1 / 32

Multisource Least-squares Migration of Marine Data

Multisource Least-squares Migration of Marine Data. Xin Wang & Gerard Schuster Nov 7, 2012. Outline. Motivation Gain high quality image by least-squares migration Improve the efficiency by multisource technique Theory Multisource Kirchhoff migration Multisource least-squares migration

ezra
Download Presentation

Multisource Least-squares Migration of Marine Data

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Multisource Least-squares Migration of Marine Data Xin Wang & Gerard Schuster Nov 7, 2012

  2. Outline • Motivation • Gain high quality image by least-squares migration • Improve the efficiency by multisource technique • Theory • Multisource Kirchhoff migration • Multisource least-squares migration • Numerical Tests • Synthetic data test • Marine data test • Conclusion

  3. Motivation of LSM KM Image Least-squares Migration Image 3.4 Z (km) 0 3.4 Z (km) 0 0 X (km) 14 0 X (km) 14 Advantages of LSM compared to KM: Decrease artifacts, balance amplitudes, increase resolution, natural anti-alias and anti-gap filter (Nemeth,1999). Drawbacks of LSM High computational, IO and memory cost.

  4. Motivation of MLSM Multisource -> Crosstalk Multisource LSM Image Multisource KM Image • MLSM To: • Increase efficiency • Remove artifacts • Suppress crosstalk Problems of LSM High computational, IO and memory cost. Solution: Multisource phase-encoding technique.

  5. Outline • Motivation • Gain high quality image by least-squares migration • Improve the efficiency by multisource technique • Theory • Multisource Kirchhoff migration • Multisource least-squares migration • Numerical Tests • Synthetic data test • Marine data test • Conclusion

  6. Generate Supergather Data Supergather CSG 1 CSG 2 d2 d1 P2 d P1 shift 1 s Encoding Time (s) Time (s) shift -0.5 s 0 0 Depth (km) + =

  7. Decoding CSG 1 CSG 2 Supergather P1T P2T d d Decoding shift -1 s Time (s) Time (s) shift 0.5 s 0 0 Depth (km) -> +

  8. Multisource Migration CSG 1 CSG 2 L2TP2Td2 L2TP2Td1 L1TP1Td1 L1TP1Td2 Time (s) 0 Depth (km) migration artifacts crosstalk noise

  9. Multisource Least-squares Migration L = N1 L1+ N2 L2 & d = N1 d1 + N2 d2 Given:Lm = d Find: m s.t. min || Lm – d ||2 Solution: m = [ LT L ]-1LTd or if L is too big m(k+1) = m(k) – a LT ( Lm – d ) = m(k) – a ( L1T ( L1m – d1) + L2T ( L2m – d2) + L2TN2T N1(L1m – d1) + L1TN1T N2 ( L2m – d2) ) Crosstalk

  10. Workflow Modeling supergather d = L m(k) MLSM calculate data residual d - dobs dobs Save IO and memory cost calculate gradient LT (d – dobs) calculate analytical step length LSM calculate numerical step length dobs dobs More IO save Update reflectivity m(k+1) and refresh Applicable to GPU

  11. Outline • Motivation • Gain high quality image by least-squares migration • Improve the efficiency by multisource technique • Theory • Multisource Kirchhoff migration • Multisource least-squares migration • Numerical Tests • Synthetic data test • Marine data test • Conclusion

  12. Numerical Results Marmousi2 Velocity Model 3.4 Z (km) 0 km/s 4.8 # of shots: 192, ds = 60 m # of receiver: 120, dg = 20 m Streamer length: 2.4 km # of supergather: 16 0 X(km) 14 1.0 # of shots in supergather: 12 Synthetic data are generated with FD solution to acoustic wave-equation.

  13. KM vs LSM KM Image 3.4 Z (km) 0 0 X(km) 14 LSM Image, 15 Iterations 3.4 Z (km) 0 0 X(km) 14

  14. KM vs LSM (zoom view) Zoom view of red box Velocity Model Reflectivity Model 1.2 Z (km) 0.6 True LSM KM LSM Image, 15 Iterations KM Image 1.2 Z (km) 0.6 6 X (km) 7 6 X (km) 7

  15. KM vs LSM (zoom view) Zoom view of blue box Reflectivity Model Velocity Model 2.6 Z (km) 2.2 KM Image LSM Image, 15 Iterations 2.6 Z (km) 2.2 5.2 X (km) 7.2 5.2 X (km) 7.2

  16. MLSM (static and hybrid encoding) MLSM Image, Static Encoding, 15 Iterations 3.4 Z (km) 0 Static encoding: Same N for all iterations. 0 X(km) 14 MLSM Image, Hybrid Encoding, 15 Iterations 3.4 Z (km) 0 Hybrid encoding: Change N for every 5 iterations. 0 X (km) 14

  17. MLSM (dynamic encoding) vs LSM MLSM Image, Dynamic Encoding, 15 Iterations 3.4 Z (km) 0 Dynamic encoding: Change N for every iteration. 0 X (km) 14 LSM Image, 15 Iterations 3.4 Z (km) 0 0 X (km) 14

  18. MSLSM (dynamic encoding) vs LSM Zoom view of red box MLSM Image, Static, 15 Iters LSM Image, 15 Iters 1.2 Z (km) 0.6 MLSM Image, Hybrid, 15 Iters MLSM Image, Dynamic, 15 Iters 1.2 Z (km) 0.6 6 X (km) 7 6 X (km) 7

  19. MLSM (dynamic encoding) vs LSM Zoom view of blue box MLSM Image, Static, 15 Iters LSM Image, 15 Iters 2.6 Z (km) 2.2 MSLSM Image, Hybrid, 15 Iters MLSM Image, Dynamic, 15 Iters 2.6 Z (km) 2.2 5.2 X (km) 7.2 5.2 X (km) 7.2

  20. Comparison of CPU and IO Cost Assumption: conventional data cannot be stored in memory, but supergather data are small enough to be kept in the memory.

  21. Marine Data Velocity Model km/s 2.2 2.5 Z (km) 0 # of shots: 496, ds = 37.5 m # of receiver: 480, dg = 12.5 m Streamer length: 6 km # of supergather: 32 1.5 # of shots in supergather: 16 0 X(km) 18.8 Velocity model is from FWI. (Boonyasiriwat et al., 2010) A 10-15-70-75 Hz bandpass filter is applied. Source wavelet is generated from stacking near offset ocean bottom reflections.

  22. Marine Data Test KM Image 1.88 Z (km) 0.6 5 X(km) 13.8

  23. Marine Data Test RTM Image 1.88 Z (km) 0.6 5 X(km) 13.8

  24. Marine Data Test LSM image, 30 iterations 1.88 Z (km) 0.6 5 X(km) 13.8

  25. Marine Data Test MLSM image, dynamic encoding, 50 iterations 1.88 Z (km) 0.6 5 X(km) 13.8

  26. KM vs RTM vs LSM vs MSLSM Zoom view of red box RTM Image KM Image 1.5 Z (km) 0.9 MLSM Image, Dynamic, 20 Iters LSM Image, 20 Iters 1.5 Z (km) 0.9 10.5 X (km) 11.5 10.5 X (km) 11.5

  27. KM vs RTM vs LSM vs MLSM KM Image 1.5 Z (km) 0.9 10.5 X (km) 11.5

  28. KM vs RTM vs LSM vs MLSM RTM Image 1.5 Z (km) 0.9 10.5 X (km) 11.5

  29. KM vs RTM vs LSM vs MLSM MLSM Image, Dynamic, 20 Iterations 1.5 Z (km) 0.9 10.5 X (km) 11.5

  30. Outline • Motivation • Gain high quality image by least-squares migration • Improve the efficiency by multisource technique • Theory • Multisource Kirchhoff migration • Multisource least-squares migration • Numerical Tests • Synthetic data test • Marine data test • Conclusion

  31. Conclusion • MLSM can improve image quality over conventional Migration. KM Image MLSM Image • MLSM can improve the efficiency, and save IO cost. • Dynamic MLSM reduces IO cost to 1/30. • Wave equation based migration method can save the computational cost.

  32. Acknowledgement We thank for the 2011 CSIM sponsors for their financial support.

More Related