Patent Application
20150234066
1. Technical Field
Embodiments of the subject matter disclosed herein generally relate to methods and devices used for processing blended seismic data. Blended seismic data is acquired when seismic receivers detect overlapping signals corresponding to seismic waves originating from different seismic sources.
2. Discussion of the Background
Seismic surveys are exploration methods that gather and record data related to seismic wave reflections from interfaces between geological layers of an explored subsurface structure. Seismic surveys are used in the oil and gas industry to search for and evaluate subterranean hydrocarbon deposits.
After a seismic source is activated to inject a seismic wave into an explored underground structure, seismic receivers detect energy (a signal) reflected from the explored underground structure to acquire and record seismic data for a recording period. The interval between sources' activation limits the seismic data acquisition rate. If seismic sources are activated so their respective recording periods overlap, acquired seismic data is blended and includes overlapping signals caused by one or more individual sources, cross-talk interference and noise. However, acquiring data with overlapping recording times reduces a survey's total acquisition time and cost. Instead of reducing acquisition time, acquiring data with overlapping recording times may be used to acquire a higher density dataset in the same survey time. Using blended data requires additional processing to extract seismic datasets focusing on individual signals.
Thus, in conventional surveying techniques, sources are activated so that a signal corresponding to one source does not overlap another signal corresponding to another source in their significant portions (i.e., when the ratio of the signals' amplitudes is substantially larger than each of the individual signal-to-noise ratios). When a source is activated, it is said that a “shot” has occurred.
Thus, for gather 30 in
Varying shot timing (known as “timing dither”), which is seismic source activation at varying intervals, yields an incoherency in cross-talk noise timing in all domains other than the shot domain. For example,
Traditionally, data sets focusing on an individual signal are extracted from blended data using methods that fall into the following three categories (all relying to some degree on randomized timing):
1. Impulsive de-noising,
2. Iterative coherency enhancement/de-noising, and
3. Full modeling of energy from all sources.
Impulsive de-noising technique (disclosed, for example, by Stefani et al., “Acquisition using simultaneous sources,”69th EAGE Conference & Exhibition, 2007, the entire content of which is incorporated herein by reference) uses the fact that when data is sorted into any domain other than the common shot, the cross-talk noise from other sources is incoherent as illustrated in
Iterative coherency enhancement/de-noising techniques (described, for example, in Abma et al., “Separating simultaneous sources by inversion,”71st EAGE Conference & Exhibition, 2009, M. Maraschini et al., “Source Separation by Iterative Rank Reduction—Theory and Applications,”74th EAGE Conference & Exhibition, 2012, and M. Maraschini et al., “An iterative SVD method for deblending: theory and examples,” SEG Technical Program Expanded Abstracts 2012, the entire contents of which are incorporated herein by reference) rely on the fact that cross-talk noise on some traces is a duplication of signal on other traces. This means that with the knowledge of the timing of all shots, a signal estimate made for one source can then be used to reduce the level of cross-talk for all other sources.
The full modeling of energy from all sources technique (described, for example, in Akerberg et al., “Simultaneous source separation by sparse Radon transform,”78th Ann. Internat. Mtg.: Soc. of Expl. Geophys, 2008, and Moore et al., “Simultaneous source separation using dithered sources,”78th Ann. Internat. Mtg.: Soc. of Expl. Geophys, 2008, the entire contents of which are incorporated herein by reference) has similarities to the iterative de-noising method, except that this formulation solves the relationship between source energy and cross-talk noise implicitly at the core of the problem formulation. Equations can be formulated as designing a transform domain for each source or spatial area (e.g., tau-p domain, Fourier domain, etc.) such that when it is reverse-transformed and re-blended, the raw input data is reconstructed as accurately as possible in a least squares sense.
This technique (i.e., full modeling of energy from all sources) uses the timings and positioning of all sources and also relies on a sparse solution to the equations. Once the transform domains have been calculated, the final step to de-blend the data requires application of reverse-transform without re-blending. While this method may result in some filtering of the original data, it removes low-amplitude cross-talk noise and preserves the primary signal. This method could be considered an alternative way to solve the same problem as the iterative coherency enhancement/de-noising technique (analogous to sparse least squares Radon versus inversion through “iterative cleaning”).
It is desirable to improve conventional methods of de-blending seismic data acquired with overlapping recording time while overcoming their drawbacks.
De-blending methods use different seismic source signatures as discriminating factors when extracting datasets corresponding to each of the overlapping signals from blended seismic data. Applying de-convolution with a target seismic source signature to the blended seismic data emphasizes a target signal among the overlapping signals, having an effect similar to a monochromatic focusing lens.
According to one embodiment, there is a method for de-blending seismic data recorded while simultaneously detecting a first signal from an explored underground structure caused by a first seismic wave characterized by a first source signature, and a second signal from the explored underground structure caused by a second seismic wave characterized by a second source signature, respectively, the second source signature being different from the first source signature. The method includes applying a designature operator corresponding to the first source signature, to first seismic data, which is extracted from the seismic data. The method further includes obtaining an estimate of seismic data related to the first signal, from the first seismic data to which the designature operator corresponding to the first source signature has been applied. The method then includes applying a source convolution based on the first source signature, to the estimate of seismic data related to the first signal, to obtain a refined estimate of the first seismic data, and extracting second seismic data from the seismic data by filtering out cross-talk contamination due to the first signal using the refined estimate of the first seismic data. The method the includes storing at least one of the second seismic data and the refined estimate of the first seismic data.
According to another embodiment, there is a seismic data processing apparatus including an input-output (I/O) interface and a data processing unit. The input-output (I/O) interface is configured to receive blended seismic data recorded while simultaneously detecting a first signal from an explored underground structure caused by a first seismic wave characterized by a first source signature, and a second signal from the explored underground structure caused by a second seismic wave characterized by a second source signature, which is different from the first source signature. The data processing unit is configured to apply a designature operator corresponding to the first source signature, to first seismic data, which is extracted from the seismic data, to obtain an estimate of seismic data related to the first signal, from the first seismic data to which the designature operator corresponding to the first source signature has been applied, to apply a source convolution based on the first source signature, to the estimate of seismic data related to the first signal, to obtain a refined estimate of the first seismic data, and to extract second seismic data from the seismic data by filtering out cross-talk contamination due to the first signal using the refined estimate of the first seismic data. At least one of the second seismic data and the refined estimate of the first seismic data are stored.
According to another embodiment, there is a method for exploring an underground structure. The method includes injecting a first seismic wave characterized by a first source signature into the explored underground structure, and injecting a second seismic wave characterized by a second source signature, which is different from the first seismic signature, into the explored underground structure. The method further includes recording blended seismic data while detecting simultaneously a first signal caused by the first seismic wave and a second signal cause by the second seismic wave emerging from the explored underground structure. The method then includes extracting at least first seismic data focusing on the first signal from the blended seismic data using a first source designature based on the first source signature to emphasize the first signal.
The accompanying drawings, which are incorporated in and constitute a part of the specification, illustrate one or more embodiments and, together with the description, explain these embodiments. In the drawings:
The following description of the embodiments refers to the accompanying drawings. The same reference numbers in different drawings identify the same or similar elements. The following detailed description does not limit the invention. Instead, the scope of the invention is defined by the appended claims. The following embodiments are discussed for blended seismic data acquired so that seismic receivers simultaneously record signals due to seismic waves successively injected into an explored underground structure. However, similar de-blending methods may also be used to analyze data sets related to electromagnetic waves or other data acquired in a similar manner.
Reference throughout the specification to “one embodiment” or “an embodiment” means that a particular feature, structure or characteristic described in connection with an embodiment is included in at least one embodiment of the subject matter disclosed. Thus, the appearance of phrases “in one embodiment” or “in an embodiment” in various places throughout the specification is not necessarily referring to the same embodiment. Further, the particular features, structures or characteristics may be combined in any suitable manner in one or more embodiments.
During seismic surveys, blended seismic data is acquired when a first seismic wave is injected into an underground structure, and then a second seismic wave is injected into the underground structure before seismic receivers have finished recording a signal caused by the first seismic wave. Although in the following description only one second seismic wave is discussed for simplicity and clarity, a person skilled in the art would understand that the second wave may be one of a plurality of second seismic waves. Blended seismic data have to be de-blended to be able to analyze the individual signals. The various de-blending methods described in this section use different source signatures characterizing the seismic waves causing the overlapping signals as discriminating factors. Conventional de-blending methods described in the “Discussion of the Background” section do not use seismic source signature information.
Generally, a seismic source signature is amplitude (e.g., pressure variation) versus time. The signal caused by a seismic wave characterized by a seismic source signature is a convolution of the explored underground structure's response and the seismic source signature. This signal is included in the amplitudes (e.g., pressure or displacement) detected in time by seismic receivers, which generate seismic data.
Blended seismic data may be acquired using land or marine data acquisition systems.
Method 400 further includes obtaining an estimate of seismic data related to the first signal, from the first seismic data to which the designature operator corresponding to the first source signature has been applied at 420. In this step, most of the second signal and the cross-talk are eliminated from the first seismic data to which the designature operator corresponding to the first source signature has been applied.
Method 400 further includes applying a source convolution based on the first source signature, to the estimate of seismic data related to the first signal, to obtain a refined estimate of the first seismic data, at 430. Then, at 440, second seismic data is extracted from the seismic data by filtering out cross-talk contamination due to the first signal using the refined estimate of the first seismic data. At 450, at least one of the second seismic data and the refined estimate of the first seismic data are stored.
A similar procedure may then be applied relative to the second seismic data usable to analyze the second signal. Thus, method 400 may further include:
Furthermore, the first seismic data and the second seismic data may be extracted iteratively until a predetermined condition is met. One example of predetermined condition is when a signal-to-noise ratio for the first signal or for the second signal in the refined estimate of the first seismic data or in the refined estimate of the second seismic data, respectively, to be less than a predetermined value. Another example of predetermined condition is performing a predetermined number of iterations.
Method 400 may also include outputting final first data and/or final second data. The final first data may be the refined first data resulting from the last iteration or the first seismic data extracted using the refined estimate of the second seismic data resulting from the last iteration. Similarly, the final second data may be the refined second seismic data resulting from the last iteration, or the second seismic data extracted using the refined estimate of the first seismic data resulting from the last iteration. The final first and/or second data may then be further processed to generate images of the explored underground structure.
The first and second seismic waves may be generated by different sources (i.e., a first seismic source and a second seismic source, respectively). Such sources are said to be simultaneously activated. However, first and second seismic waves may be generated by the same source (e.g., in the case of vibrators). In other words, “simultaneously-activated sources” means one or more sources are activated at an interval so that seismic data includes portions with overlapping signals, i.e., blended seismic data. The interval between injecting the first and second seismic waves into the explored underground structure may vary from shot to shot (i.e., successive seismic waves). For example, for a listening time of 6 s, a later shot (i.e., generating and injecting the second wave) may occur between 3 and 4 s after the first shot.
Depending on the type of data acquisition system and seismic sources used, different source signatures may be achieved by controlling different parameters. For example, in a marine survey, if two gun arrays, each including plural airguns (e.g. 48 airguns), are fired in a flip-flop mode, the source signatures depend on the overall arrangement and individual characteristics (e.g., the respective pneumatic chamber volumes) of the individual airguns. Additionally, for each gun array, the air guns are fired according to a vector of small delays (e.g., a few milliseconds or tens of milliseconds) from a generic firing time. Any change in this vector of small delays alters the gun array's signature.
Seismic vibrators are a type of seismic source used in both marine and land seismic surveys. As opposed to airgun arrays and dynamite, which are impulsive-type seismic sources (generating seismic waves lasting less than 1 s at one location), seismic vibrators generate seismic waves over an extended period (e.g., a few seconds or dozens of seconds). A vibrator source may consist of one or plural seismic vibrators. The vibrator's source signature may be characterized by a frequency versus time pattern (known as a sweep) while seismic waves are generated. Seismic data acquired using vibrators may be correlated with the sweep to convert the extended detected signal into an impulse-type response. Marine or land vibrators may use pseudo-random sweeps, random sweeps, optimal sweeps, regular (linear) sweeps, step sweeps, or each source may use a different type of sweep to obtain different source signatures.
Seismic waves may also be generated by hybrid sources. In a marine seismic survey, hybrid sources may include airguns and vibrators. In land surveys, hybrid sources may include a combination of vibrator trucks and dynamite.
Source signatures may be constant during the seismic survey, i.e., seismic waves generated by seismic source 1 are always characterized by source signature 1, seismic waves generated by seismic source 2 are always characterized by source signature 2, etc. However, the methods according to various embodiments may be applied when the source signatures vary at each activation:
The manner of activating seismic sources may include an element of randomized source timing, partial randomized source timing, or optimal source timing (where activation times are designed to optimize separation with a given algorithm) to modify the source signature at each activation. Vibrating sources may be modified to each have a different sweep from one source activation to the next. Airgun arrays may fire the air guns in a synchronized, random, optimal or desynchronized fashion. The vibrator or desynchronized air-gun arrays may operate with energy being emitted continuously over a period of several minutes or hours.
Looking closer now at the manner of extracting the first and second data, impulsive noise attenuation methods (e.g., similar to the impulsive de-noising techniques mentioned in the Background section) may be used to filter the data represented in a common receiver domain, a common mid-point domain, or a common offset domain. This approach to extracting first data (and, similarly, second data) may be used whether or not at least one of the first and second source signatures are modified at each activation (e.g., with random delays for air-gun arrays), but it is even more efficient when the source signatures are modified. Also, this approach may be used with or without the dithering time. This approach can be modified to include blended acquisition generated by only one source, or to deblend seismic surveys that uses continuous recording acquisition.
In another embodiment, de-blending is implemented within an iterative method based on dithering time. As a consequence of the dithering time, reflections related to each source response can be viewed as coherent in one time configuration and as random noise in a different time configuration. For example (as reproduced from the previously referred to article of Hampson et al.),
Considering two sources (although these methods can be applied to several simultaneous sources or to one source firing with short time intervals) and the activation time of the second source varying randomly with respect to the first source's activation time, the blended data (actually measured) is D=D1+T(D2), where D1 represents signal in response to the first waves generated by the first source, and D2 represents a theoretical response to the second waves (generated by the second source) without time delays. Operator T is a time-delay operator that shifts traces by the delay time of the second source. D is conceptually a data cuboid whose dimensions are t-x-y or time-offset-shot (e.g., may be a single source-cable combination from one sail-line of a marine survey). In any common offset slice of D, D1 appears coherent, while T(D2) is random because each trace belongs to a different shot, hence the imprint of the random delay sequence randomizes the second source component. In a different time representation in which the random delays are removed from D, data is T−1 (D)=T−1(D1)+D2; in this representation, the source 1 component now appears random in the common offset domain while the source 2 component becomes coherent. Operator T−1 is a time-delay operator that shifts traces by the negative delay time of the second source, i.e. T−1(T(D))=D. Exploiting this duality by continually switching between these two time configurations D and T−1(D), first data and second data may be extracted from blended data D as illustrated in
At each iteration, noise (e.g., due to cross-talk) is attenuated for conservative estimates of each signal (D1c and D2c) to obtain estimates of the first and second data (D′1 and D′2). The conservative estimates are obtained by subtracting from the blended data D time aligned estimate of the other source(s) data. Thus, the conservative estimate of the first data D1c=D−T(D2′) is obtained at 610, and the conservative estimate of the second data D2c=T−1(D−D1′) is obtained at 620. Enhanced estimates of the first and second data (D1′ and D2′) are obtained at 615 and 625, respectively, using, for example, Cadzow filtering (applied to small time-offset-shot cubes of D), rank reduction filtering, filtering or thresholding in the curvelet domain, POCS (projection onto convex sets) algorithm (i.e., a method of iterative thresholding/filtering in the Fourier domain), FX de-convolution, projection filtering, dip filtering, x-t de-convolution, sparse radon in time or frequency domain, filtering using the anti-leakage Fourier transform, filtering using the tau-p version of the anti-leakage Fourier transform, etc. Thus, seismic data is represented in a time configuration aligned with the first seismic wave in 610, and 615 (as in
The iterative process ends when a predetermined condition is met (e.g., a predetermined number of iterations, or a signal-to-removed-noise ratio becomes smaller than a predetermined threshold), to output final estimates of the first and second data at 730 and 740, respectively. These final estimates of first and second seismic data may be the refined estimates (D1′ and D2′) obtained at the last iteration or the conservative estimates D−T(D2′) and T−1(D−D1′) calculated using the refined estimates obtained at the last iteration. The designature and resignature operations may be single operators or vary with directivity. If they vary with directivity, the source designature and resignature operations may be applied in the FK or tau-p domain. The source designature step converts the energy emitted from a source to an impulsive wavelet (within the constraints of any energy deficiency of the original signal, e.g. in marine the free surface ghost). Source directivity relates to a source whose energy varies with angle and azimuth leaving the source. An example of directional source designature may be found in Poole et al, “Shot-to-shot directional designature using near-field hydrophone data,” in SEG 2013 conference proceedings. The designature and resignature operations may be designed so that when the designature and resignature operators are convolved, the result is a spike at zero lag.
The iterative loop processing illustrated in
In one embodiment, steps (A)-(D) above may be used to obtain one set of seismic data (e.g., first seismic data), while other methods may be used to obtain the other(s) individual seismic data (e.g., second seismic data usable to analyze the second signal(s)).
The use of different seismic source signatures as discriminating factors provides the advantage that, during each filtering, the events of interest are focused, while the cross-talk energy is defocused or spread, improving the efficiency of most filters. Using different seismic source signatures as discriminating factors may optionally be combined with time dithering. Time delays between sources may no longer be required, in particular, if the source signatures vary from shot to shot. Instead of applying time dithering as a separate process to designature/resignature, the shifts may instead be applied directly to the designature and resignature operators.
In another embodiment, de-blending is implemented within an iterative method based on dithering time without the need to specify the source that is activated in each position. This method allows 1, 2 or more sources to be considered. This strategy can also be applied to dataset obtained using continuous recording, and independent simultaneous shooting acquisition. The dataset is arranged in a 3D data cube, each trace generated by the source activation Si and recorded by the receiver Rj is placed in the mid-point position and is aligned in time according to the firing time of Si. As a consequence of the dithering time, reflections related to Si are coherent, and reflections related to the other source activations are incoherent.
The blended data (actually measured) is D=D0+Q(D0), where D0 represents signal in response to the injected waves, and operator Q is an operator that allows calculating the crosstalk noise generated by the data D0 based on the firing time of each source. D0 is coherent, while Q(Do) is not. Several deblending strategies can be used to obtain D0 from D. Similar to the previous described embodiment, at each iteration an estimate of D0 can be obtained, and the corresponding cross-talk Q(D0) calculated and subtracted from the input, the filter used to obtain D0 becoming milder with iterations. In this embodiment, source signature varying with source activation can improve the deblending process.
In another embodiment, full modeling of energy from all sources (as described in the Background section) is used to extract first seismic data from the data emphasizing the first signal. This step also includes the source resignature operation as part of the reverse transform. The source resignature may be applied in the data domain or in the transform domain. The sparseness constraint works better if the seismic waves have different signatures, or if the signatures vary from shot to shot.
A schematic diagram of a seismic data processing apparatus 800 configured to perform the methods according to various above-discussed embodiments is illustrated in
Processor 802 may communicate with other internal and external components through input/output (I/O) circuitry 808 and bussing 810. Input-output (I/O) interface (808) is configured to receive blended seismic data recorded while simultaneously detecting a first signal from an explored underground structure caused by a first seismic wave characterized by a first source signature, and a second signal from the explored underground structure caused by a second seismic wave characterized by a second source signature, which is different from the first source signature.
Processor 802 carries out a variety of functions as are known in the art, as dictated by software and/or firmware instructions. Processor 802 is configured to apply a designature operator corresponding to the first source signature, to first seismic data, which is extracted from the seismic data, thereby to obtain data emphasizing the first signal corresponding to the first seismic wave. Processor 802 is further configured to obtain an estimate of seismic data related to the first signal, from the first seismic data to which the designature operator corresponding to the first source signature has been applied. Processor 802 unit is further configured to apply a source convolution based on the first source signature, to the estimate of seismic data related to the first signal, to obtain a refined estimate of the first seismic data, and to extract second seismic data from the seismic data by filtering out cross-talk contamination due to the first signal using the refined estimate of the first seismic data.
Processor 802 may extract the second seismic data by filtering out cross-talk contamination due to the first signal, using an impulsive-noise attenuation method, in a common receiver domain, a common mid-point domain, or a common offset domain.
Processor 802 may be further configured:
Processor 802 may be configured to iteratively extract the first seismic data and the second seismic data until a predetermined condition is met.
Server 801 may also include one or more data storage devices, including disk drives 812, CD-ROM drives 814, and other hardware capable of reading and/or storing information, such as a DVD, etc. The first seismic data, second seismic data, refined first data and/or refined second data may be stored in these data storage devices. In one embodiment, software for carrying out the above-discussed methods may be stored and distributed on a CD-ROM 816, removable media 818 or other forms of media capable of storing information. The storage media may be inserted into, and read by, devices such as the CD-ROM drive 814, disk drive 812, etc. Server 801 may be coupled to a display 820, which may be any type of known display or presentation screen, such as LCD, plasma displays, cathode ray tubes (CRT), etc. Server 801 may control display 820 to exhibit images of the explored subsurface structure generated using first and/or second seismic data. A user input interface 822 may include one or more user interface mechanisms such as a mouse, keyboard, microphone, touch pad, touch screen, voice-recognition system, etc.
Server 801 may be coupled to other computing devices, such as the equipment of a vessel, via a network. The server may be part of a larger network configuration as in a global area network such as the Internet 828, which allows ultimate connection to the various landline and/or mobile client/watcher devices.
The above-described methods create the opportunity to acquire seismic data more efficiently (in shorter time or with higher density) without degrading the quality of the explored subsurface image.
Method 900 further includes recording blended seismic data while detecting simultaneously a first signal caused by the first seismic wave and a second signal cause by the second seismic wave emerging from the explored underground structure at 930. Method 900 then includes extracting at least first seismic data focusing on the first signal from the blended seismic data using a first source designature based on the first source signature to emphasize the first signal at 940. Step 940 is part of seismic data analysis, while Steps 910-930 are directed to seismic data acquisition.
The disclosed embodiments provide methods and devices that extract data sets usable to de-blend blended seismic data using the different seismic source signatures of seismic waves causing overlapping signals as discriminating factors. These de-blending methods may be used for seismic data acquired both in land and marine seismic surveys (including transition zone surveys). It should be understood that this description is not intended to limit the invention. On the contrary, the exemplary embodiments are intended to cover alternatives, modifications and equivalents, which are included in the spirit and scope of the invention as defined by the appended claims. Further, in the detailed description of exemplary embodiments, numerous specific details are set forth in order to provide a comprehensive understanding of the claimed invention. However, one skilled in the art would understand that various embodiments may be practiced without such specific details.
Although the features and elements of the present exemplary embodiments are described in particular combinations, each feature or element may be usable alone without the other features and elements of the embodiments or in other various combinations with or without other features and elements disclosed herein.
The written description uses examples of the subject matter disclosed to enable any person skilled in the art to practice the same, including making and using the described devices or systems and performing any of the described methods. The patentable scope of the subject matter is defined by the claims, and may include other examples that occur to those skilled in the art. Such examples are intended to be within the scope of the claims.
This application claims priority and benefit from U.S. Provisional Patent Application No. 61/843,024, filed Jul. 4, 2013, for “De-blending for Simultaneous Source Acquisition Using Source Signature,” the content of which is incorporated in its entirety herein by reference.
| Filing Document | Filing Date | Country | Kind |
|---|---|---|---|
| PCT/EP2014/064257 | 7/3/2014 | WO | 00 |
| Number | Date | Country | |
|---|---|---|---|
| 61843024 | Jul 2013 | US |
