интерактивной графики", из-во Addison-Wesley, Reading, MA).In FIG. 7A shows discretization of apparent inclination using equal increments Δp and Δq in a rectangular lattice of 69 angles. FIG. 7B shows a discretization using equal increments Δd and Δφ in the radial lattice 97 of the angles. It is clear that we have no desire to sample the inclination (dip) d = 0 ms / m for ten different azimuths. The Chinese Checker mosaic of FIG. 7C more accurately displays the equal and therefore more economical quantization (d, φ) of a surface with a minimum number of points (for example, 61 angles). Each mosaic of FIGS. 7A-7C displays an approximately equal trajectory of the solid angle ΔΩ. For angular sampling, shown in figs, and for the radius of the circular analysis, and, the increment dip Δd is chosen in accordance with the formula

Indication
Although it is possible to display similarity, inclination, and azimuth independently, it is clear that the last two attributes are related to each other. Moreover, the confidence we have regarding these estimates is proportional to coherence / similarity. Other authors (see US Pat. No. 4,970,699 to Bushehr et al., “Method for color displaying geophysical data”, applicant, Amoko Corporation) have shown that the HLS color model (color tone, brightness, saturation) can be quite effective in displaying multi-component seismic attributes (see also Foley, JD & Van Dam, A., 1981, Fundamentals of Interactive Graphics, from Addison-Wesley, Reading, MA).

Turning now to FIGS. 8A-8D, where you can directly plot the azimuth φ on the axis of the color tone H:
H = φ
It should be noted that both H (commonly known as the "color disk") and φ can vary in the range from -180 to +180 degrees (see FIG. 8B). The blue color corresponds to the North in azimuth, the orange-pink - to the East, yellow - to the South, and the color of forest green to the West. The azimuths, which correspond to the zero inclination, are arbitrarily assigned the value 0 degrees (North), so they are plotted as blue on the graph.

After that, we produce (see Fig. 8C) averaged similarity / coherence c on the brightness axis L
L = αs
Where
0 ≤ L ≤ 100
0 ≤ s ≤ 1.0,
and α is a scale factor less than 100, since changes in hue and saturation near L = 0 (black) and L = 100 (white) are difficult to distinguish. White or L = 100 corresponds to high similarity or c = 1, while black or L = 0 corresponds to low similarity, c = 0. Intermediate degrees of similarity correspond to intermediate grades of gray (such as silver, gray, and charcoal). Brightness (sometimes called luminosity) reflects the degree of illumination. It is displayed with a gray scale ranging from black to white.

Finally, on the saturation scale S, we postpone the inclination d
S = 100 / d _max .

 Saturation (S = 0) and hue are chosen arbitrarily; we also have the opportunity to display this attribute for the value (H = 0, S = 100), which would allow us to indicate the similarity as white, pastel white, pure blue, midnight blue and black. Saturation expresses the absence of color dilution with white light. Fully saturated color does not have the addition of white light; with the addition of white, the color is “washed out” without changing its color tone (see Fig. 8D).

 Fig. 9 shows four constant surfaces through the 3D (H, L, S) color hemisphere (φ, s, d) shown in Fig. 8A, which correspond to c = 1.00, c = 0.75, c = 0 , 50 and c = 0.00.

 In Appendix 1 at the end of the description, the color scheme is described in more detail. Some of the advantages of the HLS color model are as follows: the azimuth is cyclical and can be delayed purely on a cyclic color wheel (color tone); azimuths that correspond to d = 0 do not make sense; all azimuths can be smoothly transformed into grayscale for small inclinations; and lower confidence in the estimation of inclination and azimuth in weak zones, and lower degrees of similarity (such as through faults) are displayed in darker colors.

The implementation of the mathematical process
Interpretation workstations of Landmark and GeoKvest firms (see FIG. 16B) can be used, for example, to visualize and interpret faults and stratigraphic characteristics if the processed data is loaded as a seismic volume. Visualization software (such as Landmark's SeisCube software) can be used to quickly perform slices through a seismic volume to help understand complex fault relationships.

Computer program
The FORTRAN 77 program is written for performing calculations and provides information for the previously described mappings. Additional details are given in Appendix 2 at the end of the description. Access to each UMN trace is made according to its longitudinal and transverse indices M and N. The user selects a rectangular or elliptical spatial analysis window or cell at each point / trace in the input data set (see FIG. 4G). The major and minor axes of this analysis window, a and b, are specified as a = aplength, b = apwidth. The orientation or azimuth of the major axis φ _{a is} given by φ _a = apazim. The rectangular analysis window (FIG. 4H) is displayed by setting R on the command line. Indicators 2J relative to the center of this analysis window (and which correspond to the traces falling into this window) are tabulated as a simple list, with m (j) and n (j) indicating the coefficient of the trace (relative to the analysis trace U _MN ) in the x and y directions, respectively. The program performs a simultaneous 2D search by visible inclinations (p, q) in the longitudinal and transverse directions, and (p ² + q ² ) ^1/2 <+ smax. The increments dp and dq are chosen in such a way that the data are quantized at four points per period = 1 / (fref) at the edge of the analysis window. For interpretation, it can be convenient to express each visible inclination pair (p, q) in spherical coordinates as the true (in time or depth) inclination d and the inclination azimuth φ.
The data in the analysis window is interpreted at time t-px-qy for each test inclination and azimuth (see figure 5), which leads mainly to the "flattening" of the data. The similarity for this test inclination and azimuth at the point of analysis is defined as the similarity of these flattened paths in the analysis window.

в котором x и y представляют собой расстояния, измеренные от центра окна анализа. Это может быть выражено как

где Dx и Dy представляют собой промежутки между трассами в продольном и поперечном направлениях.For data in the time domain, we flatten (align) the jth trace with respect to the analysis point (M, N) using the expression

in which x and y are the distances measured from the center of the analysis window. It can be expressed as

where Dx and Dy are the gaps between the tracks in the longitudinal and transverse directions.

После этого подобие может быть вычислено для всех последующих наклонений и азимутов с использованием выражения

Как и для анализа скорости, подобие для каждого наклонения, азимута и точки анализа сглажено путем временной интеграции текущего окна для частичных сумм от -К до +К, где К= apheight/dt. В этом случае когерентность с (τ, p, q) может быть определена как

Те пары наклонения и азимута Ω = (d,φ), которые имеют максимальную когерентность с (в интегрированном текущем окне), принимаются в качестве оценки когерентности, с, наклонения и азимута (d, φ) для данной точки анализа.For data in the depth region, we flatten (align) the jth trace using the expression

After that, the similarity can be calculated for all subsequent inclinations and azimuths using the expression

As for velocity analysis, the similarity for each inclination, azimuth and analysis point is smoothed out by temporarily integrating the current window for partial sums from -K to + K, where K = apheight / dt. In this case, coherence with (τ, p, q) can be defined as

Those inclination and azimuth pairs Ω = (d, φ) that have maximum coherence c (in the integrated current window) are accepted as estimates of coherence, s, inclination and azimuth (d, φ) for a given point of analysis.

Examples
In FIG. 11A-11C are 3D images of seismic attributes (φ, s, d) that correspond to FIGS. 10A-10C and are constructed using a similarity-based coherence algorithm in accordance with expression (6) and the color display technique described in FIG. .8 and 9. The input data is sampled at 4 ms, the gap in the longitudinal direction of the trace is Δx = 12.5 mm, and the gap in the transverse direction of the trace is Δy = 25 mm, and the information acquisition line is oriented along the NS axis. In FIG. 11A-11C used a circular (round) analysis window or cell with dimensions a = b = 60 mm (see Fig. 4A), which in the calculation includes 11 traces. The maximum search inclination (see FIG. 7C) was d _max = 0.25 ms / m, which led to the presence of 61 search angles. The used time (current) integration time is w = 16 ms or K = 4, which allows averaging of the similarity calculation for 9 samples.

The lines AA 'and BB' in FIGS. 10A and 10B are selected so that vertical sections through the center of the salt dome extend from the South to the North and from West to East (SN, W-E). The line CC 'is offset from the line SN and illustrates the appearance of vertical shifts in a vertical section. On figa-11C the inner part of the salt dome is displayed in dark colors that correspond to the zone of mainly low coherence. Zones of low coherence correspond to radial shifts that are visible along the CC 'line. Coherent planar inclinations are represented as light gray regions and dominate in the section at a distance from the salt dome, in particular along the CC 'line. The blue color on the north side of the salt dome (which is visible on the NS line AA ') corresponds to deposits having a steep slope (d = d _max ) to the North. These inclinations become gradually smaller when moving away from the salt dome and are first indicated as blue (at saturation S = 100.0), then as blue “cadet” (S = 75.0) and blue steel (S = 50.0) and then the inclinations are aligned and displayed as gray (S = 0,0). The yellow color on the south side of the salt dome (which can be seen along the line AA ') corresponds to deposits having a steep slope to the south. The orange-pink color on the eastern slope of the salt dome (shown on the E-W line BB ') corresponds to sediments having a steep slope to the East. These inclinations also become gradually smaller when moving away from the salt dome and are first indicated as an orange-pink color (S = 100.0), then as a siena (reddish-brown color) (S = 50.0) and, finally, as gray color that matches flat inclination. The color of forest green on the western slope of the salt dome (shown on line AA ') corresponds to sediments with a steep slope to the West. These inclinations are also flattened away from the salt dome and displayed using the colors shown in the western part of the image in Fig. 9. The NS line SS 'is not radially aligned with the salt dome. Thus, rotation outside the plane of various fault blocks is shown, with the gray block corresponding to inclinations to SW, and the cyan block corresponding to inclinations to NW.

 Since these 3D attributes are calculated for each point in the input seismic volume, they can be displayed as horizontal time slices of the attribute (see FIGS. 12A and 12B); they correspond to a time slice of raw seismic data. Both the inside of the salt dome and the radial faults are displayed in dark colors, which corresponds to incoherent data zones. Due to the purely radial symmetry of the salt diapir at t = 1,200 ms (see Fig. 12A), the inclined deposits that flank the diapir also come out azimuthally, so that their azimuths almost exactly correspond to the color legend on the left side of Fig. 9. This picture is somewhat less symmetrical at t = 1,600 ms (see FIG. 12B), and the inclination to the South is smaller than to the North. In addition, inside the salt dome, internal blocks of coherent data can be seen.

 Referring to FIG. 9 color legend allows you to have only four "buckets" of coherence. In order to study coherence in more detail, it should be postponed on the graph as the only attribute. This is shown in FIGS. 13A and 13B, where all 184 colors are applied to the simple gray scale shown in FIG. 8C. In this image, the maximum coherence (c = 1.0) is displayed as white, and the minimum coherence (c = 0.0) is displayed as black. Although the interior of the salt diapir is shown as a highly incoherent zone, this image better shows the subtle details in the radial fracture pattern. In particular, faults emerging from the salt dome are shown, with some branching as they are removed. In addition to more continuous binding of the coherence attribute, part of this difference in perception (perception) is due to the fact that the observation of colors and black and white in the human retina is performed using various receptors (cones and rods). There is also a psychological difference in the ability to differentiate between green and blue colors in the male and female population. For this reason, male interpreters often prefer the image of a simple single coherence attribute shown in FIGS. 13A, 13B and 15A to the multi-attribute image (φ, c, d) shown in FIGS. 11A-12B and 15B. In reality, these images are complementary: the image of the 3D component is useful in recognizing the occurrence of conflicting azimuths of inclination between adjacent rotated fault blocks; and a single-component image is useful for emphasizing the contours (boundaries) or incoherent gaps that separate them.

Consideration of the features of the data processing process
A careful examination of FIGS. 13A and 13B reveals the area of incoherent energy that surrounds the salt dome. To clarify the reasons for the appearance of such effects, vertical sections were made through a single-component coherence cube, which corresponds to the seismic data of Figures 10A-10C. The results are shown in figa-14C. The interior of the salt dome is clearly incoherent. The incoherent underwater characterization of the canyon (described by Nissen et al. In the article "3D seismic coherence technique used to identify and contour landslide parameters", 1995, SEG Expanded Abstracts, pp. 1532-1534) is shown in the northern part of the salt dome. If the seismic data shown in FIGS. 10A-10C are superimposed on the coherence cross-sections shown in FIGS. 14A-14C, then one can clearly see the low coherence zones of FIGS. 14A-14C with zero intersection of the seismic reflection events of FIGS. 10A-10C. This can be easily understood if we assume that there is a fixed but not coherent seismic noise level for all the data. For analysis points where the visible inclinations are combined with the peaks and troughs of large amplitude seismic reflectors (for which the signal energy estimate is high relative to incoherent noise), a high signal-to-noise ratio can be expected, which leads to the appearance of a high coherence estimate. However, if our analysis point is such that the visible inclinations are aligned with the zero intersections of the same seismic reflectors, then the signal will be low relative to our incoherent noise, so we can expect a low signal-to-noise ratio, which leads to an estimate of low coherence .

 We found three ways to increase the signal-to-noise ratio: the first is more suitable for conducting structural analysis; the second is more suitable for stratigraphic analysis; and the third is suitable for both types of analysis.

 Consider the first way to increase the signal-to-noise ratio. Due to the steep inclination (less than 45 degrees from the vertical) of the faults, the signal-to-noise ratio can be increased by simply increasing the size of our vertical analysis window w, which is included in expression (2). In this case, two effects are observed. The first is associated with a decrease in structural leakage, which corresponds to the zero crossing points of the reflectors as the vertical size of the integration window increases. The second is associated with a decrease in lateral resolution of faults as the vertical window size increases, since only a few of faults are purely vertical. It turned out that a good compromise is the choice of the analysis window with w = 16 ms (which includes the full cycle (period) of the peak of 30 Hz data energy).

 The second way to increase the signal-to-noise ratio (which is equally suitable for both stratigraphic and structural analysis) is to extract coherence along the interpreted stratigraphic horizon. If this stratigraphic horizon is combined with an extremum of seismic data, such as a peak or a trough, then only data that has a relatively high signal to noise ratio is displayed. It is clear that the extraction of coherence data, which correspond to the zero intersection, significantly exacerbates the image of coherence. A more economical version of this approach is to preliminarily flatten the data along the horizon of interest, followed by the calculation of seismic attributes only for the selected horizon. This approach is somewhat more sensitive to operator errors for automatic (and human!) Sampling (taking readings), since cyclic omissions of the interference signal during sampling are somewhat random and therefore always manifest as incoherent.

 Small characteristics (for example, small channels; small characteristics of tidal channels that correspond to processed delta sands; and small layered faults) do not exist at any distance above or below the interpreted stratigraphic horizon; therefore, the inclusion of any data above or below the horizon, where they are localized, adds uncorrelated changes in amplitude, which makes such gaps (sections) more coherent and, therefore, washes them out. If time samples above or below the interpreted horizon contain independent, there may be powerful gaps, then these gaps will seep into the analysis data for large windows, while this stratigraphic horizon will contain parameters mixed with other stratigraphic horizons obtained at other geological times.

Вычисления σ(τ,p,q) и c(τ, p, q) полностью аналогичны выражениям (1) и (2), причем следует отметить, что v_j ² определено выражением
v $\genfrac{}{}{0ex}{}{\text{2}}{\text{j}}$ ≡ v_jv $\genfrac{}{}{0ex}{}{\text{*}}{\text{j}}$ ≡ (u_j+iv_j)(u_j-iv_j).
Третий способ повышения отношения сигнал/шум позволяет избежать цифровых нестабильностей в оценке подобия в выражении (1) у "нулевых пересечений" мощных, в другом отношении, рефлекторов.The third way to increase the signal-to-noise ratio is to generalize the initial set of seismic traces u _j for the analytical trace v _j , which is defined as
v _j (t) ≡ v _j (t) + iu $\genfrac{}{}{0ex}{}{\text{H}}{\text{j}}$ (t)
where u _j ^H (t) is the quadrature or Hilbert transform for u _j (t), ai equals

The calculations of σ (τ, p, q) and c (τ, p, q) are completely analogous to expressions (1) and (2), and it should be noted that v _j ^{2 is} defined by the expression
v $\genfrac{}{}{0ex}{}{\text{2}}{\text{j}}$ ≡ v _j v $\genfrac{}{}{0ex}{}{\text{*}}{\text{j}}$ ≡ (u _j + iv _j ) (u _j -iv _j ).
The third way to increase the signal-to-noise ratio avoids digital instabilities in the similarity assessment in expression (1) for the “zero crossings” of powerful, in another respect, reflectors.

The effect of the horizontal analysis window
In the study of expression (2), it becomes clear that the cost of calculations during the analysis increases linearly with an increase in the number of traces included in the analysis. However, when comparing the similarity based on the time slice of coherence with 11 traces with the data of the time slice with 3 cross-correlation traces (with an identical vertical analysis window with w = 32 ms), it can be assumed that adding traces during the calculations can lead to increasing signal to noise ratio. In general, the signal-to-noise ratio increases with increasing size of the analysis window. However, the total coherence decreases somewhat (which appears to be less white when observed), since as the window size increases, the approximation of a possible bent reflector by a constant (p, q) planar event is violated. As a rule, the signal-to-noise ratio of the inclination / azimuth estimates increases as the number of traces increases during the calculations until the approximation of the locally planar reflector is no longer observed.

Conclusions
The 3D similarity technique proposed in this patent application is an excellent means of measuring seismic coherence. Through the use of an arbitrary size analysis window, we are able to balance conflicting requirements for obtaining maximum lateral resolution and maximum signal to noise ratio, which is not possible using the cross-correlation technique with three fixed paths. Accurate coherence measurements can be achieved using a short time (vertical) integration window of the order of the shortest data periods, while a cross-correlation technique with a zero average value is mainly used with an integration window that is larger than the longest data period. Therefore, the results of the similarity determination process have less vertical blurring (blurring) of geology than the results of the cross-correlation process, even for wide spatial analysis windows (see FIGS. 15A and 15B). What is of equal importance for assessing coherence, the similarity determination process provides (gives) a direct means of estimating the 3D solid angle (inclination and azimuth) for each reflector event. These solid angle maps may or may not be associated with conventional temporal structure maps that define the boundaries of formations. Similar to the basic process of Bagorich and Farmer (for example, cross-correlation), an estimate of the instantaneous inclination / azimuth cube can be obtained before any interpretation of the data, which can be used for rough (initial) viewing of the geological deposition. In reconnaissance mode, coherence and inclination / azimuth cubes allow the user to select key inclination and strike lines that intersect important structural or sediment characteristics at a very early stage in the project interpretation phase. In interpretation mode, these inclinations and azimuths can be associated with the formation and / or sequence of boundaries, so that it becomes possible to map the patterns of progression or transgression of the internal structure in 3D. Finally, if we estimate the instantaneous inclination and azimuth at any point in the data cube, then we can use the usual attributes of the seismic trace to localize planar reflectors, resulting in significantly increased signal-to-noise ratios.

и

где U (τ, p, q) равняется
(см. числитель выражения (1)).Although a preferred embodiment of the invention has been described, it is clear that various changes, additions and modifications may be made by those skilled in the art that do not, however, fall outside the scope of the following claims. In particular, other algorithms can be used to measure the similarity of nearby areas of seismic data or to generate a “cube of discontinuity”. Moreover, the calculations shown and described may be replaced by their equivalents. For example, instead of conducting a search by the apparent inclinations p and q, you can search by the inclination and azimuth (d, φ). An inversion of the calculated similarity can be used, which allows you to get an image similar to the negative of a photograph. In addition, some characteristics of the invention can be used independently of its other characteristics. For example, after the estimation of the solid angle (inclination and azimuth), a smoother and more reliable multi-path estimate of the attributes of a common complex path can be obtained (Taner, M.T., Koehler, F., & Sheriff, RE: 1979, " Analysis of the integrated seismic route ", Geophysics, 44, 1041-1063). Instead of calculating these attributes on a single trace, you can calculate the attributes of the corner stack (set) of traces inside the analysis window. You can calculate
a _i (τ, p, q) = {[U (τ, p, q)] ² + [U ^H (τ, p, q)] ² } ^1/2 ,
ψ _i (τ, p, q) = tan ^-1 {U ^H (τ, p, q) / U (τ, p, q)},

and

where U (τ, p, q) equals
(see the numerator of expression (1)).

U ^H (τ, p, q) represents the Hilbert transform or the quadrature U (τ, p, q);
and _i (τ, p, q) is a shell or instantaneous amplitude;
Ψ _i (τ, p, q) represents the instantaneous phase;
f _i (τ, p, q) represents the instantaneous frequency; and
b _i (τ, p, q) represents the instantaneous bandwidth. (See Cohen, L.; 1993; Instant Nothing; Proc. IEEE Int. Conf. Acoust. Speech Signal Processing, 4, 105-109).

компонент фазы девяносто градусов

а также асимметрию, время нарастания и длительность отклика (реакции). Так как вдоль направления истинного наклонения происходит перемешивание, то медленные изменения амплитуды, фазы, частоты и ширины полосы компонентов события должны быть сохранены. Более того, вычисление когерентности/сходства/подобия позволяет произвести "анализ текстуры" аналогичных сейсмических областей. В сочетании с "анализом кластеров" анализ текстуры позволяет провести сегментационный анализ. Среди прочего, это дает возможность производить геологические корреляции и экстраполировать геологический характер нижнего горизонта. Кроме того, определение когерентности может быть использовано для наложения априори ограничений как на постстековую (post-stack), так и на престековую сейсмическую инверсию. Следует иметь в виду, что могут быть проведены различные модификации, вариации и изменения, а также использованы различные альтернативы, что не выходит за рамки настоящего изобретения и соответствует его духу, как это определено в приложенной формуле изобретения. Само собой разумеется, что все такие модификации перекрываются объемом патентных притязаний, изложенным в формуле изобретения.In addition to these “instantaneous” attributes, other attributes can be proposed for characterizing the signal in a given path envelope lobe, for example, an attribute at the peak of the envelope τ _e . These include (see Bodine, J. N.; 1994, “Waveform Analysis with Seismic Attributes,” presented at the 54th annual Mtg. SEG conference, Atlanta, California, USA):
pulse shell
a _r (τ, p, q) = a _i (τ _e , p, q),
pulse phase
Ψ _r (τ, p, q) = Ψ _i (τ _e , p, q),
pulse frequency
f _r (τ, p, q) = f _i (τ _e , p, q),
pulse width
b _r (τ, p, q) = b _i (τ _e , p, q),
zero phase component

ninety degree phase component

as well as asymmetry, rise time and duration of the response (reaction). Since mixing occurs along the true inclination direction, slow changes in the amplitude, phase, frequency, and bandwidth of the event components must be preserved. Moreover, the calculation of coherence / similarity / similarity allows a “texture analysis” of similar seismic regions. In combination with “cluster analysis” texture analysis allows for segmentation analysis. Among other things, this makes it possible to produce geological correlations and extrapolate the geological nature of the lower horizon. In addition, the definition of coherence can be used to impose a priori constraints on both post-stack and pre-stack seismic inversions. It should be borne in mind that various modifications, variations and changes may be made, as well as various alternatives used, which is not beyond the scope of the present invention and is consistent with its spirit, as defined in the attached claims. It goes without saying that all such modifications overlap with the scope of patent claims set forth in the claims.

Claims (8)

 1. A method for localizing underground characteristics, faults and contours, which includes the following operations: (a) obtaining 3D seismic data covering a predetermined volume of the earth’s thickness, (b) dividing at least part of the indicated volume into a grid of relatively small three-dimensional analysis cells, and (c) performing calculations for the traces localized in the cells, characterized in that (1) operation (b) is carried out by localizing at least five laterally displaced from each other and mainly vertical seismic traces, (2) operation (c) is carried out by determining in each of these cells one criterion of similarity of at least five paths with respect to at least two predetermined directions, the land volume being described by a three-dimensional lattice of criteria for similarity of paths, and (3) registering these criteria similarities in the form of an image of a two-dimensional map of underground characteristics, wherein said two-dimensional map passes through said three-dimensional lattice.  2. The method according to claim 1, characterized in that when performing operation (2), these predetermined directions are mutually perpendicular, and the indicated similarity of these traces inside each cell is measured at least as a function of time, the number of seismic traces within the specified analysis cell as well as the apparent inclination and the visible azimuth of the inclination of the indicated paths within the specified analysis cell.  3. The method according to p. 2, characterized in that the indicated similarity of the indicated traces within each cell is determined by conducting a plurality of similarity measurements of these traces within each cell and selecting the largest measurement result of the indicated similarity of each cell, and operation (2) further includes the operation of determining the apparent inclination and the apparent azimuth of the inclination corresponding to the largest of these measurements, which will be an estimate of the true inclination and the assessment of the true azimuth of the inclination I seismic traces within said cell analysis.  4. The method according to claim 3, characterized in that each of the specified set of measurements of the indicated similarity is at least a function of the energy of these traces, and the specified energy of these traces is a function of time, the number of seismic traces within the specified analysis cell, and the apparent inclination and the apparent azimuth of the inclination of the indicated traces within the specified analysis cell.  5. The method according to claim 3, characterized in that the map indicated in operation (3) is a color map, which is characterized by color tone, saturation and brightness, moreover, one of these estimates of the true azimuth of the inclination, these estimates of the true inclination and the largest the calculated similarity criteria is applied to one of the scales of brightness, color tone and saturation, while the other of the indicated estimates of the true azimuth of the inclination, the specified estimates of the true inclination and the specified largest of the crit riev similarity applied to other scales of luminance, hue and saturation, and the remaining one of said estimates of true azimuth of inclination, these estimates of true dip and the largest of said calculated similarity criteria applied to the remaining one of the gray scale, hue and saturation.  6. The method according to p. 5, characterized in that the operation (3) includes the following operations: mapping the indicated estimates of the true azimuth of the inclination along the specified axis of the color tone, mapping the estimates of the true inclination along the specified axis of the saturation, and applying to the map of the indicated largest calculated similarity criteria along the specified brightness axis. 7. The method according to claim 4, characterized in that each similarity criterion is at least a function

where each cell is characterized by lateral dimensions (-m _x , + m _x , -m _y , + m _y ), where x and y are the distances measured from the center of the analysis cell, p and q are visible inclinations in the x and y directions respectively, and the expression u _f (t, p, q, x, y) represents a seismic trace within the specified analysis cell, while the true inclination d and the azimuth of the inclination φ are associated with p and q expressions: p = dsinφ and q = dcosφ .
8. The method according to claim 7, characterized in that each similarity criterion is a function

9. The method according to claim 1, characterized in that when performing operation (2), said criterion is at least a function

in which J represents the number of traces in the indicated analysis cell, while u _j (τ, p, q) is the mapping of seismic traces in the indicated analysis cell, where τ is the time, p is the apparent inclination in the x direction, and q is visible inclination in the y direction, while p and q are measured in ms / s, and the x and y directions are mutually perpendicular. 10. The method according to claim 9, characterized in that when performing operation (2), said criterion is also a function

11. A method for searching for hydrocarbon deposits, according to which the colored seismic attributes of 3D seismic data for a given three-dimensional volume of the earth’s thickness are displayed by: (a) using a computer to convert the volume into a grid of relatively small three-dimensional cells, and (b) determining image colors using the hue scale, saturation scale and brightness scale, characterized in that it includes the following operations: (1) operation (a) due to localization in each of the specified cells of the plot of at least five seismic traces, (2) taking multiple similarity measurements using a computer for at least five seismic traces inside each of these cells, each measurement being at least a function of time, the number of seismic traces inside the specified cell, and also the visible inclination of the indicated paths and the apparent azimuth of the inclination of the indicated paths, (3) the choice of one of the many indicated similarity measurements for each cell, (4) use as an estimate of the true inclination and as estimates of the true azimuth of the inclination in each cell, the visible inclination and the visible azimuth of the inclination, which correspond to one selected from the indicated similarity measurements in the specified cell, (5) for each cell, mapping the indicated estimate of the true azimuth of the inclination on the color scale, (6) for each cell, plotting the indicated estimate of the true inclination on the saturation scale, (7) for each cell, plotting one of the calculated similarity measurements on the brightness scale for the map, and (8) using color image to identify structural and sedimentological parameters that are typically associated with the capture and accumulation of hydrocarbons.

Images