Device and method for removal of multiples from seismic data vintages

Patent metadata

Field	Value
Publication number	US-9234976-B2
Application number	US-201313739107-A
Country	US
Kind code	B2
Filing date	Jan 11, 2013
Priority date	Jan 13, 2012
Publication date	Jan 12, 2016
Grant date	Jan 12, 2016

Abstract

A method, an apparatus and a computer readable medium for generating an image of a subsurface based on seismic data corresponding to at least two different times, for a same surveyed area are provided. A cost function, which is a sum over the seismic data vintages of a norm of differences between data and model predicted multiples, is minimized subject to minimizing residual multiples that are differences of corresponding multiples belonging to different vintages among the seismic data vintages.

First claim

What is claimed is:

1. A method for generating an image of a subsurface based on seismic data vintages, the method comprising:
collecting seismic data vintages corresponding to at least two different times, for a same subsurface;
minimizing in a data processing unit a cost function (CF), which is a sum, over a number of the seismic data vintages, of a norm of differences between each seismic data vintage (d i ) and a corresponding model predicted multiples (m i f i ), subject to minimizing residual multiples, wherein the residual multiples are differences of corresponding multiples belonging to different vintages among the seismic data vintages; and
generating the image of the subsurface based on a result of minimizing the cost function (CF).

2. The method of claim 1 , wherein the norm used in the cost function is one of an L2-norm, an L1-norm or a hybrid norm.

3. The method of claim 1 , wherein the minimizing of the differences between corresponding multiples includes minimizing a sum over combinations of the seismic data vintages of a norm of the differences between corresponding multiples in different seismic data vintages, the norm of the differences between corresponding multiples in different seismic data vintages and the norm used in the cost function being each one of a L2-norm, a L1-norm or a hybrid norm.

4. The method of claim 1 , wherein the model predicted multiples in a seismic data vintage (i) are obtained by multiplying a vintage specific model (m i ) with a vintage specific filter (f i ).

5. The method of claim 1 , wherein, in the cost function, weight matrices (W i ) multiply differences between the seismic data vintages and the model predicted multiples for each seismic data vintage (i).

6. The method of claim 1 , wherein the model predicted multiples are obtained using plural models and filters for each seismic data vintage.

7. The method of claim 1 , wherein the minimizing of the cost function, subject to the minimizing across the seismic data vintages of the differences between the corresponding multiples is performed by minimizing a function that combines the cost function with the minimizing across seismic data vintages condition by using Lagrange multipliers (λ).

8. The method of claim 1 , wherein models used to predict the multiples are data-driven models or model-driven models or a combination thereof.

9. The method of claim 1 , further comprising:
imaging differences between two of the seismic data vintages after correcting the seismic data vintages according to the minimizing.

10. The method of claim 1 , wherein the minimizing of the cost function, subject to the minimizing across the seismic data vintages of the differences between the corresponding multiples is performed in time domain, frequency domain, or tau-p domain.

11. The method of claim 1 , wherein the minimizing achieves de-noising and de-convolution of the data simultaneously.

12. The method of claim 1 , wherein the CF to be minimized is

CF
=

∑

i
=
1

No
.

⁢
Vintages

⁢

⁢



W
i

⁡

(

d
i

-

m
i

⁢

f
i

)



p
p

→
min

subject to minimal residual of the multiples from the 4D difference

∑

i
=
1

(

No
.

⁢
Vintages

-
1

)

⁢

⁢

∑

j
=

i
+
1

No
.

⁢
Vintages

⁢

⁢



residual
⁢

⁢
multiple



q
q

→
min

where q, a norm for the constraints and p the norm used for subtracting predicted multiples are any one of an L2-norm, an L1-norm or a hybrid norm.

13. A computer readable medium non-transitory storing executable codes that when executed on a computer make the computer perform a method for generating an image of a subsurface based on seismic data vintages, the method comprising:
collecting seismic data vintages corresponding to at least two different times, for a same subsurface;
minimizing a cost function, which is a sum, over a number i of the seismic data vintages, of a norm of differences between each seismic data vintage (d i ) and a corresponding model predicted multiples (m i f i ), subject to minimizing residual multiples, wherein the residual multiples are differences of corresponding multiples belonging to different vintages among the

Assignees

• Cggveritas Services Sa

Inventors

• Zabihi Naeini Ehsan
• Hoeber Henning

Classifications

• G01V1/38 Primary

  specially adapted for water-covered areas (G01V1/28 takes precedence) · CPC title

• G01V2210/56

  De-ghosting; Reverberation compensation · CPC title

• G01V1/364 Primary

  Seismic filtering (G01V1/37 takes precedence) · CPC title

• G01V1/36 Primary

  Effecting static or dynamic corrections on records, e.g. correcting spread; Correlating seismic signals; Eliminating effects of unwanted energy · CPC title