Q-compensated full wavefield inversion

Abstract

A method, including: obtaining a velocity model generated by an acoustic full wavefield inversion process; generating, with a computer, a variable Q model by applying pseudo-Q migration on processed seismic data of a subsurface region, wherein the velocity model is used as a guided constraint in the pseudo-Q migration; and generating, with a computer, a final subsurface velocity model that recovers amplitude attenuation caused by gas anomalies in the subsurface region by performing a visco-acoustic full wavefield inversion process, wherein the variable Q model is fixed in the visco-acoustic full wavefield inversion process.

Claims

1. A method, comprising: obtaining a velocity model generated by an acoustic full wavefield inversion process; generating, with a computer, a variable Q model by applying pseudo-Q migration on processed seismic data of a subsurface region, wherein the velocity model is used as a guided constraint in the pseudo-Q migration; generating, with a computer, a final subsurface velocity model that recovers amplitude attenuation caused by a gas anomaly in the subsurface region by performing a visco-acoustic full wavefield inversion process, wherein the variable Q model is fixed in the visco-acoustic full wavefield inversion process; generating a subsurface image based on the final subsurface velocity model, wherein the subsurface image includes geological structure beneath the gas anomaly; and drilling a well to extract the hydrocarbons, wherein the well is disposed at a location determined by analysis of the subsurface image.

2. The method of claim 1, further comprising: generating the processed seismic data, wherein the generating includes applying an acoustic ray-based pre-stack depth migration to the velocity model and outputting common image gathers.

3. The method of claim 2, wherein the generating the variable Q model includes flattening the common image gathers in accordance with the guided constraint.

4. The method of claim 2, wherein the guided constraint defines a zone, within the velocity model, that contains the gas anomaly.

5. The method of claim 4, wherein the pseudo Q migration is only applied to the zone that contains the gas anomaly.

6. The method of claim 4, further comprising limiting application of the pseudo-Q migration to the zone that contains the gas anomaly.

7. The method of claim 1, wherein the variable Q model is kept fixed through an entirety of the visco-acoustic full wavefield inversion process.

8. The method of claim 1, wherein the generating the final subsurface velocity model includes applying pseudo-Q migration to construct another variable Q model via flattening visco-acoustic common image gathers, and the velocity model generated from the visco-acoustic full wavefield inversion process is used as a guided constraint in the pseudo-Q migration.

9. The method of claim 1, further comprising iterative repeating (a) performance of the visco-acoustic full wavefield inversion process, (b) then generation of visco-acoustic common image gathers from visco-acoustic ray-based pre-stack depth migration, and (c) then generation of another variable Q model via flattening the visco-acoustic common image gathers, said iterative repeating occurring until a predetermined stopping condition is reached, wherein the velocity model generated from the visco-acoustic full wavefield inversion process is used as a guided constraint in the pseudo-Q migration.

10. The method of claim 1, further comprising conducting a seismic survey, wherein at least one source is used to inject acoustic signals into the subsurface and at least one receiver is used to record the acoustic signals reflecting from subsurface features.

11. The method of claim 1, wherein the guided constraint is guided by geological structures inverted from the acoustic full wavefield inversion process.

Description

BRIEF DESCRIPTION OF THE DRAWINGS

(1) While the present disclosure is susceptible to various modifications and alternative forms, specific example embodiments thereof have been shown in the drawings and are herein described in detail. It should be understood, however, that the description herein of specific example embodiments is not intended to limit the disclosure to the particular forms disclosed herein, but on the contrary, this disclosure is to cover all modifications and equivalents as defined by the appended claims. It should also be understood that the drawings are not necessarily to scale, emphasis instead being placed upon clearly illustrating principles of exemplary embodiments of the present invention. Moreover, certain dimensions may be exaggerated to help visually convey such principles.

(2) FIG. 1 is a graphical representation of a subsurface region that is useful in explaining the operation of an exemplary embodiment of the present invention

(3) FIG. 2 illustrates an exemplary method of generating a starting velocity model and starting Q model.

(4) FIG. 3 illustrates an exemplary method of visco-acoustic FWI.

(5) FIG. 4A illustrates an exemplary velocity model.

(6) FIG. 4B illustrates an exemplary Q model to match a gas cloud in FIG. 4A.

(7) FIG. 5A illustrates exemplary common offset data generated by acoustic forward modeling.

(8) FIG. 5B illustrates exemplary common offset data generated by visco-acoustic forward modeling.

(9) FIG. 6A illustrates an exemplary velocity update generated from acoustic FWI.

(10) FIG. 6B illustrates an exemplary velocity update generated from visco-acoustic FWI with Q values fixed.

DETAILED DESCRIPTION

(11) Exemplary embodiments are described herein. However, to the extent that the following description is specific to a particular embodiment, this is intended to be for exemplary purposes only and simply provides a description of the exemplary embodiments. Accordingly, the invention is not limited to the specific embodiments described below, but rather, it includes all alternatives, modifications, and equivalents falling within the true spirit and scope of the appended claims.

(12) Exemplary embodiments described herein provide a comprehensive model building workflow, which effectively compensates for the Q-effect in FWI without suffering energy leakage and is capable of generating high-resolution property profiles with much improved subsurface fidelity.

(13) Seismic waves propagating through gas clouds often result in distorted phase, dim amplitude and lower frequency. Acoustic FWI does not compensate for such Q effects, and thus cannot recover amplitude and bandwidth loss beneath gas anomalies. Non-limiting embodiments of the present technological advancement compensate for the Q-effect by combining ray-based Q-model building with visco-acoustic FWI. In ray-based Q-model building, pseudo-Q migration is used to efficiently scan all possible Q-values for the optimum Q-effect. Compared with other Q estimation methods like Q tomography or full wavefield Q inversion, pseudo-Q migration can scan within a target-oriented local gas area and the scan process is both efficient and stable.

(14) In visco-acoustic FWI, the Q-effect is compensated for in both the forward modeling and the adjoint computation. Application of Q-compensated FWI to complex synthetic data has revealed clearly improved structures beneath the gas zone and the results definitely can benefit geological interpretation. The present technological advancement can be, in general, applied to any field data as long as the Q-effect is considered to be an issue. The present technological advancement is most applicable on the datasets where strong gas-anomalies exist in the subsurface and a conventional acoustic model building workflow cannot recover amplitude and bandwidth loss for the potential reservoir targets beneath the gas.

(15) An embodiment of the present technological advancement provides a gas-friendly model building workflow that can include: ray-based pseudo-Q migration for efficient estimation of Q values; and wave-based visco-acoustic FWI for a Q compensated velocity update.

(16) In visco-acoustic FWI, the Q-values are fixed and only the velocity gradient needs to be computed. Fixing the Q values throughout the entire FWI velocity inversion blocks the energy leakage for Q into velocity, which makes the inversion stable and the inverted velocity model reliable. Applications of Q-compensated FWI to synthetic and field data with gas clouds have shown improved structures and well-ties beneath the gas and thus can positively impact geological interpretation.

(17) FIG. 1 is a graphical representation of a subsurface region that is useful in explaining the operation of an exemplary embodiment of the present technological advancement. The graph is generally referred to by the reference number 100. The graph 100 shows a water region 101 and a sediment region 102. Those of ordinary skill in the art will appreciate that the water region 101 exhibits a very high Q value, typically represented by a large number such as 999. Accordingly, seismic waves travel through the water region 101 with relatively little attenuation. The sediment region 102 may have a much lower Q value than the water region 101. For example, the Q value of the sediment region 102 may be in the range of about 150.

(18) Subsurface anomalies such as gas caps or the like may have extremely low Q values. In the graph 100, an anomaly A 103 exhibits a Q value in the range of about 20. Similarly, an anomaly B 104 also exhibits a Q value in the range of about 20. The low Q values of the anomaly A 103 and the anomaly B 104 result in attenuated seismic data corresponding to deeper subsurface structures. By way of illustration, the anomaly A 103 negatively affects the integrity of seismic data in an anomaly attenuation region 105 that extends below the anomaly A 103. Any seismic energy that travels through the anomaly A 103 will be significantly attenuated when it returns to the surface and is measured. If the anomaly A 103 is disposed above a deposit of hydrocarbons, seismic data that could identify the presence of the deeper reservoir of hydrocarbons could be obscured. This phenomenon is shown in the graph 100 by a series of reflectors 106, 107, 108 and 109. Portions of the reflectors 106, 107, 108 and 109 that are likely to be represented by significantly attenuated seismic data are shown as dashed lines. Data corresponding to portions of the reflectors 106, 107, 108 and 109 that are unlikely to be significantly attenuated by the presence of the anomaly A 103 and the anomaly B 104 are shown as solid lines in FIG. 1. The performance of pseudo Q migration in accordance with an exemplary embodiment of the present invention is intended to restore accurate amplitude, frequency and phase data for seismic energy that is adversely affected by passing through the anomaly A 103 and the anomaly B 104.

(19) FIG. 2 illustrates an exemplary method of generating a starting velocity model and starting Q model, which are then used as inputs to the visco-acoustic FWI method (see FIG. 3). In step 201, an initial velocity model V.sub.0.sup.TOMO is generated. The initial velocity model can be generated through ray-based tomography (Liu et al., 2008). However, other techniques could be used.

(20) In step 203, acoustic FWI is applied using the initial velocity model. In step 205, a high resolution velocity distribution V.sub.1.sup.FWI is generated from the acoustic FWI. The crux of any FWI algorithm can be described as follows: using a starting subsurface physical property model, synthetic seismic data are generated, i.e. modeled or simulated, by solving the wave equation using a numerical scheme (e.g., finite-difference, finite-element etc.). The term velocity model or physical property model as used herein refers to an array of numbers, typically a 3-D array, where each number, which may be called a model parameter, is a value of velocity or another physical property in a cell, where a subsurface region has been conceptually divided into discrete cells for computational purposes. The synthetic seismic data are compared with the field seismic data and using the difference between the two, an error or objective function is calculated. Using the objective function, a modified subsurface model is generated which is used to simulate a new set of synthetic seismic data. This new set of synthetic seismic data is compared with the field data to generate a new objective function. This process is repeated until the objective function is satisfactorily minimized and the final subsurface model is generated. A global or local optimization method is used to minimize the objective function and to update the subsurface model. Further details regarding FWI can be found in U.S. Patent Publication 2011/0194379 to Lee et al., the entire contents of which are hereby incorporated by reference.

(21) In step 207, based on the FWI inverted velocity model V.sub.1.sup.FWI, an acoustic ray-based pre-stack depth migration (PSDM) is applied to compute common-image-gathers whose flatness reflects the accuracy of the velocity model and the Q model. Several pre-stack migration methods can be used to perform step 207, and include, for example, Kirchoff PSDM, one-way wave equation migration, and reverse time migration, each of which is known to those of ordinary skill in the art.

(22) In step 209, the common image gathers (CIG) are provided as an input to the pseudo-Q migration. A common image gather is a collection of seismic traces that share some common geometric attribute, for example, common-offset or common angle.

(23) In step 211, low velocity zones, which normally represent gas anomalies, are determined from V.sub.1.sup.FWI, and provided as an input to the pseudo-Q migration. At later stage where Q is being determined (see step 215), these low velocity zones are used as FWI guided constraints. Particularly, the FWI guided constraints delineate the regions of probable gas (targeted gas zones), and are used in the subsequent pseudo-Q migration to limit the pseudo-Q migration to only those regions likely to contain gas (as indicated by regions of low velocity in V.sub.1.sup.FWI). The guided constraints are guided by geological structures inverted from the acoustic full wavefield inversion process.

(24) In step 213, Q.sub.o is obtained and provided as an input to the pseudo-Q migration. Q.sub.o can be an initial homogeneous Q model.

(25) In step 215, pseudo-Q migration is applied to the regions or zones delineated by the FWI constraints. Pseudo-Q migration is different from Q tomography. Q tomography is a very tedious process which requires carefully preparing different input data. In addition, Q tomography is unstable when the signal/noise ratio of the common-image-angle is low. Pseudo-Q migration, however, is not only efficient, but is also stable since it is similar to a Q scan. Below is a brief overview of the theory of pseudo-Q migration. A fuller description of pseudo Q migration is found in International Patent Application Publication WO 2009/123790, the entire content of which is hereby incorporated by reference.

(26) An aspect of pseudo-Q migration involves the building of a Q integration table that can be used to restore amplitude, frequency and phase values of data corresponding to a migrated trace. Q integration data is computed using the multiplication of a matrix (derivatives of Q integration with respect to a given Q model) and a vector (update of the Q model). The derivatives of Q integration may be computed in part by determining the rate of change in a velocity model. In an exemplary embodiment, the derivatives of the Q integration are desirably represented as a sparse matrix.

(27) The derivative matrix is independent of Q model; therefore, it can be pre-calculated and stored for use with subsequently developed Q models. A table of Q integration is then calculated at each specific image point and for its reflection angle in a trace. Basically, pseudo Q migration in accordance with an exemplary embodiment includes a trace-in and trace-out operation; so the computation is essentially a 1D processing operation. Furthermore, pseudo Q migration can be implemented in target orientation. An input trace will be remigrated for amplitude and phase restoration only if it is predicted to be affected by a low Q zone such as the anomaly A attenuation region 105 (see FIG. 1).

(28) By way of example, let c(x) be a complex velocity in a visco-acoustic velocity field versus the frequency variable . Under these assumptions, c(x) can be represented, as follows:

(29) $\begin{array}{cc}c\left(x\right)=c_o\left(x\right)\left(1+\frac{1}{2}i{Q}^{-1}\left(x\right)+\frac{1}{}{Q}^{-1}\left(x\right)\ln \frac{}{{}_o}\right)& \left(6\right)\end{array}$
where c.sub.0 is the acoustic part of the complex velocity, Q is the quality factor representing attenuation, and .sub.0 is a reference frequency.

(30) The complex travel time can be calculated by

(31) $\begin{array}{cc}{T}_c\left(x\right)=T\left(x\right)-\frac{1}{2}i{T}^{*}\left(x\right)-\frac{1}{}{T}^{*}\left(x\right)\ln \frac{}{{}_o}& \left(7\right)\end{array}$
where (x) is the travel time in the acoustic medium c.sub.0 and

(32) $\begin{array}{cc}{T}^{*}\left(x\right)={}_L\frac{1}{C_{o}Q}\mathrm{ds}& \left(8\right)\end{array}$

(33) The first term in equation (6) contains the primary kinematic information in migration imaging and can be calculated by ray tracing in an acoustic medium. The second term in equation (6) permits migration to compensate for amplitude loss due to attenuation, and the third term in equation (6) permits migration to compensate for phase distortion due to dispersion. Both the second and third terms depend on T*, the integral of Q.sup.1 along the ray path L, as defined in equation (8). T* can be calculated on the same ray paths as used to calculate T. When Q is updated, T remains the same and the change of T* is
T*=c.sub.o.sup.1(Q.sup.1)ds(9)

(34) Moreover, equation (9) may be rewritten into a matrix form:
T*=D*(Q.sup.1)(10)

(35) In equation (10), D is the matrix of derivatives of T* with respect to Q.sup.1 (the derivative values). As examples of terminology used herein, D contains the derivatives of Q integration values based on a velocity model, where the term Q integration values is represented by equation (8) and c.sub.0 refers to the velocity model. The matrix D can be pre-calculated and stored because it does not depend on a particular Q model.

(36) Exemplary processing steps for pseudo Q migration can be stated as follows: (1) given migrated traces (common image gather or the processed seismic data), the velocity model used in the migration, and an initial Q model; (2) select reflection points and estimate reflection angles for those points; (3) compute the derivatives of Q integration with respect to the Q model (dT*/dQ) and output those derivatives; and (4) multiply the derivatives of Q integration by the initial Q model to obtain the table of Q integration (Q.sub.1.sup.Pseudo-Q).

(37) With a simple homogeneous Q available as the initial Q.sub.o model, the process in FIG. 2 will determine and output an optimized variable Q model, Q.sub.1.sup.Pseudo-Q(step 217), which is judged by most-effectively reducing the image gather's curvature (i.e., flatness). In each iteration of the pseudo-Q optimization, the ray path matrix is fixed and the Q-values are perturbed. Different from any inversion-based Q model building methods, pseudo-Q migration scans for available Q models within a reasonable range of Q values and performs the scan within an FWI guided target area (i.e., the areas identified by the FWI velocity model as being gas). Therefore, the resulted Q.sub.1.sup.Pseudo-Q lacks local minima and is thus preferred by the later stage of visco-acoustic wavefield propagation (see FIG. 3).

(38) A common feature of FIG. 2 is that every process can be acoustic. In FIG. 2, the whole process is intentionally kept independent of the Q factor. This acoustic process helps to build a more reliable starting Q model, which is able to improve over an acoustic model from the very beginning.

(39) Outputs from the method in FIG. 2, are obtained and used as the starting models (steps 301 and 303) for the first iteration of the method in FIG. 3. In step 305, visco-acoustic FWI is performed with fixed Q (fixed does not mean constant as Q.sub.1.sup.Pseudo-Q is variable, fixed rather refers to the use of the same variable Q model throughout each iteration i of the method shown in FIG. 3). As illustrated in FIG. 3, which uses both the velocity V.sub.1.sup.FWI and the Q model Q.sub.1.sup.Pseudo-Q, visco-acoustic FWI automatically includes Q in its wavefield propagation and is able to effectively account for the attenuation effect. The Q model, however, is fixed in the visco-acoustic FWI process to avoid energy leakage between Q inversion and velocity inversion. Accordingly, only velocity gradient needs to be computed in the visco-acoustic FWI process.

(40) In step 307, the updated velocity model is output for subsequent use in the visco-acoustic PSDM.

(41) In step 309, the Q model Q.sub.1.sup.Pseudo-Q is obtained for subsequent use in the visco-acoustic PSDM.

(42) In step 311, visco-acoustic PSDM is applied using the updated velocity model from step 307 and the Q model from step 309. Visco-acoustic Kirchoff migration is the most widely used method to generate visco-acoustic PSDM gathers and stacks.

(43) In step 313, common image gathers (CIG) are output from the visco-acoustic PSDM.

(44) Similar to step 211 in FIG. 2, step 315 includes using the FWI inverted velocity model (from step 307) to determine gas anomalies (based on regions of attenuated velocity), said anomalies being useable as guided constraints for the pseudo-Q inversion.

(45) In step 317, the Q model Q.sub.1.sup.Pseudo-Q is obtained for subsequent use in the pseudo-Q migration.

(46) In step 319, pseudo-Q migration stably scans for optimum Q values to further minimize the image gather curvatures. By scan, the process can cycle through a plurality of Q values and determine which of the plurality is the optimum solution. Pseudo-Q migration is discussed supra, and that discussion is applicable to the performance of this step.

(47) The workflow in FIG. 3 updates velocity and Q models iteratively and stops at the predetermined iteration i=N, which can be specified by the user. Alternatively, other stopping criteria could be used.

(48) It is noted that the index i is for the big hybrid loop of the method in FIG. 3, and not the iteration number of the FWI step 305. As those of ordinary skill in the art will understand, FWI is itself an iterative process.

(49) A common feature of FIG. 3 is that every process can be visco-acoustic. In FIG. 3, both depth migration and FWI use a visco-acoustic engine to guarantee that the wave-propagation is consistent.

(50) Once the iterations over the overall process in FIG. 3 are completed (i.e., N iterations are completed), the output from FIG. 3 is a final subsurface velocity model (or physical property model) and a final Q model (steps 321 and 323. The final physical property subsurface model can be used to generate a subsurface image for interpretation of the subsurface and/or management of hydrocarbon exploration (step 325). Application of Q-compensated FWI to complex synthetic data has revealed clearly improved structures beneath the gas zone and the results definitely can benefit geological interpretation. As used herein, hydrocarbon management includes hydrocarbon extraction, hydrocarbon production, hydrocarbon exploration, identifying potential hydrocarbon resources, identifying well locations, determining well injection and/or extraction rates, identifying reservoir connectivity, acquiring, disposing of and/or abandoning hydrocarbon resources, reviewing prior hydrocarbon management decisions, and any other hydrocarbon-related acts or activities.

(51) The following describes a non-limiting example of an application of the present technological advancement.

(52) The methods of FIGS. 2 and 3 have been applied to 2-D synthetic streamer data. FIG. 4A illustrates an exemplary velocity model. FIG. 4A is a true underground model from which forward modeling can be applied to generate synthetic data. The velocity model in FIG. 4A was modified from a deep-water field geology scenario where a gas zone is located close to the water bottom and has made conventional depth imaging and model building very challenging in the beneath reservoir area.

(53) To parallel with the low velocity gas zone 401, a Q anomaly 402 is created as shown in FIG. 4B. Similar to FIG. 4A, FIG. 4B is also used in visco-acoustic modeling to compute synthetic data.

(54) FIG. 5A shows the near-offset data generated from acoustic forward modeling (common offset data, with offset=100 m) using the velocity model in FIG. 4A, whereas FIG. 5B shows similar data generated from visco-acoustic modeling using both velocity and Q anomalies. In the circled areas 501 and 502, events are attenuated due to the absorption effect of the gas. Compared with FIG. 5A, the only difference in FIG. 5B is in the gas area where phases are distorted and amplitudes are attenuated. The difference between FIGS. 5A and 5B demonstrates the Q-attenuation effect. FIG. 5B clearly has low quality because of Q.

(55) FIG. 6A shows the velocity update inverted from acoustic FWI (i.e., step 203) and FIG. 6B shows the velocity update inverted from visco-acoustic FWI (i.e., step 305) with Q values fixed. In both FWI, the synthetic visco-acoustic data were used as the observed data. FIGS. 6A and 6B demonstrates that including Q or not in FWI has significant impacts on the velocity inversion. For this specific case, including fixed Q in FWI clearly improves the velocity update in the target gas area (compare 601 and 602). The same workflow has also been applied to 3-D marine field data and FWI with fixed Q generated visiably better geological structures compared to acoustic FWI.

(56) In all practical applications, the present technological advancement must be used in conjunction with a computer, programmed in accordance with the disclosures herein. Preferably, in order to efficiently perform FWI, the computer is a high performance computer (HPC), known to those skilled in the art. Such high performance computers typically involve clusters of nodes, each node having multiple CPU's and computer memory that allow parallel computation. The models may be visualized and edited using any interactive visualization programs and associated hardware, such as monitors and projectors. The architecture of system may vary and may be composed of any number of suitable hardware structures capable of executing logical operations and displaying the output according to the present technological advancement. Those of ordinary skill in the art are aware of suitable supercomputers available from Cray or IBM.

(57) The present techniques may be susceptible to various modifications and alternative forms, and the examples discussed above have been shown only by way of example. However, the present techniques are not intended to be limited to the particular examples disclosed herein. Indeed, the present techniques include all alternatives, modifications, and equivalents falling within the spirit and scope of the appended claims.

REFERENCES

(58) The following references are hereby incorporated by reference in their entirety: Bai, J., Yingst, D., Bloor, R., Leveille, J., 2014, Viscoacoustic waveform inversion of velocity structures in the time domain, Geophysics, 79, R103-R119; Liu, J., 2007, Method for performing pseudo-Q migration of seismic data, International Patent Publication WO2009/123790; Liu, J., Bear, L., Krebs, J., Montelli, R., and Palacharla, G., 2008, Tomographic inversion by matrix transformation, Geophysics, 73(5), VE35-VE38; Pratt, R. G., Shin, C., and Hicks, G. J., 1998, Gauss-Newton and full Newton methods in frequency-space seismic waveform inversion: Geophysical Journal International, 133, 341-362; Tarantola, A., 1984, Inversion of seismic reflection data in the acoustic approximation: Geophysics, 49, 1259-1266; and Zhou, J., Wu, X., Teng, K., Xie, Y., Lefeuvre, F., Anstey, I., Sirgue, L., 2014, FWI-guided Q tomography for imaging in the presence of complex gas clouds, 76.sup.th EAGE conference.