CWP-722

Unfaulting and unfolding 3D seismic images

Simon Luo & Dave Hale
Center for Wave Phenomena, Colorado School of Mines, Golden, CO 80401, USA
a) b) c)
Figure 1. A seismic image (a) is rst unfaulted (b) and then attened (c) using fault throw vectors and attening shift vectors
computed automatically.
ABSTRACT
One limitation of automatic interpretation methods such as seismic image at-
tening is their inability to handle geologic faults. To address this limitation,
we propose to combine a method for automatic image unfaulting with seismic
image attening. First, using fault surfaces and fault throw vectors estimated
from an image, we interpolate throw vectors to produce a throw vector eld,
which we use to unfault the image. Then, we atten the unfaulted image to ob-
tain a new image in which sedimentary layering is horizontal and also aligned
across faults. From this attened unfaulted image, we can automatically extract
geologic horizons.
Key words: seismic image unfaulting attening interpretation interpolation
1 INTRODUCTION
Extracting isochronal geologic surfacesgeologic hori-
zons of the same ageis a common problem in geo-
physics and geology. Such horizons are useful for inter-
pretation of stratigraphic features and analysis of struc-
tural deformation, as well as interpolation and correla-
tion of subsurface properties. A geologic horizon is as-
sumed to have been initially deposited as a horizontal
layer and subsequently subjected to faulting and fold-
ing. So in order to extract such a horizon, it is necessary
to quantify faulting and folding apparent in images such
as the one shown in Figure 1a.
Perhaps the most straightforward way to extract a
geologic horizon is by manual picking. Manual picking
is often used in conjunction with autotracking methods
(e.g., Howard, 1990), which track seismic events by fol-
lowing local extrema or zero crossings in amplitude in a
seismic image. Most autotracking methods are not fully
automatic, and an experienced interpreter is often re-
quired, for example, to identify seed points of coherent
events to track, or to correlate events across faults. An
obvious disadvantage of manual interpretation is that
the process can be slow, because human interaction is re-
quired. The advantage, however, is that an experienced
interpreter can pick horizons in areas in which a fully au-
tomatic method might have diculty. Such areas could
arise from a combination of geological (e.g., faults, un-
conformities, and complex stratigraphy) and geophys-
ical (e.g., imaging and processing artifacts, noise, and
multiples) complications (R. Howard, personal commu-
nication).
Alternatives to horizon autotracking methods in-
clude volume interpretation methods (e.g., Stark, 1996),
which, rather than tracking single events, process si-
multaneously an entire seismic volume. Automatic seis-
mic image attening (Stark, 2004; Lomask et al., 2006;
Parks, 2010) is an example of a volume interpreta-
222 S.Luo & D.Hale
w
h
w
f

t(w
h
)
Figure 2. A fault throw vector. At w
f
on the footwall side
of the fault, the fault throw is zero. At w
h
on the hanging
wall side of the fault, the fault throw vector

t(w
h
) species
the location of the image sample that aligns with the sample
on the footwall side.
tion method. Automatic seismic image attening could
potentially identify all horizons in an image, but the
method is unable to match horizons across faults unless
additional information (e.g., fault throw) is provided.
Moreover, most automatic attening methods are lim-
ited to only vertical shearing of an image, but images
with non-vertical faults such as the one shown in Figure
1a clearly cannot be attened by vertical shear alone.
A fully automatic method (e.g., Tnacheri and
Bearnth, 2007) for extracting geologic horizons is ideal.
Toward this end, we propose an automatic method that
can be used to extract all geologic horizons in an im-
age, consisting of two steps: image unfaulting followed
by image unfolding (i.e., image attening). To unfault
an image, we rst use the method described by Hale
(2012) to estimate fault locations and fault throw vec-
tors, displacement vectors along the dip direction of a
fault surface. Then, using estimated fault throw vectors,
we unfault the image. For example, we used fault throw
vectors estimated from the seismic image shown in Fig-
ure 1a to obtain the unfaulted image shown in Figure
1b. To unfold an unfaulted image, we use non-vertical
image attening (e.g., Luo and Hale, 2011). By unfault-
ing and then attening an image, we obtain an image
such as the one shown in Figure 1c in which a surface of
constant relative geologic time (i.e., a horizontal slice)
maps to a geologic horizon in the original image shown
in Figure 1a.
2 IMAGE UNFAULTING
To unfault an image, we must rst estimate fault lo-
cations and fault slip. For the examples shown in this
paper, we use the method described by Hale (2012) to
automatically compute fault surfaces and fault throws
from a 3D seismic image. Although we choose to use
Hales (2012) method, other methods (e.g., Borgos et al.,
2003; Carrillat et al., 2004; Skov et al., 2004; Aurnham-
mer and T onnies, 2005; Admasu, 2008; Liang et al.,
2010) could also be used to estimate fault locations and
fault throw.
For an image f(x), where x = (x1, x2, x3) are co-
ordinates in the present-day space, the estimated fault
throw vectors

t(w), where w = (w1, w2, w3) are coor-
dinates in the unfaulted space, can be used to compute
an image

h(w) = f(w+

t(w)) (1)
in which seismic events are aligned across faults where

t(w) is specied. An example of a fault throw vector for

a synthetic 2D seismic image is shown in Figure 2. In the
gure, w
f
indicates the location of an image sample on
the footwall side of the fault, while w
h
indicates the lo-
cation of the corresponding sample on the hanging wall
side of the fault. The fault throw vector

t(w
h
) speci-
es the location of the image sample that, once shifted
to w
h
on the hanging wall side, aligns with the image
sample at w
f
on the footwall side. Note from equation
1 that events are shifted only at locations where the
fault throw

t(w) is specied. Because we estimate fault
throw only at locations where we have identied a fault
surface, we must interpolate fault throw vectors at lo-
cations between faults to avoid creating new disconti-
nuities in an image when unfaulting.
Our convention is that fault throw vectors

t(w)
specify throws on the hanging wall side of a fault. Be-
cause we will interpolate these throw vectors (e.g., Fig-
ure 4a) between faults, we must also specify fault throw
vectors on the footwall side of a fault so that the rela-
tive throws on opposing sides of a fault do not change
after interpolation. Because fault throw vectors spec-
ify throws on only the hanging wall side of a fault, the
fault throws on the footwall side must be zero (see Fig-
ure 2). For example, Figure 3a shows a subsection of
a 3D seismic image from oshore Netherlands with the
vertical component of the estimated fault throw vectors
overlaid. Notice in Figure 3a that on the hanging wall
sides of faults, the fault throws are nonzero, while on
the footwall sides, the fault throws are zero. Note that
Figure 3a and Figure 4a show the same fault throws
on the hanging wall sides of faults, but only Figure 3a
shows the zero-valued fault throws on the footwall sides
of faults.
To interpolate fault throw vectors at locations be-
tween faults, we use blended neighbor interpolation
(Hale, 2009a). For Euclidean distances, blended neigh-
bor interpolation is similar to natural neighbor inter-
polation (Sibson, 1981) and discrete Sibson interpola-
tion (Park et al., 2006). Although we use blended neigh-
bor interpolation, in principle any smooth interpolation
would suce. However, it is important that the inter-
polation satises the interpolation condition, which re-
quires that the interpolant at known locations matches
exactly the known values, because we must make sure
Unfaulting and unfolding 3D seismic images 223
a)
b)
Figure 3. A seismic image (a) overlaid with the vertical component of fault throw vectors, and the blended neighbor interpo-
lation (b) of the vertical component of fault throw vectors.
224 S.Luo & D.Hale
that interpolation does not change the fault throws es-
timated at fault locations.
To interpolate a vector eld, we interpolate sepa-
rately each vector component. For example, Figure 3b
shows the blended neighbor interpolation of the vertical
component of the fault throw vectors shown in Figure
3a. Using the interpolated throw vectors t(w) estimated
from an input image f(x), the unfaulted image h(w) is
computed as
h(w) = f(w+t(w)) . (2)
Figure 4b shows the unfaulted image computed accord-
ing to equation 2 using the interpolated throw vectors
shown in Figure 3b and the input image shown in Figure
3a. Similarly, Figure 6b shows the unfaulted image com-
puted from the input image shown in Figure 6a and the
blended neighbor interpolation of the fault throw vec-
tors whose vertical component is overlaid on the image
in Figure 6a.
3 IMAGE FLATTENING
Automatic image attening (Stark, 2004; Lomask et al.,
2006; Parks, 2010) is a process for computing seismic im-
ages in which sedimentary layering is horizontal. Most
methods for automatic attening are limited to only ver-
tical shearing of an image. Flattening by vertical shear,
however, can signicantly distort image features (Luo
and Hale, 2011). Moreover, for seismic images such as
the one shown in Figure 4a with non-vertical faults, the
true geologic deformation clearly is not vertical. For this
reason, we do not limit our attening method to vertical
shear only, but instead allow for non-vertical attening
shift vectors.
To atten an (unfaulted) image h(w), we must nd
a mapping w(u), where u = (u1, u2, u3) are coordinates
in the attened space, such that the image
g(u) = h(w(u)) (3)
is at. We write the mapping w(u) in terms of a shift
vector eld r(u):
w(u) = u r(u) , (4)
which has a corresponding Jacobian matrix J = w/u:
J =

1 r1/u1 r1/u2 r1/u3

r2/u1 1 r2/u2 r2/u3
r3/u1 r3/u2 1 r3/u3

.
(5)
Next, given normal vectors n = (n1, n2, n3), which we
compute from an image using structure tensors (van
Vliet and Verbeek, 1995; Fehmers and H ocker, 2003),
we can write the Jacobian matrix for rotation:
Jr =

n3 + n
2
2
/(1 + n3) n1n2/(1 + n3) n1
n1n2/(1 + n3) n3 + n
2
1
/(1 + n3) n2
n1 n2 n3

.
(6)
Normal vectors transform with the transpose of the Ja-
cobian (Parks, 2010; Luo and Hale, 2011):
J

n = m, (7)
where m are normal vectors in the attened space, and
if all normal vectors mpoint downward in an image, i.e.,
m = [ 0 0 1 ]

, then the image is at. It is straightfor-

ward to check that J

r
n = [ 0 0 1 ]

, so that if J = Jr,
then applying the shifts r(u) will atten the image from
which the normal vectors n were computed.
To atten an image, we solve for an approximately
isometric mapping w(u) with Jacobian matrix J that
satises
J

Jr = I , (8)
where I is the identity matrix. Isometric mappings are
desirable because they preserve metric properties. Thus,
if we could isometrically map an image to a attened im-
age, then all metric properties (e.g., length, angle, area,
and volume) of features in the original image would be
preserved in the attened image. Isometric mappings,
however, exist only in special cases (Floater and Hor-
mann, 2005), so in general, we solve for a mapping w(u)
that is only approximately isometric.
The columns of J contain vectors w1(u), w2(u),
and w3(u). That is,
J = [ w1(u) w2(u) w3(u) ] , (9)
where
w1(u) =
w(u)
u1
=

1
r1
u1

r2
u1

r3
u1

,
w2(u) =
w(u)
u2
=

r1
u2
1
r2
u2

r3
u2

,
w3(u) =
w(u)
u3
=

r1
u3

r2
u3
1
r3
u3

.
(10)
Vectors w1(u) and w2(u) are tangent to a surface (e.g.,
a horizon) at u and thus are orthogonal to a vector n(u)
normal to the surface at u (see Figure 8). The vector
w3(u) is tangent to the line for which the horizontal co-
ordinates u1 and u2 in the attened space are constant,
i.e., the line in coordinates w that maps to a vertical line
in coordinates u (Mallet, 2004). For an exactly isomet-
ric mapping w(u), tangent vectors w1(u), w2(u), and
w3(u) are orthonormal vectors, and the corresponding
Jacobian matrix is orthogonal.
Next, if we denote the columns of Jr as w1(u),
w2(u), and w3(u), then
Jr = [ w1(u) w2(u) w3(u) ] , (11)
and equation 8 states

1
w1 w

1
w2 w

1
w3
w

2
w1 w

2
w2 w

2
w3
w

3
w1 w

3
w2 w

3
w3

1 0 0
0 1 0
0 0 1

. (12)
Unfaulting and unfolding 3D seismic images 225
a)
b)
Figure 4. A seismic image (a) overlaid with the vertical component of fault throw vectors, and the unfaulted image (b).
226 S.Luo & D.Hale
a)
b)
Figure 5. The seismic image shown in Figure 4a is unfaulted and attened (a) using the composite shift vectors (b).
Unfaulting and unfolding 3D seismic images 227
a)
b)
Figure 6. A seismic image (a) overlaid with the vertical component of fault throw vectors, and the unfaulted image (b).
228 S.Luo & D.Hale
a)
b)
Figure 7. The seismic image shown in Figure 4a is unfaulted and attened (a) using the composite shift vectors (b).
Unfaulting and unfolding 3D seismic images 229
The matrix J

Jr is a metric tensor characterizing lo-

cal metric properties such as length, angle, area, and
volume (Mallet, 2002, 2004), so by setting this matrix
equal to the identity, we constrain the type of deforma-
tion parameterized by the mapping w(u). Equation 12
gives nine equations for the partial derivatives of the
shift vector eld r(u):
n1

1
r1
u1

n2
r2
u1
n3
r3
u1
= 0 ,
n1
r1
u2
+ n2

1
r2
u2

n3
r3
u2
= 0 ,
(13)
and

1
r1
u1

r2
u1
+ n1
r3
u1
= 1 ,

1
r1
u1

r2
u1
+ n2
r3
u1
= 0 ,

r1
u2
+

1
r2
u2

+ n1
r3
u2
= 0 ,

r1
u2
+

1
r2
u2

+ n2
r3
u2
= 1 ,
(14)
and

r1
u3

r2
u3
n1

1
r3
u3

= 0 ,

r1
u3

r2
u3
n2

1
r3
u3

= 0 ,
n1
r1
u3
n2
r2
u3
+ n3

1
r3
u3

= 1 ,
(15)
where
= n3 + n
2
2
/(1 + n3) ,
= n3 + n
2
1
/(1 + n3) ,
= n1n2/(1 + n3) .
(16)
We solve equations 13, 14, and 15 for the components
of the shift vector eld r(u) by weighted least-squares
using conjugate gradient iterations.
Equation 8 describes an isometric mapping of an
image to a attened image, but in general, we cannot
expect to nd an exactly isometric mapping for all im-
ages. In fact, the only image for which we can nd an
exactly isometric mapping is one in which the normal
vectors are constant. For all other images, equations 13,
14, and 15 cannot be satised exactly, and we must de-
cide which equations to emphasize.
Equations 13, 14, and 15 correspond to entries in
the metric tensor on the left side of equation 12, and
they characterize the lengths of and angles between
tangent vectors w1(u), w2(u), and w3(u). Specically,
the diagonal entries of the metric tensor characterize
lengths of tangent vectors, while the o-diagonal en-
tries characterize angles between tangent vectors. For
image attening, we give most weight to equations 13,
which determine the angle between the surface tan-
gent vectors w1(u) and w2(u) and the normal vector
u
w
1
w
2
w
w(u) u(w)
n
Figure 8. At any point w on a smooth surface, there exist
vectors w
1
and w
2
tangent to the surface and orthogonal to
a normal vector n. The mapping w(u), with inverse map-
ping u(w), determines local metric properties around points
w and u. For seismic image attening, we solve for an ap-
proximately isometric mapping w(u) that attens geologic
horizons.
n(u). If these equations are satised, then the image
g(u) = h(w(u)) = h(u r(u)) obtained by applying
the shifts r(u) will be at.
We give less weight to equations 14. The four
corresponding entries in the metric tensor (equation
12) form what is referred to as the rst fundamental
form (Floater and Hormann, 2005), which character-
izes lengths, areas, and angles measured on a surface. If
the rst fundamental form equals the identity, then the
surface is said to be locally developable, meaning it is
isometric to a plane. Because we assume that a geologic
horizon was initially deposited as a horizontal layer (i.e.,
a plane), a developable geologic horizon would indicate
that metric properties measured on the horizon have
been preserved since its initial deposition.
Finally, we give least weight to equations 15, which
determine the length of the tangent vector w3(u) and
the angles it forms with the surface tangent vectors. Re-
call that w3(u) is tangent to the line in coordinates w
that maps to a vertical line in the attened coordinates
u. Thus, if w3(u) is a unit vector parallel to the normal
vector n, then the thickness of sedimentary layers mea-
sured perpendicular to bedding will be preserved in the
attening process. For most images, we cannot preserve
thickness while attening, so we give the corresponding
equations least weight.
Figures 5a and 7a show attened unfaulted images
computed from the unfaulted images shown in Figures
4a and 6a, respectively. For both examples, normal vec-
tors were computed using structure tensors with Gaus-
sian smoothing lters (Hale, 2009b) with a vertical half-
width of 32 ms and horizontal half-widths of 50 m. In
addition, when solving equations 13, 14, and 15, we used
230 S.Luo & D.Hale
1.5 s
1.6 s
1.7 s
1.8 s
1.1 s
1.2 s
1.3 s
1
.1
s
1
.2
s
1
.3
s
b)
1

k
m
1
.5
s
1
.6
s
1
.7
s
1
.8
s
a)
1

k
m
Figure 9. Geologic horizons extracted using the composite shift vector elds shown in Figure 5b (a) and Figure 7b (b).
smoothing preconditioners as described by Parks (2010)
to speed convergence. The smoothing lter in the verti-
cal direction had a half-width of 24 ms, while the lters
in the horizontal directions had half-widths of 150 m.
4 HORIZON EXTRACTION
Once we obtain a attened unfaulted image, the corre-
sponding attening shift vectors, and the interpolated
throw vectors, we can extract geologic horizons such as
those shown in Figure 9.
We extract a horizon by rst selecting a horizontal
slice of constant u3 in a attened unfaulted image g(u).
Next we form a composite mapping x(u) by combining
the mapping x(w) = w+t(w) used to unfault an image
with the mapping w(u) = u r(u) used to atten an
image to obtain
x(u) = u s(u) , (17)
where s(u) is the composite shift vector eld:
s(u) = r(u) t(u r(u)) , (18)
which allows for a direct mapping from an image f(x)
to a attened unfaulted image g(u) with
g(u) = f(u s(u)) . (19)
For example, we use equation 19 to compute the at-
tened unfaulted image shown in Figure 5a from the
composite shift vector eld shown in Figure 5b and the
original image shown in Figure 4a. Similarly, Figure 7a
shows the attened unfaulted image computed from the
composite shift vector eld and original image shown in
Figures 7b and 6a, respectively. Using the composite
shift vector eld s(u), we map a surface of constant u3,
which corresponds to constant geologic time or constant
depositional time, to a geologic horizon in present-day
coordinates. For example, for u = (u1, u2, k3) where k3
is constant, the coordinates x of the horizon in present-
day space are simply
x = x( u) = u s( u). (20)
Figures 9a and 9b show geologic horizon surfaces ex-
tracted from the composite shift vector elds shown in
Figures 5b and 7b, respectively. In the horizon in Fig-
ure 9a, notice the en echelon faults that can be clearly
seen in the seismic image in Figure 4a. In the horizon
in Figure 9b, notice the roughly circular fault polygons,
which correspond to the conical fault surfaces described
by Hale (2012). Note that although we show only a sin-
gle horizon for each image, it is trivial to extract another
horizon by simply choosing a dierent horizontal slice
of constant u3 from a attened unfaulted image.
5 CONCLUSION
We have presented a method to automatically unfault
and atten seismic images. The method requires an es-
timate of fault locations and fault throw vectors, which,
for the examples shown in this paper, we obtain us-
ing the method described by Hale (2012). Flattened un-
faulted images are images in which horizontal slices cor-
respond to a constant geologic time, and the geologic age
of horizontal slices increases with increasing time.
The attening method we describe solves, in a least-
squares sense, for an approximately isometric mapping
of an image to a attened image. For such a mapping,
it is assumed that geologic folding follows the exural
slip style of deformation (Suppe, 1985) with constant
sedimentation velocity (Mallet, 2002). This assumption
might not always be correct, but we can evaluate it using
the composite shift vector eld used to compute a at-
tened unfaulted image. For example, we can compute
the tangent vectors given in equation 9 to determine
if they are orthonormal. Also, we can compute proper-
ties of geologic horizons such as Gaussian curvature to
Unfaulting and unfolding 3D seismic images 231
determine which horizons are locally developable. Note
that these computations require partial derivatives of
the mapping x(u) = u s(u). For images in which ge-
ologic horizons are horizontal (i.e., attened unfaulted
images), these derivatives can be computed with a sim-
ple nite-dierence scheme, but for images in which
horizons are deformed, derivatives are less straightfor-
ward to compute because they require knowledge of the
surface parameterization.
Some limitations remain in our method. One is that
we are currently using only estimated fault throw vec-
tors to unfault an image. In reality, fault slip consists of
fault heave as well as fault throw. Fault heave, however,
is more dicult to estimate not only because fault heave
tends to be parallel to sedimentary layering, but also
because seismic images have lower lateral than vertical
resolution. Another limitation arises from the way in
which we compute normal vectors. Because normal vec-
tors are computed in local windows (and, moreover, are
constrained to point in the positive time or depth direc-
tion), they cannot distinguish overturned or recumbent
horizons. This limitation could be overcome by using a
dierent method for computing normal vectors or by l-
tering normal vectors computed using structure tensors
to distinguish overturned or recumbent horizons.
Although the method described in this paper con-
sists of two stepsimage unfaulting followed by image
atteningit is possible to combine these two steps into
one, by solving for shifts that atten an image while con-
straining the shift vectors to be equal to fault throw vec-
tors estimated at fault locations. Recall that when at-
tening, we try to preserve surface metric tensors (equa-
tion 14) and sedimentary layer thickness (equation 15),
while emphasizing image atness (equation 13). The
attening process, however, can only preserve surface
metric tensors and layer thickness with respect to the
input image, which, for the examples shown in this pa-
per, was an unfaulted image. Preserving surface metric
tensors and layer thickness with respect to the original
faulted and folded image would be more appropriate,
but to do so would require the one-step image unfault-
ing and attening process described above.
ACKNOWLEDGMENTS
Thanks to dGB Earth Sciences for providing the 3D
seismic images shown in this report. This research was
supported by the sponsors of the Center for Wave Phe-
nomena at the Colorado School of Mines.
REFERENCES
Admasu, F., 2008, A stochastic method for auto-
mated matching of horizons across a fault in 3D seis-
mic data: PhD thesis, Otto-von-Guericke-University
Magdeburg.
Aurnhammer, M., and K. D. T onnies, 2005, A genetic
algorithm for automated horizon correlation across
faults in seismic images: IEEE Transactions on Evo-
lutionary Computation, 9, 201210.
Borgos, H., T. Skov, T. Randen, and L. Sonneland,
2003, Automated geometry extraction from 3D seis-
mic data: 73rd Annual International Meeting, SEG,
Expanded Abstracts.
Carrillat, A., H. Borgos, T. Randen, L. Sonneland, L.
Kvamme, and K. Hansch, 2004, Fault system analy-
sis with automatic fault displacement estimates a
case study: EAGE 66th Conference and Exhibition,
Expanded Abstracts.
Fehmers, G. C., and C. F. W. H ocker, 2003, Fast struc-
tural interpretations with structure-oriented ltering:
Geophysics, 68, 12861293.
Floater, M. S., and K. Hormann, 2005, Surface pa-
rameterization: A tutorial and survey: Advances in
Multiresolution for Geometric Modelling, 157186.
Hale, D., 2009a, Image-guided blended neighbor inter-
polation: CWP Report 634.
, 2009b, Structure-oriented smoothing and sem-
blance: CWP Report 635.
, 2012, Fault surfaces and fault throws from 3D
seismic images: CWP Report 721.
Howard, R. E., 1990, Method for attribute tracking in
seismic data: US Patent 5,056,066.
Liang, L., D. Hale, and M. Maucec, 2010, Estimating
fault displacements in seismic images: 80th Annual
International Meeting, SEG, Expanded Abstracts,
13571361.
Lomask, J., A. Guitton, S. Fomel, J. F. Claerbout, and
A. Valenciano, 2006, Flattening without picking: Geo-
physics, 71, P13P20.
Luo, S., and D. Hale, 2011, Non-vertical deformations
for seismic image attening: CWP Report 694.
Mallet, J.-L., 2002, Geomodeling: Oxford University
Press.
, 2004, Space-time mathematical framework for
sedimentary geology: Mathematical Geology, 36, 1
32.
Park, S. W., L. Linsen, O. Kreylos, J. D. Owens, and B.
Hamann, 2006, Discrete Sibson interpolation: IEEE
Transactions on Visualization and Computer Graph-
ics, 12, 243253.
Parks, D., 2010, Seismic image attening as a linear
inverse problem: Masters thesis, Colorado School of
Mines.
Sibson, R., 1981, A brief description of natural neigh-
bor interpolation, in Interpreting Multivariate Data:
John Wiley & Sons, 223233.
Skov, T., M. Oygaren, H. Borgos, M. Nickel, and L.
Sonneland, 2004, Analysis from 3D fault displacement
extracted from seismic data: EAGE 66th Conference
and Exhibition, Expanded Abstracts.
Stark, T. J., 1996, Surface slice generation and inter-
pretation: A review: The Leading Edge, 15, 818819.
232 S.Luo & D.Hale
, 2004, Relative geologic time (age) volumes
Relating every seismic sample to a geologically rea-
sonable horizon: The Leading Edge, 23, 928932.
Suppe, J., 1985, Principles of structural geology:
Prentice-Hall.
Tnacheri, N. O., and R. E. Bearnth, 2007, Method of
seismic interpretation: US Patent 7,519,476.
van Vliet, L. J., and P. W. Verbeek, 1995, Estima-
tors for orientation and anisotropy in digitized im-
ages: Proceedings of the rst annual conference of
the Advanced School for Computing and Imaging, 95,
442450.