260 likes | 447 Views
Migration Deconvolution vs Least Squares Migration. Jianhua Yu, Gerard T. Schuster University of Utah. Outline. Motivation MD vs. LSM Numerical Tests Conclusions. Footprint. Migration noise and artifacts. Migration Noise Problems. Time. Aliasing. Recording footprints.
E N D
Migration Deconvolution vs Least Squares Migration Jianhua Yu, Gerard T. Schuster University of Utah
Outline • Motivation • MD vs. LSM • Numerical Tests • Conclusions
Footprint Migration noise and artifacts Migration Noise Problems Time
Aliasing Recording footprints Amplitude distortion Limited resolution Migration Problems
Improve resolution Suppress migration noise Computational cost Robustness Motivation Investigate MD and LSM:
Outline • Motivation • MD vs. LSM • Numerical Tests • Conclusions
-1 T T m = (L L ) Ld Reflectivity Seismic data Modeling operator Migration operator Least Squares Migration
-1 T T m = (L L ) Ld m’ Reflectivity Migrated data Modeling operator Migration Deconvolution
-1 T T LSM: m = (L L ) Ld T -1 m = (L L ) m’ MD: Migrated image Data Solutions of MD Vs. LSM
Relative samll cube I/O of 3-D MD Vs. LSM Huge volume LSM: MD:
Outline • Motivation • MD Vs. LSM • Numerical Tests • Conclusions
Numerical Tests • Point Scatterer Model • 2-D SEG/EAGE overthrust model poststack MD and LSM
Scatterer Model Krichhoff Migration 1.0 1.0 0 0 0 Depth (km) 1.8
MD LSM Iter=10 1.0 1.0 0 0 0 Depth (km) 1.8
LSM Iter=20 1.0 0 LSM Iter=15 1.0 0 0 Depth (km) 1.8
Numerical Tests • Point Scatterer Model • 2-D SEG/EAGE Overthrust Model Poststack MD and LSM
X (km) 0 7.0 0 KM Depth (km) 4.5 X (km) 0 7.0 0 LSM 15 4.5
X (km) 0 7.0 0 KM Depth (km) 4.5 X (km) 0 7.0 0 MD 4.5
X (km) 0 7.0 0 LSM 15 Depth (km) 4.5 X (km) 0 7.0 0 MD 4.5
Zoom View KM LSM 15 2 Depth (km) 3.5 LSM 19 MD 2 Depth (km) 3.5
Why does MD perform better than LSM ? X (km) 0 7.0 0 Depth (km) LSM 19 4.5 0 MD 4.5
Outline • Motivation • MD Vs. LSM • Numerical Tests • Conclusions
Function Resolution Performance MD < LSM (?) Conclusions Efficiency MD >> LSM Suppressing noise MD = LSM (?) Robustness MD < LSM
Acknowledgments • Thanks UTAM (http://utam.gg.utah.edu) sponsors for the financial support