You are on page 1of 12


3D seismic image processing for unconformities

Xinming Wu & Dave Hale

Center for Wave Phenomena, Colorado School of Mines, Golden, CO 80401, USA

a) b)

Figure 1. Unconformity likelihood (a) computed from a seismic image highlights unconformities, and can be used as constraints
to more accurately estimate seismic normal vectors and better flatten (b) the seismic image.

We propose a 3D seismic unconformity attribute to detect complete unconfor-
mities, highlighting both their termination areas and correlative conformities.
These detected unconformities are further used as constraints to more accu-
rately estimate seismic normal vectors at unconformities. Then, using seismic
normal vectors and detected unconformities as constraints, we can better flatten
seismic images containing unconformities.
Key words: unconformity seismic normal vectors flattening

1 INTRODUCTION tor terminations. Second, automatic seismic flattening

methods cannot correctly flatten reflectors at unconfor-
An unconformity is a non-depositional or erosional sur-
mities with geologic time gaps. To address these chal-
face separating older strata below from younger strata
lenges, we first detect unconformities and then use them
above, and thus represents a significant gap in geologic
as constraints for seismic normal vector estimation and
time (Vail et al., 1977). In seismic images, an unconfor-
image flattening.
mity can be first identified by seismic reflector termi-
nations (i.e., truncation, toplap, onlap or downlap) and
then be traced to its corresponding correlative confor- 1.1 Unconformity detection
Unconformity detection is a significant aspect of Seismic coherence (Bahorich and Farmer, 1995), which
seismic stratigraphic interpretation, because unconfor- highlights reflector discontinuities, is used as a seismic
mities represent discontinuities in otherwise continuous attribute to detect faults and termination areas of un-
depositions and hence serve as boundaries when inter- conformities. However, this coherence attribute is better
preting seismic sequences that represent successively de- suited for detecting faults than unconformities, because
posited layers. reflector discontinuities across a fault are usually more
Moreover, unconformities pose challenges for other apparent than discontinuities across an unconformity.
automatic techniques used in seismic interpretation. Hoek et al. (2010) propose a better unconformity at-
First, it is difficult to accurately estimate normal vec- tribute that measures the degree of seismic reflector con-
tors or slopes of seismic reflectors at an unconformity vergence (or divergence), and thereby highlights the ter-
with multiply-oriented structures due to seismic reflec- mination areas of an unconformity. Both of these meth-
ods process a seismic image locally (Ringdal, 2012) to
264 X.Wu & D.Hale
compute unconformity attributes that can highlight an Extracting horizons terminated by faults or uncon-
unconformity within its termination area, but cannot formities is generally difficult for these methods. Luo
detect its correlative conformity. and Hale (2013) extract horizons across faults by first
Ringdal (2012) proposes a global method that first unfaulting a seismic image; Wu and Hale (2014) do
extracts a 2D flow field that is everywhere tangent to the same by placing control points on opposite sides
reflectors in a 2D seismic image. Then the flow field of faults. However, none of these methods correctly flat-
is used to compute an unconformity probability image tens a seismic image with unconformities, which should
by repeating the following processing for each sample: produce gaps in the flattened image (e.g., Figure 1b).
four seeds are first placed at the four neighbors of the
sample in the 2D flow field; the four seeds then move
along the flow field to produce trajectories; the sepa- 1.4 This paper
ration rate of the trajectories is finally calculated and In this paper, we first propose a method to automati-
used as unconformity probability for that sample. The cally detect an unconformity, complete with its termi-
advantage of this method is that it can use long tra- nation area and correlative conformity, as shown in Fig-
jectories to detect the correlative conformity of an un- ure 1a. We then describe how to more accurately esti-
conformity. The disadvantage is that, to detect such a mate seismic normal vectors at unconformities by using
correlative conformity, the trajectories are required to the detected unconformities as constraints. Finally, we
start from the parallel area (correlative conformity) and discuss how to better flatten (Figure 1b) seismic images
end in the non-parallel area (termination). For 3D seis- containing unconformities by using these estimated seis-
mic images, this method processes inline and crossline mic normal vectors and again using constraints derived
slices separately throughout the volume to compute an from detected unconformities.
unconformity probability volume.

1.2 Seismic normal vector estimation 2 UNCONFORMITY DETECTION

Orientation vector fields, such as vectors normal to or In manual 3D seismic stratigraphic interpretations, an
slopes of seismic reflectors, are useful for seismic inter- unconformity is first recognized as a border at which
pretation. For example, estimated orientation informa- seismic reflectors terminate (i.e., truncation, toplap, on-
tion is used to control slope-based (Fomel, 2002) and lap or downlap), and then is traced to its correlative
structure-oriented (Fehmers and H ocker, 2003; Hale, conformities where reflectors are parallel. Therefore, to
2009) filters so that they smooth along reflectors to obtain a complete unconformity, an automatic method
enhance their coherencies. Seismic normal vectors or should be able to detect both the termination areas
slopes are also used to flatten (Lomask et al., 2006; (green ellipse in Figure 2a) and correlative conformities
Parks, 2010) or unfold (Luo and Hale, 2013) seismic (dashed blue ellipse in Figure 2a) within the unconfor-
images, or to generate horizon volumes (Wu and Hale, mity.
2014). We propose an unconformity attribute that mea-
Structure tensors (van Vliet and Verbeek, 1995; sures dierences between two seismic normal vector
Fehmers and H ocker, 2003) or plane-wave destruction fields computed from two structure-tensor fields, one is
filter (Fomel, 2002) have been proposed to estimate computed using a vertically causal smoothing filter, and
seismic normal vectors or slopes. These methods can the other using a vertically anti-causal smoothing filter.
accurately estimate orientation vectors for structures This attribute can highlight both the termination areas
with only one locally dominant orientation. This means and correlative conformities of an unconformity.
that they can correctly estimate the normal vectors (or
slopes) of the reflectors in conformable areas of a seis- 2.1 Structure tensor
mic image, but for an angular unconformity where two
dierent structures meet, these methods yield smoothed The structure tensor (van Vliet and Verbeek, 1995;
vectors that represent averages of orientations across the Fehmers and H ocker, 2003) can be used to estimate seis-
unconformity. mic normal vectors that are perpendicular to seismic
reflectors. For a 2D image, the structure tensor T for
each sample is a 2 2 symmetric positive-semidefinite
1.3 Seismic image flattening matrix constructed as the smoothed outer product of
Seismic image flattening (Lomask et al., 2006; Parks, image gradients:

2010; Wu and Hale, 2014) or unfolding (Luo and Hale, hg1 g1 ih,v hg1 g2 ih,v
2013) methods are applied to a seismic image to obtain T = hgg> ih,v = , (1)
hg1 g2 ih,v hg2 g2 ih,v
a flattened image in which all seismic reflectors are hor-
izontal. From such a flattened seismic image, all seismic where g = [g1 g2 ]> represents the image gradient vec-
horizons can be identified by simply slicing horizontally. tor computed for each image sample; hih,v represents
Image processing for unconformities 265

a) normal vectors (cyan segments in Figure 2b) are discon-

tinuous across the unconformity.
At an unconformity where seismic reflectors termi-
termination nate (dashed green ellipse in Figure 2a), structures of re-
flectors above the unconformity are dierent from those
of reflectors below. Therefore, if we compute structure
correlative conformity tensors using vertically causal smoothing filters, which
b) average structures from above, we will obtain normal
vectors at the unconformity that are dierent from those
obtained using vertical anti-causal filters, which average
structures from below.
With a vertically causal filter, the structure tensor
computed for each sample represents structures aver-
aged using only samples above. We define such a struc-
Figure 2. A 2D synthetic seismic image (a) with an un- ture tensor as
conformity (red curve) that is manually extended from its

hg1 g1 ih,vc hg1 g2 ih,vc
termination area to its correlative unconformity. The esti- Tc = , (3)
hg1 g2 ih,vc hg2 g2 ih,vc
mated seismic normal vectors (magenta segments in (b) )
are smoothed within the termination area, and therefore are where hih,vc represents horizontal Gaussian (subscript
incorrect, compared to the true seismic normal vectors (cyan h) and vertically causal (subscript vc) smoothing filters.
segments in (b)) that are discontinuous within that area. With a vertically anti-causal filter, the structure
tensor computed for each sample represents structures
averaged using only samples below. We define such a
smoothing in both horizontal (subscript h) and verti-
structure tensor as
cal (subscript v) directions. These horizontal and ver-
tical smoothing filters are commonly implemented with hg1 g1 ih,va hg1 g2 ih,va
Ta = , (4)
Gaussian filters with corresponding half-widths h and hg1 g2 ih,va hg2 g2 ih,va
v. where the subscript va denotes a vertically anti-causal
As shown by Fehmers and H ocker (2003), the seis- smoothing filter.
mic normal vector for each image sample can be es-
timated from the eigen-decomposition of the structure
tensor T: 2.2.1 Vertical smoothing

T= > > To compute two structure-tensor fields that dier signif-

u uu + v vv , (2)
icantly at an unconformity, the causal smoothing filter
where u and v are unit eigenvectors corresponding to that averages from above should smooth along the direc-
eigenvalues u and v of T. tion perpendicular to the structures above the uncon-
We choose u v , so that the eigenvector u, formity, while the anti-causal filter should smooth along
which corresponds to the largest eigenvalue u , indicates the direction perpendicular to the structures below the
the direction of highest change in image amplitude, and unconformity. Here, we simply use vertically causal and
therefore is perpendicular to locally linear features in anti-causal filters because unconformities are usually
an image, while the orthogonal eigenvector v indicates tend to be horizontal in seismic images. We implement
the direction that is parallel to such features. In other these two filters with one-sided exponential smoothing
words, the eigenvector u is the seismic normal vector filters, which are efficient and trivial to implement.
that is perpendicular to seismic reflectors in a seismic A one-sided causal exponential filter for input and
image, and the eigenvector v is parallel to the reflectors. output sequences x[i] and y[i] with lengths n can be
implemented in C++ (or Java) as follows:
2.2 Smoothing float b = 1.0f-a;
float yi = y[0] = x[0];
The structure tensor T given in equation 1 can be used for (int i=1; i<n; ++i)
to accurately estimate the local orientation of struc- y[i] = yi = a*yi+b*x[i];
tures in an image where there is only one locally domi-
nant orientation present. However, for multiply-oriented Similarly, a one-sided anti-causal exponential filter can
structures such as an unconformity where seismic re- be implemented as follows:
flectors terminate (dashed green ellipse in Figure 2a), float b = 1.0f-a;
this structure tensor provides a local average of the float yi = y[n-1] = x[n-1];
orientations of structures. The seismic normal vectors for (int i=n-2; i>=0; --i)
(magenta segments in Figure 2b) estimated from T are y[i] = yi = a*yi+b*x[i];
smoothed near the termination area, whereas the true The parameter a in these two one-sided exponential fil-
266 X.Wu & D.Hale
a) a)

b) b)


correlative conformity

Figure 3. Two dierent seismic normal vector fields es- Figure 4. Unconformity likelihoods, an attribute that evalu-
timated using structure tensors computed with vertically ates dierences between two estimated seismic normal vector
causal (yellow segments) and anti-causal (green segments) fields (yellow and green segments in Figure 3b), before (a)
smoothing filters. In (a), the vector fields dier only within and after (b) thinning highlight both the termination area
the termination area of the unconformity; in (b), these vector and correlative conformity of the unconformity.
dierences are extended to the correlative conformity.

puted with a laterally structure-oriented filter and ver-

ters controls the extent of smoothing. tically causal and anti-causal filters, are defined by
From structure-tensor fields Tc and Ta computed
hg1 g1 is,vc hg1 g2 is,vc
for the same seismic image using vertically causal and Ts,c = , (5)
hg1 g2 is,vc hg2 g2 is,vc
anti-causal smoothing filters, respectively, we estimate
two seismic normal vector fields uc and ua . As shown and
in Figure 3a, the two seismic normal vector fields uc
hg1 g1 is,va hg1 g2 is,va
(green segments in Figure 3a) and ua (yellow segments Ts,a = , (6)
hg1 g2 is,va hg2 g2 is,va
in Figure 3a) are identical in conformable areas with
parallel seismic reflectors, because orientation of struc- where the subscript s represents a structure-oriented
tures locally averaged from above (used to compute Tc ) filter that smoothes along reflectors in a seismic im-
are identical to orientation of structures averaged from age. Note that the structure-oriented smoothing is gen-
below (used to compute Ta ). However, at the termina- erally more expensive than the vertically causal and
tion area of an unconformity, the two vector fields are anti-causal smoothing. We therefore first apply the
dierent, because the structure tensors Tc computed structure-oriented smoothing filter to each element of
with structures locally averaged from above, should be gg> to obtain Ts = hgg> is , which then is smoothed
dierent from Ta computed with structures locally av- separately by vertically causal and anti-causal filters to
eraged from below. obtain Ts,c and Ts,a , respectively. By doing this, we ap-
Therefore, as shown in Figure 3a, the dierence be- ply the relatively expensive structure-oriented smooth-
tween estimated normal vector fields uc and ua pro- ing only once. However, if we first apply the vertically
vides a good indication of the termination area of an causal and anti-causal smoothing to compute two dif-
unconformity. However, a complete unconformity, that ferently smoothed outer products hggic and hgg> ia ,
is, a curve (in 2D) or surface (in 3D) with geologic time we then need to apply the structure-oriented smooth-
gaps, extends from its termination area its correlative ing twice to obtain two structure-tensor fields Ts,c and
conformity with. Thus we should extend normal vec- Ts,a .
tor dierences from the termination area, where these As discussed by Hale (2009, 2011), to obtain a
dierences originate, to the correlative conformity. smoothed output image q(x) from an input p(x), the
structure-oriented smoothing method solves a finite-
dierence approximation to the following partial dier-
2.2.2 Structure-oriented smoothing ential equation:
To detect a correlative conformity, we extend vector dif-
q(x) r D(x) r q(x) = p(x), (7)
ferences (between uc and ua ) at an unconformity from 2
its termination area to its correlative conformity, by where D(x) is a diusion-tensor field that shares the
replacing the horizontal Gaussian smoothing filter in eigenvectors of the structure tensor computed from an
equations 3 and 4 with a structure-oriented smoothing image, and therefore orients the smoothing along image
filter (Hale, 2009) when computing structure tensors. structures. Similar to the half-width in a Gaussian
Then, the structure tensors Ts,c and Ts,a , com- smoothing filter, the parameter controls the extent of
Image processing for unconformities 267
(a) (b)

Figure 5. Applying our method to the synthetic image cut from Hoek et al. (2010), we obtain unconformity likelihoods before
(a) and after (b) thinning.

a) b)

Figure 6. Unconformity likelihoods before (a) and after (b) thinning.

smoothing. Tc and Ta computed with a horizontal Gaussian

In 2D, we use the eigenvectors u(x) and v(x), esti- filter and vertically causal and anti-causal filters, the
mated using the structure tensors shown in equation 1, estimated seismic normal vectors uc (green segments in
to construct our diusion-tensor field Figure 3a) and ua (yellow segments in Figure 3a) dier
> > only within the termination area of the unconformity.
D(x) = u (x)u(x)u (x) + v (x)v(x)v (x). (8)
Using structure tensors Ts,c and Ts,a computed with
Then, because eigenvectors u(x) and v(x) are perpen- a structure-oriented smoothing filter instead of a
dicular and parallel to seismic reflectors, respectively, horizontal Gaussian filter, the dierences between the
we can control the structure-orient filter to smooth estimated seismic normal vectors us,c (green segments
along reflectors by setting the corresponding eigenval- in Figure 3b) and us,a (yellow segments in Figure 3b)
ues u (x) = 0 and v (x) = 1 for all tensors in D(x). are extended from the termination area to the correla-
As indicated by the seismic normal vectors shown tive area.
in Figure 2b, normal vectors (magenta segments) In summary, by first applying a structure-oriented
estimated using the structure tensors computed in filter to each structure-tensor element of gg> , we
equation 1 are inaccurate at unconformities. However, extend any structure dierences, which originate within
they are accurate in conformable areas, including the termination area of an unconformity, to its correl-
the area near correlative conformity of the unconfor- ative conformity. Then, applying vertically causal and
mity. Thus the structure tensors in equation 1 are anti-causal filters for each structure-tensor element,
adequate for constructing diusion-tensors D(x) for we compute two dierent structure-tensor fields Ts,c
structure-oriented smoothing along seismic reflectors, and Ts,a with seismic normal vector fields us,c and
including those near the correlative conformity of an us,a that dier within both the termination area and
unconformity. By applying such a structure-oriented correlative conformity of an unconformity. Finally, the
filter to the elements of the structure tensors Ts,c dierences between the two estimated vector fields us,c
and Ts,a , we extend structural dierences, originating and us,a can be used as an unconformity attribute that
within the termination area of an unconformity, to the highlights the complete unconformity.
corresponding correlative conformity.
As shown in Figure 3, using structure tensors
268 X.Wu & D.Hale



Figure 7. Unconformity likelihoods before (a) and after (b) thinning. Thinned unconformity likelihoods form unconformity
surfaces (b).
Image processing for unconformities 269

a) b)

c) c) d)

Figure 8. Vertical (u1 ) and horizontal (u2 ) components of seismic normal vectors estimated using structure tensors computed
with (b, d) and without (a, c) unconformity constraints.

2.3 Unconformity likelihood highlight two unconformities apparent in the seismic im-
As shown in Figure 3b, the vectors us,c (green segments)
For a 3D seismic image, following the same process
and us,a (yellow segments) are identical everywhere ex-
as above, we compute an unconformity-likelihood vol-
cept at the unconformity, including its termination area
ume as shown in Figure 7, which correctly highlights
and correlative conformity. Therefore, we define an un-
two apparent unconformities. In the time slices of un-
conformity likelihood attribute g, that evaluates the dif-
conformity likelihoods before and after thinning, we ob-
ferences between us,c and us,a , to highlight unconfor-
serve that samples in the lower-left and upper-right ar-
eas, separated by high unconformity likelihoods, suggest
g1 (us,c us,a )p . (9) dierent seismic facies. This indicates that they belong
to two dierent depositional sequences that have dier-
A large power p (p 1) increases the contrast between ent geologic times.
samples with low and high unconformity likelihoods. For From ridges of unconformity likelihoods (Fig-
the example shown in Figure 4a, the unconformity like- ure 7b), we connect adjacent samples with high un-
lihoods are computed with p = 200. conformity likelihoods to form unconformity surfaces as
Using a process similar to that used by Hale (2012) shown in upper-right panel of Figure 7b.
for extracting ridges of fault likelihoods, we extract
ridges of unconformity likelihood by simply scanning
each vertical column of the unconformity likelihood im-
age (Figure 4a), preserving only local maxima, and set-
ting unconformity likelihoods elsewhere to zero. Fig- We first use unconformity likelihoods as constraints to
ure 4b shows that ridges of unconformity likelihood co- more accurately estimate seismic normal vectors at un-
incide with the unconformity that appears in the syn- conformities. Then, using more accurate normal vectors
thetic seismic image. and unconformity likelihoods as constraints in our seis-
Figure 5 shows a more complicated 2D synthetic mic image flattening method, we are able to better flat-
image used by Hoek et al. (2010). The geometric at- ten an image containing unconformitities.
tributes they compute highlight only the termination ar-
eas of unconformities apparent in this synthetic image.
In comparison, unconformity likelihoods before (Fig- 3.1 Estimation of seismic normal vectors at
ure 5a) and after (Figure 5b) thinning, computed us- unconformities
ing our method, highlight the complete unconformities, Using structure tensors computed with horizontal and
including their termination areas as well as correlative vertical Gaussian filters as shown in equation 1, we find
conformities. smoothed seismic normal vectors (magenta segments in
Figure 6 shows an example of a real 2D seismic Figure 2b) in the termination area, because discontin-
image, in which generated unconformity likelihoods be- uous structures across the unconformity are smoothed
fore (Figure 6a) and after (Figure 6b) thinning correctly
270 X.Wu & D.Hale

a) b)

c) d)

Figure 9. RGT (a) and flattened (c) images generated with inaccurate seismic normal vectors (Figures 8a and 8c) and without
unconformity constraints. Improved RGT (b) and flattened (d) images with more accurate seismic normal vectors (Figures 8b
and 8d) and constraints from unconformity likelihoods (Figure 6).

by symmetric Gaussian filters. Therefore, to obtain cor- 3.2 Seismic image flattening at unconformities
rect normal vectors (cyan segments in Figure 2b) that
Seismic normal vectors or dips can be used to flatten
are discontinuous in the termination area, we must use
(Lomask et al., 2006; Parks, 2010) or unfold (Luo and
more appropriate filters to compute structure tensors.
Hale, 2013) a seismic image to generate a horizon vol-
To preserve structure discontinuities, we compute
ume (Wu and Hale, 2014), that allows for the extraction
the structure tensors using horizontal and vertical fil-
of all seismic horizons in the image. Faults and uncon-
ters that do not smooth across unconformities:
formities, which represent discontinuities of reflectors in
hg1 g1 ish,sv hg1 g2 ish,sv a seismic image, present challenges for these methods.
T= , (10)
hg1 g2 ish,sv hg2 g2 ish,sv Luo and Hale (2013) and Wu and Hale (2014) have ex-
where the hish,sv represent horizontal (subscript sh) tended their methods to handle faults by first unfaulting
and vertical (subscript sv) filters that vary spatially, the seismic image or by placing control points on oppo-
and for which the extent of smoothing is controlled by site sides of a fault. However, neither of these methods
the thinned unconformity likelihoods. correctly handles seismic images with unconformities,
The horizontal and vertical filters are similar to because estimated seismic normal vectors or dips are
the edge-preserving smoothing filter discussed in Hale inaccurate at unconformities, and because unconformi-
(2011): ties are not automatically detected and then used as
constraints in these methods.
In this paper we have proposed methods to auto-
q(x) r c2 (x) r q(x) = p(x). (11)
2 matically detect unconformities and more accurately es-
We compute c(x) = 1 gt (x) to prevent this fil- timate seismic normal vectors at unconformities. There-
ter from smoothing across unconformities. gt (x) is a fore, we can easily extend the flattening method de-
thinned unconformity likelihood image as shown in Fig- scribed in Wu and Hale (2014), to better flatten a
ure 6b, which has large values (close to 1) only at un- seismic image at unconformities, by using seismic nor-
conformities and zeros elsewhere. mal vectors estimated from structure tensors computed
Figure 8 shows seismic normal vectors estimated for with equation 10, and by incorporating constraints de-
the image with two unconformities shown in Figure 6. rived from unconformity likelihoods into the flatten-
Both the vertical (Figure 8a) and horizontal (Figure 8c) ing method. We incorporate unconformity constraints
components of seismic normal vectors, estimated from in our flattening method by weighting the equations for
structure tensors computed as in equation 1, are smooth flattening using unconformity likelihoods, and then us-
at the unconformities; those estimated from structure ing the unconformity likelihoods to construct precondi-
tensors computed as in equation 10 preserve disconti- tioner in the conjugate gradient method used to solve
nuities at unconformities (Figures 8b and 8d). those equations.
Image processing for unconformities 271



Figure 10. Generated RGT volume (a) and flattened (b) 3D seismic image. Discontinuities in the RGT volume correspond to
vertical gaps in the flattened image at unconformities.
272 X.Wu & D.Hale
3.2.1 Weighting as in equation 11, to preserve discontinuities in shifts
s(x, y, z) at unconformities.
To generate a horizon volume or to flatten a seismic
image, we first solve for vertical shifts s(x, y, z) as dis-
cussed in Wu and Hale (2014): 3.2.3 Results
2 @s @s 3 2 3
w( @x p @z ) wp With the computed shifts s(x, y, z), we first generate
6 7 6 7 a relative geologic time (RGT) volume (x, y, z) =
6 7 6 7
6w( @s p @s )7 6wq 7 , (12) z + s(x, y, z) (Figures 9a and 10a). We then use the
6 @y @z 7 6 7
4 5 4 5 RGT volume to map a seismic image f (x, y, z) (Fig-
@s 0
@z ures 6 or 7) in the depth-space domain to a flattened
image f(x, y, ) (Figures 9b or 10b) in the RGT-space
where p(x, y, z) and q(x, y, z) are inline and crossline
reflector slopes computed from seismic normal vectors;
From the 2D example shown in Figure 9, the RGT
w(x, y, z) represent weights for the equations; and the
@s (Figure 9a) and flattened (Figure 9c) images, gener-
third equation @z 0, scaled by a small constant , is
ated with inaccurate seismic normal vectors (Figure 8a
used to reduce rapid vertical variations in the shifts.
and 8c) and without unconformity constraints, are in-
For a seismic image with unconformities, we in-
correct at unconformities, where we expect discontinu-
corporate constraints derived from unconformity like-
ities in the RGT image and corresponding gaps in the
lihoods into the equations 12 by setting w(x, y, z) =
flattened image. With more accurate seismic normal
1 gt (x, y, z) and we use a spatially variant (x, y, z)
vectors (Figure 8b and 8d) and with constraints derived
instead of a constant value:
from unconformity likelihoods (Figure 6), we obtain an
(x, y, z) = 0 (1 gt (x, y, z)), (13) improved RGT image (Figure 9b) with discontinuities
at unconformities. Using this RGT image, we obtain an
where 0 is a small constant number (we use 0 = 0.01 improved flattened image (Figure 9c), in which seismic
for all examples in this paper), and gt (x, y, z) denotes reflectors are horizontally flattened and unconformities
the thinned unconformity likelihoods, such as those appear as vertical gaps.
shown in Figure 7b. Figure 10 shows a 3D example with two unconfor-
This spatially variant (x, y, z), with smaller values mity surfaces, highlighted by unconformity likelihoods
(nearly 0) at unconformities, helps to generate more rea- in Figure 7. We observe obvious discontinuities in RGT
sonable shifts with gradual variations everywhere except at unconformities in our generated RGT volume (Fig-
at unconformities. ure 10a). These RGT discontinuities result in vertical
gaps in the corresponding flattened seismic image (Fig-
3.2.2 Preconditioner ure 10b). The time slice of an RGT image shows large
RGT variations between samples in the lower-left and
As discussed in Wu and Hale (2014), to obtain the shifts upper-right areas that are separated by an unconfor-
s(x, y, z) in equation 12 for a 3D seismic image with N mity. This indicates that the sediments, represented by
samples, we solve its corresponding least-squares prob- the samples in the two dierent areas, are deposited in
lem expressed in a matrix form: two dierent sedimentary sequences occurring at dier-
(WG)> WGs = (WG)> Wv, (14) ent geologic times.

where s is an N 1 vector containing the unknown shifts

s(x, y, z), G is a 3N N sparse matrix representing 4 CONCLUSION
finite-dierence approximations of partial derivatives,
W is a 3N 3N diagonal matrix containing weights We have proposed a method to obtain an unconfor-
w(x, y, z) and (x, y, z), and v is a 3N 1 vector with mity likelihood attribute from the dierences between
2N slopes p and q, and N zeros. two seismic normal vector fields estimated from two
Because the matrix (WG)> WG is symmetric structure-tensor fields, one is computed using a verti-
positive-semidefinite, we can solve the linear system of cally causal smoothing filter, and the other using a ver-
equation 14 using the preconditioned conjugate gradi- tically anti-causal filter. From a seismic image, we first
ent method, with a preconditioner M 1 as in Wu and compute smoothed outer products of image gradients by
Hale (2014): applying a structure-oriented smoothing filter to each
element of these outer products. These smoothed outer
M 1
= Sx S y Sz S> > >
z Sy Sx , (15) products are then smoothed using vertically causal and
where Sx , Sy and Sz are filters that smooth in the x, y anti-causal filters to compute two dierent structure-
and z directions, respectively. tensor fields, and their corresponding normal vector
For a seismic image with unconformities, the fil- fields.
ters Sx , Sy and Sz are spatially variant filters designed Using structure-oriented smoothing filters for the
Image processing for unconformities 273
outer products, we extend structure variations from a Ringdal, K., 2012, Flow-based segmentation of seismic
termination area to the corresponding correlative con- data: Masters thesis, University of Bergen.
formity. In doing this, we assume that the correlative Vail, P. R., R. G. Todd, and J. B. Sangree, 1977, Seis-
conformity is not dislocated by faults. If faults appear mic stratigraphy and global changes of sea level: Part
in the seismic image, we could perform unfaulting (Luo 5. chronostratigraphic significance of seismic reflec-
and Hale, 2013) before attempting to detect unconfor- tions: Section 2. application of seismic reflection con-
mities. figuration to stratigraphic interpretation: M 26: Seis-
We use separate vertically causal and anti-causal mic StratigraphyApplications to Hydrocarbon Ex-
filters to obtain structure tensors that dier at uncon- ploration, AAPG Memoirs, 99116.
formities. Unconformity likelihoods might be further im- van Vliet, L. J., and P. W. Verbeek, 1995, Estima-
proved by instead using causal and anti-causal filters tors for orientation and anisotropy in digitized im-
that smooth in directions orthogonal to unconformities. ages: ASCI95, Proc. First Annual Conference of the
As examples of applications, we have shown how to Advanced School for Computing and Imaging (Hei-
estimate more accurate seismic normal vectors and bet- jen, NL, May 16-18), ASCI, 442450.
ter flatten seismic reflectors at unconformities by using Wu, X., and D. Hale, 2014, Horizon volumes with in-
unconformity likelihoods as constraints. terpreted constraints: CWP Report XXX.

This research is supported by the sponsor companies
of the Consortium Project on Seismic Inverse Methods
for Complex Structures. All of the real seismic images
shown in this paper were provided by the US Depart-
ment of Energy and by dGB Earth Sciences. The 2D
synthetic seismic image used in the example shown in
Figure 5 is that shown in Hoek et al. (2010).

Bahorich, M., and S. Farmer, 1995, 3-D seismic dis-
continuity for faults and stratigraphic features: The
coherence cube: The Leading Edge, 14, 10531058.
Fomel, S., 2002, Applications of plane-wave destruction
filters: Geophysics, 67, 19461960.
Hale, D., 2009, Structure-oriented smoothing and sem-
blance: CWP Report 635.
, 2011, Structure-oriented bilateral filtering: 81st
Annual International Meeting, SEG, Expanded Ab-
stracts, 35963600.
, 2012, Methods to compute fault images, extract
fault surfaces, and estimate fault throws from 3D seis-
mic images: Geophysics, 78, O33O43.
Hoek, T. V., S. Gesbert, and J. Pickens, 2010, Geomet-
ric attributes for seismic stratigraphic interpretation:
The Leading Edge, 29, 10561065.
Lomask, J., A. Guitton, S. Fomel, J. Claerbout, and
A. A. Valenciano, 2006, Flattening without picking:
Geophysics, 71, 1320.
Luo, S., and D. Hale, 2013, Unfaulting and unfolding
3D seismic images: Geophysics, 78, O45O56.
Fehmers, G. C., and C. H ocker, 2003, Fast structural
interpretation with structure-oriented filtering: Geo-
physics, 68, 12861293.
Parks, D., 2010, Seismic image flattening as a linear
inverse problem: Masters thesis, Colorado School of
274 X.Wu & D.Hale