This action might not be possible to undo. Are you sure you want to continue?

**estimation using time-migration velocity
**

Maria Cameron

∗

, Sergey Fomel

†

, and James Sethian

‡

ABSTRACT

The objective of this work is to build an eﬃcient algorithm (a) to estimate seismic

velocity from time-migration velocity, and (b) to convert time-migrated images to

depth. We establish theoretical relations between the time-migration velocity and

the seismic velocity in 2-D and 3-D using paraxial ray tracing theory. The relation

in 2-D implies that the conventional Dix velocity is the ratio of the interval seismic

velocity and the geometrical spreading of the image rays. We formulate an inverse

problem of ﬁnding seismic velocity from the Dix velocity and develop a numerical

procedure for solving it. This procedure consists of two steps: (1) computation

of the geometrical spreading of the image rays and the true seismic velocity in

the time-domain coordinates from the Dix velocity; (2) conversion of the true

seismic velocity from the time domain to the depth domain and computation of

the transition matrices from time-domain coordinates to depth.

For step 1, we derive a partial diﬀerential equation (PDE) in 2-D and 3-D re-

lating the Dix velocity and the geometrical spreading of the image rays to be

found. This is a nonlinear elliptic PDE. The physical setting allows us to pose

a Cauchy problem for it. This problem is ill-posed. However we are able to

solve it numerically in two ways on the required interval of time. One way is a

ﬁnite diﬀerence scheme inspired by the Lax-Friedrichs method. The second way

is a spectral Chebyshev method. For step 2, we develop an eﬃcient Dijkstra-like

solver motivated by Sethian’s Fast Marching Method.

We test our numerical procedures on a synthetic data example and apply them to

a ﬁeld data example. We demonstrate that our algorithms give signiﬁcantly more

accurate estimate of the seismic velocity than the conventional Dix inversion.

Our velocity estimate can be used as a reasonable ﬁrst guess in building velocity

models for depth imaging.

Seismic Velocity Estimation

Cameron, Fomel, Sethian 2 Velocity estimation from time migration

INTRODUCTION

Time-domain seismic imaging is a robust and eﬃcient process routinely applied to

seismic data (Yilmaz, 2001; Robein, 2003). Rapid scanning and determination of

time-migration velocity can be accomplished either by repeated migrations (Yilmaz

et al., 2001) or by velocity continuation (Fomel, 2003). Time migration is considered

adequate for seismic imaging in areas with mild lateral velocity variations. However,

even mild variations can cause structural distortions of time-migrated images and

render them inadequate for accurate geological interpretation of subsurface structures.

To remove structural errors inherent in time migration, it is necessary to convert

time-migrated images into the depth domain either by migrating the original data

with a prestack depth migration algorithm or by depth migrating post-stack data

after time demigration (Kim et al., 1997). Each of these options requires converting

the time migration velocity to a velocity model in depth.

The connection between the time- and depth-domain coordinates is provided by

the concept of image ray, introduced by Hubral (1977). Image rays are seismic rays

that arrive normal to the Earth’s surface. Hubral’s theory explains how a depth ve-

locity model can be converted to the time coordinates. However, it does not explain

how a depth velocity model can be converted to the time-migration velocity. More-

over, image-ray tracing is a numerically inconvenient procedure for achieving uniform

coverage of the subsurface. This may explain why simpliﬁed image-ray-tracing algo-

rithms (Larner et al., 1981; Hatton et al., 1981) did not ﬁnd widespread application in

practice. Other limitations of image rays are related to the inability of time migration

to handle large lateral variations in velocity (Bevc et al., 1995; Robein, 2003).

The objective of the present work is to ﬁnd an eﬃcient method for building a

velocity model from time-migration velocity. We establish new ray-theoretic connec-

tions between time-migration velocity and seismic velocity in 2-D and 3-D. These

results are based on the image ray theory and the paraxial ray tracing theory (Popov

and Pˇsenˇcik, 1978;

ˇ

Cerven´ y, 2001; Popov, 2002). Our results can be viewed as an

extension of the Dix formula (Dix, 1955) to laterally inhomogeneous media. We show

that the Dix velocity is seismic velocity divided by the geometrical spreading of the

image rays. Hence, we use the Dix velocity instead of time migration velocity as a

more convenient input. We develop a numerical approach to ﬁnd (a) seismic velocity

from the Dix velocity, and (b) transition matrices from the time-domain coordinates

to the depth-domain coordinates. We test our approach on synthetic and ﬁeld data

examples.

Our approach is complementary to more traditional velocity estimation methods

and can be used as the ﬁrst step in a velocity model building process.

Seismic Velocity Estimation

Cameron, Fomel, Sethian 3 Velocity estimation from time migration

TIME MIGRATION VELOCITY

Kirchhoﬀ prestack time migration is commonly based on the following travel time

approximation (Yilmaz, 2001). Let s be a source, r be a receiver, and x be the

reﬂection subsurface point. Then the total travel time from s to x and from x to r is

approximated as

T(s, x) + T(x, r) ≈

ˆ

T(x

0

, t

0

, s, r) (1)

where x

0

and t

0

are eﬀective parameters of the subsurface point x. The approximation

ˆ

T usually takes the form of the double-square-root equation

ˆ

T(x

0

, t

0

, s, r) =

t

2

0

+

|x

0

−s|

2

v

2

m

(x

0

, t

0

)

+

t

2

0

+

|x

0

−r|

2

v

2

m

(x

0

, t

0

)

, (2)

where x

0

and t

0

are the escape location and the travel time of the image ray (Hubral,

1977) from the subsurface point x. Regarding this approximation, let us list four

cases depending on the seismic velocity v and the dimension of the problem:

2-D and 3-D, velocity v is constant. Equation 2 is exact, and v

m

= v.

2-D and 3-D, velocity v depends only on the depth z. Equation 2 is a conse-

quence of the truncated Taylor expansion for the travel time around the surface

point x

0

. Velocity v

m

depends only on t

0

and is the root-mean-square velocity:

v

m

(t

0

) =

1

t

0

t

0

0

v

2

(z(t))dt. (3)

In this case, the Dix inversion formula (Dix, 1955) is exact. We formally deﬁne

the Dix velocity v

Dix

(t) by inverting equation 3, as follows:

v

Dix

(t) =

d

d t

0

(t

0

v

2

m

(t

0

)) . (4)

2-D, velocity is arbitrary. Equation 2 is a consequence of the truncated Taylor

expansion for the travel time around the surface point x

0

. Velocity v

m

(x

0

, t

0

) is

a certain kind of mean velocity, and we establish its exact meaning in the next

section.

3-D, velocity is arbitrary. Equation 2 is heuristic and is not a consequence of

the truncated Taylor expansion. In order to write an analog of travel time

approximation 2 for 3-D, we use the relation (Hubral and Krey, 1980)

Γ = [v(x

0

)R(x

0

, t

0

)]

−1

, (5)

where Γ is the matrix of the second derivatives of the travel times from a

subsurface point x to the surface, R is the matrix of radii of curvature of the

emerging wave front from the point source x, and v(x

0

) is the velocity at the

Seismic Velocity Estimation

Cameron, Fomel, Sethian 4 Velocity estimation from time migration

surface point x

0

. For convenience, we prefer to deal with matrix K ≡ Γ

−1

,

which, according to equation 5 is

K(x

0

, t

0

) ≡ v(x

0

)R(x

0

, t

0

). (6)

The travel time approximation for 3-D implied by the Taylor expansion is

ˆ

T(x

0

, t

0

, s, r)

=

t

2

0

+ t

0

(x

0

−s)

T

[K(x

0

, t

0

)]

−1

(x

0

−s) (7)

+

t

2

0

+ t

0

(x

0

−r)

T

[K(x

0

, t

0

)]

−1

(x

0

−r).

The entries of the matrix

1

t

0

K(x

0

, t

0

) have dimension of squared velocity and

can be chosen optimally in the process of time migration. It is possible to show,

however, that one needs only the values of

det

∂

∂t

0

K(x

0

, t

0

)

(8)

to perform the inversion. This means that the conventional 3-D prestack time

migration with traveltime approximation 2 provides suﬃcient input for our in-

version procedure in 3-D. The determinant in equation 8 is well approximated

by the square of the Dix velocity obtained from the 3-D prestack time migration

using the approximation given by equation 2.

One can employ more complex and accurate approximations than the double-square-

root equations 2 and 7, i.e. the shifted hyperbola approximation (Siliqi and Bousqui´e,

2000). However, other known approximations also involve parameters equivalent to

v

m

or K.

SEISMIC VELOCITY

In this section, we will establish theoretical relationships between time-migration

velocity and seismic velocity in 2-D and 3-D.

The seismic velocity and the Dix velocity are connected through the quantity Q,

the geometrical spreading of image rays. Q is a scalar in 2-D and a 2 × 2 matrix in

3-D. The simplest way to introduce Q is the following. Trace an image ray x(x

0

, t).

x

0

is the starting surface point, t is the traveltime. Call this ray central. Consider

a small tube of rays around it. All these rays start from a small neighborhood dx

0

of the point x

0

perpendicular to the earth surface. Thus, they represent a fragment

of a plane wave propagating downward. Consider the fragment of the wave front

deﬁned by this ray tube at time t

0

. Let dq be the fragment of the tangent to the

front at the point x(x

0

, t

0

) reached by the central ray at time t

0

, bounded by the

ray tube (Figure 1). Then, in 2-D, Q is the derivative Q(x

0

, t

0

) =

dq

dx

0

. In 3-D, Q

is the matrix of the derivatives Q

ij

(x

0

, t

0

) =

dq

i

dx

0j

, i, j = 1, 2, where derivatives are

Seismic Velocity Estimation

Cameron, Fomel, Sethian 5 Velocity estimation from time migration

Figure 1: Illustration for the deﬁnition of geometrical spreading.

taken along certain mutually orthogonal directions e

1

, e

2

(Popov and Pˇsenˇcik, 1978;

ˇ

Cerven´ y, 2001; Popov, 2002).

The time evolution of the matrices Q and P is given by

d

dt

Q

P

=

0 v

2

0

I

−

1

v

0

V 0

Q

P

, (9)

where v

0

it the velocity at the central ray at time t, V =

∂

2

v

∂q

i

∂q

j

i,j=1,2

, and I is the

2 × 2 identity matrix. The absolute value of det Q has a simple meaning: it is the

geometrical spreading of the image rays (Popov and Pˇsenˇcik, 1978;

ˇ

Cerven´ y, 2001;

Popov, 2002). The matrix Γ, introduced in the previous section, relates to Q and P

as Γ = PQ

−1

. Hence, K = QP

−1

.

In (Cameron et al., 2007), we have proven that

v

Dix

(x

0

, t

0

) ≡

∂

∂t

0

(t

0

v

2

m

(x

0

, t

0

)) =

v(x(x

0

, t

0

), z(x

0

, t

0

))

|Q(x

0

, t

0

)|

(10)

in 2-D, where v

m

(x

0

, t

0

) is the time-migration velocity, and

∂

∂t

0

(K(x

0

, t

0

)) = v(x(x

0

, t

0

))

Q(x

0

, t

0

)Q

T

(x

0

, t

0

)

−1

(11)

in 3-D, K is deﬁned by equation 6 and can be determined from equation 7.

Seismic Velocity Estimation

Cameron, Fomel, Sethian 6 Velocity estimation from time migration

PARTIAL DIFFERENTIAL EQUATIONS FOR THE

GEOMETRICAL SPREADING OF IMAGE RAYS

In this section, we derive the partial diﬀerential equations for Q in 2-D and 3-D.

From now on, we will denote the square of the Dix velocity by f in 2-D and the

corresponding matrix by F in 3-D, to avoid the subscript:

F ≡

∂

∂t

0

(K(x

0

, t

0

)) . (12)

Furthermore, we imply that t

0

denotes the one-way traveltime along the image rays.

Finally, we assume that our domain does not contain caustics, i.e., the image rays do

not cross on the interval of time we consider.

2-D case

Consider a set of image rays coming to the surface. Suppose we are tracing them

all backwards in time together with the quantities Q and P. Let us eliminate the

unknown velocity v in system 9 using equation 10. Moreover, let us eliminate the

diﬀerentiation in q using the deﬁnition of Q and rewrite it in the time-domain coor-

dinates x

0

, t

0

). Indeed, Q =

dq

dx

0

, hence

d

dq

=

d

dx

0

dx

0

dq

= Q

−1 d

dx

0

. Therefore, system 9

becomes

Q

t

0

= (fQ)

2

P, P

t

0

= −

1

fQ

(fQ)

x

0

Q

x

0

. (13)

Eliminating P in system 13, we get the following partial diﬀerential equation (PDE)

for Q

Q

t

0

f

2

Q

2

t

0

= −

1

fQ

(fQ)

x

0

Q

x

0

. (14)

The initial conditions are Q(x

0

, 0) = 1, Q

t

0

(x

0

, 0) = 0. Equation 14 simpliﬁes in

terms of the negative reciprocal of Q. Introduce y = −

1

Q

. Then equation 14 becomes

y

t

0

f

2

t

0

=

y

f

f

y

x

0

y

x

0

. (15)

In the expanded form, equation 15 is

y

t

0

t

0

f

2

−2

y

t

0

f

t

0

f

3

= y

f

x

0

x

0

f

−y

x

0

f

x

0

f

−y

x

0

x

0

+

y

2

x

0

y

. (16)

3-D case

Equation 11 can be rewritten in the following form

v =

4

det F(det Q)

2

, (17)

Seismic Velocity Estimation

Cameron, Fomel, Sethian 7 Velocity estimation from time migration

where F is the left-hand side of equation 11. As in 2-D, we rewrite system 9 in the

time-domain coordinates (x

0

, t

0

). Then we get

Q

t

0

= v

2

P, (18)

P

t

0

= −

1

v

Q

−1

∇

Q

−1

∇v

T

Q, (19)

where v is given by equation 17, and the gradients are taken with respect to x

0

. Then

the PDE for Q is

1

v

2

Q

t

0

t

0

= −

1

v

Q

−1

∇

Q

−1

∇v

T

Q. (20)

The initial conditions are Q(x

0

, 0) = I

2

, Q

t

0

(x

0

, 0) = 0. The required input

√

det F is

well-approximated by the squares of the Dix velocity obtained from the 3-D prestack

time migration. We emphasize that despite the fact that Q is a matrix in 3-D, scalar

data are enough for its computation.

Cauchy problem for elliptic equations

Equations 14 and 20 reveal the nature of the instabilities in the problem in hand.

These PDEs are elliptic, while the physical setting allows us to pose only a Cauchy

problem for them, which is well-known to be ill-posed. Furthermore, the fact that the

PDEs involve not only the Dix velocity itself, but also its ﬁrst and second derivatives,

leads to high sensitivity of the solutions to input data.

Nonetheless, we found two ways for solving these PDEs numerically on the re-

quired, and relatively short, interval of time: namely, a ﬁnite diﬀerence scheme in-

spired by the Lax-Friedrichs method and a spectral Chebyshev method. A detailed

analysis of the problem shows that our methods work thanks to

1. the special input v

Dix

, corresponding to a positive ﬁnite seismic velocity;

2. the special initial conditions Q(x

0

, t

0

= 0) = 1, Q

t

(x

0

, t

0

= 0) = 0 corresponding

to the image rays;

3. the fact that our ﬁnite diﬀerence method contains error terms which damp the

high harmonics; truncation of the polynomial series in the spectral Chebyshev

method which is similar to truncation of the high harmonics; and

4. the short interval of time, in which we need to compute the solution, so that

the growing low harmonics fail to develop signiﬁcantly.

Items 1 and 2 say that the exact solutions of our PDEs for the hypothetical perfect Dix

velocity given by equations 10 and 11 are ﬁnite and nonzero. Items 3 and 4 say that

the numerical methods take care of the imperfection of the data and computations

on a short enough time interval.

Seismic Velocity Estimation

Cameron, Fomel, Sethian 8 Velocity estimation from time migration

INVERSION METHODS

Our numerical reconstruction of true seismic velocity v(x) in depth-domain coordi-

nates from the Dix velocity given in the time-domain coordinates (x

0

, t

0

) consists of

two steps:

Step 1. Compute the geometrical spreading of the image rays in the time-domain

coordinates from the Dix velocity by solving equation 14 in 2-D and 20 in 3-D.

Then ﬁnd v(x

0

, t

0

) from equation 10 in 2-D and equation 17 in 3-D.

Step 2. Convert the seismic velocity v(x

0

, t

0

) in the time-domain coordinates to

the depth-domain coordinates x using the time-to-depth conversion algorithm,

which was presented by Cameron et al. (2007). It is a fast and robust Dijkstra-

like solver motivated by the Fast Marching method (Sethian, 1996, 1999).

We performed step 1 in two ways: a ﬁnite diﬀerence method and a spectral Cheby-

shev method.

Finite diﬀerence method

This method was inspired by the Lax-Friedrichs method for hyperbolic conservation

laws Lax (1954) due to its total variation diminishing property. We use the “Lax-

Friedrichs averaging” and the wide 5-point stencil in space. The scheme is given

by

P

n+1

j

=

P

n

j+1

+ P

n

j−1

2

−

∆t

4∆x

1

v

n

j

v

n

j+2

−v

n

j

Q

n

j+1

−

v

n

j

−v

n

j−2

Q

n

j−1

, (21)

−

1

Q

n+1

j

= −

1

Q

n

j

+

∆t

2

(f

n

j

)

2

P

n

j

+ (f

n+1

j

)

2

P

n+1

j

, (22)

where v ≡ fQ. We impose the following boundary conditions Q

n

0

= Q

n

nx−1

= 1, P

n

0

=

P

n

nx−1

= 0 corresponding the straight boundary rays. We set the initial conditions

Q

0

j

= 1, P

0

j

= 0 corresponding to the initial conditions for the image rays traced

backward: Q = 1, P = 0.

Spectral Chebyshev method

Alternatively, we solve our PDE in the form given by equation 15 by a spectral

Chebyshev method Boyd (2001). Using cubic splines, we deﬁne the input data at

N

coef

Chebyshev points. We compute the Chebyshev coeﬃcients and the coeﬃcients

of the derivatives in the right-hand side of equation 15. Then we use a smaller number

N

eval

of the coeﬃcients for function evaluation. We need to do such Chebyshev

Seismic Velocity Estimation

Cameron, Fomel, Sethian 9 Velocity estimation from time migration

diﬀerentiation twice. Finally we perform the time step using the stable third-order

Adams-Bashforth method Boyd (2001), which is

u

n+1

= u

n

+ ∆t

23

12

F

n

−

4

3

F

n−1

+

5

12

F

n−2

, (23)

where F

n

≡ F(u

n

, x, t

n

) is the right-hand side. In numerical examples, we tried

N

coef

≥ 100 and N

eval

≤ 25. This method allows larger time steps than the ﬁnite

diﬀerence, and has the adjustable parameter N

eval

.

For step 2, we use a Dijkstra-like solver introduced in Cameron et al. (2007). It

is an eﬃcient time-to-depth conversion algorithm motivated by the Fast Marching

Method (Sethian, 1996). The input for this algorithm is v(x

0

, t

0

) and the outputs are

the seismic velocity v(x, z) and the transition matrices from time-domain to depth-

domain coordinates x

0

(x, z) and t

0

(x, z). We solve the eikonal equation with an

unknown right-hand side coupled with the orthogonality relation

|∇t

0

| =

1

v(x

0

(x, z), t

0

(x, z))

, ∇t

0

· ∇x

0

= 0. (24)

The orthogonality relation means that the image rays are orthogonal to the wave-

fronts. Such time-to-depth conversion is very fast and produces the outputs directly

on the depth-domain grid.

EXAMPLES

Synthetic data example

Figure 2(a) shows a synthetic velocity model. The model contains a high velocity

anomaly that is asymmetric and decays exponentially. The corresponding Dix velocity

mapped from time to depth is shown in Figure 2(b). There is a signiﬁcant diﬀerence

between both the value and the shape of the velocity anomaly recovered by the Dix

method and the true anomaly. The diﬀerence is explained by taking into account

geometrical spreading of image rays. Figure 2(c) shows the velocity recovered by our

method and the corresponding family of image rays. An analogous 3-D example is

provided in (Cameron et al., 2007).

Field data example

Figure 3, taken from (Fomel, 2003), shows a prestack time migrated image from

the North Sea and the corresponding time-migration velocity obtained by velocity

continuation. The most prominent feature in the image is a salt body which causes

signiﬁcant lateral variations of velocity. Figure 4 compares the Dix velocity converted

to depth with the interval velocity model recovered by our method. As in the syn-

thetic example, there is a signiﬁcant diﬀerence between the two velocity caused by

Seismic Velocity Estimation

Cameron, Fomel, Sethian 10 Velocity estimation from time migration

(a)

(b)

(c)

Figure 2: Synthetic test on interval velocity estimation. (a) Exact velocity model. (b)

Dix velocity converted to depth. (c) Estimated velocity model and the correspond-

ing image rays. The image ray spreading causes signiﬁcant diﬀerences between Dix

velocity and estimated velocity.

Seismic Velocity Estimation

Cameron, Fomel, Sethian 11 Velocity estimation from time migration

Figure 3: (a) Seismic image from North Sea obtained by prestack time migration using

velocity continuation (Fomel, 2003). (b) Corresponding time-migration velocity.

Seismic Velocity Estimation

Cameron, Fomel, Sethian 12 Velocity estimation from time migration

(a)

(b)

Figure 4: Field data example of interval velocity estimation. (a) Dix velocity con-

verted to depth. (b) Estimated velocity model and the corresponding image rays.

The image-ray spreading causes signiﬁcant diﬀerences between Dix velocity and true

velocity.

the geometrical spreading of image rays. The middle part of the velocity model may

not be recovered properly. The true structure should include a salt body visible in

the image. The inability of our method to recover it exactly shows the limitation

of the proposed approach in the areas of signiﬁcant lateral velocity variations, which

invalidate the assumptions behind time migration (Robein, 2003). Figure 5 com-

pares three images: post-stack depth-migration image using Dix velocity, post-stack

depth-migration image using the velocity estimated by our method, and prestack

time-migration image converted to depth with our algorithm. The evident structural

improvements in Figure 5(b) in comparison with Figure 5(a), in particular near salt

ﬂanks, and a good structural agreement between Figures 5(b) and 5(c) serve as an

indirect evidence of the algorithm success. An ultimate validation should come from

prestack depth migration velocity analysis, which is signiﬁcantly more expensive.

CONCLUSIONS

We have applied the recently established theorem that the Dix velocity obtainable

from the time-migration velocity is the true interval velocity divided by the geomet-

rical spreading of image rays to pose the corresponding inverse problem. We have

suggested a set of numerical algorithms for solving the problem numerically. We have

tested these algorithms on a synthetic data example with laterally heterogeneous ve-

Seismic Velocity Estimation

Cameron, Fomel, Sethian 13 Velocity estimation from time migration

(a)

(b)

(c)

Figure 5: Migrated images of the ﬁeld data example. (a) Poststack migration using

Dix velocity. (b) Poststack migration using estimated velocity. (c) Prestack time

migration converted to depth with our algorithm.

Seismic Velocity Estimation

Cameron, Fomel, Sethian 14 Velocity estimation from time migration

locity and demonstrated that they produce signiﬁcantly better results than simple Dix

inversion followed by time-to-depth conversion. Moreover, the Dix velocity may qual-

itatively diﬀer from the output velocity. We have also tested our algorithm on a ﬁeld

data example and validated it by comparing prestack time migration image mapped

to depth with post-stack depth migrated images. Our approach is complementary to

velocity estimation methods that work directly in the depth domain. Therefore, it

can serve as an eﬃcient ﬁrst step in seismic velocity model building.

REFERENCES

Bevc, D., J. L. Black, and G. Palacharla, 1995, Plumes: Response of time migration

to lateral velocity variation: Geophysics, 60, 1118–1127.

Boyd, J. P., 2001, Chebyshev and Fourier spectral methods, 2nd ed. (revised): Dover

Publications.

Cameron, M., S. Fomel, and J. Sethian, 2007, Seismic velocity estimation from time

migration: Inverse Problems, 23, 1329–1369.

ˇ

Cerven´ y, V., 2001, Seismic ray method: Cambridge University Press.

Dix, C. H., 1955, Seismic velocities from surface measurements: Geophysics, 20,

68–86.

Fomel, S., 2003, Time-migration velocity analysis by velocity continuation: Geo-

physics, 68, 1662–1672.

Hatton, L., L. K. Larner, and B. S. Gibson, 1981, Migration of seismic data from

inhomogeneous media: Geophysics, 46, 751–767.

Hubral, P., 1977, Time migration - Some ray theoretical aspects: Geophysical

Prospecting, 25, 738–745.

Hubral, P., and T. Krey, 1980, Interval velocities from seismic reﬂection time mea-

surements: Soc. of Expl. Geoph.

Kim, Y. C., W. B. Hurt, L. J. Maher, and P. J. Starich, 1997, Hybrid migration: A

cost-eﬀective 3-D depth-imaging technique: Geophysics, 62, 568–576.

Larner, K. L., L. Hatton, B. S. Gibson, and I. S. Hsu, 1981, Depth migration of

imaged time sections: Geophysics, 46, 734–750.

Lax, P. D., 1954, Weak solutions of hyperbolic equations and their numerical compu-

tation: Communications in Pure and Applied Mathematics, 7, 159–193.

Popov, M. M., 2002, Ray theory and Gaussian beam method for geophysicists:

EDUFBA.

Popov, M. M., and I. Pˇsenˇcik, 1978, Computation of ray amplitudes in inhomogeneous

media with curved interfaces: Studia Geophysica et Geodaetica, 22, 248–258.

Robein, E., 2003, Velocities, time-imaging and depth-imaging in reﬂection seismics:

EAGE.

Sethian, J., 1996, Fast marching level set method for monotonically advancing fronts:

Proceedings of the National Academy of Sciences, 93, 1591–1595.

——–, 1999, Fast marching methods: SIAM Review, 41, 199–235.

Siliqi, R., and N. Bousqui´e, 2000, Anelliptic time processing based on a shifted hy-

perbola approach: 70th Annual International Meeting, Soc. of Expl. Geophys.,

Seismic Velocity Estimation

Cameron, Fomel, Sethian 15 Velocity estimation from time migration

2245–2248.

Yilmaz, O., 2001, Seismic data analysis: Soc. of Expl. Geophys.

Yilmaz, O., I. Tanir, and C. Gregory, 2001, A uniﬁed 3-D seismic workﬂow: Geo-

physics, 66, 1699–1713.

Seismic Velocity Estimation

. we use the Dix velocity instead of time migration velocity as a more convenient input. To remove structural errors inherent in time migration.Cameron.and depth-domain coordinates is provided by the concept of image ray. These results are based on the image ray theory and the paraxial ray tracing theory (Popov ˇ and Pˇenˇik. However. it is necessary to convert time-migrated images into the depth domain either by migrating the original data with a prestack depth migration algorithm or by depth migrating post-stack data after time demigration (Kim et al. Robein.. We develop a numerical approach to ﬁnd (a) seismic velocity from the Dix velocity. 1981) did not ﬁnd widespread application in practice. Moreover. it does not explain how a depth velocity model can be converted to the time-migration velocity. 2003). The objective of the present work is to ﬁnd an eﬃcient method for building a velocity model from time-migration velocity. even mild variations can cause structural distortions of time-migrated images and render them inadequate for accurate geological interpretation of subsurface structures. Hatton et al. We show that the Dix velocity is seismic velocity divided by the geometrical spreading of the image rays. 1997). Seismic Velocity Estimation . We establish new ray-theoretic connections between time-migration velocity and seismic velocity in 2-D and 3-D. 2003). The connection between the time. Robein. 2003). Our results can be viewed as an s c y extension of the Dix formula (Dix. Each of these options requires converting the time migration velocity to a velocity model in depth. Hence. Image rays are seismic rays that arrive normal to the Earth’s surface. Hubral’s theory explains how a depth velocity model can be converted to the time coordinates. However. Fomel. and (b) transition matrices from the time-domain coordinates to the depth-domain coordinates. We test our approach on synthetic and ﬁeld data examples. 2001) or by velocity continuation (Fomel. Our approach is complementary to more traditional velocity estimation methods and can be used as the ﬁrst step in a velocity model building process. Popov. image-ray tracing is a numerically inconvenient procedure for achieving uniform coverage of the subsurface. 2001. 1978. 1995. Sethian 2 Velocity estimation from time migration INTRODUCTION Time-domain seismic imaging is a robust and eﬃcient process routinely applied to seismic data (Yilmaz. Other limitations of image rays are related to the inability of time migration to handle large lateral variations in velocity (Bevc et al. Time migration is considered adequate for seismic imaging in areas with mild lateral velocity variations. 1981. Cerven´... 1955) to laterally inhomogeneous media. introduced by Hubral (1977). Rapid scanning and determination of time-migration velocity can be accomplished either by repeated migrations (Yilmaz et al.. 2001. 2002). This may explain why simpliﬁed image-ray-tracing algorithms (Larner et al.

3-D. velocity v depends only on the depth z. Fomel. and x be the reﬂection subsurface point. r be a receiver. t0 )]−1 . Sethian 3 Velocity estimation from time migration TIME MIGRATION VELOCITY Kirchhoﬀ prestack time migration is commonly based on the following travel time approximation (Yilmaz. In order to write an analog of travel time approximation 2 for 3-D. Equation 2 is heuristic and is not a consequence of the truncated Taylor expansion. 2-D and 3-D. Let s be a source. s.Cameron. r) ≈ T (x0 . The approximation ˆ T usually takes the form of the double-square-root equation ˆ T (x0 . Velocity vm (x0 . Equation 2 is a consequence of the truncated Taylor expansion for the travel time around the surface point x0 . Equation 2 is exact. t0 . t0 ) t2 + 0 |x0 − r|2 . d t0 (4) 2-D. Then the total travel time from s to x and from x to r is approximated as ˆ T (s. 1980) Γ = [v(x0 )R(x0 . velocity is arbitrary. as follows: vDix (t) = d 2 (t0 vm (t0 )) . Equation 2 is a consequence of the truncated Taylor expansion for the travel time around the surface point x0 . and we establish its exact meaning in the next section. 1977) from the subsurface point x. 1955) is exact. We formally deﬁne the Dix velocity vDix (t) by inverting equation 3. Velocity vm depends only on t0 and is the root-mean-square velocity: vm (t0 ) = 1 t0 t0 v 2 (z(t))dt. t0 ) (2) where x0 and t0 are the escape location and the travel time of the image ray (Hubral. and v(x0 ) is the velocity at the Seismic Velocity Estimation . velocity is arbitrary. the Dix inversion formula (Dix. Regarding this approximation. 2001). t0 . r) = t2 + 0 |x0 − s|2 + 2 vm (x0 . we use the relation (Hubral and Krey. let us list four cases depending on the seismic velocity v and the dimension of the problem: 2-D and 3-D. s. (5) where Γ is the matrix of the second derivatives of the travel times from a subsurface point x to the surface. R is the matrix of radii of curvature of the emerging wave front from the point source x. velocity v is constant. t0 ) is a certain kind of mean velocity. x) + T (x. 0 (3) In this case. r) (1) where x0 and t0 are eﬀective parameters of the subsurface point x. 2 vm (x0 . and vm = v.

we will establish theoretical relationships between time-migration velocity and seismic velocity in 2-D and 3-D. e 2000). r) t2 + t0 (x0 − s)T [K(x0 . Q dqi is the matrix of the derivatives Qij (x0 . t0 ) ∂t0 (8) to perform the inversion. The travel time approximation for 3-D implied by the Taylor expansion is ˆ T (x0 . bounded by the dq ray tube (Figure 1). Sethian 4 Velocity estimation from time migration surface point x0 . Call this ray central. The determinant in equation 8 is well approximated by the square of the Dix velocity obtained from the 3-D prestack time migration using the approximation given by equation 2. In 3-D. t0 ). All these rays start from a small neighborhood dx0 of the point x0 perpendicular to the earth surface. according to equation 5 is K(x0 . i. t0 . the shifted hyperbola approximation (Siliqi and Bousqui´. s. The seismic velocity and the Dix velocity are connected through the quantity Q. Consider the fragment of the wave front deﬁned by this ray tube at time t0 . j = 1. Consider a small tube of rays around it. t0 )]−1 (x0 − s) 0 t2 0 + t0 (x0 − r)T [K(x0 . SEISMIC VELOCITY In this section. 2. The simplest way to introduce Q is the following. (6) = + (7) The entries of the matrix t1 K(x0 . t0 )]−1 (x0 − r). t). This means that the conventional 3-D prestack time migration with traveltime approximation 2 provides suﬃcient input for our inversion procedure in 3-D. x0 is the starting surface point. other known approximations also involve parameters equivalent to vm or K. however. t0 ) reached by the central ray at time t0 . the geometrical spreading of image rays. Thus.e. Then. that one needs only the values of det ∂ K(x0 . It is possible to show. For convenience. Let dq be the fragment of the tangent to the front at the point x(x0 . Q is a scalar in 2-D and a 2 × 2 matrix in 3-D. i. where derivatives are Seismic Velocity Estimation . in 2-D. Trace an image ray x(x0 . t0 ) = dx0 . Fomel. One can employ more complex and accurate approximations than the double-squareroot equations 2 and 7. Q is the derivative Q(x0 . t is the traveltime. which. t0 ) ≡ v(x0 )R(x0 . t0 ) have dimension of squared velocity and 0 can be chosen optimally in the process of time migration. However. they represent a fragment of a plane wave propagating downward.Cameron. t0 ) = dx0j . we prefer to deal with matrix K ≡ Γ−1 .

s c y Popov. s c ˇ Cerven´. taken along certain mutually orthogonal directions e1 . The absolute value of det Q has a simple meaning: it is the ˇ geometrical spreading of the image rays (Popov and Pˇenˇik. Hence. 1978. and I is the i. t0 ) ∂t0 −1 (11) in 3-D.2 where v0 it the velocity at the central ray at time t. relates to Q and P as Γ = PQ−1 .j=1. introduced in the previous section. The matrix Γ. 2007). e2 (Popov and Pˇenˇik. K = QP−1 . Cerven´. ∂2v ∂qi ∂qj (9) . t0 )| (10) in 2-D. and ∂ (K(x0 . Sethian 5 Velocity estimation from time migration Figure 1: Illustration for the deﬁnition of geometrical spreading. 2002). Seismic Velocity Estimation . Fomel. 2001. y The time evolution of the matrices Q and P is given by d dt Q P = 0 − v10 V 2 v0 I 0 Q P . t0 )) = v(x(x0 . where vm (x0 . t0 ) is the time-migration velocity. K is deﬁned by equation 6 and can be determined from equation 7. t0 )QT (x0 . Popov. t0 )) = ∂t0 |Q(x0 . z(x0 . we have proven that vDix (x0 . V = 2 × 2 identity matrix. 1978.. In (Cameron et al.Cameron. t0 )) Q(x0 . 2001. t0 ). t0 )) 2 (t0 vm (x0 . t0 ) ≡ ∂ v(x(x0 . 2002).

Introduce y = − Q . (15) In the expanded form. hence dq = dx0 dx0 = Q−1 dx0 .e. i. Suppose we are tracing them all backwards in time together with the quantities Q and P . Q = dx0 . f2 f f f y (16) 3-D case Equation 11 can be rewritten in the following form v= 4 det F(det Q)2 . we get the following partial diﬀerential equation (PDE) for Q (f Q)x0 Qt0 1 . From now on. ∂t0 (12) Furthermore. we will denote the square of the Dix velocity by f in 2-D and the corresponding matrix by F in 3-D. (13) Pt0 = − fQ Q x0 Eliminating P in system 13. equation 15 is y2 yt0 t0 y t ft fx x fx − 2 0 3 0 = y 0 0 − yx0 0 − yx0 x0 + x0 . we assume that our domain does not contain caustics. Let us eliminate the unknown velocity v in system 9 using equation 10. Then equation 14 becomes yt0 f2 = t0 y f f y y x0 x0 . t0 ). (17) Seismic Velocity Estimation .. to avoid the subscript: F≡ ∂ (K(x0 . Moreover. Qt0 (x0 . Finally. Fomel. system 9 dq becomes (f Q)x0 1 Qt0 = (f Q)2 P. 0) = 1. let us eliminate the diﬀerentiation in q using the deﬁnition of Q and rewrite it in the time-domain coordq d d d dinates x0 . (14) =− f 2 Q2 t0 fQ Q x0 The initial conditions are Q(x0 .Cameron. we imply that t0 denotes the one-way traveltime along the image rays. t0 )) . Sethian 6 Velocity estimation from time migration PARTIAL DIFFERENTIAL EQUATIONS FOR THE GEOMETRICAL SPREADING OF IMAGE RAYS In this section. Indeed. Therefore. . the image rays do not cross on the interval of time we consider. 0) = 0. 2-D case Consider a set of image rays coming to the surface. Equation 14 simpliﬁes in 1 terms of the negative reciprocal of Q. we derive the partial diﬀerential equations for Q in 2-D and 3-D.

A detailed analysis of the problem shows that our methods work thanks to 1. Cauchy problem for elliptic equations Equations 14 and 20 reveal the nature of the instabilities in the problem in hand. 1 Pt0 = − Q−1 v (18) Q−1 v T Q. Fomel. Qt (x0 . and 4. scalar data are enough for its computation. in which we need to compute the solution. We emphasize that despite the fact that Q is a matrix in 3-D. Qt0 (x0 . we rewrite system 9 in the time-domain coordinates (x0 . interval of time: namely. t0 = 0) = 1. These PDEs are elliptic. but also its ﬁrst and second derivatives. As in 2-D. and the gradients are taken with respect to x0 . Sethian 7 Velocity estimation from time migration where F is the left-hand side of equation 11. 0) = 0. the special initial conditions Q(x0 . a ﬁnite diﬀerence scheme inspired by the Lax-Friedrichs method and a spectral Chebyshev method. the special input vDix . corresponding to a positive ﬁnite seismic velocity. 3. (19) where v is given by equation 17. 2. Nonetheless. we found two ways for solving these PDEs numerically on the required. Items 3 and 4 say that the numerical methods take care of the imperfection of the data and computations on a short enough time interval. t0 = 0) = 0 corresponding to the image rays. 0) = I2 . Items 1 and 2 say that the exact solutions of our PDEs for the hypothetical perfect Dix velocity given by equations 10 and 11 are ﬁnite and nonzero. truncation of the polynomial series in the spectral Chebyshev method which is similar to truncation of the high harmonics. the fact that the PDEs involve not only the Dix velocity itself. Furthermore. so that the growing low harmonics fail to develop signiﬁcantly. the fact that our ﬁnite diﬀerence method contains error terms which damp the high harmonics. t0 ). while the physical setting allows us to pose only a Cauchy problem for them. Then the PDE for Q is 1 1 T Q−1 v Q. and relatively short. Seismic Velocity Estimation . which is well-known to be ill-posed.Cameron. Then we get Qt0 = v 2 P. leads to high sensitivity of the solutions to input data. The required input det F is well-approximated by the squares of the Dix velocity obtained from the 3-D prestack time migration. (20) Q = − Q−1 2 t0 v v t0 √ The initial conditions are Q(x0 . the short interval of time.

Step 2. we solve our PDE in the form given by equation 15 by a spectral Chebyshev method Boyd (2001). Using cubic splines. We impose the following boundary conditions Qn = Qn 0 nx−1 = 1. Then ﬁnd v(x0 . 1999). t0 ) consists of two steps: Step 1. (2007). Compute the geometrical spreading of the image rays in the time-domain coordinates from the Dix velocity by solving equation 14 in 2-D and 20 in 3-D. Qj 2 . The scheme is given by Pjn+1 = − 1 Qn+1 j n n n n n n vj − vj−2 Pj+1 + Pj−1 ∆t 1 vj+2 − vj − − n 2 4∆x vj Qn Qn j+1 j−1 1 ∆t = − n+ (fjn )2 Pjn + (fjn+1 )2 Pjn+1 . It is a fast and robust Dijkstralike solver motivated by the Fast Marching method (Sethian. t0 ) from equation 10 in 2-D and equation 17 in 3-D. Fomel. Then we use a smaller number Neval of the coeﬃcients for function evaluation. Finite diﬀerence method This method was inspired by the Lax-Friedrichs method for hyperbolic conservation laws Lax (1954) due to its total variation diminishing property. we deﬁne the input data at Ncoef Chebyshev points. P0 = n Pnx−1 = 0 corresponding the straight boundary rays.Cameron. We set the initial conditions Q0 = 1. 1996. Sethian 8 Velocity estimation from time migration INVERSION METHODS Our numerical reconstruction of true seismic velocity v(x) in depth-domain coordinates from the Dix velocity given in the time-domain coordinates (x0 . (21) (22) n where v ≡ f Q. Spectral Chebyshev method Alternatively. We performed step 1 in two ways: a ﬁnite diﬀerence method and a spectral Chebyshev method. We use the “LaxFriedrichs averaging” and the wide 5-point stencil in space. which was presented by Cameron et al. Convert the seismic velocity v(x0 . t0 ) in the time-domain coordinates to the depth-domain coordinates x using the time-to-depth conversion algorithm. We compute the Chebyshev coeﬃcients and the coeﬃcients of the derivatives in the right-hand side of equation 15. We need to do such Chebyshev Seismic Velocity Estimation . P = 0. Pj0 = 0 corresponding to the initial conditions for the image rays traced j backward: Q = 1.

1996). The input for this algorithm is v(x0 . we use a Dijkstra-like solver introduced in Cameron et al. taken from (Fomel. We solve the eikonal equation with an unknown right-hand side coupled with the orthogonality relation | t0 | = 1 . which is un+1 = un + ∆t 5 23 n 4 n−1 F − F + F n−2 . t0 (x. EXAMPLES Synthetic data example Figure 2(a) shows a synthetic velocity model. v(x0 (x. Figure 2(c) shows the velocity recovered by our method and the corresponding family of image rays. Sethian 9 Velocity estimation from time migration diﬀerentiation twice. The most prominent feature in the image is a salt body which causes signiﬁcant lateral variations of velocity. Fomel. 2007). tn ) is the right-hand side. There is a signiﬁcant diﬀerence between both the value and the shape of the velocity anomaly recovered by the Dix method and the true anomaly. there is a signiﬁcant diﬀerence between the two velocity caused by Seismic Velocity Estimation . z) and t0 (x. (24) The orthogonality relation means that the image rays are orthogonal to the wavefronts. The corresponding Dix velocity mapped from time to depth is shown in Figure 2(b). Such time-to-depth conversion is very fast and produces the outputs directly on the depth-domain grid. z)) t0 · x0 = 0. Finally we perform the time step using the stable third-order Adams-Bashforth method Boyd (2001). As in the synthetic example. z) and the transition matrices from time-domain to depthdomain coordinates x0 (x. An analogous 3-D example is provided in (Cameron et al.Cameron. z). shows a prestack time migrated image from the North Sea and the corresponding time-migration velocity obtained by velocity continuation. and has the adjustable parameter Neval . x. This method allows larger time steps than the ﬁnite diﬀerence. The model contains a high velocity anomaly that is asymmetric and decays exponentially. It is an eﬃcient time-to-depth conversion algorithm motivated by the Fast Marching Method (Sethian. For step 2. 2003). z). The diﬀerence is explained by taking into account geometrical spreading of image rays. (2007). Field data example Figure 3.. we tried Ncoef ≥ 100 and Neval ≤ 25. 12 3 12 (23) where F n ≡ F (un . Figure 4 compares the Dix velocity converted to depth with the interval velocity model recovered by our method. t0 ) and the outputs are the seismic velocity v(x. In numerical examples.

(b) Dix velocity converted to depth. Fomel. (c) Estimated velocity model and the corresponding image rays.Cameron. Seismic Velocity Estimation . Sethian 10 Velocity estimation from time migration (a) (b) (c) Figure 2: Synthetic test on interval velocity estimation. (a) Exact velocity model. The image ray spreading causes signiﬁcant diﬀerences between Dix velocity and estimated velocity.

2003). Fomel. (b) Corresponding time-migration velocity. Sethian 11 Velocity estimation from time migration Figure 3: (a) Seismic image from North Sea obtained by prestack time migration using velocity continuation (Fomel. Seismic Velocity Estimation .Cameron.

post-stack depth-migration image using the velocity estimated by our method. The inability of our method to recover it exactly shows the limitation of the proposed approach in the areas of signiﬁcant lateral velocity variations. We have suggested a set of numerical algorithms for solving the problem numerically. in particular near salt ﬂanks. The evident structural improvements in Figure 5(b) in comparison with Figure 5(a). (b) Estimated velocity model and the corresponding image rays. An ultimate validation should come from prestack depth migration velocity analysis. We have tested these algorithms on a synthetic data example with laterally heterogeneous veSeismic Velocity Estimation . The true structure should include a salt body visible in the image. which invalidate the assumptions behind time migration (Robein. Sethian 12 Velocity estimation from time migration (a) (b) Figure 4: Field data example of interval velocity estimation. which is signiﬁcantly more expensive. and a good structural agreement between Figures 5(b) and 5(c) serve as an indirect evidence of the algorithm success. The middle part of the velocity model may not be recovered properly. and prestack time-migration image converted to depth with our algorithm. Fomel. (a) Dix velocity converted to depth. 2003). The image-ray spreading causes signiﬁcant diﬀerences between Dix velocity and true velocity. CONCLUSIONS We have applied the recently established theorem that the Dix velocity obtainable from the time-migration velocity is the true interval velocity divided by the geometrical spreading of image rays to pose the corresponding inverse problem. Figure 5 compares three images: post-stack depth-migration image using Dix velocity. the geometrical spreading of image rays.Cameron.

Fomel. Sethian 13 Velocity estimation from time migration (a) (b) (c) Figure 5: Migrated images of the ﬁeld data example. Seismic Velocity Estimation . (b) Poststack migration using estimated velocity. (c) Prestack time migration converted to depth with our algorithm.Cameron. (a) Poststack migration using Dix velocity.

46. Time-migration velocity analysis by velocity continuation: Geophysics. 1329–1369. Fomel. ——–. Fomel. Gibson. 734–750. 751–767. C. Y. 2007. 248–258. M. Hatton. 2003. y Dix.. P. Migration of seismic data from inhomogeneous media: Geophysics.. Hubral. 1999. P. the Dix velocity may qualitatively diﬀer from the output velocity. 2001. Starich. Weak solutions of hyperbolic equations and their numerical computation: Communications in Pure and Applied Mathematics.. 62. Hurt. Seismic velocities from surface measurements: Geophysics. P. E. Hubral. P. Popov. 738–745. Cameron. 2000.. M. Geophys. 1978. We have also tested our algorithm on a ﬁeld data example and validated it by comparing prestack time migration image mapped to depth with post-stack depth migrated images. Depth migration of imaged time sections: Geophysics. J. 2nd ed. 1591–1595. B. 41. ˇ Cerven´. Fomel. 1980. and J. J. V.. W. Seismic Velocity Estimation . 1995. Therefore. 1954. Anelliptic time processing based on a shifted hye perbola approach: 70th Annual International Meeting. Hsu. 22. L. Ray theory and Gaussian beam method for geophysicists: EDUFBA. 1997. S. S. D. Bousqui´. and G. 1996. M. 68. J. 199–235.. (revised): Dover Publications.. 7. Siliqi. Geoph. L. Hybrid migration: A cost-eﬀective 3-D depth-imaging technique: Geophysics.. Palacharla. K. 2001. L.. Robein. Kim.. M. Krey.. 1981. 68–86. 2002. and T. 1981. L. Gibson. of Expl. 1662–1672. 60. S. Black. L. Computation of ray amplitudes in inhomogeneous s c media with curved interfaces: Studia Geophysica et Geodaetica.. Plumes: Response of time migration to lateral velocity variation: Geophysics. Our approach is complementary to velocity estimation methods that work directly in the depth domain. Pˇenˇik..Some ray theoretical aspects: Geophysical Prospecting. Moreover. B. Velocities. Seismic ray method: Cambridge University Press. 93. 46. 25. R.. 20. and P. J. H. Maher. Larner. and I.. REFERENCES Bevc. Popov. Interval velocities from seismic reﬂection time measurements: Soc.. Larner. S.. 23. 2003. S. 568–576.Cameron. and B. and N. Fast marching methods: SIAM Review. M. 1977. D. Boyd. Sethian. Chebyshev and Fourier spectral methods. of Expl. time-imaging and depth-imaging in reﬂection seismics: EAGE. 1955. it can serve as an eﬃcient ﬁrst step in seismic velocity model building. Time migration . Hatton. Sethian. Lax. Seismic velocity estimation from time migration: Inverse Problems. C. K. Sethian 14 Velocity estimation from time migration locity and demonstrated that they produce signiﬁcantly better results than simple Dix inversion followed by time-to-depth conversion.. Fast marching level set method for monotonically advancing fronts: Proceedings of the National Academy of Sciences. J. 159–193. L. 1118–1127. Soc. and I.

.Cameron. Yilmaz. Gregory.. Seismic Velocity Estimation . Yilmaz. 2001. O. Geophys. Seismic data analysis: Soc. and C. 66. I. 1699–1713. Fomel. O. Tanir. A uniﬁed 3-D seismic workﬂow: Geophysics. Sethian 15 Velocity estimation from time migration 2245–2248. of Expl. 2001.

- Petrological and Geochemical Study
- Petrological and Geochemical Study
- Marine Environment
- 25371860 Sedimentary Basins
- Nelp Offshore
- Basics of Petroleum Geology
- Basic Well Log Interpretation
- Density Porosity
- Geothermal gradient and Formation Temperature Calculation
- Bikaner Nagur Basin
- (Www.entrance-exam.net)-UPSC Geologists Exam Sample Paper 1
- Is Water a Mineral
- Geologist Cover Letter
- New ONGC Week-5
- Sub Surface Geological Mapping
- Bikaner Nagaur
- Calculation of TVD
- Dolomite drilling fluid

Sign up to vote on this title

UsefulNot useful- time2depth_08
- ion to 3D Seismic Survey Design Problem
- Basics of Seismic Interpretation
- 2. SEISMIC SURVEY DESIGN-carlos
- Seismic Interpretation Badly
- Petrel2008presentation[1]
- Seismic Interpretation
- Seismic Sequence Stratigraphy_presentation
- Russell - Introduction to Seismic Inversion Methods
- seismic interpretation
- Seismic Data Processing
- Quantitative Seismic Interpretation
- Seismic data Processing
- umesh final dissertation
- 2d and 3d Seismic Interpret a Ion
- Seismic Data Acquisition
- 2D and 3D Land Seismic Data Acquisition
- Seismic Data ProcessingOZDOGAN YILMAZ
- Fast Solvers for Complex Problems
- Slides 09
- Sofia Kovalevskaya. Unified Approach ...
- On State Space Models for Systems Described by PDE Zak
- PR KALUT
- A Matlab Toolbox For
- Qualifying Exam Syllabus
- CISE301-Topic9
- CONVERGENCE CRITERIA FOR ITERATIVE METHODS IN SOLVING CONVECTION-DIFFUSION EQUATIONS ON ADAPTIVE MESHES
- Lecture 10
- 38706.pdf
- Diff Equation 9 PDE 2012 FALL
- Time to depth conversion

Are you sure?

This action might not be possible to undo. Are you sure you want to continue?

We've moved you to where you read on your other device.

Get the full title to continue

Get the full title to continue reading from where you left off, or restart the preview.

scribd