A Crash Course in Seismic Attributes

Kurt J. Marfurt (The University of Oklahoma)

Spectral Decomposition

2-1
 Course Outline
Geometric Attributes

Spectral Decomposition

Multiattribute Display

Poststack Seismic Data Conditioning

2-2
 Spectral Decomposition

After this section you will be able to:

• Identify the geological features highlighted by spectral

decomposition and wavelet transforms,

• Interpret spectral anomalies in the context of thin bed tuning

• Use CWT spectral components or spectral ‘ridges’ to

cosmetically broaden the spectrum,

• Use discontinuities in the magnitude and phase spectra as

indicators of unconformities, and

• Use spectral ratio techniques to estimate Q (1/attenuation).

2-3
 Fourier components of a broadband seismic pulse

2-4 (Aarre et al., 2012)

 Fourier components of a broadband seismic pulse

2-5 (Aarre et al., 2012)

 Thin bed tuning and the wedge model
0 thickness (ms) 50
0

Time (ms)
50
impedance 100

150

0
-0.1

Time (ms)
50

reflectivity +0.1 100

150

Time (ms)
50
seismic
100

150 Env
2
0

Time (ms)
50
envelope 100

150
0

2-6 (Partyka, 2001)

 Thin Bed Resolution

2-7 (Kallweit and Wood, 1982)

 A review of some math

3/4
/16

/2
/4
/8
0
Time (ms)

u
50 T  lim [  u(t  T / 2)  u (t  T / 2)]
T T 0

100

Tuning Data Complex spectrum

thickness
u (t )  U ( )
Note!
We can DETECT an amplitude u
difference between T=λ/8 and λ/16 !  iU ( )
But there is no change between peak- t
trough thickness!
Rotate each Increase amplitude of
frequency by 900 high frequencies
2-8
 Alternative Basis Functions

2-9
 SWDFT wavelets SWDFT spectra
1
Tapered window
1
0.5 0.8
0.6
fc=10 Hz 0
0.4
-0.2 -0.1 0.0 0.1 0.2
-0.5 0.2
0
-1 20 Hz10 20 30 40 50 60 70 80 127
90 100 Hz

1
1
0.5 0.8
0.6
0
fc=20 Hz 0.4
-0.2 -0.1 -0.5 0.0 0.1 0.2 0.2
0
-1 0 10 20 30 40 50 60 70 80 90 100

1.0

1
0.5
0.8
0.6
fc=40 Hz -0.2 -0.1
0.0
0.0 0.1 0.2 0.4

-0.5 0.2
0
-1.0 0 10 20 30 40 50 60 70 80 90 100

2-10 t (s) f (Hz)

 Morlet wavelets Morlet spectra
1.0 1
0.8
0.5 Tapered windows
0.6
0.0 0.4
fc=10 Hz -0.2 -0.1 0 0.1 0.2 0.2
-0.5 0
0 10 20 30 40 50 60 70 80 90 10
-1.0 0

1.0
1
0.8
0.5
0.6
0.0 0.4
fc=20 Hz -0.2 -0.1 0 0.1 0.2 0.2
-0.5 0
0 10 20 30 40 50 60 70 80 90 10
-1.0 0

1.0
1
0.8
0.5
0.6
0.0 0.4
fc=40 Hz -0.2 -0.1 0 0.1 0.2 0.2
-0.5 0
0 10 20 30 40 50 60 70 80 90 10
-1.0 0

2-11 t (s) f (Hz)

 Wavelets for spectral ratios Spectra
1.0 1
0.8
0.5
0.6

fc=10 Hz 0.0 0.4

0.2
-0.2 -0.1 0 0.1 0.2
-0.5 0
0 10 20 30 40 50 60 70 80 90 10
-1.0 0

1.0
1
0.8
0.5
0.6
0.4
fc=20 Hz -0.2 -0.1
0.0
0 0.1 0.2 0.2
-0.5 0
0 10 20 30 40 50 60 70 80 90 10
-1.0 0

1.0 1
0.8
0.5
0.6

fc=40 Hz 0.0 0.4

0.2
-0.2 -0.5 0 0.2
0
0 10 20 30 40 50 60 70 80 90 10
-1.0 0

2-12 t (s) f (Hz)

 Spectral Decomposition using the
Short Window Discrete Fourier Transform
(SWDFT)

2-13
 Source Seismic
Reflectivity wavelet Noise data
r(t) s(t) + n(t) u(t)
*
Long window

Time
spectral
decomposition
and the
convolutional Fourier Transform

model
Amplitude Amplitude Amplitude Amplitude
Frequency

× + =
Bandlimited white
White spectrum
spectrum
2-14 (Partyka et al., 1999)
 Spectral balancing

Peak amplitude, amax

a(f)

Noise threshold=εamax

Frequency, f

Peak amplitude, 1.0

abalanced(f)=a(f)/[a(f) +amax]
Abalanced(f)

Noise threshold=0.5

Frequency, f
2-15
 Source Seismic
Reflectivity wavelet Noise data
r(t) s(t) + n(t) u(t)
*
Short window
spectral
* + =

Time
decomposition
and the
convolutional Fourier Transform

model
Amplitude Amplitude Amplitude Amplitude
Frequency

× + =

Colored spectrum Bandlimited colored

spectrum
2-16 (Partyka et al., 1999)
 Spectral decomposition work flow

Time (s)
Time (s)

Interpret the horizons Window the data

Frequency (Hz)
Frequency (Hz)

Compute spectral components Analyze spectral slices

2-17 (Johann et al., 2003)

 Sensitivity of the Spectral Components to
Formation Thickness

2-18
 Thin bed tuning variation with thickness

A A 30 Hz
Time (s) 15 Hz

A A

A A

15 Hz Map 30 Hz Map
2-19 (Laughlin et al., 2002)
 Thin bed tuning variation with thickness

10 Hz 30 Hz 50 Hz

thalweg moderate thickness channel edges

2-20 (Courtesy of Apache Corp.)

 Thin bed tuning variation with thickness
50 Hz
10
30 Hz

thalweg
moderate
channelthickness
edges
2-21 (Courtesy of Apache Corp.)
 The thin bed tuning model

r1

r2=-r1

2-22 (Marfurt and Kirlin, 2001)

 Spectral response of the thin bed tuning model

r1=-2r2 r1=+2r2
r1=-r2 r1=+r2
r1=-0.7r2 r1=+0.7r2
r1=-0.5r2
r1=+0.5r2

2-23 (Marfurt and Kirlin, 2001)

 A A’ B B’
1.0
Time (s)

1.1

Synthetic data from picks

Amplitude Amplitudedata from picks
Synthetic
1.2

2 km
Freq
Time
A’ B’ (Hz)
(ms)
0

30
15

60
30

Amplitude
Peak frequency
Temporal fromfrom
thickness synthetic
picks data
A B

2-24 (Marfurt and Kirlin, 2001)

 Application of spectral decomposition and
coherence: Midcontinent, USA

2-25 (Peyton et al., 1998)

 Geologic cross section (from wells)

Flatten Flatten

2-26 (Peyton et al., 1998)

 Pick a continuous horizon parallel to eroded formation

A΄
A
South North
Wharton #1 Pester #2 Griffin #1 Jameson #1

L. Skinner

Flattened
on easy-
Novi
to-pick
reflector
320 ft 1.6 km
50 ms
100 m

2-27 (Peyton et al., 1998)

 Apply spectral decomposition
between phantom horizons
A A΄
South North
Wharton #1 Pester #2 Griffin #1 Jameson #1

L. Skinner
analysis
window
Novi

320 ft
50 ms 1.6 km
100 m

2-28 (Peyton et al, 1998)

 36 Hz spectral
component over
Red Fork

Bow River, 20 km SE of Calgary, AB

2-29 (Peyton et al., 1998)

 Geology 101: “What you see is what’s preserved!”

36 Hz spectral
component over
Red Fork

2-30 (Peyton et al., 1998)

 Coherence
extraction
along a
‘phantom
horizon’
36 ms below
Skinner

2-31 (Peyton et al., 1998)

 Light Source

Shaded
illumination
map of
Skinner-Novi
isochron

Stage V shows
up due to
differential
compaction
Stage V

2-32 (Peyton et al., 1998)

 Why do spectral components look ‘better’ than time-
thickness maps and amplitude slices?
• we use more information (~50 samples!) in the analysis! (we
only use 2 samples in the time-thickness map)
(response attributes, weighted attributes, and coherence
behave similarly)

• many of images actually don’t look like much of anything!

(rather a human interpreter chooses those that looked
geologically reasonable!)

2-33
 Red-Green-Blue color blending of three discrete
spectral magnitude volumes

2-34 (Leppard et al., 2010)

Blending of 3 spectral components for the wedge model

Z1<Z2>Z3 Z1<Z2<Z3

SWDFT SWDFT

CWT CWT

Matching Pursuit Matching Pursuit

2-35 (McArdle and Ackers., 2012)

 Simultaneous
display of 3
attributes will
therefore show
more of the data!

40 Hz  Blue
50 Hz  Green
60 Hz  Red

2-36 (Bahorich et al., 2002)

 Spectral Balancing to compensate for source
wavelet and energy loss with depth

amp
18 Hz  Red
24 Hz  Green
36 Hz  Blue

f (Hz)

amp
f (Hz)

1. For a window (e.g. 100ms) about each time slice, assume that the
geology is random
2. Calculate average spectrum at each time slice
3. Rescale spectral components such that the new average value is 1.0
4. Plot relative spectral amplitudes on each time slice

2-37
 Spectrum of a 100 ms analysis window
100
Mode

Amplitude
(peak) 50
Mean
(average)
0
20 40 60 80
Frequency (Hz)
Magnitude

Noise level

trough

Frequency

2-38
 Definition of peak spectral magnitude
and peak frequency
High
Peak Magnitude
Magnitude

0
0 10 20 30 40 50

Frequency (Hz)
Peak freq

2-39 (Blumentritt, 2008)

 Horizon slices along paleo Mississippi River
Freq
(Hz)
0

5 km 30

60

Coh
1.0

0.6

Peak Frequency
blended
Peak with Coherence
Frequency
Coherence
2-40 (Marfurt and Kirlin, 2001)
 Spectral Decomposition using the
Continuous Wavelet Transform
(CWT)

2-41
 1
0.8 Mother wavelet

Continuous Wavelet Transform

0.6
Amplitude

0.4

using Morlet wavelets, ψ(t)

0.2
0
-0.2
-0.4
-5 -4 -3 -2 -1 0 1 2 3 4 5
Time (s)
6
18
5 16
4 14
3 12

Amplitude
Amplitude

2 10
1 8
0 6
-1 4
-2 2
-3 0
-3 -2 -1 0 1 2 3 20 40 60 80 100 120
Time (s) Frequency (Hz)

2-42 (Matos and Marfurt, 2011)

 Continuous Wavelet Transform (CWT)
f  t       t 


1   t u  1 1  t u t 
(u,s )  fdt,u ,fs s  u   sf t(t )     dt
( )
CWT
Wf(u , xs )  (u, sf )(t) Wf 
 s  s  s  s s s 
The CWT can be interpreted as
60 ˆ s     sˆ   s  a band pass filter response at
50
each scale s
E s c a la

40
30
60 20
120
10
50 100
100 200 300 400 500 600 700 800 900 1000
Amostras

Frequency
40 80
Escala
Scales

A m p lit u d e

30 0.4
60
0.2

20 0 40
10 -0.2
20
100 200 300 400 500 600 700 800 900 1000
Time (ms)
Amostras

100 200 300 400 500 600 700 800 900 1000
Amostras
Samples

0.4
Amplitude

0.2

-0.2

100 200 300 400 500 600 700 800 900 1000
2-43 Amostras
Time (ms) (Matos and Marfurt, 2011)
 Forward CWT
Σ
Reflectivity Synthetic CWT Magnitude Voices ICWT

CWT
magnitude
pos

Le Nozze di Figaro

2-44 (Matos and Marfurt, 2011)

 Spectral Decomposition
using Matching Pursuit

2-45
 Read seismic trace Matching pursuit flow
Precompute seismic chart
wavelets Generate complex seismic trace
and their spectra
Set residual=original complex trace
Set complex spectrum=(0.0,0.0)

Calculate instantaneous envelope and frequency

of residual

Pick times and frequencies of the strongest envelopes

Least-squares fit complex wavelets to residual

Subtract complex wavelets from previous

residual to compute new residual

no Residual
energy < threshold value?

yes
Sum spectra of
2-46 component wavelets (Liu and Marfurt, 2005)
 Pennsylvanian Age Channels, CBP, Texas, USA
0.6 A A
Time (s)

1.060 s

1.6 Original data

2-47 (Liu and Marfurt, 2005)

 Convergence using a modified matching-pursuit algorithm
0.6 A A
Time (s)

1.6 Modeled
Modeleddata
data
after
after
1
4
8
16iterations
iteration
2iterations
iterations

2-48 (Liu and Marfurt, 2005)

 Convergence using a modified matching-pursuit algorithm
0.6 A A
Time (s)

1.6 Residual after 16

1 iterations
2
4
8 iteration
iterations

2-49 (Liu and Marfurt, 2005)

 Convergence using a modified matching-pursuit algorithm
0.6 A A
Amp
high

0
Time (s)

1.6 Wavelet amplitudes after 8

1
2
16iterations
4 iteration
iterations

2-50 (Liu and Marfurt, 2005)

 Convergence using a modified matching-pursuit algorithm
0.6 A A
Amp
high

0
Time (s)

1.6 40 Hz component after 8

2
4
1
16iterations
iteration
iterations

2-51 (Liu and Marfurt, 2005)

 Time Slice through Seismic Data
A
Pennsylvanian Channels
Time Slice t=1.060 s

2 km

Original data
2-52 A (Liu and Marfurt, 2005)
 Time Slice through Spectral Components
A
Pennsylvanian Channels
Time Slice t=1.060 s

2 km

20
10Hz
40
80
30
60
70
50
90 Hzcomponent
Hz component
component
2-53 A (Liu and Marfurt, 2005)
 Peak frequency modulated 10
Peak frequency (Hz)
90
by peak magnitude high

Mag above average

A

low

2 km

2-54
A (Liu and Marfurt, 2005)
 2 km

Peak frequency
modulated by
peak amplitude
(Time slices)

30
Peak Mag

010 20 40 60 80 100
Peak Freq (Hz)

Time (ms)

2-55 (Liu and Marfurt, 2005)

 100 100 100
After balancing

Amplitude

Amplitude
Amplitude

50 50 50

0 0 0
20 40 60 80 20 40 60 80 20 40 60 80
Frequency (Hz) Frequency (Hz) Frequency (Hz)

A Low High

Horizon slice
through peak
A’ frequency and
peak amplitude
above Atoka
unconformity
2 km

2-56 (Liu and Marfurt, 2007)

 Vertical slice through seismic
A A’

High
1.0
Atoka
Time (s)

1.2
Low
(a)

100 100 100

After balancing

Amplitude
Amplitude
Amplitude

50 50 50

0 0 0
20 40 60 80 20 40 60 80 20 40 60 80
Frequency (Hz) Frequency (Hz) Frequency (Hz)
(b) (c) (d)

Low High

2-57 (Liu and Marfurt, 2007)

 Motivation

The mode is a good first-order approximation of the spectrum:

• peak magnitude,
• peak frequency, and
• Peak phase

Moment-based measures are useful for nearly-Gaussian spectra.

The spectra of balanced data are NOT Gaussian.

We have developed several statistical attributes including:

• bandwidth,
• range-trimmed mean,
• spectral slope, and
• roughness

2-58 (Zhang et al., 2008)

 Theory
High

roughne
ss
Magnitude

slope

bandwidth
0.00
Jlow Jhigh Frequency index

1.00
Percentile

phigh

0.50
plow

0.00
Jlow J50 Jhigh Frequency index

Schematic illustration of spectral attributes

2-59 (Zhang et al., 2008)
 Theory

 Bandwidth
The frequency distance between Jhigh and Jlow

Range trimmed mean

J
1 high

mRTM   mj
J high  J low  1 j  J low
Spectral slope
The slope that best fits the spectrum between Jlow and Jhigh

Spectral Roughness
A measure of how well the spectrum is fit by a linear
variation between Jlow and Jhigh .

2-60 (Zhang et al., 2008)

 A A’
0.8

0.9

1.0
Time

1.1

1.2

1.3

Slope 1 > Slope 2 < Slope 3

2-61 5 f (Hz) 120 (Zhang et al., 2008)

 N
N

Peak mag
1 High
km

1
km

Peak magnitude (Zhang et al.,

2-62 (Phantom horizon slice 40 ms above the picked horizon) 2008)
 N

RT-mean
1 High
km

Range trimmed mean (Zhang et al.,

2-63 (Phantom horizon slice 40ms above the picked horizon) 2008)
 N

RT-mean
1 High
km

Range trimmed mean (Zhang et al.,

2-64 (Phantom horizon slices 65-15 ms above picked horizon 2008)
 N
Spectral
slope
1 Positive
km
0

Negative

Slope (Zhang et al.,

2-65 (Phantom horizon slice 40ms above the picked horizon) 2008)
 N

Roughness
1 High
km

Roughness (Zhang et al.,

2-66 (Phantom horizon slice 40ms above the picked horizon) 2008)
 Peak phase Peak frequency

Coherence

2-67 (Liu and Marfurt, 2006)

 Alternative spectral decomposition algorithms

Continuous Wavelet
Short Window DFT Transform (CWT)

2-68 (Leppard et al., 2010)

 Comparison of Short Window DFT
SWDFT, CWT, and
matching pursuit
spectral
decomposition
algorithms

Continuous Wavelet Transform (CWT)

2-69 (Leppard et al., 2010)

 SWDFT CWT Matching Pursuit

Comparison of
constant
bandwidth, CWT,
and matching
pursuit spectral
decomposition
algorithms
SWDFT CWT Matching Pursuit

SWDFT CWT Matching Pursuit

2-70 (Leppard et al., 2010)

 Spectral Balancing

2-71
 Average time-frequency spectrum (106 traces)

Mag
High

(Computed using CWT)

2-72 (Qi and Marfurt, 2014)
 Average time-frequency spectrum

Mag
High

(After spectral balancing and bluing)

2-73 (Qi and Marfurt, 2014)
 Original seismic amplitude
A A’
Amp
Positive
0.6 Top Large scale
Marble karst doline 0
Fall
Negative
Top
0.8 Ellenburger
Collapse
features
Collapse
features
Time (s)

1.0

1.2

1.4

400 600 800 1000 1200 1400

CDP no.
2-74 (Qi and Marfurt, 2014)
 Balanced and blued seismic amplitude
A A
Amp

0.6 Top
’ Large scale
Positive

Marble karst doline 0

Fall
Negative
Top
0.8 Ellenburger
Collapse
features
Collapse
features
Time (s)

1.0

1.2

1.4

400 600 800 1000 1200 1400

CDP no.
2-75 (Qi and Marfurt, 2014)
 Bandwidth Extension

2-76
 Inverse CWT

Reflectivity Synthetic CWT Magnitude Voices ICWT

CWT
magnitude
pos

2-77 (Matos and Marfurt, 2011)

 ICWT “deconvolution” workflow

CWT CWT ICWT

Shrunken
Morlet MML Morlet
Reflectivity Synthetic CWT Magnitude CWT MML voices

CWT
magnitude

X
pos

2-78 (Matos and Marfurt, 2011)

 Barnett Shale (Original Seismic Amplitude)

2-79 (Matos and Marfurt, 2011)

 “Bandwidth extension” using the inverse CWT

2-80 (Matos and Marfurt, 2011)

 Sparse-spike frequency domain inversion

2-81 (Matos and Marfurt, 2011)

 “Bandwidth extension” using the inverse CWT
Marble Falls
Amplitude
60

0
Upper Barnett Lm
Upper Barnett Sh
60

Forestburg
Lower Barnett Sh

Viola
2-82 (Matos and Marfurt, 2011)
 Sparse-spike frequency domain inversion
Marble Falls
Amplitude
10

0
Upper Barnett Lm
Upper Barnett Sh
10

Forestburg
Lower Barnett Sh

Viola
2-83 (Matos and Marfurt, 2011)
 Geotrace’s bandwidth extension using the
inverse CWT
before after

Synthetic from well

Synthetic from well

2-84 (Smith et al., 2008)

 before after
te
r bona

Geotrace’s
a
To pc

bandwidth
Vertical extension
slice AA’ using inverse
CWT
Phantom horizon slice
20 ms below top
carbonate’

A A’ A A’

2-85 (Smith et al., 2008)

 Migrated CRPs before and after bandwidth extension
before after

2-86 (Smith, 2011)

 Angle stack power spectra before and after bandwidth
extension

before

after

2-87 (Smith, 2011)

 Seismic inversion products after bandwidth extension

2-88 (Smith, 2011)

 Lumina’s “bandwidth extension”
(through sparse spike spectral inversion)

2-89
Thickness estimate for an even impulse function
calculation
Impulse Function

g(t, θ) = r1*δ(t - t1)+r2*δ(t - t1 - T)

Center at t=0

g(t, θ) = r1*δ(t – T/2) + r2*δ(t + T/2)

Fourier Transform
G(f) = r1 (θ)exp(-i2πf [t + T/2]) + r2 (θ)exp(-i2πf [t - T/2])

Real part If we divide out the Wavelet

Re[G(f)] = (r1e + r2e)cos(πfT)

Take the derivative

d/df{Re[G(f)]} = - (r1e + r2e)πTsin(πTf)

An ugly transcendental equation

Divide derivative by real part
d/df {Re[G(f)]} / Re[G(f)] = - πTtan(πTf)
2-90 Solve for T (Puryear and Castagna, 20
Lumina’s bandwidth extension through spectral inversion
Poisson’s Ratio estimate through a Haynesville survey

Without spectral inversion

2-91 (Courtesy Lumina Geophys
Lumina’s bandwidth extension through spectral inversion
Poisson’s Ratio estimate through a Haynesville survey

With spectral inversion

2-92 (Courtesy Lumina Geophys
 Q – Quality Factor

A silver bell A rubber bell

High quality sound Low quality sound
High price ( $65) Low price ( $1)

2-93
 Seismic Estimation of Q

Spectral Ratio (SR) Method

and are the spectra of source and
 target wavelets, respectively.

  ¿
Reference
Magnitude

 𝜋 𝑓 (𝑡 1−𝑡 0)/𝑄=−ln¿
Target

ln[B(t1,f)/B(t0,f)]
0 Frequency Slope~-1/Q

Frequency

2-94 (White, 1992)

Q Estimation -
Log Spectral Ratio Method

1/Q

2-95 (Singleton, 2008)

Q Estimation -
Log Spectral Ratio Method

1/Q

2-96 (Singleton, 2008)

 Seismic Estimation of Q

Peak Frequency Shift (PFS) Method

andare the peak frequencies of the

 source and target wavelets,
respectively.
Magnitude

Reference
2
  𝜋 (𝑡 1 −𝑡 0)𝑓 𝑓
𝑝 𝑝
𝑄= 1 0

¿¿
Target

0 Peak Peak Frequency

Frequency Frequency

2-97 (Zhang and Ulrych, 2002)

 Seismic Estimation of Q

Centroid Frequency Shift (CFS) Method

 ofandare the centroid frequencies

source and target wavelets,
respectively. is the spectral
variance of source wavelet.
Magnitude

  𝜋 (𝑡 1 −𝑡 0) 𝜎 2𝑏
𝑄= 0

𝑓 𝑐 −𝑓𝑐
0 1

0 Centroid Centroid Frequency

Frequency Frequency

2-98 (Quan and Harris, 1997)

 Seismic Attenuation “Attributes”

Low-frequency
Attenuation Slope

Attenuation Slope Attributes

Magnitude

High-frequency
Attenuation Slope

0 Frequency

2-99 (Courtesy of Huailai Zhou and Fangyu Li)

 Seismic Attenuation “Attributes”

High-
frequency Energy Difference Attribute
Magnitude

Energy Decay

Low-frequency
Energy
Augmentation
0 Frequency

2-100 (Courtesy of Huailai Zhou and Fangyu Li)

 GR Hydrocarbon Detection (China)
AC
RT Well A DEN Well B Well C

分1井-双庙101井-双庙1井-铁北1井-双庙102井-雷北1井须家河组含气性对比图
High-frequency
地层 新分
层
0

2
GR
分1井
150

RT 100000
深度 岩性

23 00
150

1
AC

DEN
20

3
0

2
GR
双庙101井
150

RT 100000
深度 岩性
150

1
AC

DEN
20

3
0

2
GR
双庙1井
150

RT 100000
深度 岩性

18 00
150

1
AC

DEN
20

3
0

2
GR
铁北1井
150

RT 100000
深度 岩性
150

1
AC

DEN
20

3
0

2
GR
双庙102井
150

RT 100000
深度 岩性
150

1
AC

DEN
20

3
0

2
GR

RT
150
雷北1井

100000
深度 岩性
150

1
AC

DEN
20

3
Attenuation Slope
80 0
TX
61 1 90 0
23 00

19 00

19 00
24 00
T3X6
TX
62 90 0

24 00
2 00 0
20 00

20 00
25 00
T3X5
10 00
TX 25 00
41 2 10 0
21 00

21 00
26 00

TX
42 11 00
T3X4 26 00
2 20 0
22 00

22 00
27 00

TX
43
12 00

27 00
2 30 0
23 00

T3X3 23 00
28 00

TX
21
13 00

28 00
2 40 0
24 00
TX
22 24 00
29 00

T3X2
14 00

29 00
2 50 0
25 00
TX
23
30 00

15 00

泥岩层 气层 干层 气水同层 水层

Depth High
Lithology
Well A Well B Well C

H2
Low
H1

2-101 (Courtesy of Huailai Zhou, Chengdu Univ.)

 Hydrocarbon Detection

Well A

H1 Horizon slice Well B

High-frequency
Attenuation Slope
High

Well C
Low

Well A Well B Well C

分1井-双庙101井-双庙1井-铁北1井-双庙102井-雷北1井须家河组含气性对比图
分1井 双庙101井 双庙1井 铁北1井 双庙102井 雷北1井
0 GR 150 150 AC 20 0 GR 150 150 AC 20 0 GR 150 150 AC 20 0 GR 150 150 AC 20 0 GR 150 150 AC 20 0 GR 150 150 AC 20
新分
地层 层 深度 岩性 深度 岩性 深度 岩性 深度 岩性 深度 岩性 深度 岩性
2 RT 100000 1 DEN 3 2 RT 100000 1 DEN 3 2 RT 100000 1 DEN 3 2 RT 100000 1 DEN 3 2 RT 100000 1 DEN 3 2 RT 100000 1 DEN 3

1800
2300

800
TX
61 1900
2300

1900

1900
2400
T3X6
TX
62 900

2400
2000
2000

2000
2500
T3X5
1000
TX 2500
41 2100
2100

2100
2600

TX
42 1100
T3X4 2600
2200
2200

2200
2700

TX
43
1200

2700
2300
2300
T3X3
2300
2800

TX
21
1300

2800
2400
2400
TX
22 2400
2900

T3X2
1400

2900
2500
2500
TX
23

(Courtesy of Huailai Zhou, Chengdu Univ.)

3000

2-102
1500

泥岩层 气层 干层 气水同层 水层
 Hydrocarbon Detection

Well A

H2 Horizon slice Well B

High-frequency
Attenuation Slope
High

Well C
Low

Well A Well B Well C

分1井-双庙101井-双庙1井-铁北1井-双庙102井-雷北1井须家河组含气性对比图
分1井 双庙101井 双庙1井 铁北1井 双庙102井 雷北1井
0 GR 150 150 AC 20 0 GR 150 150 AC 20 0 GR 150 150 AC 20 0 GR 150 150 AC 20 0 GR 150 150 AC 20 0 GR 150 150 AC 20
新分
地层 层 深度 岩性 深度 岩性 深度 岩性 深度 岩性 深度 岩性 深度 岩性
2 RT 100000 1 DEN 3 2 RT 100000 1 DEN 3 2 RT 100000 1 DEN 3 2 RT 100000 1 DEN 3 2 RT 100000 1 DEN 3 2 RT 100000 1 DEN 3

1800
2300

800
TX
61 1900
2300

1900

1900
2400
T3X6
TX
62 900

2400
2000
2000

2000
2500
T3X5
1000
TX 2500
41 2100
2100

2100
2600

TX
42 1100
T3X4 2600
2200
2200

2200
2700

TX
43
1200

2700
2300
2300
T3X3
2300
2800

TX
21
1300

2800
2400
2400
TX
22 2400
2900

T3X2
1400

2900
2500
2500
TX
23

(Courtesy of Huailai Zhou, Chengdu Univ.)

3000

2-103
1500

泥岩层 气层 干层 气水同层 水层
 GR AC Hydrocarbon Detection
RT Well A DEN Well B Well C

分1井-双庙101井-双庙1井-铁北1井-双庙102井-雷北1井须家河组含气性对比图

High-frequency
分1井 双庙101井 双庙1井 铁北1井 双庙102井 雷北1井
0 GR 150 150 AC 20 0 GR 150 150 AC 20 0 GR 150 150 AC 20 0 GR 150 150 AC 20 0 GR 150 150 AC 20 0 GR 150 150 AC 20
新分
地层 层 深度 岩性 深度 岩性 深度 岩性 深度 岩性 深度 岩性 深度 岩性
2 RT 100000 1 DEN 3 2 RT 100000 1 DEN 3 2 RT 100000 1 DEN 3 2 RT 100000 1 DEN 3 2 RT 100000 1 DEN 3 2 RT 100000 1 DEN 3

18 00
23 00

80 0

Energy Decay
TX
61 1 90 0
23 00

19 00

19 00
24 00
T3X6
TX
62 90 0

24 00
2 00 0
20 00

20 00
25 00
T3X5
10 00
TX 25 00
41 2 10 0
21 00

21 00
26 00

TX
42 11 00
T3X4 26 00
2 20 0
22 00

22 00
27 00

TX
43
12 00

27 00
2 30 0
23 00

T3X3 23 00
28 00

TX
21
13 00

28 00
2 40 0
24 00
TX
22 24 00
29 00

T3X2
14 00

29 00
2 50 0
25 00
TX
23
30 00

15 00

泥岩层 气层 干层 气水同层 水层

Depth
Lithology
Well A Well B Well C

High
H2

H1

Low

2-104 (Courtesy of Huailai Zhou, Chengdu Univ.)

 GR AC
Hydrocarbon Detection
Well A Well B Well C
RT DEN
分1井-双庙101井-双庙1井-铁北1井-双庙102井-雷北1井须家河组含气性对比图

Low-frequency
分1井 双庙101井 双庙1井 铁北1井 双庙102井 雷北1井
0 GR 150 150 AC 20 0 GR 150 150 AC 20 0 GR 150 150 AC 20 0 GR 150 150 AC 20 0 GR 150 150 AC 20 0 GR 150 150 AC 20
新分
地层 层 深度 岩性 深度 岩性 深度 岩性 深度 岩性 深度 岩性 深度 岩性
2 RT 100000 1 DEN 3 2 RT 100000 1 DEN 3 2 RT 100000 1 DEN 3 2 RT 100000 1 DEN 3 2 RT 100000 1 DEN 3 2 RT 100000 1 DEN 3

18 00
23 00

80 0

Energy
TX
61 1 90 0
23 00

19 00

19 00
24 00
T3X6
TX
62 90 0

24 00
2 00 0
20 00

20 00
25 00
T3X5

Augmentation
10 00
TX 25 00
41 2 10 0
21 00

21 00
26 00

TX
42 11 00
T3X4 26 00
2 20 0
22 00

22 00
27 00

TX
43
12 00

27 00
2 30 0
23 00

T3X3 23 00
28 00

TX
21
13 00

28 00
2 40 0
24 00
TX
22 24 00
29 00

T3X2
14 00

29 00
2 50 0
25 00
TX
23
30 00

15 00

泥岩层 气层 干层 气水同层 水层

Depth Lithology
Well A Well B Well C

High
H2
H1

Low

2-105 (Courtesy of Huailai Zhou, Chengdu Univ.)

 Application of Q “attributes”

 Intrinsic attenuation is currently used in

 direct hydrocarbon detection.
 identifying high TOC source rock.
 estimating pore pressure.

 Can we develop attributes that are sensitive to

geometric attenuation due to
 Natural fractures?
 Hydraulically induced fractures?

2-106
 Attenuation from Natural and Induced Fractures
Attributes for fracture delineation
Barnett Shale

The Barnett shale lies

between the Marble Falls
Limestone and Viola
unconformity and is
separated into Upper Barnett (a) Time structure of Upper Barnett Shale and (b)
and Lower Barnett. Lower Barnett Shale, (c) Most negative curvature, (d)
Most positive curvature, (e-f) energy ratio similarity of
Upper Barnett Shale and Lower Barnett Shale.
2-107
 Attenuation from Natural and Induced Fractures

Spectral ratio Peak frequency shift Centroid frequency

(SR) method (PFS) method shift (CFS) method

Alternative estimates of attenuation between the Base Forestburg and Top

Viola Limestones (> 400 hydraulically fractured wells)

2-108(Courtesy of Fangyu Li and Huailai Zhou. Data courtesy of Devon Energy.)

 Attenuation from Natural and Induced Fractures
Most negative curvature

Most positive curvature

Centroid frequency shift (CFS) method

2-109(Courtesy of Fangyu Li and Huailai Zhou. Data courtesy of Devon Energy.)

 Attenuation from Natural and Induced Fractures
(Time-Lapse Seismic)

2-110 (Goodway et al., 2012)

 Attenuation from Natural and Induced Fractures
(Time-Lapse Seismic)

Δamp ΔT

2-111 (Goodway et al., 2012)

 Baseline Monitor Difference
Change in
spectra

Unstimulated zone

Stimulated zone

2-112 (Cho et al., 2013)

 Baseline Monitor Difference
Change in
spectra
With Stimulation
Baseline Baseline Without Stimulation
Monitor Monitor

Stimulated zone Unstimulated zone Differences

2-113 (Cho et al., 2013)

 Map showing
changes in 35
Hz component
suggesting
fluid mobility

2-114 (Cho et al., 2013)

 Unresolved Questions
 Can we use changes in attenuation as a function of
azimuth to identify the orientation of natural or
hydraulically induced fractures?

 Can we use attenuation as a function of offset to map

properties of rugose vs. specular reflectors?

 Can we develop simple models of geometric

attenuation (scattering) for heterogeneous media to
explain the “negative-Q” estimates from the Barnett
Shale?

2-115
 Stabilized Inverse-Q filter

2-116 (Wang, 2006)

 Original migrated data

2-117 (Wang, 2006)

 After inverse-Q filter on amplitude and phase

2-118 (Wang, 2006)

 Q Responds to Effective Stress

Frequency Normal Pressure Frequency

Effective Pore High Q

Frequency stress Over Pressure Pressure Frequency

Low Q
2-119 (Courtesy of eSeis)
 Q-Based Pore-Pressure Procedure
1. Compute Q 2. Correlate mud
from spectral weights to Q
3. Predict pore pressure
components Mud weights Q

Weight
2 lbs/gal
Time (s)

20
3

15
4
10
5
2 km

2-120 (Courtesy of eSeis)

 Q-based shale pore-pressure prediction
Weight
2 km
Well A lbs/gal
Well B
0 20

15
1

10
2
Time (s)

Pore pressure calibrated to spectral analysis Q at wells

2-121 (Courtesy of eSeis)

 Velocity-based shale pore-pressure prediction
Weight
2 km
Well A lbs/gal
Well B
0 20

15
1

10
2
Time (s)

Pore pressure calibrated to velocity at wells

2-122 (Courtesy of eSeis)

 Gulf of Mexico Application of Pore-Pressure Prediction
0

Shale
2

Sandy

Depth (km)
3
Shale
4

Carbonate
5

Fro
y
cit
1.0 1.5 2.0

lo Pressure (specific gravity)

mQ
ve
rom
2-123 F (Courtesy of eSeis)
 Impact of dry gas on spectra

2-124
 90

80
70

60 Time
Spectra 50
Domain
Before 40
Wavelet
Filtering 30

20
Before
10 Entering
0 Reservoir
0 20 40 60 80 100

90.00%

80.00%

Compressible 70.00%

Fluid 60.00%
Energy Loss

Energy 50.00%

Loss Function
40.00%
Compressible
Fluid -
30.00%

20.00%

10.00%
Gas Reservoir
0.00% Filter
0 10 20 30 40 50 60 70 80 90

Hz

90

80

70

Spectra 60

After 50 Time
Filtering 40 Domain
30
Wavelet
20

10
After
0
Exiting
0 10 20 30 40 50 60 70 80 90
Filter
2-125 (courtesy of Apex Spectral Technology, Inc.)
 Dry Gas Attenuation Signature

90

80

70
Dominant
60 Frequency or
“DF” drops due
50 to the filtering
effect of gas as
40 the traveling
wavelet passes
30
through the
reservoir
20

10

0
0 10 20 30 40 50 60 70 80 90

2-126 (courtesy of Apex Spectral Technology, Inc.)

 Dry Gas Attenuation Signature

90 The distance in Hz from Dominant to a user

–defined fractional energy level is called
80 Delta-Frequency or δf

70

60

50 Delta-
Frequency
shortens due to
40 the filtering
effect of gas as
30 the traveling
wavelet passes
20 through the
δf reservoir
10
δf
0
0 10 20 30 40 50 60 70 80 90

2-127 (courtesy of Apex Spectral Technology, Inc.)

 Wharton County, TX: TRI C Webernick-Goff No 1; Cum. Gas Prod:
2.9 BCF

δ f curve
US Patent
Application No.
11/903,602

Wilcox reservoir
exhibits strong δf
signature

2-128 (courtesy of Apex Spectral Technology, Inc.)

 Sensitivity of the Low Frequency Part of
the Spectrum to Hydrocarbons

2-129
 Estimating fluid mobility (permeability and viscosity)

  R r()

R0 Permeable
tn Impermeable
t

t t t n = i f n

r() = R0 - R0(1- R0 )[exp(-it)] )[1- R1nn exp(-itn)] + R1nn exp(-itn)

Top Bottom Lenses

2-130 (Courtesy Gennady Goloshubin)

 Fluid mobility attributes
1. Low frequency amplitude

2. Low frequency slope

r
3. High frequency decay
f
Balanced amplitude, r(f)

fluid
mobility reservoir geometry dominant
dominant r
f

Impermeable
Permeable

0 10 20 30 40 50 60 70
Frequency (Hz)

2-131 (Courtesy Gennady Goloshubin)

 Estimating fluid mobility
(Western Siberia)

1.5
Time (s)

sandstone
2.0 reservoir

fractured
shale
2.5 reservoir

2-132 (Goloshubin et al. 2002)

 Estimating fluid mobility

Oil
Oil and Water
Water
Oil-Water contact predicted by
low frequency analysis
101
63

77 86
Blind Test 73
6
4 Calibration wells 93

• 3 Oil 74
3
• 1 Water
75 78

2-133
(Goloshubin et al, 2002)
 Estimating fluid mobility
1200

ry
Production rate (m3/day)

eo
Th
800

400

0
0 4 8 12 16 20
2
 fluid  r 
mobility     f
  f 
2-134
(Goloshubin et al, 2002)
 Frequency-Dependent AVO
(Gulf of Mexico, U.S.A.)

Low Frequency AVO Conventional

Conventional AVO

2-135 (Courtesy Gennady Goloshubin)

 Exploiting Waveform Singularities

2-136
 Earlier recognition of waveform singularities
Original data Envelope

component)
(real
component)
Quadrature
(imaginary

+180
Phase

-180
Weighted average frequency
Frequency

Singularity
2-137 (Taner et al, 1979)
 Instantaneous Frequency
0.0

1.0
Time (s)

2.0

3.0

2-138
(Courtesy Rock Solid Images)
 Envelope weighted Instantaneous Frequency
0.0

1.0
Time (s)

2.0

3.0

2-139
(Courtesy Rock Solid Images)
 Thin bed indicator (finst-<finst>)
0.0

1.0
Time (s)

2.0

3.0

2-140
(Courtesy Rock Solid Images)
 SPectral Imaging of Correlative Events (SPICE)
Fit the spectral amplitudes r(s,t=t0) in a log-log plot with a
straight line. The slope of this line is the Hölder exponent, h:

h=0.684
log2(r(s,t=t0))

log2(s=scale)
2-141 (Li & Liner, 2004)
The Hölder exponent, h, measures the strength of a
singularity

h = ∞ (An analytical function)

h=0 (a Heaviside function)

h = –1 (a Dirac distribution)

The higher the value of h, the more regular the function

2-142
 Hölder exponent on Well Data
x104
1.6

Acoustic Impedance (AI)

0.6

+1.0
Hölder exponent from AI
0.0

+0.1
0.0 Reflection Coefficient (RC)
-0.1
0.0

-0.1 Hölder exponent from RC

+1.0

0.0 40 Hz Synthetic (S)

-1.0
+1.0

0.0 Hölder exponent from S

-1.0

Time (samples)
2-143 (Smyth et al., 2004)
 1 km Mapping unconformities
1.0
Time (s)

Seismic amplitude

1.5

1.0
Time (s)

Hölder exponent

1.5
2-144 (Liner et al, 2004)
 Seismic through a Miocene channel
Gulf of Mexico, USA
A A

1.0
Time (s)

1.5

2.0

2-145 (Liner et al, 2004)

 Horizon slices along the Miocene horizon
A A

A A

2-146 Coherence SPICE (Liner et al, 2004)

 Magnitude and phase of a single spike

2-147 (Matos et al., 2010)

 Magnitude and phase of a double spike

2-148 (Matos et al., 2010)

 The SPICE algorithm maps discontinuities in spectral
magnitude

What about discontinuities in spectral phase?

Amplitude
Positive

Zero

Negative
Thickness (ms)

a) b) c) d)

0 50
0
Time (s)

0.125

0.250

2-149 (Matos et al., 2010)

 a ) 18 ms CWT Magnitude CWT Phase
0 0
Amplitude Phase

Positive +π

-π
Zero

Time (s)
Time (s)

0.25 0.25
4 Frequency (Hz) 84 4 Frequency (Hz) 84

b) 23 ms
0 0 Phase
Amplitude +π
Positive
0

-π
Zero
Time (s)

Time (s)

0.25 0.25
2-150
4 Frequency (Hz) 84 4 Frequency (Hz) 84
 c ) 25 ms CWT Magnitude CWT Phase
0 0 Phase
Amplitude +π
Positive
0

-π
Zero

Time (s)
Time (s)

0.25 0.25
4 Frequency (Hz) 84 4 Frequency (Hz) 84

d ) 48 ms
0 0 Phase
Amplitude +π
Positive
0

-π
Zero
Time (s)

Time (s)

0.25 0.25
2-151
4 Frequency (Hz) 84 4 Frequency (Hz) 84
 Computation of phase residues in time-frequency domain

Phase
0 (radians)
+π

0   r     dr    r0 
-π 
Time (s)

Frequency (Hz)

35 40 45
0.100

0.25 Δφ4
4 Frequency (Hz) 84
Δφ1 A Δφ3
Time (s)

Δφ2 Δφ8
0.125
ГA Δφ5 B Δφ7
Δφ6

ГB
0.150
2-152
 Computation of phase residues in time-frequency domain
Frequency (f) Frequency (f)

Time (t)
Time (t)

Δφ4 Δφ8
0.2π 0.4π 0.6π 0.2π 0.4π 0.6π

Δφ1 Δφ3 Δφ5 Δφ7

Δφ2 Δφ6
-0.2π -0.4π -0.8π -0.2π -0.4π -0.8π

-0.4π -0.4π -0.6π -0.4π -0.4π -0.6π

Δφ1=-0.4π/2π Δφ5=-0.8π/2π

Δφ2=-0.2π/2π Δφ6=-0.4π/2π

Δφ3=0.8π/2π Δφ7=W(1.4π)=-0.6π/2π

Δφ4=-0.2π/2π Δφ8=-0.2π/2π

I= Δ φ1+ Δφ2+Δφ3+ Δφ4=0 continuous I= Δ φ5+ Δφ6+Δφ7+ Δφ8=-1 residue

2-153
 Phase residues of a pinchout

Amplitude
0
Positive

Zero

Negative
Time (s)

0.25
0 50 0
Thickness (ms)
2-154 (Matos et al., 2010)
 Mapping clinoforms associated with
Progradation-Regression
0.0
Time (s)

0.5

1.0

2-155 (Matos et al., 2012)

 Mappig clinoforms associated with
Progradation-Regression
Seismic Amplitude CWT Relative Impedance Inversion
0.5
Time (s)

0.7

0.9

CWT Deconvolution CWT Phase Residues

0.5
Time (s)

0.7

0.9

2-156 (Matos et al., 2012)

 Incised Valleys (Anadarko Basin)

2-157
 Incised Valleys (Anadarko Basin)

2-158 (Davogustto et al., 2012)

 Incised Valleys (Anadarko Basin)

2-159 (Davogustto et al., 2012)

 Incised Valleys (Anadarko Basin)

2-160 (Davogustto et al., 2012)

 Incised Valleys (Anadarko Basin)

2-161 (Davogustto et al., 2012)

 Time-structure map – top of Red Fork

The Watonga data set revisited

2-162 (Davogustto et al., 2012)

 Incised Valleys (Anadarko Basin)

Regional Red Fork Stages I - V

2-163 (Davogustto et al., 2012)

 Spectral Decomposition
In Summary:
• Constructive and destructive interference from the top and bottom of a thin bed give rise to
changes in the seismic amplitude and phase spectra

• Components of these spectra can be used to detect lateral changes in layer thickness and
heterogeneity, well below the limits of classic /4 seismic resolution

• Phase components delineate faults

• Peak spectral frequency (the mode of the spectrum) is a good zero order representation of
the seismic spectral response. Other statistical measures may help delineate upward-fining
and upward-coarsening sequences.

• Matching pursuit spectral decomposition provides less vertical mixing of stratigraphy than
that based on fixed-length window discrete Fourier transforms.

• Low frequency anomalies are often associated with low GOR – some interpret this as a
viscosity effect, others as complex waveform healing…

• Complex trace attributes degenerate when multiple reflectors interfere with each other – the
SPICE algorithm and phase residues use this interference as signal

2-164
 Pattern-recognition Math

sinx
 six
?
?
n

2-165