Full-Wavefield Inversion of Primaries and Multiples in Marine Environment

Abstract

Method for using the full wavefield (primaries, internal multiples and free-surface multiples) in inversion of marine seismic data, including both pressure and vertical velocity data (21), to infer a subsurface model of acoustic velocity or other physical property. The marine seismic data are separated (22) into up-going (23) and down-going (24) wavefields, and both wavefields are inverted in a joint manner, in which the final model is impacted by both wavefields. This may be achieved by inverting both wavefields simultaneously (25), or one after the other, i.e. in a cascaded approach (35→37, or 45→47), for the subsurface properties (26, 38, 48).

Description

FIELD OF THE INVENTION

The invention relates generally to the field of geophysical prospecting, and more particularly to geophysical data processing. Specifically, the invention is a method to invert seismic data containing primaries and multiples in a marine environment.

BACKGROUND OF THE INVENTION

Full wavefield inversion (FWI) is a computer-implemented geophysical method that is recently being used to invert for subsurface properties such as velocity or acoustic impedance. FWI is known to estimate the subsurface properties more accurately than, for example, inversion of the recorded wavefield after being processed to eliminate multiple reflections. The crux of any FWI algorithm can be described as follows: using a starting subsurface 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 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. The accuracy of any FWI method is in general dictated by its two important components: the numerical algorithm used for solving wave equation to generate synthetic seismic data and the optimization scheme. Depending on the type of optimization scheme employed, a FWI method may get stuck in a local minimum while updating the subsurface model.

There are several numerical methods such as finite-difference, finite-element etc. available for solving the wave equation. The finite-difference methods [1] which are the most popular numerical scheme for solving the wave equation suffer from the interface error generated by the misalignment between numerical grids and numerical interfaces [2]. Although all types of reflection (primaries, free-surface multiples, internal multiples etc.) suffer from the interface error, the free-surface multiples are affected the most due to multiple bounces between the free surface and reflectors in subsurface. Given that free-surface multiples are some of the strongest arrivals in a seismic record, including free-surface multiples in a FWI workflow may result in erroneous inverted subsurface properties.

Although in any seismic experiment, full wavefield (primaries, internal multiples and free-surface multiples) are acquired, due to inability of accurately modeling free-surface multiples, in most of the FWI methods only primaries and internal multiples are used to invert for subsurface models. Given that the free-surface multiples carry additional information about the subsurface model and complements to the information being carried by primaries and internal multiples, it is expected that including free-surface multiples in inversion will improve the accuracy of the inverted subsurface model. The present invention is a method that permits circumventing the direct modeling and subtraction of free-surface multiples.

SUMMARY OF THE INVENTION

In one embodiment, the invention is a method for inverting marine seismic data to infer a subsurface physical property model, said seismic data including pressure data and vertical velocity data, said method comprising (a) separating the pressure data and the vertical velocity data into an up-going wavefield and a down-going wavefield; and (b) inverting the up-going wavefield and the down-going wavefield together, either simultaneously or cascaded, to infer a subsurface physical property model; wherein the separating and the inverting are performed using a computer.

In a second embodiment, the invention is a method for inverting marine seismic data to infer a subsurface physical property model, said seismic data including pressure data and vertical velocity data, said method comprising: (a) separating the pressure data and the vertical velocity data into an up-going wavefield and a down-going wavefield; and (b) iteratively inverting the up-going wavefield and the down-going wavefield together, meaning that each iteration cycle inverts the up-going wavefield or the down-going wavefield or simultaneously inverts both up-going and down-going wavefields, with each iteration cycle resulting in a model update which is used to generate simulated data in a next iteration cycle, resulting finally in a final physical property model; wherein the separating and the inverting are performed using a computer.

BRIEF DESCRIPTION OF THE DRAWINGS

The advantages of the present invention are better understood by referring to the following detailed description and the attached drawings, in which:

FIG. 1 illustrates decomposition of pressure and vertical particle velocity into up-going and down-going wavefields, where the up-going wavefield contains primaries and associated internal multiples and the down-going wavefield carries free-surface multiples and associated internal multiples;

FIG. 2 is a flowchart showing basic steps in a simultaneous inversion embodiment of the present inventive method;

FIG. 3 is a flowchart showing basic steps in one cascaded inversion embodiment of the present inventive method;

FIG. 4 is a flowchart showing basic steps in another cascaded inversion embodiment of the present inventive method; and

FIGS. 5A-5C are data displays showing test results for the present inventive method.

The invention will be described in connection with example embodiments. However, to the extent that the following detailed description is specific to a particular embodiment or a particular use of the invention, this is intended to be illustrative only, and is not to be construed as limiting the scope of the invention. On the contrary, it is intended to cover all alternatives, modifications and equivalents that may be included within the scope of the invention, as defined by the appended claims.

DETAILED DESCRIPTION OF EXAMPLE EMBODIMENTS

The invention is a method for using the full wavefield (primaries, internal multiples and free-surface multiples) in inversion. The marine seismic data are separated into up-going and down-going wavefields, which is possible for marine surveys because of the pressure data that can be collected underwater. Then, both wavefields are inverted in some joint manner, in which the final model is impacted by both wavefields. This may be achieved by inverting both wavefields simultaneously, or one after the other (in a cascaded approach) for the subsurface properties.

Wong et al. (2010) [3] disclosed a method for performing joint least-squares migration of up- and down-going wavefields. In contrast, the present invention inverts for the subsurface properties (by implementing a method of full waveform non-linear inversion). Inversion is quite different from migration. The objective in migration, i.e. imaging, is to move the reflectors to their correct subsurface positions, assuming a velocity model of the subsurface. The velocity model is not updated in migration. Inversion is a process, usually iterative, where the objective is to improve and update the initial assumed velocity model.

The present invention is designed for marine seismic data and requires acquisition of at least two-components of wavefield namely pressure and particle velocity. Any marine seismic acquisition can be categorized into one of the following three categories: towed streamer acquisition, ocean bottom cable (OBC) acquisition and ocean bottom node (OBN) acquisition. Towed streamer acquisitions have traditionally been used to record only the pressure component of the wavefield. In the past few years, however, it has become increasingly common to record both pressure and velocity components of the wavefield using a towed streamer acquisition system. OBC and OBN acquisitions almost always record pressure and velocity components of the wavefield.

Both pressure (P) and vertical velocity components (Z) have primaries, internal multiples and free-surface multiples. In principle, pressure and velocity components can be used together in a FWI method to estimate subsurface properties. However, as mentioned previously, accurate numerical modeling of free-surface multiples is challenging. Hence, any FWI method that requires accurate modeling of seismic data is prone to error if the free-surface multiples are used in inversion. To circumvent this problem, instead of using field-recorded pressure and velocity components, the present inventive method uses up-going and down-going wavefields in inversion.

There are several published methods available for deriving up-going and down-going wavefields from pressure and vertical velocity components. Some of those are described in [4, 5]. The up-going wavefield contains primaries and associated internal multiples while the down-going wavefield contains free-surface multiples and the associated internal multiples. (The terms up-going and down-going refer to the direction of the wavefield as it arrives at the receiver.) In terms of seismic inversion, both wavefields have their own advantages and disadvantages. Reference may be had to the self-explanatory schematic diagrams of FIG. 1. The up-going wavefields have larger reflection angle and usually have better signal-to-noise ratio. Down-going wavefields provide larger aperture but have less fold than the up-going wavefield. Since the up-going wavefield does not contain free-surface multiples, free-surface multiples do not need to be modeled in order to perform FWI. The down-going wavefield, which contains the free-surface multiples, can be modeled by using mirror geometry [6, 7] without needing a free-surface condition on the top of the model. In mirror geometry, the top of the model is padded (see FIG. 1) with a water layer as thick as the water depth and the receiver location is moved upward (directly above the original location) with a distance 2 g, where g is the depth of the original receiver location.

Up-going and down-going wavefields are used to perform FWI to invert for subsurface parameters. FWI may be performed using, for example, any of the three following approaches which are all embodiments of the present inventive method. The first four steps (21-24) are the same in each approach.

Simultaneous inversion: FIG. 2 describes basic steps in the workflow. Pressure and vertical velocity components of marine seismic data, including towed streamer, OBC, OBN/OBS (21), are decomposed (22) into up-going (23) and down-going (24) wavefields. These two wavefields are simultaneously inverted (25) for subsurface properties (26). A simultaneous inversion can be characterized by the optimizing of a single, combined objective function measuring misfit between simulated and actual data of both wavefields, resulting in a single model update, which updated model is then used in the next iteration cycle to simulate the up-going and down-going wavefields, etc.

Cascaded Inversion I: FIG. 3 describes the workflow. Pressure and vertical velocity components of marine seismic data, including towed streamer, OBC, and OBN/OBS (21), are decomposed (22) into up-going (23) and down-going (24) wavefields. Next, FWI is performed on up-going wavefield (35) to invert for subsurface properties, which may be called intermediate subsurface properties (36). Thereafter, FWI is applied on down-going wavefield (37) to invert for final subsurface properties (38) and the intermediate subsurface properties is used as the starting model for FWI.

Cascaded Inversion II: FIG. 4 describes the workflow. Pressure and vertical velocity components of marine seismic data, including towed streamer, OBC, and OBN/OBS (21), are decomposed (22) into up-going (23) and down-going (24) wavefields. Next, FWI is performed on down-going wavefield (45) to invert for subsurface properties, which may be called the intermediate subsurface properties (46). Thereafter, FWI is applied on up-going wavefield (47) to invert for final subsurface properties (48) and the intermediate subsurface properties is used as the starting model for FWI.

FIGS. 5A-5C show test results for the present inventive method applied to actual data. FIG. 5A shows the initial velocity model used to start the iterative inversion process. For complex scenarios such as the high velocity geo-body of this example, a reasonably close starting model is important to successful inversion. FIG. 5B shows the inverted velocity model using the simultaneous inversion method of FIG. 2. FIG. 5C shows the inverted velocity model using the cascaded inversion I method of FIG. 3. Both inverted models show many details not present in the starting model.

An alternative to the cascaded inversion embodiments described above might be to invert the up-going wavefield and the down-going wavefield separately, and then reduce the two resulting models to a single, best model by some sort of averaging process or least-squares fitting.

The foregoing description is directed to particular embodiments of the present invention for the purpose of illustrating it. It will be apparent, however, to one skilled in the art, that many modifications and variations to the embodiments described herein are possible. For example, an iterative inversion cycle where in the first cycle the up-going wavefield is inverted resulting in a model update, then use the updated model to invert the down-going wavefield, update the model again, then use that updated model to invert the up-going wavefield, and so on alternating the two wavefields from one cycle of the iterative process to the next. All such modifications and variations are intended to be within the scope of the present invention, as defined by the appended claims.

REFERENCES

1. Virieux, J., “P-SV wave propagation in heterogenous media,” Geophysics 51, 889-901 (1986).

2. Symes, W. W. and Vdovina, T., “Interface error analysis for numerical wave propagation,” Computational Geosciences 13, 363-371 (2009).

3. Wong, M., Biondi, B., Ronen, S., “Joint least-squares inversion of up- and down-going signal for ocean bottom data sets,” SEG Expanded Abstracts 29, 2752 (2010).

4. Barr, F. J. and Sanders, J. I., “Attenuation of water-column reverberations using pressure and velocity detectors in a water-bottom cable,” 59^th annual SEG meeting, Expanded Abstracts, 653 (1989).

5. Amundsen, L., “Elimination of free-surface related multiples with need of a source wavelet,” Geophysics 66, 327-341 (2001).

6. Godfrey, R. P. K., Armstrong, B., Cooper, M. and Thorogood, E., “Imaging the Foinaven ghost,” SEG Expanded Abstracts, 1333-1335 (1998).

7. Ronen, S., Comeaux, L. and Miao J., “Imaging downgoing waves from ocean bottom stations,” SEG Expanded Abstracts, 963-966 (2005).

Claims

1. A computer-implemented method for inverting marine seismic data to infer a subsurface physical property model, said seismic data including pressure data and vertical velocity data, said method comprising: separating the pressure data and the vertical velocity data into an up-going wavefield and a down-going wavefield; andinverting the up-going wavefield and the down-going wavefield together, either simultaneously or cascaded, to infer a subsurface physical property model;

2. The method of claim 1, wherein the inversion is cascaded, the up-going wavefield being inverted first to infer an intermediate physical property model, then, using the intermediate physical property model as an initial model, the down-going wavefield is inverted to infer a final physical property model.

3. The method of claim 1, wherein the inversion is cascaded, the down-going wavefield being inverted first to infer an intermediate physical property model, then, using the intermediate physical property model as an initial model, the up-going wavefield is inverted to infer a final physical property model.

4. The method of claim 1, wherein the physical property is velocity or acoustic impedance.

5. The method of claim 1, wherein the down-going wavefield is simulated in the inversion using mirror geometry.

6. A method of claim 1, wherein the pressure and vertical velocity data are full wavefield data, not processed to eliminate or reduce multiple reflections.

7. A computer-implemented method for inverting marine seismic data to infer a subsurface physical property model, said seismic data including pressure data and vertical velocity data, said method comprising: separating the pressure data and the vertical velocity data into an up-going wavefield and a down-going wavefield; anditeratively inverting the up-going wavefield and the down-going wavefield together, meaning that each iteration cycle inverts the up-going wavefield or the down-going wavefield or simultaneously inverts both up-going and down-going wavefields, with each iteration cycle resulting in a model update which is used to generate simulated data in a next iteration cycle, resulting finally in a final physical property model;wherein the separating and the inverting are performed using a computer.

8. The method of claim 7, wherein each inversion comprises using a current physical property model to simulate data, comparing the simulated data to corresponding actual data, measuring a degree of misfit, and using the misfit to generate an update to the current physical property model; wherein the up-going wavefield and the down-going wavefield are separately simulated.

9. The method of claim 8, wherein the down-going wavefield is simulated using mirror geometry.