Fluvial facies multi-phase superposed sand body multi-dimensional identification and description method

Abstract

The invention discloses a fluvial facies multi-phase superposed sand body multi-dimensional identification and description method, and relates to the technical field of petroleum and natural gas exploration and development. The method comprises a synthetic record well seismic calibration step, a three-dimensional isochronous model establishment step, a river channel period determination step, a river channel sand body main body identification step, a river channel fine carving step and a single river channel early and late period judgment step. According to the fine riverway boundary identification and sand body thickness description method based on the high-density isochronous sequence slice, in the identification process, the top and bottom range of the riverway boundary and the sand body is judged on the automatically tracked high-density isochronous plane, and the description result isochronism, accuracy and working efficiency are all higher than those of a conventional method.

Description

Technical Field

[0001] The invention relates to the technical field of oil and gas exploration and development, and more specifically to a multi-dimensional identification and characterization method for multi-stage superimposed sand bodies of fluvial phases. Background Art

[0002] The commonly used fluvial sand body characterization method includes the following steps:

[0003] Step 1: Use seismic horizons, sand body thickness, sedimentary facies and development horizons at all levels to perform well-seismic calibration and obtain well-seismic calibration results;

[0004] Step 2: Based on the well-seismic calibration results, establish the spatial positions of sandstone groups, sub-layers and sedimentary units in the time domain;

[0005] Step 3: Obtain the type of sedimentary facies, and combine the spatial layers of the above three vertically according to the type of sedimentary facies to obtain the framework model;

[0006] Step 4: Use the framework model to identify the distribution characteristics of the channel sand bodies in the target work area.

[0007] In the division of the stratigraphic framework of the sand layer group, the stratigraphic framework is first divided according to the development characteristics of the sand bodies in the target area, and the sub-layer division based on the sedimentary cycle on the well curve is completed based on the sedimentary cycle characteristics. On this basis, combined with the well-seismic calibration, the reflection characteristics of each interface on the seismic data are clarified, and the sedimentary cycle sequence interpretation work in the three-dimensional seismic is completed. Then, the sand layer group is matched with the sedimentary cycle to divide the single sand body period.

[0008] In terms of sand body characterization, first determine the extension direction of the single-phase channel sand body based on whether the well location encounters a single-phase channel sand body, and then carry out channel top and bottom interface tracking work in 3D seismic based on the results of drilling interpretation. Common tracking methods include manual tracking and seed point diffusion tracking methods. In the characterization of single-phase channel sand bodies, it can be determined whether the well location encounters a single-phase channel sand body based on the well location data. If it is encountered, the boundary of the single-phase channel sand body is characterized based on the actual drilling data of the well location. If it is not encountered, the boundary is characterized based on the width-to-thickness ratio of the single-phase channel sand body. Finally, the vertical staging results of each layer of single-phase channel sand bodies are integrated to obtain the channel sand body carving results.

[0009] The above method can identify the distribution characteristics of river channel sand bodies of different thicknesses, thereby improving the accuracy of identifying the distribution characteristics of river channel sand bodies.

[0010] However, when the thickness of the sand body in a single channel decreases or there are differences in the deposition time of a single channel, the corresponding seismic reflection characteristics will weaken or be disturbed, resulting in interruption or inequality in the tracking of the sand layer interface, affecting the description of the planar distribution characteristics of the channel. There are still problems with the reliability of the sand body boundary characterization and the isochronism of the longitudinal thickness, and the work efficiency of the entire operation process also needs to be improved. Summary of the invention

[0011] In order to overcome the defects and shortcomings in the above-mentioned prior art, the present invention provides a multi-dimensional identification and characterization method for multi-stage superimposed sand bodies in river phases. The purpose of the present invention is to provide a fine channel boundary identification and sand body thickness characterization method based on high-density isochronous sequence slices. The characterization method of the present invention relies on judging the channel boundary and the top and bottom range of the sand body on the automatically tracked high-density isochronous surface. The isochronism, accuracy and work efficiency of the characterization results are higher than those of conventional methods.

[0012] In order to solve the above problems existing in the prior art, the present invention is implemented through the following technical solutions.

[0013] The present invention provides a method for multi-dimensional identification and characterization of multi-stage superimposed sand bodies in fluvial facies, the method comprising the following steps:

[0014] S1, synthetic log well seismic calibration step, obtaining drilling data, logging data and seismic data of the research area to perform synthetic log well seismic calibration;

[0015] S2, the three-dimensional isochronous model establishment step, performs fault interpretation on the three-dimensional seismic data, and establishes a seismic sequence stratigraphic framework model, i.e., a three-dimensional isochronous model, with the help of the interpreted faults; and automatically completes the horizon tracking interpretation;

[0016] S3, a step of determining the channel stage, establishing a certain number of stratum slices through the obtained three-dimensional isochronous model, assigning seismic attributes to the stratum slices, observing the channel distribution characteristics through the combination analysis of plane sections, and determining the channel stage;

[0017] S4, the main body identification step of the river channel sand body, by matching pursuit frequency division processing on the original seismic data, the three frequency band data obtained after the frequency division processing are RGB fused; finally, the river channel contour is determined by the attribute fusion representing the river channel boundary;

[0018] S5, a river channel fine carving step, combining the three-dimensional isochronous model obtained in step S3 and the river channel contour obtained in step S4 to perform fine carving on the river channel;

[0019] S6. Carry out a combined plan and section analysis on the river channel carving results of step S5, split the superimposed river channels into single channels, and determine the early and late stages of the single channels.

[0020] Further preferably, in the synthetic record well seismic calibration step, acoustic wave synthetic record calibration is adopted, and the acoustic wave synthetic record calibration is performed according to set principles; the set principles include extracting seismic traces along the trajectory, estimating sub-waves from well side traces, giving priority to strong phases, and jointly comparing and calibrating the main survey line and the connecting line.

[0021] More preferably, in the synthetic log seismic calibration step, before the synthetic log seismic calibration, the logging curve is optimized, and the optimization includes manual editing, filtering or correction; wherein the manual editing refers to point-by-point correction of abnormal values ​​in the curve, adding a constant value to the amplitude of a certain depth section or the entire well section of the curve, and multiplying the amplitude of a given depth section by the constant value;

[0022] The filtering process refers to eliminating singular points in the curve by using median filtering and bandpass filtering to achieve the purpose of smoothing the curve;

[0023] The correction process refers to calculating the error using the seismic logging data and the overall sonic logging data, and then obtaining the error curve, and then correcting the original sonic logging curve with the error curve.

[0024] Further preferably, in step S2, the PaleoScan software is used to perform fault interpretation on the three-dimensional seismic data.

[0025] Further preferably, in step S2, global model-controlled structural modeling technology is used to interpret the faults of the seismic data in the research area; in the global model-controlled framework, automatic identification and manual judgment are combined to jointly interpret each fault from three directions to form a fault plane that is highly consistent with the directions, and to implement each fault; finally, the interpreted faults are used to improve the global model, so that the finely interpreted faults participate in the establishment of the three-dimensional isochronal model.

[0026] More preferably, in step S2, the specific process of establishing the three-dimensional isochronous model is as follows:

[0027] On the basis of fault interpretation, based on the cost function of seismic data similarity and geological consistency, the idea based on the global optimal method is adopted to spatially deconstruct the seismic data body, select the optimal structure from a large number of possible sequence stratigraphic frameworks, and directly establish a seismic sequence stratigraphic framework model, namely a three-dimensional isochronal model, from the seismic data to automatically complete the stratigraphic tracking interpretation.

[0028] Further preferably, in step S3, the set seismic attribute includes a root mean square amplitude attribute.

[0029] Further preferably, in step S4, matching pursuit frequency division processing is performed on the original seismic data, specifically, firstly, the complex seismic trace of the original seismic trace is calculated, and the instantaneous attribute at the maximum value of the instantaneous amplitude envelope is obtained; then, a small-range atom library is established with the instantaneous attribute as the center; then, the matching pursuit algorithm is used to find the best matching atom from the atom library, and it is continuously iterated until a given threshold is reached; finally, low-frequency, medium-frequency, and high-frequency band data are obtained based on the matching pursuit frequency division processing to highlight the river channel.

[0030] More preferably, the instantaneous properties include displacement, instantaneous frequency and instantaneous phase.

[0031] More preferably, in step S4, the three frequency band data obtained after the frequency division process are subjected to RGB fusion, specifically, the three frequency band data of low frequency, medium frequency and high frequency obtained based on the matching pursuit frequency division process are respectively assigned RGB colors, and then the energy attribute bodies of the three frequency bands that do not overlap with each other are mixed and displayed in RGB mode to form an RGB mixed frequency data body with passband information; specifically,

[0032] First, each energy attribute body is normalized so that each energy attribute body has a completely consistent dynamic range of change; then the three energy attribute bodies are mixed with RGB colors, and each sampling point obtains a composite color, which is related to the proportion of each energy attribute body at this point.

[0033] More preferably, the attributes characterizing the river channel boundary include amplitude attributes, coherence volume attributes and fracture detection attributes.

[0034] Further preferably, in step S5, the step of fine carving of the river channel specifically refers to determining the contour of the river channel on the plane through the judgment result of step S4, and on this basis, controlling the longitudinal thickness of the river channel through a three-dimensional isochronal model to carry out fine carving of the thickness distribution characteristics of the river channel sand.

[0035] Compared with the prior art, the beneficial technical effects brought by the present invention are as follows:

[0036] 1. The most similar method in the prior art is a sand body carving method that combines conventional attributes with well control. The advantage of this method is that the channel boundary and the top and bottom range of the sand body can be determined on the automatically tracked high-density isochronous surface during the identification process. The isochronism, accuracy and work efficiency of the carving results are higher than those of conventional methods.

[0037] 2. In the global model control framework, the present invention combines automatic identification and manual discrimination, and interprets each fault from three directions to form a fault plane that is highly consistent with the direction, ensuring the rigor and efficiency of the fault interpretation work and implementing each fault. Finally, the interpreted faults are used to improve the global model, and the fine interpretation of the faults is used to participate in the establishment of a three-dimensional isochronal model, which helps to better identify and determine the river sand from both the plane and the section.

[0038] 3. Based on the fault interpretation, the present invention uses the idea of ​​global optimization methods to spatially deconstruct the seismic data body based on the cost function of seismic data similarity and geological consistency, selects the optimal structure from a large number of possible sequence stratigraphic frameworks, and directly establishes a seismic sequence stratigraphic framework model, that is, a three-dimensional isochronal model, from the seismic data; and automatically completes the horizon tracking interpretation. The horizons extracted based on the seismic sequence stratigraphic framework model have improved accuracy compared to the horizons obtained by traditional interpretation. At the same time, the cross-axis, jump point, and non-closure phenomena that exist in the traditional horizon interpretation results are basically non-existent in the extraction results of the new method. The seismic sequence stratigraphic framework model accurately reflects the sedimentary characteristics of the river channel of the stratum, and can more accurately identify and determine the river channel stage. The arbitrary subdivided layer layers extracted from the seismic sequence stratigraphic framework model match well with the seismic reflection characteristics, providing a solid foundation for the subsequent extraction of along-layer and inter-layer attributes.

[0039] 4. In order to meet the needs of river channel sand characterization in different regions, after investigation, we use the matching pursuit method (MPD) for spectral decomposition. This method has a very high time-frequency resolution. The wavelet decomposition method does not perform time-frequency analysis through Fourier transform. Therefore, this method has a very high time-frequency resolution and can obtain a high-resolution single-frequency body. Conventional spectral decomposition methods are mainly based on short-time Fourier transform and wavelet transform. Both methods are based on Fourier transform, and the time-frequency resolution is not high, while the wavelet decomposition method based on matching pursuit can solve this problem well.

[0040] 5. The present invention uses the instantaneous attribute information of the complex seismic trace to reduce the atom library, thereby reducing the time used to search for the best atom and greatly improving the calculation efficiency. The specific method can be divided into three steps as a whole: first, calculate the complex seismic trace of the original seismic trace and obtain the instantaneous attribute at the maximum value of the instantaneous amplitude envelope; then use the instantaneous attributes such as displacement, instantaneous frequency, instantaneous phase of the point as the center to establish a small-scale atom library; the third step is to use the matching pursuit algorithm to find the best matching atom from the atom library, and iterate continuously until the given threshold is reached. The advantage of this method is that it uses the instantaneous attributes of the seismic signal as prior information, which greatly reduces the number of atoms in the atom library, saves the time of searching for atoms, and also makes the algorithm more accurate, making the result have practical significance. For the comparison of the frequency division volume profiles of the conventional frequency division algorithm and the MPD frequency division algorithm for data from different regions, the longitudinal resolution of the latter is significantly improved than that of the former. Finally, based on the MPD algorithm, low frequency, medium frequency and high frequency are obtained to highlight the river channel.

[0041] 6. The present invention introduces the RGB mixed mode into the seismic frequency division attribute analysis, and mixes and displays the energy attribute bodies of the three frequency bands that do not overlap each other in the RGB mode to form an RGB mixed data body with frequency information. In practical applications, the low-frequency band energy attribute is selected as red (R), the medium-frequency band energy attribute is selected as green (G), and the high-frequency band energy attribute is selected as blue (B). First, the energy of each attribute body is normalized so that the energy of each body has a completely consistent dynamic range, that is, it is standardized to a color intensity value within the range of 0 to 255. Among them, 0 is the smallest attribute value, which appears as black; 255 is the largest attribute value, which appears as a pure color, such as red. Then these three attribute bodies are mixed by RGB color, and each sampling point obtains a composite color, and the composite color is related to the proportion of each body at the point. If all three are pure colors at this point (that is, the energy attribute values ​​are all maximum values), the point will appear white after mixing; if the low-frequency band R and the medium-frequency band G are all pure colors (that is, the energy attribute values ​​are all maximum values), and the high-frequency band is black (the energy attribute value is the smallest), the point will appear yellow after mixing. This RGB mixed display method has a good effect on highlighting the energy-approximate characteristic areas in each frequency division attribute, which can highlight the commonality and weaken the difference. For geological body identification, if the different frequency band attribute bodies show strong energy characteristics after frequency division, the geological body will be clearly presented as a near-white feature on the color data body after RGB mode mixing, which is obviously different from the surrounding rock; if the geological body also shows weak characteristics on the energy attribute body after frequency division, it is difficult to distinguish its distribution characteristics with a single frequency energy attribute body, but after using multiple frequency division energy attribute RGB modes for mixing, this type of reservoir will show obvious differences from the surrounding rock on the mixed body, which highlights the reservoir instead.

[0042] 7. The present invention adopts the method of primary color fusion of amplitude attributes, coherence body attributes, and fracture detection attributes for analysis. The combination of coherence body attributes and fracture detection attributes can help us identify the river channel contour more clearly, while the amplitude attribute can determine whether there is a sand body response in these existing "contours" - that is, whether it is a bright spot reflection, as well as the thickness of the sand body and the increase or decrease of physical properties. In areas where there are bright spot reflections but the coherence body attributes and fracture detection attributes are weak, we can combine the frequency division fusion slices obtained in the previous step to identify the river channel contour.

[0043] 8. The present invention has the following advantages in controlling the thickness of sand bodies based on the three-dimensional isochronal model: ① It makes up for the limitation of seismic resolution in identifying thin sand layers; ② It makes full use of well data to control the quality of prediction results, and determines the isochronal surfaces of the top and bottom of a single-period river channel with multiple wells; ③ It uses isochronal surfaces to cut the sand bodies vertically, following the occurrence of the sequence where the river channel is located, and avoiding time penetration; at the same time, when the sand body thickness is less than one-eighth of the wavelength and the top and bottom of the sand body cannot be accurately determined by the seismic waveform, determining the top and bottom of the sand body based on the combination of wells and seismic under the high-density sequence model is also a further supplement to the limit of thin sand layers predicted by seismic prediction.

[0044] 9. The present invention combines the results of river channel carving with planar section analysis, splits the superimposed river channels into single channels, and determines the early and late stages of the single channels. Through the seismic identification of the dominant offset and the layer-by-layer analysis combined with high-density isochronous slices, the development sequence of each single channel of the sand group is sorted out by combining the vertical and horizontal directions and the well-seismic combination. The sequence of river channel deposition in the study area can be clearly seen in the high-density isochronous slices. Through the channel morphology, trend, source direction, energy changes, and early and late appearance order from bottom to top, it can not only distinguish between different periods of river channels, but also has a good ability to distinguish different channel monomers of the same period of river channels. The same period of river channels are divided into two situations: independent single channel channels and multiple channel superpositions. The early and late order of the single channel can be determined according to the early and late order of its appearance and the direction of the source water system. For the situation of multiple overlapping river channels, further determination is required through profiles. By pulling a river channel profile perpendicular to the direction of the overlapping river channel area, the position of the bright spots in the river channel can be observed on the profile. According to its position, the order of the two overlapping river channels can be further determined. For the situation where two river channels have multiple intersection areas, river channel profiles can be pulled from multiple areas for supporting evidence and judgment. By pulling profiles from all the overlapping river channel areas, the river channel monomers can be classified and sorted according to their early and late orders. Through the fine carving of single river channels, the longitudinal evolution characteristics of each river channel can be clarified. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] Figure 1 It is a flow chart of the multi-dimensional identification and characterization method of multi-stage superimposed sand bodies of fluvial facies of the present invention;

[0046] Figure 2 It is a flow chart of the global model-controlled structure interpretation and slice extraction of the present invention;

[0047] Figure 3 A schematic diagram of the present invention for combining plane and section to identify and divide river channel stages;

[0048] Figure 4 This is the RGB fusion display of the sand body of the present invention;

[0049] Figure 5 This is a sand body multi-attribute fusion display diagram of the present invention;

[0050] Figure 6 This is a schematic diagram of determining the top and bottom of a sand body by combining wellbore seismic analysis under the three-dimensional isochronal model of the present invention;

[0051] Figure 7 This is a three-dimensional display diagram of river sand of the present invention;

[0052] Figure 8 This is a cross-section of the overlapping area of ​​the river crossing channel of the present invention. DETAILED DESCRIPTION

[0053] The following is an exemplary embodiment of the present invention that helps to fully understand the claims and their equivalents in conjunction with the accompanying drawings, wherein the specific details will be considered as exemplary only, and do not limit the scope of the present invention. Therefore, those of ordinary skill in the art can make various changes and modifications to the embodiments without departing from the scope and spirit of the present invention.

[0054] Example 1

[0055] As a preferred embodiment of the present invention, this embodiment discloses a method for multi-dimensional identification and characterization of multi-stage superimposed sand bodies of fluvial facies, the method comprising the following steps:

[0056] S1, synthetic log well seismic calibration step, obtaining drilling data, logging data and seismic data of the research area to perform synthetic log well seismic calibration;

[0057] S2, the three-dimensional isochronous model establishment step, performs fault interpretation on the three-dimensional seismic data, and establishes a seismic sequence stratigraphic framework model, i.e., a three-dimensional isochronous model, with the help of the interpreted faults; and automatically completes the horizon tracking interpretation;

[0058] S3, a step of determining the channel stage, establishing a certain number of stratum slices through the obtained three-dimensional isochronous model, assigning seismic attributes to the stratum slices, observing the channel distribution characteristics through the combination analysis of plane sections, and determining the channel stage;

[0059] S4, the main body identification step of the river channel sand body, by matching pursuit frequency division processing on the original seismic data, the three frequency band data obtained after the frequency division processing are RGB fused; finally, the river channel contour is determined by the attribute fusion representing the river channel boundary;

[0060] S5, a river channel fine carving step, combining the three-dimensional isochronous model obtained in step S3 and the river channel contour obtained in step S4 to perform fine carving on the river channel;

[0061] S6. Carry out a combined plan and section analysis on the river channel carving results of step S5, split the superimposed river channels into single channels, and determine the early and late stages of the single channels.

[0062] The closest existing method is the sand body carving method that combines conventional attributes with well control. The advantage of this method is that it determines the channel boundary and the top and bottom range of the sand body on the automatically tracked high-density isochronous surface during the identification process. The isochronism, accuracy and work efficiency of the carving results are higher than those of conventional methods.

[0063] Example 2

[0064] As another preferred embodiment of the present invention, this embodiment further elaborates and supplements the technical solution of the present invention on the basis of the above-mentioned embodiment 1. In this embodiment, in the synthetic record well seismic calibration step, acoustic wave synthetic record calibration is adopted, and the acoustic wave synthetic record calibration is performed according to the set principles; the set principles include extracting seismic traces along the trajectory, estimating wavelets from well side traces, giving priority to strong phases, and jointly comparing and calibrating the main survey line and the tie line.

[0065] Furthermore, in the synthetic record seismic calibration step, before the synthetic record seismic calibration, the logging curve is optimized, and the optimization includes manual editing, filtering or correction; wherein the manual editing refers to point-by-point correction of abnormal values ​​in the curve, adding a constant value to the amplitude of a certain depth segment or the entire well segment of the curve, and multiplying the amplitude of the given depth segment by the constant value; the filtering refers to the use of median filtering and bandpass filtering to eliminate singular points in the curve to achieve a smooth curve; the correction refers to using seismic logging data and overall acoustic logging data to calculate errors, and after obtaining the error curve, the original acoustic logging curve is corrected by the error curve.

[0066] Example 3

[0067] As another preferred embodiment of the present invention, this embodiment further elaborates and supplements the technical solution of the present invention on the basis of the above-mentioned embodiment 1 or embodiment 2. In this embodiment, in step S2, the PaleoScan software is used to perform fault interpretation on the three-dimensional seismic data.

[0068] Furthermore, in step S2, the global model-controlled structural modeling technology is used to interpret the faults of the seismic data in the study area; in the global model-controlled framework, automatic identification and manual judgment are combined to jointly interpret each fault from three directions to form a fault plane that is highly consistent with the direction, and to implement each fault; finally, the interpreted faults are used to improve the global model, so that the finely interpreted faults can participate in the establishment of the three-dimensional isochronal model.

[0069] The specific process of establishing the three-dimensional isochronous model is as follows:

[0070] On the basis of fault interpretation, based on the cost function of seismic data similarity and geological consistency, the idea based on the global optimal method is adopted to spatially deconstruct the seismic data body, select the optimal structure from a large number of possible sequence stratigraphic frameworks, and directly establish a seismic sequence stratigraphic framework model, namely a three-dimensional isochronal model, from the seismic data to automatically complete the stratigraphic tracking interpretation.

[0071] Example 4

[0072] As another preferred embodiment of the present invention, this embodiment is a further detailed elaboration and supplement to the technical solution of the present invention on the basis of the above-mentioned embodiment 1, embodiment 2 or embodiment 3. In this embodiment, in step S3, the set seismic attributes include root mean square amplitude attributes. In step S4, by performing matching pursuit frequency division processing on the original seismic data, specifically, firstly, the complex seismic trace of the original seismic trace is calculated, and the instantaneous attribute at the maximum value of the instantaneous amplitude envelope is obtained; then, using the instantaneous attribute as the center, a small-scale atom library is established; then, the matching pursuit algorithm is used to find the best matching atom from the atom library, and it is continuously iterated until a given threshold is reached; finally, based on the matching pursuit frequency division processing, low-frequency, medium-frequency, and high-frequency three-band data are obtained for highlighting the river channel. The instantaneous attributes include displacement, instantaneous frequency, and instantaneous phase.

[0073] In step S4, the three frequency band data obtained after the frequency division process are subjected to RGB fusion, specifically, the three frequency band data of low frequency, medium frequency and high frequency obtained by the matching pursuit frequency division process are respectively assigned RGB colors, and then the energy attribute bodies of the three frequency bands that do not overlap with each other are mixed and displayed in RGB mode to form an RGB mixed frequency data body with passband information; specifically,

[0074] First, each energy attribute body is normalized so that each energy attribute body has a completely consistent dynamic range; then the three energy attribute bodies are mixed by RGB color, and each sampling point obtains a composite color, which is related to the proportion of each energy attribute body at this point. The attributes characterizing the river boundary include amplitude attributes, coherence body attributes and fracture detection attributes.

[0075] Further preferably, in step S5, the step of fine carving of the river channel specifically refers to determining the contour of the river channel on the plane through the judgment result of step S4, and on this basis, controlling the longitudinal thickness of the river channel through a three-dimensional isochronal model to carry out fine carving of the thickness distribution characteristics of the river channel sand.

[0076] Example 5

[0077] As another preferred embodiment of the present invention, this embodiment is based on the above-mentioned embodiment 1, embodiment 2, embodiment 3 or embodiment 4, and provides a detailed description of the specific implementation steps of a multi-dimensional identification and characterization method of multi-stage superimposed sand bodies in river phases of the present invention.

[0078] The first step is the calibration of synthetic record well seismic

[0079] Synthetic record calibration is the basic work of seismic interpretation. It is to organically combine drilling data, logging data and seismic data to establish a bridge for the corresponding relationship between lithology, electrical properties, physical properties and wellside seismic traces. The main methods of layer calibration are VSP calibration, acoustic synthetic seismic record calibration and comprehensive velocity calibration. Among them, acoustic synthetic seismic record calibration is one of the most commonly used means of layer calibration in the process of seismic interpretation. In the calibration process, make full use of the full-well logging curve to calibrate the main marker layer of the whole well section. If there is a density curve in the drilling, use the density curve. Follow the following principles of acoustic synthetic record calibration: extract seismic traces along the trajectory as much as possible; estimate sub-waves of wellside traces; give priority to strong phases; and calibrate the main survey line and the connecting line by joint comparison.

[0080] On the basis of the optimization of logging curves, the well positions in the work area are finely calibrated, which is helpful for the establishment of sequence interfaces and subsequent river channel identification and carving. The optimization process includes manual editing, filtering or correction processing; wherein, the manual editing refers to point-by-point correction of abnormal values ​​in the curve, adding a constant value to the amplitude of a certain depth section or the entire well section of the curve, and multiplying the amplitude of a given depth section by a constant value; the filtering process refers to the use of median filtering and bandpass filtering to eliminate singular points in the curve to achieve the purpose of smoothing the curve; the correction process refers to the use of seismic logging data and overall acoustic logging data to calculate errors, and after obtaining the error curve, the original acoustic logging curve is corrected by this error curve.

[0081] Step 2: 3D isochronous model building steps

[0082] Globally controlled structural modeling builds a seismic sequence stratigraphic framework model with the help of interpreted faults, automatically completes the horizon tracking interpretation, and completes the establishment of a three-dimensional isochronal model.

[0083] In the global model control framework, the automatic identification and manual discrimination are integrated to jointly interpret each fault from three directions, forming a fault plane that is highly consistent with the direction, ensuring the rigor and efficiency of the fault interpretation work and implementing each fault. Finally, the interpreted faults are used to improve the global model, and the fine interpretation of the faults is used to participate in the establishment of a three-dimensional isochronal model, which helps to better identify and determine the river sand from both the plane and the section.

[0084] On the basis of fault interpretation, based on the cost function of seismic data similarity and geological consistency, the idea of ​​global optimization methods is used to spatially deconstruct the seismic data body, select the optimal structure from a large number of possible stratigraphic frameworks, and directly establish a seismic stratigraphic framework model from the seismic data to automatically complete the horizon tracking interpretation. The horizons extracted based on the seismic stratigraphic framework model are more accurate than those obtained by traditional interpretation. At the same time, the cross-axis, jump point, and non-closure phenomena that exist in the traditional horizon interpretation results are basically non-existent in the results extracted by the new method.

[0085] The seismic sequence stratigraphic framework model accurately reflects the sedimentary characteristics of the river channel in the stratum and can more accurately identify and determine the river channel stage. The arbitrary subdivided layer positions extracted from the seismic sequence stratigraphic framework model match well with the seismic reflection characteristics, providing a solid foundation for the subsequent extraction of along-layer and inter-layer attributes.

[0086] Step 3: Determine the river channel period

[0087] Through the obtained 3D isochronous model, a certain number of layer slices (such as Figure 2 ), assign various attributes to the slices (taking the root mean square amplitude as an example), and through the combination of planimetric and sectional methods, the channel distribution characteristics can be better observed and the channel stage can be determined. (e.g. Figure 3 ).

[0088] Step 4: Identification of the main body of the river sand body

[0089] After determining the corresponding characteristics of the river channel, the forward modeling response of the special sand body and the fine division of the sand body pattern, it is necessary to carve the sand body within the threshold range. The specific workflow and technical principles are as follows:

[0090] 1) Matching pursuit frequency division processing:

[0091] Due to the wavelet superposition effect and ringing effect generated by multiple wave impedance interfaces, seismic traces cannot directly reflect the characteristics of interface positions and stratum combinations, and different geological targets have different sensitivities to different frequency components of seismic data. Therefore, using the specific frequency and frequency band information of seismic signals to highlight the imaging effect of seismic targets has always been an important part of geophysical exploration research.

[0092] Spectral decomposition is to decompose a seismic trace into a collection of sub-waves of different shapes (main frequencies) and amplitudes. It is an effective technology to understand geological bodies by using the different response characteristics of different frequencies to geological bodies of different scales, thereby performing seismic interpretation.

[0093] In order to meet the needs of river channel sand characterization in different regions, after investigation, we use the matching pursuit method (MPD) for spectral decomposition. This method has a high time-frequency resolution. The wavelet decomposition method does not perform time-frequency analysis through Fourier transform, so this method has a high time-frequency resolution and can obtain a high-resolution single-frequency body. Conventional spectral decomposition methods are mainly based on short-time Fourier transform and wavelet transform. Both methods are based on Fourier transform, and the time-frequency resolution is not high. The wavelet decomposition method based on matching pursuit can solve this problem well.

[0094] The basic idea of ​​the MPD (Matching Pursuit, MP) matching pursuit algorithm is to project the original signal onto a series of orthogonal atoms and represent the original signal as a linear combination of these atoms. These atoms form an atom library. When this atom library is overcomplete, these atoms can be used to accurately express the original signal. The matching pursuit algorithm is actually a sparse decomposition of the signal.

[0095] The MPD matching pursuit algorithm is an iterative process that can accurately decompose and reconstruct the signal. However, this algorithm does not consider the overall signal, but only the local best match of the signal. This algorithm is somewhat "greedy". Therefore, in the first few iterations, the remaining residual will converge quickly, but as the number of iterations increases, the convergence speed of the error will become slower and slower. Moreover, if the entire overcomplete atomic library is searched once in each iteration, the amount of calculation is very huge.

[0096] We use the instantaneous attribute information of the complex seismic trace to narrow the atom library, thereby reducing the time used to search for the best atom and greatly improving the computational efficiency. The specific approach can be divided into three steps as a whole: first, calculate the complex seismic trace of the original seismic trace and obtain the instantaneous attribute at the maximum value of the instantaneous amplitude envelope; then use the instantaneous attributes of the point such as displacement, instantaneous frequency, instantaneous phase as the center to establish a small-scale atom library; the third step is to use the matching pursuit algorithm to find the best matching atom from the atom library, and iterate continuously until the given threshold is reached. The advantage of this method is that it uses the instantaneous attributes of the seismic signal as prior information, which greatly reduces the number of atoms in the atom library, saves the time of searching atoms, and also makes the algorithm more accurate and the results have practical significance.

[0097] Comparison of the frequency-division volume profiles of the conventional frequency-division algorithm and the MPD frequency-division algorithm for data from different regions shows that the longitudinal resolution of the latter is significantly higher than that of the former. Finally, based on the MPD algorithm, low frequency, medium frequency, and high frequency are obtained to highlight the river channel.

[0098] 2) RGB fusion highlights the river boundary

[0099] The multi-attribute fusion display technology is to mix and display the above-mentioned low-frequency band, medium-frequency band, and high-frequency band energy attribute bodies in RGB mode to form a color data body. RGB is the most commonly used color mode in the industry, where R represents red, G represents green, and B represents blue. Red, green, and blue are the three most basic colors. The rich and colorful colors in nature are all obtained by superimposing and mixing these three primary colors.

[0100] RGB images use only three colors, and can mix them in different proportions to reproduce up to 160,000 colors on the screen. In RGB mode, each RGB component can use a value from 0 (black) to 255 (white). When all three component values ​​are equal, a gray shade is produced. When the value of all components is 255, the result is pure white; when the value is 0, the result is pure black.

[0101] Similarly, the RGB mixed mode is introduced into the seismic frequency division attribute analysis, and the energy attribute bodies of the three non-overlapping frequency bands are mixed and displayed in RGB mode to form an RGB mixed frequency data body with frequency information. In practical applications, the low-frequency band energy attribute is selected as red (R), the medium-frequency band energy attribute is selected as green (G), and the high-frequency band energy attribute is selected as blue (B). First, the energy of each attribute body is normalized so that the energy of each body has a completely consistent dynamic range, that is, it is standardized to a color intensity value in the range of 0 to 255. Among them, 0 is the smallest attribute value, which is presented as black; 255 is the largest attribute value, which is presented as a pure color, such as red. Then these three attribute bodies are mixed by RGB color, and each sampling point obtains a composite color, which is related to the proportion of each body at that point. If all three are pure colors at that point (that is, the energy attribute values ​​are all maximum), the point will appear white after mixing; if the low-frequency band R and the medium-frequency band G are both pure colors (that is, the energy attribute values ​​are all maximum), and the high-frequency band is black (the energy attribute value is the minimum), the point will appear yellow after mixing.

[0102] This RGB mixed display method has a good effect on highlighting the energy-approximate characteristic areas in each frequency division attribute, which can highlight the commonality and weaken the difference. For geological body identification, if the different frequency band attribute bodies show strong energy characteristics after frequency division, the geological body will be clearly presented as a near-white feature on the color data body after RGB mode mixing, which is obviously different from the surrounding rock; if the geological body also shows weak characteristics on the energy attribute body after frequency division, it is difficult to distinguish its distribution characteristics with a single frequency energy attribute body, but after using multiple frequency division energy attribute RGB modes for mixing, this type of reservoir will show obvious differences from the surrounding rock on the mixed body, which highlights the reservoir instead.

[0103] 3) Comprehensive identification of river contours using multi-attribute fusion technology

[0104] The traditional method of characterizing sandstone reservoirs is to use rock physics analysis and select the attribute with the strongest reservoir sorting ability for reservoir prediction, while other attributes are difficult to fully play their role; this method has a relatively simple attribute selection and only one threshold value. However, fine stratigraphic slices extracted based on stratigraphic models, combined with seismic attribute extraction, can finely carve reservoirs and predict sandstone reservoirs. The multi-attribute fusion (three-parameter) reservoir prediction and carving technology method can simultaneously use three reservoir-sensitive attributes and inversion results to characterize the reservoir; eliminate non-reservoir layers from different angles to reduce reservoir multi-solution; and is a significant improvement over the previous method of using a single attribute and a single threshold value to calibrate sandstone reservoir distribution prediction.

[0105] In most cases, the river channel outline can be determined only through the frequency division fusion display, but due to factors such as the vertical resolution of the slice and seismic sampling, the river channel boundary often has afterimages above and below the actual location. This afterimage will cause us to deviate from the river channel carving and misjudge the correct location and thickness of the river channel. In order to deal with such situations, we use multi-attribute fusion technology to conduct comprehensive analysis from multiple angles and finally determine the accurate river channel outline and location.

[0106] The analysis is performed by fusing the amplitude attribute, coherence body attribute, and fracture detection attribute in primary colors. The combination of coherence body attribute and fracture detection attribute can help us identify the river channel contour more clearly, while the amplitude attribute can determine whether there is a sand body response in these existing "contours" - that is, whether it is a bright spot reflection, and the thickness of the sand body and the increase or decrease of physical properties. In areas where there are bright spot reflections but the coherence body attribute and fracture detection attribute are weak, we can combine the frequency division fusion slices obtained in the previous step to identify the river channel contour.

[0107] Step 5: Detailed river carving

[0108] The river channel is finely carved through the sand body thickness control technology of high-density isochronous sequence stratigraphic model (3D isochronous model).

[0109] 1) Response of pre-stack data to sand bodies with different impedances

[0110] In the pre-stack data volumes with different offset distances, high-impedance and low-impedance sand bodies have different energy intensities, that is, the high-impedance sand body has the strongest energy in the near track, showing continuous strong bright spots, has stronger energy in the middle track, and has the weakest energy in the far track, and its shape is discontinuous.

[0111] Therefore, we can draw a simple conclusion: the stacks of different offset distances (near, medium, and far) - low-impedance sand bodies respond most strongly in far-offset stacks; high-impedance sand bodies respond most strongly in near-offset stacks. Therefore, we use far-offset stacks to identify and carve low-impedance channels, and near-offset stacks to identify and carve high-impedance channels.

[0112] 2) Fine carving of river channels

[0113] With the continuous deepening of oil and gas exploration and development, the development target is no longer structural oil and gas reservoirs dominated by thick reservoirs, but more complex oil and gas reservoirs with thinner reservoir thickness (single sand body thickness is mostly less than 10 meters).

[0114] Generally speaking, for sandstones thicker than 10 meters, the thicker the sandstone, the stronger the earthquake amplitude. When the thickness exceeds 20 meters, the amplitude gradually stabilizes. This type of sandstone is easy to identify, with obvious characteristics of strong peaks at the top and strong troughs at the bottom. The forward modeling results are consistent with this view. For sandstones less than 10 meters thick, the ability of earthquakes to identify thin sand layers depends on the tuning relationship between the seismic frequency and the thickness of the sand body, that is, H = (V / F) / 4 = 4450 / F / 4 = 26 to 10.6 meters. Based on this, it is necessary to increase the frequency within the effective frequency band, improve the resolution of the original seismic data, and carry out technical research on the identification of sand bodies below 10 meters.

[0115] Reservoirs with a thickness greater than 1 / 4 wavelength have a tuning effect, and the impedance inversion and pseudo-acoustic inversion combined with logging data can meet the needs of reservoir identification. Affected by the current seismic resolution, when the main frequency of the seismic wavelet is around 30hz, it is difficult to distinguish the top and bottom seismic reflections of thin sand bodies with a thickness less than 1 / 4 wavelength (less than 20 meters). However, the thickness of the river channel sand bodies in the current exploration and development layers is often less than 1 / 4 wavelength, and the sedimentary phase changes rapidly laterally. The identification and prediction of river channel thin sand bodies are becoming increasingly prominent.

[0116] Through the previous steps of the workflow, we determined the channel contour on the plane. On this basis, we controlled the longitudinal thickness of the channel through a high-density isochronal sequence model and carried out detailed carving of the channel sand thickness distribution characteristics.

[0117] The fine carving of river channel sand is a sand body thickness control technology based on a high-density sequence stratigraphic model, which is a technology unique to the PaleoScan software. This method forms an isochronal surface with extremely high vertical density on the global geological model body, combines well and seismic data, and refers to the top and bottom of the measured sand body and the predicted sand body characteristics on the single-period river channel to determine the top and bottom isochronal surfaces and extract the predicted attribute body within the time window. Finally, the spatial distribution of the single-period river channel is determined in combination with the horizontal pre-boundary identification range of the previous river channel. This method is applied to the sand body extraction of each period.

[0118] Sand body thickness control based on high-density sequence stratigraphic model has the following advantages:

[0119] ① It makes up for the limitation of seismic resolution in identifying thin sand layers;

[0120] ② Make full use of well data to control the quality of prediction results and use multiple wells to determine the top and bottom isochronal surfaces of a single-period river channel;

[0121] ③ Using isochronous planes to cut the sand body longitudinally follows the occurrence of the sequence where the river channel is located and avoids diachrony;

[0122] At the same time, when the sand body thickness is less than one-eighth of the wavelength and the top and bottom of the sand body cannot be accurately determined by the seismic waveform, determining the top and bottom of the sand body based on the combination of well and seismic data under the high-density sequence model is also a further supplement to the seismic prediction of the limit of thin sand layers.

[0123] The sixth step is to split the superimposed river channels into individual units by combining horizontal section and determine the early and late stages of the individual units.

[0124] Through seismic identification of advantageous offset distances, combined with layer-by-layer analysis of high-density isochronous slices, vertical and horizontal integration, and well-seismic integration, the development sequence of each individual river channel of the sand group is sorted out.

[0125] In high-density isochronal slices, the order of river channel deposition in the study area can be clearly seen. Through the channel morphology, trend, source direction, energy changes, and order of appearance from bottom to top, not only can the different periods of river channels be distinguished, but also the different channel monomers of the same period of river channels can be well resolved. The same period of river channels can be divided into two situations: independent single channel and multiple channel superposition. The order of single channel can be determined according to the order of their appearance and the direction of source water system.

[0126] For the case of multiple overlapping channels, it is necessary to further determine the cross section. By pulling the cross section perpendicular to the direction of the channel in the overlapping area, the position of the bright spot of the channel can be observed on the cross section. According to its position, the order of the two overlapping channels can be further determined ( Figure 4), if there are multiple crossing areas between two river channels, we can also pull river channel sections from multiple areas for supporting judgment. By pulling sections from all overlapping areas of river channels, we can classify and sort the river channel monomers by early and late times, and through the detailed carving of single river channels, we can clarify the longitudinal evolution characteristics of each river channel.

[0127] Example 6

[0128] As another preferred embodiment of the present invention, this embodiment takes river sand carving in a certain area of ​​Sichuan Basin as an example to illustrate the specific scheme of the present invention.

[0129] The first step is to calibrate the synthetic record well seismic data to prepare for the subsequent establishment of sequence interfaces and river channel identification and carving.

[0130] There are 57 wells of various ages in this area, 42 wells have been completed in the Jurassic system, and 15 wells have reached the Xujiahe Formation. Among them, 43 wells were drilled with the Da'anzhai oil layer as the target layer from 1985 to 2013, and 13 wells were drilled with the Xujiahe Formation as the target layer from 2008 to 2013. However, due to the long drilling age of some wells, the acoustic wave, density and other curves are seriously missing, and synthetic record calibration is impossible. Therefore, this round of research selected 14 wells for fine well-seismic calibration. From the calibration results, the bottom boundary of the Shaximiao Formation is a relatively continuous trough reflection, which can be traced and compared throughout the area. Therefore, it can be used as a marker layer for calibration and interpretation. This round of calibration uses new processing results, the sand body reflection is clear, the waveform fit is good, and the calibration results are accurate and reliable.

[0131] From the calibration, we can clearly know that the bottom of the first section of the Shahe Formation is the second weak trough reflection located above the strong trough of Lianggao Mountain. The bottom of the second section of the Shahe Formation, i.e. the top of the first section of the Shahe Formation, is marked by a set of interlaminar shale in the drilling. The top of the first section of the Shahe Formation in the 3D octagonal field near Longgang 3D is a set of continuous medium-amplitude trough reflections, which are significantly different from the seismic reflection characteristics of the surrounding rock, and have the same reflection characteristics in the Qiulin area. The top boundary of the Shaximiao Formation is the first discontinuous weak reflection trough under the continuous wide peak at the bottom of the Suining Formation.

[0132] At the same time, in the synthetic record, the synthetic record trace calculated using the corrected acoustic wave density curve is almost consistent with the seismic reflection characteristics of each marker layer, which indirectly proves the accuracy of the synthetic record production process.

[0133] After determining the specific position of the layer on the earthquake, the slices corresponding to the position of each marker layer in the seismic volume under the high-density isochronal model volume are extracted, that is, the determined stratigraphic framework is expanded from points to surfaces. Figure 4-6 It is a three-dimensional display of the bottom boundaries of each layer of the Shaximiao Formation, which, from bottom to top, are the bottom of the first section of the Shaxi Temple Formation, the bottoms of the first, second, third and fourth small layers of the second section of the Shaxi Temple Formation, and the top of the Shaxi Temple Formation.

[0134] Among them, the structure of the bottom boundary of the Shaxi Temple Formation is relatively flat, with a structural depth difference of less than 500m. The middle part is a strip-shaped uplift along the connecting line. The highest point of the structure is near the Qiu 1 and Qiu 8 wells, and the structural low position is near the Qiu 36 well. The middle part of the bottom boundary of the second section of Sha is similar to the bottom boundary of the first section of Sha. There is also a strip-shaped uplift along the connecting line. The highest point is in the southeast of the middle of the work area. The overall structure in the southeast of the work area is relatively high. At the top of Shaxi Temple, the strip-shaped uplift along the connecting line in the middle is no longer obvious. The overall work area is characterized by low structure on the northwest side and high structure on the southeast side; the overall structure is relatively flat, with a vertical height difference of 400m.

[0135] The second step is to establish a 3D isochronous model. PaleoScan software is used to interpret the faults of the 3D seismic data. With the help of the interpreted faults, a seismic sequence stratigraphic framework model is established, and the horizon tracking interpretation is automatically completed to complete the establishment of a 3D isochronous model.

[0136] This technology is the applicant's unique fault fracture interpretation technology, which is based on the fusion of multiple predicted fracture attributes (amplitude, phase, inclination) to comprehensively predict fractures. It can avoid the uncertainty and multi-solution of fracture prediction with a single attribute, improve the precision and accuracy of prediction, and can simultaneously detect large faults that control structures and small faults within the sequence. In this round, a total of 15 faults were interpreted in Shaximiao, including 12 faults that broke to the sand bottom (hydrocarbon source faults) and 3 faults that did not break to the sand bottom (non-hydrocarbon source faults).

[0137] In order to build an accurate geological model, seismic interpreters must have a clear understanding of the structural information in the seismic data, especially the fault information. Traditionally, this is achieved by manually picking faults on the seismic data, which is a tedious and time-consuming task. Therefore, great efforts have been made in the past two decades to improve the efficiency of fault interpretation and realize automatic fault detection. At present, seismic attributes such as coherence, variance and chaos have been used to capture the fault location, but the results are limited due to the influence of the inhomogeneity of the seismic signal near the fault on the attribute value. The seismic data in this area are of good quality and have strong lateral comparability. Therefore, the global model-controlled structural modeling technology is used to complete the interpretation of the seismic reflection interface of the Shaximiao Formation, the fine interpretation of faults, and the identification and carving of sand bodies.

[0138] Under the constraints of the top and bottom positions of the Shaximiao Formation, the PaleoScan software was used to adjust, establish and improve the seismic sequence framework model of the Shaximiao Formation. The seismic sequence stratigraphic framework model accurately reflects the continuous, stable and locally thickened sedimentary characteristics of the Shaximiao Formation. The extraction of any subdivided layer positions and seismic morphology from the seismic sequence stratigraphic framework model provides a good basis for the subsequent extraction of along-layer attributes.

[0139] The third step is to determine the river stage. Through the obtained three-dimensional isochronous model, a certain number of stratigraphic slices are established and various attributes of the slices are assigned (taking the root mean square amplitude as an example). By combining flat sectioning and sectioning, the river distribution characteristics can be better observed and the river stage can be determined.

[0140] The global modeling and interpretation technology of structural modeling must first determine the continuous and traceable seismic reflection interface for model building. In this area, there are four clearly characterized seismic reflection interfaces in the Shaximiao Formation, namely the bottom of the Shaximiao Formation (J2s1), the bottom of the first subsection of the second section of the Shaximiao Formation (J2s21), the bottom of the second subsection of the second section of the Shaximiao Formation (J2s22), and the top of the Shaximiao Formation (J3s). The seismic phases above and below the interface are significantly different. Within different sequences, the seismic reflection strength on the profile changes laterally, reflecting the development of the river channel.

[0141] The fourth step is the identification of the main body of the river sand body. The original seismic data is processed by matching pursuit frequency division, and the three frequency band data obtained after the frequency division processing are fused with RGB. Finally, the river contour is determined by the fusion of attributes representing the river boundary.

[0142] Under the global isochronous stratigraphic framework, seismic frequency division fusion technology and multi-dimensional three-parameter fusion technology are combined to comprehensively identify the river boundary, greatly improving the reliability of the lateral boundary of the river; six attributes are used at the same time. The six attributes are combined into two sets of plane fusion results, namely seismic frequency division fusion and multi-dimensional three-parameter fusion.

[0143] Due to the wavelet superposition effect and ringing effect generated by multiple wave impedance interfaces, seismic traces cannot directly reflect the characteristics of interface positions and stratigraphic combinations, and different geological targets have different sensitivities to different frequency components of seismic data. In order to meet the needs of channel sand characterization, the matching pursuit method (MPD) is used for spectral decomposition.

[0144] Spectral decomposition is to decompose a seismic trace into a collection of sub-waves of different shapes (main frequencies) and amplitudes, and to use different frequencies to understand the different response characteristics of geological bodies of different scales, so as to interpret the seismic body effectively. Figure 5 ).

[0145] In order to meet the needs of river channel sand characterization, after investigation, we use the matching pursuit method (MPD) for spectral decomposition. This method has a high time-frequency resolution. The wavelet decomposition method does not perform time-frequency analysis through Fourier transform, so this method has a high time-frequency resolution and can obtain a high-resolution single-frequency body. Conventional spectral decomposition methods are mainly based on short-time Fourier transform and wavelet transform. Both methods are based on Fourier transform, and the time-frequency resolution is not high. The wavelet decomposition method based on matching pursuit can solve this problem well.

[0146] The basic idea of ​​the MPD (Matching Pursuit, MP) matching pursuit algorithm is to project the original signal onto a series of orthogonal atoms and represent the original signal as a linear combination of these atoms. These atoms form an atom library. When this atom library is overcomplete, these atoms can be used to accurately express the original signal. The matching pursuit algorithm is actually a sparse decomposition of the signal.

[0147] The MPD matching pursuit algorithm is an iterative process that can accurately decompose and reconstruct the signal. However, this algorithm does not consider the overall signal, but only the local best match of the signal. This algorithm is somewhat "greedy". Therefore, in the first few iterations, the remaining residual will converge quickly, but as the number of iterations increases, the convergence speed of the error will become slower and slower. Moreover, if the entire overcomplete atomic library is searched once in each iteration, the amount of calculation is very huge.

[0148] We use the instantaneous attribute information of the complex seismic trace to narrow the atom library, thereby reducing the time used to search for the best atom and greatly improving the computational efficiency. The specific approach can be divided into three steps as a whole: first, calculate the complex seismic trace of the original seismic trace and obtain the instantaneous attribute at the maximum value of the instantaneous amplitude envelope; then use the instantaneous attributes of the point such as displacement, instantaneous frequency, instantaneous phase as the center to establish a small-scale atom library; the third step is to use the matching pursuit algorithm to find the best matching atom from the atom library, and iterate continuously until the given threshold is reached. The advantage of this method is that it uses the instantaneous attributes of the seismic signal as prior information, which greatly reduces the number of atoms in the atom library, saves the time of searching atoms, and also makes the algorithm more accurate and the results have practical significance.

[0149] Our method adopts Ricker wavelet set for wavelet decomposition.

[0150] Ricker sub-wave function: W R (t,f j )=(1-2π 2 f 2 j t 2 )exp(-π 2 t 2 f 2 j )

[0151] The spectrum of the wavelet:

[0152] Before wavelet decomposition, the seismic data needs to be Hilbert transformed.

[0153] The result after transformation is:

[0154] U(t)=u(t)+iu H (t)

[0155] The Hilbert value of the wavelet is: W(t,f j )=w(t,f j )+iw H (t,f j )

[0156] The wavelet decomposition is as follows:

[0157] Wavelet decomposition is to find the solution of Aj. The above equation is a large linear matrix and is solved by iterative method. The calculation process includes the following steps:

[0158] 1) Construct a wavelet set

[0159] D={d γ1 ,d γ2 ...,d γj ....,d γn}

[0160] 2) Set the maximum number of iterations Kstop, the minimum error δstop, and assign initial values.

[0161] R x 0 [n]=x[n],E x 0 =||R 0 x || 2

[0162] 3) Start iteration and calculate the inner product of Rn and the wavelet set:

[0163] a rj k =<R k x [n],d rj k >

[0164] 4) Calculate the new Rxk[n]

[0165] R x k [n] = R x k-1 [n]-a rj k d k rj [n]

[0166] Calculation error: E x k =||Rx k ||2

[0167] 5) If the number of iterations K < Kstop or Ex > δstop, return to (2) and continue; otherwise, jump out of the iteration and output the calculation result.

[0168] The above algorithm is slow in calculation speed. During the implementation process, the instantaneous attribute information of complex seismic traces is used to narrow the wavelet library, thereby reducing the time used to search for the best atom and greatly improving the calculation efficiency.

[0169] The RGB mixing mode is introduced into the seismic frequency division attribute analysis. The energy attribute volumes with non-overlapping three frequency bands are mixed and displayed in the RGB mode to form an RGB mixed frequency data volume with broadband information. In practical applications, the low-frequency band energy attribute is selected as red (R), the middle-frequency band energy attribute is selected as green (G), and the high-frequency band energy attribute is selected as blue (B). First, the energy of each attribute volume is normalized so that the energies of each volume have exactly the same dynamic range of change, that is, normalized to the color intensity values in the range of 0-255. Among them, 0 is the minimum attribute value and appears as black; 255 is the maximum attribute value and appears as a pure color, such as red. Then, these three attribute volumes are mixed in RGB colors. At this time, a synthesized color is obtained at each sampling point, and the synthesized color is related to the proportion of each volume at this point. For example, if all three are pure colors at this point (that is, the energy attribute values are all the maximum values), the point will appear white after mixing; if the low-frequency band R and the middle-frequency band G are both pure colors (that is, the energy attribute values are all the maximum values), and the high-frequency band is black (the energy attribute value is the minimum), the point will appear yellow after mixing.

[0170] This RGB mixing display method has a good effect on highlighting the energy approximate feature regions in each frequency division attribute, which can highlight the commonalities and weaken the differences. For geological body identification, if strong energy characteristics are shown on the attribute volumes of different frequency bands after frequency division, the geological body will clearly appear as a near-white feature on the color data volume after RGB mode mixing, with obvious differences from the surrounding rock; if the geological body also shows weak characteristics on the energy attribute volume after frequency division processing, it is very difficult to distinguish its distribution characteristics from a single frequency energy attribute volume. However, after mixing multiple frequency division energy attributes in the RGB mode, obvious differences will appear between this type of reservoir and the surrounding rock on the mixed body, which instead highlights the reservoir.

[0171] In this experiment, we used the original color fusion method of amplitude attributes, coherence body attributes, and fracture detection attributes in PaleoScan for analysis. The combination of coherence body attributes and fracture detection attributes can help us identify the river channel contour more clearly, while the amplitude attribute can determine whether there is a sand body response in these existing "contours" - that is, whether it is a bright spot reflection, and the thickness of the sand body and the increase or decrease of physical properties. In areas with bright spot reflections but weak coherence body attributes and fracture detection attributes, we can combine the frequency division fusion slices obtained in the previous step to identify the river channel contour.

[0172] The fifth step is the fine carving of the river channel, in which the river channel is finely carved through the sand body thickness control technology of the high-density isochronal sequence stratigraphic model.

[0173] The fine carving of the river channel sand in this embodiment is based on the sand body thickness control technology of the high-density sequence stratigraphic model, which is a technology unique to the PaleoScan software. This method is to form an isochronous surface with extremely high vertical density on the global geological model body, combine well and seismic data, refer to the measured top and bottom of the sand body and the predicted sand body characteristics on the single-period river channel, determine the top and bottom isochronous surfaces, and extract the predicted attribute body within the time window. Finally, the spatial distribution of the single-period river channel is determined in combination with the horizontal pre-boundary identification range of the previous river channel. This method is applied to the sand body extraction of each period.

[0174] The above method was used to carve the river channel, and finally 24 sets of sand layers were obtained. Four river channels were identified in the first section of the sandstone layer, and 20 river channels were identified in the second section of the sandstone layer. According to the previous division of sand body response patterns, we divided the identified river channels into high-impedance sandstone channels and low-impedance sandstone channels. Among them, the first section of the sandstone layer is all high-impedance sandstone, and the second section of the sandstone layer has 7 high-impedance river channels and 13 low-impedance river channels.

[0175] Based on the above results of fine carving of river sand, a comprehensive analysis of the obtained sand body was conducted. Overall:

[0176] 1. The river channels of the Shaximiao Formation are generally oriented from north to south and west to east, with a small number of them being oriented from east to west. The direction of the rivers reflects that the main source direction is from the northeast of the work area, which is consistent with the overall geological understanding of the area.

[0177] 2. The development patterns of the first and second sections of the Shaxi Temple Formation are different. The river channel of the first section is generally wider, and its shape is similar to that of a beach or fan, which is similar to the sedimentary characteristics of the delta front. The average width of the river channel of the second section gradually narrows from bottom to top, and the river channel area decreases from large to small. The overall river channel characteristics show a more obvious braided river morphology.

[0178] 3. From the perspective of sand body pattern, the lower part of the Shaximiao Formation, including the first and bottom of the second member of the Shaxi Formation, mainly develops high-impedance channels, while the middle and upper parts of the second member of the Shaxi Formation are mainly low-impedance channels.

[0179] 4. In terms of thickness, the river channels of the Shaximiao Formation show a trend of gradually decreasing thickness from bottom to top. The river channels of the first section of Sha are generally thicker. There are several large-scale river channels with thick thickness in the middle of the second section of Sha. The river channels gradually tend to be undeveloped near the top of the sand, and their number, area and thickness gradually decrease until they disappear.

[0180] 5. From the plane view of the river distribution in the whole area, it can be seen from the multi-stage river superimposed pipe system that the river development stages in the middle of the work area are the most, followed by the southern part, and there are fewer river development stages in the northern part of the work area.

[0181] The sixth step is to split the superimposed river channels into individual units by combining horizontal sectioning and determine the early and late stages of the individual units.

[0182] The superposition of river channel sand bodies is very important for the deployment of horizontal wells in the later stage of development. Recently, Qiulin 205 well and the horizontal section Qiulin 205-H1 were drilled in the No. 5 river channel sand body of Shaximiao Formation northeast of Qiulin 3 well in the northern part of Qiulin area. From the drilling situation, the well encountered two superimposed sand bodies (2349-2408m), with a total thickness of 59m. Among them, the top sandstone reservoir is good, with a thickness of 30.1m, of which the reservoir thickness is 21.1m. The sand body is thick and the reservoir is developed, which is the target of this horizontal well drilling. However, since the original seismic data is a full offset stacking, the resolution is not high, and the superimposed relationship of the sand bodies is difficult to distinguish. In order to more accurately depict the distribution of sand bodies and ensure that the horizontal section design trajectory can enter the target sand body and travel along the sand body, it is necessary to distinguish the superimposed river channels and find the plane position and burial depth of the target river channel.

[0183] In this study, the data gathers were divided into three offset distances: near, medium and far for superposition. The stacking results of data gathers with different offset distances were calibrated and compared, with a focus on the analysis of the seismic reflection characteristics of the No. 5 sand.

[0184] Through comparative analysis, it was found that in the mid-offset gather, the seismic reflection at the No. 5 sand body showed complex waves, and there were conditions for further differentiation. In order to further improve the resolution of seismic data, we conducted a post-stack frequency-boosting test on the mid-offset stacked profile. After the frequency boosting, the No. 5 sand body showed the characteristics of two peaks and two valleys in the seismic, and the river channel superposition phenomenon was compounded with geological knowledge.

[0185] PaleoScan software was used to finely carve the channel sand body encountered in the No. 5 sand body on the frequency data. It is believed that the channel overlap is serious in the QL205 well area, with a total of 3 main channels and 2 small channels. Among them, the 3 main channels can be roughly divided into two phases according to the order of deposition: the first phase channel is lower in space and has a northeast direction; the second phase channel is located above the first phase sand body in seismic, and there are two channels, but because the channel development period is close, it is difficult to distinguish on the profile. The target channel of the Qiulin 205_H1 well is the second phase channel, which extends in the northeast direction and overlaps with the northeast-to-north-south channel.

[0186] Based on this understanding, the designed horizontal section of the Qiulin 205_H1 well was successfully drilled in May 2019, encountering 580m of sand layer and successfully completing the designed drilling task.

[0187] While the inventive concept has been particularly shown and described with reference to exemplary embodiments thereof, it will be understood by those skilled in the art that various changes in form and details may be made therein without departing from the spirit and scope of the invention as defined by the claims.

Claims

1. A multi-dimensional identification and characterization method for multi-stage superimposed sand bodies in fluvial facies, characterized by: The method comprises the following steps, S1, synthetic log well seismic calibration step, obtaining drilling data, logging data and seismic data of the research area to perform synthetic log well seismic calibration; S2, the three-dimensional isochronous model establishment step, performs fault interpretation on the three-dimensional seismic data, and establishes a seismic sequence stratigraphic framework model, i.e., a three-dimensional isochronous model, with the help of the interpreted faults; and automatically completes the horizon tracking interpretation; S3, a step of determining the channel stage, establishing a certain number of stratum slices through the obtained three-dimensional isochronous model, assigning seismic attributes to the stratum slices, observing the channel distribution characteristics through the combination analysis of plane sections, and determining the channel stage; S4, the main body identification step of the river channel sand body, by matching pursuit frequency division processing on the original seismic data, the three frequency band data obtained after the frequency division processing are RGB fused; finally, the river channel contour is determined by the attribute fusion representing the river channel boundary; S5, a river channel fine carving step, combining the three-dimensional isochronous model obtained in step S3 and the river channel contour obtained in step S4 to perform fine carving on the river channel; S6. Carry out a combined plan and section analysis on the river channel carving results of step S5, split the superimposed river channels into single channels, and determine the early and late stages of the single channels.

2. A multi-dimensional identification and characterization method for multi-stage superimposed sand bodies of fluvial facies as claimed in claim 1, characterized in that: In the synthetic record well seismic calibration step, acoustic wave synthetic record calibration is adopted and performed according to the set principles; the set principles include extracting seismic traces along the trajectory, estimating wavelets from wellside traces, giving priority to strong phases, and jointly comparing and calibrating the main survey line and the tie line.

3. A multi-dimensional identification and characterization method for multi-stage superimposed sand bodies of fluvial facies as claimed in claim 2, characterized in that: In the synthetic log seismic calibration step, before the synthetic log seismic calibration, the logging curve is optimized, and the optimization includes manual editing, filtering or correction; wherein the manual editing refers to point-by-point correction of abnormal values ​​in the curve, adding a constant value to the amplitude of a certain depth section or the entire well section of the curve, and multiplying the amplitude of a given depth section by the constant value; The filtering process refers to eliminating singular points in the curve by using median filtering and bandpass filtering to achieve the purpose of smoothing the curve; The correction process refers to calculating the error using the seismic logging data and the overall sonic logging data, and then obtaining the error curve, and then correcting the original sonic logging curve with the error curve.

4. A multi-dimensional identification and characterization method for multi-stage superimposed sand bodies of fluvial facies according to any one of claims 1 to 3, characterized in that: In step S2, PaleoScan software is used to perform fault interpretation on 3D seismic data.

5. A multi-dimensional identification and characterization method for multi-stage superimposed sand bodies of fluvial facies according to any one of claims 1 to 3, characterized in that: In step S2, the global model-controlled structural modeling technology is used to interpret the faults of the seismic data in the study area; in the global model-controlled framework, automatic identification and manual judgment are combined to jointly interpret each fault from three directions to form a fault plane that is highly consistent with the directions and to implement each fault; finally, the interpreted faults are used to improve the global model so that the finely interpreted faults can participate in the establishment of the three-dimensional isochronal model.

6. A multi-dimensional identification and characterization method for multi-stage superimposed sand bodies of fluvial facies as claimed in claim 5, characterized in that: In step S2, the specific process of establishing the three-dimensional isochronous model is as follows: On the basis of fault interpretation, based on the cost function of seismic data similarity and geological consistency, the idea based on the global optimal method is adopted to spatially deconstruct the seismic data body, select the optimal structure from a large number of possible sequence stratigraphic frameworks, and directly establish a seismic sequence stratigraphic framework model, namely a three-dimensional isochronal model, from the seismic data to automatically complete the stratigraphic tracking interpretation.

7. A multi-dimensional identification and characterization method for multi-stage superimposed sand bodies of fluvial facies according to any one of claims 1 to 3, characterized in that: In step S3, the set seismic attributes include root mean square amplitude attributes.

8. A multi-dimensional identification and characterization method for multi-stage superimposed sand bodies of fluvial facies according to any one of claims 1 to 3, characterized in that: In step S4, the original seismic data is processed by matching pursuit frequency division. Specifically, the complex seismic trace of the original seismic trace is first calculated, and the instantaneous attribute at the maximum value of the instantaneous amplitude envelope is obtained; then, a small-scale atomic library is established with the instantaneous attribute as the center; the matching pursuit algorithm is used to find the best matching atom from the atomic library, and it is continuously iterated until a given threshold is reached; finally, low-frequency, medium-frequency, and high-frequency band data are obtained based on the matching pursuit frequency division processing to highlight the river channel.

9. A multi-dimensional identification and characterization method for multi-stage superimposed sand bodies of fluvial facies as claimed in claim 8, characterized in that: The instantaneous properties include displacement, instantaneous frequency and instantaneous phase.

10. A multi-dimensional identification and characterization method for multi-stage superimposed sand bodies of fluvial facies as claimed in claim 8, characterized in that: In step S4, the three frequency band data obtained after the frequency division process are subjected to RGB fusion, specifically, the three frequency band data of low frequency, medium frequency and high frequency obtained by the matching pursuit frequency division process are respectively assigned RGB colors, and then the energy attribute bodies of the three frequency bands that do not overlap with each other are mixed and displayed in RGB mode to form an RGB mixed frequency data body with passband information; specifically, First, each energy attribute body is normalized so that each energy attribute body has a completely consistent dynamic range of change; then the three energy attribute bodies are mixed with RGB colors, and each sampling point obtains a composite color, which is related to the proportion of each energy attribute body at this point.

11. A multi-dimensional identification and characterization method for multi-stage superimposed sand bodies of fluvial facies according to any one of claims 1 to 3, characterized in that: The attributes characterizing the river channel boundary include amplitude attributes, coherence volume attributes and fracture detection attributes.

12. A multi-dimensional identification and characterization method for multi-stage superimposed sand bodies of fluvial facies according to any one of claims 1 to 3, characterized in that: In step S5, the step of fine carving of the river channel specifically refers to determining the outline of the river channel on the plane through the judgment results of step S4. On this basis, the longitudinal thickness of the river channel is controlled through a three-dimensional isochronal model to carry out fine carving of the thickness distribution characteristics of the river channel sand.

Images