Methods and data processing apparatus for deblending seismic data

1. A deblending method for seismic data, the method comprising:

receiving

the seismic data recorded by a receiver detecting simultaneously reflections from an explored underground formation due to distinct signals emitted at different times, and

emitted signal information;

performing a first inversion of the seismic data to obtain an underground formation response in a transform domain, the first inversion using a source operator built based on the emitted signal information, and the first inversion being formulated to minimize a number of non-zero coefficients of a first inversion result expressed in a transform domain basis;

performing a second inversion of the seismic data seeking the underground formation response in a sub-domain of the transform domain, the second inversion also using the source operator, and the sub-domain being defined by vectors of the transform domain basis for which the first inversion result has the non-zero coefficients; and

using the underground formation response obtained in the second inversion, to extract a deblended dataset corresponding to one of the distinct signals, from the seismic data.

2. The method of claim 1 , wherein the transform domain is a curvelet domain.

3. The method of claim 1 , wherein the emitted signal information includes source signature, source activation time and source position for each of the distinct signals.

4. The method of claim 1 , wherein the first inversion iteratively determines a vector x in the transform domain which minimizes ∥Aϕx−b∥ 2 2 +λ∥x∥ 1 , where A is the source operator built based on the emitted signal information, ϕ is an inverse transform operator, b is the seismic data, and λ is a constraint weight.

5. The method of claim 2 , wherein the first inversion iteratively determines a vector x w in the transform domain which minimizes ∥Aϕx w −b∥ 2 2 +λ∥Wx w ∥ 1 , where W is a diagonal weight matrix distinguishing among curvelets, A is the source operator built using the emitted signal information, ϕ is an inverse transform operator, b is the seismic data, and λ is a constraint weight.

6. The method of claim 1 , wherein the first inversion is solved using a Fast Iterative Shrinkage-Thresholding Algorithm.

7. The method of claim 1 , wherein the second inversion iteratively determines a vector x s in the sub-domain, the vector x s minimizing ∥Aϕ s x s −b∥ 2 2 , where A is the source operator built based on the emitted signal information, ϕ s is an inverse transform operator in the sub-domain, and b is the seismic data.

8. The method of claim 1 , wherein second inversion is solved using a conjugate gradient algorithm.

9. The method of claim 1 , wherein sources generating the distinct signals are vibroseis, and the emitted signals information includes force measurements acquired when the sources emitted the distinct signals.

10. The method of claim 1 , wherein the using of the underground formation response to extract the deblended dataset includes:

generating a noise model;

obtaining correlated receiver data by correlating the seismic data with the emitted signal information; and

subtracting the noise model from the correlated receiver data.

11. The method of claim 1 , further comprising:

performing another first inversion on one of the deblended seismic dataset corresponding to the one of the distinct signals, the other first inversion being performed in the transform domain;

performing another second inversion on the one of the deblended seismic datasets in another sub-domain of the transform domain, the other sub-domain being defined by other vectors of the transform domain basis for which a result of the other first inversion has other non-zero coefficients; and

using a solution of the other second inversion to extract another deblended dataset corresponding to another one of the distinct signals, from the seismic data.

12. A seismic data processing apparatus, comprising:

an interface configured to receive

seismic data recorded by a receiver detecting simultaneously reflections from an explored underground formation due to distinct signals emitted at different times, and

emitted signals information; and

a data processing unit configured

to perform a first inversion of the seismic data seeking an underground formation response in a transform domain, the first inversion using a source operator built based on the emitted signal information, and the first inversion being formulated to minimize a number of non-zero coefficients of a first inversion result expressed in a transform domain basis,

to perform a second inversion of the seismic data seeking the underground formation response in a sub-domain of the transform domain, the second inversion also using the source operator, and the sub-domain being defined by vectors of the transform domain basis for which the first inversion result has the non-zero coefficients, and

to extract a deblended dataset corresponding to one of the distinct signals, from the seismic data, using the underground formation response obtained in the second inversion.

13. The seismic data processing apparatus of claim 12 , wherein the transform domain is a curvelet domain.

14. The seismic data processing apparatus of claim 12 , wherein the emitted signal information includes source signature, source activation time and source position for each of the distinct signals.

15. The seismic data processing apparatus of claim 12 , wherein the first inversion iteratively determines a vector x in the transform domain which minimizes ∥Aϕx−b∥ 2 2 +λ∥x∥ 1 , where A is the source operator built based on the emitted signal information, ϕ is an inverse transform operator, b is the seismic data, and λ is a constraint weight.

16. The seismic data processing apparatus of claim 13 , wherein, in the first inversion, the data processing unit iteratively determines a vector x w in the transform domain which minimizes ∥Aϕx w −b∥ 2 2 +λ∥Wx w ∥ 1 , where W is a diagonal weight matrix distinguishing among curvelets, A is the source operator built using the emitted signal information, ϕ is an inverse transform operator, b is the seismic data, and λ is a constraint weight.

17. The seismic data processing apparatus of claim 12 , wherein, in the second inversion, the data processing unit iteratively determines a vector x s in the sub-domain, the vector x s minimizing ∥Aϕ s x s −b∥ 2 2 , where A is the source operator built based on the emitted signal information, ϕ s is an inverse transform operator, and b is the seismic data.

18. The seismic data processing apparatus of claim 12 , wherein the data processing extracts the deblended dataset by:

generating a noise model;

obtaining correlated receiver data by correlating the seismic data with the emitted signal information; and

subtracting the noise model from the correlated receiver data.

19. The seismic data processing apparatus of claim 12 , wherein the data processing unit is further configured:

to perform another first inversion on the deblended dataset corresponding to the one of the distinct signals, the other first inversion being performed in the transform domain;

to perform another second inversion on the one of the at least one deblended dataset in another sub-domain of the transform domain, the other sub-domain being defined by other vectors of the transform domain basis for which a result of the other first inversion has other non-zero coefficients; and

to use a solution of the other second inversion to extract another a-deblended dataset corresponding another one of the distinct signals, from the seismic data.

20. A non-transitory computer readable medium storing executable codes which, when executed by a data processing unit make the data processing unit to perform a seismic data deblending method, the method comprising:

receiving

seismic data recorded by a receiver detecting simultaneously reflections from an explored underground formation due to distinct signals emitted at different times, and

emitted signal information;

performing a first inversion of the seismic data seeking an underground formation response in a transform domain, the first inversion using a source operator built based on the emitted signal information, and the first inversion being formulated to minimize a number of non-zero coefficients of a first inversion result expressed in a transform domain basis;

performing a second inversion of the seismic data seeking the underground formation response in a sub-domain of the transform domain, the second inversion also using the source operator, and the sub-domain being defined by vectors of the transform domain basis for which the first inversion result has the non-zero coefficients; and

using the underground formation response obtained in the second inversion, to extract a deblended dataset corresponding to one of the distinct signals, from the seismic data.