Methods and systems for seismic inversion and related seismic data processing

Patent metadata

Field	Value
Publication number	US-10353099-B2
Application number	US-201514922704-A
Country	US
Kind code	B2
Filing date	Oct 26, 2015
Priority date	Oct 24, 2014
Publication date	Jul 16, 2019
Grant date	Jul 16, 2019

Abstract

A method and system of processing seismic data is presented. The method may include, for each of a plurality of seismic traces, generating a respective intermediate set of reflectivity coefficients and a partial deconvolution of an estimated wavelet from the respective seismic trace. The method may also include decomposing a model into a plurality of orthogonal components, and projecting each of a plurality of eigenvectors corresponding to one of the orthogonal components onto intermediate reflectivity coefficients corresponding with all of the plurality of seismic traces at each of a plurality of times to generate a plurality of eigen-coefficients associated with each of the plurality of times. The eigen-coefficients may be used to generate a plurality of basis coefficients, which may then be used to generate a respective updated set of reflectivity coefficients for each of the seismic traces.

First claim

What is claimed is:

1. A method of processing seismic data, comprising:
receiving a plurality of seismic traces [S(t)] in a common midpoint (CMP) gather representing reflected seismic energy in a portion of ground;
for each of the plurality of seismic traces [S(t)] in the CMP gather, generating a respective intermediate set of reflectivity coefficients [fi(t)] associated with each respective seismic trace [Si(t)] based at least in part on a respective previous set of reflectivity coefficients [Ri(t)] associated with the respective seismic trace [Si(t)] and a partial deconvolution of an estimated wavelet [W(t)] from the respective seismic trace [Si(t)];
decomposing a model [G] into a plurality of orthogonal components [U, Σ, V], the model [G] relating reflectivity to seismic properties [vp, vs, ρ] corresponding to a plurality of angles [θ] associated with the plurality of seismic traces [Si(t)];
projecting each of a plurality of eigenvectors [um] corresponding to one of the orthogonal components [U] of the model [G] onto intermediate reflectivity coefficients [fi] corresponding with all of the plurality of seismic traces at each of a plurality of times [t] to generate a plurality of eigen-coefficients associated with each of the plurality of times [t];
soft-thresholding the plurality of eigen-coefficients associated with each of the plurality of times [t];
summing the plurality of eigen-coefficients for each of the plurality of times [t];
generating a plurality of basis coefficients [x(t)] for each of the plurality of times [t] by multiplying the sum of the plurality of eigen-coefficients with a basis vector [vm] corresponding to a second of the orthogonal components [U] of the model [G]; and
generating a respective updated set of reflectivity coefficients for each of the plurality of seismic traces [S(t)] using the plurality of basis coefficients [x(t)] and the model [G]; and
generating, based on the respective updated set of reflectivity coefficients, a seismic image of the portion of ground.

2. The method of claim 1 , further comprising:
normalizing the plurality of eigen-coefficients prior to said soft-thresholding by multiplication with an eigenvalue [σ] corresponding to the model [G].

3. The method of claim 1 , further comprising:
iteratively repeating said generating the respective intermediate sets of reflectivity coefficients, said projecting, said soft-thresholding, said summing, said generating the pluralities of basis coefficients, and said generating the updated sets of reflectivity coefficients, wherein the updated sets of reflectivity coefficients from each iteration are used as the respective previous sets of reflectivity coefficients in each subsequent iteration.

4. The method of claim 1 , wherein the estimated wavelet [W(ω)] is partially deconvolved from each respective seismic trace [Si(ω)] in a frequency domain to generate a frequency domain representation of the respective intermediate set of reflectivity coefficients [fi(ω)], and the respective intermediate set of reflectivity coefficients [fi(t)] is generated by transforming the frequency domain representation of the respective intermediate set of reflectivity coefficients [fi(ω)] into a time domain.

5. The method of claim 4 , wherein the partial deconvolution acts to reduce influence of the estimated wavelet [W(t)] from the respective seismic trace [Si(t)].

6. The method of claim 1 , further comprising:
for each of the plurality of seismic traces, generating a respective second intermediate set of reflectivity coefficients associated with each respective seismic trace based at least in part on a respective previous second set of reflectivity coefficients associated with the respective seismic trace and a partial deconvolution of the estimated wavelet from the respective seismic trace;
projecting each of the plurality of eigenvectors onto second intermediate reflectivity coefficients corresponding with all of the plurality of seismic traces at each of the plurality of times to generate a plurality of second eigen-coefficients associated with each of the plurality of times;
hard-thresholding the plurality of second eigen-coefficients associated with each of the plurality of times;
summing the plurality of second eigen-coefficients for each of the plurality of times;
generating a plurality of second basis coefficients for each of the plurality of times by multiplying the sum of the plurality of second eigen-coefficients with the basis vector; and
generating a respective second updated set of reflectivity coefficients for each of the plurality of seismic traces using the plurality of second basis coefficients and the model.

7. The method of claim 6 , further comprising:
iteratively repeating said generating the respective second intermediate sets of reflectivity coefficients, said projecting, said soft-thresholding, said summing, said generating the pluralities of second basis coefficients, and said generating the second updated sets of reflectivity coefficients, wherein the second updated sets of reflectivity coefficients from each iteration are used as the previous second set of reflectivity coefficients in each subsequent iteration.

8. The method of claim 6 , wherein the respective updated sets of reflectivity coefficients are used as the respective previous second sets of reflectivity coefficients.

9. The method of claim 6 , wherein said hard-thresholding acts to remove bias in the model.

10. The method of claim 1 , further comprising:
building a prior model based at least in part on the plurality of eigen-coefficients for each of the plurality of times, uncertainties associated with the updated sets of reflectivity coefficients, and prior statistical knowledge;
for each of the plurality of seismic traces, generating a respective Bayesian intermediate set of reflectivity coefficients based at least in part on the respective updated set of reflectivity coefficients associated with each respective seismic trace and a partial deconvolution of the estimated wavelet from the respective seismic trace; and
updating the prior model using a Bayesian least squares technique constrained by the uncertainties associated with the updated sets of reflectivity coefficients and the Bayesian intermediate sets of reflectivity coefficients.

11. The method of claim 10 , wherein said building of the prior model is an automated process.

12. The method of claim 1 , further comprising:
determining layer properties of a subsurface region corresponding to the CMP gather based at least in part on the plurality of basis coefficients for each of the plurality of times and the updated sets of reflectivity coefficients for each of the plurality of seismic traces.

13. The method of claim 12 , further comprising:
constraining the determination of the layer properties using a Bayesian least-squares technique considering uncertainties associated with the updated sets of reflectivity coefficients and prior statistical knowledge of the subsurface region.

14. The method of claim 12 , wherein the layer properties include one or both of P-impedance or S-impedance.

15. The method of claim 12 , further comprising detrending the layer properties based at least in part on a background model.

Assignees

• Ion Geophysical Corp

Inventors

• Sassen Douglas Spencer

Classifications

• G01V2210/67

  Wave propagation modeling · CPC title

• G01V1/368 Primary

  Inverse filtering · CPC title

• G01V2210/614

  Synthetically generated data · CPC title

• G01V1/282 Primary

  Application of seismic models, synthetic seismograms · CPC title

• G01V2210/23

  Wavelet filtering · CPC title