By:
M. Turhan Taner
Introduction
Classification of Attributes
Pre-Stack Attributes
Post-Stack Attributes
Instantaneous Attributes
Wavelet Attributes
Physical Attributes
Geometrical Attributes
Reflective Attributes
Transmissive Attributes
Computational Procedures
Frequency Domain Computation
Discrete Time Domain Computation
Gabor-Morlet Decomposition
Simple Harmonic Motion Method
Wavelet Attributes
Geometrical Attributes
Computation of Geometrical Attributes
Event Continuity
Instantaneous (Phase) Dip
Instantaneous Lateral Continuity
Attributes Computed by Dip Scanning
2
Maximum Lateral Semblance and corresponding Dips as Lateral Continuity Attributes
Dip of Maximum Lateral Semblance
Smoothed dips of maximum lateral semblance
Dip Variance (Local Dip versus average Dip)
Lateral Continuity
Smoothed Maximum Lateral Semblance
Variance of maximum coherency, instantaneous versus average
Hybrid Attributes
Parallel Bedding Indicator
Chaotic zone indicator
Zones of Unconformities
Shale Indicator
FILTER DESIGN
Band Pass Filters
Butterworth Filter
Low Pass Butterworth Filter
High Pass Butterworth Filter
Band Pass Butterworth Filter
Minimum Phase Butterworth Filters
Taner Filter
Low Pass Filter
High Pass Filter
Band Pass Filter
CONVOLUTIONAL HILBERT/BUTTERWORTH FILTER
CONVOLUTIONAL DIFFERENTIATION FILTER
3
Preface:
This is the third edition of the "Attributes Revisited" report. We have added a new classification of the
attributes, as we understand them at the present time. This classification may well change as our understanding
of the use of attributes improves. We give a full description of each attribute and include their projected use in
interpretation. The whole report is put in an html format to be viewed interactively by the user with any Web
browser. There will also be a printed copy available. This report contains all 2-D and 3-D post-stack and 2-D pre-
stack attributes.
Definition:
Seismic Attributes are all the information obtained from seismic data, either by direct measurements or by logical
or experience based reasoning.
Introduction
Based on their definition, the computation and the use of attributes go back to the origins of seismic exploration
methods. The arrival times and dips of seismic events were used in geological structure estimation. Frank Rieber
in the 1940's introduced the Sonograms and directional reception. This method was extensively used in noise
reduction and time migration. The introduction of auto-correlograms and auto-convolograms (Anstey and
Newman) led to better estimates of multiple generation and more accurate use of the later developed
deconvolution. NMO velocity analysis gave better interval velocity estimates and more accurate subsurface
geometries. Bright spot techniques led to gas discoveries, as well as to some failures. This was improved by the
introduction of AVO technology. Each of these developments has helped our understanding of the subsurface
and reduced the uncertainties. Unfortunately, one of the principal failures of any of the individual techniques was
our implicit dependence on it. Finally, the power of the combined use of a number of attributes is being
recognized and successful techniques are being introduced. The attribute discussed in this paper is the outcome
of the work relating to the combined use of several attributes for lithology prediction and reservoir
characterization..
Complex seismic trace attributes were introduced around 1970 as useful displays to help interpret the seismic
data in a qualitative way. Walsh of Marathon published the first article in the 1971 issue of Geophysics under the
title of " Color Sonograms". At the same time Nigel Anstey of Seiscom-Delta had published “Seiscom 1971” and
introduced reflection strength and mean frequency. He also showed color overlays of interval velocity estimates
for lithological differentiation. The new attributes were computed in the manner of radio wave reception. The
reflection strength was the result of a low pass filtered, rectified seismic trace. The color overlays showed more
information than was visible on the black and white seismic sections. Realizing the potential for extracting useful
instantaneous information, Taner, Koehler and Anstey turned their attention to wave propagation and simple
harmonic motion. This led to the recognition of the recorded signal as representing the kinetic portion of the
energy flux. Based on this model, Koehler developed a method to compute the potential component from its
kinetic part. Dr. Neidell suggested the use of the Hilbert transform. Koehler proceeded with the development of
the frequency and time domain Hilbert transform programs, which made possible practical and economical
computation of all of the complex trace attributes. In the mid 70's three principal attributes were pretty well
established. Over the years a number of others were added.
The study and interpretation of seismic attributes give some qualitative information of the geometry and the
physical parameters of the subsurface. It has been noted that the amplitude content of the seismic data is the
4
principal factor for the determination of physical parameters, such as the acoustic impedance, reflection
coefficients, velocities, absorption etc. The phase component is the principal factor in determining the shapes of
the reflectors, their geometrical configurations etc. Our objective is to bring the interpretation of attributes from a
qualitative manner to a more quantitative manner. In this paper we will first discuss the several computational
methods of conventional attributes, basically the computation of the analytic trace. In the second part we will
present computation of the conventional attributes and their derivatives. One point that must be brought out is
that we define all seismically driven parameters as the Seismic Attributes. They can be velocity, amplitude,
frequency, rate of change of any of these with respect to time or space and so on. We will classify the attributes
based on their computational characteristics. They can be computed from pre-stack or post stack data sets.
Some of the attributes computed from the complex trace such as envelope, phase etc. correspond to the various
measurements of the propagating wave front. We will call these the 'Physical Attributes'. Others, computed from
the reflection configuration and continuity, we will call 'Geometrical Attributes'. The principal objectives of the
attributes are to provide accurate and detailed information to the interpreter on structural, stratigraphic and
lithological parameters of the seismic prospect.
Classification of Attributes
This paper is written to provide the background information to the RSI_ATTRIB3D interactive seismic attribute
computation program developed by Seismic Research Corporation, which later became a part of Rock Solid
Images. The project was sponsored by a number of oil companies, initially by the Italian Oil Company ENI-
AGIP. The initial objective was to develop as many physical attributes as possible in order to define the
lithological parameters and reservoir characteristics from different points of view. In the development we
established a general classification of attributes based on their input data and their usage. Attributes can be
computed from pre-stack or from post-stack data before or after time migration. The procedure is the same in all
of these cases. Attributes can be classified in many different ways. Several authors have given their own
classification (please see references). Here we give a classification based on the characteristics of the attributes.
Pre-Stack Attributes
Input data are CDP or image gather traces. They will have directional (azimuth) and offset related information.
Computations generate huge amounts of data; hence they are not practical for initial studies.
Post-Stack Attributes
Stacking is an averaging process, losing offset and azimuth related information. Input data could be CDP
stacked or migrated. One should note that time migrated data will maintain their time relations, hence temporal
variables, such as frequency, will retain their physical dimensions. For depth migrated sections, frequency is
replaced by wave number, which is a function of propagation velocity and frequency. Post-stack attributes are
better for observing large amounts of data in initial investigations. For detailed studies, pre-stack attributes may
be incorporated.
Class I attributes are computed directly from traces. This data could be pre- or post-stack, 2-D or 3-D, before or
after time migration. Trace envelope and its derivatives, instantaneous phase and its derivatives, bandwidth, Q,
dips etc. are some of the attributes computed this way.
Class II attributes are computed from the traces with improved S/N ratios after lateral scanning and semblance-
weighted summation. Details of the computation are given in the Maximum Semblance Computation section of
the Geometrical attributes. All of the Class I attributes are computed in Class II. In addition lateral continuity and
dips of maximum semblance are computed from the scanning procedure.
Based on the information content, attributes are divided into two general categories:
5
Instantaneous Attributes
Instantaneous attributes computed sample by sample, representing instantaneous variations of various
parameters. Instantaneous values of attributes such as trace envelope, its derivatives, frequency and phase may
be determined from complex traces. Both Class I and Class II attributes are computed.
Wavelet Attributes
Instantaneous attributes computed at the peak of the trace envelope have a direct relation to the Fourier
transform of the wavelet in the vicinity of the envelope peak. For example, Instantaneous frequency at the peak
of the envelope is equal to the mean frequency of the wavelet amplitude spectrum. Instantaneous phase
corresponds to the intercept phase of the wavelet. This attribute is also called the "response attribute". Both
Class I and Class II attributes are computed.
Physical Attributes
Physical attributes relate to physical qualities and quantities. The magnitude of the trace envelope is proportional
to the acoustic impedance contrast, frequencies relate to the bed thickness, wave scattering and absorption.
Instantaneous and average velocities directly relate to rock properties. Consequently, these attributes are mostly
used for lithological classification and reservoir characterization.
Geometrical Attributes
Geometrical attributes describe the spatial and temporal relationship of all other attributes. Lateral continuity
measured by semblance is a good indicator of bedding similarity as well as discontinuity. Bedding dips and
curvatures give depositional information. Geometrical attributes were initially thought to help the stratigraphic
interpretation. However, further experience has shown that the geometrical attributes defining the event
characteristics and their spatial relations, quantify features that directly help in the recognition of depositional
patterns, and related lithology.
Most of the attributes, instantaneous or wavelet, are assumed to study the reflected seismic wavelet
characteristics. That is, we are considering the interfaces between two beds. However, velocity and absorption
are measured as quantities occurring between two interfaces, or within a bed. Therefore, we can divide the
attributes into two basic categories based on their origin.
Reflective Attributes
Attributes corresponding to the characteristics of interfaces. All instantaneous and wavelet attributes can be
included under this category. Pre-stack attributes such as AVO are also reflective attributes, since AVO studies
the angle dependent reflection response of an interface.
Transmissive Attributes
Transmissive attributes relate to the characteristics of a bed between two interfaces. Interval, RMS and average
velocities, Q, absorption and dispersion come under this category.
We will define all of the available attributes in the following sections and indicate their categories and their
possible relation to lithology, reservoir characteristics and depositional settings. In most instances individual
attributes may indicate several possible conditions, hence their logically combined use to minimize the inherent
6
uncertainty. We call individual attributes measuring only one quantity "Primitive" attributes. These primitive
attributes may be logically combined to form "Hybrid" attributes. This combination is knowledge based. We
have several attributes of this form, which are described later.
Gabor-Morlet type joint Time-Frequency analysis allows us to study frequency-varying attributes. Instantaneous
spectra, spectral ratio and phase differences provide measurements for bed thickness variation, absorption and
dispersion estimates.
7
Supervised Training of, and Classification by, Neural
Networks
There have been a number of Neural Networks developed within the last couple of decades. Supervised
trainable networks are used in many different fields. In this case, the user provides some examples for the neural
network to learn, and then the network is tested with another data set to check the success of training. One
important point to remember is that the network, if trained properly, will recognize and correctly classify only
those cases included in the training set. Any new conditions not included in the training set will be misclassified
or not recognized. Feed forward, fully connected perceptron artificial neural networks (ANN), Learning Vector
Quantization (LVQ), Probabilistic Neural Networks (PNN), and Radial Basis Function Networks (RBF) are
some of the available networks. Each of the methods has its advantages and limitations.
Computational Procedures
We will investigate several methods of analytic trace computation. They will essentially give the same result.
However, one must keep in mind that the Hilbert transform is only valid for band limited data. For example, the
Hilbert transform of a spike is the Hilbert transform filter itself and it is infinitely long with decay of 1/t. This is
contrary to the definition of a spike. For this reason, we prefer to use band pass filter shaped, Hilbert transform
time domain filters, which will be described later.
a)Transfer the seismic trace to a complex array and place it into the real part, leaving the imaginary part equal to
zero.
c)Zero out negative frequency components, double the positive side, but leave zero and folding frequency
components as are. This will create the causal Fourier transform of the Analytic trace.
d)The inverse Fourier transform will give an input trace that is unaltered in the real part and the imaginary part of
the output will contain the Hilbert transform of the input trace.
The frequency domain method preserves the original spectrum while generating the quadrature trace. Since
almost all computers have standard fast FFT routines, this method represents the fastest and most convenient
procedure.
8
with identical band-pass characteristics, is used to generate the imaginary part. This insures the identical
amplitude spectra of both parts and excessive distortions around zero and folding frequencies are avoided.
Gabor-Morlet Decomposition
One of the problems associated with the Hilbert transform is that it is only valid for narrow band signals. We will
have an increasing degree of uncertainty with increasing bandwidth. For example, in the extreme case, the spike
is the widest bandwidth signal and its Hilbert transform is the time domain response of the transform, which is
infinitely long. This is contrary to the locality of the spike. Luckily our data is somewhat band limited by its very
nature, so we can use the Hilbert transform to some degree. In Gabor-Morlet decomposition we divide the signal
band of the original data into smaller Gabor-Morlet bands. Gabor-Morlet filters are exponentially weighted
complex cosine wavelets:
(2.1)
Decomposition is done by convolving the data by a series of Gabor-Morlet wavelets generated for a sequential
series of values. Gabor wavelets were first introduced to seismic processing by Morlet et al (1982). I am
enclosing a discussion of Gabor-Morlet decomposition by Koehler at the end of this report. Since the wavelets
are complex valued and Analytic, their output will also be Analytic, complex valued. These sub-bands are
summed to form the real and imaginary parts of the wide-band Analytic trace. The sub-bands are generated
equally spaced in the octave scale; hence they do not cover the zero frequency vicinity. Usually 7-21 sub-bands
are sufficient. We use this decomposition in the spectral balancing program `SBAL'.
In the following pages we will cover the conventional and the new attributes. We assume that the data
going into the attribute computation have been adequately processed to contain mainly the subsurface
reflection characteristics. It is important to point out that, we highly recommend the use of 32 bit floating
point seismic data as input. 8 bit fixed point data, used regularly for visual interpretation, do not contain
sufficient dynamic range (+/-128) to produce any reliable results. Some of the recommended processing
9
steps have been discussed in a Rock Solid Images technical paper "Seismic Data Processing with Post-
Stack Attribute Computation as a Consideration" by: Jago and Taner.
We further assume that the trace and its quadrature (Hilbert transformed component) have been computed
previously, using the original data trace for Class I and the signal to noise improved trace for Class II attributes.
The rest of the attributes are computed from this complex trace. Complex trace generation methods are covered
elsewhere in this report.
Envelope
Let the Analytic trace be given by:
(3.1)
'>Where f(t) is the real part corresponding to the recorded seismic data and g(t), the imaginary part of the
complex trace, is the Hilbert transform of f(t). Then the envelope is the modulus of the of the complex function;
(3.2)
E(t) represents the total instantaneous energy and its magnitude is of the same order as that of the input traces.
It varies approximately between 0 and the maximum amplitude of the trace. As indicated on equation (3.2), the
envelope is independent of the phase and it relates directly to the acoustic impedance contrasts. It may
represent the individual interface contrast or, more likely, the combined response of several interfaces,
depending on the seismic bandwidth. Trace envelope is a physical attribute and it can be used as an effective
discriminator for the following characteristics:
•Bright spots,
•Sequence boundaries,
•Unconformities,
•Indicates the group, rather than phase component of the seismic wave propagation,
(3.3)
Events with a sharp relative rise also imply a wider bandwidth, hence less absorption effects. This attribute is
also a physical attribute and it can be used to detect possible fracturing and absorption effects.
•It is used in the computation of group propagation direction. When compared with phase propagation direction,
it may indicate dispersive waves.
•Large changes of the depositional environment, even when the corresponding envelope amplitude may be low.
•Very good presentation of the image of the subsurface within the seismic bandwidth.
Instantaneous Phase
The argument of the complex function is the instantaneous phase:
(3.4)
We display instantaneous frequency in degrees and use the color wheel to display the phase continuously. Zero
degree is displayed as yellow, +120 degrees is displayed as red (magenta) and -120 degrees (+240 degrees)
displayed as blue (cyan). All phase angles between these are color interpolated. In the default color selection we
are using 64 different colors on the color wheel. This gives a rather smooth color variation in the phase display.
11
In a number of displays we have found that 8 different colors are sufficient. These colors represent 45-degree
phase increments. The phase information is independent of trace amplitudes and it relates to the propagation
phase of the seismic wave front. Since, most of the time, wave fronts are defined as lines of constant phase, the
phase attribute is also a physical attribute and can be effectively used as a discriminator for geometrical shape
classifications:
•Shows discontinuity, but may not be the best. It is better for showing continuity.
•Sequence boundaries,
Instantaneous Frequency
Time rate of change of phase is the instantaneous frequency:
(3.5)
12
Since the phase function is multi-valued with jumps, it is better to compute the time rate of change as the
derivative of the arctangent function, which avoids the discontinuities:
(3.6)
Instantaneous frequency is displayed by a color table which starts with red as the lowest frequency and it
gradually changes to yellow to green and finally to blue shades for higher frequencies. The computed output is
given in units of cycles per second. Instantaneous phase represents the phase of the resultant vector of
individual simple harmonic motions. While individual vectors will rotate in clockwise motion, their resultant vector
may, in some instances, form a cardioid pattern and appear to turn in the opposite direction. We interpret this as
the effect of interference of two closely arriving wavelets. This can also be caused by the noise interference in
the low amplitude zones. Because of these reversals, the instantaneous frequency will have unusual magnitudes
and fluctuations. Since instantaneous frequencies are influenced by the bed thickness, we would like to observe
them without too much interference. This we accomplish by using several adjacent traces to form a consistent
output. It has been shown that instantaneous frequency, computed as the time derivative of instantaneous
phase, relates to the centroid of the power spectrum of the seismic wavelet.
Instantaneous frequency computation, due to its interpretational importance, has been a subject of a number of
papers. O'Doherty suggests a different way to compute the instantaneous frequency. Consider the Analytic trace
F(t) and its autocorrelation function . Let the Fourier transform of the analytic trace be represented by
and the autocorrelation function be given by . Therefore, the normalized
autocorrelation function time and frequency responses relate as:
(3.7)
The time derivative of the autocorrelation function corresponds to the multiplication of the power density
spectrum by iw. Therefore, the derivative computed at the zero lag represents the centroid of the power density
spectrum of the seismic event (Note that this is not the same as the instantaneous frequency computed as the
time derivative of phase):
(3.8)
Since the derivative of the real part of the complex autocorrelation at time zero is equal to zero, the value at the
first time lag represents the phase equal to the rate of change of phase per sample. The autocorrelation function
is computed over a number of samples, which represent the averaging window. This computation, therefore, will
be less affected by superimposed reflections. Unbiased mean frequency is similar to the carrier frequency of
radio signals.
Instantaneous frequencies relate the wave propagation and depositional environment, hence they are physical
attributes and they can be used as effective discriminators:
•Corresponds to the average frequency (centroid) of the power spectrum of the seismic wavelet.
13
•Hydrocarbon indicator by low frequency anomaly. This effect is some times accentuated by unconsolidated
sands due to the oil content of the pores.
•Bed thickness indicator. Higher frequencies indicate sharp interfaces or thin shale bedding, lower frequencies
indicate sand rich bedding.
(3.9)
This frequency attribute is less influenced by short wavelength effects. Longer wavelength factors, such as
absorption due to thick beds or massive sand bodies, will change the propagating wavelet characteristics that
can be observed on the weighted mean frequency attribute. It is a physical attribute, indicating longer
wavelength variations.
(3.10)
This attribute shows the interference zones in phase. It is a physical attribute since it relates to closely spaced
events. It can be used in detailed studies:
•Indicates non-reflecting zone, when it appears laterally at random, like ‘salt and pepper’,
14
•Shows fine detail of bedding patterns.
Instantaneous Acceleration
The time derivative of instantaneous frequency, by definition, gives the instantaneous acceleration. We can
compute this both from instantaneous frequency and from time averaged instantaneous frequency. It is obvious
that the time derivative of instantaneous frequency will accentuate the local frequency jumps. Consequently, it
will make the thin bed indicators more prominent. It should also indicate, to some degree, the effect of absorption
by showing the frequency dispersion of seismic signals going through unconsolidated or quickly deposited
layers.
(3.11)
The acceleration computation can also be made by the O'Doherty method. The second derivative of the complex
autocorrelation function corresponds to the multiplication of the power density spectrum by the square of w.
Therefore, the second derivative computed at zero lag divided by the modulus of the autocorrelation function at
the zero lag will be equal to the acceleration;
(3.12)
It is interesting to note that this equation is same as equation 3.15 given below. Equation 3.15 gives the square
of the RMS frequency (second moment of the power density spectrum). Here (equation 3.12) we compute the
instantaneous acceleration. The derivative of the time averaged frequency will be subtler, however, these
displays need further investigation.
•Higher resolution, may have somewhat higher noise level due to differentiation,
(3.13)
15
(3.14)
And RMS frequency (the second moment of the Power spectrum) is given by the expression:
(3.15)
(3.16)
We can now examine these statistical measurements of the power spectrum in the form of useful attributes.
These computations represent the statistics of the seismic wavelet computed over some time window. Hence,
they are more closely associated with the time smoothed instantaneous attributes. We will, however, compute
and display them continuously for all of the data samples. By definition , the centroid frequency is the mean
frequency where an equal amount of energy exists on either side of this frequency.
The variance with respect to the mean frequency (standard deviation) indicates the width of power spectral
density distribution over a band of frequencies; hence we can use it as an indication of the spectral bandwidth.
Barnes (1992) suggests instantaneous bandwidth can be computed by:
(3.17)
This equation measures the absolute value of the rate of change of envelope amplitude. We could also compute
the instantaneous bandwidth from the geometric equation shown above (equation 3.16) using centroid and RMS
frequency measurements. However, Barnes’ expression is simple enough to be practical. We compute and
display the bandwidth in terms of octaves.
Instantaneous bandwidth is a statistical measure of the seismic wavelet, but it relates to various physical
conditions:
•Represents seismic data bandwidth sample by sample. It is one of the high-resolution character correlators.
16
the second derivative of the normalized complex autocorrelation function will give the required results. However,
we use a second method of computation as suggested by equation 3.16. Since instantaneous frequency,
computed as the time derivative of instantaneous phase, represent the mean frequency (centroid of the power
spectrum), then the centroid of the second moment of the power spectrum, or the RMS frequency, is obtained
by:
(3.18)
Barnes calls this the dominant frequency. The display for this is similar to the instantaneous frequency display, in
units of cycles per second.
(3.18)
Where decay(t) is the instantaneous decay rate, which is defined as the derivative of the instantaneous
envelope divided by the envelope.
Except for a factor of , this is similar to the instantaneous bandwidth. The decay rate can take both positive
and negative values. Hence, the instantaneous quality factor is the ratio of instantaneous frequency to twice the
instantaneous bandwidth. Barnes points out that this definition is consistent with the standard definitions of the
quality factor (Close, 1966 and Johnson and Toksöz, 1981). We must point out that this Q computation is the
short wavelength variation of the Q value, hence it gives relative values. It is a transmissive attribute, similar to
the interval and instantaneous velocities. It is also a physical attribute with a strong relation to porosity,
permeability and fracture.
•Indicates local variation of Q factor, similar to the relative acoustic impedance computation from the seismic
trace. Longer wavelength variation should be computed by spectral division and added to this attribute.
•May indicate liquid content by ratioing pressure versus shear wave section Q factors.
•It is a transmissive attribute; its various wavelength components should be estimated in a similar way to the
average velocity and velocity inversion procedures.
(3.19)
Which is:
17
(3.20)
Therefore, by integrating the zero phase trace, we will get the band-limited estimate of the natural log of the
acoustic impedance. Since it is band limited, the impedance will not have absolute magnitudes and the stack
section is usually the estimate of zero offset reflectivity; hence it is called relative acoustic impedance.
(3.21)
In practice, however, due to noise and imperfect spectral content of the seismic data, relative acoustic
impedance computed by integration will develop arbitrary long wavelength trends. Since the seismic data does
not contain any viable information at very low frequencies (due to band pass filtering in the field and during
processing) these long wavelength trends cannot be utilized. We remove these with a low cut filter.
Computation is a simple integration followed by a low pass filtering, without any exhaustive inversion. It reflects
physical property contrast, hence it is a physical attribute effectively utilized in many calibration procedures.
•Relative acoustic impedance shows band limited apparent acoustic impedance contrast,
•Indicates discontinuities.
Definition: We assume that a seismic wavelet, or a compound wavelet, occupies the time span between two
adjacent envelope minima. In reality, wavelets will extend beyond these minima, however, for display purposes
we consider the minima as the boundary. Furthermore the part between the minima represents a significant
portion of the wavelet.
Therefore, we search for the envelope minima and maxima positions. All envelope maxima positions are used to
obtain the instantaneous attributes and these values are assigned over the time zone between the minima on
either side of the maximum position. It has been found, however, that the computation of the attributes at the
18
sample points is not sufficiently accurate. For example, if we consider 4 millisecond-sampled data, for a 40 Hz.
wavelet each sample interval will have a 60-degree phase difference. This is much too large to be useful. In
order to be accurate, therefore, we compute the position of the maximum of the envelope by quadratic
interpolation. All of the other wavelet attributes are also computed at the interpolated maximum position of the
envelope.
•Wavelet envelope
•Maximum positive time derivative of wavelet envelope. This attribute is computed at the onset of the wavelet at
the position where the time derivative is a maximum. It is obvious the time derivative at the envelope maximum
will be zero, hence useless. The maximum time derivative scaled by the envelope maximum will have
implications of absorption.
•Wavelet second derivative of envelope. This indicates the sharpness of the peak, hence some indication of
bandwidth.
•Apparent polarity. Polarity is assigned based on the Wavelet phase. If the phase is between -90 and +90
degrees a positive polarity is assigned, otherwise a negative polarity is assigned. Its magnitude is equal to the
interpolated envelope magnitude.
•Wavelet instantaneous frequency. Similar to phase, a good indicator of lateral continuity and bandwidth.
•Wavelet acceleration
•Wavelet bandwidth
•Wavelet dominant (RMS) frequency. This is direct computation of the RMS frequency of the wavelet power
spectrum.
Wavelet attributes use the same color table as the instantaneous attributes in order keep their relationships
apparent.
All of these attributes have similar discriminatory significance, but they relate to individual events, rather than to
individual samples.
Geometrical Attributes
Introduction
The phase component of seismic data contains an expanse of useful information. This information can be
obtained if the seismic trace is considered either as a complex entity, or as an analytic function consisting of the
recording of the potential and the kinetic components of the energy flux at the surface of the earth. While the
measurement of the phase and its time derivative gives direct information to the state and variation of energy in
a temporal sense, the measurement, extended to include the spatial information, yields information on the wave
number and the visible direction of propagation. It is well known that, wave propagate in two separate mode,
19
phase and group. In dispersive medium these two modes will have different propagation velocities. Here, we
assume a non-dispersive mode and consider the phase component only. Separate computation of phase and
group propagations have been discussed by Barnes (1994).
A further, and perhaps more important benefit, comes from the redundant information contained along the wave
front which helps to reduce the effects of noise. One of the more useful attributes, the instantaneous frequency,
suffers from the influence of noise, which breaks the spatial continuity. This appears, even in time averaged
frequency displays, as trace wide color streaks. O'Doherty (1992) has shown that by including the spatial
information in the averaging process, the instantaneous frequency and dip computations can be stabilized to the
degree that the events can be tracked with more confidence.
Let f(t,x) be the recorded seismic trace which is the real part of the complex trace, then g(t,x) is computed by
Hilbert or Gabor filtering of the f(t,x) trace, then the complex traces are defined by (now in two dimensional
sense)
(4.1)
(4.2)
(4.3)
(4.4)
One of the main problems is the discontinuity in the phase function. It is not a continuously increasing function.
This gives rise to negative instantaneous frequencies, most likely due to the interference caused by two or more
overlapped wavelets. The main hypothesis is ``Simple seismic wavelets are such that their amplitudes rise to a
maximum at a rate proportional to their general spectral content and their phase increases continuously without
any local reversal''. It has been shown that the rise time is affected by the energy loss at higher frequencies or is
due to dispersion effects. One method of Q computation is based on the wavelet envelope rise time. According
to our proposed condition of continuous phase increase, we consider the phase reversals are due to the
interference of closely spaced wavelets with opposite polarities. If we plot the hodogram (the trace vector in the
complex domain) the end of the vector rotates in a counter clockwise direction, in a somewhat circular orbit with
a time varying radius. The interference appears as a cardioid pattern that shows the resultant vector going in the
opposite direction for a short time period and then proceeding in the original direction. The average angular
velocity of the vector is the mean frequency of the corresponding wavelets. This is similar to the carrier
frequencies of radio waves. In our computation we will have to compute and remove the slowly time varying
component of the angular velocity (instantaneous frequency) or the mean slope of the phase function. The
remainder will show the local deviations. These local deviations, when displayed in a sectional form, will show
the thin bed structure of the seismic data. Local deviations can also be due to the influence of the noise vector,
therefore, we have to separate the effects of the noise from those of the wavelet interference. Usually the noise
effects are visible in the zones of low signal amplitude. Interference occurs at the arrival of a second signal with
larger envelope magnitudes. This condition has led us to develop a weighted filtering scheme for smoothed
instantaneous frequency determination.
20
This weighted frequency is a slowly varying function of time and represents, essentially, the Wavelet frequency.
Since the reflectivity function of the earth is finely structured and the wavelets used to illuminate it have,
relatively, a much longer wavelength, we are essentially observing the interference patterns, rather than a direct
observation of images. In the future we will need to develop some new methods to take advantage of the
interference patterns and make better and finer detailed estimates of the subsurface images. Thin bed
indicators, instantaneous and average frequencies were covered in the second section.
Instantaneous frequency was one of the earliest used attributes. It is the rate of change of the instantaneous
phase with respect to time, and is a temporal measurement. Similarly, spatial measurements lead to the wave
numbers, which will be useful in indicating the direction of the phase component of the propagating wave field.
The Eikonal equation gives us the relationships between the temporal and spatial variables.
Let be the wave-number (radians/meter) in the x direction and the wave number in the vertical z direction
and be the angular velocity (radians per second) and V the propagation velocity (meters/second), then a
plane harmonic wave represented in the form:
(4.5)
will satisfy the scalar wave equation. Substituting this into the wave equation we will obtain the Eikonal
relationship:
(4.6)
(4.7)
The time gradient on the seismic section is , then from Eikonal equation, this is equal to:
(4.8)
In the case of finely sampled data sets ' and ` ' can be computed directly from the lateral and temporal
derivatives of the instantaneous phase function. If the data is coarsely sampled we will be subjected to spatial
aliasing, hence we cannot obtain a good measurement of dips this way. In these cases local dip scanning gives
a little coarser but more accurate results. An alternative way is to interpolate the data to finer spatial sampling
intervals and then compute the dips. However, this is not an efficient method because, during the interpolation,
there is an implied step of dip estimation in order to interpolate the aliased data properly.
21
Ron O'Doherty showed that the ' method of computing dips suffers in the same way as the anomalous
variations in the instantaneous frequencies. For the same reason he recommended temporal and spatial
smoothing in order to get sectional consistent dips. We are using amplitude weighted smoothing similar to the
frequency smoothing in order to minimize the effects of the low amplitude noise.
O'Doherty proposes a slightly different method of smoothing, which is done during the computation of the
instantaneous frequencies and wave numbers.
It was shown by a number of authors that the instantaneous frequency computed at the maximum of a non-
complicated wavelet (like zero or nearly zero phase wavelets) corresponds to the centroidal frequency of the
power spectrum of the wavelet:
(4.9)
We know that the autocorrelation function is the inverse Fourier transform of the Power density spectrum:
(4.10)
(4.11)
Which is equal to the denominator of the above expression 4.9. On the other hand, the derivative of the
autocorrelation function in the frequency domain is its Fourier transform multiplied by :
(4.12)
Which is essentially the numerator of the equation 4.9. At the zero lag the exponential portion of the above
expression vanishes to give us the derivative of the complex autocorrelation function at the zero lag. We can
therefore compute the instantaneous frequency by the ratio of:
(4.13)
In order for this equation to make sense, must be uniquely defined at all frequencies and must be a
complex valued function. And, since it is a complex autocorrelation function, it should have Hermitian type
symmetry - positive lags are the complex conjugate of the negative lags. Its first derivative at the zero lag will be
purely imaginary. The imaginary part of the autocorrelation function will be zero and the first derivative of its real
part will also be zero. Therefore, the instantaneous phase of the complex autocorrelation function at time lag one
will represent the amount of phase change per sample time unit. So the instantaneous frequency, defined as
radians per sample, will be given by:
22
(4.15)
Since instantaneous frequency is the time derivative of the instantaneous phase, equation 4.13 gives the
evaluation of this derivative at the peak of the envelope, which is described as the mean frequency:
(4.16)
It is interesting to note here that, since the autocorrelation function is complex valued, we will have 4 samples
per cycle at the folding frequency. In order to compute the instantaneous frequency correctly we have to
compute the phase as shown in equation 4.14 from zero to 180 degrees, not between -90 to +90.
O'Doherty suggests using time windows long enough to suppress the effects of the noise. A time window of the
order of an average period should be sufficient. This computation should produce instantaneous frequencies
similar to the weighted mean frequencies described above. In the computation of horizontal wave numbers
however, this method may be more convenient, since it uses less samples.
O'Doherty also suggests spatial averaging of the autocorrelation function as well as the temporal averaging. This
way the influence of noise is minimized and laterally coherent wavelet information is strengthened.
The second derivative of phase with respect to the spatial variable x shows lateral (linear) continuity. If the
second derivative is zero or near zero then the lateral change of phase is linear, hence the event is linearly
continuous. Any large second differences indicate some form of discontinuity or curvature of the arrival times of
the event. For the sake of practicality we will use linear continuity over 3 adjacent traces as the measure of
lateral continuity.
However, before we consider the lateral continuity, we will look into the other lateral measurement, the wave
number in the spatial x direction. The wave number is computed in the same manner as in the temporal
direction. We have found, as mentioned above, that O'Doherty’s method, using the first derivative of the
autocorrelation function in spatial direction, has advantages and it uses less spatial samples. Therefore the wave
number is given by:
(4.17)
Where is the first space lag of the autocorrelation function computed in the lateral direction. The number
of traces that are included in the autocorrelation computation controls the degree of smoothing.
According to the Eikonal equation, the ratio of temporal frequency to the spatial wave number is the time
gradient of the wave front arriving at the surface.
(4.18)
And,
(4.19)
Where is surface arrival angle of the wave front and is the velocity at the layer where the receivers are
implanted. Scheuer and Oldenburg (1988) have used this ratio to determine the local phase and the apparent
velocity from common shot data. This measurement is displayed as the apparent dip. By assigning the surface
velocity, the apparent dip can be converted into the exit angle of the surface arriving wave front.
23
(4.20)
This computation can easily be extended to 3-D data sets. Similar to the expressions given above, by
introducing the component in the y direction, we will have:
(4.21)
And;
(4.22)
Equations 4.19 and 4.22 give the time gradients measured in x and y directions. Then the maximum gradient is
given by:
(4.23)
(4.24)
One of the important stratigraphic measurements is the lateral continuity of the reflecting horizons (Sangree,
1988 and Vail, 1988). We are also interested in the concordance or the consistency of the bed thickness (if they
are parallel or not). As we have discussed earlier, we will measure the linear continuity by the second order
partial differences of the phase in the lateral direction. Increasing values of second difference will indicate an
increasing amount of reflection curvature. The direction of continuity will be obtained from the dip
measurements. A smoothed version of the lateral continuity measurement can be obtained by the inclusion of
more samples in the spatial and temporal directions. This is discussed in the Dip Scan section.
Another important indicator is the location of event terminations in the form of on-lap, top-lap truncation and etc.
(Sangree and Vail). These event terminations are used to determine the stratigraphic sequence boundaries.
Determination of abrupt phase changes, as in the case of faults on a migrated section, is simple. The main
difficulty is locating gentle dip differences that might be associated with down-lap or top-lap conditions. In un-
migrated data, faults have diffraction hyperbolae, which makes them appear continuous. Migration process
focuses these diffraction hyperbolae to their origination points, thus making the discontinuities more visible. At
present, we will use the computed lateral semblance to indicate the lateral event termination and continuity. We
will also consider sequence boundary detection as one of the principal research subjects for the near future.
One way to indicate event termination is to highlight the opposite of continuity as discontinuity. That is, we will
highlight the larger values of the lateral second difference of instantaneous phase. This method will probably be
less diagnostic for the beds that merge gradually. A second method is to laterally track lines of constant phase
and to indicate their terminations. This tracking could also be carried out on narrow frequency band filtered data.
We would have to observe the consistency of truncation over several bands in order to have higher redundancy.
This method will be adversely influenced by noise and by zones of chaotic reflection. Therefore a simple method
to check consistency will be necessary.
24
method will give aliased results. A third method of dip scanning will overcome this difficulty. In this section we will
continue to discuss the last two methods.
Event Continuity
This is an intermediate result of a hybrid attribute computation. The objective of this attribute is to develop the
lateral continuity of peaks and troughs, and to classify the type of discontinuity. At this time all real trace positive
peaks are output as +1. and negative peaks output as -1. Everything else will be output as zero. The display
therefore will show only the peaks and troughs, all with same magnitude. The development phase will connect
like peaks and establish discontinuities. Based on the type of discontinuity, it will attempt to classify the
terminations as top-lap, on-lap etc.
Will give the maximum dip, and its direction (Instantaneous Dip Azimuth) will be:
As mentioned above, this computation is valid for dips of up to 180-degree (exclusive) phase differences. Actual
dips beyond this will produce aliased results, hence they will be misleading. In such cases dip scanning with a
greater number of traces becomes necessary. As a side note, when time migrated sections are used, it is highly
recommended that the image trace spacing should be designed to eliminate any possible aliasing problems. In
most instances, conventional surface recording intervals of 12.5 or 25 meters will be sufficient to generate
images at target levels below one second at less than half of the surface interval, without violating any sampling
laws.
(4.25)
Linearly continuous events will give zero curvature. Beds with a hummocky appearance will have non-zero
curvature values. Non-reflecting zones will have highly variable curvature values in time and space. This
attribute highlights the zones of large lateral dip variation; hence it can be a good indicator of faults and
fractures.
25
improved signal to noise ratio output. To measure the similarity, semblance, which was developed in 1965 by
Seiscom, was used. This information was also used in the velocity analysis to provide the surface arrival dip as
part of the stacking velocity estimate. These estimates were later used for dip, depth and interval velocity
computations. Since discontinuities are the inverse of continuities, dip scan sections have highlighted
discontinuity as well as continuity. In the late 1980's, lateral maximum semblance values were introduced as
additional geometrical attributes.
In some instances, to further improve the lateral continuity of the traces, the scanned output is scaled by a
percentage of the semblance value. Since the semblance values lie between zero (corresponding to complete
lateral dissimilarity) and 1.0 (100 percent lateral continuity), multiplying with semblance values will enhance
further the laterally continuous events while suppressing zones of discontinuity. However, this additional process
produces rather strong effects, hence it should be used with care.
The semblance is a measure of coherent power existing between a number of traces versus the total power of
all traces, as given by;
(4.26)
Where is the m'th trace of the gather, and N is the length of computation window.
The lateral "coherency" indication is computed as semblance that is determined by considering a user-selected
number of traces. The number of traces depends on the signal to noise ratio (poor S/N, more traces), bed
curvature (more curvature, less traces), and higher lateral discontinuity resolution (better resolution, less traces).
The semblance is generally computed over a running 40 - to - 80-millisecond time window (shorter time windows
are appropriate for wide band data). For good S/N data, 3 adjacent traces for scanning will give higher
resolution. 5 or higher numbers of traces will tend to lose lateral discontinuities and start producing longer linear
components of existing events. The dip scan range and dip increments are input in milliseconds per trace,
regardless of inline and cross line distances. However, if the actual inline and cross-line spacing has been
entered, then the dip attribute output will be correctly scaled in milliseconds per 100 meters.
26
In 3-D data sets, dip scanning is done for in-line and cross-line traces separately. Each scan will output S/N
improved trace, maximum semblance and dip. These are combined to produce one dip, maximum semblance
and S/N improved trace. Computation of maximum dip and its azimuth is given earlier in this report.
In 2-D cases the output from subroutine DIPSCN is used as direct input for Class II attribute computation. In 3-D
cases the output from in-line and cross-line scans are combined by semblance weighting as:
(4.26)
Where and are in-line and cross-line semblance scanned trace outputs,
respectively. Equation 4.26 shows that the final trace is generated by proportional summation of two traces
weighted by their percent semblance values. This output is used as the input trace for Class II computation in 3-
D data sets.
Let D(t) be the average dip, d(t) instantaneous dips and w(i) are the weights, the average dip is computed by:
(4.27)
And the weights are computed iteratively after each pass by:
(4.28)
Where K is a constant controlling the severity of rejecting outliers. Normally K=4 could be used.
Lateral Continuity
27
The maximum lateral semblance as output from DIPSCN is used directly in 2-D cases. In 3-D cases the output
from in-line and cross-line scans are vectorially summed to produce the lateral continuity attribute. This attribute
shows the regions of lateral consistency, or the similarity of depositional environment. It will also show the areas
where the continuity has been disturbed.
Hybrid Attributes
Most of the attributes discussed previously are defined as "Primitive" attributes. They basically measure a
particular physical or geometrical condition or statistic. They are primitive, but they can also be used as building
blocks and, by combining them logically, can be used to detect more complicated conditions. The logic of the
combination is experience and knowledge based. For example, the fluid factor computed by the AVO analysis is
a hybrid attribute based on experience. In this section I will give descriptions of several hybrid attributes. These
are our initial attempts to combine attributes with a knowledge-based logic. Since their introduction they have
been shown to be effective discriminators. We will continue in developing additional hybrid attributes as we
incorporate an additional knowledge of Rock Physics, Geology and Reservoir characteristics.
(4.29)
Zones with parallel bedding will have zero, or close to zero, variance or standard deviation. Increasing
values of standard deviation will indicate increasing dip variation within the window. This attribute,
computed as dip statistics, provides a direct indication of the geometrical configuration of beds and
interfaces.
28
(4.30)
Zones of chaotic bedding will have high attribute values. Massive carbonate zones with incoherent noise, like
weak reflections, will show lower values of chaotic attributes.
Zones of Unconformities
This is a hybrid attribute for detecting possible surfaces of unconformity. The routine first detects the peaks of
the seismic trace envelope. The location of the peak of the envelope represents the seismically visible bedding
interface, a possible unconformity surface. Its magnitude is relative to the degree of impedance contrast. The
next process is to compute additional contrasts. Three samples of dips of maximum lateral semblance,
instantaneous frequency, lateral semblance and variance of dip attributes are picked from both sides of the
interface. A normalized vector dot product is formed between the set of attributes on two sides. Since the vector
dot product is normalized, (the components are the attributes), its values will lie between + /- 1. +1 represents
the highest similarity, indicating there is no difference between the sides from the dip, frequency and continuity
points of view. A smaller value will indicate some degree of difference. A dissimilarity factor is formed by
subtracting the normalized vector dot product from 1.0, which is then multiplied by the envelope peak magnitude
and output as the indicator of the Zones of Unconformity. A larger magnitude output represents a higher degree
of contrast.
Shale Indicator
This a hybrid attribute that combines several primitive attributes to detect possible shale zones in a clastic
environment. Shale beds are identified by their geometrical configuration as; thin parallel beds with high lateral
continuity. This attribute uses Instantaneous higher frequencies as the thin bed indicator, lower values of
standard deviation of maximum semblance dips as the parallel bed indicator and semblance and its variance as
the lateral continuity indicator. The highest output value indicates the highest possibility of shale occurrence.
Lower magnitudes indicate the possibility of lithologies other than shale, such as sand or carbonate beds.
FILTER DESIGN
In this chapter we will present the design of several filters used in seismic data processing and Attribute
computation. These will include two types of band pass filter, a Hilbert transform filter in convolutional form and a
special differentiation filter and a specific process to transform from the cepstrum domain to the time domain by
a simple recursion. All of these filters are explained in the text and the corresponding Fortran subroutines are
included in the appendix.
We have found that the amplitude or power spectrum of band-pass filters has to be continuous and continuously
differentiable; that is smooth in the frequency domain. This characteristic produces the least undulation of the
envelope of the filter wavelet in the time domain, a most desirable property for high resolution and stratigraphic
29
processing. Another point is that the filters described here pass all of the frequency band to some degree. At no
time do they completely suppress any frequency, as is the case with Ormsby type filters. Therefore they can be
inverted, and their inverses will have no undesirable poles in the frequency domain or corresponding
reverberations in the time domain.
We define the filters by their low and high cut frequencies, or their 1/2 power points at the low and high
frequency sides. For example, a filter with 8 to 60 Hz pass band and with slopes of 18 dB/Oct. on the low side
and 24 dB/Oct. on the high side means that the data will be suppressed by less than 1/2 power from 8 Hz. to 60
Hz. Outside these frequencies, suppression will gradually increase beyond 1/2 power or -3 dB. On the low side,
one octave lower than 8 Hz, at 4 Hz., the suppression will reach -21 dB. or 1/125'th power of the input. On the
high side, at 120 Hz. (one octave higher than 60 Hz.) it will reach -27 dB. Or 1/500'th of input power.
We will use and as the low and high cut frequencies, and and as the low and high frequency
side roll-off rates respectively.
Butterworth Filter
Butterworth band-pass filters are composed of the multiplication of two sigmoid functions. The following
computations are performed in the SRCBAND subroutine.
(5.1)
Where f is the frequency given in Hz. We compute the constant N from the roll-off rate requirement, since the
half power point at is automatically satisfied by eq. 5.1; If the roll-off rate is , then by definition, the
power spectrum at frequency will be minus ‘3.0103 ’ decibel down;
Let , then
(5.2)
30
(5.3)
The power ratio at 1/2 of the low cut frequency will be down from 1/2 power at frequency, then
Let , then
(5.4)
(5.5)
(5.6)
Since we have used only the real part with the imaginary part equal to zero, then the will be symmetrical.
In order to form the cepstrum of the minimum phase wavelet we will set the negative lags to zero:
for,
And double all values for lags larger than zero, leaving the zero lag as is:
31
for,
We now have the cepstrum of the minimum phase filter. Once the cepstrum of the minimum phase wavelet is
obtained, then its time domain response is computed by a simple recursion (Oppenheim and Schafer, 1975, p.
504). This is done in the CEPTIM subroutine. Let be the n'th element of the cepstrum and the m'th element
of the filter in the time domain, then by recursion we compute later elements of the filter from earlier computed
elements as:
(5.7)
Recursion will produce the rest of the filter weights. In order to have gradual reduction of filter
amplitudes down to zero at the tail end, we use Butterworth type sigmoidal tapers as used in the TAPER
subroutine.
Taner Filter:
We have found that Ricker wavelets have the most desirable shape; their envelope is a decaying exponential
function. All of the filters designed by specifying four corners of the spectrum, like Ormsby filters, will have
envelopes with many lobes. These lobes are very disagreeable for the fine definitions required by stratigraphic
processing, as well as regular processing. In order to achieve smoothness of the wavelet envelopes, I
experimented with exponential functions in the log and octave domain. Low and high pass filters are simple
exponential functions, whose characteristics are controlled by the user given band limits and roll-off rates.
(5.8)
Where x is the frequency given in octaves as . If then we will have 1/2 power or
in log scale. Therefore;
We compute the constant from the roll-off rate requirement, since half power point at x=xH is automatically
satisfied by eq.5.8.; If the roll-off rate is , then by definition the power spectrum at 1 octave higher than
frequency will be ' decibel down;
Or,
32
(5.9)
(5.10)
The power ratio at a frequency one octave lower than the low cut frequency will be lower than the 1/2
power at frequency. Therefore;
Or,
(5.11)
(5.12)
Computation of zero and minimum phase Taner filters are given in TANER subroutine.
33
filter’s real and imaginary parts are also Hilbert pairs, then their amplitude spectrum will be causal. Since both
the real and imaginary parts have the same amplitude spectrum, we double the computed Butterworth amplitude
spectrum for all positive frequency values, leaving the zero frequency as is. An inverse Fourier transform will
form the complex Hilbert filter in the time domain. In another way, a cosine transform of the amplitude spectrum
will generate the symmetrical real part of the Hilbert filter and a sine transform will form the antimetric imaginary
part. The HILBAND subroutine generates the pair of Hilbert filters with a Butterworth spectrum controlled by the
user’s bandwidth and roll-off rate requirements. Hilbert filters can be designed with Taner filter type spectral
controls by replacing the spectral design portion of the subroutine.
We design the filter weights so that the derivative at time zero will be:
(5.13)
If , then ,
And
, then ,
should result by this convolution. Therefore;
(5.14)
Or;
(5.15)
Therefore,
(5.16)
(5.17)
And,
34
As we can see, this operator decays by a slow 1/t rate, making a very long filter. Secondly, it will have a very
sharp edge at the folding frequency, which will create undesirable side effects on the output. In order to minimize
these side effects we have designed differentiation filters with gentler slopes on the high frequency side.
Ackroyd, M.H., 1970, Instantaneous spectra and instantaneous frequency: Proc. IEEE, 58, 141.
Adamick, J. H., D., Skoyles, D., DeWildt, J. and Erickson, J., 1994, AVO as an exploration tool: Gulf of Mexico
case studies and examples: 64th Annual Internat. Mtg., Soc. Expl. Geophys., Expanded Abstracts, , 94, 1107-
1111.
Addy, S. K., 1997, Attribute analysis in Edwards limestone in Lavaca county, Texas: Annual Meeting Abstracts,
Society Of Exploration Geophysicists, 737-740.
Adler, F., 1998, Attribute Analysis Using Locally Scaled Regression : 61st Mtg. Eur. Assoc. Expl Geophys.,
Extended Abstracts, European Association Of Geophysical Exploration, Session:P139.
Akbar, N., Dvorkin, J., Nur, A., 1993, Relating P-wave attenuation to permeability: Geophysics, 58, 20-29.
Alam, A., Caragounis, P., Matsumoto, S. and Hurst, C., 1995, Reservoir Classification with Seismic Attributes:
57th Mtg. Eur. Assoc. Expl Geophys., Extended Abstracts, , 95, Session:A037.
Alam, A., Matsumoto, S., Hurst, C. and Caragounis, P., 1995, Qualitative porosity prediction from seismic
attributes: 65th Annual Internat. Mtg., Soc. Expl. Geophys., Expanded Abstracts, , 95, 313-315.
Anstey, N.A., 1977, Seismic Interpretation: The Physical Aspects. Internal. Human Res. Develop. Corp.
Antonini, M. and Schweitzer, C., 1993, Inverse modelling vs. attribute analysis - 3D seismic reservoir
description of Rotliegend sandstones: 55th Mtg. Eur. Assoc. Expl Geophys., Abstracts, , 93, Session:A026.
Arts, R. and Ampilov, Y., 1999, Enhancement of a Flat Spot From Seismic Data Using Multi-Attribute Analysis:
61st Mtg. Eur. Assoc. Expl Geophys., Extended Abstracts, European Association Of Geophysical Exploration,
Session:5010.
Bahorich, M. S. and Bridges, S. R., 1992, The seismic sequence attribute map (SSAM): 62nd Annual
Internat. Mtg., Soc. Expl. Geophys., Expanded Abstracts, , 92, 227-230.
Bahorich, M. S. and Farmer, S. L., 1995, 3-D seismic discontinuity for faults and stratigraphic features: The
coherence cube: 65th Annual Internat. Mtg., Soc. Expl. Geophys., Expanded Abstracts, , 95, 93-96.
Bahorich, M. S., 1994, The seismic sequence attribute map (SSAM): 56th Mtg. Eur. Assoc. Expl Geophys.,
Extended Abstracts, , 94, Session:H042.
Bahorich, M. S., Lopez, J., Haskell, N. L., Nissen, S. E. and Poole, A., 1995, Stratigraphic and structural
interpretation with 3-D coherence: 65th Annual Internat. Mtg., Soc. Expl. Geophys., Expanded Abstracts, , 95,
97-100.
35
Ballesteros, M. W., 1991, Harriet field - A two dimensional modeling study: 8th Aust. Soc. Expl. Geophys.
Conf., Expl. Geophys., 22, 23-26.
Barnes, A. E., 1994, Theory of two-dimensional complex seismic trace analysis: 64th Annual Internat. Mtg.,
Soc. Expl. Geophys., Expanded Abstracts, , 94, 1580-1583.
Barnes, A.E., 1990, Analysis of temporal variations in average frequency and amplitude of COCRP deep seismic
reflection data: 60th Ann. Internal. Mtg., Soc. Expl. Geophys., Expanded Abstracts, 1553-1556.
Barnes, A.E., 1991, Instantaneous frequency and amplitude at the envelope peak of a constant-phase wavelet:
Geophysics, 56, 1058-1060
Barnes, A.E., 1992, Another look at NMO stretch: Geophysics, 57, in press.
Barnes, A.E., 1992, Instantaneous spectral bandwidth and dominant frequency with applications to seismic
reflection data: Geophysics,
Barrett, A. V., 1994, Risk analysis of AVO attributes: 64th Annual Internat. Mtg., Soc. Expl. Geophys.,
Expanded Abstracts, , 94, 950-951.
Barton, K., 1997, Delineation of oil and gas fields in Bredasdorp Basin, offshore south Africa, using
seismic attribute mapping, seismic inversion and AVO techniques: Annual Meeting Abstracts, Society
Of Exploration Geophysicists, 631-634.
Behrens, R. A., MacLeod, M. K., Tran, T. T. and Alimi, A. O., 1996, Incorporating seismic attribute maps in 3-
D reservoir models: Annual Meeting Abstracts, Society Of Exploration Geophysicists, 1933.
Boashash, B., O'Shea, P., and Ristic, B., 1991, Statistical/computation comparison of some estimators for
instantaneous frequency: Proc. IEEE ICASSP-91, 3193-3196.
Bodine, J.H., 1984, Waveform analysis with seismic attributes: Presented at the 54th Annual International
Meeting of the S.E.G. in Atlanta, Georgia.
Bossert, A., Blanche, J. P., Capelle, P., Marrauld, J. and Torheim, E., 1993, Seismic monitoring on the Frigg
gasfield (Norway) using AVO attributes and inversion: 55th Mtg. Eur. Assoc. Expl Geophys., Abstracts, , 93,
Session:A009.
Bracewell, R.N., 1978, The Fourier transform and its applictions: McGraw-Hill Book Co.
Brown, A.R., 1996, Interpreter’s Corner - Seismic attributes and their classification: 15, 10, no. 1090, .
Chen, Q. and Sidney, S., 1997, Advances in seismic attribute technology: Annual Meeting Abstracts, Society Of
Exploration Geophysicists, 730-733.
Chen, Q. and Sidney, S., 1997, Seismic attribute technology for reservoir forecasting and monitoring: The
Leading Edge, 16, no. 05, 445,447-448,450,453-456.
Chung, H. and Lawton, D. C., 1990, AVO analysis and complex attributes for a Glauconitic gas sand bar: J. Can.
Soc. Expl. Geophys., 26, 72-86.
36
Clark, V.A., Tittman, B.R., and Spencer, T.W., 1980, Effect of volatiles on attenuation, Q and velocity in
sedimentary rocks: Journal of Geophysical Research, 85, 5190-5198.
Clearbout,J.F., 1976, Fundamentals of geophysical data processing: With applications to petroleum prospecting:
McGraw-Hill Book Co.
Close, C.M., 1966, The analysis of linear circuits: Harcourt, Brace & World, Inc.
Cohen, L., 1989, Time-frequency distributions-A review: Proc. IEEE, 77, 941-981.
Cohen, L., and Lee, C., 1990, Instantaneous bandwidth for signals and spectogram: Proc. IEEE ICASSP-90,
2451-2454.
Cooke, D., 1999, Techniques for improving seismic attribute versus well-log property cross plots with synthetic
and real data examples from Prudhoe Bay Field: Annual Meeting Abstracts, Society Of Exploration
Geophysicists, 69-72.
Cooke, D., Ball, V., Muryanto, T., O'Donnell, G. and Sena, A., 1999, What is the best seismic attribute
for quantitative seismic reservoir characterization?: Annual Meeting Abstracts, Society Of Exploration
Geophysicists, 1588-1591.
Dasios, A., Astin, T. and McCann, C., 1998, Weighted Peak Instantaneous Frequency (WPIF): an Attribute for
Reservoir Fluid Detection and Monitoring: 61st Mtg. Eur. Assoc. Expl Geophys., Extended Abstracts, European
Association Of Geophysical Exploration, Session:P136.
de Voogd, N., and Rooijen, H., 1983, Thin-layer response and spectral bandwidth, Geophysics, 48, 12-18.
Donelick, M., Bahorich, M. and Farmer, S., 1993, Prediction of gas-oil and oil-water contacts in Norwegian
block 30/8 by means of seismic attribute mapping - A new tool for the interpreters' arsenal: 55th Mtg. Eur.
Assoc. Expl Geophys., Abstracts, , 93, Session:A012.
Fuh, S., Wu, M. and Hsiuan, T., 1995, Seismic characters of subtle stratigraphic traps in erchungchui formation:
Application for evaluation of hydrocarbon potential of future prospects: Annual Meeting Abstracts, Society Of
Exploration Geophysicists, 525-528.
Futterman, W.I., 1962, Dispersive body waves: Journal of Geophysical Research, 67, 5279-5291.
Gallagher, J. W. and Heggland, R., 1995, Shallow gas evaluations based on conventional 3-D seismic data: 65th
Annual Internat. Mtg., Soc. Expl. Geophys., Expanded Abstracts, , 95, 518-520.
Gallagher, J. W., Kumpas, M. G. and Kemp, A. C., 1996, Seismic attribute analysis - A prestack and
poststack interpretation tool: 61st Mtg. Eur. Assoc. Expl Geophys., Extended Abstracts, European
Association Of Geophysical Exploration, Session:P036.
Ganley, D.C., 1981, A method for calculating synthetic seismograms which include the effects of absorption and
dispersion: Geophysics, 46, 1100-1107.
Ganley, D.C., and Kanasewich, E.R., 1980, Measurement of absorption and dispersion from check shot surveys:
Journal of Geophysical Research, 85, 5219-5226.
37
Goldman, S., 1948, Frequency analysis, modulation and noise: McGraw-Hill Book Co., 434.
Grace, O.D., 1981, Instantaneous power spectra: J. Acoust. Soc. Am., 69(1), 191.
Ha, S.T.T., Sheriff, R.E., and Gardner, G.H.F., 1991, Instantaneous frequency, spectral centroid, and even
wavelets: Geophys. Res. Lett., 18, 1389-1392.
Hamilton, E.L., 1972, Compressional-wave attenuation in marine sediments: Geophysics, 37, 620-646.
Hardage, B. A., Carr, D. L., Lancaster, D. E., Simmons, J. L., Jr., Hamilton, D. S., Elphick, R. Y., Oliver, K. L.
and Johns, R. A., 1996, 3-D seismic imaging and seismic attribute analysis of genetic sequences deposited in
low-accommodation conditions : Geophysics, 61, no. 05, 1351-1362. (* Discussion and reply in GEO-62-6-
1996)
Haskell, N. L., Nissen, S. E., Lopez, J. A. and Bahorich, M. S., 1995, 3-D seismic coherency and the imaging of
sedimentological features: 65th Annual Internat. Mtg., Soc. Expl. Geophys., Expanded Abstracts, , 95, 1532-
1534.
Hauge, P.S., 1981, Measurement of attenuation from vertical seismic profiles: Geophysics, 46, 1548-1558.
Heggland, R., 1995, Detection of ancient morphology and potential hydrocarbon traps using 3-D
seismic data and attribute analysis: 65th Annual Internat. Mtg., Soc. Expl. Geophys., Expanded
Abstracts, , 95, 316-318.
Hesthammer, J. and Fossen, H., 1997, The influence of seismic noise in structural interpretation of
seismic attribute maps: First Break, 15, no. 6, 209-219.
Hocker, C. and Dalley, D., 1997, Structural vector field - A volume attribute for seismic interpretation: 61st Mtg.
Eur. Assoc. Expl Geophys., Extended Abstracts, European Association Of Geophysical Exploration,
Session:B025.B025.
Hoetz, H. L. J. G. and Watters, D. G., 1992, Seismic horizon attribute mapping for the Annerveen gasfield, The
Netherlands: First Break, 10, no. 02, 41-51.
Hooge, C. D., Hajnal, Z. and Schreiner, B., 1994, High-resolution reflection imaging of a groundwater aquifer in
glacial tills, central Saskatchewan: 64th Annual Internat. Mtg., Soc. Expl. Geophys., Expanded Abstracts, , 94,
620-623.
Jannsen, D., Voss, J., and Thielen, F., 1983, Comparison of methods of determine Q in shallow marine
sediments from vertical reflection seismograms: Geophys. Prosp., 33, 479-497.
Johnston, D. H., 1993, Seismic attribute calibration using neural networks: 63rd Annual Internat. Mtg., Soc.
Expl. Geophys., Expanded Abstracts, , 93, 250-253.
Johnston, D.H. and Toksoz, M.N., 1981, Definitions and terminology, in Toksoz, M.N. and Johnston, D.H.,Eds.,
Seismic Wave Attenuation: Society of Exploration Geophysics, Geophysics Reprint Series no. 2, 1-5.
Johnston, D.H., and Toksoz, M.N., 1980, Ultrasonic P- and S-wave attenuation in dry and saturated rocks under
pressure: Journal of Geophysical Research, 85, 925-936.
38
Jones, G., and Boashash, B., 1990, Instantaneous frequency, instantaneous bandwidth, and the analysis of
multicomponent signals: Proc. IEEE ICASSP-90, 2467-2470.
Jones, I. F., Baud, H. and Strachan, A., 1999, Velocity as an Attribute - Continuous Velocity
Estimation from PreSDM CRP Gathers: 61st Mtg. Eur. Assoc. Expl Geophys., Extended Abstracts,
European Association Of Geophysical Exploration, Session:1042.
Kallweit, R.S., and Wood, L.C., 1982, The limits of resolution of zerophase wavelets: Geophysics, 47,
1035-1046.
Kemp, L. F., Threet, J. R. and Veezhinathan, J., 1992, A neural net branch and bound seismic horizon tracker:
62nd Annual Internat. Mtg., Soc. Expl. Geophys., Expanded Abstracts, , 92, 10-13.
Kerr, J. D. and Jones, G. L., 1995, Time-slice verses statistical slice (stat-slice): A comparison of their use for 3-
D seismic interpretation: 65th Annual Internat. Mtg., Soc. Expl. Geophys., Expanded Abstracts, , 95, 319-321.
Kjartansson, E., 1979, Constant Q-wave propagation and attenuation: J. Geophys. Res., 84, 4737-4748.
Knize, S., 1991, Seismic attributes in borehole full waveform sonic logging: 53rd Mtg. Eur. Assoc. Expl
Geophys., Abstracts, , 91, 178-179.
Krehbiel, S. C., 1999, Trivariate attribute displays: The Leading Edge, 18, no. 11, 1248, 1250, 1252-
1253, 1256-1257.
Kreyszig, E., 1072, Advanced engineering mathematics: John Wiley & Sons, Inc.
Lefeuvre, F. and Chanet, A., 1993, Reservoir characterization: A seismic attributes approach: 63rd Annual
Internat. Mtg., Soc. Expl. Geophys., Expanded Abstracts, , 93, 289-293.
Lefeuvre, F. E., Wrolstad, K. H., Zou, K. S., Smith, L. J., Maret, J-P. and Nyein, U. K., 1995, Sand-
shale ratio and sandy reservoir properties estimation from seismic attributes: An integrated study: 65th
Annual Internat. Mtg., Soc. Expl. Geophys., Expanded Abstracts, , 95, 108-110.
Lefeuvre, F., 1994, Fracture related anisotropy detection and analysis: and if the P-waves were enough? : 64th
Annual Internat. Mtg., Soc. Expl. Geophys., Expanded Abstracts, , 94, 942-945.
Lendzionowski, V., Walden, A.T., and White, R.E., 1990, Seismic character mapping over reservoir intervals:
Geophys. Prosp., 38, 951-969.
Leslie, R. B., 1994, Digital image processing of 3-D seismic attributes benefits exploration and production
interpreters: 64th Annual Internat. Mtg., Soc. Expl. Geophys., Expanded Abstracts, , 94, 813.
Levin, M. J., 1964, Instantaneous spectra and ambiguity functions: IEEE Trans. Information Theory, IT-10, 95-
97.
Lewis, C., 1995, Seismic attributes for reservoir monitoring: A feasibility study using forward modeling: 65th
Annual Internat. Mtg., Soc. Expl. Geophys., Expanded Abstracts, , 95, 309-312.
Li, X. and Ulrych, T. J., 1996, Multi-scale attribute analysis and trace decomposition: Annual Meeting Abstracts,
Society Of Exploration Geophysicists, 1634-1637.
39
Lundquist, G.M., and Cormier, V.C., 1980, Constraints on the absorption band model of Q: Journal of
Geophysical Research, 85, 5244-5256.
Luo, Y., and Schuster, G. T., 1992, Wave packet transform and data compression: 62nd Annual Internat. Mtg.,
Soc. Expl. Geophys., Expanded Abstracts, , 92, 1187-1190.
MacKay, S. and Abma, R., 1993, Incorporating wavefront curvature as a new attribute for depth- focusing
analysis: 55th Mtg. Eur. Assoc. Expl Geophys., Abstracts, , 93, Session:B036.
Mandel, L., 1974, Interpretation of instantaneous frequencies: Am. J. of Physics, 42, 840-846.
Mann, J., Hocht, G., Jager, R. and Hubral, P., 1999, Common-Reflection-Surface Stack - an Attribute
Analysis : 61st Mtg. Eur. Assoc. Expl Geophys., Extended Abstracts, European Association Of
Geophysical Exploration, Session:P140.
Marfurt, K. J., Scheet, R. M., Sharp, J. A. and Harper, M. G., 1998, Suppression of the acquisition footprint for
seismic sequence attribute mapping: Geophysics, 63, no. 03, 1024-1035.
Marfurt, K. J., Scheet, R. M., Sharp, J. A., Cain, G. J. and Harper, M. G., 1995, Suppression of the
acquisition footprint for seismic sequence attribute mapping: 65th Annual Internat. Mtg., Soc. Expl.
Geophys., Expanded Abstracts, , 95, 949-952.
Matteucci, G., 1996, Seismic attribute analysis and calibration: A general procedure and a case study: Annual
Meeting Abstracts, Society Of Exploration Geophysicists, 373-376.
Mazzotti, A., 1991, Amplitude, phase and frequency versus offset applications: Geophys. Prosp., 39, 863-886.
McCann, C., and McCann, D.M., 1969, The attenuation of compressional waves in marine sediments:
Geophysics, 34, 882-892.
McClarley, L.A., 1985, An autoregressive filter model for constant Q attenuation: Geophysics, 50, 749-758.
Meunier, J., Denis, M., Baliguet, P. and Burger, P., 1996, Quality control and attribute analysis on a 3-D
Vibroseis crew: Annual Meeting Abstracts, Society Of Exploration Geophysicists, 1067-1068.
Morlet,J., Arens, G., Fourgeau, E., and Giard, D., 1982, Wave propagation and Sampling theory- Part II:
Sampling theory and complex waves: Geophysics, 47, 222-236.
Morozov, I.B. and Smithson, S.B., 1996, Instantaneous polarization attributes and directional filtering: 61, 3, no.
872-881,
Morris, R. and MacGregor, M. D., 1991, An examination of amplitude vs. offset attributes of a middle Devonian
carbonate bank margin at the Caroline giant gas pool, Alberta, Canada: 61st Annual Internat. Mtg., Soc. Expl.
Geophys., Expanded Abstracts, , 91, 1058-1061.
Müller, B. and Van Den Berg, C., 1996, Color-coded crossplotting of AVO attributes: 15, 11, no. 1259-1260, .
Nelson, H. R. Jr., Mastoris, S. and Huxohl, C., 1991, Visualization of map and seismic attributes: 53rd Mtg. Eur.
Assoc. Expl Geophys., Abstracts, , 91, 86-87.
40
Nester, D. C. and Duncan, P. M., 1994, Target characterization by seismic attribute analysis--A case history
from West Texas: 64th Annual Internat. Mtg., Soc. Expl. Geophys., Expanded Abstracts, , 94, 1672-1673.
Newman, P.J., and Worthington, M.H., 1982, In-situ investigation of seismic body wave attenuation in
heterogeneous media: Geophys. Prosp., 30, 377-400.
Nickerson, R. and Cambois, G., 1998, Interpreter's Corner - AVO attribute analysis on marginal 3-D land data
improved target selection in the Sacramento Basin: The Leading Edge, 17, no. 12, 1672, 1674-1677.
Nissen, S. E., Haskell, N. L., Lopez, J. A., Donlon, T. J. and Bahorich, M. S., 1995, 3-D seismic
coherency techniques applied to the identification and delineation of slump features: 65th Annual
Internat. Mtg., Soc. Expl. Geophys., Expanded Abstracts, , 95, 1535-1536.
O'Brien, P.N.S., and Lucas, A.L., 1971, Velocity dispersion of seismic waves: Geophys. Prosp., 19, 1-26.
O'Doherty, R.F., and Taner, M.T.,1992, A method of computing Instantaneous frequency and dip: Presented at
the 54th Annual Meeting of E.A.E.G. in Paris, France.
Oppenheim, A.V., and Schafer, R.W., 1975, Digital Signal Processing: Prentice-Hall Inc.
Ormerod, D., 1995, Attempting to define a subtle trap by seismic attribute mapping of thin sands - a
Timor Sea example: 12th Aust. Soc. Expl. Geophys. Conf., Expl. Geophys., 26, 379-382.
Ortmann, K. A. and Wood, L. J., 1995, Successful application of 3-D seismic coherency models to predict
stratigraphy, offshore eastern Trinidad: 65th Annual Internat. Mtg., Soc. Expl. Geophys., Expanded Abstracts, ,
95, 101-103.
Papoulis, A., 1984, Probability, random variables, and stochastic processes: McGraw-Hill Book Co.
Parks, T.W., and Burrus, C.S., 1987, Digital filter design: Join Wiley & Sons, Inc.
Pearson, R. and Hart, B., 1999, Convergence of 3-D seismic attribute-based reservoir property prediction and
geologic interpretation as a risk reduction tool: A case study from a Permian intraslope basin: Annual Meeting
Abstracts, Society Of Exploration Geophysicists, 896-899.
Possato, S., Saito, M., Curtis, M. P. and Martinez, R. D., 1983, Interpretation of three-dimensional seismic
attributes contributes to stratigraphic analysis of Pampo oil field: 53rd Annual Internat. Mtg., Soc. Expl.
Geophys., Expanded Abstracts, , 83, Session:S16.2.
Priestley, M.B., 1965, Evolutionary spectra and non-stationary process: J. R. Stat. Soc. Ser. B27, 204-
229.
Raeuchle, S. K., Hamilton, D. S. and Uzcategui, M., 1997, Integrating 3-D seismic imaging and seismic attribute
analysis with genetic stratigraphy: Implications for infield reserve growth and field extension, Budare field,
Venezuela: The Leading Edge, 16, no. 09, 1338-1339.
Raikes, S.A., and White, R.E., 1984, Measurements of earth attenuation from downhole and surface seismic
recordings: Geophys. Prosp., 32, 892-919.
Ratliff, T. L. and Watkins, J. S., 1989, Characterization of the L-1 sand using well logs and amplitude attribute
analysis: 59th Annual Internat. Mtg., Soc. Expl. Geophys., Expanded Abstracts, , 89, 745.
41
Rietveld, W. E. A., Marfurt, K. J. and Kommedal, J. H., 1997, The effect of 3-D prestack seismic migration on
seismic attribute analysis: Annual Meeting Abstracts, Society Of Exploration Geophysicists, 1367-1370.
Rihaczek, A.W., 1968, Signal energy distribution in time and frequency: IEEE Info. Theory, IT-14, 369.
Rijks, E. J. H. and Jauffred, J. C. E. M., 1991, Attribute extraction: An important application in any
detailed 3-D interpretation study: The Leading Edge, 10, no. 9, 11-19.
Rijks, E. J. H., 1990, Attribute extraction: An important application in any detailed 3-D interpretation study:
Annual Meeting Abstracts, Society Of Exploration Geophysicists, 774.
Robertson, J. D. and Fisher, D. A., 1987, Complex seismic trace attributes, in Brown, Alistair R., Ed., Seismic
interpretation series, 1: Soc. Expl. Geophys., 49-53. (* Reprinted from The Leading Edge)
Robertson, J. D. and Fisher, D. A., 1988, Seismic interpretation 9 - Complex seismic trace attributes:
The Leading Edge, 7, no. 6, 22-26.
Robertson, J. D. and Nogami, H. H., 1981, Thin bed stratigraphy from complex trace attributes, 51st Annual
Internat. Mtg., Soc. Expl. Geophys., Reprints, 81: , Session:S19.2.
Robertson, J.D., and Fisher, D.A., 1988, Complex seismic trace Attributes: The Leading Edge, 7, no. 6, 22-26.
Robertson, J.D., and Nogami, H.H., 1984, Complex seismic trace analysis of thin beds: Geophysics, 49, 344-
352.
Ronen, S., Hattori, M., Geoltrain, S., and Schultz, P., 1992, Lateral prediction of reservoir properties using
seismic attributes with well log data: 54th Mtg. Eur. Assoc. Expl Geophys., Abstracts, , 92, 18-19.
Ronen, S., Hoskins, J., Schultz, P. S., Hattori, M. and Corbett, C., 1994, Seismic-guided estimation of log
properties, part 2: Using artificial neural networks for nonlinear attribute calibration *: The Leading Edge, 13,
no. 6, 674-678. (* Erratum in TLE-13-8-822)
Santoso, D., Alam, S., Hendraya, L., Alfian., S., and Munadi, S., 1995, Estimation of limestone reservoir
porosity by seismic attribute and AVO analysis: 12th Aust. Soc. Expl. Geophys. Conf., Expl. Geophys., 26, 437-
443.
Scheuer, T. E., and Oldenburg, D. W., 1988, Local phase velocity from complex seismic data: Geophysics, 53,
1503-1511.
Schoenberger, M., and Levin, F.K., 1974, Apparent attenuation due to intrabed multiples: Geophysics, 39, 278-
291.
Scholtz, P., 1996, Coherence calculation of different seismic attribute traces for fault interpretation: 61st Mtg.
Eur. Assoc. Expl Geophys., Extended Abstracts, European Association Of Geophysical Exploration,
Session:P065.
Schroot, B. M., Stavenga, T. and Duin, E. J. T., 1997, Modelling Zechstein interval velocities using a 3D seismic
attribute: 61st Mtg. Eur. Assoc. Expl Geophys., Extended Abstracts, European Association Of Geophysical
Exploration, Session:B015.B015.
42
Schuelke, J. S., Quirein, J. A., Sarg, J. F., Altany, D. A. and Hunt, P. E., 1997, Reservoir architecture and
porosity distribution, Pegasus field, West Texas -- an integrated sequence stratigraphic -- seismic attribute study
using neural networks: Annual Meeting Abstracts, Society Of Exploration Geophysicists, 668-671.
Sengupta, M. K. and Peddy, C. P., 1995, AVO analysis for the high-impedance gas sands: Improved
visualization, and rock-property estimation: 65th Annual Internat. Mtg., Soc. Expl. Geophys., Expanded
Abstracts, , 95, 581-584.
Sheriff, R.E., 1991, Encyclopedic dictionary of exploration geophysics, third edition: Soc. Expl. Geophys.
Sheriff, R.E., and Geldart, L.P., 1989, Exploration seismology volume 1: History, theory, and data acquisition:
Cambridge University Press.
Shumway, G., 1960, Sound speed and absorption studies of marine sediments by resonance method, part II:
Geophysics, 25, 659-682.
Signer, C., Kvia, P., Randen, T., Saeter, T., Schlaf, J. and Sonneland, L., 1999, Novel Seismic Attribute Analysis
Procedure for Reservoir Characterisation : 61st Mtg. Eur. Assoc. Expl Geophys., Extended Abstracts, European
Association Of Geophysical Exploration, Session:5011.
Smith, T. A. and Neidell, N. S., 1990, Improved seismic resolution of stratigraphically complex reservoirs
through modeling and color displays: 60th Annual Internat. Mtg., Soc. Expl. Geophys., Expanded Abstracts, , 90,
357.
Smith, W. H., Reeves, J. J. and Bacon, S. D., 1997, The geostatistical mapping of effective porosity utilizing
petrophysics and seismic attribute analyses, Grayburg and San Andres shallow shelf carbonate reservoirs,
Permian basin, USA: Annual Meeting Abstracts, Society Of Exploration Geophysicists, 734-736.
Sonneland, L., Barkved, O. and Hagenes, O., 1990, Construction of reservoir maps from seismic classifier maps:
60th Annual Internat. Mtg., Soc. Expl. Geophys., Expanded Abstracts, , 90, 241-244.
Sonneland, L., Barkved, O., Olsen, M. and Snyder, G., 1989, Application of seismic wave-field attributes in
reservoir characterization: 59th Annual Internat. Mtg., Soc. Expl. Geophys., Expanded Abstracts, , 89, 813.
Spencer, T.W., Sonnad, J.R., and Butler, T.M., 1982, Seismic Q-Stratigraphy or dissipation: Geophysics, 47, 16-
24.
Steeghs, P., 1999, Fast computation of the sliding-window radon transform applied to 3-D seismic
attribute extraction: Annual Meeting Abstracts, Society Of Exploration Geophysicists, 1146-1149.
Steeghs, P., Fokkema, J. T. and Diephuis, G., 1998, 3-D local radon power spectra for seismic attribute
extraction: Annual Meeting Abstracts, Society Of Exploration Geophysicists, 645-648.
Steeghs, T. P. H., Drijkoningen, G. G., Peet, W. E. and Fokkema, J. T., 1995, A New Method for the Extraction
of Seismic Facies Attributes: 57th Mtg. Eur. Assoc. Expl Geophys., Extended Abstracts, , 95, Session:A029.
Steentoft, H. and Wende, S., 1991, 3-component colored 'attribute seismograms': 53rd Mtg. Eur. Assoc. Expl
Geophys., Abstracts, , 91, 468-469.
43
Strick, E., 1967, The determination of Q, dynamic viscosity and transient creep curves from wave propagation
measurements: Geophysical Journal of the Royal Astronomical Society, 13, 197-218.
Talley, D., Davis, T. L. and Benson, R. D., 1997, San Andres carbonate reservoir characterization from four-
dimensional, multi-component seismic attribute analysis: Annual Meeting Abstracts, Society Of Exploration
Geophysicists, 883-885.
Taner, M. T., Schuelke, James S., O'Doherty, Ronen and Baysal, Edip, 1994, Seismic attributes revisited: 64th
Annual Internat. Mtg., Soc. Expl. Geophys., Expanded Abstracts, , 94, 1104-1106.
Taner, M.T., Koehler, F., and Sheriff, R.E., 1979, Complex seismic trace analysis: Geophysics, 44, 1041-1063.
Tatham, R.H., and McCormack, M.D., 1991, Multicomponent seismology in petroleum exploration: Soc. Expl.
Geophys.
Thomas, R. D., Castoro, A. and White, R. E., 1998, QC of Prestack Seismic Processing and Attribute Extraction
: 61st Mtg. Eur. Assoc. Expl Geophys., Extended Abstracts, European Association Of Geophysical Exploration,
Session:02-54.
Thomas, R., White, R. and Castoro, A., 1998, Quantitative controls on 3-D seismic processing for prestack
attribute analysis: The Leading Edge, 17, no. 11, 1596-1599.
Thon, K. T. and Reksnes, P. A., 1993, Troll west - Seismic attributes and optimal locations of horizontal wells in
a thin oil leg: 55th Mtg. Eur. Assoc. Expl Geophys., Abstracts, , 93, Session:A010.
Tilbury, L. A. and Bush, D., 1991, Image processing of interpreted 3-D seismic data to enhance subtle structural
features/lineations: 8th Aust. Soc. Expl. Geophys. Conf., Expl. Geophys., 22, 391-396.
Toksöz, M.N., and Johnston, D.H., 1981, Seismic wave attenuation: Soc. Expl. Geophys.
Toksöz, M.N., Johnston, D.H., and Timur, A., 1979, Attenuation of flexural waves in dry and saturated rocks. I:
Laboratory measurements: Geophysics, 44, 681-190.
Tonn, R., 1989, Comparison of seven methods for the computation of Q: Phys. Earth Plan. Int., 55, 259-269.
Tonn, R., 1991, The determination of the seismic quality factor Q from VSP data: A comparison of different
computational methods: Geophys. Prosp., 39, 1-27.
Tonn, R., 1992, Statistical approach to correlate reservoir parameters and 3-D seismic attributes: 62nd Annual
Internat. Mtg., Soc. Expl. Geophys., Expanded Abstracts, , 92, 272-274.
Urosevic, M., Evans, B. J. and Hatherly, P. J., 1992, The improvement in seismic resolution by map and trace
attribute analysis: ASEG 13th Geophysical Conference, Australian Society Of Exploration Geophysicists, 23,
387-391.
Urosevic, M., Evans, B. J. and Hatherly, P. J., 1994, New developments for coal exploration in Australia: 64th
Annual Internat. Mtg., Soc. Expl. Geophys., Expanded Abstracts, , 94, 537-539.
44
Vidaillac, Th, Blanche, J. P., Nguyen, T., Marrauld, J., Capelle, P. and Grinde, P., 1993, How AVO attributes
can improve seismic calibration and interpretation on the Heimdal gasfield (Norway): 55th Mtg. Eur. Assoc.
Expl Geophys., Abstracts, , 93, Session:A014.
Ville, J., 1948, Théorie et applications de la notion du signal analytique: (in French), Cables et Transmission, 2,
61-74.
Vossier, D. A., 1989, Automatic delineation of lateral facies changes in clastic environments: 59th Annual
Internat. Mtg., Soc. Expl. Geophys., Expanded Abstracts, , 89, 803.
White, J. E, and Walsh, D.J., 1972, Proposed attenuation-dispersion pair for seismic waves: Geophysics, 37,
456-461.
White, R. E., 1991, Properties of instantaneous seismic attributes *: The Leading Edge, 10, no. 7, 26-
32. (* Discussion and reply in TLE-11-8-45-46 and TLE-11-10-10-12)
Yang, C. H. and Hsiuan, T. H., 1990, Identification of seismic anomalies using seismic attributes and
reconstructed images: 60th Annual Internat. Mtg., Soc. Expl. Geophys., Expanded Abstracts, , 90, 226-228.
Zhang, X. L., Yanda, H. Q. and Wang, L., 1995, Estimating reservoir's lithological parameters from
seismic data using neural network: 65th Annual Internat. Mtg., Soc. Expl. Geophys., Expanded
Abstracts, , 95, 606-608.
Zhang, X. L., Yanda, H. Q. and Feng, D., 1995, Early-stage reservoir analysis with SOMA: A neural network
approach: 65th Annual Internat. Mtg., Soc. Expl. Geophys., Expanded Abstracts, , 95, 138-141.
Zimmerman, C. J. and Duran, R. E., 1992, Using the seismic range equation for AVO analysis to detect low gas
saturation: 62nd Annual Internat. Mtg., Soc. Expl. Geophys., Exp
45