Vous êtes sur la page 1sur 12

AVO Attribute Stacks2175

ProMAX Reference

AVO Attribute Stacks


AVO Attribute Stacks analyzes the offset-dependence of
amplitudes within a prestack CDP gather.

Theory
With proper data preconditioning, this analysis can give
indications of lithology and/or pore content. Reconnaissance
analysis can be done with AVO Attribute Stacks using leastsquares linear fits of the amplitudes with offset or with AVO
Weighted Stacks using weighted stacking of amplitudes to
determine elastic parameters. More detailed investigations
should include angle limited CDP gathers and angle gathers.
See helpfile for AVO Analysis Gathers. AVO analysis is
successfully used to discriminate between the presence of
gas and lithologic variations in Bright Spot analysis.
The theory behind AVO analysis finds its roots in the
Zoeppritz equations. These equations are the plane wave
approximations that describe reflection coefficients for
varying angles of incidence. Qualitatively, seismic wave mode
conversion is responsible for the dependence of amplitude
anomalies in seismic data with respect to wavefront incidence
angle or offset. This mode conversion is most evident with a
large change in Poissons ratio, and this occurs when gas
saturated rock meets water saturated rock.
The rock properties associated with gas saturated formations
and hydrocarbon presence are manifested in seismic
amplitudes. An increase in the absolute reflection coefficient
and the resulting reflection amplitude with offset can be
indicative of hydrocarbon presence.
This implementation of AVO attribute analysis is done across
isotime sample values or within gated intervals defined by
stack polarity. Input gathers are assumed to be NMO
corrected to align primary events. Multiple energy and other
forms of coherent noise, as well as random noise, will
degrade the AVO analysis. The problems caused by noise can
be reduced by the input of offset-mixed supergathers as well
as by preconditioning with an offset-consistent amplitude
correction.

Other Docs

Search Page

Known Problems

AVO Attribute Stacks2176

ProMAX Reference

A common problem with AVO attribute stacks is residual


NMO on the CDP gathers resulting from imperfect velocity
specification. In order to compensate for this problem,
amplitudes may be selected within polarity gates, relieving
the requirement for exact NMO application. Polarity gates are
determined by a partial stack of the data in the CDP gather.
Contiguous stack amplitudes with a common polarity define
a polarity gate. The data within each polarity gate is
condensed by the use of either the maximum or average
amplitude on each trace within the gate. A robust least
squares line is fitted to the amplitudes of each trace, and
various attributes are computed from the fit to the data. The
output attributes are blocked as constant values over each
polarity gate.
Stack
Trace 400

Input Gather Offset

4200

Polarity
Gate

Usage
Input gathers must have primary sort order CDP and
secondary sort key AOFFSET. Angle-of-incidence gathers
from AVO Analysis Gathers (secondary sort key ANG_VALU)
can be input as well, but the output cannot be expected to be
the same as when inputting offset gathers. NMO must be
applied to the input data. For screen displays, trace scaling
should be set to Scale to Entire Screen.

Other Docs

Search Page

Known Problems

AVO Attribute Stacks2177

ProMAX Reference

This process outputs an ensemble of nine traces for each


input gather. The traces are flagged for subsequent sorting by
the AVO_ATTR trace header word:

AVO_ATTR

Description

Intercept

Gradient

Intercept * Gradient

Intercept * Gradient *
Correlation Coefficient

Angle of Crossover

Sign(Intercept) * Gradient

Correlation Coefficient

Runs Statistic

Fluid Factor

These intercept and gradient attributes characterize a linear


fit to amplitude versus sine-squared of the angle of incidence.
See figure below. This line is the two term Shuey
approximation to the Zoeppritz equations.
The angles of incidence are calculated using a onedimensional representation of the input velocity field at each
CDP. Either of two methods may be selected: Linear velocity
gradient or Interval/RMS velocity ratio. The linear velocity
gradient method is exact if there is a linear gradient, but may
deviate from the correct value if this is not the case. The
second method is valid if the two-term NMO equation holds
and does not have the restriction of a linear velocity gradient.

Other Docs

Search Page

Known Problems

AVO Attribute Stacks2178

ProMAX Reference

Linear Fit to AVO Data (6 fold CMP)

Amplitude

*
I

*
*

*
*

*
sin2 (incidence angle)

Amp = I + G sin2 (angle)


I = Intercept

G= Gradient

* = observed data at a specific time


The attributes in the table, other than intercept and gradient,
are computed as they are expected to be useful. AVO
Attribute Computations allows more flexible manipulation
of the primary intercept and gradient attributes, as well as
the correlation coefficient and runs statistic attributes.
The following describes the attribute from the previous table:
Intercept
The first attribute is the intercept of the best fit line with
the zero offset or zero angle axis. This is the zero offset
amplitude directly related to the reflection coefficient
calculated using sonic and density logs. This zero-offset
section may also be useful for trace inversion.
Gradient
The second attribute is the slope of the best fit line and
describes the relative change in amplitude with angle
and offset. Although the resulting section shows the
change in amplitude over the specified offset range, this
section, when viewed alone, does not indicate amplitude
increase or decrease with offset. Use this attribute in
conjunction with other attributes (AVO_ATTR 1, 3, and
6) to quantify the effect.
Intercept * Gradient
The third attribute is a good prospecting tool since it

Other Docs

Search Page

Known Problems

AVO Attribute Stacks2179

ProMAX Reference

highlights events with strong zero offset reflection


coefficients also having appreciable AVO characteristics.
An increase in absolute amplitude with offset is a peak
and a decrease in AVO is a trough on this seismic
section.
Intercept * Gradient * Correlation Coefficient
AVO Attribute Stacks also quantifies the statistical
reliability of the aforementioned attributes. The AVO
attribute Intercept*Gradient* Correlation Coefficient is a
statistical measure of the quality of fit to the amplitude
data. This is a reliability factor for the AVO pick. In other
words, a strong zero offset amplitude with an increase in
absolute amplitude with offset and a high correlation
coefficient will be a strong peak on this stack. Lower
correlation coefficients will weaken an events response.
Angle of Crossover
The fifth attribute corresponds to the angle of incidence
at which a polarity reversal occurs. This attribute is
calculated as long as the crossover angle is determined
to be less than 30 degrees. If the crossover angle is
greater than 30 degrees, a value of 31 degrees is output.
Absolute amplitude decreases with offset until it reaches
the crossover angle, flips polarity, and subsequently
increases. Events with this response typically appear
weak on a stacked section. Nonetheless, this polarity
reversal effect has been known to indicate rock
properties consistent with pay zones.
Sign(Intercept) * Gradient
The sixth AVO attribute is the product of the sign of the
intercept and the gradient. This retains the value of the
gradient but the polarity varies with the combined

Other Docs

Search Page

Known Problems

AVO Attribute Stacks2180

ProMAX Reference

polarities of the gradient and intercept.


Gradient vs. sign (I) * G
AVO
+

Case
1

Sin2

Ampl
-

Case

sign(I) * G = AVO_ATTR#6

AVO

AVO_ATTR#6 = AVO effect and G = AVO in all cases


Correlation Coefficient
The seventh AVO attribute is the Correlation Coefficient.
A display of this attribute may be useful in determining
the relative reliability of different zones of the data. It
may also be used to weight the other attributes, by
multiplication in Trace Math.
The Correlation Coefficient has an absolute value range
of 0.0 to 1.0, 1.0 indicating a perfect fit of the points to
the least squares straight line. Low absolute values of
the correlation coefficient highlight regions where the fit
of the data to the straight line is poor.
Runs Statistic
The eighth AVO attribute is the Runs Statistic, the
statistical estimate of the validity of the least-squares fit.
Large absolute values of the runs statistic, greater than
3, indicate that the data does not trend in a straight line
and therefore a straight line fit to the data is not
appropriate.
Other Docs

Search Page

Known Problems

AVO Attribute Stacks2181

ProMAX Reference

The Runs Statistic roughly indicates the number of


standard deviations the least squares fit deviates from a
straight line with random noise. Data which oscillate
rapidly, such as a high-frequency sinusoid, will give high
runs statistics. Data which oscillate very slowly, such as
a low-frequency sinusoid, will give a large negative runs
statistics. Data which have randomly sequenced points
either side of the fit line will have small absolute runs
statistics. Small runs statistics are therefore more
desirable since this fits our model of data with random
noise addition.
Fluid Factor
The ninth AVO attribute is the Fluid Factor. This is
computed using Castagnas mudrock equation as an
estimate of the P-wave to S-wave velocity ratio. Gardners
relationship is used to determine density. Deviation of
intercept and gradient from a regional trend will show up
as a Fluid Factor anomaly. The Fluid Factor equation is:
Vs dVs
dVp
dF = ------------- 1.16 ------- -----------Vp dVsa
dVpa
Where:
dVp
------------- = 1.6 Intercept
dVpa
and

dVs
------------ = Intercept Gradient
dVsa

The mudrock trend used to compute the background


shear wave velocity for the Vs/Vp ratio is:

Vs = 0.8621 Vp 1.1724

Where the units of velocity are km/s.

Other Docs

Search Page

Known Problems

AVO Attribute Stacks2182

ProMAX Reference

Practically, we know that shales and wet sands tend to


lie along the mudrock trend. Gas sands will have a larger
perpendicular offset from the mudrock trend, thus tend
to appear as anomalies in this attribute.
The gradient and intercept are calculated using a robust least
squares algorithm which reduces the influence of anomalous
points in the calculation. Two statistical quantities
determined from the least squares analysis can also be used
to distinguish between regions where the amplitude trend of
the data is linear and those regions where the trend is nonlinear or random.
You have the option to reject analyses at samples that do not
meet defined thresholds based on the above quantities. This
will produce a set of output attribute traces that only have
analyses at samples where a good statistical fit to the data is
obtained.

References
Castagna, J.P, 1993, AVO analysis - tutorial and review: in Castagna, J. P. and
Backus, M. M., Eds., Offset-dependent reflectivity - Theory and Practice of AVO
analysis: Soc. Expl. Geophys.
Koefoed, O., 1955, On the effect of Poissons ratios of rock strata on the reflection
coefficients of plane waves: Geophysical Prospecting, 3, 381-387.
Ostrander, W.J., 1984, Plane-wave reflection coefficients for gas sands at nonnormal angles of incidence: Geophysics, 49, 1637-1648.
Shuey, R.T.,1985, A Simplification of the Zoeppritz Equations: Geophysics, 50,
609-614.
Smith, G. C., and Gidlow, P.M., 1987, Weighted stacking for rock property estimation and detection of gas: Geophysical Prospecting, 35, 993-1014.
Walden, A.T., 1991, Making AVO sections more robust. Geophysical Prospecting,
39, 915-942.

Parameters
Beginning time for AVO analysis
Enter the minimum time for AVO analysis.

Other Docs

Search Page

Known Problems

AVO Attribute Stacks2183

ProMAX Reference

Ending time for AVO analysis


Enter the ending time for AVO analysis. Enter 0 for the
maximum trace length.
Minimum incidence angle to be used
Enter the minimum incidence angle to use when fitting the
gradient and intercept line. This value is usually 0, but if
near offset amplitudes have been modified, a value of 1 to 5
degrees can be appropriate.
Maximum incidence angle to be used
Enter the maximum incidence angle to use when fitting the
gradient and intercept line. The angle should be set to the
maximum angle of incidence for which a linear fit is valid in
the region where the data were collected. Thirty degrees is a
reasonable value.
Minimum offset to be included in AVO analysis
Enter the minimum offset to include in AVO analysis. Input
gathers can be offset limited to include only absolute offsets
greater than or equal to this value. If the input data are
angle-of-incidence gathers, this parameter does not apply.
Maximum offset to be included in AVO analysis
Enter the maximum offset to include in AVO analysis. Input
gathers can be offset limited to include only absolute offsets
less than or equal to this value. If the input data are angle-ofincidence gathers, this parameter does not apply.
Apply final datum statics?
Select Yes to shift the data to final datum using the header
entry FNL_STAT.
Method of amplitude selection for AVO fit
Select the method of amplitude selection for AVO fit:
Constant sample gives a least-squares fit of amplitudes
at each sample of the traces. Errors in NMO which cause
misalignment of the amplitudes will give erroneous gradients.

Other Docs

Search Page

Known Problems

AVO Attribute Stacks2184

ProMAX Reference

Maximum in gate uses a quadratic spline to interpolate


the maximum amplitude in the polarity gate.
Average in gate uses the average of all amplitudes
within the gate in the least squares linear fit procedure.
The second two options result in the trace being divided into
polarity gates, defined between zero-crossings, with leastsquares fit of amplitudes occurring only once per polarity
gate. These two gated methods are expected to run in one half
to one third the CPU time required for the constant sample
method.
Minimum offset in polarity gate stack
This appears if Maximum in gate or Average in gate to
Method of amplitude selection for AVO fit. Enter the
minimum absolute offset to compute a stack which will
define trace polarities. If the input data are angle-ofincidence gathers, all input angles will be used in forming the
polarity gate stack.
Maximum offset in polarity gate stack
This appears if Maximum in gate or Average in gate to
Method of amplitude selection for AVO fit. Enter the
maximum absolute offset to compute a stack which define
trace polarities. Enter 0, the default, to use all offsets in the
stack. If the input data are angle-of-incidence gathers, all
input angles will be used in forming the polarity gate stack.
Is this a spatially variant velocity calculation?
Select Yes if this is a spatially variant velocity calculation.
Select No if this is not a spatially variant velocity calculation.
If No is selected, only a single velocity function is allowed in
the input velocity table or in the type-in velocity function
parameter. Selecting No allows minor speed-up for spatial
invariance of the velocity field.
Get velocities from the database?
Select Yes if the velocity functions already exist in a velocity
parameter file. Select No to enter the velocity function.

Other Docs

Search Page

Known Problems

AVO Attribute Stacks2185

ProMAX Reference

Select velocity parameter file


This appears if Yes to Get velocities from the database.
Enter a velocity parameter file.
Velocity function
This appears if No to Get velocities from the database.
Time, and possibly spatially, variant velocity field indexed by
CDP.
Method of computing angle of incidence
Select how the angles of incidence for each data sample are
computed. If the input data are angle-of-incidence gathers,
this parameter does not apply. The choices are:
Linear velocity gradient: This method assumes that a
linear velocity gradient exists from the surface to the
current sample time. All interval velocities between the
surface and the current sample time are ignored in the
computation. This method is only suitable for linear gradients of interval velocity.
Interval/RMS ratio: This method is valid for all velocities, offsets and times where the two-term NMO equation
is valid. It uses a ratio of the RMS velocity to the sample
time and the local interval velocity to compute the angle
of incidence. The equation used here is:

Vint
X
sin ( ) = ------------- ---------------------------------------------------
Vrms X 2 + ( Vrms T ) 2
0

Minimum correlation coefficient allowed


Enter a correlation coefficient threshold value. The default
value of 0.0 results in no editing based on the correlation
coefficient. Values greater than 0.5 will result in strong
editing. Correlation coefficients with absolute values less
than the submitted threshold value will result in the output
gradient values at those time samples being zeroed.
Maximum runs test statistic allowed
Enter a runs statistic threshold value. The default value of
100 results in no editing based on the runs statistic. Values
greater than 5 will result in little editing. Values less than 3

Other Docs

Search Page

Known Problems

AVO Attribute Stacks2186

ProMAX Reference

may result in strong editing. Editing with such low runs


statistics may result in valid data being removed due to
expected variations in the distribution of seismic amplitudes.
Runs statistics with absolute values less than the submitted
threshold value will result in the output gradient values at
those time samples being zeroed.
Note: Interval velocities in time are required by this tool.
Previous versions used RMS velocity and converted to
interval velocity internally. In order to ensure that the this
conversion does not introduce artifacts due to rapid
variations in interval velocity, the velocity conversion must
now be done using Velocity Manipulation. It is strongly
recommended that the interval velocities be carefully
examined before attempting AVO analysis.

Other Docs

Search Page

Known Problems