You are on page 1of 33

An Introduction to Seismic Interpretation

Chapter 2: Principles of the Seismic Method

About this Disc

Disc Contents (PDF format)

Show/Hide Bookmarks

Previous Page | Next Page

Search

Print

Close

Navigation
Show Hide

2.1 Overview
Conventional reflection seismic technology uses acoustic waves (sound) to image the subsurface.
Conceptually, as shown in FIGURE 2.1, we begin by generating a bang at the earths surface.
Following the bang, the sound travels down into the earth where some of it is reflected at buried
interfaces where there is a change in a physical property known as the acoustic impedance. As
described in detail later in this chapter (SECTION 2.4) acoustic impedance is a function of a
materials P-wave velocity (i.e., how fast sound travels through the material) and its density. The
acoustic energy reflected at those interfaces is recorded at the surface as echoes. The distance
from the surface to the buried horizons that generated the reflections is measured in units of time
the two-way traveltime (TWT) being a measure of how long it took for the energy to pass from
the surface source to the reflecting horizon and return back to a receiver located at the surface. The
strength (amplitude) of the reflection is also recorded.

Usually we are interested in knowing how deep (in meters or feet) an interface is, rather than how
long it takes for sound to travel to it and back to the surface. If we know the velocity of sound in the
propagating medium we can derive true depths using a simple equation:
D=VxT

[2.1]

Where D is the distance (depth) to a horizon, V is the average velocity and T is the time it takes for
the sound to travel from the surface to the horizon (a one-way traveltime).
Most people have an intuitive understanding of how Equation 1 works. For example, if we know
that it takes approximately 2.5 hours to drive between City A and City B, and we know that our
average velocity is 120 km/h, we can easily calculate that the distance between those two points is
approximately 300 km1. Likewise, suppose we know that the two-way traveltime to a reflection is
2.302 seconds, and we know that the average velocity was 2022 m/s 2. We calculate the one-way
__________________________________________________________________
1

Of course, our velocity may not have been constant for the entire length of the voyage. For example, we may have
sped up or slowed down because of traffic or (perhaps) the presence of law-enforcement officers.

Again, the velocity may be variable over the path traveled by the sound.

Page 1

About this Disc

Disc Contents (PDF format)

Show/Hide Bookmarks

Previous Page | Next Page

Search

Print

Close

Navigation

An Introduction to Seismic Interpretation Chapter Two Principles of the Seismic Method


traveltime as 1.151 seconds, and subsequently the depth to
the horizon that generated the reflection is 2327 m. This
is the principle used during conventional bathymetric
profiling. A ship-borne source emits acoustic waves
(pings) that travel through the water column, reflect off
the seafloor, and are recorded at the surface. A constant
water velocity (e.g., 1450 - 1500 m/s) is then used to
convert the two-way traveltime to a water depth. In world
of reflection seismology, one of our challenges will be to
determine how fast the sound has traveled in the subsurface
between source and receiver because, as described below,
velocity usually changes vertically (i.e. with depth) and
laterally.
FIGURE 2.2 shows a simple example of how seismic
imaging (i.e. vertical axis in units of time) can distort our
view of the subsurface geology. In this example, the water
depth changes from one side to the other. The velocity of
sound in water in this instance is 1450 m/s, and the velocity
of the rocks below the seafloor is 2500 m/s. A horizontal
surface, Horizon A, 500 m below the seafloor appears to
bow down to the right in the seismic data as the water gets
deeper. Clearly the seismic image is not showing the true
subsurface structure. The fundamental problem is that the
seismic image shows how long it takes for sound to travel
from the surface to a horizon and be reflected back up to
the surface. On the left, the relatively slow water is 50 m
deep whereas on the right the water depth is 200 m. As
such, it takes the sound longer for sound to travel down
to Horizon A and back on the right than on the left. We
will return to velocity-related imaging problems in later
chapters.
In practice, reflection seismology is more complicated than
the simple cases described above. The optimal source of
acoustic energy for the study being performed needs to be
determined, there are many interfaces in the subsurface
that generate reflections, velocities change laterally and
vertically, and we need to find cost-effective ways to collect
2-D seismic profiles or 3-D seismic volumes. Practical
aspects of seismic acquisition and processing are described
in CHAPTER 3. In this chapter the focus is on the
physical basis of the seismic method, including topics such
as the controls on the velocity at which sound propagates,
the strength of the reflection, and seismic resolution.
We can represent the paths of the energy generated by our

Page 2

FIGURE 2.1:
Schematic representation of
how sound is used to image
the subsurface

FIGURE 2.2:
Simple example of how
seismic imaging can distort
our view of the subsurface
geology

Show Hide

About this Disc

Disc Contents (PDF format)

Show/Hide Bookmarks

Previous Page | Next Page

Search

Print

Close

Navigation

An Introduction to Seismic Interpretation Chapter Two Principles of the Seismic Method


bang in two different ways (Figures 2.1, 2.3). The first
is to depict a wavefront that expands out from the energy
source. A commonly used analog is to consider the ripples
that spread out on the surface of a pond when a pebble is
dropped into the water. The ripples are the wavefront that
expands away from the energy source. A second way to
depict the expansion of energy away from the source is
to show raypaths that connect the source point with the
wavefront, i.e. the direction of energy propagation. If the
velocity of wave propagation is the same in all directions,
the wavefront spreads out in a spherical form and the
raypaths are straight as shown in FIGURE 2.3A. When
the velocity varies in different directions (the usual case in
the real earth), the raypaths bend and the wavefront is no
longer spherical, as shown schematically in FIGURE 2.3B.
It is common practice to illustrate concepts of reflection
seismology using raypaths, partly because the mathematics
of raypaths are simpler than those of propagating
wavefronts, but also because raypaths are easier to draw.

FIGURE 2.3:
Raypaths connect the source
point with the wavefront,
depicting the expansion of
energy away from the source

2.2 Characteristics of Waves


There are several types of waves that can propagate through
the Earth or at its surface. As noted in the PREVIOUS
SECTION, most seismic datasets are collected using
acoustic waves, also known as compressional waves,
P-waves, or simply sound waves. Another type of
wave used in reflection seismology is a shear wave, also
known as an S-wave. Most geologists are introduced
to compressional and shear waves in the context of
earthquakes. Compressional waves are referred to as
primary or P-waves in earthquake seismology because they
are the first waves recorded by a seismograph following
an earthquake. The shear waves are the second waves
(secondary waves) recorded by a seismograph. Because
of the differences in traveltime, it should be clear that
compressional waves travel at a higher velocity than
shear waves. Furthermore, shear waves cannot propagate
through fluids (this is one of the lines of evidence that has
been used by geophysicists to determine that the earths
outer core is in a liquid state). Other types of waves
(e.g., surface waves known as Rayleigh waves) are also
generated by earthquakes and in reflection seismology,
however we will not discuss them here.
FIGURE 2.4 shows the particle motions induced in the
earth as P-waves and S-waves propagate through it. The
Page 3

FIGURE 2.4:
Representation of particle
deformation during the
passage of a compressional
wave (P-wave) and a shear
wave (S-wave)

Show Hide

About this Disc

Disc Contents (PDF format)

Show/Hide Bookmarks

Previous Page | Next Page

Search

Print

Close

Navigation

An Introduction to Seismic Interpretation Chapter Two Principles of the Seismic Method


passage of a compressional wave causes a portion of rock
to alternately contract and then dilate. The rock body
changes shape and volume. On the other hand, the shearing
motions associated with the passage of a shear wave cause
changes in shape, but not in volume.
This text focuses primarily on the use of P-wave seismic
methods because they are the most common type of wave
used in reflection seismology. It should be noted however
that: a) S-wave seismic methods are seeing increasing use,
and these uses will be discussed later (CHAPTER 8), and
b) the geologic interpretation of S-wave seismic data has
many similarities to P-wave interpretation. P-wave and
S-wave seismic data are sometimes recorded together as
multicomponent seismic data.
Waves, seismic or otherwise, can be described in
various ways. A simple cosine, although not necessarily
representative of an actual wave used in reflection
seismology, can be used to help define these terms
(FIGURE 2.5). The wave consists of changes in amplitude
that can be represented by a series of positive and negative
values. A peak represents a local maximum amplitude
value and a trough represents a local minimum value.
As seismic waves travel through the earth, they have a
physical length (meters or feet; FIGURE 2.5A). The
wavelength is the distance between successive repetitions
of the waveform (e.g., the distance between two successive
troughs or two successive peaks). Note that the cosine
curve shown in the figure is a repeating waveform.
A different way of showing a wave is to show how
amplitudes change at a given point through time. For
example, imagine that you are standing in the surf zone of
a beach, and that the waves generated offshore pass you on
their way to the beach. You could record changes in the
water surface elevation as a function of time. If the waves
are cosine curves (an unlikely situation in a surf zone),
then the plot of amplitude (water surface elevation) versus
time might look like the one shown in FIGURE 2.5B. In
this case, the separation (in time) between successive wave
crests is known as the wave period (T).
The key variables for defining a waves shape are its
amplitude, frequency and phase. For example, the equation
for a cosine curve can be expressed:

Page 4

FIGURE 2.5:
A simple cosine representing
traveling waves in distance
(A) and time (B)

FIGURE 2.6:
Examples showing the effects
of changing the amplitude,
frequency and phase of a
cosine curve

Show Hide

About this Disc

Disc Contents (PDF format)

Show/Hide Bookmarks

Previous Page | Next Page

Search

Print

Close

Navigation

An Introduction to Seismic Interpretation Chapter Two Principles of the Seismic Method


f(x) = a cos(bx + c)

Show Hide

[2.2]

where: a controls the amplitude of the curve, b controls the frequency, and c the phase of the curve.
Examples showing the effects of changing the amplitude, frequency and phase of a cosine curve are
shown in FIGURE 2.6. The cosine curves shown in FIGURE 2.5 are monochromatic, i.e. they
consist of only one frequency.

Frequency Content of Seismic Data


The frequency (F) of a wave train is the number of cycles (e.g., peaks) that pass by a given point
in a given time. Frequency is measured in cycles/second or Hertz (abbreviated as Hz). For example,
40 Hz describes a case where 40 cycles (e.g., peaks) will pass by a given point in one second. The
frequency can be expressed as the inverse of the period:

1
T

[2.3]

For example, a period of 0.05 seconds corresponds to a frequency of 20 Hz. Typical petroleumindustry seismic data have frequencies that are in the 10s of Hertz range (e.g., 10 80 Hz). Shallow
land surveys might have data in the 100s of Hz range, whereas some high-resolution marine profiling
systems use frequencies that extend up into the kHz range.
A key relationship for seismic interpreters relates the wavelength () to the wave propagation
velocity and the frequency:

V /F

[2.4]

Two examples show how this equation might be used. Assume that the P-wave velocity is 5000 m/s
and that the frequency is 50 Hz. The wavelength will be 5000/50 = 100 m. Now assume that the
velocity remains 5000 m/s but the frequency is now 25 Hz. The wavelength will be 5000/25 = 400
m. As discussed below, the wavelength places fundamental limits on seismic resolution and the limit
of seismic detectability.
Seismic wavelets do not repeat like the cosine curves presented in FIGURE 2.5. The simplest
way of thinking about a seismic wavelet is to consider it to be the acoustic pulse that was generated
by the seismic source (FIGURE 2.1). Simm and White (2002) discuss various aspects of seismic
wavelets. A more realistic seismic wavelet, but still highly simplified, is shown in FIGURE 2.7A.
Note that this wavelet does not repeat itself. Furthermore, this wavelet contains a range of
frequencies, rather than just one like the monochromatic waves of FIGURE 2.5. FIGURE 2.7B
shows how several frequencies can be combined to generate a wavelet such as the one shown in part
A of that figure. Four different cosine curves are shown in FIGURE 2.7B, all four having identical
amplitudes, but different frequencies. Additionally, in all four cases the location of the central peak
amplitude is the same (i.e., they all have the same phase). Summation of those four curves produced
the wavelet shown in part A. In the middle of the wavelet, the positive values add constructively to
generate a strong central peak. On either side of that peak, positive and negative values are being
summed, and they tend to cancel each other out, especially as the distance from the central peak
increases. Non-perfect cancelling leads to the generation of side lobes. FIGURE 2.7B only shows
four discrete frequencies being added. In practice, the broader the range of frequencies being added

Page 5

About this Disc

Disc Contents (PDF format)

Show/Hide Bookmarks

Previous Page | Next Page

Search

Print

Close

Navigation

An Introduction to Seismic Interpretation Chapter Two Principles of the Seismic Method


(i.e., the broader the bandwidth), the sharper the central
peak will be and the smaller the side lobes. The importance
of reducing the side lobes will be discussed later.
The world of music provides an analogy for how a wavelet
might contain a range of frequencies. Imagine a small
child playing with a piano. Using one finger, the child
can play one note (frequency) at a time, including high
notes (high frequencies), low notes (low frequencies) and
notes (frequencies) in the middle. Then the child uses her
forearm to smash on 15 piano keys at once. The child has
produced a wave containing several frequencies that sounds
more like a bang than an individual note. In the same
way, the bang produced by an explosive source (airgun,
dynamite, etc.) creates a wavelet that consists of a broad
range of frequencies (e.g., 10 100 Hz).
As shown by EQUATION 2.4, the frequency content
of the seismic data partially controls (along with the
velocity) the wavelength. Although it is possible to directly
measure the frequency content of the source pulse, this
is not routinely done (except for vibroseis sources; see
CHAPTER 3). Furthermore, and for reasons described
below, the frequency content of seismic data generally
varies with depth (time). Interpreters therefore seek to
determine the frequency content of a portion of seismic
data in a variety of ways.
One method requires the interpreter to define a portion
of the seismic data for the analysis (e.g., a subset of the
seismic data defined by a range of two-way travel times
and areal extent; some seismic interpretation software
packages allow the user to choose this window by clicking
and dragging the mouse to define an analysis window). A
Fourier transform (FT) can be used to convert the input
data (amplitude versus time) into an amplitude spectrum
(amplitude versus frequency) such as the two examples
shown in FIGURE 2.8. The x-axis shows the frequency (in
Hz) and the y-axis shows a measure of the relative strength
of the signal at that frequency (commonly in either decibels
or power).
The amplitude spectrum in FIGURE 2.8A is from some
seismic data acquired on land using a dynamite source.
It shows a broad range of frequencies that extends from
approximately 10 to 100 Hz. The apparent presence of

Page 6

FIGURE 2.7:
Simplified wavelets
representing various aspects
of seismic data

FIGURE 2.8:
Amplitude spectra for two
different seismic data sets
(from two different seismic
interpretation packages)
showing the range of
frequencies in the data

Show Hide

About this Disc

Disc Contents (PDF format)

Show/Hide Bookmarks

Previous Page | Next Page

Search

Print

Close

Navigation

An Introduction to Seismic Interpretation Chapter Two Principles of the Seismic Method


frequencies below 8 or 10 Hz is a mathematical artefact
(those frequencies are generally either not recorded during
data acquisition or may be filtered out during processing
to remove a surface wave called ground roll) as is the
presence of frequencies above ~100 Hz. The amplitude
spectrum in FIGURE 2.8B is from a different land 3-D
seismic data set that has a much more restricted range of
useable frequencies, extending from approximately 15 40
Hz. Note too that the strength of the signal for different
frequencies is quite variable in that range3. In this case,
frequencies above 40 Hz were digitally removed (filtered
out) during data processing in an attempt to reduce noise
(i.e., processors thought that frequencies recorded above
this level were primarily related to noise, not useful seismic
reflections).

FIGURE 2.9:
Comparison of an idealized
amplitude spectrum and a real
amplitude spectrum

Because the seismic wavelet contains a range of


frequencies, rather than a single frequency, we need to be
able to characterize it using a single dominant frequency
in order to calculate the wavelength. As shown in
FIGURE 2.7A, the dominant frequency is represented by
the separation in time between the first side lobes. This
measure was called the wavelet breadth by Kallweit and
Wood (1982). The dominant frequency is used to calculate
the wavelength. One common way to estimate the
dominant frequency from an amplitude spectrum is to take
the peak value on an amplitude spectrum (Liner, 2004).
Alternatively, if the amplitude spectrum is relatively flat,
the maximum and minimum useable frequencies (referred
to as the upper terminal frequency fu - and lower terminal
frequency fl respectively by Kallweit and Wood,
1982, also known as low-pass and high-pass frequencies,
respectively) are estimated, and the dominant frequency is
given as the midpoint between those two values.
FIGURE 2.9A shows an idealized amplitude spectrum
and how the dominant frequency would be derived from
it. FIGURE 2.9B shows a real amplitude spectrum. The
definition of the dominant frequency becomes somewhat
more ambiguous in this case, because the midway point
between the low- and high-pass frequencies (approximately
35 Hz) does not correspond to the maximum signal strength
(approximately 42 Hz).
3.8 peaks(cycles) 10 38cycles
x
38 H z
100m s
10 1000m s
__________________________________________________________________
3

In the piano case, this would mean that the child is hitting some of the notes harder than others.

Page 7

Show Hide

About this Disc

Disc Contents (PDF format)

Show/Hide Bookmarks

Previous Page | Next Page

Search

Print

Close

Navigation

An Introduction to Seismic Interpretation Chapter Two Principles of the Seismic Method


Another way to estimate, this time by inspection, the
dominant frequency of some seismic data is to use a
procedure known as counting peaks. This quick-look
method does not require us to mathematically transform
the seismic data to the frequency domain via a Fourier
transform. Instead, the interpreter identifies a narrow
time interval in the data, perhaps 100 ms, and counts the
number of cycles (e.g., peaks) in that interval. The portion
of seismic data chosen for this analysis should be relatively
noise free and consist of relatively equally spaced peaks.
If a 100 ms time window is used, the dominant frequency
of the seismic data in that interval can be estimated by
multiplying that number by 10. For example, suppose 3.8
peaks are counted in a 100 ms interval at a location below
the red arrow in FIGURE 2.10 (the number of peaks will
not always be a whole number). We can convert this count
to an equivalent frequency by multiplying the numerator
and denominator by 10:

FIGURE 2.10:
Counting peaks to estimate
the dominant frequency

3.8 peaks(cycles) 10 38cycles


x
38 H z
100m s
10 1000m s

Finally, we might identify a single reflection in the data


and carefully measure the distance (in two-way traveltime)
between two successive peaks (or troughs). This will be an
estimate of the period at that point in the data. With this
information we can calculate the dominant frequency using
EQUATION 2.3. FIGURE 2.11A shows how this might
be done. The time separation between two peaks at the red
arrow in the image can be measured as approximately 14
ms. This estimate of the period can be used to estimate the
frequency at that level:

1
1

7 1H z
T 0.014

An amplitude spectrum derived from approximately the


same level is shown in FIGURE 2.11B. Note that the 71
Hz frequency derived from estimating the period is high,
when compared to the dominant frequency (approximately
50 Hz) that might be estimated using the maximum and
minimum useable frequencies, although it is closer to the
dominant frequency that would be estimated by picking
the peak amplitude in the spectrum. This is because the
spacing between two peaks is actually controlled by the
interaction between two variables: the seismic frequencies
and the spacing between beds. As such, estimates of
frequency content derived from too a narrow an analysis
Page 8

FIGURE 2.11:
Measuring wave period
to estimate the dominant
frequency

Show Hide

About this Disc

Disc Contents (PDF format)

Show/Hide Bookmarks

Previous Page | Next Page

Search

Print

Close

Navigation

An Introduction to Seismic Interpretation Chapter Two Principles of the Seismic Method


window (e.g., measuring the period of a single reflection)
may not be representative of the true frequency content of
the data.
These reservations aside, with time and practice, an
experienced interpreter can estimate the dominant
frequency of some seismic data visually with
approximately the same accuracy as can be derived from an
amplitude spectrum. However, the amplitude spectrum has
an advantage in that it provides information about the range
of frequencies in the data. From a practical perspective,
the dominant frequency estimated using any of these
methods is exactly that an estimate. As such, it might be
more honest to give a possible range of values (e.g., FD is
between 25 and 30 Hz) than to assume that an exact value
(e.g., 27.3 Hz) is more accurate.
The range of frequencies embedded in the seismic wavelet
is known as the bandwidth. Ideally, the seismic wavelet
contains a broad range of frequencies, including both low
and high frequencies. In this way, the pulse is sharper (i.e.
narrower) and wavelet sidelobes are reduced, as shown in
FIGURE 2.12. The useable bandwidth is typically defined
in Hz using the low- and high-pass frequencies (e.g., 10-90
Hz).
The wavelets used in FIGURE 2.12 are known as bandpass
wavelets. They are constructed mathematically, using
frequency filters, and hopefully they resemble the wavelets
that are embedded in our seismic data. Butterworth and
Ormsby wavelets are bandpass wavelets that are named
after the mathematicians who defined them. This type of
wavelet is defined using four frequencies, known as the
low-cut, low-pass, high-pass and high-cut frequencies,
as shown in FIGURE 2.9A. No energy is present below
the low-cut frequency or above the high-cut frequency.
All frequencies between the low-pass and high-pass
frequencies have equal amplitude in one of these wavelets
(although this is not necessarily true in actual seismic data,
e.g., FIGURE 2.8B), and amplitudes ramp up between the
low-cut and low-pass frequencies, and ramp down between
the high-pass and the high-cut frequencies.

Seismic Phase
The phase of the seismic wavelet is also a key variable to
a seismic interpreter. The effects of changing the phase
Page 9

FIGURE 2.12:
Effects of changing the
frequency content of a
wavelet on wavelet shape

Show Hide

About this Disc

Disc Contents (PDF format)

Show/Hide Bookmarks

Previous Page | Next Page

Search

Print

Close

Navigation

An Introduction to Seismic Interpretation Chapter Two Principles of the Seismic Method


of a cosine curve were displayed in FIGURE 2.6. For a
seismic wavelet, let us begin by considering two different
types of phase.
A minimum phase wavelet (FIGURE 2.13A) approximates
the type of wavelet generated by an airgun or dynamite
energy source. It is called a minimum phase wavelet
because the highest amplitude has a minimum delay
from the beginning of the wavelet (the reference point,
time = 0, at which we begin to generate the wave). After
that the amplitude oscillates a few times then it dies away
to nothing (Liner, 2004). Consider a bell when struck by a
hammer. The amplitude (volume) of the signal emitted by
the bell builds up rapidly as the hammer begins to hit it (t
= 0), eventually reaching a maximum. The bell rings after
the impact, with the amplitude of the ringing diminishing
with time.
Although minimum phase wavelets are wavelets that can be
produced using seismic sources, they have disadvantages
during the interpretation stage. One such disadvantage
is the presence of the oscillations following the initial
amplitude maximum. A reflection generated from a single
interface using a minimum phase wavelet might consist of
a peak (an amplitude maximum) followed by a series of
smaller peaks (FIGURE 2.14A). During the interpretation
phase, it might be tempting to interpret the trailing peaks
as weaker amplitude reflections. Alternatively, they might
interfere with reflections from underlying horizons. A
second disadvantage of a minimum phase wavelet is that
the delay between the maximum amplitude and the onset
of the wavelet (i.e. the reference point) is variable, and
depends on the frequency content of the wavelet. The
delay is greater for a low-frequency minimum phase
wavelet than for a high-frequency minimum phase wavelet
(FIGURE 2.15). From an interpretation perspective,
we would like the maximum amplitude to coincide with
the interface that generates the reflection. The delay,
and variability of the delay with frequency content, adds
unwanted ambiguity about where the interface truly is.
In contrast, a zero phase wavelet has a maximum, or a
minimum, amplitude (peak or trough respectively) that
is centered on the reference point (FIGURE 2.13B).
From an interpreters perspective, this means that the
interface generating the reflection should correspond to the
maximum (or minimum) amplitude value of the reflection
Page 10

FIGURE 2.13:
Comparison of a minimumphase wavelet with a zerophase wavelet that has the
same frequency content

FIGURE 2.14:
Comparison of the seismic
response of a minimum-phase
wavelet (A) and a zerophase wavelet (B) to a single
interface

FIGURE 2.15:
Comparison of delays in a
low-frequency minimum
phase wavelet and a highfrequency minimum phase
wavelet

Show Hide

About this Disc

Disc Contents (PDF format)

Show/Hide Bookmarks

Previous Page | Next Page

Search

Print

Close

Navigation

An Introduction to Seismic Interpretation Chapter Two Principles of the Seismic Method


(FIGURE 2.14B), providing that interference from
adjacent beds can be neglected (see below). Furthermore,
this will be true regardless of the frequency content of the
wavelet (FIGURE 2.16). As such, the interpreter can map
an interface with greater confidence. A problem with zero
phase wavelets is that they cannot be produced in nature.
Consider the zero phase wavelet shown in FIGURE 2.13B.
Amplitudes change before the reference point (t = 0),
which would be akin to our bell making a sound before it is
struck. Clearly this is impossible. In practice, zero phase
wavelets are generated mathematically during a processing
step called deconvolution. Discussion of this processing
step will be deferred until the NEXT CHAPTER.
A third type of wavelet would be one in which the phase
has been rotated away from zero (possible range of rotation
between 180). FIGURE 2.17 shows four different
wavelets, each with the same range of frequencies but
different phases. Note that the symmetrical shape of the
wavelet at 0 changes as the phase is rotated. The wavelet
used in this example is known as a Ricker wavelet. A zerophase Ricker wavelet consists of a peak that is flanked,
symmetrically, by a single pair of sidelobes. The amplitude
spectrum in the lower part of that figure shows that, unlike
bandpass wavelets, Ricker wavelets are characterized
by a single peak frequency value. Ricker wavelets are
commonly used during seismic modeling, because of
their simplicity, even though they do not capture the full
complexity (e.g., number of side lobes) of real seismic
wavelets.
In the same way that it is important for an interpreter to
determine the frequency content of a seismic dataset, it is
also important to determine the datas phase. Acquisition
and processing parameters that affect phase are discussed in
CHAPTER 3, the concept of data polarity is discussed
in CHAPTER 4, and means for quantitatively determining
data polarity are discussed in CHAPTER 5. Further
discussion of phase is deferred to these chapters.

2.3 Rock Properties


We now turn our attention to the factors that control rock
(or sediment) physical properties. As mentioned at the
outset of this chapter, and discussed in detail below in
SECTION 2.4, the primary physical property of interest
for P-wave seismic methods is the acoustic impedance (Z),
Page 11

FIGURE 2.16:
Effect of frequency content
on positioning of a zerophase wavelet

FIGURE 2.17:
Effects of changing wavelet
phase on the shape of the
wavelet

Show Hide

About this Disc

Disc Contents (PDF format)

Show/Hide Bookmarks

Previous Page | Next Page

Search

Print

Close

Navigation

An Introduction to Seismic Interpretation Chapter Two Principles of the Seismic Method

Show Hide

the product of a materials P-wave velocity () and its bulk density (b)4:

Z b

[2.5]

In turn, and as described in this section, velocity and density are controlled by variables such as
porosity, mineralogy and pressure. These topics are covered in greater detail, and with greater
physical rigor, by Sheriff and Geldart (1995), Mavko et al. (1998), Liner (2004), Schn (2004),
Avseth et al. (2005) and others. For the purposes of this text, where the focus is on the geologic
interpretation of seismic data, the treatment will not be in as much detail.

Elastic Properties and Velocities


SECTION 2.2 discussed P-waves and S-waves. As shown in FIGURE 2.4A, the passage of a
P-wave induces changes in the volume and shape of a body. FIGURE 2.4B shows that the passage
of an S-wave induces changes in the shape of a body, but not its volume. In principle, the body
returns to its original shape and volume following the passage of the wave. In both cases, the body
is subjected to a stress (e.g., compressive stress and/or shear stress) and undergoes strain (i.e.,
deforms) and then recovers and returns to its original shape after the stress ceases. The amount of
strain induced in a body by a given amount of stress is controlled by properties known as the elastic
modulii. We focus here on two of these modulii, the bulk modulus and the shear modulus.
The bulk modulus () is a measure of the compressibility of a body. It is the stress-strain ratio under
simple hydrostatic pressure, and measures the bodys propensity to change volume, not shape. The
bulk modulus is the inverse of the compressibility (e.g., the more compressible a body is, the lower
the bulk modulus) and so it is sometimes called the incompressibility. The term hydrostatic
pressure in the definition above means that the body is squeezed equally in all directions. If we take
a sample in the laboratory and squeeze it this way (FIG. 2.18A), we can define the bulk modulus via
the following measured properties:
V

P
V

[2.6]

where P is the measured change in pressure, V is the measured change in volume, and V is the
original volume. Inspection of this equation shows that the larger the change in volume for a given
pressure (i.e, the more compressible the body), the smaller the bulk modulus will be.
In principle, the bulk modulus can be broken down into three components, one related to the
incompressibility of the mineral grains (e.g., quartz sand grains), one related to the framework (e.g.,
how well cemented or compacted the rock is), and the third related to the pore-filling fluids (water
being incompressible, gas being compressible). Avseth et al. (2005) suggested that the bulk modulus
can be approximated using the following expression:

rock

mineral

[2.7]

where rock is the bulk modulus of the saturated rock, mineral is the bulk modulus of the minerals
__________________________________________________________________
4

Units of acoustic impedance are strange, and might be meters per second x kilograms per meter cubed, or perhaps feet
per second x grams per cubic centimeter, depending on the input values used to measure velocity and density.

Page 12

About this Disc

Disc Contents (PDF format)

Show/Hide Bookmarks

Previous Page | Next Page

Search

Print

Close

Navigation

An Introduction to Seismic Interpretation Chapter Two Principles of the Seismic Method


comprising the rock, is the porosity and is the bulk
modulus of the saturated pore space. This formulation
emphasizes that the rocks bulk modulus is related not
only to the mineralogy but also to the porosity and
the nature of the pore-filling fluids. Note the inverse
relationship between bulk modulus and porosity suggested
by EQUATION 2.7. All else being equal, the higher the
porosity, the more compressible a rock will be.
The shear modulus (), also known as the rigidity, is the
stress-strain ratio for simple shear, and measures a bodys
ability to resist shear deformation. It is associated with a
change in shape not volume. To measure the shear modulus
of a body in the laboratory, we might take our sample
and apply a shearing stress (FIGURE 2.18B). The shear
modulus will be defined using:

F/A
x / l

FIGURE 2.18:
Schematic illustration of the
types of strain measured
by the bulk modulus () and
the shear modulus ()

[2.8]

where F is the force, A is the area on which the force acts


(F/A is the shear stress), x is the transverse displacement,
and L is the initial length (x/l is the shear strain). An
important point to note about the shear modulus is that its
value is unaffected by the nature of any pore-filling fluids.
A gas-filled sand will have the same shear modulus as an
equivalent sand (mineralogy, cementation, porosity, etc.)
having its pores filled with water, or oil, or any combination
of these fluids. This is because fluids have no shear
strength (no rigidity), and therefore they do not add to the
rigidity of the matrix.
The equations for the bulk and shear modulii presented
above assume two things. First, there is no permanent
distortion of the sample (i.e., the deformation is truly
elastic). Second, the equations assume that the material
is isotropic. This second assumption might mean, for
example, that it is not easier to squeeze a rock parallel
to bedding than perpendicular to bedding. Anisotropy
can be important in some rocks, for example shales or
fractured rocks. Because physical properties are not
uniform in all directions, the elastic modulii (and hence
velocities, as described below) might be different parallel
versus perpendicular to bedding (shale), or parallel versus
perpendicular to fractures.

Page 13

FIGURE 2.19:
Plot of experimentally
measured P-wave velocity ()
versus density (b) for various
sedimentary rock types

Show Hide

About this Disc

Disc Contents (PDF format)

Show/Hide Bookmarks

Previous Page | Next Page

Search

Print

Close

Navigation

An Introduction to Seismic Interpretation Chapter Two Principles of the Seismic Method

Show Hide

An additional property that we need to consider is the bulk density of a rock or sediment. Density
is a mass per unit volume (e.g., g/cm3 or kg/m3). The bulk density of a rock (b) is related to the
density of the mineral matrix (ma), the porosity () and the density of the pore-filling fluids (f) via
the following mass-balance relationship:

b = f + (1 - ) ma

[2.9]

Values of density for some common minerals and fluids are listed in TABLE 2.1. Ji et al. (2002) and
Schn (2004) presented a more comprehensive list of mineral and fluid densities.
Note how, like the bulk modulus, the bulk density is a function of the mineralogy, the porosity and
the nature of the pore-filling fluids. Because the density of fluids (f) is lower than the density of
the mineral constituents (ma), higher porosity rocks will have lower densities than lower porosity
rocks of the same lithology. Similarly, a gas-charged sandstone will have a lower density than an
equivalent sandstone that is wet (i.e. pore spaces filled with water) because the pore-filling gas is
less dense than water.
Mineral/Fluid
Quartz
Calcite
Dolomite
Anhydrite
Gypsum
Salt
Water (Pure)

P-Wave Velocity (m/s)


5490
6710
7010
6100
5790
4570
1400

Density (g/cc)
2.65
2.71
2.875
2.98
2.35
2.03
1

Table 2.1 Compressional wave velocity and density for common materials in
sedimentary deposits.
The main reason for our interest in these physical properties is that they control P- and S-wave
velocities. The relationships between elastic modulii and rock velocity (P-wave and S-wave)
were developed by Gassman (1951) and Biot (1956) and are commonly simply referred to as the
Gassman equations. There are different formulations for relating elastic modulii to velocity, but a
simple equation for predicting P-wave velocity () as a function of the bulk modulus, shear modulus
and bulk density can be formulated as:

4( )

[2.10]

Inspection of this equation shows several things. First, the P-wave velocity is proportional to the
bulk modulus. The harder it is to compress a rock, the higher the P-wave velocity. Second, the
P-wave velocity is also proportional to the shear modulus. Third, although there appears to be an
inverse relationship between bulk density and velocity, the same factors that control bulk density
also control the bulk modulus. For example, increasing the porosity decreases the bulk density and
the bulk modulus. As such, there is a positive correlation between P-wave velocity and bulk density,
Page 14

About this Disc

Disc Contents (PDF format)

Show/Hide Bookmarks

Previous Page | Next Page

Search

Print

Close

Navigation

An Introduction to Seismic Interpretation Chapter Two Principles of the Seismic Method

Show Hide

as noted experimentally by Gardner et al. (1974; FIGURE 2.19).


The relationship between velocity and density for most of the sedimentary rock types shown in
FIGURE 2.19 can be approximated using the following empirical equation, commonly referred to
as the Gardner Equation:
(

b
0.23

)4

[2.11]

The constants in this equation were derived to fit the typical reservoir rocks (sandstone, shale,
limestone, dolomite) in that figure and were derived when velocity units are in feet per second and
density is expressed in grams per cubic centimeter. The constant in the denominator of the fraction
on the right hand side is 0.31 (instead of 0.23) when velocity is in m/s. If velocity and density logs
are available, the constants can be adjusted to give a better fit for a particular rock type in a particular
basin.
Another equation that is sometimes used, primarily in log analysis, to relate porosity and P-wave
velocity of a rock is the time-average equation (Wyllie et al. 1956):
1

Vf

(1 )
Vm a

[2.12]

where Vf is the velocity of the pore-filling fluids and Vma is the velocity of the rock matrix (see
TABLE 2.1 for typical values; Schn, 2004, presented a more comprehensive list of mineral
velocities). This equation suggests that the travel time of the acoustic signal through the rock is the
sum of the partial travel time through the solid matrix and the partial travel time through pore space
filled by a pore fluid. Although this equation appears to be physically based, it is in fact empirically
derived and its general applicability is questioned in seismic studies (e.g., Mavko et al., 1998; Eberli
et al., 2003; Saleh and Castagna, 2003; Avseth et al., 2005). Gardner et al. (1974) however indicated
that the time-average equation is adequate for relatively deeply buried brine-filled sedimentary
rocks where the influence of pressure on velocity is minimal. In these cases velocity is essentially a
function of porosity and mineralogy. There is however a general consensus that EQUATION 2.12
is not adequate for predicting how changes in the composition of pore-filling fluids will affect
seismic response (see discussion of time-lapse seismic methods in CHAPTER 8).
The S-wave velocity () can be defined as:

[2.13]

Note that the S-wave velocity does not depend on the bulk modulus because there is no compression
involved in the propagation of an S-wave (FIGURE 2.4). Note also that, because the shear modulus
is unaffected by the nature of pore-filling fluids, the effect of changing fluids is only felt through the
density term. As such, pore filling fluids have a minimal impact on shear wave velocity.
TABLE 2.2 illustrates qualitatively how changing various physical parameters (temperature,
overburden pressure, pore pressure, porosity, clay content and gas saturation) will affect the PPage 15

About this Disc

Disc Contents (PDF format)

Show/Hide Bookmarks

Previous Page | Next Page

Search

Print

Close

Navigation

An Introduction to Seismic Interpretation Chapter Two Principles of the Seismic Method

Show Hide

and S-wave velocities and density of a sedimentary deposit. Note that many of these parameters
(porosity, pore-filling fluids, lithology) are of direct interest to the petroleum industry. These
relationships illustrate why reflection seismology has become such a staple of that industry the
method responds to factors that are of direct interest.
P-Wave
Velocity
With Increasing: Temperature
Decreases
Pressure
Increases
Pore Pressure Decreases
Porosity
Decreases
Clay Content Decreases
Gas Saturation Decreases

S-Wave Velocity Density


Slight Decrease
Increases
Slight Decrease
Decreases
Decreases
Slight Increase

Unchanged
Increases
Unchanged
Decreases
Unchanged
Decreases

Bulk
Modulus
Decreases
Increases
Increases
Decreases
Decreases
Decreases

Shear Modulus
Slight Decrease
Increases
Decreases
Decreases
Decreases
Unchanged

Table 2.2 Changes in rock properties with changes in reservoir conditions.


Compiled from various sources.

Changes with Burial and Diagenesis


FIGURE 2.20 illustrates how diagenesis (i.e. porosity reduction due to cementation, compaction
and pressure solution) and sorting (including clay content) affect P-wave velocity. In that figure, the
Suspension Line represents freshly deposited sediments of different sorting. Well-sorted sands have
an initial porosity of approximately 40% (lower right) and more poorly sorted deposits have lower
porosities. Porosity reduction through diagenesis causes the velocity to increase along the red lines
for a given initial porosity. Porosity reduction through cementation and other processes increases the
rigidity and incompressibility of the rock, causing the velocity to increase. The red line joining the
mineral point (zero porosity) with the clean, well-sorted sand point represents a theoretical upper
bound to velocity. The magenta lines show the trends for sediments of similar states of diagenesis
(i.e., age) but different sorting (clay content). Increasing the clay content (not shown on the graph)
softens the rock (clay has lower elastic modulii than quartz) resulting in a slower velocity for shaly
sandstones than for clean sandstones for any given porosity.
Recall from TABLE 2.2 how physical properties such as porosity control velocity and density.
Porosity tends to decrease with depth due to the combined effects of compaction and cementation.
Inasmuch as velocity, density and acoustic impedance are inversely related to porosity, this means
that these three variables typically increase with depth5. Sands and shales typically follow different
compaction trends in sedimentary basins. FIGURE 2.21 shows typical acoustic impedance trends
for these lithologies in a thick (few km), normally compacting sedimentary basin. In the shallow
part of the section the acoustic impedance of unconsolidated, but somewhat cohesive mud is higher
than that of the wholly unconsolidated sands. Mud compacts earlier in response to burial than
sand, explaining why their impedance is higher than sand at shallow burial depths. With depth,
both the sands and shales compact and so their velocities and densities increase, but at different
rates. At a certain point in the burial profile, the acoustic impedances of the two lithologies may be
approximately equal. Finally, at depth the acoustic impedance of the lithified sandstone is greater
than that of the shale. This discussion assumes that the pore space of the sand is entirely filled with
__________________________________________________________________
5

Although obviously overburden pressure, lithology, temperature and other variables affect velocity too, as per TABLE 2.2.

Page 16

About this Disc

Disc Contents (PDF format)

Show/Hide Bookmarks

Previous Page | Next Page

Search

Print

Close

Navigation

An Introduction to Seismic Interpretation Chapter Two Principles of the Seismic Method


water. If part of that pore space is filled with hydrocarbons,
different acoustic impedance relationships might apply (see
CHAPTER 8).
The preceding discussion has focused primarily on
siliciclastic deposits. Carbonate diagenesis differs from
clastic diagenesis in many respects (e.g., Choquette and
Pray, 1970). As just noted, porosity generally decreases
with depth in clastic systems because compaction is
driven by the weight of the overlying sediments. Simple
relationships between porosity and depth may be
absent in carbonate systems because early cementation
(sometimes within months of deposition) may occlude
porosity at shallow depths and/or inhibit subsequent
burial compaction. Carbonate rocks and sediments are
also different from siliciclastic deposits in that whereas
their elastic properties are related to porosity, the type of
porosity plays a significant role in controling rock velocity
in carbonates. As such, it becomes important to understand
carbonate diagenesis, how diagenetic processes vary
with depth, and how diagenesis affects rock properties in
carbonates. For example, carbonate sediments can become
cemented shortly after deposition and, in this case, the loss
of porosity with depth due to mechanical compaction may
be hindered. Carbonate minerals are also more succeptible
to dissolution (secondary porosity development) and
precipitation than silicate minerals such as quartz.
Anselmetti and Eberli (1997), Wang (1997), Eberli et al.
(2003), Fournier and Borgomano (2007), and Weger et al.
(2009) showed how the porosity fraction, pore type (e.g.,
moldic, intercrystalline) and cement type affect velocity
in carbonate rocks. Differences in mineralogy (e.g.,
calcite versus dolomite) have a relatively minor influence
on velocity. FIGURE 2.22 summarizes how diagenesis
affects the p-wave velocity of carbonates. Cementation
and compaction increase acoustic impedance, whereas
dissolution initially has a relatively minor influence
because the development of pores does not significantly
affect the rigidity or compressibility of the rock. Sucrosic
dolomitization (a fabric-destroying process) increases
porosity and decreases velocity.

Igneous and Metamorphic Rock Properties


Rock properties of igneous and metamorphic rocks
were discussed by Salisbury et al. (1996), Milkereit and
Page 17

FIGURE 2.20:
Generalized model for
relationship between P-wave
velocity () and porosity ()
in clastic deposits

FIGURE 2.21:
Schematic representation of
compaction trends for brinefilled (wet) sands and shale
in a normally compacting
sedimentary basin

FIGURE 2.22:
Summary diagram illustrating
how diagenetic processes
affect relationships between
P-wave velocity (), porosity
() and acoustic impedance
trends of carbonates

Show Hide

About this Disc

Disc Contents (PDF format)

Show/Hide Bookmarks

Previous Page | Next Page

Search

Print

Close

Navigation

An Introduction to Seismic Interpretation Chapter Two Principles of the Seismic Method

Show Hide

Eaton (1998) and Schn (2004). FIGURE 2.23 compares velocity-density trends for igneous
and sedimentary rocks. Crystalline silicate rocks extend trends seen in sedimentary rocks to
higher velocities and densities along a Nafe-Drake curve (Ludwig et al., 1970) that relates to
b for velocities ranging from 1.5 to 8.5 km/s. Christensen and Mooney (1995) also presented
a relationship between and b for crystalline rocks. Ultramafic rocks have higher acoustic
impedance than mafic or felsic igneous rocks. Comparison of Figures 2.19 and 2.23 shows overlap
in acoustic impedance between carbonates and felsic igneous rocks (velocities of 6,000 to 7,000 m/s
and densities of 2.7 to 2.8 g/cm3). Note the relative position of sulphide and iron-oxide minerals.
These ores have acoustic impedances that are significantly in excess of most crustal rocks, with
much of the difference being related to differences in density (mineral densities ranging from 4.1 to
7.6 g/cm3).

Rock Properties and Lithology - Summary


Figures 2.19, 2.22 and 2.23 all show lines of equal acoustic impedance superimposed on velocitydensity or velocity-porosity relationships. From those figures, we note that: 1) Different types of
(sedimentary) rock can have acoustic impedances that are different to, or the same as, the acoustic
impedance of other types of rocks. As a general rule though, in most settings different rock types
have different acoustic impedances. 2) A single rock type (e.g., sandstone, shale) can have a
range of acoustic impedances. This should not be surprising because of, for example, variability
in the porosity, mineralogy, pore-filling fluids, burial depth (pressure) and other variables for
sandstones or other lithologies.

2.4 Waves and Interfaces


SECTION 2.1 discussed raypaths and wavefronts and stated that, if only for simplicity, it is
common to draw raypaths when representing seismic waves. SECTION 2.3, established the
controls on P- and S-wave velocity. We have not yet discussed why reflections occur. This section
examines what happens when downgoing seismic energy encounters an interface such as a bed
boundary.
Two terms used by interpreters, and occasionally throughout this book, need to be defined here.
First, interpreters sometimes refer to a reflection produced at the level of a particular horizon as an
event. For example, if the top of the (hypothetical) Smith Formation generates a single reflection,
it might be referred to as the Smith event, much to the chagrin of stratigraphers who use the
term event to denote something that took place geologically instantaneously (e.g., a volcanic
eruption or an extinction event). A geologist recognizes that stratigraphic units such as formations,
members, parasequences, etc., are unlikely to have been deposited as a single event (for example, by
a single river flood). The term is most likely to be used by geophysicists with a broad background
in seismology who view seismic traces as amplitude excursions that occur at a particular time
(remember, the vertical axis on a seismic trace is time). A second term that may be encountered is
loop. This term is sometimes used to refer to a single peak or trough between two zero crossings
(a positive or negative loop respectively). For example, an interpreter might state that the Smith
Formation corresponds to a single positive loop. Both of these terms, event and loop, can be useful
when describing seismic reflections.

Page 18

About this Disc

Disc Contents (PDF format)

Show/Hide Bookmarks

Previous Page | Next Page

Search

Print

Close

Navigation

An Introduction to Seismic Interpretation Chapter Two Principles of the Seismic Method

Geometry of Reflections
Except for some single-channel seismic profiling systems
(such as those used to study modern marine sediments), the
energy source and receivers are generally at some distance
apart during seismic acquisition. As a result, the energy
will hit an interface at an angle, rather than perpendicular
to it. If some of the energy is reflected, then the angle of
incidence (1) will be equal to the angle of reflection as
shown in FIGURE 2.24A. Burger et al. (2006) provide a
readable account of how basic physics (Huygens Principle
or Fermats Principle) can be used to prove this.

FIGURE 2.23:
Velocity-density relationships
for crystalline igneous rocks
and sedimentary rocks

Rays that are transmitted across the interface will be


refracted (bent) according to Snells Law:
sin 1 v1

sin 2 v2

[2.14]

where 2 is the angle of refraction, and v1 and v2 are


the velocities of the overlying and underlying medium
respectively6. FIGURE 2.25 shows schematically how
raypaths will be refracted depending on whether the
velocity of the underlying medium is higher (blue), lower
(red) or the same (black) as the velocity of the overlying
medium. CHAPTER 3 discusses how these types of
velocity contrasts can affect seismic imaging.
Note that if the angle of incidence exceeds a certain critical
angle (C) the downgoing energy will not be reflected.
Instead, the rays are refracted along the interface and
will eventually return up to the surface at the critical
angle (FIGURE 2.26). The energy refracted back up to
the surface this way is referred to as a head wave, or a
first-break refraction. Additionally, in the presence of a
velocity gradient, raypaths can bend and eventually move
horizontally or even upward without having generated a
reflection. This energy is associated with turning waves,
and analyses of these waves is sometimes used to generate
velocity profiles or for imaging of steeply dipping features.
Not all reflecting horizons in the subsurface are planar and
horizontal (FIGURE 2.24B) but the angle of incidence will
still be equal to the angle of reflection (as measured with
respect to a plane normal to the reflecting surface), and
Snells Law will still apply. As such reflections can take

FIGURE 2.24:
Acoustic response associated
with reflection of a P wave

FIGURE 2.25:
Changes in refraction angle
according to whether the
velocity of the lower medium
is higher, the same, or lower
than the overlying medium

__________________________________________________________________
6

Note that this relationship is independent of bulk density.

Page 19

Show Hide

About this Disc

Disc Contents (PDF format)

Show/Hide Bookmarks

Previous Page | Next Page

Search

Print

Close

Navigation

An Introduction to Seismic Interpretation Chapter Two Principles of the Seismic Method


unusual paths through areas of dipping strata, especially if
velocity contrasts are strong and/or velocity reversals are
present. As discussed in CHAPTER 3, dipping interfaces
will have an effect on seismic imaging.
FIGURE 2.24 shows the acoustic response associated with
reflection of a P wave. In reality, when a P-wave hits an
interface at an angle, some of the energy will be reflected
back as a P-wave and some will be transmitted as a P-wave,
as shown in those figures. However, because the incident
P-wave hits the interface at an angle, some of the waves
energy will be mode converted into a reflected S-wave
and a transmitted S-wave (FIGURE 2.27). This is known
as the elastic response, and we will come back to it when
we discuss multi-component seismic data and amplitude
variation with offset (AVO) in CHAPTER 8.

FIGURE 2.26:
Rays refracted along an
interface

Reflection Amplitude
We now need to discuss how much energy will be
reflected at an interface. For simplicity, we will begin by
considering only P-wave energy and vertically incident
rays.
P-wave energy will be reflected at an interface when there
is a change in acoustic impedance. The relative amplitude
of the reflection generated at an interface can be predicted
using the zero-offset reflection coefficient (R0):

R0

Z 2 Z1
Z 2 Z1

[2.15]

where Z1 and Z2 are the acoustic impedances of the layer


above and below the interface respectively
(FIGURE 2.28). The term zero offset effectively means
that the sound is hitting the interface at a right angle (i.e.
straight down for a horizontal interface). The zero-offset
reflection coefficient can be approximated using three other
formulae:

R0

R0

1 Z

2 Z *

1 1

2 * 2 *

1
1
R0 ln( 2 2 ) ln( 11 )
2
2

FIGURE 2.27:
Demonstration of the elastic
response

[2.15a]

[2.15b]

[2.15c]
Page 20

FIGURE 2.28:
The zero-off set reflection
coefficient

Show Hide

About this Disc

Disc Contents (PDF format)

Show/Hide Bookmarks

Previous Page | Next Page

Search

Print

Close

Navigation

An Introduction to Seismic Interpretation Chapter Two Principles of the Seismic Method

Show Hide

where Z is the difference in acoustic impedance between the upper and lower layer (i.e. Z2 Z1)
and Z* is the average acoustic impedance (

Z 2 Z1
2

), is the difference in density between the


2 1

upper and lower layer (i.e. 2 1) and * is the average acoustic impedance ( 2 ), is the
difference in P-wave velocity between the upper and lower layer (i.e. 2 1) and * is the average
2 1

velocity ( 2 ), and ln is the natural logarithm. A significant advantage of EQUATION 2.15B is


that it separates the contributions of density and velocity to R0 and this formulation will be helpful in
amplitude-variation-with-offset studies (CHAPTER 8).
The reflection coefficient is a relative number, with theoretical limits of -1 to 1, that is unitless. It
represents the amplitude of the reflected wave (Arfl) divided by the amplitude of the incident wave
(Ai)7:
2 A1rfl
[2.16]
R0
2 Ai
Inspection of EQUATION 2.15 leads to some interesting conclusions. First, if the acoustic
impedance of the underlying layer is greater than the acoustic impedance of the overlying layer
(Z2 > Z1), R0 is a positive number. In most North American seismic images, a positive reflection
coefficient will be displayed as a peak (but see CHAPTER 4 for a discussion of seismic polarity).
On the other hand, if the acoustic impedance of the underlying layer is less than the acoustic
impedance of the overlying layer (Z2 < Z1), R0 is a negative number which will be represented by
a trough in the seismic data. The amplitude of the peaks or troughs, corresponding to the absolute
value of R0, can be large or small depending on the relative contrast in acoustic impedance. A large
acoustic impedance contrast will generate a strong reflection, whereas a small acoustic impedance
contrast will generate a weak reflection.
Recall from SECTION 2.3 (FIGURE 2.21) that sands and shales follow different compaction trends
in subsiding basins. In the shallow part of the section the acoustic impedance of unconsolidated
sand is less than the acoustic impedance of the encasing muds. At depth, the situation is reversed. In
the shallow part of such a basin, the top of a sand will correspond to a negative reflection coefficient,
whereas deeper in the section it will correspond to a positive reflection coefficient8. Predicting or
identifying the depth at which this reversal in acoustic impedance occurs, or whether it occurs at all
in the section being studied, can be an important part of a seismic interpretation in thick sedimentary
successions (Avseth et al. 2005). This is because an interpreter needs to know how to pick the tops
of sand layers (i.e., potential hydrocarbon reservoirs) in the seismic data (i.e. do they correspond to
peaks or troughs?).
An important corollary of this discussion is that seismic data image interfaces. In the simplest case
(thick bed, zero-phase wavelet) the reflections correspond to the top (and base) of a bed, and the
reflection amplitude is proportional the magnitude of the acoustic impedance between the layers,
not the physical properties of the layers themselves. Of course, predicting the physical properties
(porosity, lithology, fluid content, etc.) of the layers can be the reason for acquiring seismic data
__________________________________________________________________
7

A waves amplitude is proportional to the square root of the waves energy

This discussion assumes that the pore space is entirely filled with water.

Page 21

About this Disc

Disc Contents (PDF format)

Show/Hide Bookmarks

Previous Page | Next Page

Search

Print

Close

Navigation

An Introduction to Seismic Interpretation Chapter Two Principles of the Seismic Method


in the first place. A variety of advanced methods (see CHAPTER 8) will be needed to convert
seismic images of interface properties into images of layer properties. A second consequence of
EQUATION 2.15 is that different combinations of layers with different lithologies can produce the
same reflection coefficients in which case the amplitude response will be the same. Consider the
following two cases:

Case 1
Case 2

Lithology
Shale
Sandstone
Limestone
Dolomite

b (kg/m3)
2300
2200
2500
2550

(m/s)
2450
2745
4900
5150

R0
Z (m/s x kg/m3)
5635000
0.0346
6039000
12250000
0.0348
13132500

Case 1 lists velocity and density values for a relatively young shale and sandstone. If the shale
overlies the sandstone, the reflection coefficient generated at the interface will be ~0.035. Case 2
shows values for a limestone overlying a dolomite of higher velocity and density. Note the large
absolute difference in velocity and density between the carbonate and clastic example, but the
reflection coefficient for both is approximately the same, i.e. the strength of the reflections from the
top of the sandstone and the top of the dolomite will be essentially the same.
For these reasons, seismic data are sometimes referred to as non-unique. That is, by itself, the
amplitude of a reflection is not diagnostic of rock properties such as lithology, porosity or type of
pore-filling fluids. Based on P-wave reflection amplitudes alone, we cannot tell whether we are
looking at a succession of young clastic sediments or indurated Paleozoic carbonates. However,
these can be exactly the types of challenges that confront seismic interpreters. Other lines of
evidence (e.g., reflection configurations, variations in amplitude with angle of incidence) will be
needed for this purpose, and these topics will be explored in later chapters.

2.5 Convolution and Vertical Resolution


Having described the seismic response for a single interface, it is now necessary to proceed to a
more geologically reasonable situation where many layers are present, each with their own velocity
and density, and some of the layers are close enough so that the reflections from the interfaces
interfere.

Convolution
Consider the simplified geologic column presented on the left in FIGURE 2.29. Each of the five
rock units shown has its own velocity and density, and so its own acoustic impedance. With five
layers, we have four interfaces, and it is possible to calculate the reflection coefficient at each
interface using the acoustic impedance of the layers above and below each of the interfaces. In
the figure, the four reflection coefficients are color coded and, in principle, each will generate a
reflection that is a scaled version of the input wavelet. The topmost reflection coefficient (blue) is a
strong positive value. This will generate a high-amplitude peak, as shown in blue to the right. Note
that the reflection includes side lobes in addition to the central peak. The next interface down is
associated with a small negative reflection coefficient (green) that will generate a low-amplitude

Page 22

Show Hide

About this Disc

Disc Contents (PDF format)

Show/Hide Bookmarks

Previous Page | Next Page

Search

Print

Close

Navigation

An Introduction to Seismic Interpretation Chapter Two Principles of the Seismic Method


trough and associated side lobes. The next interface down
corresponds to a moderate-amplitude negative reflection
coefficient (brown) that generates a moderate amplitude
trough. The final interface (red) corresponds to a moderateamplitude positive reflection coefficient that generates
a moderate-amplitude peak and associated side lobes.
Unfortunately the individual reflections are not recorded
separately. Instead, the seismic trace recorded at the
surface can be considered to consist of the algebraic sum
of all the reflections (four, in this case). This collective
response is shown to the right.

FIGURE 2.29:
Schematic representation of
the convolutional theorem

Mathematically, a better approach than adding the


reflections together is to convolve the wavelet with
the series of reflection coefficients. Convolution is a
mathematical operation that combines two functions to
produce a third function (a multiplication operation for
functions). For our purposes, the two functions are: a) the
seismic wavelet (our acoustic pulse), and b) the series of
reflection coefficients (the geology). We can express this
relationship using the following equation:

h(t ) f g (t )

[2.17]

where h(t) is the seismic trace, f is the source wavelet,


g(t) is the series of reflection coefficients, is noise
and * is the convolutional operator. The convolutional
theorem is a fundamental concept in reflection seismology
which essentially views a seismic trace as the sum of
all individual reflections. It is generally assumed that
convolution represents how seismic traces are generated
when seismic data are collected in the field.
As an interpreter, it could be our task to look at the trace
at the right side of FIGURE 2.29 (a small portion of
one of perhaps several hundred or several thousand in a
seismic dataset) and define: a) the number of layers, and
b) the physical properties of those layers. However there
is clearly no simple relationship between the number of
peaks or troughs (events) and the number of interfaces.
The seismic trace includes peaks and troughs generated
at the interfaces, but also events generated by side lobes.
Additionally, there is no simple relationship between the
amplitude of an event and the reflection coefficient. This
is because the reflections (and their side lobes) can either
constructively or destructively interfere with overlying or
underlying reflections (and their side lobes). There are
Page 23

FIGURE 2.30:
Seismic modeling experiment
to illustrate the importance
of wavelet phase on seismic
response

FIGURE 2.31:
Schematic representation of a
wedge model

Show Hide

About this Disc

Disc Contents (PDF format)

Show/Hide Bookmarks

Previous Page | Next Page

Search

Print

Close

Navigation

An Introduction to Seismic Interpretation Chapter Two Principles of the Seismic Method

Show Hide

places where amplitude values add constructively (for example a positive reflection from one level
and the positive side lobe from another level) and others where they tend to cancel each other out
(e.g., a positive side lobe and a trough). This discussion emphasizes the need to keep side lobes as
small as possible, by having a broad-bandwidth wavelet.
In addition to the frequency content of a seismic dataset, the convolutional response is also affected
by wavelet phase. Some examples show the importance of wavelet phase on seismic character.
The image in the upper left of FIGURE 2.30 shows a simple seismic model with five sand layers
that thin and pinch out towards the right side of the model. The other parts of that figure show
what the geological model would look like if imaged using different wavelets that have the same
range of frequencies (5-10-50-70 Hz). Only the zero-phase wavelet has a number of strong peaks
that corresponds to the number of sand beds i.e., the definition of stratigraphy is better. The
stratigraphy is more poorly defined in the minimum phase example because of the prominent
reverberations of the input wavelet. The wavelet with the 55 phase also does not properly image
the geology. These images illustrate the interpretational advantages of zero phase wavelets.
Remember, zero phase wavelets are produced by seismic processing (see section on deconvolution
in CHAPTER 3), and cannot be produced with any seismic sources (e.g., dynamite, airguns).
Therefore, no clear relationship between the number of peaks (or troughs) and the number of bedding
interfaces is to be expected when interpreting seismic records when deconvolution has not been
applied (or when deconvolution yields an unstable, reverberating wavelet). This will commonly
be the case for shallow (engineering) seismic surveys and high-resolution marine profiling (e.g.,
boomer sources).

Vertical Resolution, Detection Limits and Tuning Phenomena


The convolutional theorem just described has interesting implications for vertical resolution. Let
us imagine a sand bed encased in shale. Both the top and basal interfaces (bed boundaries) will
produce a reflection of different polarity (perhaps a trough at the top and a peak at the base). If the
reflections are widely enough spaced, i.e. the bed is thick enough, each reflection will be distinct. As
the bed thins, the top and base reflections get closer, and they start to interfere with each other. At a
certain bed thickness, the adjacent reflections become so close that they completely cancel each other
out and the bed can no longer be detected seismically. The changes in thickness result in two interrelated phenomena: 1) changes in separation between the reflections from the top and base of the
bed, and 2) changes in amplitude of these reflections.
Geophysicists investigate these types of relationships using wedge models. For example, imagine
a wedge of low acoustic impedance sandstone encased in relatively high acoustic impedance shale
such as that shown in FIGURE 2.31A. When the bed is thick enough (e.g., at right), the reflections
from the top and base of the sandstone are separate and the peak/trough separation (in time) is
proportional to the bed thickness. Here, the reflections from the top and base are far enough apart
that they do not interfere. As the bed starts to thin, the side lobes begin to interfere with each other.
Eventually, the bed thins to a point where the side lobes from the upper reflection begin to interfere
with the main reflection from the base and vice versa. According to the convolutional theorem
(FIGURE 2.31B), this will mean adding negative side lobe amplitudes to the trough at the top,
increasing the amplitude of that reflection, and adding positive sidelobe amplitudes to the peak at
the base of the wedge, increasing the amplitude of that reflection. Although the amplitude of the
reflections changes, the time separation between the two reflections remains a reliable indicator of
Page 24

About this Disc

Disc Contents (PDF format)

Show/Hide Bookmarks

Previous Page | Next Page

Search

Print

Close

Navigation

An Introduction to Seismic Interpretation Chapter Two Principles of the Seismic Method


bed thickness down to /4. For a zero-phase wavelet, the
side lobes align perfectly with the main reflections when the
bed is /4 thick and, because of constructive interference,
the amplitude of the reflections reaches a maximum. This
increase in amplitude is known as a tuning effect. Below
this thickness, the trough and peak from the top and base
of the bed start to interfere destructively and amplitude
decreases. Eventually the bed becomes so thin that the
reflections effectively cancel each other out completely.
The amplitude of the reflections decreases. Furthermore,
below /4 the composite waveform looks like a 90 phase
wavelet, similar to a derivative of the input wavelet. The
peak/trough separation no longer changes, i.e. it is no
longer an indicator of bed thickness.
The /4 criterion defines the vertical resolution of the
seismic data. In this case, the term resolution has a
meaning from physics that signifies that the reflections from
the top and base of the bed can be resolved separately and
their separation can be accurately related to bed thickness.
Note that it is still possible to detect a bed that is below
tuning, even if technically it cannot be resolved. The
difference between detection and resolution is a common
source of confusion. The detection limit depends on factors,
such as signal-to-noise ratio. For example, look at the
two wedges shown in FIGURE 2.32. The wedge on the
right in FIGURE 2.32A has a higher acoustic impedance
contrast with the surrounding medium than the wedge on
the left. As such, the amplitudes of the reflections from the
top and base are stronger on the right than on the left. The
models shown in FIGURE 2.32A were generated without
any noise. FIGURE 2.32B shows the same two wedges,
but this time with noise added to simulate poorer data
quality. Note that it is now more difficult to detect the lowimpedance contrast wedge where it is thin. As such, we
characterize the signal-to-noise ratio in terms of the signal
being related to the strength of the reflection from a bed
(a function of acoustic impedance contrast), and the noise
being related to seismic data quality.
FIGURE 2.33 shows amplitude and thickness changes
from a wedge model, generated with a simple (i.e. few
side lobes) 30 Hz Ricker wavelet, rather than a conceptual
wedge such as that shown in FIGURE 2.31. The input
geological model is shown below, and the seismic modeling
result is shown in the middle. Notice how the color bar

Page 25

FIGURE 2.32:
Effects of noise and acoustic
impedance contrast on
seismic detectability

FIGURE 2.33:
Wedge model showing the
effects of bed thickness on
seismic response

Show Hide

About this Disc

Disc Contents (PDF format)

Show/Hide Bookmarks

Previous Page | Next Page

Search

Print

Close

Navigation

An Introduction to Seismic Interpretation Chapter Two Principles of the Seismic Method

Show Hide

emphasizes the increase in amplitudes near /4. In reality, we generally do not have perfectly
wedge-shaped beds like the one depicted in the figure. Instead, this type of modeling is used to
illustrate the predicted seismic response for beds of any given thickness.
Sheriff (2002) defined resolution as the ability to separate two features that are close together, such
as the top and bottom of a bed. Liner (2004) defined vertical resolution as the ability to associate
peaks (or troughs) on a seismic trace with the top and base of a bed. Kallweit and Wood (1982)
discussed various definitions of resolution and examined the effects of different wavelets, different
bandwidths and different types of wedges (e.g., equal but opposite reflection coefficients at the top
and base of the wedge, such as shown in Figures 2.31 and 2.33, versus equal reflection coefficients
of the same polarity at the top and base) on seismic resolution. They concluded that although there
is variability in the predicted seismic response for these cases, the practical limit of resolution
is /4. Although this resolution is commonly cited in the petroleum industry (e.g., Sheriff and
Geldart, 1995), Burger et al. (2006, p. 218) stated that in shallow seismic work (e.g., for engineering
purposes) the practical limit is more likely /2.
The graph at the top of FIGURE 2.33 shows (in blue) how the isochron (i.e. the time separation)
between the reflection from the top and base of the wedge varies from one end of the wedge to
the other. Note that the isochron decreases nearly linearly down to ~ /4. Below that thickness,
the isochron effectively does not change. Note too how amplitude changes along the wedge.
Amplitude does not vary on the right of the graph because there are no interference effects when
the bed is thick. As the bed begins to thin, the amplitude starts to increase because of constructive
interference between the reflections from the top and base of the wedge, reaching a maximum value
at /4 the tuning thickness. Below tuning, amplitude decreases to zero. Compare these trends with
those shown schematically in FIGURE 2.31. Note that the amplitude and isochron trends seen in
FIGURE 2.33 are very simplified compared to amplitude and thickness trends that are observed in
real data. The reasons for these discrepancies are: 1) Only thickness is changing in the noise-free
model, whereas in reality noise, porosity, shaliness and other variables are likely to change along
the length of a sandstone bed, and 2) The modeling was done using a Ricker wavelet that has fewer
sidelobes than a real seismic wavelet (e.g., FIGURE 2.17).
Castagna et al. (2003) noted that the idea of defining a discrete tuning thickness for a seismic dataset
may be obsolete. This is because current interpretation technology allows interpreters to view the
data at any frequency (within the recorded bandwidth). Because beds of various thickness are likely
to be present, each one will have a certain frequency (and associated wavelength, via
EQUATION 2.4) at which the amplitude reaches a maximum because of the tuning effect.
Remember that a seismic dataset contains a range of frequencies. By viewing a seismic dataset
at different frequencies, it might be possible to see reflections change amplitude as we change the
frequency. We will return to this idea in CHAPTER 7 when discussing an analysis technique called
spectral decomposition.
Brown (2004) provided some approximate detection limits (limit of visibility) for different
scenarios. He suggested that in areas of outstanding signal-to-noise ratio (the signal-to-noise ratio
being a function of the strength of the reflection and data quality), the detection limit might be 1/30
of the wavelength. Detection limits for high, moderate and low signal-to-noise datasets might be

Page 26

About this Disc

Disc Contents (PDF format)

Show/Hide Bookmarks

Previous Page | Next Page

Search

Print

Close

Navigation

An Introduction to Seismic Interpretation Chapter Two Principles of the Seismic Method

Show Hide

20 , 12 , and 8 respectively. Sheriff (2002) and Liner (2004) suggested a single detectability limit

of 25 . Note that these are not absolute values, but rather should be thought of as rough estimates
that could be used to make (hopefully) intelligent guesses about detectability limits. Recall that
the wavelength can be defined using an estimate of the dominant frequency combined with some
estimate of velocity (EQUATION 2.4).
Armed with this information, we are in a position to start evaluating questions like: How thick does a
bed need to be before it can be seen seismically? Lets walk through an example. Suppose we know
(or suspect) from wireline logs or outcrop analogs that sand beds 2.5 m thick are likely to be present
in a certain stratigraphic unit. Should we be able to see them in our seismic data? Perhaps we know
from sonic logs that the P-wave velocity of the sands is 3000 m/s, and (from an amplitude spectrum)
that the dominant frequency at the target level is 50 Hz. We can prepare the following table:
Velocity Frequency Wavelength Resolution Detection
(m/s)
(Hz)
(m)
/4 (m)
/8 (m) /20 (m) /30 (m)
3000
50
60
15
7.5
3
2
If, following Brown (2004), the detection limits for poor, good and outstanding signal-to-noise
ratios can be approximated as 1/8, 1/20 and 1/30 of the wavelength respectively, we could predict
that we would need good to outstanding data quality in order to see the 2.5 m thick bed. Again, it is
important to stress that the numbers listed for the detection limit are approximations designed to help
us calculate ballpark, not absolute, estimates.
Another utility for this type of analysis would be to determine what types of frequencies would be
needed in order to image a specific type of target. For example, if we seek to image a 8 - 12 m thick
porosity zone in a carbonate reservoir, we might set 10 m as a lower limit for detection. If we know
that we can only get good (not excellent) quality seismic data from an area (perhaps because of
surface conditions and/or our budget), we could set 8 m as the lower limit of detection, i.e. /20 =
8 m, and the needed wavelength would be 160 m. If the P-wave velocity of the carbonates is 5500
m/s, we can rearrange EQUATION 2.4 to solve for the dominant frequency needed to detect the 8 m
thick bed:

160

5500
FD

We conclude that we would need a dominant frequency of at least 34 Hz in our good-quality


seismic data in order to have a chance of detecting the porosity zone when it is at its thinnest. This
type of knowledge is useful when designing a seismic survey, or when assessing the ability of a
particular dataset to image our target.
Before leaving the wedge model, it is important to note that the changes in thickness produce
changes in amplitude of the reflection from the top and base of the wedge shown in FIGURE 2.33.
Remember that the physical properties (acoustic impedance) of the wedge and the surrounding shale
do not change laterally. These changes in amplitude are known as tuning effects, and they need to

Page 27

About this Disc

Disc Contents (PDF format)

Show/Hide Bookmarks

Previous Page | Next Page

Search

Print

Close

Navigation

An Introduction to Seismic Interpretation Chapter Two Principles of the Seismic Method


be understood when using seismic amplitudes to predict
physical properties.
Beds less than /4 are known as seismic thin beds. Many
reservoirs are seismic thin beds. As such, predicting the
thickness of these reservoirs (and so, perhaps estimating
hydrocarbons in place) from seismic data requires
additional information. We noted earlier that the number of
reflections may not equal the number of beds if the wavelet
has not been converted to zero phase during processing.
However, even if the data are truly zero phase, the number
of reflections may not correspond to the number of beds
when working with thin beds, or beds that are so closely
spaced that their reflections interfere with each other. We
will return to this point when talking about the stratigraphic
interpretation of seismic data in CHAPTER 7.
In reality, it is not just the thickness of the bed that
determines whether it will be visible or not. As shown in
FIGURE 2.34 it is a combination of the beds thickness
(with respect to the wavelength) and the degree of acoustic
impedance contrast between the bed and the adjacent
strata. As labelled at upper right, thick beds that have
high acoustic impedance contrasts with the adjacent rocks
will be visible seismically. However, relatively thin beds
might be visible if the acoustic impedance contrast is high
enough, and so might be thick beds with low impedance
contrasts. The zone at lower left represents thin beds that
have low acoustic impedance contrasts. A bed might have
a very large impedance contrast with the adjacent rocks, but
if it is too thin it will not be visible (e.g., cm- to dm-scale
carbonate concretion layers in shale). A very thick bed
might not generate a reflection if the impedance contrast is
essentially zero. Separating the visible region from the
not visible region is a diffuse grey zone. Whether or
not a bed is visible in this area will depend on factors such
as the data quality, and the geoscientists interpretation
skills.

Changes in Resolution with Depth


One problem that we may encounter during an
interpretation is that the frequency content of the seismic
dataset commonly changes with depth. The data will
typically have more high frequencies in the shallow part of
our data than in the deeper parts, especially when working
with thick successions represented by several seconds of
Page 28

FIGURE 2.34:
Schematic graph that predicts
whether a bed will generate a
detectable reflection

Show Hide

About this Disc

Disc Contents (PDF format)

Show/Hide Bookmarks

Previous Page | Next Page

Search

Print

Close

Navigation

An Introduction to Seismic Interpretation Chapter Two Principles of the Seismic Method

Show Hide

data. This problem arises because the farther the sound travels through the earth, the more it loses
high frequencies through absorption - the conversion of the wave energy to heat as the waveform
passes through the rock and strains it. The energy intensity (I) of a seismic wave at a distance (r)
from the source location is related to the source energy intensity (I0) via the following equation:

I I 0 e r

[2.18]

where the absorption coefficient has units of decibels per wavelength. Recall that the higher the
frequency the shorter the wavelength (EQUATION 2.4). If the velocity of a rock is 2500 m/s, a
20 Hz wavelet will have a 125 m wavelength and a 50 Hz wavelet will have a 50 m wavelength.
To reach a depth of one kilometre and return to the surface, the 20 Hz wavelet will have traveled a
distance equal to 16 wavelengths whereas the 50 Hz wavelet will have traveled a distance equal to
40 wavelengths. The higher frequency portion of the signal will therefore lose more energy due to
absorption than the lower frequency portion because the absorption coefficient in EQUATION 2.18
is in units of decibels per wavelength. The net result is that the farther the sound travels through the
earth, the more the high frequencies are lost. This phenomenon is referred to as attenuation.
Another music analog is helpful. Most people have heard the boom, boom of music coming
through a wall from an adjacent room (e.g., from an annoying neighbor). It is the low notes (low
frequencies) that preferentially come through the wall while the high notes (high frequencies) are
muffled out. The wall acts like a low-pass filter, in the same way that the earth preferentially lets the
low frequencies pass while filtering out the high frequencies.
The absorption coefficient is inversely related to a quantity known as Q, the quality factor, which
is dimensionless. The quality factor is the ratio of a waves energy to the energy dissipated in one
cycle of an oscillation. A material that loses no energy has a Q of infinity, whereas a completely
lossy material (i.e., one which dissipates all seismic energy) has a Q of zero. Rocks have Q
values that range from 50 300, with most sedimentary rocks having values near 100 (a seismic
wave passing through a rock with a Q of 100 loses 1/100th of its energy every cycle). Grain-tograin friction causes some absorption in sedimentary rocks, however absorption is primarily caused
when motion of the rock matrix and the pore-filling fluids becomes decoupled as the seismic signal
propagates through the subsurface (Pride et al., 2003). Much current research is focused on defining
means of estimating or predicting the quality factor for two primary reasons: 1) it is useful during
processing to recover amplitudes or to help during deconvolution (see CHAPTER 3), and 2) it
is a physical property that might be useful for estimating permeability or defining the presence of
hydrocarbons.
We noted previously that velocity tends to increase with depth, and we just noted that the dominant
frequency of our seismic data decreases with depth. With depth, the numerator on the right side of
EQUATION 2.4 is increasing and the denominator is decreasing. As such, the wavelength increases
with depth and the resolution of the seismic data decreases. These effects are shown schematically
in FIGURE 2.35. Note that the exact way in which the wavelength increases with depth will vary
from basin to basin because of differences in the absorption coefficient (absorption tends to be
greatest in unconsolidated sediments) and velocity-depth trends. There is effectively nothing that
can be done during acquisition to fix this problem. We might try to use a seismic source rich in high
frequencies but, if the earth will not let those frequencies travel down to our target depth and back

Page 29

About this Disc

Disc Contents (PDF format)

Show/Hide Bookmarks

Previous Page | Next Page

Search

Print

Close

Navigation

An Introduction to Seismic Interpretation Chapter Two Principles of the Seismic Method


up, we are wasting our time (and probably money). Ideally,
we would like to know how the earth will respond during
the planning stages of our survey when selecting a seismic
source. This may be known if seismic data have already
been collected in an area, but it may be an unknown if no
seismic data have been collected there.
It should be noted too that time-variant filters are
sometimes applied to seismic data during processing, with
higher frequencies being filtered out deeper in the section in
an effort to reduce noise. As such, the change in frequency
content with depth of a seismic image is controlled both by
attenuation and processing.
The wedge model shown in Figures 2.31, 2.32, and 2.33 is
one very simplified example of a stratigraphic succession
(a bed with sharp basal and upper contacts). The seismic
response of other types of stratigraphic successions
(e.g., coarsening-upward, fining upward) was modeled
by Meckel and Nath (1977), Neidel and Poggiagliolmi
(1977) and others. FIGURE 2.36 shows how the seismic
expression of three simple stratigraphic successions,
each consisting of a single layer, changes as a function of
thickness. If more than one layer is present, the seismic
response will be more complicated because of interference
effects (e.g., FIGURE 2.30). In these cases, building
seismic models is often helpful for decoding observed
seismic responses.

2.6 Lateral Resolution


Until now, we have focused on vertical resolution how
thick something needs to be in order to be detected. We
must also think about lateral resolution how wide an
object needs to be in order to be visible seismically.
Lateral resolution is defined by the Fresnel Zone. A seismic
reflection is generated from an area, rather than from a
single point as might be conceptualized from examination
of raypath diagrams. This is because the seismic signal
being sent down into the earth forms an expanding
wavefront (Figures 2.1, 2.37). The diameter of the Fresnel
zone (DF) depends upon three key variables: 1) the average
velocity down to the horizon of interest (v), 2) the two-way
travel time (t), and 3) the frequency (f):

DF v

t
f

[2.19]
Page 30

FIGURE 2.35:
Schematic illustration of
changes in velocity, dominant
frequency of the seismic
wavelet and resulting
wavelength

FIGURE 2.36:
Graphic table showing how
changes in bed thickness
and abruptness of vertical
contacts affects the seismic
response for three simple
stratigraphic geometries

FIGURE 2.37:
Schematic representation of a
Fresnel Zone

Show Hide

About this Disc

Disc Contents (PDF format)

Show/Hide Bookmarks

Previous Page | Next Page

Search

Print

Close

Navigation

An Introduction to Seismic Interpretation Chapter Two Principles of the Seismic Method

Show Hide

The width of an object needs to be equal to or greater than the Fresnel Zone diameter in order to
be resolved seismically9. As an example, suppose we are interested in predicting the Fresnel Zone
diameter at 2.5 s into the data in an area where the average velocity to that level is 3000 m/s, with a
dominant frequency of 25 Hz:

DF 3000

2.5
300m
25

Although this limitation implies bad news for our ability to detect channels or other features that
might be a few 10s of meters wide, in reality the Fresnel zone can be shrunk by a processing step
known as migration. Properly migrated seismic data has a Fresnel Zone that is equal to /4 (i.e. 30
m for a velocity of 3000 m/s and a dominant frequency of 25 Hz).

2.7 References
Anselmetti, F.S., and G.P. Eberli, 1997, Sonic velocity in carbonate sediments and rocks, in, I. Palaz
and K.J. Marfurt, eds., Carbonate Seismology: SEG Geophysical Developments No. 6, p. 53-74.
Avseth, P., T. Mukerji, and G. Mavko, 2005, Quantitative seismic interpretation: Cambridge
University Press, 359 p.
Biot, M.A., 1956, Theory of propagation of elastic waves in fluid-saturated porous solid: Parts I and
II: Journal of the Acoustical Society of America, v. 28, p. 168-191.
Brown, A.R., 2004, Interpretation of 3-D seismic data (6th ed.): AAPG Memoir 42, 541 p.
Burger, H.R., A.F. Sheehan, and C.H. Jones, 2006, Introduction to applied geophysics, exploring the
shallow subsurface: W.W. Norton and Company, 554 p.
Castagna, J.P., S. Sun, and R.W. Siegfried, 2003, Instantaneous spectral analysis: Detection of lowfrequency shadows associated with hydrocarbons: The Leading Edge, v. 22, p. 120-127.
Choquette, P.W., and L.C. Pray, 1970, Geological nomenclature and classification of porosity in
sedimentary carbonates: AAPG Bulletin, 54, p. 207-250.
Christensen, N. I., and W. D. Mooney, 1995, Seismic velocity structure and composition of the
continental crust: a global view: Journal of Geophysical Research, v 100, p. 9761-9788.
Eberli G. P., G. T. Baechle, F.S. Anselmetti and M.L. Incze, 2003, Factors controlling elastic
properties in carbonate sediments and rocks: The Leading Edge, v. 22, p. 654-660.
Fournier, F., and J. Borgomano, 2007, Geological significance of seismic reflections and imaging
of the reservoir architecture in the Malampaya gas field (Philippines): AAPG Bulletin, v. 91, p.
235-258.
__________________________________________________________________
9

Note that the Fresnel zone diameter is defined independently of seismic acquisition parameters such as bin size (3-D
data) or line spacing (2-D data). Having more closely spaced lines, or smaller bins, does not solve the lateral resolution
problem. We return to this issue in CHAPTER 3.

Page 31

About this Disc

Disc Contents (PDF format)

Show/Hide Bookmarks

Previous Page | Next Page

Search

Print

Close

Navigation

An Introduction to Seismic Interpretation Chapter Two Principles of the Seismic Method

Show Hide

Gassmann, F., 1951, Ueber die Elastizitt porser Medien: Vierteljahrsschrift der Naturforschenden
Ges., v. 96, p. 1-23.
Gardner, G.H.F., L.W. Gardner, and A.R. Gregory, 1974, Formation velocity and density the
diagnostic basics for stratigraphic traps: Geophysics, v. 39, p. 770-780.
Hart, B.S., 2008, Stratigraphically significant attributes: The Leading Edge, v. 27, p. 320-324.
Ji, S., Q. Wang, and B. Xia, 2002, Handbook of seismic properties of minerals, rocks and ores:
Polytechnic International Press, Montreal, 630 p.
Kallweit, R. S., and L.C. Wood, 1982, The limits of resolution of zero-phase wavelets: Geophysics,
v. 47, 1035-1046.
Liner, C.L., 2004, Elements of 3-D seismology: PennWell, 608 p.
Ludwig, W. J., J. E. Nafe, and C. L. Drake, 1970, Seismic refraction, in, A. E. Maxwell, ed.,The Sea,
Vol. 4: Wiley-Interscience, New York, p. 5384.
Mavko, G., T. Mukerji, and J. Dvorkin, 1998, The rock physics handbook: Tools for seismic analysis
in porous media: Cambridge University Press, 329 p.
Meckel, L.D., and A.K. Nath, 1977, Geologic considerations for stratigraphic modeling and
interpretation. in C.E. Payton, ed., Seismic stratigraphyApplication to hydrocarbon
exploration: AAPG Memoir, 26, p. 417-438.
Milkereit, B., and D. Eaton, 1998, Imaging and interpreting the shallow crystalline crust:
Tectonophysics, v. 286, p. 5-18.
Neidell, N.S. and E. Poggiagliolmi, 1977, Stratigraphic modeling and interpretation geophysical
principles and techniques, in, C.E. Payton, ed., Seismic stratigraphyApplication to
hydrocarbon exploration: AAPG Memoir, 26, p. 389-416.
Pride, S. R., J. Harris, D.L. Johnson, A. Mateeva, K. Nihei, R.L. Nowack, J. Rector III, H. Spetzler,
R. Wu, T. Yamomoto, J. Berryman and M. Fehler, 2003, Permeability dependence of seismic
amplitudes: The Leading Edge, v. 22, p. 518-525.
Saleh, A.A., and J.P. Castagna, 2003, Revisiting the Wyllie time average equation in the case of nearspherical pores: Geophysics, v. 69, p. 45-55.
Salisbury, M.H., B. Milkereit, and W. Bleeker, 1996, Seismic imaging of massive sulfide deposits, 1.
Rock Properties: Economic Geology, v. 91, 821-828.
Schn, J.H., 2004, Physical properties of rocks: fundamentals and principles of petrophysics:
Elsevier, 583 p.
Sheriff, R.E., 2002, Encyclopedic dictionary of applied geophysics: Society of Exploration
Geophysics, Geophysical References Series, 13, 429 p.
Sheriff, R.E., and L.P. Geldart, 1995, Exploration Seismology (2nd Ed.). Cambridge University Press,
592 p.

Page 32

About this Disc

Disc Contents (PDF format)

Show/Hide Bookmarks

Previous Page | Next Page

Search

Print

Close

Navigation

An Introduction to Seismic Interpretation Chapter Two Principles of the Seismic Method

Show Hide

Simm, R., and R. White, 2002, Phase, polarity and the interpreters wavelet: First Break, v. 20, p.
277-281.
Wang, Z., 1997, Seismic properties of carbonate rocks, in, I. Palaz and K.J. Marfurt, eds., Carbonate
seismology: SEG Geophysical Developments No. 6, p. 29-52.
Weger, R.J., G.P. Eberli, G.T. Baechle, J.L. Massaferro, and Y.-F. Sun, 2009, Quantification of pore
structure and its effect on sonic velocity and permeability in carbonates: AAPG Bulletin, v. 93, p.
1297-1317.
Wyllie, M.R.J., A.R. Gregory, and G.H.F. Gardner, 1956, Elastic wave velocities in heterogeneous
and porous media: Geophysics, v. 21, p. 4170.

Page 33

You might also like