Transcript No Slide Title

Improve Migration Image Quality
by 3-D Migration Deconvolution
Jianhua Yu, Gerard T. Schuster
University of Utah
Outline
Motivation
Migration Deconvolution
Implementation of MD
Examples
Conclusions
Seismic Migration Noise
0.5
Footprint
Weak
illumination
Migration noise and artifacts
3.5
What Affects Seismic Migration Quality
Limit recording aperture
Irregular acquisition geometry
Bandlimited wavelet
Incorrect velocity
High order phenomenon:
anisotropy, attenuation etc.
Migration Image suffers from
Noise and artifacts
Poor spatial resolution
Non-uniform illumination
Objective :
Develop 3-D migration deconvolution
to deblur the influence of
Limit recording aperture
Irregular acquisition geometry
Objective :
Suppressing migration noise
and artifacts
Improving spatial resolution
Enhancing illumination
Outline
Motivation
Migration Deconvolution (MD)
Implementation of MD
Examples
Conclusions
Migration Deconvolution Theory
Migration operator
T
m = G G
d Rbut d = G R
Migrated Section
Data
m = PSF(R)
Migration image = Blurred image of
true reflectivity model
Migration Deconvolution Theory
Reflectivity
G G R=
T
=
Migration
Green’s
function
Migration image
m
Migration Deconvolution Theory
-1
-1
[G G] G G R = [G G] m
T
1
T
T
=
1
Migration Deconvolution Theory
-1
R = [G G] m
T
=
1

   
*   * 
mig (r ro ) =  G (rg r )G (r rs )G(rg ro )G(ro rs )drg drs
Migration Green’s function
(Schuster et al., 2000)
Migration Deconvolution Theory


 
m( r ) =  mig ( r ro ) R( ro )dro

   
*   * 
mig (r ro ) =  G (rg r )G (r rs )G(rg ro )G(ro rs )drg drs
Reduction of MD cost
Lateral shift invariant migration Green’s function

m(r ) =  mig ( x  xo , y  yo , z x p , y p , zo )
R( xo , yo , zo )dxodyodzo
( x p , y p ) --- Reference position of migration Green’s function
In wavenumber-space domain:
 

m = R
Outline
Motivation
Migration Deconvolution (MD)
Implementation of MD
Examples
Conclusions
MD Implementation Steps:
Step 1:
Prepare traveltime table
Acquisition
geometry
information
Step 2:
Calculate the migration Green’s function
at the depth Zi ( x, y, z j x p , y p , zi )
MD Implementation Steps:
Step 3: Obtain MD image at the depth Zi by
solving following equation
 

m = R
Step 4: Repeat Steps 2-3 until the maximum
depth is finished
Outline
Motivation
Migration Deconvolution (MD)
Implementation of MD
Examples: Synthetic data
Conclusions
Recording Geometry
: Sources
: Receivers
MIG
Depth Slices
0
0
MD
0
0
3 3
3 3
Z=1 km
0
0
0
0
3 3
3 3
Z=3 km
0
0
3 3
0
0
Z=5 km
3 3
MIG
Depth Slices
0
0
MD
0
0
3 3
3 3
Z=7 km
0
0
0
0
3 3
3 3
Z=9 km
0
0
3 3
0
0
Z=10 km
3 3
Meandering Stream Model
2.5 km 2.5 km
0
3.5 km
0
5 x 1 Sources;
11 x 7 Receivers
Z=3.5 KM
Model
Mig
0
MD
2.5 0
Y (km)
2.5
VSP Geometry: source 21 x 21; geophone: 12
Migration
Depth=1.75 km
MD
GOM Velocity Model
2
Depth (km)
0
12
X (km)
10
Migration
4
Depth (km)
8
10
X (km)
Migration+MD
10
4
X (km)
10
Migration
4
Depth (km)
8.5
10
X (km)
10
4
Migration+M
D X (km)
10
3-D SEG/EAGE Salt Model
12.2 km 12.2 km
0
0
Imaging:
dx=dy=20 m
9 x 5 Sources; 201 x 201 Receivers
dxshot=dyshot=1 km
3-D SEG/EAGE Salt Model
Y=7.12 km
Y (km)
X (km)
Mig
Y (km)
3
10
5
X (km)
z = 1.4 km
9.8
5
MD
X (km)
9.8
Mig (z=1.2 km)
Y (km)
3
10
5
X (km)
9.8
MD (z=1.2 km)
5
X (km)
9.8
Sigsbee2B Model
Depth (km)
3
10
0
X (km)
20
Depth (km)
2.5
10
0
X (km)
20
Mig
Depth (km)
2.5
10
0
X (km)
20
MD
Depth (km)
5
10
Mig.
MD
Outline
Motivation
Migration Deconvolution (MD)
Implementation of MD
Examples: 2-D field data
Conclusions
0
Time (s)
0
8
X (km)
6
PS PSTM Image ( by Unocal)
X (km)
0
Time (s)
0
8
PSTM(courtesy of Unocal)
6
PSTMD
MD
0
Time (s)
3 PSTM(courtesy of Unocal)
8
X (km)
6
PSTMD
MD
Time (s)
Mig (courtesy of Aramco)
MD
Time (s)
Mig (Courtesy of Aramco)
MD
Mig (Courtesy of Aramco)
MD
Outline
Motivation
Migration Deconvolution (MD)
Implementation of MD
Examples: 3-D field data
Conclusions
3D PSTM (courtesy of Unocal)
Crossline
Inline
1.6 s
MD
Crossline
3D PSTM (courtesy of Unocal)
2.0 s
MD
Mig in Inline (Courtesy of Unocal)
Times (s)
1.2
3.0
MD
Mig (Courtesy of Unocal)
Inline Number
90
1
Crossline Number
1
1
300
(2 kft)
MD
Inline Number
90
MD
Mig
Inline Number
90
1
Crossline Number
11
265
(3.6 kft)
Inline Number
90
Mig (courtesy of Unocal)
Inline Number
90
1
Depth (kft)
1.1
1
7.0
(Crossline=50)
MD
Inline Number
90
Mig (courtesy of Unocal)
Inline Number
90
1
Depth (kft)
1.1
1
8.0
(crossline 200)
MD
Inline Number
90
Outline
Motivation
Migration Deconvolution (MD)
Implementation of MD
Examples
Conclusions
Conclusions
Improve spatial resolution
Suppress migration noise
MD cost is related to acquisition
geometry
V(z) assumption for moderately
complex models
Acknowledgements
UTAM Sponsors
Aramco, BP and Unocal
SMAART Joint Venture