M49: Seismic Attributes for
Exploration and Reservoir
Characterization
John Castagna
University of Houston
jpcastagna@uh.edu
Seismic Attributes
• Numbers extracted from seismic traces
used to quantify rock and fluid properties
and to visualize geological relations. A
useful attribute is a SUBSET of the data.
Sheriff Dictionary
• Seismic attributes allow one to view 3D
seismic data from a variety of
perspectives.
• They allow one to obtain improved
geological visualization and determination
of rock/fluid properties
Coherence
Various AttributesCube TM Attributes
from Core Labs School
TIME COHERENCE AZIMUTH DIP INSTANTANEOUS AMPLITUDE
RESPONSE AMPLITUDE INSTANTANEOUS PHASE RESPONSE PHASE INSTANTANEOUS BANDWIDTH RESPONSE BANDWIDTH
INSTANTANEOUS FREQUENCY RESPONSE FREQUENCY 00 PHASE COMPONENT 900 PHASE COMPONENT CARRIER
Course Outline
• Introduction • Part II: DHI’s
– Post-stack Indicators
• Part I: Seismic Attributes and
Reservoir Characterization – AVO
• Theory
– Introduction
• Interpretation
– Fundamentals
• Problems and Pitfalls
– 3D Seismic Attributes
• Applications
– Seismic Inversion • Part III: Spectral Analysis
– Reservoir – Spectral Decomposition
Characterization – Spectral Inversion
– Time-Lapse Seismic • Part IV: Advanced Topics
Reservoir Geophysics: The use of
surface seismic and borehole geophysical
data if available to quantitatively determine
reservoir lithology, porosity, pore fluid
content, lateral extent and continuity,
volumetrics, and internal architecture.
This process may be part of exploration or
development operations.
Reservoir geophysics differs from exploration
geophysics, primarily due to the assumptions of
increased well control and a well-designed,
detailed geophysical survey.
A good question is: “Will the geophysical technique
being proposed be able to differentiate between the
competing reservoir models sufficiently well to be
worth the effort and cost?” (Pennington, 2001)
The answer lies in the rock physics of the reservoir
rock and neighboring formations, not just in the
geophysical model. (Pennington, 1997)
Textbooks
• Sheriff (1987): Reservoir Geophysics
• Brown: Interpretation of 3D Seismic Data
• Chopra and Marfurt
• Yilmaz: Seismic Data Analysis
– Volume II, last chapter
•
Definition
Attribute - (Latin attributus - to associate, from
ad- to + tribuere assign (originally) to divide
among the tribes < tribus tribe) A quality
associated with a person or thing; characteristic:
Patience is the attribute of a good teacher.
(from Marfurt course notes) World Book Dictionary (1989)
Archaic View
Calvin and Hobbs
Walter Johnson
AAPG Explorer
Devonian Pinnacle Reef
Core Labs
Core Labs
A series of Coherence Slices through the Reef
red = low, blue = high coherence
More from Sheriff….
Question
What are the common components of ALL seismic
attributes?
All seismic attributes are based on the arrival time,
amplitude, and spacing of seismic reflection
coefficients as convolved with the seismic
wavelet.
All seismic attributes are thus related to seismic
velocities, contrasts in seismic impedances
and/or intrinsic attenuation.
Classification of Attributes
• Arrival time related
– structure, lateral variability and velocity
• Amplitude related – rock properties
• Signature (waveform shape) related
– layering
– instrinsic attenuation
• Combination of the above
Signature
Signature
Are these signatures the same?
More from Sheriff….
Question
• Do attributes add information?
Question
• Do attributes add information?
NO!!!!
At best, they help you organize, visualize, and
analyze your data.
At worst, they can be very misleading – especially
when correlated to minimal well control.
Why use attributes?
• Attributes allow you to view your prospect
or reservoir through rose (or other colored)
glasses.
• Allow you to look at the same feature from
many perspectives.
• Different attributes direct your attention to
different features of the data
• Aid in quantitative parameter estimation.
Crossplotting (scattergram)
Property
Attribute
Which is the independent variable and which the dependent one?
Crossplotting (scattergram)
Property
Attribute
What happens if the sample size is too small?
What steps can be taken to guard against this?
Crossplotting (scattergram)
Property
Attribute
No correlation
Crossplotting (scattergram)
Property
Attribute
No correlation
Crossplotting (scattergram)
Property
Attribute
Less likely but possible – especially if outliers are discarded
More from Sheriff….
It is crucial to ask the right questions!
It’s not about the math.
It’s about asking the right
questions.
Geology is the driver.
Asking the right question…
For a simple thin layer:
• How does peak amplitude vary with
thickness?
• How does amplitude at peak frequency
vary with thickness?
BlueSkies
Raw Data
Initial Interpretation
Final Results
Raw Data
Boucher, 2009
Boucher, 2009
Where is the pay?
Paradigms
• Mental constructs that provide a context for data
recognition.
Paradigms are EXTREMELY powerful filters. Once you
have one you cannot escape being subjectively led down
that path. (If you are expecting to see channels you WILL
see channels).
Quantitative attribute analysis can potentially restore
objectivity to interpretation – but this is generally not
realized. (ie; what quantitative attributes must a feature
have to be considered a channel?).
Amplitude
extraction
base
Integrated interpretation
top (Rijks and Jauffred, 1991)
Namibia
Kidd, 1999
Deep-water Contourites
Kidd, 1999
Question
• What NUMBERS can be used to
distinguish dunes from deep-water
sediments?
Lena
Lena
Alistair Brown
Landmark
Are attributes related to data quality?
Seismic Facies
• A distinctive set of characteristics for a “window”
of seismic data that is presumably related in
some manner to the depositional system.
• We are told by geologists that “chaotic” seismic
facies are an indication of “high energy”
depositional environments.
• How would you explain a “chaotic” seismic
facies occuring on 1983 vintage data that has
“weak parallel continuous reflections” on 2008
vintage data?
• What attribute would distinguish these two types
of seismic characteristics?
Without a geological context what is
the meaning of these amplitude
variations?
Can physics give us a unique answer?
Without a geological context what is
the meaning of these amplitude
variations?
Geology is the driver!
Quantitative Seismic Analysis
• Old use of seismic: Structural mapping
and stratigraphic patterns – qualitative
• State-of-the art use: Seismic amplitudes
measured and quantitatively translated
into rock characteristics
How will the amplitudes of these sands vary?
Reflections are not isolated interfaces. Vertical shape affects seismic amplitudes.
If vertical shape varies laterally it is difficult to establish a good correlation between
seismic amplitude and rock properties. Multiattribute analysis may help account for such
variation.
Questions
• What is the relationship between seismic
impedance and porosity?
• What is the relationship between contrasts
in seismic impedance and porosity?
• What is the relationship between seismic
amplitude and porosity?
Seismic Porosity Cube
Neural Network Resistivity Inversion of Seismic Section
WELLS WITH CONFIRMED PRODUCTION FROM THIS SAND
WELLS WITH POSSIBLE PRODUCTION FROM THIS SAND
P*G for S11with 40ms window
Radon and no AGC
Flat Spots
GOC
OWC
GOC
OWC
Offshore Nigeria
Alister Brown, 5th Edition
Spectral Decomposition
Fahmy et al, 2008
Vp/Vs Ratio Can Be Correlated To Effective Stress,
Even At Very Low Effective Stresses
10
9
8
7
Vp/Vs
6
pre Vp/VS
5
8% clay Vp/Vs
4
3
2
1
0
0 500 1000 1500 2000 2500 3000
Effective Pressure (psi)
Figure from Huffman & Castagna (2001)
Vp/Vs Section for Shallow Water Flow
Figure from Moreno et al. (2003)
Permian Basin Spectral Inversion
Input Data -90 Spectrally Inverted Data -90
Note the small “high” and channel not seen in input data
Original Seismic Time Slice Spectrally Inverted RC’s
Spectrally Inverted Impedance
Note how spectrally
inverted data
emphasize subtle
structural features on
a time slice
o
Original PSTM -90 rotation
Gas
Fractured Carbonate Reservoir
o
Spectral Inversion with -90 phase shift
(pseudo-impedance)
Gas Production
Minor Faulting? Unproductive
o
Spectral Inversion with -90 phase shift
(pseudo-impedance)
Gas Production
Minor Faulting? Unproductive
Attribute Workflow
Definition
Computation
Evaluation
Combination
Integration
Alternate Attribute Workflow
Definition
Evaluation
Computation
Combination
Integration
Definition
• Define the problem
– Geological visualization
– Anomaly hunting
– Quantitative analysis
Definition
Geological Visualization
• Resolution of the attributes should match
the feature to be visualized
• Multitrace attributes (dip, coherency)
• Instantaneous Phase
• Spectral Decomposition
Definition
Anomaly Hunting
• Amplitudes and related attributes
• Inversion-based attributes
• AVO
• Spectral Decomposition
Definition
Quantitative Analysis
• Reservoir Characterization
• All attributes potentially useful
• AVO
• Spectral Decomposition
Computation
• Compute appropriate attributes
– various attribute engines
– computation time not a critical factor
– consider multi-trace attributes, inversion,
AVO, spectral decomposition
– too many attributes better than too few
Evaluation
• Evaluate attributes
– Eliminate redundant and unstable attributes
– Consider physical significance of attributes
– Scan attribute maps and cross-sections; is
geology revealed?
– Crossplot attributes, look at correlation tables
and tests of significance
Combination
• Combine to form mega-attributes
– Combine attributes (principle component or
other analysis)
– Consider external attributes
– Perform multiattribute analysis
• Step-wise regression
• Multiple linear regression
• Neural networks
Integration
• Integrate with well information
– Validate results
• Compare to validation wells
• Scan predicted images
– Perform geostatistical simulation
F-test
K = number of parameters
n = number of data points
R = correlation coefficient
mean squared model/K
F=
mean squared residual/n
F-test
2
[R /K]
F= 2
[(1-R )/(n-K-1)]
K= number of parameters
n= number of data points
R= correlation coefficient
Note: R ↑ = F ↑ ; n ↑ = F ↑ ; K ↑ = F ↓
Assumptions
• Each well has one average property for one
reservoir interval that corresponds to one
seismic reflector.
• Each well is a data point
• Each attribute is a parameter
Problem
• I want to estimate average porosity from
average impedance. I have two calibration
wells. What is the correlation coefficient?
What is the significance of the correlation?
• What is the significance if I have 3
calibration wells and a correlation
coefficient of .7? .4?
• What if I have 6 calibration wells?
We want F to be greater than 1.
Can we be confident in a correlation if F is less than 1?
What is F if R=1 and n > K+1 ?
• Given K parameters and R, how many wells
are needed to achieve F=1?
n=1+ K(1+(1-R2)/(R2))
If R=1, how many wells are needed?
If R=0, how many wells are needed?
Wells needed
F=1
for F=1
40 R
0.2
30
Wells Needed
0.3
0.4
20
0.5
0.7
10
0.9
0
1 3 5 7 9
Number of Attributes
Advice
• Think hard about the questions you are
asking
• Evaluate significance of correlations
• Consider physical basis for correlations
Attribute Workflow
• Define the problem
– Geological visualization
– Anomaly hunting
– Quantitative analysis
• Compute appropriate attributes
• Evaluate attributes
– Eliminate redundant and unstable attributes
– Consider physical significance of attributes
– Scan attribute maps and cross-sections
– Crossplot attributes, look at correlation tables and tests of significance
• Compute mega-attributes
– Combine attributes (principle component or other analysis)
– Consider external attributes
– Perform multiattribute analysis
• Step-wise regression
• Multiple linear regression
• Neural networks
– Validate results
• Compare to validation wells
• Scan images
• Integrate with well information
– Perform geostatistical simulation
Review Questions
1. How does one determine if a geophysical technique will be able to differentiate
between competing reservoir models? (a) studying the rock physics of the relevant
rock formations (b) perform seismic modeling (c) estimate seismic signal-to-noise
ratio (d) all of the above (e) none of the above
2. T or F___ A seismic attribute is any measurable characteristic of seismic data.
3. Seismic attributes may be classified as being related to (a) amplitude (b) arrival time
(c) signature (d) all of the above (e) none of the above
4. T or F___ Seismic attributes are useful because they add new independent
information to the analysis.
5. Seismic attributes (a) allow you to look at the same feature from many perspectives
(b) direct your attention to different features (c) aid in quantitative parameter
estimation (d) all of the above (e) none of the above
6. T or F___ A common problem in attribute analysis is false positive correlations
resulting from too few calibration data points.
7. How does one draw valid inferences from too few data points for statistical
significance? (a) perform non-linear transformation (b) rely on an underlying physical
model (c) use neural networks (d) none of the above
8. Paradigms are (a) anomalous information (b) mental constructs that provide a
context for data recognition (c) a collection of data (d) all of the above (e) none of the
above
