Professional Documents
Culture Documents
2 n 1
dl
l
X
i
r
Minimize T ( x, y , z )
Xs
X v( x, y, z )
v
i
2
i
s
l2
dl is the differential
1
Minimize f (v ) || Tobs T (v ) ||22
2
l vu
l 2j
l5j
j
3
l 4j
li
T j (v )
i 2 v i
f ( v ) J T ( v )T (T ( v ) Tobs )
1
p ' (v), p (v)
2
where,
Vp
is a
1 p ( x, t ; xs )
( x).
p ( x , t ; x s ) s ( x , t ; x s )
2
2
v( x)
t
( x)
p ( x,0; xs ) 0
p ( x, t ; xs )
p ( x,0; xs )
0
t
is a function described by
the source, (x ) is the density
of the medium.The full
wave equation is solved with
the staggered-grid finite
difference scheme (Luo and
Schuster 1991, Savic 1995)
v cte.
v
2
v
(
x
,
y
,
z
)
3
x
z
For example:
v2
vi
li
( zi' ) 2
( xi' ) 2
( yi' ) 2
,
2
2
2
((v z , j ) i )
((v x , j ) i )
((v y , j ) i )
Xr
Xs
( X ,Y , Z )
2 n 1
i 2
(zi ' ) 2
(xi ' ) 2
(yi ' ) 2
2
2
((vZ , j ) i ) ((v x , j ) i ) ((v y , j ) i ) 2
where,
H. Jacob (1970)
Ivansson (1983)
Using multiscale
descomposition techniques
for findind long wavelength
components first and then
recursively refine them to get
shorter scales.
Kolb, Collino and Lailly (1986)
Pica, Diet and Tarantola (1990)
Bunks et al. (1995)
1. Given
2. If || P( x
x0 n
, 0
g k ) xk || 0
3. Compute
l xu
, stop
and set
f ( xk j ) g kT d k
:
, then
k , xk 1 xk k d k , yk g k 1 g k , sk xk 1 xk
5.
f (x)
and M 0
d k P ( xk k g k ) xk
4. If f ( xk 1 ) max
Min
s. t.
go to step 5
skT sk
k 1 T
sk y k
n
Where: P is the projection on {x / l x u} and
g k f ( xk )
5 layer synthetic model where P-S converted waves velocities are considered
1.
2.
3.
Convergence to the
global minimum is
obtained.
5 layer synthetic model where P-S converted wave velocities are considered
Cores, Fung and Michelena, A fast and global two point low storage optimization technique for tracing rays
in 2D and 3D isotropic media, Journal of Applied geophysics 45, 273-278, 2000.
2.
Lateral heterogeneous
model :
v( x, y ) ax by c,
a (0,1.7,1.5,1.3,0.8,0.8,1.3,1.5,1.7)T ,
b (0,1,1,1,1,1,1,1,1)T ,
c (0,800,700,500,150,150,500,700,800)T
3.
4.
,
vi li ((v z , j ) i ) 2 ((v x , j ) i ) 2 ((v y , j ) i ) 2
xi' xi cos( i ) cos(i ) yi sin( i ) cos( i ) zi sin( i ),
yi' xi sin( i ) yi cos(i ),
zi' xi sin( i ) cos(i ) yi sin(i ) sin( i ) zi cos( i ),
Where i
(v z , p ) i 1500 100 * i m / s,
(v x , p ) i 1200 50 * i m / s,
(v y , p ) i 1350 80 * i m / s,
(v z , s ) i 1400 100 * (n 3 i ) m / s,
(v x , s ) i 1000 50 * (n 3 i ) m / s,
(v y , s ) i 1150 80 * (n 3 i ) m / s
for i=2,...,n+1
We used a 20x20
grid size to measure
Real velocities
Initial velocities
The initial velocities have an error of 50% from the real velocities
The quality of the solution by the 2 methods are almost the same
Castillo, Cores and Raydan, Low cost optimization techniques for solving the nonlinear seismic reflection
tomography, Optimization and Engineering 1, 155-169, 2000.
0.2km ( vZ , j )i ( v z )i 5km
0.2km (v y , j ) i (v y )i 5km
0.2km (v x , j ) i (v x ) i 5km
10 i 30
2 i 9
Stopping criteriun:
P ( X k f ( X k )) X k
M=8 (SPG)
1006
Square mesh
Radial mesh
i
2
3
4
5
6
7
( v x )i
1.5
2
3
2.7
1.8
1.3
(v )
( v y )i
(v y )iap
1.5
1.91
2.86
2.83
1.89
1.29
1.7
2.3
2.8
2.9
2
1.6
1.69
2.22
2.85
2.85
2.07
1.61
ap
x i
( v z )i
1.9
2.5
3.3
3.1
2.3
1.8
(v z ) iap
(v x ) iap
(v y ) iap
(v z ) iap
1.86
2.37
3.21
3.19
2.42
1.84
1.49
1.81
3.01
2.73
2.01
1.29
1.69
2.06
2.81
2.88
2.27
1.6
1.9
2.21
3.55
2.91
2.59
1.81
2
3
4
5
6
7
5
7
3
3
7
5
iap
5.84
7.36
6
5.78
7.47
5.01
20
15
25
25
15
20
ap
i
21.49
15.46
20.31
19.14
14.29
18.94
.
iap
8.46
4.93
2
6
4.47
7.64
iap
20.1
13.46
25.13
20.43
12
19.86
The SPG optimization method gets a good precision for estimating the velocities
using small number of rays.
The problem for obtaining a better estimate of the polar angle vector is not the
optimization scheme used, it depends on the seismic data acquisition.
Increasing the number of rays, the error in the velocity vector and in the azimuthal
angle vector can be reduced, but the CPU time increase.
None of the mesh distribution used here give enough information for obtaining a
good estimate of the polar angle vector ( ). May be the travel time information is
not appropiate for estimating fracture orientation.
Meza and Cores, Seismic velocity estimation and fracture orientation in orthorombic media, in
preparation
Zeev, Savasta and Cores, Non monotone spectral projeted gradient method applied to full
waveform inversion, Geophysical Prospecting 54, 1-10, 2006.
Conclusions
Conclusions
Ellipsoidal Velocity
Solving the eigenvalue problem: (G (1 , 2 ) WI )U 0
Where 1 and 2 are the polar and azimuthal phase angles.
Gik
j ,l 1
ijkl j l
1 cos( 2 ) sin(1 )
2 sin( 2 ) sin(1 )
3 sin(1 )
Ellipsoidal Velocity
Approximating the eigenvalues of the Christoffel equation
and using the Byun Transformation, Contreras et al. in
1997 obtained an ellipsoidal group velocity:
1
1
1
2
2
2
cos
(
cos
(
)
sin
(1i )
1i
2i
2
2
2
( vi )
(( vZ , j )i )
(( v NMO , j )i )
[ X ;Z ]
1
(( v NMO[Y ;Z ] , j )i )
vi
2
2
sin
(
)
sin
(1i )
2i
2
Ellipsoidal velocity
f i ( xi , yi ) f i 1 ( xi 1, yi 1 )
cos(1i )
li
( xi xi 1 )2 ( yi yi 1 ) 2
sin(1i )
li
sin(2i )
cos(2i )
yi yi 1
( xi xi 1 ) 2 ( yi yi 1 ) 2
xi xi 1
( xi xi 1 ) 2 ( yi yi 1 ) 2
cos i
Rp 0
sin
i
cos i
Ra sin i
0 sin i
1
0
0 cos i
sin i
cos i
0
0
1
Ellipsoidal Velocity
For j=P,SV,SH and
2
i=2,,2n+1
2
1 1
zi
xi
yi2
2
2
2
vi li (( vZ , j )i )
(( v NMO[ XZ ] , j )i )
(( v NMO[YZ ] , j )i )
where,
yi yi yi 1
zi zi zi 1
xi xi xi 1
2.
3.
The step size k is not the classical choice for the steepest
descent method. It speeds up the convergence of the PSG
method.
4.
5.
6.
Conclusions
1.
SIRT has low computational cost per iteration but requires too many
iterations and therefore consumes more CPU time.
2.
PSG, PR+ and CONMIN reach quickly a good precision (10e-03) when
compared to SIRT and Gauss Newton methods.
3.
Gauss Newton is fast, in CPU time, for very small size of the grid.
4.
The PSG and PR+ methods outperform CONMIN for very large
problems.
5.
The PSG method is always slightly faster , in CPU time, than PR+.