Migration Deconvolution vs. Least Squares
Migration
Jianhua YuUniversity of Utah
OutlineOutline• MotivationMotivation
• MD vs. LSMMD vs. LSM
• Numerical TestsNumerical Tests
• ConclusionsConclusions
Migration Noise ProblemsMigration Noise Problems
Migration noise and artifacts
Footprint Amplitude distortion
Migration ProblemsMigration Problems
AliasingAliasing
Limited ResolutionLimited Resolution
MotivationMotivation
Investigate MD and LSM:
Improving resolution
Suppressing migration noiseComputational cost
Robustness
OutlineOutline• MotivationMotivation
• MD vs. LSMMD vs. LSM
• Numerical TestsNumerical Tests
• ConclusionsConclusions
m = (m = (L L L L )) L L ddTTTT -1
Least Squares Migration
Reflectivity
Modeling operator
Seismic data
Migration operator
TTmm = ( = (L LL L ) m’ ) m’
-1-1
ReflectivityReflectivity
MD deblurring operator
Migration SectionMigration Section
Migration Deconvolution
Solutions of MD vs. LSMSolutions of MD vs. LSM
m = (m = (L L L L )) L L ddTTTT -1LSM:
TTmm = ( = (L LL L ) ) mm’’
-1-1 MD:
Migrated image
Data
OutlineOutline• MotivationMotivation
• MD vs. LSMMD vs. LSM
• Numerical TestsNumerical Tests
• ConclusionsConclusions
Numerical TestsNumerical Tests
• Point Scatterer ModelPoint Scatterer Model
• 2-D SEG/EAGE overthrust model 2-D SEG/EAGE overthrust model poststack MD and LSMpoststack MD and LSM
Scatterer Model Kirchhoff MigrationD
epth
(k
m)
1.8
01.00 1.00
MD LSM Iter=15D
epth
(k
m)
1.8
01.00 1.00
• Point Scatterer ModelPoint Scatterer Model
• 2-D SEG/EAGE Overthrust Model 2-D SEG/EAGE Overthrust Model Poststack MD and LSMPoststack MD and LSM
Numerical TestsNumerical Tests
KM
Dep
th (
km
)
4.5
00 7.0
0 7.0
X (km)
X (km)
4.5
0
LSM 10
KM
Dep
th (
km
)
4.5
00 7.0
0 7.0
X (km)
X (km)
4.5
0
LSM 15
KM
Dep
th (
km
)
4.5
00 7.0
0 7.0
X (km)
X (km)
4.5
0
MD
Dep
th (
km
)
4.5
00 7.0
0 7.0
X (km)
X (km)
4.5
0
MD
LSM 15
LSM 15
MD
KM2
3.5
Dep
th (
km
)
LSM 192
3.5
Dep
th (
km
)Zoom View
Dep
th (
km
)
4.5
00 7.0
Why does MD perform better than LSM ?
4.5 MD
LSM 19
0
X (km)
OutlineOutline• MotivationMotivation
• MD vs. LSMMD vs. LSM
• Numerical TestsNumerical Tests
• ConclusionsConclusions
ConclusionsConclusions
Efficiency MD >> LSM
FunctionFunction PerformancPerformanceeResolutionResolution MD = LSMMD = LSM
.
Suppressing noise MD > LSM
Robustness MD < LSM
AcknowledgmentsAcknowledgments
• Thanks to 2001 UTAM sponsors Thanks to 2001 UTAM sponsors for their financial supportfor their financial support