lsmf for suppressing multiples
DESCRIPTION
LSMF for Suppressing Multiples. Jianhua Yu. University of Utah. Contents. Motivation. LSMF Inversion. Numerical Examples. Conclusions. Contents. Motivation. LSMF Inversion. Numerical Examples:. Conclusions. Demultiple Methods. Radon transform. Inverse scattering theory. - PowerPoint PPT PresentationTRANSCRIPT
LSMF for Suppressing MultiplesLSMF for Suppressing Multiples
Jianhua YuJianhua Yu
University of UtahUniversity of Utah
Contents Contents
Motivation
LSMF Inversion
Numerical Examples
Conclusions
Contents Contents
Motivation
LSMF Inversion
Numerical Examples:
Conclusions
Demultiple Methods Demultiple Methods
Radon transformRadon transform
Inverse scattering theoryInverse scattering theory
Prediction+subtractionPrediction+subtraction
Benefits: Benefits:
Demultiple for coarse acquisition Demultiple for coarse acquisition geometrygeometry
Use both primary and multiple Use both primary and multiple informationinformation
Contents Contents
Motivation
LSMF Inversion
Numerical Examples:
Conclusions
Assuming that seismic data can be written mathematically as
D : seismic data
L : Primary forward operatorp
L : Multiple forward operatorm
R : Primary model p
R : Multiple model m
mmpp LLD RR
D : seismic dataobs
R : Primary model p
R : Multiple model m
LSMF equation (Nemeth, 1996) :
Minimize the misfit function||LLD||E mmppobs RR
d
T
T
TT
TT
m
p
mmpm
mppp
m
p
L
L
LLLL
LLLL
R
R1
LSMF Inversion:LSMF Inversion:
Algorithm: Conjugate Gradient (CG)
twtxRtxx pprs
p (),(),,( 0 d
Primary Modeling OperatorPrimary Modeling Operator
000 )),|,(),|,( dxdttxzxzxtx rrss
Wp
a weight
Rp
A primary model
dp
primary reflections
twtxRtxx mmmrs
m (),(),,( 0 d
Multiple Modeling OperatorMultiple Modeling Operator
000 )),|,(),|,( mmrr
mss
mm dxdttxzxzxtx
Wm
a weight
Rm
A multiple model
dm
multiples reflections
Operators for primary and Operators for primary and multiple migration are the multiple migration are the transpose of modeling operators transpose of modeling operators
Multiple initial modelMultiple initial model
Wang (Geophys, 2003)Wang (Geophys, 2003)
Demultiple Using LSMF Demultiple Using LSMF
Input CMP gathers
d
T
T
TT
TT
m
p
mmpm
mppp
m
p
L
L
LLLL
LLLL
R
R1
Solving the following equation by CG algorithm
Demultiple using LSMF Demultiple using LSMF
Predicted multiple M
Subtract multiple M from raw data D and get primary P
P=D-M
Contents Contents
Motivation
LSMF Inversion
Numerical Examples
Conclusions
Model
P+M P M
P+M P M
LSMFTim
e (s
)
P+M M P
CMP 300 (NS) T
ime
(s)
P+M M P
CMP 1700 (NS) T
ime
(s)
P+M M P
CMP 1300 (NS)T
ime
(s)
Before LSMF After LSMF
VelocityT
ime
(s)
CMP 1300 (NS)
Velocity
3.5
1.4 40
0
Before LSMF After LSMF
VelocityT
ime
(s)
CMP 1300 (NS)
Velocity
3.5
1.4 4
P+M M P
CMP 800 (Unocal)T
ime
(s)
P+M M P
CMP 900 (Unocal)T
ime
(s)
P+M M P
CMP 1100 (Unocal)T
ime
(s)
Tim
e (s
)
4
01.4 3.2 1.4 3.2
Velocity Velocity
Before LSMF After LSMF
Contents Contents
Motivation
LSMF Inversion
Numerical Examples:
Conclusions
Conclusions Conclusions
Works for synthetic data and real dataNo limit to coarse geometry
Straightforward to extend to 3D
Regularization strategy required for shallow reflections1-D model
ACKNOWLEDGMENTSACKNOWLEDGMENTS
2003 UTAM Sponsors
Unocal and Mobil for 2-D field data
CHPC