Sparse hyperbolic inversion of seismic data
First Claim
1. A method for suppressing water-bottom multiples in a gather of seismic data traces, said method comprising the steps of:
- (a) inverting said gather of seismic data traces from the data domain to the model domain to obtain a model domain representation of said gather, each data sample in said model domain representation being defined in terms of a set of parameters including at least zero-offset traveltime, moveout velocity, two-way water-bottom reflection traveltime, and water-bottom reflection coefficient;
(b) determining a forward modeling matrix that relates the model domain to the data domain; and
(c) using said model domain representation and said forward modeling matrix to identify and remove said water-bottom multiples from said gather of seismic data traces.
2 Assignments
0 Petitions
Accused Products
Abstract
A method of sparse hyperbolic inversion is used to suppress multiples in marine seismic data. By explicitly including the two-way water-bottom reflection traveltime and water-bottom reflection coefficient as part of an augmented model domain, the inversion may be successfully carried out even for intermediate water depths and sloping water bottoms. Linear noises can also be suppressed by appropriate definition of the model. The invention also has the capability of handling time-varying wavelets or several different wavelets simultaneously. This latter capability is useful in suppressing dispersive ground roll on land seismic data.
62 Citations
22 Claims
-
1. A method for suppressing water-bottom multiples in a gather of seismic data traces, said method comprising the steps of:
-
(a) inverting said gather of seismic data traces from the data domain to the model domain to obtain a model domain representation of said gather, each data sample in said model domain representation being defined in terms of a set of parameters including at least zero-offset traveltime, moveout velocity, two-way water-bottom reflection traveltime, and water-bottom reflection coefficient;
(b) determining a forward modeling matrix that relates the model domain to the data domain; and
(c) using said model domain representation and said forward modeling matrix to identify and remove said water-bottom multiples from said gather of seismic data traces. - View Dependent Claims (2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17)
(a) modifying said forward modeling matrix to model only said water-bottom multiples;
(b) forward modeling said model domain representation of said gather to the data domain using said modified forward modeling matrix to obtain a data domain representation of said water-bottom multiples; and
(c) subtracting said data domain representation of said water-bottom multiples from said gather of seismic data traces.
-
-
9. The method of claim 8, wherein said forward modeling matrix contains an estimate of the seismic wavelet for said gather of seismic data traces.
-
10. The method of claim 9, wherein said estimate of the seismic wavelet is obtained from measurement of the actual seismic wavelet used to generate said seismic data traces.
-
11. The method of claim 9, wherein said estimate of the seismic wavelet is obtained from an isolated pulse in said seismic data traces.
-
12. The method of claim 9, wherein said estimate of the seismic wavelet is a time-varying wavelet.
-
13. The method of claim 9, wherein said estimate of the seismic wavelet comprises a set of wavelets.
-
14. The method of claim 1, wherein said step of inverting said gather of seismic data traces from the data domain to the model domain comprises an iterative optimization procedure.
-
15. The method of claim 14, wherein said iterative optimization procedure comprises the steps of:
-
(d) selecting an initial size for said model domain representation and an initial set of input data traces to be modeled;
(e) iteratively (i) generating synthetic data corresponding to said model domain representation, (ii) comparing said synthetic data to said input data traces using a residual error objective function, and (iii) updating said model domain representation, until the error calculated according to said residual error objective function is less than a preselected maximum error;
(f) iteratively (i) increasing the number of input data traces, (ii) reducing the size of said model domain representation, (iii) updating said residual error objective function, and (iv) repeating step (e), until all of said seismic data traces have been modeled.
-
-
16. The method of claim 15, wherein said residual error objective function is updated using an Lp-norm.
-
17. The method of claim 1, wherein said set of parameters includes water bottom dip.
-
18. A method for suppressing noise in a gather of seismic data traces, said method comprising the steps of:
-
(a) inverting said gather of seismic data traces from the data domain to the model domain to obtain a sparse model domain representation of said gather, each data sample in said sparse model domain representation being defined in terms of a set of parameters including at least zero-offset traveltime and stacking velocity, each primary reflection in said gather being represented by a single data sample in said sparse model domain representation;
(b) performing a filtering operation on said sparse model domain representation of said gather to suppress data samples relating to said noise; and
(c) forward modeling said filtered sparse model domain representation of said gather to the data domain. - View Dependent Claims (19, 20, 21, 22)
(d) selecting an initial size for said model domain representation and an initial set of input data traces to be modeled;
(e) iteratively (i) generating synthetic data corresponding to said model domain representation, (ii) comparing said synthetic data to said input data traces using a residual error objective function, and (iii) updating said model domain representation, until the error calculated according to said residual error objective function is less than a preselected maximum error;
(f) iteratively (i) increasing the number of input data traces, (ii) reducing the size of said model domain representation, (iii) updating said residual error objective function, and (iv) repeating step (e), until all of said seismic data traces have been modeled.
-
Specification