Sei sulla pagina 1di 128

Attributes for Resource Plays

Kurt J. Marfurt (The University of Oklahoma)


Satinder Chopra (Arcis)

A Short Overview of Geometric Attributes

2-1
Course Outline
A short overview of spectral decomposition
A short overview of geometric attributes
Attribute prediction of fractures and stress

Interactive multiattribute analysis


Statistical multiattribute analysis
Unsupervised multiattribute classification
Supervised multiattribute classification
Case Study Prediction of rate of penetration in the Mississippi Lime Play

Impact of acquisition and processing on seismic attributes


Post-stack data conditioning
Inversion for acoustic and elastic impedance

Attributes and hydraulic fracturing of shale reservoirs


Quantitative interpretation case studies

1-2
Observation:
Given sufficient skill and time, geometric attributes do not extract any
additional information over traditional map-based interpretation
workflows,
but
Geometric attributes

Are excellent in extracting subtle 3D features that are easily overlooked

Highlights all the features, not just those on horizons you had time to pick,
including those between horizons that can be picked

Provides results for large surveys in hours or days, which may be critical in
making acreage decisions

Provide quantitative measurements that can be used in


Risk analysis
Geostatistics, multilinear regression, or artificial neural network
prediction of fracture density, lithology, porosity, frackability,
2-3
Learning Objectives

After this section you will be able to:

Use co-rendered coherence and curvature volumes to rapidly


image structural style

Use co-rendered coherence and rotation with reflector


convergence to map syntectonic deposition

Use image logs to correlate curvature to fractures, allowing it to be


used as a fracture proxy

Integrate curvature anomalies and concepts of differential


compaction to estimate lithology

Use volumetric shapes to quantify carbonate mounds, collapse


features, and increased accommodation space

2-4
2-5 (Bureau of Economic Geology, Texas)
Attributes based on volumetric dip and azimuth

Seismic amplitude

Inline dip Crossline dip

Coherence

2-6
Coherence
Amp
Positive

Negative
Coh
1.0

0.6

2-7 (Seismic data courtesy of Devon Energy)


Sign convention for 2D curvature attributes:
Anticline: k > 0
Plane: k=0 d 2z
Syncline: k < 0 1 dx 2
k
R s dz 2 3 / 2
1
dx
k=0
x

k<0 Flat Plane


z

R
Syncline
3D Curvature
and Biometric
k2< 0
k2 > 0
Identification of
k1 < 0 Suspicious
Travelers
k1 > 0

k2 = 0
Circles in perpendicular planes tangent
to a quadratic surface

|kmax|=1/Rmin

|kmin|=1/Rmax

2-10 (Mai et al., 2009)


k1

k2

(Wolfram demonstration project)


Radius of Curvature
3 km

1.0
Time (s)

1.2
Attributes based on volumetric dip and azimuth

Seismic amplitude

Inline dip Crossline dip

Quadratic
surface? Coherence
yes

Strike of Principal
Principal curvatures
curvatures

2-13
Most negative principal curvature, k2
Curv
Positive

Negative

2-14 (Seismic data courtesy of Devon Energy)


Most negative principal curvature, k2,
co-rendered with coherence
Amp
Positive

Negative
Coh
1.0

0.6

Curv Curv Opacity


Positive

Negative
0 1

2-15 (Seismic data courtesy of Devon Energy)


Most positive principal curvature, k1
Curv
Positive

Negative

2-16 (Seismic data courtesy of Devon Energy)


Most positive principal curvature, k1, co-
rendered with coherence
Amp
Positive

Negative
Coh
1.0

0.6

Curv Curv Opacity


Positive

Negative
0 1

2-17 (Seismic data courtesy of Devon Energy)


Both principal curvatures, k1 and k2,
co-rendered with coherence
Amp
Positive

Negative
Coh
1.0

0.6

Curv Curv Opacity


Positive

Negative
0 1

2-18 (Seismic data courtesy of Devon Energy)


Reflector Shape

2-19
Attributes based on volumetric dip and azimuth

Seismic amplitude

Inline dip Crossline dip

Quadratic
surface? Coherence
yes

Strike of Principal
Principal curvatures
curvatures

Shape components

2-20
s=-1.0
The shape index, s: bowl

2 k 2 k1
s ATAN ( ) s=-0.5 Valley
k 2 k1

k1 k2 s=0.0 Saddle

Principal curvatures
s=+0.5 Ridge

s=+1.0 Dome

2-21 (Courtesy of Ha Mai)


Shape index modulated by curvedness

s
-1.0 0.0 +1.0
High
C

0.0

2-22 (Seismic data courtesy of Devon Energy)


Shape index modulated by curvedness,
co-rendered with coherence
Amp
Positive

Negative
Coh
1.0

0.6

s
-1.0 0.0 +1.0
High
50 %Transparent
C

100 %Transparent
0.0

2-23 (Seismic data courtesy of Devon Energy)


Filters to enhance structural shape components

Bowl Valley Saddle Ridge Dome


1.0
Filter Amplitude

0.5

0.0
-1.0 -0.5 0.0 0.5 1.0
Shape index, s

2-24
Bowl component
co-rendered with coherence
Amp
Positive

Negative
Coh
1.0

0.6

Bowl Component
Positive

0
0 1
Opacity

2-25 (Seismic data courtesy of Devon Energy)


Correlation of bowl shape component
with collapse features

Bowl and
Coherence
coherence

2-26 (data courtesy of Devon Energy)


Correlation of bowl shape component
with collapse features

Bowl and coherence

2-27 (data courtesy of Devon Energy)


Structural Lineaments

2-28
Attributes based on volumetric dip and azimuth

Seismic amplitude

Inline dip Crossline dip

Quadratic
surface? Coherence
yes

Strike of Principal
Principal curvatures
curvatures

Shape components Lineament volumes

2-29
Orientation of lineaments

Fractures are often stronger near the fold axis (sometimes parallel, often
at an angle associated with Mohrs circle), and hence to the strike of the
curvature anomalies

2-30 (Chopra and Marfurt, 2011)


Strike, k2
N -90 0 +90
W E strong

Most-negative principal
S

curvature, k2
N
N W E

N
weak
W E

2-31
Strike modulated by most-negative
principal curvature

k2
k2

2-32 (Seismic data courtesy of Devon Energy)


Strike modulated by most-negative principal
curvature, co-rendered with coherence
Amp
Positive

Negative
Coh
1.0

0.6

k2 vs. k2 Opacity
255

0 k2
0 1
k2

2-33 (Seismic data courtesy of Devon Energy)


Time slices through strike modulated by
most-negative principal curvature
Amp
Positive

Negative
Coh
1.0

0.6

k2 vs. k2 Opacity
255

0 k2
0 1
k2

Transparent

2-34 (Seismic data courtesy of Devon Energy)


Box probe through strike modulated by
most-negative principal curvature
Amp
Positive

Negative
Coh
1.0

0.6

k2 vs. k2 Opacity
255

0 k2
0 1
k2

Transparent

2-35 (Seismic data courtesy of Devon Energy)


Structural lineaments displayed as roses
Amp
Positive

Negative
Coh
1.0

0.6

2-36 (Seismic data courtesy of Devon Energy)


2-37 (Bureau of Economic Geology, Texas)
Reflector Convergence

2-38
Attributes based on volumetric dip and azimuth

Seismic amplitude

Inline dip Crossline dip

Quadratic
surface? Coherence
yes no

Strike of Principal Reflector


Principal curvatures
curvatures convergence

Shape components Lineament volumes

2-39
Reflector convergence co-rendered with coherence
t=1.1 s

0.6 A A

0.8
Amp
Positive
1.0
Time (s)

0
Negative 1.2
Coh
1.0
1.4

0.6 1.6
N

W E 1.8
2-40 S
Reflector convergence co-rendered with coherence
t=1.1 s 0.6 B B

0.8

1.0

Time (s)
1.2

1.4

1.6

1.8

Amp
Positive
0
Negative
Coh
1.0

0.6
N

W E

2-41 S
Reflector convergence co-rendered with coherence
t=1.1 s 0.6 C C

0.8

1.0

Time (s)
1.2

1.4

1.6

1.8

Amp
Positive
0
Negative
Coh
1.0

0.6
N

W E

2-42 S
Reflector convergence co-rendered with coherence
0.6 A A
t=1.0 s
0.8
1.0

Time (s)
1.2
1.4
1.6
1.8

Amp
Positive
0
0.6 B B
Negative
Coh 0.8
1.0
1.0
Time (s)

1.2
0.6
N 1.4

W E
1.6
1.8
2-43 S
Reflector convergence co-rendered with coherence
0.6 A A
t=1.1 s
0.8
1.0

Time (s)
1.2
1.4
1.6
1.8

Amp
Positive
0
0.6 B B
Negative
Coh 0.8
1.0
1.0
Time (s)

1.2
0.6
N 1.4

W E
1.6
1.8
2-44 S
Reflector convergence co-rendered with coherence
0.6 A A
t=1.2 s
0.8
1.0

Time (s)
1.2
1.4
1.6
1.8

Amp
Positive
0
0.6 B B
Negative
Coh 0.8
1.0
1.0
Time (s)

1.2
0.6
N 1.4

W E
1.6
1.8
2-45 S
Reflector convergence co-rendered with coherence
0.6 A A
t=1.3 s
0.8
1.0

Time (s)
1.2
1.4
1.6
1.8

Amp
Positive
0
0.6 B B
Negative
Coh 0.8
1.0
1.0
Time (s)

1.2
0.6
N 1.4

W E
1.6
1.8
2-46 S
Attributes based on volumetric dip and azimuth

Seismic amplitude

Inline dip Crossline dip

Quadratic
surface? Coherence
yes no

Strike of Principal Reflector Reflector rotation


Principal curvatures
curvatures convergence about normal

Shape components Lineament volumes

2-47
The simplest model
positive rotation (down to the right) across the fault

2-48
The simplest model
negative rotation (up to the right) across the fault

2-49
The next simplest model
A rotating down-dropped graben

2-50
Reflector rotation co-rendered with coherence
t=1.1 s
Amp
Positive

Negative
Coh
1.0

0.6

Rot Rot Opacity


Positive

Negative
0 1

2-51
Attributes based on volumetric dip and azimuth

Seismic amplitude

Inline dip, p Crossline dip, q

T p Coherence

T q

Maximum aberrancy Minimum aberrancy


mag and azimuth mag and azimuth

2-52
Aberrancy

Coherence Aberrancy

Can we accurately map flexures that correspond to sub-seismic resolution faults?

(Courtesy Xuan Qi, OU)


Internal Steps

Inline dip, p crossline dip, q

2 p 2 p 2 p 2q 2q 2q
2 2
x xy xz x xy xz
2 p 2 p 2 p 2q 2q 2q
T p T q
xy y 2 yz xy y 2 yz
2 p 2 p 2 p 2q 2q 2q

xz yz z 2 xz yz z 2

Rotate to prime system (p=0,


q=0)

Find extrema of third derivatives as


function of azimuth,
' 'T p '( )

Rotate back to original system

Azimuth and Azimuth and


magnitude, amax magnitude, amin
(Courtesy Xuan Qi, OU)
Time-structure

Top Marble Falls (Courtesy Xuan Qi, OU)


Vertical slices

Aberrancy

Coherence

(Courtesy Xuan Qi, OU)


Vertical slices

Aberrancy

Coherence

(Courtesy Xuan Qi, OU)


Vertical slices

Aberrancy

Coherence

(Courtesy Xuan Qi, OU)


Coherence

Top Marble Falls (Courtesy Xuan Qi, OU)


Co-rendered curvature and coherence

Top Marble Falls (Courtesy Xuan Qi, OU)


Aberrancy amax and amin

Top Marble Falls (Courtesy Xuan Qi, OU)


Co-rendered coherence and aberrancy amax and amin

Top Marble Falls (Courtesy Xuan Qi, OU)


Figuring out the meaning of the cubic roots

Three roots scenario One root scenario

(Courtesy Xuan Qi, OU)


Amplitude Gradients
Attributes based on volumetric amplitude, envelope, or
energy gradients

Seismic amplitude

Seismic envelope

Normalized inline Normalized crossline


envelope gradient envelope gradient

2-65
Artifacts on a simple NS gradient

(Barnes, 2011)
Artifacts suppressed using a median filter

(Barnes, 2011)
Energy-weighted coherent amplitude gradients
1. Calculate the wavelet that best represents the vertical variation of the
data within the analysis window.
Analysis
window

(Marfurt, 2006)
Attributes based on volumetric dip and azimuth

Seismic amplitude

Inline dip Crossline dip

Structure-oriented
filtered data

Inline energy Crossline energy


gradient gradient

2-69
2. Fit the coherent wavelet to the data within the analysis window. This
represents the coherent component of the seismic data.

Lateral gradient of coherent amplitude


Coherent amplitude, uc
Analysis
window

Coherent energy of the J=5 traces:


J K
ec (t ) c (t kt, jx)
u 2

j 1 k K (Marfurt, 2006)
Energy-weighted coherent amplitude gradients
1v1/x
1v1/y
v1/x

9 Components of eigenvector, v1, describe the amplitude variation of the


coherent wavelet along a local surface, v(x,y)

(Marfurt, 2006)
Time-structure map of horizon A - 120 ft
(a phantom horizon)

(Sarkar et al. 2009)


Phantom horizon slice through the
coherent energy

(Sarkar et al. 2009)


Phantom horizon slice through the
inline coherent energy gradient

(Sarkar et al. 2009)


Phantom horizon slice through
co-rendered attributes

(Sarkar et al. 2009)


Central Basin Platform, Texas, USA
Devonian Thirtyone Horizon

Grad
Positive

Negative

coherence EW coherent energy


gradient

(Blumentritt et al.,1998)
Midcontinent, USA
Time slices

2 km

Grad
Positive

Negative

coherence EW coherent NS coherent energy


energy gradient gradient
Energy-weighted coherent amplitude gradients
So. Marsh Island, Gulf of Mexico, USA
Grad
Positive

Negative

(Data courtesy of Fairfield)


Amplitude curvature

Amplitude, u Envelope, e

(Oliveros and Radovich, 1997)


Lateral variation in amplitude

Amplitude profile
u

1st derivative
du/dx

profile
d2u/dx2

2nd derivative
profile

x
Structural curvature versus Amplitude curvature

Inline (NS) dip Crossline (EW) dip Inline (NS) energy Crossline (EW) energy
gradient gradient

1 km

Coherence

Principal structural Principal structural Amplitude positive Amplitude negative


positive curvature (LW) negative curvature (LW) curvature (LW) curvature (LW)
(Chopra and Marfurt, 2011; Data courtesy: Fairborne Energy Ltd., Calgary)
Horizon slice from the seismic amplitude volume
Most-positive curvature of the amplitude
(short wavelength)
Most-positive curvature of the amplitude
(long wavelength)
Deformation of highly competent rocks (Edwards Group)

(Ferrill and Morris, 2008)


Deformation of mixed competency rocks (Glen Rose fm)

negative
negative
positive

(Ferrill and Morris, 2008)


Deformation of less competent rocks (e.g. Eagleford fm)

negative

positive

(Ferrill and Morris, 2008)


Example 1: Co-rendering multiple attributes

Alberta, Canada

2-88
High

2 km

Low

Horizon slice
through the
coherence
volume

(Chopra and
2-89 Marfurt, 2010)
Pos

2 km 0

Neg

Horizon slice
through the most-
positive curvature
volume

(Chopra and
2-90 Marfurt, 2010)
Pos

2 km 0

Neg

Horizon slice
through the most-
negative
curvature volume

(Chopra and
2-91 Marfurt, 2010)
High Pos

2 km 0

Low Neg

Color stack of
coherence, most-
horst
graben
positive
curvature, and

graben
most negative
curvature
Half graben

Half graben

Half graben
(Chopra and
2-92 Marfurt, 2010)
Example 2: Incised canyons and channels
on a carbonate shelf

NW Australia

2-93
Most negative curvature

Amp
0.0 Pos

Neg
Time (s)

1.0 k2
0.02

Opacity
0

-0.02

2.0
5000 m

2-94 (Wallet, 2014)


Geobody extraction workflow

2-95 (Wallet, 2014)


Reflector convergence

Az
0.0 +180

-180
Mag
Max
Time (s)

Opacit
1.0

y
0
Amp
Pos

Opacity
0

2.0
5000 m Neg

96
2-96 (Wallet, 2014)
Reflector convergence

2-97 (Wallet, 2014)


Reflector convergence

2-98 (Wallet, 2014)


Reflector convergence

2-99 (Wallet, 2014)


Example 3: Rotation of fault blocks and
lateral variation of accommodation space

Alberta, Canada

2-100
Reflector convergence co-rendered with coherence
t = 1.330 s

Amp
Positive

0
Negative
Coh
1.0

convergence
N
0.6

W E

2-101 (Chopra and Marfurt, 2011)


Reflector convergence co-rendered with coherence
t = 1.500 s

Amp
Positive

0
Negative
Coh
1.0

convergence
N
0.6

W E

2-102 (Chopra and Marfurt, 2011)


Reflector convergence co-rendered with coherence
t = 1.550 s

Amp
Positive

0
Negative
Coh
1.0

convergence
N
0.6

W E

2-103 (Chopra and Marfurt, 2011)


Reflector convergence co-rendered with coherence
t = 1.710 s

Amp
Positive

0
Negative
Coh
1.0

convergence
N
0.6

W E

2-104 (Chopra and Marfurt, 2011)


Coh
Reflector vector convergence Amp N

Positive 1.0
Azimuth of convergence, ,
0 W
modulated by its strength, c. E
Negative
(Alberta, Canada) 0.6

2-105
Vector dip
co-rendered
with
coherence

Amp
Positive
0
Negative t=1.610 s
Coh
1.0

0.6
N

W E

2-106S
Reflector
rotation about
the average
normal
co-rendered
Rot
with
Positive coherence
0

Negative

Amp
Positive
0
Negative t=1.610 s
Coh
1.0

0.6

2-107
Reflector
convergence
co-rendered
with
coherence

Amp
Positive
0
Negative t=1.610 s
Coh
1.0

0.6
N

W E

2-108S
Coh
Reflector rotation about the average normal Amp Rot
Positive
Positive 1.0

0 0

Negative Negative
0.6

(Chopra and
Marfurt, 2012)
2-109
Coh
Reflector rotation about the average normal Amp Rot
Positive
Positive 1.0

0 0

Negative Negative
0.6

(Chopra and
Marfurt, 2012)
2-110
Example 4: Pinnacle reefs

Horseshoe Atoll, TX

2-111
Ridge and dome shapes volumetrically rendered with a coherence slice
(Diamond M, Horseshoe Atoll, west Texas)

2-112 (Seismic data courtesy of Parallel Petroleum LLC)


Ridge and dome shapes volumetrically rendered with a coherence slice

2-113 (Seismic data courtesy of Parallel Petroleum LLC)


Ridge and dome shapes volumetrically rendered using a box probe.

2-114 (Seismic data courtesy of Parallel Petroleum LLC)


Picking a geobody

2-115 (Seismic data courtesy of Parallel Petroleum LLC)


Picking multiple geobodies

2-116 (Seismic data courtesy of Parallel Petroleum LLC)


Example 5: Differential Compaction

Alberta and Saskatchewan, Canada

2-117
Incised channels - Alberta, Canada
Coherence

Coh Curv Curv Opacity


1.0 Positive

0.6 Negative (Chopra and Marfurt, 2010;


2-118 0 1 Seismic data courtesy of Arcis)
Incised channels - Alberta, Canada

Coh Curv Curv Opacity


1.0 Positive

0.6 Negative (Chopra and Marfurt, 2010;


2-119 0 1 Seismic data courtesy of Arcis)
Differential compaction over a channel - Canada

Most Positive Curvature


Coherence
Coh Curv Curv Opacity
1.0 Positive

0.6 Negative (Chopra and Marfurt, 2010;


2-120 0 1 Seismic data courtesy of Arcis)
Two independent attributes illuminate the same channel

Coh Curv Curv Opacity


1.0 Positive

0.6 Negative (Chopra and Marfurt, 2010;


2-121 0 1 Seismic data courtesy of Arcis)
Differential compaction over Winnipegosis reefs

Neg Pos
Curvature

(Chopra and Marfurt, 2010;


2-122 Seismic data courtesy of Fairborne)
Differential compaction over Alberta Channels
Coherence

Coh
1.0

0.6

(Chopra and Marfurt, 2015)


Coherence

Most-negative
curvature

Most-positive
curvature

Coh
1.0

0.6

Curv Curv Opacity


Positive

Negative
0 1

(Chopra and Marfurt, 2015)


Geometric Attributes
In Summary:
Co-rendered coherence, reflector rotation, and curvature volumes provide
enhanced images of faults, conjugate faults, wrench faults and ramps

Co-rendering coherence and reflector vector convergence provides


quantitative images of syntectonic deposition and unconformities.

Curvature is not a measure of fractures but of strain. Fractures can be


inferred from curvature through a deformation model.

If image logs correlate curvature to fractures, curvature can be used as a


fracture proxy

Co-rendered coherence and curvature anomalies can be coupled with


concepts of differential compaction to estimate lithology

The volumetric shape of carbonate mounds, collapse features, and bars


can be combined with attributes such as impedance to form crossplot
geobodies or input to a geostatistics workflow
1-125
Computing the dip and azimuth of faults, axial
planes, and unconformities

The 2nd order moment tensor, C, allows us to compute


the moments of inertia of an asteroid.

x x
im jm m Asteroid
v3

Cij m1 M image


m 1
m

v1
(Sebago.mit.com)
v2

(Machado et al., 2015)


The Laplacian of a Gaussian filter
Directional smoothing and sharpening

x3

2 2
12 22 23
= + 4x3 exp + +
23 3 1 2 3

(Machado et al., 2015)


Dip and azimuth of fault planes, Great South Basin, NZ

(Machado et al., 2015)

Potrebbero piacerti anche