Patent Application
20140200812
Survey data can be collected and processed to produce a representation (e.g., image) of a subsurface structure. In some implementations, survey data includes seismic survey data collected using seismic survey equipment. The seismic survey equipment includes one or more seismic sources that are activated to produce seismic wavefields propagated into the subsurface structure. A part of the seismic wavefields is reflected from the subsurface structure and detected by seismic receivers that are part of the survey equipment.
Seismic surveying can be performed in a marine environment. An issue associated with marine seismic surveying is the presence of ghost data. Ghost data refer to data in measurement data resulting from reflections from an air-water interface of the marine environment. A seismic wavefield generated by a seismic source is propagated generally downwardly into the subsurface structure. A reflected seismic wavefield (that is in response to the seismic wavefield propagated by the seismic source) propagates generally upwardly toward an arrangement of seismic receivers. In the marine environment, where receivers are generally positioned beneath the water surface, the seismic wavefield reflected from the subsurface structure continues to propagate upward past the receivers towards the air-water interface, where the seismic wavefield is reflected back downwardly.
This reflected, generally downwardly traveling seismic wavefield from the air-water interface is detected by the seismic receivers as ghost data, which appears in measurement data collected by the seismic receivers. The presence of ghost data can result in reduced accuracy when generating a representation of the subsurface structure based on the measurement data.
In general, according to some implementations, survey data corresponding to a subsurface region of interest is received. A wavefield is determined by iteratively performing until a specified condition is satisfied: selecting, for a current iteration based at least in part on a current residual representing an approximation error, an element that includes a representation of at least one portion of the wavefield, where the element is determined from the received survey data; computing, for the current iteration, a respective data structure from the selected element; and orthogonally projecting the data structure onto a space spanned by a plurality of data structures including the computed data structure. The current residual is updated based at least in part on the orthogonal projection.
Other or alternative features will become apparent from the following description, from the drawings and from the claims.
Some embodiments are described with respect to the following figures.
It will also be understood that, the terms first, second, etc., are used to distinguish one element from another, and should not be construed to imply any ordering of the elements. For example, a first element or step could be termed a second element or step, and, similarly, a second element or step could be termed a first element or step.
As used herein, the singular forms “a,” “an” and “the” are intended to include the plural forms as well, unless the context clearly indicates otherwise. It will also be understood that the term “and/or” as used herein refers to and encompasses any possible combination of one or more of the associated listed items. It will be further understood that the terms “includes,” “including,” “comprises,” “comprising,” “has,” and/or “having” when used herein, specify the presence of stated features, integers, steps, operations, elements, and/or components, but do not preclude the presence or addition of one or more other features, integers, steps, operations, elements, components, and/or groups thereof.
As used herein, the term “if” may be construed to mean “when” or “upon” or “in response to determining” or “in response to detecting,” depending on the context. Similarly, the phrase “if it is determined” or “if [a stated condition or event] is detected” may be construed to mean “upon determining” or “in response to determining” or “upon detecting [the stated condition or event]” or “in response to detecting [the stated condition or event],” depending on the context.
In the ensuing discussion, reference is made to performing deghosting according to some implementations in a marine survey environment. Note, however, that techniques or mechanisms according to some implementations can also be applied in land-based survey environments or wellbore-based survey environments in which ghost data can appear in measured survey data, as measured by one or more survey receivers. In addition, techniques or mechanisms according to some implementations can be applied in other contexts, such as based on data collected by cables or streamers that are in slanted acquisition profiles (cables or streamers including survey receivers and/or survey sources are slanted rather than horizontal) and/or towed in turning configurations (e.g., data acquired by survey arrangements that shoot in turns or that perform coil-based acquisition).
Moreover, although reference is made to performing surveying to characterize a subsurface structure, techniques or mechanisms according to some implementations can also be applied to perform surveys of other structures, such as human tissue, a mechanical structure, plant tissue, animal tissue, a solid volume, a substantially solid volume, a liquid volume, a gas volume, a plasma volume, a volume of space near and/or outside the atmosphere of a planet, asteroid, comet, moon, or other body, and so forth. In addition, the following describes seismic sources and seismic receivers that are part of seismic survey equipment. In other implementations, other types of survey equipment can be used, which can include other types of survey sources and survey receivers.
Deghosting attempts to remove ghost data from measured survey data. Ghost data (or ghost reflections) can result in gaps or notches in the amplitude spectra of recorded survey data, where the notches can reduce the useful bandwidth of the survey data. Generally, deghosting is applied to the total wavefield (the sum of the upgoing and downgoing wavefields); the deghosting produces the upgoing portion (the portion reflected from a subsurface structure) of the total wavefield. In a deghosting procedure, a given component of the recorded total wavefield can be expressed mathematically as the combination of a ghost operator (which corresponds to the given component) and the upgoing wavefield.
Generally, an upgoing wavefield refers to a wavefield that travels in a direction that has at least one directional component that is in the vertical up direction. Similarly, a downgoing wavefield refers to a wavefield that travels in a direction that has at least one directional component that is in the vertical down direction.
In accordance with some implementations, techniques or mechanisms are provided to determine a target wavefield that can be used for performing deghosting or for some other operation. The determined target wavefield can be the upgoing wavefield (or any other target wavefield). The target wavefield can be determined by using an iterative process that includes an orthogonal generalized matching pursuits (OGMP) technique (discussed further below).
Although reference is made to using the OGMP technique for determining a target wavefield for purposes of deghosting, it is noted that the OGMP technique can be applied for performing other operations, such as to perform crossline interpolation of survey data. In a survey arrangement, such as a towed marine survey arrangement or land-based survey arrangement, multiple lines (e.g., streamers or arrays) of survey receivers can be provided. Although the spacing between survey receivers along a line can be relatively small (to provide finer sampling of survey data along the direction of the lines), the spacing between the lines can be relatively coarse, which provides for coarse crossline survey receiver separations. In other words, in the crossline direction (direction that is generally perpendicular to the direction of the lines), coarser sampling of survey data is achieved. To provide finer sampling of survey data in the crossline direction, crossline interpolation can be performed to produce survey data at interpolated points (points where survey receivers do not exist) between the lines.
The OGMP technique according to some implementations can also be applied for performing other types of operations.
As discussed in further detail below, the OGMP technique according to some implementations uses dictionary elements that are vectors whose elements are the product of a ghost operator and a complex exponential, in the context of deghosting. In other contexts, a dictionary element can be a vector having elements that are the product of an operator and a complex exponential. A dictionary element represents a part of a total wavefield at the locations of the respective survey receivers.
The OGMP technique applies orthogonal matching pursuits to derive an approximation to components of a measured multicomponent wavefield, in the form of a weighted sum (series expansion), or other aggregate, of dictionary elements. A matching pursuits procedure uses the theory of acoustic wave propagation to formulate mapping of a target wavefield (e.g., upgoing wavefield), which is the desired output, onto components of the measured multicomponent wavefield. The matching pursuits procedure is an iterative process that iteratively determines an improved-fit (e.g., best-fit) target wavefield that can be mapped by ghost operators to respective components. The resulting target wavefield can be output at an arbitrary location (even at a location where a survey receiver does not exist), which allows for performing crossline interpolation as discussed above.
The target wavefield (e.g., an upgoing wavefield) can be estimated by omitting the ghost operators of the dictionary elements from the weighted sum approximation. The estimated (interpolated) downgoing wavefield and hence also the estimated (interpolated) total wavefield can be obtained by modifying expansion coefficients of the weighted sum approximation.
The marine vessel 100 tows the streamer 102 and seismic source assembly 114 through a body of water 108 above a bottom surface 118 (e.g., seafloor). A subsurface structure 110 is located below the bottom surface 118, and the subsurface structure 110 includes at least one subsurface element 112 of interest. Examples of the subsurface element 112 can include a hydrocarbon-bearing reservoir, a freshwater aquifer, a gas injection zone or other subsurface element of interest.
The upgoing seismic wavefield (122) continues to travel upwardly until the wavefield reaches the air-water interface (106), where the seismic wavefield is reflected generally downwardly (as indicated by arrow 124). The reflected downgoing seismic wavefield (124) is also detected at the seismic receivers 104, which causes ghost data to appear in the measurement data collected by the seismic receivers 104. The reflected downgoing wavefield interacts with the upgoing wavefield, which causes constructive and destructive interference that result in the ghost data. This interference is detrimental to the seismic data since it causes amplitude and phase distortions and can result in total elimination of frequencies near the so-called ghost notch frequency.
For simplicity,
The process 300 receives (at 302) survey data acquired by survey receivers (e.g., 104 or 204), where the survey data corresponds to a subsurface region of interest. The process then determines (at 304) the target wavefield by using an iterative process that iterative performs tasks 306-312 performing until a specified condition (stopping condition) is satisfied.
The iterative determining process (304) includes selecting (at 306), for a current iteration based at least in part on a current residual representing an approximation error, a dictionary element that includes a representation of at least one portion of the wavefield, where the selected dictionary element is determined from the received survey data. In some examples, the residual is the sum of the errors between a measured component and the corresponding modeled estimate. As discussed further below, the residual is used for converging the iterative determining process, by using the residual as part of the stopping condition of the iterative determining process (304). In some implementations, selecting the dictionary element is according to a criterion that reduces a residual for a next iteration.
As noted above, a dictionary element, expressed as Eq. 5 below in some examples, is a vector including multiple elements, where an element in the vector is the product of a ghost operator and a complex exponential, for example. A dictionary element represents part of a total wavefield at the locations of the respective survey receivers.
The iterative determining process (304) further computes (at 308), for the current iteration, a data structure (e.g., orthonormal vector) from the selected dictionary element. An example orthonormal vector is expressed as Eq. 1 below. The orthonormal vectors collectively provide an orthonormal basis of a space that is spanned by a dictionary element.
The iterative determining process (304) then orthogonally projects (at 310) the orthonormal vector onto a space spanned by the orthonormal basis. An example of such orthogonal projection is represented as Eq. 2 below.
Next, the iterative determining process (304) updates (at 312) the current residual based at least in part on the orthogonal projection. The updated current residual represents an updated approximation error of the wavefield estimation process 300. An example of updating the current residual is expressed by Eq. 3 below.
As the iterative determining process (304) proceeds through multiple iterations, the residual is continually updated, and eventually will reach a sufficiently low value (e.g., less than a predetermined threshold). The current residual (as computed at 312) being less than the predetermined threshold is an example of a stopping condition that causes the iterative determining process (304) to stop.
Once the residual is small enough, the total wavefield can be derived (at 314) by computing a weighted sum (or other aggregation) of dictionary elements, such as expressed by Eq. 4 below. The target wavefield (e.g., upgoing wavefield when performing deghosting) can be derived from the total wavefield by omitting the ghost operators of the dictionary elements.
The measured survey data acquired by survey receivers (e.g., 104, 204) can include components in multiple directions, including the horizontal directions such as the x and y directions, as well as the vertical direction, which can be referred to as the z direction. The measured survey data can include particle motion data, including velocities, accelerations, and so forth.
An approximation for the measured P, Vy, Vz (pressure and particle velocity) components of the total wavefield may be derived in the form of a linear sum of complex exponentials (indexed by spatial wavenumber), each multiplied by the respective ghost operator, such as expressed by Eq. 4 below.
The following describes a difference between a matching pursuits procedure and an orthogonal matching pursuits procedure. For illustrative purposes, assume that matching pursuits is being used to approximate a function using a weighted sum of basis functions (e.g., dictionary elements di) that are selected from a larger dictionary of such elements. At each iteration, the matching pursuits procedure selects the element from the dictionary giving the largest absolute projection onto the current residual. The projection gives the value of an expansion coefficient, and hence the contribution of the element to the approximation. The residual is then updated by subtracting the contribution. The matching pursuits procedure then proceeds to the next iteration.
The rationale of the orthogonal matching pursuits procedure is that although the matching pursuits procedure will give a residual that eventually reduces to zero or other low value, the residual at each iteration is not the smallest obtainable with the set dictionary elements so far selected at the current and previous iterations. To improve upon matching pursuits, the orthogonal matching pursuits procedure forms an orthonormal basis out of the selected dictionary elements and derives an expansion in terms of this new orthonormal basis (which is according to the orthonormal vectors discussed above). At a given iteration, the approximation computed using the orthogonal matching pursuits procedure is then the orthogonal projection of the desired function onto the space spanned by the orthonormal basis. The residual is orthogonal (perpendicular) to this space and is therefore a minimum. In some implementations, the orthonormal basis may be computed using a Gram-Schmidt algorithm; in other examples, other techniques for forming the orthonormal basis can be used.
The orthonormal vectors that make up the orthonormal basis can be denoted by u. In iteration n+1, the vector (un) added to the orthonormal basis is given by:
In the foregoing, dn represents a dictionary element as expressed by Eq. 5 below. For example, after three iterations, the following three respective orthonormal vectors are constructed:
u
0
=d
0
u
1
=d
1
−<d
1
,u
0
>u
0
u
2
=d
2
−<d
2
,u
1
>u
1
−<d
2
,u
0
>u
0
In general, the orthonormal vector u for a current iteration is orthogonal to previous orthonormal vectors u's computed in previous iterations. The orthonormal vectors u's computed for the multiple iterations are included in an orthonormal basis of the space spanned by the dictionary elements di. For a given iteration, the current u is projected onto the current residual, and the residual for the subsequent iteration is computed. The projection gives the coefficient bi of ui from
R
i
f=<R
i
f,u
i
>u
i
+R
i+1
f=b
i
u
i
+R
i+1
f, (Eq. 2)
where the updated residual Ri+1 is used to select the next orthonormal vector from the dictionary (d1). The criterion used to select a dictionary element d1 is that it maximizes
|<Ri+1f,dk>|, (Eq. 3)
where the index k ranges over the entire dictionary (i.e., all dictionary elements). When the residual is small enough, the expansion in terms of the di is recovered by back substitution such that the following total wavefield PT is derived:
Eq. 4 expresses the total wavefield as a weighted sum of dictionary elements,
where the weights are represented by coefficients ai. The foregoing weighted sum can also be equivalently computed by
which is the weighted sum of orthonormal vectors derived in the iterative determining process (304) of
The target wavefield (e.g., an upgoing wavefield) can be estimated from the total wavefield of Eq. 4 by omitting the ghost operators (see Eq. 5 below) of the dictionary elements from the weighted sum. In other words, estimating the upgoing wavefield can be performed by using a modified version of Eq. 4, in which di is replaced with elements without the ghost operators GP, GY, and GZ in Eq. 5.
The OGMP technique can be applied to dictionary elements (dj) that are finite-dimensional vectors constructed out of the products of complex exponential functions and ghost operators. The vector elements of the dj are the values of these products at survey receiver locations.
In Eq. 5, j indexes a dictionary element, k is a spatial wavenumber, f is frequency, and z is the streamer depth (in the vertical direction) that is used to define the ghost operators GP, GY, and GZ. The suffixes P, Y, Z denote the different components of the multi-component wavefield.
The d(k) elements in Eq. 5 are vectors whose components are the values of the complex exponential function:
d(k)=eik.x, (Eq. 6)
which has spatial wavenumber k and spatial coordinate x, evaluated at the survey receiver locations xi. If the input data is recorded at NY survey receivers, where three components are used, the vectors dj have length 3NY. For example, for the example case NY=2, the OGMP dictionary element corresponding to the wavenumber ki would be the vector
wherein in the most general case k and x are two-dimensional vectors (i.e., (kx,ky) and (x,y) respectively for survey receivers located on a two-dimensional surface). However, the dictionary elements can be placed in the one-dimensional form of dj above.
The processor(s) 404 is (are) connected to a storage medium (or storage media) 406, which can store measurement data 408 collected by the survey receivers 104 or 204 depicted in
The storage medium (or storage media) 406 can be implemented as one or more non-transitory computer-readable or machine-readable storage media. The storage media can include different forms of memory including semiconductor memory devices such as dynamic or static random access memories (DRAMs or SRAMs), erasable and programmable read-only memories (EPROMs), electrically erasable and programmable read-only memories (EEPROMs) and flash memories; magnetic disks such as fixed, floppy and removable disks; other magnetic media including tape; optical media such as compact disks (CDs) or digital video disks (DVDs); or other types of storage devices. Note that the instructions discussed above can be provided on one computer-readable or machine-readable storage medium, or alternatively, can be provided on multiple computer-readable or machine-readable storage media distributed in a large system having possibly plural nodes. Such computer-readable or machine-readable storage medium or media is (are) considered to be part of an article (or article of manufacture). An article or article of manufacture can refer to any manufactured single component or multiple components. The storage medium or media can be located either in the machine running the machine-readable instructions, or located at a remote site from which machine-readable instructions can be downloaded over a network for execution.
In general, according to some implementations, survey data corresponding to a subsurface region of interest is received. A wavefield is determined by iteratively performing until a specified condition is satisfied: selecting, for a current iteration based at least in part on a current residual representing an approximation error, an element that includes a representation of at least one portion of the wavefield, where the element is determined from the received survey data; computing, for the current iteration, a respective data structure from the selected element; orthogonally projecting the data structure onto a space spanned by a plurality of data structures including the computed data structure; and updating the current residual based at least in part on the orthogonal projection.
In general, according to further or other implementations, selecting the element comprises selecting an element from a dictionary of elements that represent respective portions of the wavefield corresponding to respective survey receiver locations.
In general, according to further or other implementations, computing the data structure comprises computing an orthonormal vector.
In general, according to further or other implementations, orthonormal vectors for respective iterations provide an orthonormal basis, the space being spanned by the orthonormal basis.
In general, according to further or other implementations, selecting the element is according to a criterion that reduces a residual for a next iteration.
In general, according to further or other implementations, determining the wavefield comprises determining a total wavefield.
In general, according to further or other implementations, determining the wavefield comprises determining an upgoing wavefield.
In general, according to further or other implementations, the specified condition includes the current residual being less than a predetermined threshold.
In general, according to further or other implementations, deghosting of the received survey data is performed using the determined wavefield.
In general, according to further or other implementations, interpolation to compute survey data at one or more interpolation points is performed using the determined wavefield.
In general, according to some implementations, a computer system includes a storage medium to store survey data corresponding to a subsurface region of interest, and at least one processor configured to iteratively determine a wavefield, based at least in part on the survey data, by performing orthogonal matching pursuits.
In general, according to further or other implementations, performing the orthogonal matching pursuits comprises performing an iterative process comprising: selecting, for a current iteration based at least in part on a current residual representing an approximation error, a dictionary element that includes a representation of at least one portion of the wavefield, where the dictionary element is determined from the received survey data; computing, for the current iteration, a respective orthonormal vector from the selected dictionary element; orthogonally projecting the orthonormal vector onto a space spanned by a plurality of orthonormal vectors; and updating the current residual based at least in part on the orthogonal projection.
In general, according to further or other implementations, the iterative process stops upon the current residual satisfying a specified condition.
In general, according to further or other implementations, the at least one processor is configured to further perform deghosting and crossline interpolation using the determined wavefield.
In general, according to further or other implementations, the at least one processor is configured to compute a total wavefield derived from a weighted aggregate of the orthonormal vectors computed for respective iterations of the iterative process.
In general, according to further or other implementations, the weighted aggregate includes a weighted sum of products of coefficients and the orthonormal vectors, wherein the coefficients are computed by the orthogonal projecting of the orthonormal vector onto the space spanned by the plurality of orthonormal vectors.
In general, according to further or other implementations, the dictionary elements are based at least in part on products of ghost operators and values derived from the survey data, and wherein the at least one processor is configured to compute an upgoing wavefield from the total wavefield by omitting the ghost operators.
In general, according to further or other implementations, selecting the dictionary element is according to a criterion that reduces a residual for a next iteration.
In general, according to some implementations, an article comprising at least one non-transitory machine-readable storage medium stores instructions that upon execution cause a system to receive survey data corresponding to a subsurface region of interest, and determine a wavefield by iteratively performing until a specified condition is satisfied: selecting, for a current iteration based at least in part on a current residual representing an approximation error, an element that includes a representation of at least one portion of the wavefield, wherein the element is derived from the received survey data; computing, for the current iteration, a respective data structure from the selected element; orthogonally projecting the data structure onto a space spanned by a plurality of data structures including the computed data structure; and updating the current residual based at least in part on the orthogonal projection.
In the foregoing description, numerous details are set forth to provide an understanding of the subject disclosed herein. However, implementations may be practiced without some of these details. Other implementations may include modifications and variations from the details discussed above. It is intended that the appended claims cover such modifications and variations.
This application claims the benefit of U.S. Provisional Patent Application Ser. No. 61/751,689 filed Jan. 11, 2013, which is incorporated herein by reference in its entirety.
| Number | Date | Country | |
|---|---|---|---|
| 61751689 | Jan 2013 | US |
