Professional Documents
Culture Documents
and Stratigraphy
Kurt J. Marfurt (University of Oklahoma)
Geometric Attributes
5-1
Course Outline
Introduction
Complex Trace, Horizon, and Formation Attributes
Multiattribute Display
Spectral Decomposition
Geometric Attributes
Attribute Expression of Geology
5-2
Volumetric dip and azimuth
5-3
Definition of reflector dip
z
θ (dip magnitude)
φ (dip azimuth)
n
a
θy
θx (crossline dip)
(inline dip)
ψ y
x (strike)
5-5
1. 3-D Complex Trace Analysis
(Instantaneous Dip/Azimuth)
Hilbert transform
−1 H
Instantaneous phase φ = tan (d /d)
∂H d ∂d H
d − d
∂φ ∂ ∂
Instantaneous frequency ω = 2π f = 2π = 2π t t
∂t d2 + dH
2
( )
∂H d ∂d H
d − d
Instantaneous kx =
∂φ
= ∂ x ∂ x
in line wavenumber ∂x d2 + dH
2
( )
∂H d ∂d H
Instantaneous d − d
∂φ ∂y ∂y
ky = =
cross line wavenumber ∂y d2 + d( )H 2
Instantaneous
kx ky
apparent dips p =
ω
;q =
ω
5-6
Seismic data
7.5 km
1500 Amp
pos
Depth (m)
neg
4500
Vertical slice Depth slice
7.5 km
1500 Dip
(deg)
high
Depth (m)
4500
Vertical slice Depth slice
d(t)
Quadrature
H
d (t)
)
+180
Phase
-1 H
0 φ(t) = tan [d (t)/d(t)]
-180
Weighted average
Frequency
frequency
f(t) = dφ(t) /dt
4500
Vertical slice Depth slice
4500
Vertical slice Depth slice
7.5 km
1500
Azim
(deg)
360
Depth (m)
180
4500
Vertical slice Depth slice
180
4500
Vertical slice Depth slice
Dip with
maximum
coherence
(+50)
Amp
pos
Time (s)
0
L R
C
neg
inline
crossline
5-18 (Marfurt, 2006)
Search for the most coherent window
containing the analysis point
crossline
Time (s)
+.2
2
0
3
smoothed inst. multi-window -.2
Inline dip inline dip scan
5-20 4 (Marfurt, 2006)
2 km
Comparison
of dip
estimates on
time slice
(t=1.0 s)
A A′′
seismic inst.dip
dip
(µs/m)
+.2
-.2
0.50
0
0.75
pos
Caddo
Time (s)
1.00
1.25
Ellenburger
1.50
Basement?
1.75
5-22
Time/structure of Caddo horizon
5 km B′′
Time (s)
0.70
0.75
0.80
0.85
B
5-23
Dip magnitude from picked horizon
5 km B′′
Dip
(s/km)
0.00
0.06
5-24
B
5 km
NS dip from picked horizon
B′′
Dip
(s/km)
+0.06
0.00
-0.06
5-25
B
NS dip from multi-window scan
5 km B′′
Dip
(deg)
+2
-2
5-26
B
5 km
EW dip from picked horizon
B′′
Dip
(s/km)
+0.06
0.00
-0.06
5-27
B
EW dip from multi-window scan
5 km B′′
Dip
(deg)
+2
-2
5-28
B
Shaded illumination
Dip (deg)
+2
-2
=0000
ϕ=150
=30
=60
=90
=120
5-30
Time slices through apparent dip
(t=1.2s)
Dip (deg)
+2
-2
0 0 00
ϕϕ=150
=30
=60
=90
=0
=120
5-31
Volumetric visualization of
reflector dip and azimuth
Dip Azimuth
Hue
0 180 360
High
Dip Magnitude
Saturation
N
1.2
e (s)
W E
Tim
1.4
(c)
5-32
Volumetric visualization of
reflector dip and azimuth
Dip Azimuth
Hue
0 180 360
High
Dip Magnitude
1 .2
Saturation
s)
Time (
1. 4
Transparent
0
W Transparent E
5-33
0.0
Towards 3-D
divergent
seismic
stratigraphy…
t (s)
convergent
divergent
1.5
convergent
In Summary:
• Dip and azimuth cubes only show relative changes in dip and azimuth,
since we do not in general have an accurate time to depth conversion
• Dip and azimuth estimated using a vertical window in general provide more
robust estimates than those based on picked horizons
• Dip and azimuth volumes form the basis for volumetric curvature,
coherence, amplitude gradients, seismic textures, and structurally-oriented
filtering
• Dip and azimuth will be one of the key components for future computer-
aided 3-D seismic stratigraphy
5-35
Coherence
5-36
Seismic Time Slice
5 km
5 km
salt
e
lin
e
lin
s
os
s
os
cr
cr
inline inline
5-39
Cross correlation of 2 traces
Shifted windows
Trace #1 of Trace #2
Cross
lag: -4 -2 0 +2 +4 correlation
40 ms
Maximum
coherence
5-40
AAA
high
coh
high
low
neg
A
coh
high
low
3 km
5-43
coherence seismic
Alternative measures of waveform
similarity
• cross correlation
• semblance, variance, and Manhattan distance
• eigenstructure
• Gradient Structural Tensors (GST)
5-44
Semblance estimate of coherence
energy of average traces
≡
5. coherence≡
energy of input traces
1. Calculate energy of input traces
2. Calculate the average
wavelet within the
analysis window.
Analysis
window
t-K∆t
dip
8 ms
5-47
Solution: calculate coherence on the analytic
trace
t-K∆t
dip
5-50
Eigenstructure coherence:
Time slice through total energy in 9 trace, 40 ms window
scour
salt
5-51
Eigenstructure coherence:
Time slice through coherent energy in 9 trace, 40 ms window
scour
salt
5-52
Eigenstructure coherence:
Time slice through ratio of coherent to total energy
faults
scour
salt
5-53
Coherence
algorithm
evolution
Seismic Crosscorrelation
Canyon
Salt
Channels
Semblance Eigenstructure
5-54 (Gersztenkorn and Marfurt, 1999)
GST dip scan
seismic coherence coherence
inline
slice
Comparison of
Gradient
Structure Tensor
and dip scan
crossline
eigenstructure
slice
coherence
slice
time
5-55 (Bakker, 2003)
Coherence artifacts due to an ‘efficient’
calculation without search for structure
Seismic 0.6 s
A A′′
0.4
0.6
t (s)
0.8
0.8 s
1.0
1.2
1.4
Seismic section
1.0 s
1.2 s
5-57 1.4 s
0.4 s
Coherence 0.6 s
t (s)
0.8
0.8 s
1.0
1.2
1.4
Coherence section
1.0 s
1.2 s
5-58 1.4 s
0.4 s
Coherence 0.6 s
t (s)
0.8
0.8 s
1.0
1.2
1.4
Coherence section
1.0 s
1.2 s
5-59 1.4 s
Impact of
lateral analysis
window
Temporal Temporal
aperture = 8 ms aperture = 32 ms
On a
structural
target
Temporal Temporal
5-61 aperture = 8 ms aperture = 40 ms
Impact of vertical analysis window
(time slice at t = 1.586 s)
+/- 6
24ms
12 ms
5-62
Impact of vertical analysis window
Fault on coherence green time slice is
shifted by a stronger, deeper event
0.5
Time (s)
1.0
1.5
salt
Figure 3.45b
5 km
Coherence
horizon slice
(better for
stratigraphic
analysis)
5-64
5 km
time (ms)
+16
Impact of +8 analysis
0
height of -8 window
-16
analysis -24
-32
window
5 km
+32
+24
time (ms)
+16
+8
0 analysis window
-8
-16
-24
-32
5-65
Coherence
In summary, coherence:
• Is an excellent tool for delineating geological boundaries (faults, lateral
stratigraphic contacts, etc.),
• Allows accelerated evaluation of large data sets,
• Provides quantitative estimate of fault/fracture presence,
• Often enhances stratigraphic information that is otherwise difficult to
extract,
• Should always be calculated along dip – either through algorithm design or
by first flattening the seismic volume to be analyzed, and
• Algorithms are local - Faults that have drag, are poorly migrated, or separate
two similar reflectors, or otherwise do not appear locally to be discontinuous,
will not show up on coherence volumes.
In general:
• Stratigraphic features are best analyzed on horizon slices,
• Structural features are best analyzed on time slices, and
• Large vertical analysis windows can improve the resolution of vertical faults,
but smears dipping faults and mixes stratigraphic features.
5-66
Volumetric curvature and reflector shape
5-67
Statistical measures based on vector dip
• Reflector divergence and/or parallelism
• angular unconformities
• stratigraphic terminations?
• Reflector curvature
• flexures and folds
• unresolved or poorly migrated faults
• differential compaction
• Reflector rotation
• data quality
• wrench faults
5-68
Sign convention for 3-D curvature attributes:
Anticlinal: k > 0
Planar: k=0
Synclinal: k < 0
5-70 (http://www.srs.fs.usda.gov/bentcreek)
3D Curvature and Molecular Docking
5-71 (http://en.wikipedia.org/wiki/Molecular_docking)
3D Curvature
and Biometric
kneg < 0 Identification
kneg > 0
kpos < 0 of Suspicious
kpos > 0 Travelers
kpos > 0
kneg = 0
5-72
Geometries of folded surfaces
kpos < 0 kpos = 0 kpos > 0
synform
kneg < 0
bowl saddle
antiform
kneg = 0
plane
kneg > 0
dome
1. Mean Curvature
2. Gaussian Curvature Measures validity of
3. Rotation quadratic surface
4. Maximum curvature Mathematical Basis
Established use in
5. Minimum curvature fracture prediction
6. Most positive curvature Most useful for structural
7. Most negative Curvature interpretation
8. Dip curvature
9. Strike curvature
10. Shape index Established use in biometric
11. Curvedness ID and molecular docking
12. Shape index modulated by curvedness
5-75
kx-ky transform of time picks
0
-2
ky (cycles/m)
00
power
0.00
0
Short
wavelength 50 Footprint!
00
00
50
x y
(m +0.02
) -0.02 +0.00 +0.02 0
0 0 kx (cycles/m)
0 00
1
The horizon exhibits different scale structures such as domes and basins
on the broad-scale, faults on the intermediate-scale, and smaller scale
undulations.
Power
Long
0.00 wavelength
5000
y (m)
Contours
0 -5
0 5000 10000
x (m)
5-78 (Bergbauer et al, 2003)
Typical workflow for curvature
calculated along picked horizons
1. pick horizon
2. smooth horizon
3. calculate curvature on tight grid for short
wavelength estimates
4. smooth horizon some more
5. calculate curvature on coarse grid for
long wavelength estimates
5-79
Multispectral estimates of volumetric
curvature
Motivation:
• Structural Geology models relate curvature to fractures
• We have very accurate dip and azimuth volumes – can we use them to
generate more robust curvature volumes?
5-80
Radius of Curvature
3 km
1.0
Time (s)
1.2
5-81
Thermal imagery with sun-shading
Red=0.75
Green=1.00
Blue=1.25
1st derivative
fractional derivative
(or 1st derivative followed by a low pass filter)
amp
amp
0.0
-1.0 0.0
0.00 0.25 0.50
-20 0 +20
distance in grid points kx wavenumber (cycles/grid point)
ive
Filter applied to 1st derivative
t
4 0.25
riva
t de
s
fir
d
3 lize
dea
I
2 0.80
1.00
0 Wavelength, λ
infinite 0
8∆x
4∆x
2∆x
32∆x π/16∆x
16∆x
Wavenumber, k
π/∆x
π/8∆x
π/4∆x
π/2∆x
5-87
Vertical
B
Slice – Fort Worth Basin, USA
B′′
0.750
0.800
Caddo
t (s)
1.000
1.250
Ellenburger
s/m2
-.25
0.0
+.25
5 km
5-89 B
kmean horizon slice – Caddo
(volumetric calculation)
B′′
s/m2
-.25
0.0
+.25
5 km
5-90 B
Coherence horizon slice – Caddo
5 km
B′′
1.0
0.9
.08
5-91 B
Attributes extracted along time slices
5-92
Vertical slice through seismic
5 km
B B′′
0.750
0.800
Caddo
t (s)
1.000
1.250
Ellenberger
1.0
0.9
0.8
B
5-94 (al-Dossary and Marfurt, 2006)
Most negative curvature (α=1.00)
5 km
t=0.800 sB′′
s/m2
+.25
0.0
-.25
α=1.00
B
5-95 (al-Dossary and Marfurt, 2006)
Spectral estimates of most negative
5 km curvatureB′′ t=0.8 s
s/m2
+.25
0.0
-.25
α=0.25
=2.00
=1.75
=1.50
=1.25
=1.00
=0.75
=0.50
B
5-96 (al-Dossary and Marfurt, 2006)
Principal component coherence
5 km t=0.8 s B′′
1.0
0.9
0.8
B
5-97 (al-Dossary and Marfurt, 2006)
Principal component coherence
5 km t=1.2 s B′′
1.0
0.9
0.8
B
5-98 (al-Dossary and Marfurt, 2006)
Most negative curvature (α=0.25)
5 km t=1.2
B′′ s
s/m2
+.25
0.0
-.25
B
5-99
(al-Dossary and Marfurt, 2006)
Most positive curvature (α=0.25)
5 km t=1.2B′′s
s/m2
-.25
0.0
+.25
B
5-100
(al-Dossary and Marfurt, 2006)
Zero-crossings are less sensitive to
noise than peaks or troughs
noise
pick error
pick error
signal
pick error
pick error
signal+noise
pick error
A Neg Pos A’
1240 ms
Picked horizon
1520 ms
2.5 km
A’ A’
Most-positive Most-negative
2.5 km
A’ A’
Most-positive Most-negative
synform
2 k 2 + k1
s = − ATAN( ) s=-0.5
π k 2 − k1
k1 ≥ k 2 s=0.0
saddle
antiform
Principal curvatures
s=+0.5
s=+1.0
dome
Shape indices
saddle
valley
dome
ridge
bowl
0.2
curvedness
plane
0.0
-1.0 -0.5 0.0 +0.5 +1.0
Shape index
5-107
(al-Dossary and Marfurt, 2006)
Shape index modulated by curvedness
5 km (α=0.25)
5 km
B’
N
1 .2
)
e (s
Tim
1. 4
Saddle
Valley
Dome
Ridge
Bowl
0.2
Curvedness
0.0 Plane
-1.0
-0.5
+0.5
+1.0
0.0
N
1 .2
)
e (s
Tim
1. 4
Saddle
Valley
Dome
Ridge
Bowl
0.2
Curvedness
Transparent
0.0 Plane
-1.0
-0.5
+0.5
+1.0
0.0
1.0
bow l
Filter response
valle y
0.5 s addle
ridge
dom e
0.0
-1.0 -0.5 0.0 0.5 1.0
Shape index
Figure
5-110 3.68f (al-Dossary and Marfurt, 2006)
Shape components (α=0.25)
5 km t=1.2 s
high
Ridge
Dome
Saddle
Valley
Bowl
5-111
(al-Dossary and Marfurt, 2006)
‘Lineament’ attribute – an attempt to use shapes to
accentuate linear features
valley
1.0
Filter response
0.5
dome
saddle
bowl
ridge
0.0
-1.00 -0.50 0.00 0.50 1.00
Shape index
weak
5-113
(al-Dossary and Marfurt, 2006)
Strike
-90 -45 0 45 90
0.1
Curvature lineaments
Valley component
colored by azimuth
0.0
1.0
1.2
1.1
0.8sss
1.0
0.9 s
5-114
Volumetric view of lineament attribute (α=0.25)
-90
Strike
0 +90
Lineament
transparent
5 km
N
0.8
Time (s)
1.0
1.2
5-116
Coherence time slice at t=1.0 s
2km
Coh
A’ 1.0
0.9
0.8
A
5-117
Most negative curvature time slice at t=1.0 s
2km
A’
+.25
0.0
A -.25
5-118
Most positive curvature time slice at t=1.0 s
2km
A’
-.25
0.0
A +.25
5-119
Reflector rotation time slice at t=1.0 s
2km
A’
1.0
0.0
A -1.0
5-120
Vertical seismic slice
A A’
0.5
2km
Time (s)
1.0
1.5
5-121
Computational vs. Interpretational curvature
Channels seen by
curvature
5-123
Curvature
In Summary:
• Volumetric curvature extends a suite of attributes previously limited to
interpreted horizons to the entire uninterpreted cube of seismic data.
• The most negative and most positive curvatures appear to be the most
unambiguous of the curvature images in illuminating folds and flexures.
• Open fractures are a function of the strike of curvature lineaments and the
azimuth of minimum horizontal stress.
5-124
Lateral Changes in Amplitude and Pattern
Recognition
5-125
Thin bed tuning and the wedge model
0 thickness (ms) 50
0
Time (ms)
50
impedance 100
150
0
-0.1
Time (ms)
50
150
Time (ms)
50
seismic
100
150 env
2
0
Time (ms)
50
envelope 100
150
0
50
Trough to peak thickness (ms)
-0.025
40 -0.020
amplitude of trough
30 -0.015
20 -0.010
10 -0.005
0 -0.000
0 10 20 30 40 50 0 10 20 30 40 50
Temporal thickness (ms) temporal thickness (ms)
5-127
Sobel edge detector
(numerical approximation to first derivative)
∂u u ( x + ∆x) − u ( x − ∆x)
= lim
∂x ∆x − >0 2∆x
5-128
ERROR: stackunderflow
OFFENDING COMMAND: ~
STACK: