2.vol 2 Sukmono

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 414

Network of Excellence in Training

Advance Seismic Interpretation


Volume 2

(Sukmono, 2007)

By : DR. Sigit Sukmono, Email : ssukmono@pgsc.or.id


Course Content – Vol.2

1. Seismic Attributes Analysis


2. Post Stack AI Seismic Inversion
3. Pre Stack AVO / AVA Seismic Inversion
4. Pre Stack LMR/EEI Seismic Inversion
5. Pitfalls in AVO Analysis
6. Neural Network Multi-Attributes Analysis
Koreksi : Print lagi slide 6-7

Seismic Attributes for Reservoir Analysis


Contents

Page

1. Definition & Classification 2


2. Amplitude Attributes 3
3. Complex Attributes 8
3.1. Mathematical background 8
3.2. Geological application 11
3.3. Application for thin bed analysis 17
4. Times Attributes 19
4.1. Dip and azimuth 19
4.2. Illumination 21
4.3. Similarity 25

Bibliography 29
Exercises 30

Seismic Attributes for Reservoir Analysis

By : Sigit Sukmono 1
1. Definition & Classification
Brown (2000) defined an attribute as a derivative of basic seismic measurement. All available horizon and formation attributes are not
independent one of another. The difference is only in their analysis data on basic information of involved seismic wave and their result
appearance. The basic information of seismic are time, amplitude, frequency and attenuation and they can be used as a basis for attribute
classification (Figure 1). Due to it’s unique computing technique, the complex attribute normally classified as a stand-alone attribute.

SEISMIC ATTRIBUTES

FREQUENCY ATTENUATION
TIME AMPLITUDE
PRE-STACK POST-STACK PRE-STACK POST-STACK
PRE-STACK POST- PRE-STACK POST- - Q Factor
- Velocity STACK - AVO Intercept STACK
- AVO Gradient
- Intercept X Gradient HORIZON WINDOW
WINDOW - Fluid Factor - Inst. Frequency - Spectral Frequency Gradient
- Coherency - etc - Response Frequency - “bandwith” Spectrum
- Variance - etc - Ins. freq. Gradient.
- etc - etc

HORIZON HORIZON WINDOW


- Isochrones - Reflection Amplitude - Total Amplitude
- Dip - Azimuth - Composite Amplitude - Total Energy
- Illumination - Acoustic Impedance - average Absolute
- Cosinus phase - Reflection Strength - average Energy
- etc - Amplitude Ratio - etc

Figure 1. Example of seismic attributes classification

Seismic Attributes for Reservoir Analysis

By : Sigit Sukmono 2
2. Amplitude Attributes

Amplitude is the most basic attributes of a seismic trace. Initially the interpreter’s interests to amplitude is just restricted to “it
presence” not its magnitude as in that time seismic is commonly used only for structural analysis.Currently the seismic data
processing is generally directed to obtain “preserve true amplitude” so stratigraphic analysis can be done. Seismic amplitude is
commonly used also for DHI, facies and reservoir properties mapping. The lateral changes of amplitude can be used to distinguish
one facies with another. For instances, the concordant beds tend to have higher amplitudes, hummocky lower and chaotic the lowest
one. The sand-rich environment usually also have a higher amplitude compared with the shale-prone ones. This differences in sand-
shale ratio can be easily analyzed in the amplitude map.

Types of primary amplitude attributes frequently used is as follows :

1. RMS amplitude 2. Average absolute amplitude


3. Maximum peak amplitude 4. Average peak amplitude
5. Maximum trough amplitude 6. Average trough amplitude
7. Maximum absolute amplitude 8. Total absolute amplitude
9. Total amplitude 10.Average energy
11.Total energy 12.Mean amplitude
13.Variance in amplitude 14.Skew in amplitude
15.Kurtosis in amplitude

Figures 2 - 5 provide examples how these attributes are computed.

Seismic Attributes for Reservoir Analysis

By : Sigit Sukmono 3
1 N 2
∑ ai
N
RMS =
N i =1 ∑ ai
i =1
1 2 Av. Abs =
RMS = (5 + 0 2 + ... ) N
8
= 20.38
RMS = 24.46

average peak
amplitude = ?

5 + 30 + 37 + 38 + 25
= = 27
5

Interpolated
maximum peak
=39

Figure 2. Principle of RMS, average absolute, average peak and interpolated maximum peak amplitude computation

Seismic Attributes for Reservoir Analysis

By : Sigit Sukmono 4
Average trough
amplitude =
Interpolated
- 18 - 10
maximum trough
= [-19] = 19
= 14
2

Total absolute
amplitude
Max. Absolute N
amplitude = ∑ ai
i =1
38
5+0+18+10+30+
37+38+25
= 163

Figure 3. Principle of max. absolute, total absolute, average trough and interpolated maximum trough amplitude
computation

Seismic Attributes for Reservoir Analysis

By : Sigit Sukmono 5
Average Energy
Total
N
= ∑ ai
amplitude 2

N i =1
N
= ∑ ai = 598.38
i =1

5+0-18-10 +
30+37+38+25
= 107

Mean amplitude =15.29


Total Energy
N N
= ∑ ai2 ∑ ai
i =1 i =1
a =
= 4787 no of a i ≠ 0

Figure 4. Principle of total amplitude, mean amplitude, average energy and total energy computation

Seismic Attributes for Reservoir Analysis

By : Sigit Sukmono 6
Variance of amplitude Average
Amplitude
= 13.38
1 N
V = ∑ (a i - a ) 2
N i =1
= 423.15

1 N
Skew = ∑ (a i - a ) 3 Kurtosis
N i =1 N 4
K = 1 ∑ a − a 

1 N N i=1 i 
= ∑ (a i - 15.29) 3
8 i =1
= - 3961.06
= 280868

Figure 5. Principle of average, variance, skew and kurtosis amplitude

Seismic Attributes for Reservoir Analysis

By : Sigit Sukmono 7
3. Complex Attributes

3.1. Mathematical Background

The computation of complex seismic attribute is )


basically a transformation to separate the information ARY
IN
on amplitude and angle (phase & frequency) in M AG
E (I
separate display. Information in the seismic section AC
TR
will be mathematically manipulated to produce a new RE
U
display which enhance the amplitude or angle- AT
component The term “complex” here does not mean ADR
QU
that the procedure and it’s results which are complex.
It refers more to the computation which assume that
conventional seismic trace is a real part of a complex
mathematical function (the imaginary part is the
product of Hilbert transform of the real part).

No new fundamental information produces from this


manipulation. The new sections obtained sometimes COMPLEX
SEISMIC TRACE
can better displaying the geologic aspect which is
subtle in conventional displays.

Complex trace F(t) can be depicted as a trace of vector


in a complex space which rotate continuously and
change its length as shown in Fig.6. The imaginary
and real trace are the projection of these rotated
vector in real and imaginary planes. The length of the Figure 6. Isometric diagram of actual seismic trace (Taner & Sheriff., 1977)
vector is A(t) and its angle with the horizontal is θ(t)..
Imaginary trace is called also as quadrature trace.

Seismic Attributes for Reservoir Analysis

By : Sigit Sukmono 8
Real seismic trace f(t), imaginary trace f*(t) and the complex trace F (t)
can be expressed as :

f(t) = A(t) cos θ(t) (1)

f* (t) = A(t) Sin θ(t) (2)

F (t) = f(t) + j f* (t) = A (t) eJφ(t) (3)

where A(t) and θ(t) is the instantaneous amplitude and phase

Figure 7a shows a simple Ricker wavelet f(t) and imaginary trace f* (t)
derived from it. It is shown also the magnitude |F(t) | = A(t) and phase θ(t).
Figure 7b is the isometric diagram of the same wavelet showing the
imaginary component f*(t) on imaginary place perpendicular to the real
component f(t).. Figure 3 shows the relationships of each component in time
and frequency domain.

Imaginary component is obtained by applying Hilbert transform to the real


seismic trace :

f*(t) = 1/Π * f(t) (4)

Imaginary trace is identical to the real trace whose phase is shifted 90°. It
Figure 7. (a) Real (F(t), quadrature (F*(t)) complex
can be assumed as representing the potential energy while the real trace is
amplitude (F (t)) and phase θ(t) component of a
representing the kinetic energy of particles which are oscillating as they
Ricker wavelet 25 Hz, (b) Isometric diagram of
responding to seismic wave.
real and quadrature components of a Ricker
wavelet 25 Hz (Taner & Sheriff., 1977).
Seismic Attributes for Reservoir Analysis

By : Sigit Sukmono 9
Imaginary trace is used as a basis for other complex attributes
computation.
(a)
If f(t) and f* (t) are known then A(t) and θ(t) can be solved :

A(t) = [ f2(t) + f*2(t) ]1/2 =  F(t)  (5)


θ (t) = tan-1 [ f* (t) / f(t) ] (6) (b)

where A(t) is the reflection strength and θ(t) is the instantaneous


phase. The time rate of phase change gives the frequency :

d θ (t) /dt = ωt (7)


.
Either real trace f(t) or quadrature trace f*(t) can be displayed in
normal ways as the normal seismic section. The analysis of (c)
complex seismic trace give instantaneous attribute which is
computed on the basis of sample by sample.

Complex trace analysis will give natural separation between (d)


amplitude and phase information. Amplitude attribute is
commonly called as the reflection strength. Besides as a stand
alone attribute, phase attribute serves also as a basis for
instantaneous frequency measurement.
Figure 8. Real (a) and quadrature (b) of an actual seismic
trace. The dash-line is the amplitude envelope. Figures c
and d are consecutively the instantaneous phase and
frequency where the dash line is the weighted-average
frequency.
Seismic Attributes for Reservoir Analysis

By : Sigit Sukmono 10
3.2. Geological Application

Reflection Strength
The reflection strength or the instantaneous amplitude is the root
of seismic’s signal total energy on a given instant time. Hence it
can be seen as amplitude which independent from the phase. It
serves as the envelope of the seismic trace for every time sample.

The value of reflection strength always positive with magnitude


close to the value of real data. It can reach maximum on any phase
points except the peak or trough on real trace. High reflection
strength often associate with a sharp lithology change, likes in the
case of unconformity or a sharp change of depositional
environment.

For reservoir characterization, the reflection strength will be


helpful in the following matters :
1. Regional correlations of major reflectors such as SB
regional, bedrock boundary, pre-rift – syn-rift – post-rift
boundary, depositional environment boundary, etc.
2. Exposing the reflector’s difference due to significant change
in depositional environment and distinguishing the massive
reflectors, such as unconformity with other cluster of
reflector composite. The constant character may also help in
distinguishing the single and composite reflectors. The
reflection strength of the top of massive layers tend to have a
constant value for wide area.
3. Early identification of DHI bright-spot or dim-spot Figure 9. Comparison between the real traces and reflection
phenomena which indicated by a significant decrease or strength trace (Landmark, 1999).
increase of AI

Seismic Attributes for Reservoir Analysis

By : Sigit Sukmono 11
Instantaneous Phase
Instantaneous phase is a measure of reflector continuity of seismic
section. In general definition, as the real trace changes from the
peak to the troughs then the instantaneous phase changes from 0° to
+ 180°. On the troughs, the instantaneous phase is sharply wrapped
from + 180° to – 180°. On instantaneous phase section in Figure
10, the sharply wrapped is obvious as the trace change from
positive phase to negative. As the real traces change from a trough
to peaks, the instantaneous phase changes from - 180°to 0°.
Independent to the magnitude of peak or trough amplitudes the
magnitude of instantaneous phase is always same (0° for peak
amplitude, +180° for troughs). It means that the instantaneous
phase tend to balance the weak and strong reflector. Therefore, in
instantaneous phase displays it is easier to interpret the weak
coherent reflector.
Since the phase is independent to the strength reflection, it also
often highlight the weak coherent reflector. Therefore, the phase
display is very effective for detecting the fault discontinuity,
wedging-out, unconformity and reflectors with different dips which
will interference each other. The sedimentary of prograding layers
and areas of onlap or offlap will be very highlighted thus makes
this display is also very effective for sequence boundary pickings. Figure 10. Relationship between amplitude trace and
By using instantaneous phase display, the detail stratigraphic instantaneous phase trace (Landmark, 1999).
interpretation of system tract is much easier compared if using the
reflectivity.

Seismic Attributes for Reservoir Analysis

By : Sigit Sukmono 12
Cosine of Phase
Cosine of phase is computed as Cos θ (t) = f(t) / A(t) . In Inst. Phase
another word, it is same as the product when the real trace is
divided with the reflection strength. Figure 11 shows a
comparison between the instantaneous phase and the cosine
of phase. A time range for a complete cycle is same both for
the cosine of phase and the instantaneous phase. The cosine
function will make the cycle form smoother and can give
significant effect on the reflector image in seismic display.

Instantaneous Frequency
The reflection is usually a composite of individual reflector
originated from a number of closely spaced reflector where
their values remain constant on any AI separation and Cosine of Phase
contrast. The superposition of individual reflector can
produce a frequency pattern which is the characteristic of
the composite reflection. The character of a composite
reflector can change gradually in line with the gradual
change of the lithology type or thickness.

Instantaneous frequency provides information about


frequency of a reflector, absorption effect, fracturing and
depositional thickness. Low frequency shadows may
associate with reflectors lied below gas zone, condensate or
sometimes oil reservoir. Generally the change to the low
frequency occurs only for the reflector located directly Figure 11. Comparison between instantaneous phase
below the zones, whereas the deeper reflectors will look display and phase cosine display (Landmark, 1999)
normal.

Seismic Attributes for Reservoir Analysis

By : Sigit Sukmono 13
Apparent Polarity

Apparent polarity is defined by Taner et al (1979) as the sign of


real trace when the reflection strength trace has a local
maximum. The computation assumes a zero-phase wavelet and
give a positive sign when the reflection coefficient is positive
and a negative sign when the reflection coefficient is negative.
The value of a time sample associates with a local maximum (m)
on the reflection strength trace is computed as :

Value (m) = reflection strength (m) × sign (m) (8)

The obtained value will continuously have a constant value until


the new local maximum reached. As a result, the apparent
polarity display have a blocky appearance.

Apparent polarity sometimes can distinguish type of amplitude


anomalies. For instances, bright-spot anomaly which associates
with gas accumulation in clastic sediments generally correlated
with a lower AI compared with the surrounding layers. It is
represented as a negative polarity on the top of reservoir, and
positive on the gas-water contact.

Figure 12 shows a comparison among the real, reflection strength


and apparent polarity on real trace. On the real traces, the gas-oil
contact is associated with a center of high amplitude zone and
positive polarity. On the apparent polarity display, this contact is Figure 12. Comparison between the real traces and reflection
more enhanced because the effect of apparent polarity is to strength trace (Landmark, 1999).
highlight the dominant amplitude and reflector polarity.

Seismic Attributes for Reservoir Analysis

By : Sigit Sukmono 14
Response Phase
The response phase is defined as the values of
instantaneous phase which is computed on the peak of
amplitude envelope or reflection strength. The
algorithm computes the instantaneous phase values on
every envelope peaks, and apply the obtained value on
every sample between the minimum of amplitude
envelope trace. As the result, trace of response phase
will be blocky. The response phase display highlight
the character of dominant phase of the reflector. Hence
it is also very useful for detecting the phase change due
to the lateral change of fluid content and even its
related lithology.

Figure 13 shows a comparison of instantaneous phase,


reflection strength and response phase traces. On the
instantaneous phase and response phase display, the
gas-oil contact generally has 0° phase (zero-crossing
from peak to through). However the phase reversal on
the top chalk reflector is not so clear in the response
phase display. The reason behind this phenomenon can
be seen in the display of reflection strength. Top chalk
does not associate with a strong local maximum of
reflection strength and therefore the response phase
information is more related to the surrounding
reflectors. Figure 13. Comparison of the instantaneous phase and
response phase sections (Landmark, 1999).

Seismic Attributes for Reservoir Analysis

By : Sigit Sukmono 15
Response Frquency

The response frequency has a similar function as the


instantaneous frequency. However, the response frequency
display is much easier to be interpreted in the case where
the instantaneous frequency display contain much noises.
The response frequency will highlight the characteristic of
dominant frequency of the reflector. It will help in the
detection of frequency absorption effect related to the fluid
content, fracturing and the change in the depositional
environment.

The computation of the response frequency is basically an


extraction of frequency information which physically
meaningful for a localized seismic wavelet. The response
frequency is defined as the instantaneous frequency value
computed on the peak of amplitude envelope (reflection
strength). The related algorithm computes the instantaneous
frequency on each envelope peaks and then apply the value
on every sample between the minimum of amplitude
envelope trace. As the result, the response frequency trace
has a blocky appearance. An example on the differences
among the instantaneous frequency, reflection strength and
response frequency values is shown in Figure 14. Notice
the presence of depletion of response frequency value below
the thicker reservoir. On the edges and outside of the
reservoir, the response frequency is still quite high due to Figure 14. Comparison of the instantaneous frquency
the decreasing effect of gas absorption inline with the and response frequency (Landmark, 1999).
thinning of the gas column.

Seismic Attributes for Reservoir Analysis

By : Sigit Sukmono 16
3.3. Application for Thin Bed Analysis When the wedge is thick, the response phase (Figure 16d) verify
the zero-phaseness of the basic wavelet (can be 00 or 1800) and the
The application of complex attributes for thin bed analysis can be
response frequency (Figure 16f) give the quantitative measure of
demonstrated by computing a simple attribute model of a low IA
average frequency (22.5 Hz) of the wavelet spectrum model
wedge (Fig. 15). The synthetic seismogram of this model is given in
(Ricker 20 Hz). The response phase changes to 90o between T and
Figure 16a. Wedge thickness is represented by dominant period unit
T/2, reflecting a basic wavelet differentiation. Same as the
T. The polarity convention used is that the negative reflection
response frequency, the instantaneous frequency change it’s color
coefficient is represented by troughs. Figure 16b, c, d, e, f
to blue and red as the wedge thins from T/2 to zero. It is important
consecutively showing the displays of amplitude envelope,
to note that the color change in every attribute section occur
instantaneous phase, response phase and instantaneous phase. On
between T/4 and 3T/4. It corresponds to a generally known
each attribute section the dashed line represent the upper and lower
reservoir’s thickness. Hence this modeling result shows that the
limit of wedge model on conventional seismic section.
attribute section is more suitable to define the reservoir’s edges
rather than for mapping the internal variation within the reservoir
The tuning amplitude ½ T still has a red color on the amplitude
body itself.
display and the related color scale is continuously-decreasing in line
with the thinning of the wedge model and the occurrence of the
destructive interference. Those interference pattern is displayed
more clearly on the instantaneous phase and frequency sections. For
instance, in the instantaneous frequency section, it can be observed
the increases of color scale from blue to red as the wedge thickness
reduce to less than T/4. It is widely understood also that a low IA
thin bed will differentiate a wavelet and therefore increasing the
average frequency of the composite reflection compared to the basic
wavelet. The instantaneous frequency display emphasizes this
phenomenon which is the tuning characteristic of a wavelet. This
increases phenomenon can be used to map the outer extent of a low
IA thin bed. Figure 15. Geometry and parameter of a wedge
seismic model (Robertson & Fisher, 1988)

Seismic Attributes for Reservoir Analysis

By : Sigit Sukmono 17
Figure 16 (a) Conventional seismic section, (b) Envelope amplitude, (c) Instantaneous phase (d). Response phase, (e)
Instantaneous frequency, (f) Response frequency (Robertson & Fisher, 1988)

Seismic Attributes for Reservoir Analysis

By : Sigit Sukmono 18
4. Times Attributes

4.1. Dip and Azimuth

The principle on how to calculate the dip and azimuth is quite simple as shown in Figure 17. Basically the dip and azimuth are
the magnitude and direction of the time gradient vector; from a local reference, which computed on every horizon sample which
is interpreted. The computations are carried out by fitting a plane though adjacent data points and posting the calculated values
at the central data point (Figure 17). The dip values can be expressed in degrees or radians, or more commonly, in miliseconds
per meter. Dip and azimuth values are normally displayed on separate maps. These maps should be studied independently
because faults, which affect the mapped horizon do not necessarily show up clearly.

N
Figure 17. The principle of dip and
azimuth computation
azimuth
Dip

Seismic Attributes for Reservoir Analysis

By : Sigit Sukmono 19
Figure 18 and 19 show the corresponding dip and azimuth display. When both maps are compared, the azimuth map clearly
provides a better definition of the faulting than dip map. The explanation for the difference in the clarity of faulting between the
two attributes can be explained as follows. A fault will be clearly defined on the azimuth map when the dip direction of the fault
plane is opposite to the dip direction of the beds. It will be poorly expressed when the fault plane dip direction is similar to that of
the horizon. A fault will best expressed on the dip map when the dip angle of the fault plane is notably different from the horizon
dip, and will be poorly expressed when the dip angle is close to that of the horizon.

Figure 18. Dip map of horizon A (Rijks & Jauffred, 1991) Figure 19. Azimuth map of horizon A (Rijks & Jauffred, 1991)

Seismic Attributes for Reservoir Analysis

By : Sigit Sukmono 20
4.2. Illumination

In a shaded relief display, a grid is produced in which the values are proportional to the brightness of the reflection when
“illuminated” by a light source. The principle is somewhat similar to looking at the map of a seismic marker which would be
illuminated by the sun shining from a certain direction at a given elevation. Figure 20 illustrates the principle. Depending on
the direction of illumination, fault F will appear as a dark (A) to a bright (B) lineament.

Figure 20. Shaded relief-


artificial illumination
principle (Rijks &
Jauffred, 1991)

Seismic Attributes for Reservoir Analysis

By : Sigit Sukmono 21
Rijks and Jauffred (1991) gave an example of the application of this method to the case of Amerveen gas field. The principle
reservoir here is the Rotliegend Sandstone (Figure 21). The southeast ward hading fault of this graben is not illuminated and
appears as a black line-up. The other fault plane, hading in the opposite direction, is brightly illuminated and appears as a white
line-up (Figure 22).

Figure 21. Seismic display showing the Figure 22. Shaded relief map of Annerveen Field
Rotliegend sandstone reservoir (Rijks & (Rijks & Jauffred, 1991)
Jauffred, 1991)

Seismic Attributes for Reservoir Analysis

By : Sigit Sukmono 22
The throws of these faults are very small, less than 10 ms. The graben shows only as two faults on the dip map with the relief
effect (Figure II.6). Other differences can be seen between shaded relief and dip display. The fault trend in the western part of
the area, which appears to be a continuous event on the dip map, has in fact a distinct en echelon configuration on the shaded
relief map (the white arrow). This observation is of great interest as it hints at fault block communication.

Figure II.6. The dip map of the


Annetveen Field (Rijks &
Jauffred, 1991)

Seismic Attributes for Reservoir Analysis

By : Sigit Sukmono 23
It can be observed that how clearly the faults are imaged, either un the dip or azimuth display is very much depends on the orientation
of the faults to the bedding plane. To overcome the problem of the difference in detectability between one display and another, dip
and azimuth map can be displayed in combination. Figure II.7 shows a combined dip and azimuth map of the reservoir referred to
above. The different colors in the wheel represent the azimuth values while color intensity is an indication of the dip angle (the
steeper the dip, the darker the color). This map shows that the synthetic faults throwing to the southwest show in dark blue, and the
antithetic faults throwing to the northeast are in deep orange.

Figure II.7. The combined dip-


azimuth map of the Nun River
Field (Rijks & Jauffred, 1991).
Interpret strike and dip
0 1 km

Seismic Attributes for Reservoir Analysis

By : Sigit Sukmono 24
4.3. Similarity

Degree of similarity or dissimilarity can be used to enhance the lateral changes of seismic data due to changes of geological
conditions. Several major algorithms are available to compute this parameter, the popular one is the using of the Manhattan
Distance / Semblance technique.

The Manhattan distance is a statistical measure of the dissimilarity. In the continuity cube, the main similarity measurement
is in the cross-correlation, which compare the center trace with neighbor traces by multiplying samples from the center trace
times those of a neighbor trace within a window and normalizing by the autocorrelations of these two traces. The resulting
values will vary between –1 and +1 where, the most discontinuous data, such as faults, resulted in number close to - 1.

The center trace window is compared to the target trace window using a Manhattan distance and this Manhattan
measurement is output. The Manhattan distance is the sum of the absolute value of the sample difference between the
window. This sum (numerator) is divided by the sum of the absolute values of each sample of the two traces within the
specified window. The resulting values are numbers between 0 and + 1, where the most discontinuous data, such as faults, is
closer to + 1. Figure 25 illustrates the example on how to compute this Manhattan distance.

The number of neighbor traces to be compared with the center trace generally varies between one to maximum of eight
surrounding traces. The length of data intended to be covered in analysis can also be determined in the sliding comparison
window.

Seismic Attributes for Reservoir Analysis

By : Sigit Sukmono 25
Semblance values are calculated for each center
reference trace to neighbor trace pair within the
scan pattern.

From all of these measurement, a single target


neighbor trace is selected. A second calculation
computes Manhattan distance dissimilarity using
the central trace and the target trace in the
comparison window.

This Manhattan distance value is output for the


middle sample of the center trace window. Fault
with detectable vertical throw is generally
associated with a narrow zones of low similarity.
The something occurs also if there is an abrupt
contrast in seismic character due to stratigraphic or
lithologic changes such as in the case of channel
sands.
Figure 25. An illustration on the algorithm used for computing the
Manhattan Distance (Landmark, 1999)
Gradual stratigraphic contrast, such as these
associated with transgressive sequences will
produce broad regions of moderate values. Highly
dipping pattern will result in broad regions of low
similarity zones with poor data quality of lack of
reflectors (e.g. salt structures) can also produce
broad region of low similarity.

Seismic Attributes for Reservoir Analysis

By : Sigit Sukmono 26
Figure 26. Types of common scan patterns (Landmark, 1999)

Figure II.9. An example of trace selection using semblance


(Landmark, 1999)

Seismic Attributes for Reservoir Analysis

By : Sigit Sukmono 27
Figure II.10. An illustration on how
the Semblance coefficient can
represent the geologic conditions
(Landmark, 1999)

Seismic Attributes for Reservoir Analysis

By : Sigit Sukmono 28
Bibliography

1. Brown, A.R., 2000, Interpretation of three-dimensional seismic data, Am. Assoc. Pet. Geol. Memoir 42.
2. Dalley, R.M., 1989. Dip and Azimuth Displays for 3-D Seismic Interpretation. First Break.
3. Dabeye, H.W.J. and Van Riel, P., Lp-norm deconvolution Geophysical Prospecting, Vol .38, No. 4 (May 1990)
4. Galbraith, J.M. and Millington, G.F., Low frequency recovery in the inversion of seismograms, CSEG Calgary, 1978
5. Jain, A.K., 1989. Fundamentals of digital image processing. Englewood Cliffs, N.J, Prentice Hall, 4569 p.
6. Landmark, 1999, Post Stack Reference Manual.
7. Rijks, E.J.H. and Jauffred, J.C.E.M., 1991, Attributes extraction : an important application in any detailed 3D
interpretation study, The Leading Edge 10, 11.
8. Robertson, J.D. and Fisher, D.A., 1988, Complex seismic trace attributes, The Leading Edge 7, 22.
9. Sheriff, R.E., 1991. Encyclopedic dictionary of exploration geophysics, 3 rd ed. Tulsa, SEG, 376 p.
10. Taner, M.T. and Sheriff, R.E., 1977, Application of amplitude, frequency and other attributes to stratigraphic and
hydrocarbon determination, in Seismic Stratigraphy – applications to hydrocarbon exploration (ed : Payton, C.E.),
AAPG Memoir 26.
11. Yilmaz, O., 1987. Seismic data processing. Tulsa, SEG, 526 p.

Seismic Attributes for Reservoir Analysis

By : Sigit Sukmono 29
Exercises

Seismic Attributes for Reservoir Analysis

By : Sigit Sukmono 30
Exercise I-1
Integrated Analysis of Amplitude Attributes for Stratigraphy Analysis and Plan of Field Development.

Objective
To understand the available types of amplitude attributes, their meaning and applications for stratigraphy analysis and plan of
field development.

Materials
Figure I.1.1. Well-seismic tie at gas well-1. The studied reservoir is within the SB1 and SB2 with meandering-stream
depositional environment. More specifically, the productive reservoir is lied within the SS1 and SS2 (Time 2180 – 2260 ms).

Figure I.1.2a.RMS amplitude map I.1.2b.Average absolute amplitude map


I.1.2c. Maximum peak amplitude map I.1.2d.Average peak amplitude map.
I.1.2e. Maximum peak amplitude map I.1.2f. Average trough amplitude map
I.1.2g. Maximum absolute amplitude map I.1.2h.Total absolute amplitude map
I.1.2i. Total amplitude map I.1.2j.Average energy map SS1 – SS2.
I.1.2k. Average energy map I.1.2l.Average amplitude map SS1 – SS2.
I.1.2m. Variance amplitude map I.1.2n.Skew amplitude map SS1 – SS2.
I.1.2o. Kurtosis amplitude map

In the above figures, the highest to lowest values are successively represented by yellow, green, blue, red and black colors.

Questions.
Select the most appropriate maps and propose future well location and area of development.

Seismic Attributes for Reservoir Analysis

By : Sigit Sukmono 31
Figure I.1.1. Well-seismic tie at Well-1

Seismic Attributes for Reservoir Analysis

By : Sigit Sukmono 32
a) RMS Amplitude b) Average Absolute c) MaximumPeak d) Average Peak

e) Maximum Trough f) Average Trough g) Maximum Absolute h) Total Absolute

Figure I.1.2. Amplitude attributes map

Seismic Attributes for Reservoir Analysis

By : Sigit Sukmono 33
I) Total Amplitude k) Total Energy l) Average Amplitude
j) Average Energy

m) Amplitude Variance n) Skew Amplitude o) Kurtosis Amplitude

Figure I.1.2 (Continued)

Seismic Attributes for Reservoir Analysis

By : Sigit Sukmono 34
Exercise I.2
Amplitude Attributes For Thin Bed Facies Mapping N
in Compartmentalized Reservoir (A Development S
Field Case, Brown 1998)
1.6

Materials 11
Figure I.2.1 shows the tie between the VSP and the
seismic data. Only productive reservoir displayed in
this figure. VSP has wider frequency coverage (10-
1.7
120 Hz) compared to seismic data (10-80 Hz). As the
reservoirs in this field is characterized by the thin-
beds then VSP is used to identify the reservoir
position in the seismic section. Figures I.2.2 to I.2.6
show the stratigraphic correlation and pressure
1.8
history of the wells in reservoirs 7, 11 and 1.
1
7
Questions
1.In Figure I.2.7 map Channel A, Channel B,
Channel C, splay and flood plain deposits of
1.9
reservoir #7 :.
2.In Figure I.2.8 map Channel A, Channel B, splay
and flood-plain deposit of reservoir #11. Figure I.2.1. An example of VSP seismic tie in Well 75 (outside studied area)
3.In Figure I.2.9 map Channel, splay and
floodplain deposits of reservoir #1.

Seismic Attributes for Reservoir Analysis

By : Sigit Sukmono 35
W E
9 11 5 7 1 40

Channel

Figure I.2.2 Stratigraphic correlation of reservoir #7 based Figure I.2.3 Pressure history of wells in Figure I.2.2
on log and initial-bottom hole pressure (BHP) data. showing the compartmentalized channel A, B and C.

Seismic Attributes for Reservoir Analysis

By : Sigit Sukmono 36
Time slice Time slice
position position

Figure I.2.4. Stratigraphic section of reservoir #11

Seismic Attributes for Reservoir Analysis

By : Sigit Sukmono 37
30 101 95 76 64
Completion C Completion B Completion A

Splay Channel-fill

Figure I.2.5. Stratigraphic section of reservoir #1 Figure I.2.6. Diagram pressure vs cumulative gas
production for completion A, B and C in reservoir.

Seismic Attributes for Reservoir Analysis

By : Sigit Sukmono 38
Figure I.2.7. Average amplitude map of reservoir #7. (White = 0, dark blue is negative maximum, dark red is
positive maximum). Map the following facies : Channel A, Channel B, Channel C, splay and flood plain deposits.

Seismic Attributes for Reservoir Analysis

By : Sigit Sukmono 39
Figure I.2.8 Average amplitude map of reservoir #11 (White = zero, dark blue = negative maximum, dark red = positive
maximum). Map Channel A and Channel B deposits.

Seismic Attributes for Reservoir Analysis

By : Sigit Sukmono 40
Figure I.2.9. Average amplitude map of reservoir #1. Map channel, splay/flood-plain facies

Seismic Attributes for Reservoir Analysis

By : Sigit Sukmono 41
Exercise I.3

The Application of Amplitude and Velocity Attributes For Meandering Stream Facies Mapping and Development Well
Delineation.

Materials
This exercise is taken from Leu et al., 1999, Integration of 3-D seismic attributes with core and wire line log data for detailed
modeling of Cretaceous fluvial reservoir, Leading Edge, p. 730-738. Figure I.3.1 Shows the TWT map of the top reservoir. Well
data suggests that the depositional environment of this facies is the meandering stream in which the main facies comprises of
channel-fill, levee/crevasse splay andoverbank/interchannel deposits. The pay zone is mainly associated with the channel-fill
deposits. Well Sha-5, Sha-7, Sha-15 and YK-1 generally penetrate the channel-fill facies and have a good net pay. Sha-6 and
Sha-4 Wells penetrate the cressave-splay and overbank facies and there fore have a small net pay. Figure I.3.2 Shows the
reservoir RMS amplitude map. Figure I.3.3 gives the relationship between the RMS amplitude with the net-pay and porosity.
Figure I.3.4 Provides the cross-plot between the interval velocity with the net pay and porosity. Figure I.3.5a Gives the RMS
amplitude map of 8 ms analysis window interval below the top of reservoir. Figure I.3.5b displays the RMS amplitude map of
16 ms analysis window interval below the top of the reservoir or approximately in the bottom of reservoir (on the peak). Figure
I.3.7 Shows the integrated map of interval velocity and amplitudes.

Questions
Select the most representative attribute and determine the 5 (five) best location of development wells.

Seismic Attributes for Reservoir Analysis

By : Sigit Sukmono 42
Figure I.3.1. TWT map of lower cretaceous reservoir-upper section (Leu et al., 1999)

Seismic Attributes for Reservoir Analysis

By : Sigit Sukmono 43
L500

L600
L400

L800
L300

L700
T300 T300

T400 T400

T500 T500

T600 T600

T700 T700

T800 T800

Figure I.3.2 RMS amplitude map of lower. Cretaceous reservoir-lower section (Leu et al., 1999)

Seismic Attributes for Reservoir Analysis

By : Sigit Sukmono 44
40.0 16

Sha-7 15
30.0 Sha-15
YK-1
Net Pay (m)

Sha-5

Porosity (%)
14

20.0

13

10.0
12
Sha-6
Sha-4

0.0 40 11
50 60 70 80 90
RMS Amplitude

Figure I.3.3 Cross-plot between RMS amplitude and net pay (white square) - porosity (black-square) (Leu et al., 1999)

Seismic Attributes for Reservoir Analysis

By : Sigit Sukmono 45
40.0 16

YK-1

Sha-7 15
30.0 Sha-15

Sha-5
14
N e t P a y (m )

P o r o s ity ( % )
20.0

13

10.0
Sha-4 12

Sha-6
0.0 11
4100 4150 4200 4250 4300 4350 4400
Interval Velocity

Figure I.3.4. Cross-plot between Interval Velocity and net pay (white squar, dash line) and porosity
(black-squar, solid-line) (Leu et al, 1999)

Seismic Attributes for Reservoir Analysis

By : Sigit Sukmono 46
Figure I.3.5a. Amplitude slice map of a horizon located 8 ms below the top of reservoir (Leu et al., 1999)

Seismic Attributes for Reservoir Analysis

By : Sigit Sukmono 47
Figure I.3.5b. Amplitude slice map of a horizon located 16 ms below the top of reservoir (Leu et al., 1999)

Seismic Attributes for Reservoir Analysis

By : Sigit Sukmono 48
L300

L400

L500

L600

L700

L800
T800 T800

T700 T700

T600 T600

T500 T500

T400 T400

T300 T300

Figure I.3.6. Interval velocity map (m/s) of the reservoir (Leu et al., 1999)

Seismic Attributes for Reservoir Analysis

By : Sigit Sukmono 49
L400

L500
L300

L700

L800
L600
T800 T800

T700 T700

T600 T600

T500 T500

T400 T400

T300 T300

Figure I.3.7. Integrated map of interval velocity and reservoir amplitude (Leu et al., 1999)

Seismic Attributes for Reservoir Analysis

By : Sigit Sukmono 50
Exercise II.1

Normal seismic Cosine phase

0 250 m

Figure II.1.1. Given above are the normal seismic and cosine phase sections. Questions : map erosional surfaces, onlaps and
downlaps. Interpret the depositional environments. Identify distinct seismic facies characteristic supporting your
interpretation. Compare the capability of normal and attribute section for answering the questions.

Seismic Attributes for Reservoir Analysis

By : Sigit Sukmono 51
Exercise II.2

0 2 km

Normal seismic Inst phase

Figure II.2.1. Given above are the normal seismic and cosine phase sections. Questions : map sequence boundaries and related
supporting evidences such as erosional surfaces, onlaps and downlaps. Interpret the depositional environments. Identify
distinct seismic facies characteristic supporting your interpretation. Compare the capability of normal and attribute section
for answering the questions.

Seismic Attributes for Reservoir Analysis

By : Sigit Sukmono 52
±20 km

Exercise II.3
Normal seismic

Figure II.3.a. Given above are the normal section. The related inst. Phase and reflection strength sections are given in Figure
II.3b, c. Questions : map sequence boundaries and related supporting evidences such as erosional surfaces, onlaps and
downlaps. Interpret the depositional environments and the system-tracts. Identify distinct seismic facies characteristic
Seismic Attributes
supporting your interpretation. Compare for Reservoir
the capability Analysis
of normal and attribute section for answering the questions.
By : Sigit Sukmono 53
±20 km

500

-1000

-1500 Exercise II.3


Reflection strength

Figure II.3b. The reflection strength of Figure 19a. Questions : map sequence boundaries and related supporting evidences
such as erosional surfaces,Seismic
onlapsAttributes for ReservoirInterpret
and downlaps. Analysis the depositional environments and the system-tracts. Identify
distinct seismic facies characteristic 54
supporting your interpretation. Compare the capability of normal and attribute section
By : Sigit Sukmono
for answering the questions.
±20 km

-1000

Exercise II.3
-1500
Inst. Phase
Figure II.3c. The Instantaneous Phase of Figure II.3a. Questions : map sequence boundaries and related supporting evidences
such as erosional surfaces, onlaps and downlaps. Interpret the depositional environments and the system-tracts. Identify
Seismic
distinct seismic facies characteristic Attributesyour
supporting for Reservoir Analysis Compare the capability of normal and attribute section
interpretation.
for answering the questions. By : Sigit Sukmono 55
0

500

Exercise II.3

Figure II.4. A closer view of the Instantaneous Phase in Figure II.3c. Questions : map sequence boundaries and related
supporting evidences such as erosional surfaces, onlaps and downlaps. Interpret the depositional environments and the
sediments filling direction of the most recent package (80-180 ms).
Seismic Attributes for Reservoir Analysis

By : Sigit Sukmono 56
Exercise II.4

Figure II.4a. Given above are the normal section. The related inst. Phase is given in Figure II.4b. Questions : map sequence
boundaries and related supporting evidences such as erosional surfaces, onlaps and downlaps. Revise the pickings if
necessary. Interpret the depositional environments and the system-tracts. Identify distinct seismic facies characteristic
supporting your interpretation. Compare the capability of normal and attribute section for answering the questions.

Seismic Attributes for Reservoir Analysis

By : Sigit Sukmono 57
Exercise II.4

Figure II.4b. The Inst. Phase of Figure II.4a. Questions : map sequence boundaries and related supporting evidences such as
erosional surfaces, onlaps and downlaps. Revise the pickings if necessary. Interpret the depositional environments and the
system-tracts. Identify distinct seismic facies characteristic supporting your interpretation. Compare the capability of normal
and attribute section for answering the questions.

Seismic Attributes for Reservoir Analysis

By : Sigit Sukmono 58
Exercise II.5

Figure II.5. Given above are the normal seismic and instantaneous phase sections. Questions : map sequence boundaries and
related supporting evidences such as erosional surfaces, onlaps and downlaps. Interpret the depositional environments.
Identify distinct seismic facies characteristic supporting your interpretation. Compare the capability of normal and attribute
section for answering the questions.

Seismic Attributes for Reservoir Analysis

By : Sigit Sukmono 59
Exercise II.6
gas-well

Figure II.6a. Given above are the normal section with SEG normal polarity. The related cosine phase, inst. freq, refl. strength
and inst. phase sections are given in Figure II.6b,c,d e. Questions : map sequence boundaries and related supporting evidences
such as erosional surfaces, onlaps and downlaps. Interpret the depositional environments and the top – base of gas colum.
Identify distinct seismic facies characteristic supporting your interpretation. Compare the capability of normal and attribute
section for answering the questions.

Seismic Attributes for Reservoir Analysis

By : Sigit Sukmono 60
Exercise II.6

Figure II.6b.. The cosine phase (left) and instantaneous frequency (right) of Figure II.6a. Questions : map sequence boundaries
and related supporting evidences such as erosional surfaces, onlaps and downlaps. Interpret the depositional environments
and the top-base of gas column. Identify distinct seismic facies characteristic supporting your interpretation. Compare the
capability of normal and attribute section for answering the questions.

Seismic Attributes for Reservoir Analysis

By : Sigit Sukmono 61
Exercise I.6I

Figure II.6c.. The reflection strength (left) and instantaneous phase (right) of Figure II.6a. Questions : map sequence
boundaries and related supporting evidences such as erosional surfaces, onlaps and downlaps. Interpret the depositional
environments and the top-base of gas column. Identify distinct seismic facies characteristic supporting your interpretation.
Compare the capability of normal and attribute section for answering the questions.

Seismic Attributes for Reservoir Analysis

By : Sigit Sukmono 62
Exercise II.7

Cos. Phase

Figure II.7a. The normal (left) and cosine phase (right). Questions : Interpret the depositional environments and the top – base
of gas colum. Identify distinct seismic characteristic supporting your interpretation.

Seismic Attributes for Reservoir Analysis

By : Sigit Sukmono 63
Exercise II.7

Figure II.7b. The apparent polarity (left) and dominant frequency (right) of Figure II.7a. Questions : Interpret the depositional
environments and the top – base of gas colum. Identify distinct seismic characteristic supporting your interpretation.

Seismic Attributes for Reservoir Analysis

By : Sigit Sukmono 64
Exercise II.7

Figure II.7c. The reflection strength of Figure II.7a. Bright colors associate with high reflection strength. Questions : Interpret
the depositional environments and the top – base of gas colum. Identify distinct seismic characteristic supporting your
interpretation.

Seismic Attributes for Reservoir Analysis

By : Sigit Sukmono 65
Exercise II.8

Figure II.8a. The cosine phase section. Questions : Interpret the depositional environments and the seismic facies. Identify
distinct seismic characteristic supporting your interpretation.
Exercise II.8

Figure II.8b. The normal section. Related instantaneous phase and cosine phase sections are given in Figure II.8c-d.
Questions : Identify a DHI signature and supporting seismic characteristic. Compare the capability of normal section and
attributes section for answering the question.

Seismic Attributes for Reservoir Analysis

By : Sigit Sukmono 67
Exercise II.8

Figure II.8c. The instantaneous phase sections of Figure II.8a. Questions : Identify a DHI signature and supporting seismic
characteristic. Compare the capability of normal section and attributes section for answering the question.

Seismic Attributes for Reservoir Analysis

By : Sigit Sukmono 68
Exercise II.8

Figure II.8d. The instantaneous phase sections of Figure II.8a. Questions : Identify a DHI signature and supporting seismic
characteristic. Compare the capability of normal section and attributes section for answering the question.

Seismic Attributes for Reservoir Analysis

By : Sigit Sukmono 69
Exercise II.9 Reflection Phase Frequency Apparent
Frequency Strength (DB) (Degree) (Hz)
Phase Polarity
(Hz)
(Degree)
Reflection
Apparent
Strength (DB)
Polarity

Figure II.9a. Color code for Figures 27b-h. (from Tarner at al., 1979)). Question : Combine all sections in Figure II.9b-h
and identify economic and non-economic gas interval

Seismic Attributes for Reservoir Analysis

By : Sigit Sukmono 70
Exercise II.9

Figure II.9b. Reflection strength section Line A Part 1 (Tarner et Figure II.9c. Reflection strength section Line A Part 2
al., 1979) (Tarner et al., 1979)

Question : Combine all sections in Figure II.9b-h and identify economic and non-economic gas interval

Seismic Attributes for Reservoir Analysis

By : Sigit Sukmono 71
Exercise II.9

Figure II.9d. Instantaneous Frequency Section Line A Part 1 Figure II.9e. Instantaneous Frequency Section Line A Part 2
(Tarner et al., 1979) (Tarner et al., 1979)

Question : Combine all sections in Figure II.9b-h and identify economic and non-economic gas interval

Seismic Attributes for Reservoir Analysis

By : Sigit Sukmono 72
Exercise II.9

Figure II.9f. Apparent polarity section Line A Part 1 Figure II.9g. Apparent polarity section Line A Part 2
(Tarner et al., 1979) (Tarner et al., 1979)

Question : Combine all sections in Figure II.9b-h and identify economic and non-economic gas interval

Seismic Attributes for Reservoir Analysis

By : Sigit Sukmono 73
Exercise II.10
Delineation of Thin Sand Reservoir using the Instantaneous
Frequency and Amplitude

Objective
To understand the application of instantaneous frequency and
amplitude attributes for delineating thin sand reservoir.
Materials
Reservoir to be studied is a gas reservoir produced from one
formation where a discontinuous thin sand reservoir is intercalated
with shales of the fluvial, delta plain and interdistributary bay
deposits. Figure II.10.1 shows the log section of line 10. The target
interval is a sediment packages with thickness vary from 46 to 61
meter comprising of intercalated sandstones, siltstones and shales. 0
Four zones of potential sandstone reservoirs are contained in this
interval where each zone has a thickness between 4-8 meter. Facies
interpretation of the log data shows that there are three log facies :
•Log Facies – 1 : Predominantly is shale to silty shales. 50

Feet
•Log Facies – 2 : A thin clean sandstone within a thick shale
1 2 3
•Log Facies – 3 : Amalgamated siltstone and sandstone.
100
The gas reservoir in the target interval is generally as log facies-2.
The log facies – 3 is not productive since there is no proper amount
of seal.
150
Question Productive
Combine the available data to map the facies – 1, 2 and 3.
Figure II.10.1. Log section of Line 10 and three facies
recognizable from SP log
Seismic Attributes for Reservoir Analysis.

By : Sigit Sukmono 74
Figure II.10.2 show cross-plots between the average amplitudes, percentage of sandstones and instantaneous frequency. Figure II.10.3 show the
seismic line 10 and 20. Figure II.10.4 provide the instantaneous frequency section overlain by real seismic of line 10.

Figure II.10.2. Cross-plot of average amplitude vs % sandstone

Seismic Attributes for Reservoir Analysis

By : Sigit Sukmono 75
S N S N
1000 1000

1250 1250

1500
1500

Figure II.10.3. Seismic section Line -20 showing the interval target.

Seismic Attributes for Reservoir Analysis

By : Sigit Sukmono 76
S N

1000

14
15 2 1200

24 6

1400

Line 10

Figure II.10.4. Cross-plot of average amplitude vs RMS Figure II.10.5. Display of instantaneous frequency overlain by real
instantaneous frequency seismic trace.

Seismic Attributes for Reservoir Analysis

By : Sigit Sukmono 77
Figure II.10.6 shows the average amplitude map within the 40 ms analysis window interval and it’s relationship with the results of lithofacies
analysis based or SP log. Figure II.10.7 provides an instantaneous RMS frequency map along a 40 ms analysis windows interval.

Figure II.10.7. RMS instantaneous


frequency in the 40 ms window interval
within the target zone

Figure II.10.6. Average amplitude map in window 40 ms around the


target interval and relationship between SP log lithofacies and average
amplitude value. Wells 6-14-23 are productive

Seismic Attributes for Reservoir Analysis

By : Sigit Sukmono 78
Exercise III.1

Figure III.1. Dip/azimuth time slice map at 1000 ms. : interpret the strike-dip and type of fault system

Seismic Attributes for Reservoir Analysis

By : Sigit Sukmono 79
Exercise III.2

M N
S

slice
2000
ms

Figure III.2.1. Cosine phase section of traverse path N-S through M well. Interpret the reef, fore-reef, back-reef and
barrier reef seismic facies

Seismic Attributes for Reservoir Analysis

By : Sigit Sukmono 80
Exercise III.2

Figure III.2.2. Variance cube time-sliced at 2000ms. Black is high


dissimilarity, white is low dissimilarity. Interpret the reef, fore-reef,
back-reef and barrier reef seismic facies. Position of line S-N is
given.

Seismic Attributes for Reservoir Analysis

By : Sigit Sukmono 81
Exercise III.3

high

low
50

45

40

35
Variance x Amplitude

Gross Sand
30

25

20

15

10

0
0 0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16 0.18 0.2

RMS of Variance x RMS of Amplitude

Kross Korelasi Linear (Kross Korelasi)

MLD-02
MLD-03

Figure III.3.1. 3D visualization of variance x amplitude. Interpret the extent of paleochannel, paleohigh and the
thickness of the sand within the channel.

Seismic Attributes for Reservoir Analysis

By : Sigit Sukmono 82
Fundamentals of AI Seismic Inversion
Page

1. Definition 2
2. AI and Reservoir Characterization 4
3. Seismic Trace Convolutional Model 8
4. Type of Inversion Methods 13
4.1. Recursive 13
4.2. Model-based 14
4.3. Sparse-spike 24

Bibliography 33
Exercises 34

Fundamentals of AI Seismic Inversion

By : Sigit Sukmono
1
1. Definition
Basically, seismic inversion is the technique for creating sub-surface geological model using the seismic data as input and well data
as controls. The recovering of seismic record is a forward modeling. In this subject the data input is the AI or reflection coefficient
(RC) series of the earth layer itself which then forward modeled into the seismic records. The forward modeling algorithm, is a
convolution process between seismic wavelet passing thru the RC series of the earth. On the other hand, the seismic inversion is a
backward modeling, where the input is the seismic record that inverse modeled into the AI section. This inverse modeling
algorithm, basically, is a deconvolution between the seismic records and seismic wave which then produce the AI section (Figures
1-2). Major types of inversion techniques is shown in Figure 3.
Forward Modeling

AI1
FORWARD MODELLING BACKWARD MODELLING
AI2
AI3
Input EARTH MODEL SEISMIC RESPONSE
AI4

AI5

AI6
MODELING MODELING
Process ALGORITHM ALGORITHM
AI7
Model
control

SEISMIC
Output EARTH MODEL
RESPONSE

Backward Modeling

Figure 1. Diagram of forward and inverse modeling.

Fundamentals of AI Seismic Inversion

By : Sigit Sukmono
2
Earth * Wavelet = Seismic Section
Seismic
Inversion Method
Bum!
The Making of
Seismic Section

Pre-Stack Pos-Stack
Inversion Inversion

Seismic Section/ Wavelet = AI

AI1
Seismic Tomography Amplitude
AI2
Inversion Time Inversion
AI3
Process Inversion (AVO, LMR, EI) Wavefield
AI4
Amplitude
Inversion Inversion

AI Earth
Seismic Inversion AI1
For Reservoir
AI2
Characterization Sparse-Spike Model-Based
AI3 Bandlimited
AI4

Figure 3. Illustration of seismic inversion process. Figure 4. Major types of seismic inversion techniques.

Fundamentals of AI Seismic Inversion

By : Sigit Sukmono
3
2. AI & Reservoir Characterization

AI is the product of rock density and P-wave velocity which means that AI is rock property and not an interface property
as the seismic reflection data. AI value controlled mostly by the P-wave velocity. Figure 5 shows the effect of many
factors in P-wave velocity.

Figure 5. Factors affecting seismic P-wave velocity (Hiltermann, 2001)

Fundamentals of AI Seismic Inversion

By : Sigit Sukmono
4
Figure 6. Illustration on the differences between normal seismic reflections and results of AI inversion for a low AI wedge
model. (a) Low AI wedge model. (b) Seismic synthetic with wiggle trace. (c) Seismic synthetic with color amplitudes. (d)
AI model. (Latimer et al., 2000)
AI inversion transforms seismic data into pseudo acoustic impedance logs at every trace. Figure 6a shows a low AI wedge model.
Figures 6b and 6c show seismic representation of the model in standard wiggle trace and color density with wiggles overlain. Notice
the tuning effect. Figure 6d shows the results of related AI inversion. The resulting inverted wedge is a more accurate spatial
representation of the original model and provide absolute values that match the original except in the location with tuning effect.

Fundamentals of AI Seismic Inversion

By : Sigit Sukmono
5
Another compelling reason for inverting seismic data is
illustrated in Figure 7. A synthetic seismic data set is shown
in panel 7b. It is constructed by convolving the AI model
and wavelet shown in panel 7a. The model contains three
interfaces : 50 ms, 135 ms and 230 ms. Note that each
interface represents the same change in absolute AI units but
in varying gradational degrees. The seismic data identify the
sharp interface at 50 ms. They identify the top of the second
interface at 135 ms, but it is not apparent that the interface is
a gradational coarsening upward sequence because the
seismic do not recognize the base of the event. The seismic
fail to identify the most gradual interface at 230 ms.
Compare the seismic response with that of the inverted
traces in panel 7c. The inverted trace data can effectively
model all these variations in rock properties because the
inverted data utilize a complete frequency range of 0-80 Hz.

Figure 7. AI model contains more information than seismic data


because they have a broader frequency content. (b) Synthetic seismic
data set based on the AI model in (d) and the wavelet in (a). Compare
the seismic responses to that of the inverted traces in (c) (Latimer,
2000)
Fundamentals of AI Seismic Inversion

By : Sigit Sukmono
6
Since AI is a layer property and seismic amplitudes are attributes of layer boundaries, then naturally AI section gives more detail
subsurface image than the normal seismic. As AI is closely related to lithology, porosity, pore-fluid then it is common to find strong
empirical relationships between acoustic impedance and one or more of these rock properties. AI model can provide the basis for the
generation of 3-D facies model and petrophysical property models. The volume results can be ported directly into reservoir simulator for
flow analysis.

Figure 8. Illustration on the differences between normal seismic reflections (wiggle) and results of AI inversion (color). In the
left panel, the reservoir is Bekasap (top Bekasap to top Bangko interval). In the right panel, the reservoir interval is as
indicated by the arrow. In both figures the inserted log is AI log. Question : delineate the most porous layer in Bekasap (Top
Bekasap – Top Bangko). Compare the reflectivity and AI section

Fundamentals of AI Seismic Inversion

By : Sigit Sukmono
7
3. Seismic Trace Convolutional Model

Seismic trace is the convolution of earth’s reflectivity


with a seismic wavelet with addition of noise component.

St = Wt * Rt + nt

Where St is the seismic trace, Wt is seismic wavelet, Rt


is reflectivity and nt = additive noise

When the noise component = zero, it can be simplified


into :
St = Wt * RCt

If we consider that the reflectivity consist of a RC at each


time sample and the wavelet to be a “smooth” function in
time, convolution can be thought of as ‘replacing’ each
reflection coefficient with a scaled wavelet and summing
the result (Figure 9).

Notice that the convolution with the wavelet tends to


‘smear’ the reflection coefficient. That is there is a total
loss of resolution, which is the ability to resolve closely
spaced reflector.
Figure 9. The convolution model in time domain

Fundamentals of AI Seismic Inversion

By : Sigit Sukmono
8
An alternative way of looking at the seismic 1.00
trace is in the frequency domain : 0.75

0.50
S(f) = W(f) x R(f)
0.25
Where S(f) = Fourier transformation of S(t) 0 Hz
W(f) = Fourier transformation of W(t) 0 50 100 150 200 250

R(f) = Fourier transformation of R(t ) (a) Wavelet Spectrum


f = frequency 1.00

0.75

Notice that the time domain problem of 0.50


resolution loss becomes one of frequency
0.25
content in the frequency domain. Both the high
and low frequencies of the reflectivity have 0 Hz
0 50 100 150 200 250
been severely reduced by the effect of the
seismic wavelet (Figure10) (b) Reflectivity Spectrum
1.00

0.75
To get a complete frequency spectrum of
0.50
geological model in the inversion results, the
low and high frequency components lost need 0.25
to be recovered, mainly using the well log data.
0 Hz
0 50 100 150 200 250

(c) Trace Seismic Spectrum

Figure 10. The convolutional model in frequency domain

Fundamentals of AI Seismic Inversion

By : Sigit Sukmono
9
Seismic are band-limited, missing the highest and lowest
frequencies. The band-limited nature of seismic data is
often considered in terms of the high frequencies and the
consequent lack of resolution. However, the low
frequencies missing from the seismic data are extremely
important if quantitative interpretation is required. This
is illustrated in Figure 11 by a single impedance layer
model, inverted for three different frequency ranges :
10-80 Hz, 10-500 Hz, and 0-80 Hz. A modeled AI layer
(well AI, in black) was used to derive a synthetic seismic
data set utilizing a Ricker wavelet comprising the
frequency range on the right. The synthetic seismic was
subsequently inverted back to AI. The resulting inverted
AI traces are red with the bandwidth of the inversion
annotated on the right.

When the seismic data are inverted using a wavelet with


frequencies of 10-80 Hz (Figure 11a), the approximate
thickness of the layer is accurately imaged, but the
absolute impedance values and the interface shape are
incorrect. When the wavelet frequency is increased to an
extreme of 500 Hz (Figure 11b), the results are capable
of resolving thinner beds but still do not accurately
represent the model. However, when low-frequency
information is included from additional sources, the
inverted data best represent the model (Figure 11c). This
demonstrates that low-frequency information is critical Figure 11. Illustration on the effect of low-frequency to
to a complete inversion results. the inversion results (Latimer, 2000).

Fundamentals of AI Seismic Inversion

By : Sigit Sukmono
10
Frequency Domain

Figure 12. The convolution process in time and frequency domains. Notice how the low frequency
component start to be affected by the sampling effect of RC and convolution of wavelet and RC (Jason
Geosystem, 1999).

Fundamentals of AI Seismic Inversion

By : Sigit Sukmono
11
Low frequency information can be derived from log data, pre-stack depth, time migration velocities, and/or a regional gradient.
Because many of these data are very low frequency (0-2 Hz), processing that preserves low frequencies is advantageous. High-
frequency information can be derived from well control or geostatistical analysis.

Figure 13. The difference of resolution content Figure 14. Illustration showing various frequency contents of
between seismic, well and inversion result (Jason well data (Jason Geosystem, 1999)
Geosystem, 1999)

Fundamentals of AI Seismic Inversion

By : Sigit Sukmono
12
4. Recursive Inversion

Recursive inversion is the simplest inversion method. The Seismic Section


basic equation to determine the AI of a layer are :

Invert to
Z − Zi 1 + Ri 
n −1
Ri = i +1 Z n = Z1 ∏ 
Reflectivity
Z i +1 + Z i 
i =1 1 − Ri  Enter Low Freq
Component
`
Question
Convert to Pseudo - IA
If we know that the Z1 = 1, R1 = 2/4, R2 =1/7, and R3 = -3/5.
Compute Z4.
Figure 14. The recursive inversion technique

Figure 15 shows the flowchart of this technique. Reflectivity series can be obtained by deconvolving the seismic data with the wavelet
(see exercise 1). If no low frequency component is recovered, this technique is also known as ‘bandlimited’ inversion because it invert the
seismic trace itself, so the AI trace result has the same frequency range as the seismic trace.

The main weakness of this technique is that it doesn’t accommodate the geology control and, therefore, it almost identical to the forward
modeling. Low and high frequency components from earth reflectivity which lost when the reflectivity is convoluted with wavelet, also
difficult to be recovered with this technique, so the ability of this technique to laterally predict the AI is not good. Because the equation is
applied recursively from top to bottom, the error effect will be accumulated. The noise on seismic trace will be interpreted as a reflection
and involved in the inversion.

Fundamentals of AI Seismic Inversion

By : Sigit Sukmono
13
5. Model-Based Inversion

On recursive method, the inversion result is affected by Seismic Extract Model Estimate
noise, bad amplitude recovery, and band limited Trace Wavelet Trace Impedance
seismic data. It means, all problems in the data itself
will be involved in final inversion result. To solve this
problem, model-based inversion technique is
developed with task-flow as follows (Figure 15): Compute
1. Make the initial model and its blocky version by Errors Revise
averaging the AI value according to the given Impedance
block size.
2. Convert the AI into reflectivity and convolute with No
Acceptable
the estimated wavelet to recover the synthetic
Error ?
model trace.
3. Subtract the seismic synthetic trace from real
Yes
seismic trace to get the trace ‘error’.
4. Update the AI model and its thickness iteratively Solution =
by using the GLI (Generalized Linear Inversion) Estimation
inversion method, so the error decreases.
5. Iterate until a good solution obtained.
Display

Figure 15. The model based inversion technique flow-chart.

Fundamentals of AI Seismic Inversion

By : Sigit Sukmono
14
The application of this technique starts by creating an
initial geology model which then updated in several stages.

The geological model is developed into three stages:


Zero-phase wavelet from seismic
1. Add the velocity control (and also density, if
necessary) on the inverted seismic line. This velocity
control can be added from well data or T-VRMS. If one
control point is added, the velocity is extrapolated on
two ways. If more than one control points are added,
the velocity is interpolated around it.
2. Stretch and squeeze the log data in the control points
to tie to the seismic data by using reflectivity Zero-phase wavelet from seismic & well data
convoluted with wavelet from seismic data (Figure
16-17).
3. Add the lateral control of main seismic reflector by
picking and develop the interpolation of well log in
such away, so it match to the reflector. This stage is
also known as the initial model development stage
(Figure 18-19). Wavelet after stretch and squeeze

Figure 16. Effect of stretch & squeeze to the wavelet (Russel, 1997)

Fundamentals of AI Seismic Inversion

By : Sigit Sukmono
15
Figure 17. Example of stretch & squeeze process in well-seismic tie (Russel, 1997)

Fundamentals of AI Seismic Inversion

By : Sigit Sukmono
16
Seismic stratigraphy concept is accommodated in lateral control
development of this initial model. Figure 18 left shows a good
well-seismic tie, but a simple extrapolation on lateral directions
will produce ‘box’ model. Figure 18 right shows a good initial
model after accommodating seismic stratigraphy model in
initial model construction.

The discussion above suggests how important is the role of


control point and horizon. For control horizon, the best way is
utilizing sharp reflector which related to a certain sequence
stratigraphy event as it represents a certain time line (for
example sequence boundaries, MCS, etc.).

After the initial model developed, it can be used for many


purposes, depending on the inversion method used. The
bandlimited inversion only use the low frequency component
model, while on model-based method, the procedure can be
summarized as follows :
1. Make the blocky version from the model by averaging the
AI along the layer. The layer could be as small as 1 sample
(a case where the blocking doesn’t happen), but normally
on the range of 3-5 samples.
2. Change the AI into reflectivity and convolute with seismic
wavelet to produce synthetic trace.
3. Subtracts the synthetic trace from real trace to produce
trace error.
4. Modify the AI and thickness of each layer so the error
decrease.
5. Iterate until a satisfying solution obtained.
Figure 18. Illustration of poor (upper panel) and good
Fundamentals of AI Seismic Inversion (lower panel) initial model construction (Russel, 1997)
By : Sigit Sukmono
17
In this technique, a direct inversion of seismic data itself is avoided. To implement the approach in figure 15, we need to answer
two main questions :
1. How is the mathematical relation between model data and seismic data ?
2. How to update the model ?

To control the effect of geological model data and seismic data to the resulted impedance model, a mathematical function is
applied by minimizing the objective function :

J = weight1 x (T-W*RC) + weight2 x ( M-H*RC)

where : T = seismic trace, W= wavelet, RC= Final Reflection Coefficient, M= Initial AI model estimation, H= integration
operator which convolute with the final reflection coefficient to get the final AI.

Weight 1 and weight2 determine how both part is balanced. In stochastic inversion, the objective function used is exactly as in the
equation. But other model-based inversions use only the second weight, or the stochastic input value changed into zero, so the
seismic trace role dominate the equation.

If these values is one, the initial model role would be dominated. The total of first and second weight must equal to one. It is
called as soft-constraint because the final model can change into any value compared to the initial model. On the hard-constraint
inversion, the algorithm is limited to keep the final AI value on given boundary by the AI maximum change. Practically, the
inversion with constraint usually more preferable than the stochastic inversion because the change of maximum impedance
parameter is more important than the change of constraint model parameter on stochastic method.

The block size affects the final inversion result. Initial estimated model is blocked into a line of blocks in the same size. The
final inversion result may change the block size, meanings that some blocks become bigger and other smaller, but the average
size is kept constant. Using the smaller block will increase the conformities between input trace and final synthetic trace.

Figure 19-21 illustrating the model-based inversion steps.

Fundamentals of AI Seismic Inversion

By : Sigit Sukmono
18
Figure 19. Illustration on how to develop the initial model

Fundamentals of AI Seismic Inversion

By : Sigit Sukmono
19
Figure 20. The example of inversion result : a) Bandlimited, b) Constrained model-based, c) Stochastic model
based, and d) sparse-spike MLH. Analyze the difference of each method and give the explanation. The example is
taken from Arief (2001)

Fundamentals of AI Seismic Inversion

By : Sigit Sukmono
20
Figure 21. Example of trace error
display a) constrained model-
based, b) Stochastic model-based,
and c) Sparse-spike MLH.
Analyze the difference of each
method and give the explanation.
The example is taken from Arief
(2001)
Figure 22a. Illustration showing the iteration number effect to the error value

Fundamentals of AI Seismic Inversion

By : Sigit Sukmono
22
The number of iteration needed for the solution to
converge depends on the block size. A method to
determine whether the iteration is already sufficient, is
by checking the plot error (Figure 22a-b). The error-plot
can also be utilized to determine the best inversion 3

technique. The error analysis normally done in key lines


passing through well controls only.
2.5

Two main problems in model-based inversion are : 2

1. Sensitive dependency to the wavelet (two different


wavelets can produce the same seismic traces).

error
1.5

2. Non-unique solution. Certain wavelet can give


appropriate solution with the trace on well location.
1

Thus, the final inversion result depends on two factors : 0.5

1. Initial model quality


2. Seismic data quality 0
0 20 40 60 80 100 120 140 160
iteration
In the best scenario, both factors will support each other
1. Model based. 2. Bandlimited. 3. Sparse spike (maximum likelihood).
and give the same result. In the worst scenario, they will
give contradicting information about the sub-surface
model, and never give a satisfying solution.
Figure 22b. Cross-plot of error vs iteration to
determine the best technique and iteration
number.
Fundamentals of AI Seismic Inversion

By : Sigit Sukmono
23
6. Sparse-Spike Inversion Method
As discussed previously, the recursive method of seismic
inversion is based on the classic deconvolution techniques,
which assume a random reflectivity and zero or minimum-
phase wavelet. They will produce higher frequency wavelet
on the output, but never recover the complete RC series.
More recent deconvolution techniques may be grouped under
the category of sparse-spike method because these methods
assume a certain model of the reflectivity and make a
wavelet estimate based on this model assumption.

These techniques includes :


1. Maximum-likelihood inversion and deconvolution
2. Norm L1 inversion and deconvolution
3. Minimum entropy deconvolution

From the point of view of seismic inversion, sparse-spike


methods have an advantage over classical methods of
deconvolution because the sparse-spike estimate, with extra
constraints, can be used as a full bandwith estimate of the
reflectivity. Figure 23 illustrates the fundamental assumption
of maximum-likelihood deconvolution, which is that the
earth’s reflectivity is composed of a series of large events
superimposed on a Gaussian background of smaller events.
This contrasts with spiking decon, which assumes a perfectly
random distribution of reflection coefficient. The sonic-log
reflectivity at the bottom of figure shows that in fact this
model is quite logical. Figure 23. The fundamental assumption of the maximum-
likelihood method (Russel, 1997)
Fundamentals of AI Seismic Inversion

By : Sigit Sukmono
24
The sparse-spike inversion assumes that only the big spike is
important. This method locate the big spike by checking the
seismic trace. Reflectivity series reconstruct one spike each
at a time. The spike added until the trace modeled accurately.
The sparse-spike inversion use the same parameter as the
model-based inversion with constraint. The additional
parameter which must be added is the parameter to determine
how may trace would be determined on each trace. The
parameter includes maximum number of spike and threshold
of spike detection. Each new spike addition, the trace will be
modeled accurately. The new spike is smaller than the
previous ones. Geologically, the large reflectors correspond
to the unconformities and major lithologic boundaries.

On the maximum-likelihood method, the main algorithm is


SLMA (“single likely most addition”). Figure 39 - 42
illustrate the principle of this method. The principle of L1
norm method, basically, is the same as the MLH method and
illustrated further in Figure 43-56.

Figure 24. The Single Most Likely Addition (SMLA)


algorithm illustrated for a simple reflectivity model
(Russel, 1997)

Fundamentals of AI Seismic Inversion

By : Sigit Sukmono
25
Initial
Wavelet
Estimate

Estimate Improve
Sparse Wavelet
Reflectivity Estimate

Figure 25. Component used for solving


both reflectivity and wavelet. Iterate
around the loop until converge.

Figure 26. The procedure for updating the wavelet in the


maximum-likelihood method (Russel, 1997)

Fundamentals of AI Seismic Inversion

By : Sigit Sukmono
26
Figure 27. The initial seismic model using extracted wavelet (Russel, 1997)

Fundamentals of AI Seismic Inversion

By : Sigit Sukmono
27
Figure 28. Final deconvolved seismic using zero-phase wavelet (Russel, 1997)

Fundamentals of AI Seismic Inversion

By : Sigit Sukmono
28
Figure 29. The philosophy of sparse spike inversion method using L1 norm which update the reflectivity until small error
between real data seismic and the model obtained (Jason Geosystem, 1999)

Fundamentals of AI Seismic Inversion

By : Sigit Sukmono
29
Figure 30. Illustration of seismic stratigraphy control in the Figure 31. The geology impedance model. Log AI from the
initial model construction (Pendrel & Riel, 2000) well is interpolated by following the control horizon in
Figure 30 (Pendrel & Riel, 2000)

Fundamentals of AI Seismic Inversion

By : Sigit Sukmono
30
Figure 32. Low frequency (0-10 Hz) component from Figure 33. Illustration on how to control the hard
impedance model in Figure 31. This frequency component will constraint. The constraint range determine how far
be united with the inversion result to the get sub-surface image the solution can be varied against the well data
with complete frequency spectrum (Pendrel & Riel, 2000) (Pendrel & Riel, 2000)

Fundamentals of AI Seismic Inversion

By : Sigit Sukmono
31
Figure 35. The illustration of final inversion result
Figure 34. The illustration of final inversion result
after combined with low frequency model (Pendrel &
compared to the AI log (Pendrel & Riel, 2000)
Riel, 2000).

Fundamentals of AI Seismic Inversion

By : Sigit Sukmono
32
Bibliography
1. Angeleri, G.P. and Carpi, R., 1982, Porosity Prediction from Seismic Data : Geophysical Prospecting, v.30, 580-607.
2. Brown, A.R., 1991, Interpretation of Three-Dimensional Seismic Data, AAPG Memoir 42, 3rd ed., AAPG.
3. Galbraith, J.M., and Millington, G.F., 1979, Low Frequency Recovery in The Inversion of Seismograms : Journal of
CSEG, v. 15, p.30-39.
4. Johnston, D.H., 1992, Introduction to Reservoir Management in Reservoir Geophysics, ed : R.E. Sheriff : SEG
5. Kallweit, R.S. and L.C. Wood, 1982, The Limits of Resolution of Zero-phase Wavelets : Geophysics, v.47, p. 1035-1046.
6. Kelkar, M., 1982, Applied Geostatistics for Reservoir Characterization, The University of Tulsa, Oklahoma.
7. Lindseth, R.O., 1979, Synthetic Sonic Logs - a Process for Stratigraphic Interpretation : Geophysics, v.44, p.3-26.
8. Lines, L.R. and Treitel, S., 1984, A Review of Least-Squares Inversion and Its Application to Geophysical Problems :
Geophysical Prospecting, v.32, p.159-186.
9. Martinez, R., 1985, Expand Abstracts of The SEG Annual Meeting in Washington, 461-464.
10. Meckel, L.D.Jr. and A.K.Nath, 1977, Geologic Considerations for Stratigraphic Modelling and Interpretation, in C.E.
Payton, ed., Seismic Stratigraphy – Applications to Hydrocarbon Exploration : AAPG Memoir 26, 417-438.
11. Neidell, N.S. and E. Poggiagliolmi, 1977, Stratigraphic Modelling and Interpretation - Geophysical Principles and
Techniques, in C.E. Payton, ed., Seismic Stratigraphy – Application to Hydrocarbon Exploration : AAPG Memoir 417-438.
12. Oldenburg, D.W., Scheuer, T and Levy, S., 1983, Recovery of The Acoustic Impedance from Reflection Seismograms,
Geophysics, v.48, p. 1318-1337.
13. Richardson, J.G., and Sneider, R.M., 1992, Synergism in Reservoir Management, in Reservoir Geophysics, ed : Sherrif,
R.E., SEG.
14. Russel, B.H., 1988, Introduction to Seismic Inversion Methods (ed : S.N. Domenico), SOC. Of Exploration Geophysicists.
15. Russel, B.H., 1995, Introduction to Seismic Inversion Methods, STRATA workshop.
16. Russel, B.H. and Lindseth, R.O., The Information Content of Synthetic Sonic Logs – A Frequency Domain Approach,
EAEG, Frances.
17. Sheriff, R.E., 1992, Reservoir Geophysics, SEG.
18. Sneider, R.M., 1990, The Economic Value of A Synergistic Organization : presented at the 1990 Archie Conference,
Houston.
19. Sukmono, S., 1999a, Interpretasi Seismic Refleksi, Jurusan Teknik Geofisika – ITB.
20. Sukmono, S., 1999b, Seismic Stratigrafi, Jurusan Teknik Geofisika – ITB.
21. Yilmaz, O., 1987, Seismic Data Processing, Society of Exploration Geophysicists.

Fundamentals of AI Seismic Inversion

By : Sigit Sukmono
33
Exercises

Fundamentals of AI Seismic Inversion

By : Sigit Sukmono
34
Exercise I. Inversion Quality Analysis

Figure I.1 shows the reflectivity and its


AI sections. The AI section recovered
by applying the sparse-spike inversion
method. The target reservoir is reef.
Question : Show one character of DHI
and propose next exploration well
location. Is the inversion result is
correct? Give the explanation.

Fundamentals of AI Seismic Inversion

By : Sigit Sukmono
35
Exercise II.
AI for Channels Sandstone Reservoir Mapping

Figure II.1 is the example of well-seismic tie on


Well-1. The reservoir mapped here is sand-1. Figure
II.2 shows the correlation between well AI and
seismic AI. Figure II.3 shows correlation between
AI and seismic amplitude, while Figure II.4 shows
correlation between well AI, depth and type of
lithology. Figure II.5 shows reservoir minimum-
amplitude map with 20 ms window width (10 ms
above and below the sand-1). On the figure also Sand-1 20 ms
shown the location of where the seismic inversion is
run and the result is shown on Figure II.6. Notice
that the seismic inversion only implemented on the
area with good well control and the result is used for
mapping the reservoir on surrounding area which
has relatively few well data control.

Question
In Figure II.5 and II.6, map the sand-1, coal-1, and
shale-1 areas. To simplify, divide by 3 to convert
from ft to m.
Figure II.1. Example of well-seismic tie
showing the targeted sand

Fundamentals of AI Seismic Inversion

By : Sigit Sukmono
36
4000

8000

AI-seismic (m/s gr/cc)


AI-well (m/s gr/cc)

3000

7000

2000
5000

3000 4000 5000 -15000 -5000 5000 15000

AI-seismic (m/s gr/cc) AI-seismic (m/s gr/cc)

Figure II.2. Correlation between Seismic AI vs Figure II.3. Correlation between Seismic AI vs
Seismic Amplitude in Well-1 Seismic Amplitude in Well-1

Fundamentals of AI Seismic Inversion

By : Sigit Sukmono
37
IA coal-1 = 9000-19000, IA sand-1 = 19000-23500, IA shale-1 : 24000-33000 ft/s * g/cc

shale-1

coal-1 sand-1

Figure II.4. Correlation between Well AI vs Depth vs.Lithology type in well-. Divide by 3 to convert from ft to m.

Fundamentals of AI Seismic Inversion

By : Sigit Sukmono
38
-20000

-17500

-15000

-12500

-10000

-7500

-5000

-2500

Figure II.5 Minimum amplitude of sand-1 with 20 ms window. The blue box shows area where the inversion process was held

Fundamentals of AI Seismic Inversion

By : Sigit Sukmono
39
Figure II.6. The average AI value of sand-1 with 20 ms window

Fundamentals of AI Seismic Inversion

By : Sigit Sukmono
40
Exercise III. AI for Lithology Mapping

Figure III.2 and III.3 show the reflectivity, AI and density sections. AI section is recovered by applying sparse-spike
inversion. The target reservoir interval also shown. Figure III.1 shows the cross-plot between AI and density.
Question : Select the best section and identify porous sandstone, tight sands and clay/tuff.

2.48

2.44

2.40

2.36

Figure III.1. The cross-plot between AI and density (Verdin, 1999) for overall compartment.
Fundamentals of AI Seismic Inversion

By : Sigit Sukmono
41
Figure III.2. The reflectivity (right) and its AI (left) sections. The Log shows sandstone fraction . This line is about 7 km
(Verdin, 1999)

Fundamentals of AI Seismic Inversion

By : Sigit Sukmono
42
Densitas
(g/cc)

2.50

2.40

Figure III.3 The density section converted from AI section. As a comparison, the AI section is given at
background. The black wiggle shows original seismic trace (Verdin, 1999). Show the porous sandstone, tight
sands and clay/tuff.

Fundamentals of AI Seismic Inversion

By : Sigit Sukmono
43
Exercise IV. AI for Development Well Delineation

Figure IV.1 shows the AI and the AI-porosity maps. Figure IV.2 show the porosity, NES and oil-isopach maps from well data.
Question : Determine 6 best development well locations. Compare if you don’t have the inversion result for this analysis.

Figure IV.1. The average seismic AI (left) and AI porosity (right). Low AI and porosity values are shown in bright colors.

Fundamentals of AI Seismic Inversion

By : Sigit Sukmono
44
Figure IV.2. a) Effective isoporosity (PhiE), b) Net Effective Sand Map, c) oil isopach from well data. High value is shown by bright
color.

Fundamentals of AI Seismic Inversion

By : Sigit Sukmono
45
Exercise V. Reservoir Quality Evaluation

Figure V.1 shows the reflectivity and related AI sections. The AI section recovered by applying the sparse-spike inversion
method. The interval of target reservoir is also shown. Figure V.2 shows the cross-plot between AI, porosity and gamma ray.

How is the inversion result quality? Give the explanation. Based on the cross-plot result, delineate a reservoir with good-
quality, e.g. the one which has high sand/shale ratio and porosity.

Fundamentals of AI Seismic Inversion

By : Sigit Sukmono
46
9400

8400

7400

Figure V.1. The reflectivity (left) and its AI section (right)

Fundamentals of AI Seismic Inversion

By : Sigit Sukmono
47
9900

8500

7100

5700

7000 8000
AI
Figure V.2. The cross-plot of AI vs porosity and gamma-ray

Fundamentals of AI Seismic Inversion

By : Sigit Sukmono
48
Exercise VI. Seismic Inversion for Reservoir Carbonate Characterization- Exploration Field Case

Figure VI.1. Example of well-seismic tie in Well-1. the target reservoir interval is shown.
Figure VI.2. Cross-plot of AI vs porosity.
Figure VI.3. Depth structure map of top X.
Figure VI.4. Map of average AI with 10 ms window below the top X.
Figure VI.5. The AI section through Well-1

Questions

1.Which part of the X reservoir is best to be developed ?


2. Suggest two best well development location !

Fundamentals of AI Seismic Inversion

By : Sigit Sukmono
49
top X 
base X 

Figure VI.1. The example of well-seismic tie in Well-1 Figure VI.2. The relationship between well AI
and well porosity

Fundamentals of AI Seismic Inversion

By : Sigit Sukmono
50
Figure VI.3. The depth structure map of top X

Fundamentals of AI Seismic Inversion

By : Sigit Sukmono
51
<7,750,000
7,750,000-
11,250,000

Figure VI.4. Map of average AI with window of 10 ms below the top X

Fundamentals of AI Seismic Inversion

By : Sigit Sukmono
52
Well-1

Top X Bottom X

Figure VI.5. The AI section through Well-1 (Line 9A)

Fundamentals of AI Seismic Inversion

By : Sigit Sukmono
53
Exercise VII. Model Trace Construction

CONSTRUCTING SYNTHETIC SEISMOGRAM or [A] *[B] = [C] = [a0b0 a0b1 +a1b0 a1b1]
The making of synthetic seismogram, basically, is the
convolution process between the RC and wavelet data. Notice that although [A] and [B] each only have two
Matrix operation is often used to do this convolution elements, but [C] has three elements. Generally, if the first
process. In physical definition, the convolution describes vector has n element, the second vector m element, then
behavior of how two energy wavelets combined. For the convolution result vector has n+m-1 element.
example if there are two vectors [A] = [a0 a1 a2 …] and [B] =
]b0 b1 b2…].
Robinson and Treitel (1980) introduced a simple graphic
method to do the two vectors convolution. For example if
Their convolution are indicated by operator *, for example two vectors, each with three elements, are convoluted,
[C] = [A] *[B] which will produce the vector [C] = [c0 c1 both are written in to a row-column product :
c2…]. The [C] element is given by :

i b0 b1 b2
ci = ∑ ajbi − j
a0  a 0b 0 a 0b1 a 0b 2 
 a1b 0 a1b 2 
j =0

For example, if we want to convolute two vectors [A] and


a1  a1b1
a 2b 0 a 2b 2 
[B]. If the [A] = [a0 a1] and [B] = [b0 b1], so the first,
second and third elements of the convolution result are : a2 a 2b1
c0 = a0b0 , c1 = a0b1 +a1b0 , c2 = a1b1

Fundamentals of AI Seismic Inversion

By : Sigit Sukmono
54
For example the vector [A] = [1 3 5 7 2], while the
The convolution product is graphically determined as
vector [B] = [6 2 4], with the graphic way it can be
the summing of diagonal elements
written as :

a 0b0 a 0b1 a 0b2  6 2 4


a b a b a b  1 6 2 4 
 1 0 1 1 1 2 3 18

6 12 

a 2 b0 a 2 b1 a 2 b2  5  30 10 20
 
7 42 14 28
and the resulting vector [C] has elements :
212 4 8 

[a0b0 a1b0 + a0b1 a2b0 + a1b1 + a0b2 a2b1 + a1b2 a2b2 ]

Thus, [A]*[B] = [C] = [6 20 40 64 46 32 8]

Fundamentals of AI Seismic Inversion

By : Sigit Sukmono
55
Question :
1. Figure below shows the log Vp, ρ, AI and wavelet data. The wavelet can be written as [W] = [-20 70 -20]
2. By using the Robinson and Treitel (1980) method, construct the synthetic seismogram (St) on the given zone in figure, where St =
RCt*Wt
3. Compare your synthetic seismogram with the computer calculation result shown below.

80

60

40
amplitude

20

0 20 40 60 80 ms
-20

-40

(a) (b)

Figure VII.1. (a) The wavelet used, simplified into a matrix form [Wt] = [-20 70 -20]; (b) Synthetic
created by computer using 2 ms sampling interval.

Fundamentals of AI Seismic Inversion

By : Sigit Sukmono
56
Figure VII.2. The velocity, density and AI logs. Construct St = Wt*RCt on the right
diagram.
Fundamentals of AI Seismic Inversion

By : Sigit Sukmono
57
Exercise VIII. Matrix Operation for Recursive Inversion

Robinson and Treitel (1990) give example on how to do the


recursive inversion using a simple matrix operation. If the
wavelet and seismic trace data are known, theoretically by using
the matrix operation we can compute the RC and AI values. For
example, in Figures 26-27, it is known that [Wt] = [-2 7 -2] and
[St] = [17 6 3]. the equation Wt* RCt = St can be written in
matrix as :
Wt RCt St

7 -2 0 RC1 17
-2 7 -2 RC2 6
0 -2 7 RC3 3

If the number of variable that we want to solve is only a few (for


example 3), the equation can be solved using common
substitution. However, if the variables is quite many, we need
special technique, such as Gauss-Jordan’s elimination method.

Figure VIII-1. The synthetic


seismogram created by computer,
simplified into [St] = [17 6 3]
Fundamentals of AI Seismic Inversion

By : Sigit Sukmono
58
Following is the discussion on the application of this Gauss- Next, normalized the second row by divide it with pivot element 25/2;
Jordan’s elimination. For example, the following equation then reduce other elements in second column into zero by subtracts -
needs to be solved : (7/2) times new second row from first row, and –(5/2) times new
2x1 - 7x2 + 4x3 = 9 second row from third row. Notice that the reduction process now
involves the sub diagonal and super diagonal elements. The result is :
x1 + 9x2 – 6x3 = 1
1 0 - 6/25 88/25 9/25 7/25 0
-3x1 + 8x2 + 5x3 = 6 0 1 − 16 / 25 - 7/25 - 1/25 2/25 0
In Gauss-Jordan’s elimination technique, the equation can  
0 0 47/5 94/5 7/5 1/5 1
be written as :

 2 −7 4 9 1 0 0
 1 9 −6 1 0 1 0
Finally, normalize the third row by divide it with pivot element 47/5;
  then, reduce the rest element in third column into zero by subtract –
− 3
 8 5 6 0 0 
1
(6/25) and-(16/25) times the new third row from first and second rows.
The resulted matrix is [I] x [B-1], where x is the solution vector and B-1
The first row is normalized by divide it using pivot element 2; is the inverse of original matrix coefficient :
and other elements in the first column is reduced into zero by
subtracting the new first row from the old second row, and also 1 0 0 4 93/235 67/235 6/235 
by subtracting –3 times the new first row from the old third row. 0 1 0 1 13/235 22/235 16/235
The result is :  
0 0 1 2 7/47 1/47 5/47 
1 − 7/2 2 9/2 1/2 0 0
0 −8 0

25/2 - 7/2 - 1/2 1
 Thus, we recover the x1 = 4, x2 = 1, x3 = 2.
0 - 5/2 11 39/2 3/2 0 1
Exercise
If [Wt] = [-2 7 -2] and [St] = [17 6 3], by using Gauss-Jordan’s
elimination technique, compute the RC1, RC2 and RC3 , and if known
that the AI0 = 6875, compute the AI1, AI2, and AI3. Compare your
result with the log data given in the Exercise VI..
Fundamentals of AI Seismic Inversion

By : Sigit Sukmono
59
AVO/AVA for Gas/Fluid Detection
Contents
Page

1. Preface 2
2. Mathematical Foundation 2
3. Offset to Angle Transformation 7
4. The Prediction of AVO Response 12
5. Ostrander’s Model 15
6. AVO Class 1, 2 and 3 24
7. AVO Anomaly Class 4 38
8. Relationship of AVO Gradient and AVO Intercept 40
9. AVO Cross-Plot 41
10.AVO Attributes 51

References 56
Exercise 57

AVO/AVA for Gas & Fluid Detection


1
By : Sigit Sukmono
1. Preface
Amplitude Variation with Offset (AVO) or Amplitude Variation
with Angle (AVA) become popular in petroleum exploration
B1 A1
industry since introduced by Ostrander (1984). The gas-sand
model used by Ostrander shows the increasing of reflection A0
amplitude with the increasing of offset or angle and the term of θ1
AVO/AVA originates from here. The need of AVO is improving
specially because of the ambiguity of amplitude anomaly to λ1
distinguish the anomaly from gas and the anomaly from 1) Low
impedance sandstone, 2) Shale, 3) Coal, 4) Porous carbonate and
Other lithology effect.

2. Mathematical Foundation
θ2
2.1. Zoeppritz Equation
One of the basic assumption in normal seismic is that the λ2
seismic wave strikes the rock layer at vertical incidence. In
this case, the reflection coefficient is given as following
A2
equation (1) : B2
AI i +1 − AI i
RCi = (1)
AI i +1 + AI i
When the seismic wave strike the boundary at non zero
incidence angles, the conversion of P to S wave will occur. As Figure 1. Illustration of how the P-wave strike the
the consequence, the reflection coefficient becomes a boundary and split into 4 waves. A1, A2, θ1 and θ2
function of the P-wave velocity, S-wave velocity and density are amplitudes and angles of reflected and
of each of the layers. Indeed, there are now four curves that transmitted P wave. B1, B2, λ1, and λ2 are amplitudes
can be derived : reflected P-wave amplitude, transmitted P- and angles of reflected and transmitted S wave.
wave amplitude, reflected S-wave amplitude, and transmitted
S-wave amplitude, as shown in Figure 1.

AVO/AVA for Gas & Fluid Detection


2
By : Sigit Sukmono
Following equation is the Zoeppritz equations which related to rays model shown in Figure 1.

 sin λr cos φr sin λt cos φt  − sin λr


− cos λ sin φr cos λt - sin λt   A  − cos λr
 r  
 α1 ρ 2 β 2 2 α1 - ρ 2 B2α1  B 
 sin 2λr β1
cos 2φr
ρ1β1 α 2
2
sin 2 λt
ρ1 β12
cos 2φ t
 C  = sin 2λr
  
β1 − ρ 2α 2 - ρ 2 B2
 cos 2φ sin 2φr cos 2 φt sin 2φt   D 
 r
α1 ρ1α1 α 2
2
ρ1 α1  = − cos 2φr

Where A = Rpp reflection, B = Rps reflection, C=Rpp transmission, D = Rps transmission, α=Vp, β=Vs, ρ=density.

The 4x4 series of linear equations shown above is a good way of deriving the exact amplitudes of a reflected P-wave as a
function of angle. But it does not give an intuitive understanding of how these amplitudes relate to the various physical
parameter. Several approximations were made for Zoeppritz equation to get better understanding on the relationships of the
observed amplitude with the physical properties.

AVO/AVA for Gas & Fluid Detection


3
By : Sigit Sukmono
2.2. Approximation to the Zoeppritz’s Equation b. The Smith/Gidlow Method
a. The Aki, Richard and Frasier Approximation Smith and Gidlow (1987) rearranged Aki-Richard’s equation in the
following way :
The Aki, Richards and Frasier approximation is appealing
1  ∆α ∆ρ  β2  ∆β ∆ρ  2 1 ∆α
because it is written as three terms, the first involving R (θ ) =  + −2 2 β + ρ sin θ + 2 α tan θ
2

density, the second involving Vp, and the third involving Vs. 2 α ρ  α 2  
Their formula can be written as the following equation (4) :
They then chose to remove the dependency on density by using
Δα Δρ Δβ Gardner’s equation :
R( θ) = a +b +c
α ρ β 1
ρ=cα 4

where : a = 1/(2 cos2 θ) = 1/2 + tan2 θ


b = 0.5 –[(2β2/α2) sin 2 θ] which can be differentiated to give :
c = -(4β2/ α 2) sin2 θ
α = (α 1+ α 2)/2 Δρ = 1 Δα
β = (β1+ β2)/2
ρ = (ρ 1+ ρ 2)/2
ρ 4 α
∆α = α 2- α 1 Substituting equations above, we can re-express Aki and Richard’s
∆ β = β 2- β 1 equation as the following weighted sum of P- and S- wave velocity
∆ ρ = ρ 2- ρ 1 variations.
θ = (θi+ θt)/2
Δα ∆β
θt = arc sin [(α 2/ α1)sin θi] R (θ ) = a +b
α β
5 β2 1
where a = - 2 sin 2 θ + tan 2 θ
8 α 2
β2
and b = - 4 2 sin 2 θ (8)
α
AVO/AVA for Gas & Fluid Detection
4
By : Sigit Sukmono
c. Shuey’s Approximation

1  Δα Δρ   1 Δα β 2 Δβ β 2 Δρ  2 1 Δα
R (θ ) = 
2 α
+ 
 + 
ρ  2 α
-4 2
α β
-2 2
α ρ 
 sin θ +
2 α
tan 2θ − sin 2θ ( )
Simplification is held by assuming that β/α = ½ (σ = ⅓) and ignore the third term(tan2θ ≈ sin2θ), which lead to :

1  Δα Δρ   1 Δα ∆β 1 Δρ 
R (θ ) =  +  +  - -  sin 2θ
2 α ρ  2 α β 2 ρ 

1  Δα Δρ  1  Δβ Δρ 
Rp =  + ; Rs =  + 
With simplification as follows : 2  α ρ  2  β ρ 

Then the previous equation can be rewritten as :

R(θ) = Rp + (Rp – 2Rs) sin2θ ; R(θ) = Rp + G sin2 θ ; where G = Rp-2Rs

AVO/AVA for Gas & Fluid Detection


5
By : Sigit Sukmono
Figure 2. The comparison of Zoepprit’s equation and Figure 3. The error values at top interface (Russel, 1998)
approximations (Russel, 1998)

AVO/AVA for Gas & Fluid Detection


6
By : Sigit Sukmono
3. Offset to Angle Transformation

For applying AVO’s equations, it is necessary to transform


seismic data from the offset domain to the angle domain. For
simple single horizontal layer, the following equation is
commonly used :

X
tan θ =
2Z

where: θ = angle of incidence, X = offset and Z = depth

When the velocity down to the layer of interest is known, the


above equation can be rewritten as :

or θ = tan -1 X 
Vt 0 X
Z = ; tan θ =
2 Vt 0 
 Vt 0

where : V = velocity (RMS or average), t0 = total zero offset


travel time

By using those equations, the offset can be transformed to


angle.
Figure 4. The illustration of a) AVO response and
b) transformation from a into AVA response
(Amplitude versus Angle) (Russel, 1998)

AVO/AVA for Gas & Fluid Detection


7
By : Sigit Sukmono
Figure 5. The schematic relationship of depth Figure 6. The correspondences of angle and offsets
and offset. (Russel, 1998)

AVO/AVA for Gas & Fluid Detection


8
By : Sigit Sukmono
An approximation that can be used for the multi layer case
involves using the ray parameter P and total travel time t :

P = sin θ/ VINT and t2 = t02 + X2/ VRMS2

where VINT = interval velocity for a particular layer, VRMS=


velocity down to the layer

Note also that P and t are related by the equation:

X X VINT
dt
= P; P= 2
; sin θ = 2
dx t VRMS t VRMS

To see that equation reduces to single layer case equation, refer


to Figure 7, notice that :

t0 = t cos θ

Thus, by substituting equation and noting that VRMS = VINT = V


for a single layer, we see that :

sinθ x Figure 7. The τ-p analysis of NMO curve


= tan θ =
cosθ Vto

AVO/AVA for Gas & Fluid Detection


9
By : Sigit Sukmono
Once we have transformed from offset to angle, we can use the
Shuey’s approximation, which were written :

R(θ) = Rp + G sin 2θ

where R(θ) = change of reflection coefficient with angle θ


Rp = P-wave reflection coefficient at normal incidence
G = gradient depending on change of Poisson’s ratio

An example of this curve plot is shown in Figure 8.

Figure 8. Example of the plot of amplitude versus sin2 θ

AVO/AVA for Gas & Fluid Detection


10
By : Sigit Sukmono
Using the values of Rp and G derived from the plot in (a)
Figure 8, a P-wave stack and an S-wave stack can be
derived from the seismic data. This was done using
Gelfand’s approximation for G :

G = Rp – 2 Rs

where Rs = normal incidence S-wave reflection


coefficient
(b)
Therefore Rs = (Rp - G)/ 2

Figure 9 shows a seismic section and the derived P and


S-wave stacks. A bright anomaly at about 1.25 seconds
indicates the possible presence of a gas sand.

(c)

Figure 9. Display of P (b) and S -wave (c) stacks which derived


from CMP stack input in figure (a) (Russel, 1998)
AVO/AVA for Gas & Fluid Detection
11
By : Sigit Sukmono
4. The Prediction of AVO Response

The Shuey’s equation of R(θ) = Rp + G sin2 θ shows that reflectivity is roughly parabolic as a function of angle and that its relative
change depends on the sign of the Poisson’s ratio change and the sign of Rp as displayed in Figure 10-11.

Figure 11. The summary of possible combination top


Figure 10. AVO response behavior due to the effects of
reservoir’s AVO response due to the AI and Poisson’s Ratio
Poisson’s Ratio and AI changes
changes (Russel, 1998)

AVO/AVA for Gas & Fluid Detection


12
By : Sigit Sukmono
Seismic Response Seismic Response
AI σ Offset AI σ Offset

Gas sand
Gas sand

Top sand relative amplitude : decreases to offset Top sand relative amplitude : ?
Top sand absolute amplitude : increases to offset Top sand absolute amplitude : ?

Figure 13. Given above is a high AI and


Figure 12. Simplified AVO response
low Poisson’s Ratio gas-fill tight sand
modeling for a low AI and Poisson’s Ratio
reservoir. Draw the simplified AVO
gas-fill very porous sand reservoir using
response using SEG reverse polarity and
SEG reverse polarity and zero-phase
zero-phase wavelet in right diagram. What
wavelet. In stack section it will likely
DHI signature will it produce in stack
produce bright-spot.
section?

AVO/AVA for Gas & Fluid Detection


13
By : Sigit Sukmono
Seismic Response Seismic Response Seismic Response
AI σ Offset AI σ Offset AI σ Offset

Gas reservoir
Gas sand

Gas sand
Top sand relative amplitude : ? Top sand relative amplitude : ? Top sand relative amplitude : ?
Top sand absolute amplitude : ? Top sand absolute amplitude : ? Top sand absolute amplitude : ?

Figure 14. Given above is a high AI Figure 15. Given above is a near-zero Figure 16. Given above is a low AI
and low Poisson’s Ratio gas-fill tight AI contrast and low Poisson’s Ratio and high Poisson’s Ratio gas-fill
sand reservoir. Draw the simplified gas-fill sand reservoir. Draw the reservoir. Draw the simplified AVO
AVO response using SEG reverse simplified AVO response using SEG response using SEG reverse polarity
polarity and zero-phase wavelet in reverse polarity and zero-phase and zero-phase wavelet in right
right diagram. What DHI signature wavelet in right diagram. What DHI diagram. What DHI signature will it
will it produce in stack section? signature will it produce in stack produce in stack section?
section?

AVO/AVA for Gas & Fluid Detection


14
By : Sigit Sukmono
5. Ostrander’s Model
Figure 17 the classic Ostrander’ss low AI porous gas sand which is a typical of young geology age and shallow sand reservoir. Here,
the gas-sandstone with Poisson’s ratio = 0.1 located under the shale with Poisson’s ratio = 0.4.In this model, P-seismic wave velocity is
reduced from shale to gas-sandstone in amount of 20 %, that is from 10,000 ft/sec to 8,000 ft/sec, and density decrease in amount of 10
% from 2.40 g/cm3 to 2.16 g/cm3. The reflection coefficient of top sandstone changes from –0.16 to –0.28, while the reflection
coefficient of base sandstone base changes from +0.16 to +0.26. The amplitude increase in the amount of 70 % in range of 40º
incidence.

Figure 17. Ostrander’s gas sand and its AVO response model. (Ostrander, 1984)

AVO/AVA for Gas & Fluid Detection


15
By : Sigit Sukmono
Figure 18 shows 24-fold CDP seismic section
passing thru the Sacramento Valley gas field.
The sandstone reservoir is located in depth about
6.700 ft which associated with seismic amplitude
anomaly in 1.75 seconds. This reservoir is a sea
fan deposited in Cretaceous which has maximum
net pay up to 95 ft. The existing traps are the
structure and stratigraphy traps. Faults exist in
SP 95, wedging reservoir is in SP 75. Seismic
wave velocity and density in gas-sandstone are
lower than the sealing-shale, which give strong
seismic reflection on top and base of gas-
sandstone. In 1.8 s between SP 115 and SP 135,
we see the flat spot phenomena. This phenomena
is also seen in SP 75 to SP 130.

CDP gather from three location, A, B, and C, are


shown in Figure 19, 20 and 21, respectively.
Both single-fold and 10-fold refer to location A
and B, and only 10-fold refers to location C.
Shot to the offset group for all gather increases
from right to the left.

Question :
Based on Figures 18 to 21, determine the extent
of gas filled sand ? Figure 18. The stacked-seismic section for line SV-1 (Ostrander, 1984)

AVO/AVA for Gas & Fluid Detection


16
By : Sigit Sukmono
Figure 19. CDP gathers for location A on line SV-1 (Ostrander, 1984)

AVO/AVA for Gas & Fluid Detection


17
By : Sigit Sukmono
Figure 20. CDP gathers for location B on line SV-1(Ostrander, 1984)

AVO/AVA for Gas & Fluid Detection


18
By : Sigit Sukmono
Figure 21. CDP gathers for location C on line SV-1 (Ostrander, 1984)

AVO/AVA for Gas & Fluid Detection


19
By : Sigit Sukmono
Figure 22 shows the anomaly of seismic
amplitude which related to gas in GM-1
seismic line, in Gulf of Mexico. The first
anomaly is located on the left of seismic
section at the time about 0.65 seconds. The
second anomaly is a deeper anomaly located
to the center of seismic section at the time
about 1.35 seconds.

Summed CDP gather is presented on Figure


23 for location A on shallow anomaly and on
Figure 24 for location B on deep anomaly.

Question :
Based on Figures 22 to 24, determine the
extent of gas filled sand ?

Figure 22. The stacked seismic section for line GM-1 (Ostrander, 1984)

AVO/AVA for Gas & Fluid Detection


20
By : Sigit Sukmono
Figure 23. CDP gather for location A on line GM-1 (Ostrander, 1984)

AVO/AVA for Gas & Fluid Detection


21
By : Sigit Sukmono
Figure 24. CDP gathers for location B on line GM-1 (Ostrander, 1984)

AVO/AVA for Gas & Fluid Detection


22
By : Sigit Sukmono
Question :

Given in Figure 25 below is the stack and related gathers section. Determine the extent of gas filled sand ?

Figure 25. The stacked seismic section for line FB-1 and related CDP gathers (Ostrander, 1984)

AVO/AVA for Gas & Fluid Detection


23
By : Sigit Sukmono
6. AVO Class 1, 2 and 3

Rutherford and Williams in 1989 classified


three classes of gas-sandstone reservoirs, i.e :

Class 1 : high impedance gas-sandstones


Class 2 : near zero impedance contrast gas-
sandstones
Class 3 : low impedance gas-sandstones

Figure 26 shows a set of AVO reflection


coefficient curves on the interface between
shale and gas-sandstone which computed in
reflection coefficient range on normal
incidence of RC0. The curves in Figure 26 are
computed based on the Poisson’s ratio of
shale and gas-sandstone, 0.38 and 0.15,
respectively, and the density of shale and gas-
sandstone, respectively, 2.4 and 2.0 g/cm3.

Figure 26. Illustration of Rutherford and Williams (1989)


AVO classification for top interface.

AVO/AVA for Gas & Fluid Detection


24
By : Sigit Sukmono
Sandstone Class 1 has relatively high impedance than its seal layer, which usually is shale. Interface between shale and this
sandstone will result relative high positive coefficient reflection (R0). This sandstone is normally a mature sandstone which have
been moderately to highly compacted. Coefficient reflection of high acoustic impedance sandstone is positive on zero offset and
decreases as the offset increases. Magnitude of amplitude change to offset (usually known as the term ‘gradient’) for sandstone
class 1 usually is bigger than gradient of sandstone class 2 or 3. The gradient depends on the RC0, generally, gradient will
decrease as the RC0 and Poisson’s ratio decreases.

Sandstone class 2 has almost the same Acoustic Impedance as the sealing rock. This sandstone is a compacted and moderately
consolidated sandstone. Gradient of sandstone class 2 usually has big magnitude, but generally it’s smaller than the magnitude of
sandstone class 1. Reflectivity of sandstone class 2 on small offset is zero. This is often blurred due to the presence of noise on
our data seismic. The reflectivity suddenly emerge on bigger offset, that is when the reflection amplitude is located on a higher
level than the noise. Polarity change happen if the RC0 is positive, but usually it’s undetected, because it happen on the near
offset where the signal level is under the noise. Sandstone class 2 might and might not be related to amplitude anomaly on stack
data. If the angular range is available, so the amplitude will rise as the offset increasing, the amplitude anomaly on stacked data
is apparent.

Sandstone class 3 has lower acoustic impedance than the seal rock. Usually this sandstone is the less compacted and
unconsolidated sandstone. On seismic stack data, sandstone class 3 has big amplitude anomaly and reflectivity in the whole
offset. Usually, the gradient is significant enough but it has lower magnitude than the sandstone class 1and 2. The RC at normal
incidence angle is always negative. In some conditions, relatively small change of amplitude to offset can cause detection
difficulties because the presence of tuning thickness, attenuation, recording array, and decreasing of signal-to-noise to the offset
ratio. Sandstone class 3 sometime has high amplitude response which relatively flat and continuous along the offset.

AVO/AVA for Gas & Fluid Detection


25
By : Sigit Sukmono
Exercise :
Modeling of AVO response using sonic and density log data.

Questions:

See the log data in the following pages :


1. In the table below, compute the reflection amplitude against the angle of incidence by using Shuey’s approximation :
R(θ) = Rp + (Rp – 2Rs)sin2 θ, for θ = 0º, 10º, 20º, 30º.
2. Plot the amplitude to the angle in the Figure 27. What type of gas sandstone is this?

Lithology Vp Vs ρ Zp Zs Rp Rs 2Rs Rp-2Rs


Shale 10000 5000 2.6
Sandstone 12000 7500 2.2
θ sin (θ) sin^2(θ) R(θ)
0
10
20
30

AVO/AVA for Gas & Fluid Detection


26
By : Sigit Sukmono
Vp = 10.000 ft/s ; ρ = 2.6 g/cc ; Vs = 5000 ft/s

Vp = 12.000 ft/s ; ρ = 2.2 g/cc ; Vs = 7.500 ft/s

Figure 27. Curves log of Sonic P, Sonic S, and Log density resulting from measurement in Well ‘S’ of field X

AVO/AVA for Gas & Fluid Detection


27
By : Sigit Sukmono
Amplitude-Incident Angle
0.02
0.015
0.01
0.005
0
-0.005 0 5 10 15 20 25 30 35
-0.01
Amplitude

-0.015
-0.02
-0.025
-0.03
-0.035
-0.04
-0.045
-0.05
-0.055
-0.06
Incident Angle

Figure 28. Diagram of amplitude-incident angle cross plot

AVO/AVA for Gas & Fluid Detection


28
By : Sigit Sukmono
Question :

Given in Figure 29 below is the stack section whose AVO modeling and related gathers section are given in Figures 30-31. Is there any gas
sand and what class if any ?

Figure 29. A stacked seismic section on Hartshorn field. The productive interval corresponds to the dim out phenomena which highlighted
in the figure. The dim out is caused by a change in polarity to with offset of the Hartshorn reflection
(Rutherford and Williams, 1989)

AVO/AVA for Gas & Fluid Detection


29
By : Sigit Sukmono
Figure 30. The CMP gather modeling. (a) P-wave sonic log through the Hartshorn interval. (b) estimated S-wave sonic
log (c) the computed CMP gather (d) stack of the traces in (c). (Rutherford & Williams, 1989)

AVO/AVA for Gas & Fluid Detection


30
By : Sigit Sukmono
Figure 31. Three CMP gathers show the change in polarity which associated to Hartshorn gas-sandstone. The
last trace on each gather of far offset were distorted by NMO stretching (Rutherford & Williams, 1989)

AVO/AVA for Gas & Fluid Detection


31
By : Sigit Sukmono
Question :
Given in Figure 32 below is the stack section whose AVO modeling and related gathers section are given in Figures 34-35. Is there any gas
sand and what class if any ?

Figure 32. Migrated-stacked section on Miocene gas- Figure 33. Diagram of Brazos gas-sandstone
sandstone in Brazos Gulf of Mexico. The reflector of interest (Rutherford & Williams, 1989)
is at about 2.1 s on the section. (Rutherford & Williams,
1989)

AVO/AVA for Gas & Fluid Detection


32
By : Sigit Sukmono
Figure 34. Synthetic model section on Figure 32 which computed by using the log from Miocen sandstone well as shown on
Figure 32. The three sections are related to 0 m, 1524 m and 3048 m offsets (Rutherford & Williams, 1989).

AVO/AVA for Gas & Fluid Detection


33
By : Sigit Sukmono
Figure 35. Panel display of constant
reflection angle sections corresponding
to Figure 32. The angles posted on the
right side of figure refer to the centers
of the angle ranges in each panel. Each
1.0 s panel displays 1.6 s to 2.4 s data
(Rutherford & Williams, 1989)

AVO/AVA for Gas & Fluid Detection


34
By : Sigit Sukmono
Question :

Given in Figure 36 below is the stack section whose related gathers section are given in Figures 37. Is there any gas sand and what class if
any ?

Figure 36. Migrated, stacked seismic section on Pliocene gas-sandstone in High Island Gulf of
Mexico. The reflector of interest is between 2.3 s and 2.5 s (Rutherford & Williams, 1989)

AVO/AVA for Gas & Fluid Detection


35
By : Sigit Sukmono
Figure 37. Panel display of constant reflection angle
sections. The angles posted on the right side of the figure
refer to the centers of of the reflection angle ranges in each
panel. Each 1.0 s panel displays 2.0 s and 2,8 s of live data
(Rutherford & Williams, 1989)

AVO/AVA for Gas & Fluid Detection


36
By : Sigit Sukmono
Exercise

Figure 38 on the right


is a CDP gather with
normal SEG polarity.
Pick the tops and
bottoms of one class III
gas-sands and three
class I gas-sands. Be
careful with NMO
stretch effect.

AVO/AVA for Gas & Fluid Detection


37
By : Sigit Sukmono
7. AVO Anomaly Class 4

The fourth class of gas-sandstone anomaly associate with


the reflection coefficient which becoming more positive as
the offset increases, but the magnitude decreased as the
offset increases. Sandstone class 4 often emerged when the
porous sandstone, bounded by lithology which has high
seismic wave velocity, such as hard shale (e.g : siliceous or
calcareous), siltstone, tightly cemented sand, or carbonate.

Sandstone class 4 has big negative Intercept (A) and positive


Gradient (B). Sandstone class 4 is the bright spot, but the
reflection magnitude decreases as the offset increases.
Shuey’s (1985) approximation of two terms is good enough
on incidence angle about 300. For smaller angle, B is
positive for brine and gas sands, and the magnitude
decreased as the offset increases.

Figure 39. Reflection coefficient of wave against the incident angle


for top gas- sandstone class IV (quadrant II), and related brine-sand
reflection. The model parameters are : shale-VP = 3.24 km/s, VS =
1.62 km/s, ρ = 2.34 gm/cm3; brine sand-VP = 2.59 km/s, VS = 1.06
km/s, ρ = 2.21 gm/cm3; gas-sand-VP = 1.65 km/s, VS = 1.09 km/s,
ρ = 2.07 gm/cm3
AVO/AVA for Gas & Fluid Detection
38
By : Sigit Sukmono
Class Relative Quadrant A B Note
Impedance

I Higher than IV + − Coefficient


the sealing Reflection (and
layer magnitude)
decreases as the
offset increases
II Almost the III or IV ± − Reflection
same with Magnitude
the sealing decreases or
layer increases to offset
and the polarity
reversal can happen
III Lower than III − − Reflection
the sealing Magnitude increases
layer to offset
IV Lower than II − + Reflection
the sealing Magnitude
layer decreases to offset

Figure 39. Cross plot of AVO intercept (A) against gradient (B) shows 4 possibilities of quadrant. For limited time window, brine-
saturated sandstones and shale present all along the background trend. Top reflection of gas-sandstone drops below the background
trend (Castagna et.al, 1998)

AVO/AVA for Gas & Fluid Detection


39
By : Sigit Sukmono
8. Relationship of AVO Gradient and AVO Intercept

The parameter Vp, Vs, and ρ are often closely related each other and their relationship affects the values of A (Intercept) and B (Gradient).
Therefore the background trend orientation depends on the (Vp)/(Vs) ratio. Figure 51 showing that as the (Vp)/(Vs) increases then the
background trend rotates counter clockwise. The figure assumes that the gradient of mud rock line is equal to 1.16 and the Intercept constant
is equal to 1.36 km/s, which follows Castagna (1985) equation. Deviation to background trend is resulted from fluid substitution on pore
space. The gas substitution to brine cause the (Vp)/(Vs) decrease and make the ∆VP and ∆ρ become more negative. Figure 53 shows the
computed AVO intercept and gradient of pair shale/gas sand and shale/brine sand. Figure 54 showing the line of brine sand-gas sand for 25
sets of sonic log insitu measurement on brine sands, gas sands, and shales (Castagna & Smith, 1994).

Figure 40. The cross plot of zero offset reflection coefficient (A)
against the AVO gradient (B) by assuming the VP trend linear to Figure 41. The top sands responses of brine and gas sand.
VS (m = 1.16; c = 1.36 km/s) and Gardner’s relationship. The modeling uses average P-wave average velocity of 3
(Castagna et.al, 1998) km/s and Gessmann’s equation (Castagna et.al, 1998)

AVO/AVA for Gas & Fluid Detection


40
By : Sigit Sukmono
9. AVO Cross-Plot

AVO cross-plot is a simple way to show


the AVO data. The amplitude variation to
the offset for top-bottom interfaces is
shown as single point on intercept and
gradient cross-plot. The advantage of this
cross-plot is it can show information from
a data which can not be shown by
standard offset appearance. Figure 42
shows the intercept (R0) to gradient (G)
cross-plot and also the gas sandstone
classes described earlier.

Figure 42. Classes of AVO and AVO cross plots (Simm et.al, 2000)

AVO/AVA for Gas & Fluid Detection


41
By : Sigit Sukmono
9.1. Signal and Noise Effects on AVO Cross-Plot

Figure 43 showing single point on down right quadrant of crossplot. This point is resulted from AVO attribute (differentiated from least
square regression) which related to single zero phase reflection of non-noise synthetic gather. This point shows the response of AVO
anomaly class 1 from brine-filled consolidated sand top which adjacent to the seal shale, in here the amplitude decreased to offset.

If the random noise is uniformly added to CMP gather (S/N decreased as the offset increases), cross-plot response will be an oval
distribution around the real location (Figure 44). It is caused by gradient sensitivity to noise. Hendrickson called this as ‘ellipse noise’.
This noise is visible on real data by plotting the limited sample on the same horizon. Parallel spreading pattern to the gradient shows the
decrease of S/N. On real data, noise trend usually has tilt about 50 or more.

Figure 43. AVO cross plot of single class 1 brine reflection (left) and the noise which affect the gradient (Simm et.al, 2000)

AVO/AVA for Gas & Fluid Detection


42
By : Sigit Sukmono
9.2. Porosity and Gas Effects to AVO Cross-Plot

Porosity increase has two effects i.e : first, decreases the AVO gradient (such as : contras of Poisson Ratio and seal shale decreased),
and second, decreases the intercept (as the impact of acoustic impedance contras decrease). Porosity change on sandstone causes the
presence of an ellipse line as shown on Figure 44. Figure 45 showing fluids/gas substitution effect on sandstone with porosity
variation. From the figure appeared the drifting phenomena to the down left of background trend as the result of gas presence in
pored sandstone. From this figure we can obviously see the gas effect to intercept and gradient that moved to negative.

Figure 44. Porosity effect on AVO cross plot (Simm et.al, 2000) Figure 45. Gas effect on AVO cross plot (Simm et.al, 2000)

AVO/AVA for Gas & Fluid Detection


43
By : Sigit Sukmono
9.3. Time Window Cross-Plot

On the right is an illustration of time window


cross-plot in which the top and base interfaces
samples are constrained in an ellipse centered
on its point of origin (Figure 46).

Wet trend establishment is critical as it will act


as the separator for anomaly and non-anomaly.

Wet trend can be established using well data


(Vp, Vs) and laboratory measurement

Figure 47 and 48 illustrate the real examples


of time window cross-plot.

Figure 46. Time window cross plot (Simm et.al, 2000)

AVO/AVA for Gas & Fluid Detection


44
By : Sigit Sukmono
Figure 47. The example of real data. (a) Stacking section describes bright spot with picked sandstone top in green.
(b) Time window cross plot which produced from 40-ms window around the top sand (Simm et.al, 2000)

AVO/AVA for Gas & Fluid Detection


45
By : Sigit Sukmono
Figure 48. Horizon cross plot. (a) R0/G cross plot for pick which shown on Figure 48 and illustrates the different between trend
which associated with bright spot and background reflectivity. (b) Near/Far cross plot illustrates that the background trend on R0/G
cross plot related to noise, not to lithology (Simm et.al, 2000)

AVO/AVA for Gas & Fluid Detection


46
By : Sigit Sukmono
Exercise
The determination of mud-rock and dry-rock lines by using laboratory rock physics measurement results.

Questions: By using data in the following Tables, make cross plot of Vp to Poisson’s ratio in the diagram below.
Typical cross-plot is shown in the left figure below. Plot the two following sandstone samples on the cross plo;.
Sandstone A : Vp = 2400 m/sec; σ = 0.16; Sandstone B : Vp = 2700 m/sec; σ = 0.36. What type are the sandstone A
and B? How much is the approximate value of the sandstone A and B porosities?

0.5

0.4

Poisson’s Ratio
0.3

0.2

0.1

0
0 1 2 3 4 5 6 7
V P (km/sec)
Typical cross-plot of Poisson’s Ratio
vs. P-wave velocity (Russel, 1998)

AVO/AVA for Gas & Fluid Detection


47
By : Sigit Sukmono
Φ, Vp and σ on measurement I Φ, Vp and σ on measurement II

Porosity (%) Vp(m/sec) Poisson Ratio Porosity (%) Vp(m/sec) Poisson Ratio
0-5 5984 0.12 0-5 5964 0.12
10 3573 0.29 10 3621 0.12
20 2911 0.32 20 2400 0.12
30 2571 0.33 30 2132 0.12
40 2368 0.33 40 2004 0.12
50 2236 0.34 50 1959 0.12
60 2157 0.34 60 1983 0.12
70 2110 0.34 70 2001 0.12
80 2093 0.34 80 2296 0.12
90 2103 0.34 90 2762 0.12
100 2142 0.34 100 4264 0.12

AVO/AVA for Gas & Fluid Detection


48
By : Sigit Sukmono
Φ, Vp and σ on measurement III Φ, Vp and σ on measurement IV

Porosity (%) Vp(m/sec) Poisson Ratio Porosity (%) Vp(m/sec) Poisson Ratio
5 3753 0.12 15 2641 0.12
5 3750 0.12 15 2633 0.12
5 3748 0.12 15 2626 0.13
5 3746 0.12 15 2619 0.13
5 3744 0.12 15 2612 0.13
5 3743 0.13 15 2607 0.13
5 3744 0.13 15 2602 0.13
5 3747 0.13 15 2600 0.13
5 3758 0.13 15 2604 0.14
5 3786 0.15 15 2629 0.15
5 4248 0.25 15 3117 0.31

AVO/AVA for Gas & Fluid Detection


49
By : Sigit Sukmono
Table L1-5 : Φ, Vp and σ on measurement V Table L1-6 : Φ, Vp and σ on measurement VI

Porositas (%) Vp(m/sec) Poisson Ratio Porositas (%) Vp(m/sec) Poisson Ratio
33 2091 0.12 50 1977 0.12
33 2074 0.12 50 1943 0.13
33 2057 0.13 50 1911 0.13
33 2040 0.13 50 1881 0.13
33 2025 0.13 50 1853 0.13
33 2010 0.13 50 1826 0.13
33 1997 0.13 50 1802 0.13
33 1986 0.13 50 1781 0.13
33 1980 0.14 50 1765 0.14
33 1993 0.16 50 1767 0.16
33 2500 0.33 50 2238 0.34

AVO/AVA for Gas & Fluid Detection


50
By : Sigit Sukmono
10. AVO Attributes

Popular AVO attributes used in the


interpretation are :
• Normal Incident P-Wave (A)
• Gradient (B)
• Product Gradient (A*B)
• Fluid Factor (F)
• Lambda-Mu Reflectivity
• Amplitude Envelope
(Far-Near)

10.1. Normal Incident of P-wave (A)

Normal Incident P-wave (A) is also


known as Intercept. It is the first term in
Shuey’s equation :

 Δσ  2
RC(θ ) = RC p + R p A 0 + 2
sin θ + ....
 (1 − σ ) 

Normal Incident P-Wav e (A) Figure 49. P-wave normal incident section (A) (source : www.vsl.com)

AVO/AVA for Gas & Fluid Detection


51
By : Sigit Sukmono
10.2. Gradient (B)

Gradient (B) is the second term on following


Shuey’s equation:

 Δσ  2
RC(θ ) = RC p + RC p A 0 + 2
sin θ + ....
 (1 − σ ) 

Gradient (B)

Figure 50. The gradient section (B) (source : www.vsl.com)

AVO/AVA for Gas & Fluid Detection


52
By : Sigit Sukmono
10.3. Gradient Product (A*B)

Gradient product is the product of


normal incidence times gradient.

This product is used to identify the


bright spot of AVO anomaly class 3,
because the positive anomaly will give
positive values.

Figure 51. The gradient product (A*B) section (source : www.vsl.com)

AVO/AVA for Gas & Fluid Detection


53
By : Sigit Sukmono
10.4. Fluids Factor (F)

The ‘fluid factor’ is based on the ‘mudrock


equation by Castagna :

α = 1360 + 1.16 β where α, β are in m/sec.

The differential from equation above is ∆α = 1.16


∆β which can be expressed in ratio form as :

Δα β Δβ
= 1.16
α α β
The equation only accurate in the case of wet non-
productive reservoir. For an anomalous reservoir,
we can define the ‘fluid factor’ error from the
following equation :

∆α β ∆β
∆F = - 1.16
α α β
In other words, if ∆F =0, the reservoir is non-
prospective, but if | ∆F | = 0, the reservoir is
prospective.
Figure 52. The fluid factor (F) section (source : www.vsl.com)

AVO/AVA for Gas & Fluid Detection


54
By : Sigit Sukmono
Figure 53. The lambda reflectivity section λ = ρ (V P2−VS2 ) µ = ρVS2

AVO/AVA for Gas & Fluid Detection


55
By : Sigit Sukmono
References

1. Aki and Richards, 1980, Quantitative seismology, W.H. Freeman.


2. Castagna and Swan, 1997, Principles of AVO cross-plotting, The Leading Edge
3. Ostrander, W.J., 1984, Plane-wave reflection coefficient for gas sands at non-normal angles of incidence, Geophysicss 49.
4. Russel, B., 1998, Amplitude versus offset analysis, HRS training manuals.
5. Rutherford and Williams, 1989, Amplitude versus offset variations in gas sands, Geophysics
6. Shuey, 1985, A simplification of the Zoeppritz equation, Geophysics 50.
7. Simm, R., White, R., Uden, R., 2000, The anatomy of AVO cross-plots, The Leading Edge 19.
8. Smith, G.L. and Gidlow, M., The fluid factor angle and the cross-plot angle, SEG expanded abstract, 22, 185.

AVO/AVA for Gas & Fluid Detection


56
By : Sigit Sukmono
Exercises

AVO/AVA for Gas & Fluid Detection


57
By : Sigit Sukmono
AVO cross plot

Instantaneous Frequency Acoustic Impedance

Exercise: Identification and classification of gas-sands reservoir based on the seismic attribute section and AVO
cross plot. Questions : Delineate the top and bottom of the gas-sandstone in the Instantaneous Frequency and
Acoustic Impedance sections, and determine the type of gas-sandstone.

AVO/AVA for Gas & Fluid Detection


58
By : Sigit Sukmono
Well #1

Exercise :
On the right is some AVO cross-plots
overlain on instantaneous frequency
section.

Question
Is well-1 a dry or gas well ? Determine
next best well position.

AVO/AVA for Gas & Fluid Detection


59
By : Sigit Sukmono
Exercise :
On the right is some AVO
cross-plots overlain on PxG
product section.

Question
Roughly determine the
position of gas and brine
levels.

AVO/AVA for Gas & Fluid Detection


60
By : Sigit Sukmono
1000

1200

1400

1600

Exercise. Display of PRODUCT section (left) and associated positive AVO cross plot (right). In the product section, red
is positive while blue is negative. The wiggle display is the Intercept on zero angle with normal polarity. Determine the
top and bottom of gas sand.

AVO/AVA for Gas & Fluid Detection


61
By : Sigit Sukmono
410

420

450

490

Exercise. Display of PRODUCT section (left) and related positive AVO cross plot (right). In the product section, red is
positive while blue is negative. The wiggle display is the Intercept on zero angle with normal polarity. Determine the top
and bottom of gas sand.

AVO/AVA for Gas & Fluid Detection


62
By : Sigit Sukmono
LMR & EEI Inversion
Page

1. Introduction 2
2. Dong Approximation 2
3. Goodway’s Approach 6
4. Estimating Rp and Rs 19
5. Elastic Impedance (EI) 21
6. Extended Elastic Impedance (EEI) 32

Exercise I 39
Exercise II 65

LMR & EEI Inversion

By. Sigit Sukmono 1


1. Introduction
Lambda-Mu-Rho (LMR) and Extended Elastic Impedance (EEI) inversion techniques were developed from various
studies. It was Smith and Gidlow (1987) who first showed that pre-stack data can be stacked using different weights to get
fluid factor and apparent Poisson ratio sections to predict the fluids and lithology.
In 1996, Dong showed that the relationships of the changes in bulk modulus, shear modulus and Lame’s constants can be
expressed in the AVO parameters. Goodway et al (1997) gave a more comprehensive discussions on the application of
Lame’ and shear modulus as lithology and fluid indicator.

Connoly (1999) introduced the EI as a generalization of AI equation for non-normal incidence angle, which basically is
the inversion of AVO data. The EI equation was then normalized by Whitcombe (2002) and David et al (2002) which give
the EEI equation.

2. Dong Approximation

Dong (1996) showed that the change in Bulk modulus, shear modulus, Lame’s constant and density can be expressed in
AVO parameters of A, B and C defined by Aki and Richard (1980)

Based on the following equations

λ + 2µ K + 4 / 3µ µ
α= α= β= (1)
ρ ρ ρ

LMR & EEI Inversion

By. Sigit Sukmono 2


Which can be rewritten as

4
K = α 2ρ − β 2ρ (2)
3
λ = α 2 ρ − 2β 2 ρ (3)
µ = β 2ρ (4)
The above equations can be integrated to become

(3 A + B + 2C )α 2 ρ
∆K = (5)
1.5
∆λ = (2 A + B + C )α 2 ρ (6)
(C − B )α 2 ρ
∆µ = (7 )
2
Where ∆K, ∆λ and ∆μ are the changes of Bulk modulus, Lame’s constant and shear modulus of the adjacent layers.

LMR & EEI Inversion

By. Sigit Sukmono 3


Previous studies show that it is very difficult to get the value of C directly from seismic, therefore people prefer to use
C/A ratio which also known as f value. The f value is 0.8 for rocks which follow Gardner et al (1974) equation.

The bulk modulus reflectivity Rk can be obtained by substituting equations (5) and (2) :

∆K  B  3 + 2 f 
Rk = =  A +   (8)
2K  3+ 2 f  3 − 4 K 

Where K = (α/β)2

The first term in equation (8) can be written in the form of first order AVO equation :

R(θ) = A + B sin2 θ (9)

Where sin2 θ = 1/ (3+2f) (10)

If f = 0.8 (Gardner’s equation), then sin2 θk = 0.22 or θk = 28o. In other words, Rk is proportional with R(28o).

LMR & EEI Inversion

By. Sigit Sukmono 4


Using the same approach, we can get the Lame reflectivity :

∆λ  B  2 + f 
Rλ = =  A +   (11)
2λ  2+ f  2 − 4 K 

Where sin2 θλ = 1/ (2+f) (12)

If f = 0.8 (Gardner’s equation), then sin2 θλ = 0.36 or θλ = 37o means that Rλ is proportional with R(37o).

∆µ  B  f 
Rµ = =  A −   (13)
2µ  f  4 K 

Where sin2 θμ = -1/ f (14)

LMR & EEI Inversion

By. Sigit Sukmono 5


3. Goodway’s Approach
P-wave tends to travel along matrix than fluid. This makes AI is not too sensitive to fluid. To overcome this weakness, from
combination of Rp and Rs, Goodway (1997) introduced three elastic parameters : lambda-rho (λρ), mu-rho (μρ) and Bulk
modulus (K), which can be combined as indicators of lithology and fluids.

λ + 2µ K + 4 / 3µ µ
α= α= β =
ρ ρ ρ

SI = ρ x β
μρ = SI2
λρ = AI2 – 2SI2
K = ρ(Vp2 – 4/3 Vs2)
Where μ = rigidity modulus, λ = incompressibility modulus, K = bulk modulus and SI = shear impedance.
Figure 1 shows the general steps for performing LMR inversion. The final output of LMR inversion is the layer properties of
Lambda, Mu and Rho. The main difference between primer AVO analysis and LMR inversion is that the former gives the
boundary properties whereas the later provides layer properties.

LMR & EEI Inversion

By. Sigit Sukmono 6


Gathers

AVO Analysis

Estimate :
Reflectivity-P (Rp)
Reflectivity-S (Rs)

Invert to :
Acoustic Imp-P
Acoustic Imp-S

Transform to
Lambda, Mu, Rho
Figure 1. General steps for performing LMR inversion
Cross-plot and
Interpretation

LMR & EEI Inversion

By. Sigit Sukmono 7


Shear modulus or rigidity modulus Mu is the capability of rocks to resist shear strain due to the shear stress which changes
the shape but retain the volume. Mu scaled by Rho can indicate the rock type. Shear-able rocks such as shale or un-
compacted sands have lower Mu-Rho, whereas rigid rocks such as compacted sands or tight carbonate associate with bigger
Mu-Rho.
Modulus Mu is the capability of rocks to resist shear strain due to the shear stress which changes the shape but retain the
volume. Mu scaled by Rho can indicate the rock type. Shear-able rocks such as shale or un-compacted sands have lower Mu-
Rho, whereas rigid rocks such as compacted sands or tight carbonate associate with bigger Mu-Rho.
Incompressibility modulus Lambda scaled with density Rho can indicate the presence of fluid as it’s value reflects the fluid
resistance to the change of volume due to the compressional stress. Rocks and fluids can not be compressed easily means
that their Lambda is relatively high. On the other hand, gas is compressible; therefore the presence of gas in rocks will
decrease its Lambda.

The small Lambda-Rho indicate a porous rock. For the same porosity, rocks filled by gas have smaller Lambda-Rho than
rocks filled by water or oil. Conversely, tight rocks have bigger Lambda-Rho.

Figures 2 to 11 provide comparison between the usage of lambda-mu-rho and other seismic properties to identify litohology
and gas.

Figure 2 shows the comparison on the changes of Ip vs Is and Lamda-rho vs Mu-Rho values for wet sands and various
degree gas saturated sands. Figure 3 and 4 illustrate the use of P and S-impedance for sandstone – shale – coal
discrimination. Figure 5 and 6 illustrate the use of Lambdha-Mu-Rho for sandstone – shale – coal discrimination.

Figure 7 shows the discrimination of Gas Charged and Brine Charged Carbonates using Vp-Vs cross-plots. Figure 8 shows
cross-plots of various seismic properties for the differentiation of shale, brine carbonate and gas carbonate. Figure 9 to 11
illustrate the identification of gas and porous sands using Lambda-Mhu-Rho display (Winardhi, 2004).

LMR & EEI Inversion

By. Sigit Sukmono 8


Zp vs Zs for Biot-Gassmann Analysis
2650

2600

2550

Zs
2500

2450

2400
3500 4000 4500 5000 5500
Zp

Lambda-mu-rho
Wet Sand for -Biot-Gassmann
Gas Sands 90% to 0% Sw

7.25

6.75

Mu-rho
6.25
Figure 2. Cross-plots of Ip vs Is and Lamda-rho vs
Mu-Rho for wet sands and various degree gas 5.75
saturated sands 0 2.5 5 7.5 10 12.5 15
Lambda-rho

Wet Sand Gas Sand - 90% to 0% Sw

LMR & EEI Inversion

By. Sigit Sukmono 9


P-impedance vs Gamma Ray

Figure 3. Sandstone – shale – coal discrimination based on P-impedance


(Winardhi,2004).

Sandstone-Shale-Coal

LMR & EEI Inversion

By. Sigit Sukmono 10


Sands

Shales

Coal
s

P-impedance vs S-
impedance
Figure 4. Sandstone – shale – coal discrimination based on P and S-impedance
cross-plot (Winardhi,2004).

LMR & EEI Inversion

By. Sigit Sukmono 11


Lambda-Rho vs Gamma Ray

Figure 5. Sandstone – shale – coal discrimination based on Lambdha - Rho


(Winardhi,2004).

LMR & EEI Inversion

By. Sigit Sukmono 12


Lambda-Rho vs Lambda-per-Mu

Figure 6. Lambda-per-Mu as coal discriminator (Winardhi, 2004)

LMR & EEI Inversion

By. Sigit Sukmono 13


S-wave S-wave
Gas Well
Dry Well

P-wave
P-wave

Dry
well
Figure 7. Discrimination of Gas Charged and Brine Charged Carbonates using Vp-Vs cross-plots (Winardhi, 2004)

LMR & EEI Inversion

By. Sigit Sukmono 14


DNG-B LamdaRho-MuRho DNG-B Ip-MuRho CROSSPLOT
CROSSPLOT
POH-SH MTW-GAS MTW-WET
POH-SH MTW-GAS MTW-WET
40 40

MuRho (GPa*g/cc)
MuRho (GPa*g/cc)
30 30

20 20

10 10

0
0
5000 8000 11000
10 20 30 40 50 60
Ip (g/cc*m/s)
LambdaRho (GPa*g/cc)

DNG-B Vp-Vs CROSSPLOT DNG-B Vp - Vp/Vs RATIO


A POH-SH MTW-GAS MTW-WET
B CROSSPLOT
POH-SH MTW-GAS MTW-WET

3000 2.5

Vs (m/s)

Vp/Vs
2000 2.0

1000
1.5
2000 3000 4000 5000
2.00 2.25 2.50 2.75
Vp (m/s)
RHOB (G/CC)

DNG-B Vp - Vp/Vs RATIO DNG-B Ip-Is CROSSPLOT


C CROSSPLOT
POH-SH MTW-GAS MTW-WET
D POH-SH MTW-GAS MTW-WET

3.0 6000

Figure 8. Cross-plots of various seismic properties

Is (g/cc*m/s)
2.5 5000
Vp/Vs

for the differentiation of shale, brine carbonate and


4000
gas carbonate (Winardhi, 2004) 2.0

3000
1.5 6000 10000
2500 3000 3500 4000 4500 5000 Ip (g/cc*m/s)
Vp (m/s)

LMR & EEI Inversion

By. Sigit Sukmono 15


Figure 9. Identification of residual gas using
Lambda-Mhu display (Winardhi, 2004).

LMR & EEI Inversion

By. Sigit Sukmono 16


Lambda -
Rho

Figure 10. Identification of gas sand and porous sand


using Lambda-Rho and Mhu-Rho displays (Winardhi,
2004).
Mu - Rho

LMR & EEI Inversion

By. Sigit Sukmono 17


Lambda - Rho

Figure 11. Identification of prospective zones using Lambda-Rho (Winardhi, 2004).

LMR & EEI Inversion

By. Sigit Sukmono 18


4. Estimating Rp and Rs
In the flow-chart in Figure 1, it is shown that to get the Lambda-Mu-Rho, we need to estimate the Rp and Rs. It can be done
using graphical analysis (see discussion in AVO) or using the mathematical derivation as discussed below.
Aki-Richard equation (Sheriff and Geldart, 1995) :

1  ∆α  β2  ∆β 1  β2  ∆ρ 
R (θ ) =  (1 + tan 2 θ )  − 4 2 sin 2 θ  +  1 − 4 sin 2
θ 
 ρ 
 2  α  α  β  
2 α 2
 

which simplified by Smith and Gidlow (1987) :

1  ∆α ∆ρ  β 2  ∆β 2 ∆ρ  2 1 ∆α 2
R(θ ) =  +  − 2 2  2 +  sin θ + tan θ
2 α ρ  α  β ρ  2 α
In the case of small to moderate impedance change :

1 ∆( ρα ) 1 ρ∆α + α∆ρ 1  ∆α ∆ρ 
Rp = = =  + 
2 ρα 2 ρα 2 α ρ 
1  ∆β ∆ρ 
Rs =  + 
2 β ρ 

LMR & EEI Inversion

By. Sigit Sukmono 19


For θ <35 and α/β between 1.5-2.0, the third terms of Aki-Richard equation is negligible:

R(θ)=Rp(1+tan2θ)-8Rs(β/α)2 sin2θ

Rp and Rs can be determined using the following equations :

a1 d 2 − a 2 d 1 a1c 2 − a 2 c1
Rp = ; Rp =
c1 d 2 − c 2 d 1 c 2 d 1 − c1 d 2
where a1 and a2 are migrated amplitudes of near and far-angle stack, θ1 is the average of near angle stack and θ2 is the average
of far angle stack, and :

c1 = (1 + tan 2 θ1 ) c2 = (1 + tan 2 θ 2 )
β  β 
2 2

d1 = −8  sin 2 θ1 d 2 = −8  sin 2 θ 2


α  α 

LMR & EEI Inversion

By. Sigit Sukmono 20


5. Elastic Impedance (EI)
Linearization of Zoeppritz equation for P-wave reflectivity is sufficiently accurate for small changes of elastic parameter
for subcritical angle :

R(θ ) = A + B sin 2 θ + C sin 2 θ tan 2 θ

where :

1 ∆V p
1  ∆V p ∆ρ  ;
C =
A=  + 2 Vp
2 V ρ 
 p 
∆V p Vs ∆Vs
2
Vs2 ∆ρ
B= −4 2 −2 2
2V p V p Vs Vp ρ

Vp =
(V p (t i ) + V p (t i −1 ) )
2
∆V p = V p (t i ) − V p (t i −1 )
Vs2 (t i ) Vs2 (t i −1 ) 
 2 + 2 
Vs2  p i
V (t ) V p (t i −1 ) 
=
V p2 2

LMR & EEI Inversion

By. Sigit Sukmono 21


By defining the function f(t) is a parameter analogue with AI, then reflectivity as function of angle or EI can be written as
follows
f (ti ) − f (ti −1 )
R (θ ) =
f (ti ) + f (ti −1 )

In the case of small to moderate impedance change, then :

1 ∆EI 1
R(θ ) ≈ ≈ ∆1n( EIθ )
2 EI 2

 ∆V p V 2
∆Vs 
 −4 2 s 
1 1  ∆V p ∆ρ   2V p V p Vs 
∆1n( EIθ ) = + +  sin 2 θ
2 2  Vp ρ   Vs2 ∆ρ 

 − 2 2 
 Vp ρ 
1 ∆V p
+ sin 2 θ tan 2 θ
2 Vp

LMR & EEI Inversion

By. Sigit Sukmono 22


Vs2
By substituting K to
2
, we get :
Vp
1 1 ∆V p ∆ρ ∆V ∆V p 2
∆1n( EI ) = (1 + sin 2 θ ) + (1 − 4 K sin 2 θ ) − s 8K sin 2 θ + sin θ tan 2 θ
2 2 Vp ρ Vs Vp

Since sin2θ tan2θ = tan2θ-sin2θ, and by substituting every form of ∆x/x with ∆lnx, then :

∆1n( EI ) = (∆1n(V p )(1 + tan 2 θ ) + ∆1n( ρ )(1 − 4 K sin 2 θ ) − ∆1n(Vs )8K sin 2 θ )

If K made to be constant, the above equation can be written as :

(
∆1n( EI ) = ∆1n V p(1+ tan
2
θ)
)− ∆1n(V s
( 8 K sin 2 θ )
)+ ∆1n(ρ (1− 4 K sin 2 θ )
)
(
∆1n( EI ) = ∆1n V p(1+ tan
2
θ)
Vs
( −8 K sin 2 θ )
ρ (1− 4 K sin
2
θ)
)

LMR & EEI Inversion

By. Sigit Sukmono 23


To eliminate the logarithm and differential functions in both sides of above equation, we can integrate and exponentiation
the above equation to become :

(1+ sin 2 θ ) ( −8 K sin 2 θ ) (1− 4 K sin 2 θ )


EIθ = V p Vs ρ
or in shorter version

EI θ = V paVs ρ c
b

where a = (1 + sin2 θ), b = -8K sin2θ, c= (1-4K sin2θ) and K = (Vs/Vp)2.


EI is generalization of AI for incident angle unequal with 0. EI values can be obtained by inverting non-zero offset seismic
data. Since S-wave is not affected by the fluid, then EI will have different values with AI when the wave is passing thru the
fluids. Then for fluid imaging, EI has better capability than AI.

Figure 12 shows that the transformation of an AI log from 0° to 30° results in a generally similar log but with lower
absolute values. The impedance decreases with an increase in oil saturation. The percentage of decrease is greater for EI
than AI (Connolly, 1999)
Figure 13 and 14 illustrate the sensitivities of gamma-ray, AI and EI(30o) logs for the identification of oil interval.

Figure 15 shows the general steps in performing EI inversion. Figures 16 to 18 illustrate angle stack at 15, 20, 24 degree
and their respective EI inversion results.

LMR & EEI Inversion

By. Sigit Sukmono 24


Figure 12. The transformation of an AI log from 0° to 30° results in a generally similar log but with lower
absolute values. The impedance decreases with an increase in oil saturation. The percentage of decrease
is greater for EI than AI (Connolly, 1999)

LMR & EEI Inversion

By. Sigit Sukmono 25


Figure 13. Comparison of gamma-ray, AI and EI(30o) logs.
Positions of oil interval is shown (Connolly, 1999).

LMR & EEI Inversion

By. Sigit Sukmono 26


Figure 14. Comparison of gamma-ray, AI and EI(30o) logs. Positions of
oil interval is shown (Connolly, 1999).

LMR & EEI Inversion

By. Sigit Sukmono 27


Gathers

Near angle stack Far angle stack

Invert to near- Invert to far-


angle EI angle EI

Figure 15. General step of EI inversion

LMR & EEI Inversion

By. Sigit Sukmono 28


Angle Stack at 15 degree

Angle Stack at 20 degree

Figure 16. Illustration of angle stack at 15 and 20 degree (Winardhi, 2004)

LMR & EEI Inversion

By. Sigit Sukmono 29


Angle Stack at 24 degree

Elastic Impedance at 15 degree

Figure 17. Illustration of angle stack at 24 degree and EI at 15 degree (Winardhi, 2004)

LMR & EEI Inversion

By. Sigit Sukmono 30


Elastic Impedance of 20 degree

Figure 18. Illustration of EI sections at 20 and 24


degree (Winardhi, 2004)

Elastic Impedance of 24 degree

LMR & EEI Inversion

By. Sigit Sukmono 31


To make EI be comparable with AI, and to not use the dimension variable, Whitcombe (2002) modified the above equation by
introducing reference constants of α0, β0 dan ρ0

 α  a  β  b  ρ  c 
EI θ = α 0 ρ 0       
 α 0   β 0   ρ 0  
where α = Vp, β = Vs. α0, β0 and ρ0 are reference values taken from the average values of well data.

6. Extended Elastic Impedance (EEI)


The weakness of EI equation is in the limited range of sin2θ which covers only values between 0 and 1, meaning that the
available angle is between 0o and 90o. To expand the capability of EI equation, Whitcombe, Connolly, Reagen and Redshaw
(2002) introduced the modified EI equation or widely known as Extended Elastic Impedance (EEI).

Two modifications were made. First modification is by replacing sin2θ with tan x, which gives range between -∞ to ∞, and
more variable x values between -90o and 90o. The second modification was made to ensure that the resulted reflectivity never
exceed unity; this was done by scaling the reflectivity with cos x. To do these modification, the first step is by replacing the
Zoeppritz linear equation :

R(θ) = A + B sin2θ

to become (see Figure 20) :

R(x) = A + B tan x

( A cos x + B sin x)
R=
cos x

LMR & EEI Inversion

By. Sigit Sukmono 32


The scaled reflectivity Rs is then Rs = R cos x, which gives
Rs = A cos x + B sin x

Then the equivalent of the EI equations is

 α 
p
 β 
q
 ρ 
r

EEI θ = α 0 ρ 0       
 α 0   β0   ρ0  

where p = (cos x + sin x), q = -8K sin x, r = (cos x – 4K sin x)


Figure 19 shows that there is a high degree of correlation between elastic parameters and equivalent EEI curves showing.

Figure 21 shows that there is also quite high correlation of EEI function with water saturation and gamma-ray.

Figure 22 shows the suggested step for performing EEI inversion.

LMR & EEI Inversion

By. Sigit Sukmono 33


Figure 19. Comparison between
elastic parameters and equivalent
EEI curves showing high degree of
correlations (Connolly, 1999).

LMR & EEI Inversion

By. Sigit Sukmono 34


Figure 20. Illustration of EEI function R =
A + B tan X (Whitcombe et al., 2002) Figure 21. Correlation of EEI function with water
saturation and gamma-ray (Whitcombe et al., 2002)

LMR & EEI Inversion

By. Sigit Sukmono 35


Gathers
Figure 22. General steps for EEI Inversion

AVO Analysis

Intercept A Gradient B

EEI Inversion

Transform to λρ, µρ or λ/µ using generalized


linear/non-linear regression

LMR & EEI Inversion

By. Sigit Sukmono 36


References

1. Connolly, P., Elastic Impedance, The Leading Edge, 18, No.4, 438-452, 1999.
2. David, N.W., 2002, Short note : Elastic impedance normalization, Geophysics v.67, 1, 60-62.
3. David, N.W., et al, 2002, Elastic impedance for fluid and lithology prediction, Geophysics v.67, 1, 63-67.
4. Dong, W., 1996, A sensitive combination of AVO slope and intercept for hydrocarbon indication, 58th
Conference and Technical Exhibition, EAGE, Amsterdam, paper M044
5. Goodway, B., Chen, T., and Downton, J., 1997, Improved AVO fluid detection and lithology discrimination
using Lame’s petrophysical parameters; λρ, μρ and λ/μ fluid stack, from P and S inversion, 67th Ann.
International Mtg : SEG, expanded abstracts, 183-186.
6. Kamal, J.M., 2005, Studi kasus karakterisasi reservoir karbonat menggunakan metoda seismic inverse Lamba-
Mu-Rho dan EEI pada lapangan X, Master thesis, Dept. of Geophysical Engineering, ITB.
7. Whitcombe, D.N., Connolly, P.A., Reagen, R.L., and Redshaw, T.C., Extended elastic impedance for fluid and
lithology prediction, Geophysics, 67, 63-67, 2002.
8. Winardhi, S., 2004, Seismic-based lithology and fluid identification.
9. Yustiana, F., 2003, Integrasi AVO, inversi, dan estimasi parameter elastic (λ, μ and ρ) untuk identifikasi fluida
dan litlogi di Lapangan Mutiah, PT. Caltex Pacific Indonesia, Master thesis, Dept. of Geophysical Engineering,
ITB.

LMR & EEI Inversion

By. Sigit Sukmono 37


Exercises

LMR & EEI Inversion

By. Sigit Sukmono 38


Exercise-I
Objective
To understand the steps in performing LMR and EEI inversion and utilize the results for delineating prospective reservoirs.
Materials

1. Figure I.1. Cross-plot between P and S-wave at well B-01 for interval top K01-K02
2. Figure I.2. Cross-plot between P and S-impedance at well B-01 for interval top K01-K02
3. Figure I.3. Cross-plot between Lambda-Rho and Mu-Rho at well B-01 for interval top K01-K02
4. Figure I.4. CDP gather data.
5. Figure I.5. Seismic stack data of line ITB-01
6. Figure I.6. Gradient section of Line ITB-01
7. Figure I.7. Equation of Vp vs Vs used for extracting Rp and Rs
8. Figure I.8. (a) Rp and (b) Rs sections of Line ITB-01
9. Figure I.9. Initial P-Impedance model of Line ITB-01
10. Figure I.10. Final P-Impedance inversion result of Line ITB-01
11. Figure I.11. Initial S-Impedance model of Line ITB-01
12. Figure I.12. Final S-Impedance inversion result of Line ITB-01
13. Figure I.13. Final Lambda-Rho inversion result of Line ITB-01
14. Figure I.14. Final Mu-Rho inversion result of Line ITB-01
15. Figure I.15. Cross-correlation between EEI and Lambda and Mu
16. Figure I.16. Comparison between (a) EEI-19 vs Lambda, and (b) EEI-58 vs Mu
17. Figure I.17. Initial EEI (19) model with high cut frequency 12 Hz of Line ITB-01

LMR & EEI Inversion

By. Sigit Sukmono 39


18. Figure I.18. Final EEI (19) inversion result of Line ITB-01
19. Figure I.19. Initial EEI (-58) model of Line ITB-01
20. Figure I.20. Final EEI (-58) inversion result of Line ITB-01
21. Figure I.21. Cross-correlation between EEI(19) and Lambda-Rho
22. Figure I.22. Cross-correlation between EEI(19) and Lambda-Rho.
23. Figure I.23. Lambda-Rho section of Line R(19) from Emerge
24. Figure I.24 Mu-Rho section of Line R(-58) from Emerge

Question :

1. Which properties cross-plot gives the best tool for identifying the gas zone?
2. Why in Zp-Zs cross-plot the HC is clustered in the left of back-ground trend?
3. What angle gather range can be used in the analysis ?
4. What is the depositional environment and lithology type of the targetted structure?
5. Combine the available cross-plots and sections to identify the porous, tight and gas intervals.

LMR & EEI Inversion

By. Sigit Sukmono 40


Reservoar thickness is
10ft under top K-01

Figure I.1. Cross-plot between P and S-wave at well B-01for interval top K01-K02 (Kamal, 2005)

LMR & EEI Inversion

By. Sigit Sukmono 41


Reservoar thickness is
10ft under top K-01

Figure I.2. Cross-plot between P and S-impedance at well B-01 for interval top K01-K02 (Kamal, 2005).

LMR & EEI Inversion

By. Sigit Sukmono 42


Figure I.3. Cross-plot between Lambda-Rho and Mu-Rho at well B-01 for interval top K01-K02 (Kamal, 2005).

LMR & EEI Inversion

By. Sigit Sukmono 43


Figure I.4. CDP gather data (Kamal, 2005). What angle range can be used ?

LMR & EEI Inversion

By. Sigit Sukmono 44


Figure I.5. Seismic stack data of line ITB-01 Position of well B-01 is given (Kamal, 2005).

LMR & EEI Inversion

By. Sigit Sukmono 45


Figure I.6. Gradient section of Line ITB-01 (Kamal, 2005).

LMR & EEI Inversion

By. Sigit Sukmono 46


Figure I.7. Equation of Vp vs Vs used for extracting Rp and Rs (Kamal, 2005).

LMR & EEI Inversion

By. Sigit Sukmono 47


Figure I.8. (a) Rp and (b) Rs sections of Line ITB-01
(Kamal, 2005).

LMR & EEI Inversion

By. Sigit Sukmono 48


Figure I.9. Initial P-Impedance model of Line ITB-01 (Kamal, 2005).

LMR & EEI Inversion

By. Sigit Sukmono 49


Figure I.10. Final P-Impedance inversion result of Line ITB-01 using sparse-spike model (Kamal, 2005).

LMR & EEI Inversion

By. Sigit Sukmono 50


Figure I.11. Initial S-Impedance model of Line ITB-01 (Kamal, 2005).

LMR & EEI Inversion

By. Sigit Sukmono 51


Figure I.12. Final S-Impedance inversion result of Line ITB-01 using sparse-spike model (Kamal, 2005).

LMR & EEI Inversion

By. Sigit Sukmono 52


Figure I.13. Final Lambda-Rho inversion result of Line ITB-01 (Kamal, 2005).

LMR & EEI Inversion

By. Sigit Sukmono 53


Figure I.14. Final Mu-Rho inversion result of Line ITB-01 (Kamal, 2005).

LMR & EEI Inversion

By. Sigit Sukmono 54


Figure I.15. Cross-correlation between EEI and Lambda and Mu (Kamal, 2005).

LMR & EEI Inversion

By. Sigit Sukmono 55


Figure I.16. Comparison between (a) EEI-19 vs Lambda, and (b) EEI-58 vs Mu (Kamal, 2005).

LMR & EEI Inversion

By. Sigit Sukmono 56


Figure I.17. Initial EEI (19) model with high cut frequency 12 Hz of Line ITB-01 (Kamal, 2005).

LMR & EEI Inversion

By. Sigit Sukmono 57


Figure I.18. Final EEI (19) inversion result of Line ITB-01 (Kamal, 2005).

LMR & EEI Inversion

By. Sigit Sukmono 58


Figure I.19. Initial EEI (-58) model of Line ITB-01 (Kamal, 2005).

LMR & EEI Inversion

By. Sigit Sukmono 59


Figure I.20. Final EEI (-58) inversion result of Line ITB-01 (Kamal, 2005).

LMR & EEI Inversion

By. Sigit Sukmono 60


Figure I.21. Cross-correlation between EEI(19) and Lambda-Rho. Lambda-Rho = (-3.50 + 0.001125 EEI-19)2
(Kamal, 2005).

LMR & EEI Inversion

By. Sigit Sukmono 61


Figure I.22. Cross-correlation between EEI(19) and Lambda-Rho. Lambda-Rho = (0.85 + 0.00036 EEI-58)2
(Kamal, 2005).

LMR & EEI Inversion

By. Sigit Sukmono 62


Figure I.23. Lambda-Rho section of Line R(19) from Emerge (Kamal, 2005)

LMR & EEI Inversion

By. Sigit Sukmono 63


Figure I.24 Mu-Rho section of Line R(-58) from Emerge (Kamal, 2005)

LMR & EEI Inversion

By. Sigit Sukmono 64


Exercise II

Objective
To understand the steps in performing LMR and EEI inversion and utilize the results for delineating prospective reservoirs.

Materials
1. Figure II.1. Normal seismic and Rp initial model of inline 153.
2. Figure II.2. Normal seismic and Rs initial model of inline 153 p
3. Figure II.3. Normal seismic and R(θ=30o) initial model of inline 153
4. Figure II.4. Ip section and the error trace of inline 153
5. Figure II.5. EEI (θ=30o) and related error sections of inline 153
6. Figure II.6. Lambda-Rho section of inline 153
7. Figure II.7. Mu-Rho section of inline 153
8. Figure II.8. Lambda section of inline 153
9. Figure II.9. Mu section of inline 153
10. Figure II.10. Intercept * Gradient section of inline 235
11. Figure II.11. Poisson ratio section of inline 235
12. Figure II.12. Mu section of inline 235
13. Figure II.13. Mu-Rho section of inline 235
14. Figure II.14. Poisson ratio map of interval top Rindu-1 ± 10 ms
15. Figure II.15. Intercept * Gradient map of interval top Rindu-1 ± 10 ms
16. Figure II.16. Lambda-Rho section of inline 235
17. Figure II.17. Lambda section of inline 235
18. Figure II.18 Mu-Rho map of interval top Rindu-1 ± 10 ms
19. Figure II.19 Mu map of interval top Rindu-1 ± 10 ms
20. Figure II.20 Lamabda-Rho map of interval top Rindu-1 ± 10 ms
21. Figure II.21 Lamabda-Rho map of interval top Rindu-1 ± 10 ms

Question :
Combine the available sections and maps to identify the porous, tight and gas intervals.
LMR & EEI Inversion

By. Sigit Sukmono 65


Figure II.1. Normal seismic and Rp initial model of inline 153 passing thru Mutiah#5299 (Yustiana, 2003)

LMR & EEI Inversion

By. Sigit Sukmono 66


Figure II.2. Normal seismic and Rs initial model of inline 153 passing thru Mutiah#5299 (Yustiana, 2003)

LMR & EEI Inversion

By. Sigit Sukmono 67


Figure II.3. Normal seismic and R(θ=30o) initial model of inline 153 passing thru Mutiah#5299 (Yustiana, 2003)

LMR & EEI Inversion

By. Sigit Sukmono 68


Figure II.4. Ip section and the error trace of inline 153 passing thru Mutiah#5299 (Yustiana, 2003)

LMR & EEI Inversion

By. Sigit Sukmono 69


Figure II.5. EEI (θ=30o) and related error sections of inline 153 passing thru Mutiah#5299 (Yustiana, 2003)

LMR & EEI Inversion

By. Sigit Sukmono 70


Figure II.6. Lambda-Rho section of inline 153 (Yustiana, 2003)

LMR & EEI Inversion

By. Sigit Sukmono 71


Figure II.7. Mu-Rho section of inline 153 (Yustiana, 2003)

LMR & EEI Inversion

By. Sigit Sukmono 72


Figure II.8. Lambda section of inline 153 (Yustiana, 2003)

LMR & EEI Inversion

By. Sigit Sukmono 73


Figure II.9. Mu section of inline 153 (Yustiana, 2003)

LMR & EEI Inversion

By. Sigit Sukmono 74


Figure II.10. Intercept * Gradient section of inline 235 (Yustiana, 2003)

LMR & EEI Inversion

By. Sigit Sukmono 75


Figure II.11. Poisson ratio section of inline 235 (Yustiana, 2003)

LMR & EEI Inversion

By. Sigit Sukmono 76


Figure II.12. Mu section of inline 235 (Yustiana, 2003)

LMR & EEI Inversion

By. Sigit Sukmono 77


Figure II.13. Mu-Rho section of inline 235 (Yustiana, 2003)

LMR & EEI Inversion

By. Sigit Sukmono 78


Figure II.14. Poisson ratio map of interval top Rindu-1 ± 10 ms (Yustiana, 2003)

LMR & EEI Inversion

By. Sigit Sukmono 79


Figure II.15. Intercept * Gradient map of interval top Rindu-1 ± 10 ms (Yustiana, 2003)

LMR & EEI Inversion

By. Sigit Sukmono 80


Figure II.16. Lambda-Rho section of inline 235 (Yustiana, 2003)

LMR & EEI Inversion

By. Sigit Sukmono 81


Figure II.17. Lambda section of inline 235 (Yustiana, 2003)

LMR & EEI Inversion

By. Sigit Sukmono 82


Figure II.18 Mu-Rho map of interval top Rindu-1 ± 10 ms (Yustiana, 2003)

LMR & EEI Inversion

By. Sigit Sukmono 83


Figure II.19 Mu map of interval top Rindu-1 ± 10 ms (Yustiana, 2003)

LMR & EEI Inversion

By. Sigit Sukmono 84


Figure II.20 Lamabda-Rho map of interval top Rindu-1 ± 10 ms (Yustiana, 2003)

LMR & EEI Inversion

By. Sigit Sukmono 85


Figure II.21 Lamabda-Rho map of interval top Rindu-1 ± 10 ms (Yustiana, 2003)

LMR & EEI Inversion

By. Sigit Sukmono 86


Processing & Pitfalls of AVO Analysis

CONTENTS
Page
1. Processing for AVO Analysis 2
2. Restoring and preserving relative amplitudes 2
3. Improving data quality for AVO interpretation 13
4. Recommended processing flow for AVO analysis 23
5. Recommended Steps for AVO Analysis 24
6. Pitfalls in AVO Analysis 24

Exercise 40

Processing & Pitfall of AVO Analysis


1
By : Sigit Sukmono
1. Processing for AVO Analysis
AVO Processing should preserve or restore relative trace amplitude within CMP gathers. Reflections must be correctly positioned in
the subsurface. Data quality should be sufficient to ensure that reflection amplitudes contain information about reflection
coefficients.

An AVO interpretation is only as good as the seismic processing behind it. Aspects of processing important to AVO analysis are
preserving relative trace amplitude, correctly positioning reflectors, improving data quality, and producing interpretation aids such
as specialized AVO display.

2. Restoring and preserving relative amplitudes

Factor that change the amplitudes of seismic trace can be grouped into the earth effects, acquisition-related effects, and noise (Dey-
Sarkar et al., 1986). Earth effects include spherical divergence, absorption, transmission losses, interbed multiples, converted phases,
tuning, anisotropy, and structure. Acquisition-related effects include source and receiver coupling variations, lateral changes in
weathered layer properties, source and receiver arrays, and receiver sensitivity. Noise can be ambient or source-generated, coherent
or random. Processing attempts to compensate for or remove these effects, and can in the process change relative trace
amplitudes.

Some processing techniques we need to consider carefully in restoring and preserving relative amplitudes are : whole trace
equalization, surface-consistent amplitude balancing, and section-dependent equalization.

Processing & Pitfall of AVO Analysis


2
By : Sigit Sukmono
2.1. Surface consistent
amplitude balancing

Variations in source and


receiver coupling and
weathered layer variations can
produce large deviations in
amplitude that do not
correspond to target reflector
properties. Surface-consistent
amplitude balancing should
compensate for these
variations.

Figure 1 illustrates the


importance of surface-
consistent amplitude
balancing for AVO. A
reflection that decreases in
amplitude with offset before
complete surface-consistent
amplitude balancing, increases
after balancing. Figure 1. Relative trace amplitudes after one pass (a) and two passes (b) of
surface-consistent amplitude balancing. The reflections at 1.65 and 1.75 s
decrease in amplitude with offset before balancing is complete and increase
after balancing. Arrows here and in all CMP gathers displayed show direction of
increasing offset. (Allen & Peddy, 1993)

Processing & Pitfall of AVO Analysis


3
By : Sigit Sukmono
2.2. Whole trace equalization

Whole trace equalization is normally applied to improve signal-to-noise (S/N) ratios; however its application on the data used in
AVO analysis should be evaluated carefully such as illustrated in the Figures 2-5.

The gathers in Figure 2 uses whole trace equalization. The mute was applied before the equalization process to remove the near-
surface high amplitudes on the far traces. Figure 3 is a close-up of Figure 2 highlighting intervals surrounding the attractive
amplitude response. Figure 4 provides a plot of amplitude vs offset for the whole trace equalization data. Figure 5 shows the same
CMP gathers but without whole trace equalization.

Question
Based on the data presented in Figures 2-5, what do you think about the high amplitudes for the reflection at 1.75 s, does it
correspond to hydrocarbon bearing sands?

Processing & Pitfall of AVO Analysis


4
By : Sigit Sukmono
Figure 2. A complete CMP gathers used to evaluate AVO prospect. Whole trace equalization was applied in this data (Allen & Peddy,
1993)
Processing & Pitfall of AVO Analysis
5
By : Sigit Sukmono
Figure 3. CMP gathers as in Figure 2 but concentrated on the prospect interval. The reflection at 1.75 s shows a strong increase
in amplitude with offset and was drilled as an AVO anomaly. ( Allen & Peddy, 1993)

Processing & Pitfall of AVO Analysis


6
By : Sigit Sukmono
Figure 4. (a) CMP gather with whole trace equalization
applied after muting. (b) Maximum amplitude of each trace
in window 1.65-1.75 s. (Allen & Peddy, 1993)

Processing & Pitfall of AVO Analysis


7
By : Sigit Sukmono
Figure 5. Gathers in Figure 3 without whole trace equalization. (Allen & Peddy, 1993)

Processing & Pitfall of AVO Analysis


8
By : Sigit Sukmono
2.3. AGC

Land data are often of


poor quality and it is
therefore difficult to
produce attractive CMP
gathers for AVO
analysis. AGC is a
simple and effective
means of improving the
appearance of the data,
but can change relative
amplitudes within a
CMP gather.

The example in Figure


6 illustrates the changes
in relative amplitude
that can occur after
long-window AGC. The
objective reflection at
1.35 s increases in
offset after AGC, while
the original AVO Figure 6. (a) CMP gathers across amplitude anomaly at 1.35 s drilled as an
response is more AVO Anomaly. (b) An AGC with a 2 window applied to the data changes the
ambiguous. AVO response of the reflection at 1.35 s to an apparent increase. (Allen &
Peddy, 1993)

Processing & Pitfall of AVO Analysis


9
By : Sigit Sukmono
2.4. Wavelet phase

Another aspect of seismic


amplitudes important to AVO
processing is the phase. In order to
tie several vintages of data, or for
other reasons, the phase of the
section can be significantly altered
during processing.

If the phase of a section is not


established by the interpreter, then
AVO anomalies that would be
interpreted as indicative of
decreasing impedance can be
produced at interfaces where the
impedance increases

Figure 7. (a) CMP gather across a single reflector with an increase


in acoustic impedance and Poisson’s ratio. The side lobe of the zero-
phase wavelet in the small trough above the mean peak. (b) the
wavelet in (a) after a 600 phase rotation. The side lobe becomes the
dominant initial deflection after rotation. The wavelet now appears
to arise from a decrease in impedance and Poisson’s ratio. From
Allen et al., 1993.
Processing & Pitfall of AVO Analysis
10
By : Sigit Sukmono
2.5, Arrays

Source and receiver


arrays used to reduce
ground roll and
surface waves can
produce offset-
dependent changes in
amplitudes within a
CMP gather. The
effect may be more
pronounced for
reflections from
shallower reflectors.
Pre-acquisition
modeling, such as is
shown in Figure 8,
can help in judging
the effect an array
will have on
reflections from the
objective interval.
Longer array lengths
increase the
attenuation at far
offsets relative to the
near offsets.

Figure 8a-b. See the complete figure caption on the next page

Processing & Pitfall of AVO Analysis


11
By : Sigit Sukmono
Figure 8c

Figure 8 .. A synthetic CMP gather was generated for a model with a 50 ft (15m) layer with Vp , Poisson’s ratio and density
lower than the surrounding layers. The layer produced a reflection that increased in amplitude with offset. (a) The gather
and attributes with the effect of an 83 ft (25m) array. (b) An array 330 ft (100m) in length produces more attenuation at the
far offsets, reducing the gradient and NI*D Sigma (Normal incidence * delta sigma, or the change in Poisson’s ratio)
response of the reflection. (c) An array 660 ft (200m) in length completely changes the response of the reflection from an
increase to a decrease in amplitude with offset. Offsets in feet and meters given above gathers. (Allen & Peddy, 1993)
3. Improving data quality for AVO interpretation
For AVO analysis, the challenge to the processor is to improve the S/N ratio without compromising the integrity of the data.
How does the interpreter determine the extent to which the processed data represent relative amplitudes? The most common
quality control is a visual inspection that relies upon the experience of the interpreter.

Logs appropriate for modelling may not be available. S-wave velocities are often not measured, and must be calculated, leading to
uncertainty as to the correctness of the model input. The blocking process needed for modelling is not always robust. Lateral
extrapolations from the well may not be valid, especially in areas where sand deposition is erratic.

AVO gradient plots can sometimes be an indicator that the processing has not maintained relative amplitudes: if all reflections
show a positive AVO response, then the data are probably improperly balanced.

In this section we look at deconvolution used to remove multiples, and two methods to increase the S/N ratio of the data,
coherency enhancement and the creation of supergathers. Other methods for improving data quality such as AGC and trace
balancing can change relative amplitudes, as shown previously.

Processing & Pitfall of AVO Analysis


13
By : Sigit Sukmono
3.1. Deconvolution

Deconvolution is a trace-by-trace process and thus conventional deconvolution may not preserve relative amplitudes. Surface-consistent
deconvolution algorithms will not change the relative amplitudes of the traces, but may not be as effective at removing multiples as conventional
deconvolution (Figure 9).
a) Spiking deconvolution b) Surface-consistent deconvolution

c) Surface-consistent wavelet processing


Figure 9. Comparison of spiking deconvolution,
surface-consistent deconvolution and surface-consistent
wavelet processing. (a) Spiking deconvolution. Standard
deconvolution is a trace-by-trace calculation, so that
relative amplitudes may not preserved. (b) Surface
consistent deconvolution preserves relative amplitudes
but may not attack the multiples effectively. (c) Surface-
consistent wavelet processing similar to surface-
consistent deconvolution. This surface-consistent
algorithm gives results intermediate between spiking
deconvolution and the surface-consistent result in (b).
(Allen & Peddy, 1993)
3.2. Supergathers
The stacking process is
perhaps the most
powerful noise reduction
device available to
seismic interpreters.
AVO analysis takes away
that power to a large
extend, and one of the
most severe difficulties
facing the AVO
interpreters is the poor
data quality of traces in a
CMP gather.
Unfortunately, the far
traces are often
significantly more noisy
than near traces,
especially on land data.
Partial stacking of CMP
traces or supergather is
an effective method for
achieving improved data
quality without changing
relative amplitude. Figure 10 (a) Raw CMP gathers across an amplitude anomaly at 2.2 s. (b) Supergathers from the
same CMP locations as in (a). The improvement in S/N ration is significant, but the relative
amplitudes of the reflections are preserved. The supergathers were made by forming offset bins
filled by traces from four adjacent CMP positions (Allen & Peddy, 1993)

Processing & Pitfall of AVO Analysis


15
By : Sigit Sukmono
3.3. Coherency enhancement
Another method for improving the S/N
ratio of prestack data is to scan the data for
coherent events (signal) and enhance these
events relative to other events on the data
traces. In many of the case studies, the use
of a coherency enhancement algorithm
significantly improved data quality (Figure
11). Coherency enhancement algorithms
vary in approach, and caution must be
exercised. Quality control of input and
output is necessary when coherency
enhancement is employed.

Figure 11. (a) CMP gather with residual statics


applied. (b) CMP gather in (a) with the addition of a
coherency enhancing algorithm. The input data were
scanned for coherency over a 10 ms window and a
synthetic trace created that contain the coherent
reflections. The synthetic trace was then added to
each original trace. The synthetic trace represents 30
percent and the data traces 70 percent in this display.
(Allen & Peddy, 1993)

Processing & Pitfall of AVO Analysis


16
By : Sigit Sukmono
3.4. NMO stretch effect a) b)
Reflection above about
1.5s are usually subjected
to a significant amount of
normal move-out (NMO)
stretch during processing.
In AVO analysis,
reflections on CMP
gathers that are stretched
can be deceiving to the
eye. A stretched reflection
appears to have a higher
amplitude at the far offsets
simply because a larger
area is blackened. This
effect is illustrated in
Figure 12. Unfortunately,
it is difficult to circumvent
the NMO effect.
Quantitative plots of
relative trace amplitudes
can be used to distinguish
between real increases in
amplitude with offset and Figure 12. (a) CMP gather with no NMO applied. The event at 0.75 s normal incident time
apparent increase, (arrow) demonstrates a decrease in amplitude with offset in this display. (b) Same gather
especially for reflections with normal moveout applied. The stretch caused by NMO produces a false appearance of
above about 1.0 s. increasing amplitudes for the event at 0.75 s. The plot of a maximum amplitude of each
trace shows that the amplitude of each trace shows that the amplitude are decreasing with
offset. (Allen & Peddy, 1993)

Processing & Pitfall of AVO Analysis


17
By : Sigit Sukmono
3.5. Reference horizons and AVO
interpretation
Many of the difficulties inherent in AVO
analysis can be sidestepped by comparing
the AVO response of the target reflection
to the AVO response of another reflection
in the CMP unrelated to hydrocarbons
(Figure 13).

Transmission losses, array effect, source-


receiver coupling variations, spherical
divergence, and many other factors that
can change the AVO response of a
reflection would be similar for both
target and reference reflections, so any
difference in the AVO responses of the
reflections would be due to a difference
in the properties across the two reflectors.

Figure 13. (a) Maximum amplitude in windows containing the target


reflection (solid line) and reference reflection (dashed line). Amplitudes
are normalized to the averages amplitude of the reference window. (b)
The ratio of the target reflection to the reference reflection for each trace
(solid line values divided by dashed line value in top figure). In this
example, the target reflection increases in amplitude with offset with
respect to the reference event. (Allen & Peddy, 1993)
Processing & Pitfall of AVO Analysis
18
By : Sigit Sukmono
3.6. AVO versus AVA
Seismic acquisition measures amplitudes as a function of source-receiver offset, while Zoeppritz’s equations and approximations of the
equations relate reflection coefficients or amplitudes to angle of incidence. Angle stacks and offset stacks both share the advantage of
S/N improvement through stacking and the ability to compare the objective reflection with other events on the section. Angle gathers
have the advantage of showing all reflections at the same incidence angle at all traveltimes, unlike conventional CMP gathers, in which
deeper reflections contain smaller incidence angles on a given trace than shallower reflections. A reconnaissance search for AVO
anomalies on large volumes of data can be conducted by simply scanning the far trace offset stacks for high-amplitude events (Figures
14-15).

Figure 14. Offset stack containing near, middle, and far


trace groups are an easy and effective means of locating
potential AVO anomalies. By displaying large positions
of the data in a stacked format, reflections with an
anomalous AVO response are easily located. (Allen &
Peddy, 1993)

Processing & Pitfall of AVO Analysis


19
By : Sigit Sukmono
1.6

1.8

2.0

0-10 degree 10-20 degree

1.6

1.8

2.0

20-30 degree
Figure 15. Angle stacks across an AVA anomaly at 1.77 s. As was the case in Figure 14, the near trace stack
exhibits very little reflectivity for the objective reflection, with the stacked amplitude anomaly coming from
the contribution of the larger angles. Drilling the anomaly resulted in a Yegua field discovery (Allen & Peddy,
1993)

Processing & Pitfall of AVO Analysis


20
By : Sigit Sukmono
Angle stacks surgically mute CMP gathers to remove time-offset samples lying outside the angle window. The sample within the
angle window are summed to give a stack containing only a specified range of angle. A similar process can be followed to
produced angle gathers (Figure 16 and Figure 17). If angle stacks or angle gather are used, the raw CMP gathers should be
available for comparison. Coherent noise trains or noisy trace bands may not be apparent on the angle displays.

Figure 16 (a). The mixing involved in the production of angle Figure 16.(b). Conventional CMP gathers for the same
gathers can compromise the veracity of the data. In this case, the location as in Figure 16 (a). The data quality is poor, but
angle gathers indicate that the reflection indicated by the narrow, the trend of the reflection indicated by the narrow is
and almost all reflections, smoothly increase in amplitude as a clearly not the increase in amplitude seen on the angle
function of incidence angle. (Allen & Peddy, 1993) gathers in Figure 16 (a). (Allen & Peddy, 1993)

Processing & Pitfall of AVO Analysis


21
By : Sigit Sukmono
Figure 17 (a) Angle gathers across an Figure 17 (b) CMP gather at the
amplitude anomaly at 2.15 s. The angle same location as Figure 23 (a). The
traces are made by surgically muting CMP S/N ratio is lower than in the angle
records, then summing the amplitudes gather as no mixing has been applied
within the mute window. Angle gathers to this gather. A field development
have a smoothed appearance due to trace well tested this anomaly and
mixing in the creation of the gathers. (Allen discovered a new pay sand. (Allen &
& Peddy, 1993) Peddy, 1993)

Processing & Pitfall of AVO Analysis


22
By : Sigit Sukmono
4. Recommended processing flow for AVO analysis
Most of the case history examples in this book were processed
using the processing flow in the diagram in Figure 18. In some
steps, a data-dependent choice is made. The case histories
presented here feature land data sets. The processing flow for a
marine data set would differ slightly, incorporating more
emphasis on multiple removal and less on improvements in S/N
ratio. Whether basic processing flow in Figure 18 is used, or a
more elaborate flow incorporating S/N enhancement is used,
the processor, and more importantly, the interpreter should be
aware of the consequences of each processing step on relative
amplitudes.

Figure 18. Processing flow chart used in most of


the case histories in this book. This processing
flow will not always produce attractive CMP
gathers, but data veracity should be maintained. If
a seismic data set requires more extensive
processing than that described here in order to
image the target horizon, then AVO analysis for
prospect risk evaluation may be unreliable.
Coherency enhancement and the creation of
supergathers are optional, and should be applied
only when data quality is poor. (Allen & Peddy,
1993)

Processing & Pitfall of AVO Analysis


23
By : Sigit Sukmono
5. Recommended Steps for AVO Analysis

1. Do not use trace-by-trace amplitude scaling such as automatic gain control (AGC, illustrated in previous section)
2. Use surface-consistent processing corrections. Iterate surface-consistent amplitude corrections until data are balanced
3. Examine specialized AVO/AVA plots in conjunction with CMP gathers
4. Prestack migrate CMP gathers to be used for AVO analysis to reduce stratigraphic complications
5. Use field analogs of either data across producing wells or dry holes on the same seismic line or similar seismic data in the area
to check the AVO response of reflections from gas-filled or wet-rocks. Analogs should be structural, as well as stratigraphic.
6. To increase the confidence level in an AVO anomaly, use reference horizons in AVO analysis. Reference horizons help identify
anomalous AVO response in noisy data even when the measured AVO response does not match the expected AVO response.
7. Unless S-wave velocities are known for wet and gas-filled reservoir rocks, apply the results of AVO modeling with care.
Incorrect S-wave velocities can result in misleading modeling results.
8. Determine the phase of the data used in prospect evaluation using a reflection with a known impedance contrast.
9. Acquire enough data to eliminate out-of-plane effects.

6. Pitfalls in AVO Analysis

The careful processing and modeling do not guarantee successful prediction of fluid content by AVO analysis. Some pitfalls which
we have to take care are :
1. Clean, highly porous sandstone can cause AVO anomaly similar to that caused by gas sand with less porosity
2. Changes in the amount of transverse isotropy present in the target zones and surrounding rocks can cause changes in the
AVO
response of the reflections.
3. Tuning can cause an increasing AVO response.

Processing & Pitfall of AVO Analysis


24
By : Sigit Sukmono
6.1. Illustration of AVO Anomalies from Clean Wet Sands

Figure 19 shows a stacked section across a bright spot with a character similar to that caused by channel sands in nearby production. The
prospect was evaluated using AVO analysis prior to drilling. Display of the CMP gathers (Fig. 20) and angle stacks (Fig. 21) showed that
the target reflection increased in amplitude with offsett (Fig.22).

Figure 19. Stacked section across a bright spot at 1.65 s with a character similar to that caused by channel sands in nearby production.
The prospect was tested and the bright spot found to correspond to a clean, highly porous, wet sand (Allen & Peddy, 1993)

Processing & Pitfall of AVO Analysis


25
By : Sigit Sukmono
Figure 20. Unmigrated CMP gathers across the
amplitude anomaly (dry well symbol in Figure 19).
The reflection at 1.65 s demonstrates a fairly strong
increase in amplitude with offset. The increase was
interpreted as being indicative of gas (Allen & Peddy,
1993).

Processing & Pitfall of AVO Analysis


26
By : Sigit Sukmono
Figure 21. Angle stack across the anomaly in Figure 19. The reflection at 1.65 s increase in amplitude with respect to other events. The
same anomalous AVO response is expressed differently in the reference horizon plot in Figure 22 (Allen & Peddy, 1993).

Processing & Pitfall of AVO Analysis


27
By : Sigit Sukmono
Figure 22. (a) CMP supergather
showing the target reflection at 1.645 s
and the reference horizon above it. (b)
Reference horizon plot above for the
target and reference events indicated in
the CMP gather. The target reflection
increase with respect to the reference
event, indicating that the increase is not
an artifact of acquisition or processing
(Allen & Peddy, 1993)

Processing & Pitfall of AVO Analysis


28
By : Sigit Sukmono
Figure 23. Logs through the
interval testing the reflector
at 1.65 s in Figure 19. The
bright spot ties the thick, wet
sand between 6480 and 6520
ft (1980-1990 m). A 3 ft
(1m) lignite stringer overlies
the sand. Note that the
density porosity of the sand
is high, about 33 percent.
Thick, clean, wet sands can
produce AVO anomalies for
the same reason as does, e.g.,
the impedance and Poisson’s
ratio of the sand is less than
the impedance and Poisson’s
ratio of the sand is less than
the impedance and Poisson’s
ratio of the surrounding
shales (Allen & Peddy,
1993).

The prospect was drilled and was dry. The AVO response was first attributed to a lignite at the top of objective sand, seen in the
formation density log in Figure 23. Subsequent analysis has shown that the 3 ft lignite stringer was too thin to be seen seismically
and that the anomaly was caused by the 42 ft thick, clean, wet sand with density varies from 33 to 37 percent. The sand has a P-
wave velocity that is higher than surrounding sands, but a density that is lower than that of the shale.

Processing & Pitfall of AVO Analysis


29
By : Sigit Sukmono
Figure 24. Reflection
coefficient as a function of
angle of incidence for the
reflection from the top of the
sand at 1.65 s in Figure 19.
Layer properties are from the
well logs in Figure 23. Note
that the sand has a higher Vp
than the shale, but a lower
density. The impedance of the
sand is 20927 and that of the
shale 23 347, making the
reflection coefficient from the
top of sand negative. Layer 1:
Vp=9728 ft/s (2965 m/s),
density=2.4 g/cc, σ=0.34.
Layer 2: Vp=10071 ft/s (3070
m/s), density=2.078 g/cc,
σ=0.26 (Allen & Peddy,
1993).

The reduction in density is sufficient to make the top sand RC negative, so that the top sand is a decrease in both AI and Poisson’s
ratio. The solution to Zoeppritz’s equations for the top sand is an increase in a negative RC, similar to that observed for gas sands in
the surrounding area (Fig.24).

Processing & Pitfall of AVO Analysis


30
By : Sigit Sukmono
6.2. Illustration of AVO Anomalies Due to Tuning

Field analogs are perhaps the most reliable means of calibrating the AVO response of a prospect. Whatever the AVO response of a
target reflection might be, if its response can be compared favorably to that from a drilled reflection, the risk associated with the
prospect is diminished.

Figures 25-26 shows that the field and prospect have a similar stacked amplitude response. The prospect is in the same stratigraphic
interval as the field, although structurally downdip and 250 ms deeper. Velocity inversion across the production and the prospect was
encouraging. The change in background color from yellow-green to blue-green corresponds to the onset of abnormal pressure
(Figs.27-28). CMP gathers show the AVO response typical of Yegua gas production, i.e. an increase in amplitude with offset, with an
onset trough (Fig. 29).

The prospect was drilled and found to correspond to a wet sand. Related possible explanation are out of the plane effects, Fresnel zone
effects, or tuning. The latter hypothesis is favored, as tuning within the wedge can account for both the velocity inversion and the
AVO anomaly (Fig. 30). Using the log information from the dry well, the Zoeppritz ray-trace modeling was carried-out to test the
AVO response of the sand without the inclusion of 2-D effects due to the wedge. The 1-D synthetic CMP gather in Figure 31 does not
show an increasing AVO response at the target depth (arrow), but does show high amplitudes at all offsets for the sand reflector.

To test the effect of the wedge on the AVO response of the sand, a wedge model was created. The wedge creates not only an amplitude
anomaly via tuning but also an AVO anomaly similar to that at the prospect (Fig.32).

Processing & Pitfall of AVO Analysis


31
By : Sigit Sukmono
Figure 25. Relative amplitude plot across a producing gas field. Bright colors correspond to high amplitudes. (Allen & Peddy,
1993)

Processing & Pitfall of AVO Analysis


32
By : Sigit Sukmono
Figure 26. Relative amplitude section showing the prospect, downdip from the field but on the same line as in Figure 25. The
amplitude response of the fields and prospect are very similar. (Allen & Peddy, 1993)

Processing & Pitfall of AVO Analysis


33
By : Sigit Sukmono
Figure 27. Velocity inversion across producing field. Change from reds and yellows to blues corresponds to the onset of
geopressure. The velocity anomaly corresponding to the field indicated that it is low relative to the surrounding rocks. (Allen &
Peddy, 1993)

Processing & Pitfall of AVO Analysis


34
By : Sigit Sukmono
Figure 28. Velocity inversion results across the prospect. The same velocity anomaly observed at the producing field is observed at
the prospect (Allen & Peddy, 1993).

Processing & Pitfall of AVO Analysis


35
By : Sigit Sukmono
Figure29. (a) Relative amplitude CMP gather from line crossing the producing field. The target reflection (arrow) exhibits a moderate
increase in amplitude versus offset. (b) Same seismic line as in (a), but across the prospect. The AVO response of the prospect is a good
match for that of the CMP gather from the field. Shooting skips left no near-offset traces at both field and prospect locations (Allen &
Peddy, 1993).

Processing & Pitfall of AVO Analysis


36
By : Sigit Sukmono
Figure 30. A wedge can cause anomalous velocities during velocity inversion, possibly explaining the anomaly the
anomaly in Figure 61 occurring in the wedge forming the prospect. (Allen & Peddy, 1993)

Processing & Pitfall of AVO Analysis


37
By : Sigit Sukmono
Figure 31. Zeoppritz ray trace modeling based on the logs from the test well. The wet sand does not cause an increase in
amplitude with offfset in this model, but does cause a strong reflection at all offsets, accounting for the bright stacked response
in Figure 29 (Allen & Peddy, 1993).

Processing & Pitfall of AVO Analysis


38
By : Sigit Sukmono
Figure 32. 2-D elastic modeling of a wedge illustrates that an AVO anomaly can be produced by tuning. (a) Twelve-fold stacked
section wedge model of prospect. AGC applied to section. (b) Selected relative amplitude CMP gathers through wedge. CMP gathers
within the wedge at the position corresponding to the case history prospect exhibit an increase in amplitude with offset due to tuning
within the wedge. CMP gathers outside the wedge tip have an unchanging AVO response. Wedge: Vp=2459 m/s, density= 2.18 g/cc,
Vs=1448 m/s. Shale: Vp=2585 m/s, density=2.21 g/cc, Vs=1231 m/s (Allen & Peddy, 1993).

Processing & Pitfall of AVO Analysis


39
By : Sigit Sukmono
Exercise I

AVO Analysis of Gas Sands Prospects

(This exercise is taken from : A.K. Srivastava, Singh, V., Tiwary, D.N., and Rangachari, V., 2006, The Leading Edge, 1344 – 1358).

In this Exercise students learn how to analyze the log data, log cross-plot, AVO modeling, AVO analysis and effect of phase variation on
the application of AVO for gas-sands prospecting.

Please answer the questions given in each figure.


.

Processing & Pitfall of AVO Analysis


40
By : Sigit Sukmono
Figure E1. Index map of Krishna Godavari Basin showing study area and the Generalized stratigraphy of Krishna Godavari Basin
(Srivastava.et al, 2006)

Processing & Pitfall of AVO Analysis


41
By : Sigit Sukmono
Figure E2. Sample logs and litho column for well A.

Question : using the logs, determine the hydrocarbon


bearing sands.

Processing & Pitfall of AVO Analysis


42
By : Sigit Sukmono
Figure E3. Cross-plot analysis between different log parameters in the interval 2070-2150 m in well A (Srivastava.et al,
2006). Question : What elastic property and seismic method are the best to distinguish gas sand and non-reservoir facies ?

Processing & Pitfall of AVO Analysis


43
By : Sigit Sukmono
Figure E4. Synthetic gathers generated with 30-Hz, zero-, and minimum phase Ricker wavelets. Question : Using the
given log data and offset gather determine the tops and bottoms of 4 gas sand intervals (Gathers section uses SEG reverse
polarity). Which one gives better modeling results : Zero phase or Minimum Phase ?

Processing & Pitfall of AVO Analysis


44
By : Sigit Sukmono
0.100

0.075

0.050

0.025

0.000

-0.025

-0.050

-0.075

-0.100

-0.125

Figure E5. AVO attributes intercept (A) and gradient (B) are computed using a least squares straight line
method from NMO-corrected gathers by cross-ploting the amplitude with offset or sin2θ corresponding
to each time sample (Srivastava.et al, 2006). These two attributes provide the basic description of the
AVO behavior.

Question : On the gather (left figure), intercept and gradient (right figures) displays determine the top
and bottom of a gas sand layer corresponding to the amplitude – offset curve given in the middle figure.

Processing & Pitfall of AVO Analysis


45
By : Sigit Sukmono
Figure E6. Panels showing AVO
attributes A∗B and their cross-
plot generated from (a) 30-Hz,
minimum and (b) zero-phase
synthetic gathers (Srivastava.et
al, 2006).

Question : Based on the


intercept-gradient cross-plot,
determine the top and bottom of
gas sand layers on the gather
displays. Which one is better, :
minimum or zero-phase gather ?

Processing & Pitfall of AVO Analysis


46
By : Sigit Sukmono
Figure E7. Logs, original real, and dephased gathers at well A. The wavelets extracted from the original real gather by the semi-deterministic
method and extracted from dephased gathers are also shown (Srivastava.et al, 2006). Question : On the real gather data, which gas sands are
detectable ? Determine tops and bottom of these gas sands. Which one is better : Zero phase or minimum phase. Compare your findings with
question in Figure E4.

Processing & Pitfall of AVO Analysis


47
By : Sigit Sukmono
Figure E8. Panels showing
AVO attributes A∗B and
cross-plot generated from
real (a) and dephased (b)
super gathers at well A.
(Srivastava.et al, 2006).

Question : Based on the


intercept-gradient cross-
plot, determine the top and
bottom of gas sand layers
on the gather displays.
Which gas sands are
detectable ? Which one is
better : minimum or zero-
phase gather? Compare
your findings with question
in Figure 6.

Processing & Pitfall of AVO Analysis


48
By : Sigit Sukmono
Figure E9. Seismic section showing a bright spot on the stacked data in an area near well A (Srivastava.et al, 2006).
.
Processing & Pitfall of AVO Analysis
49
By : Sigit Sukmono
Figure E10. Panels
showing AVO attributes
A∗B and cross-plot
generated from real (a) and
90o phase-rotated (b) super
gathers close to the bright
spot (Srivastava.et al,
2006).

Question : Can you


identify the differences in
the amplitude response
before and after phase
rotation? Pick top and
bottom of gas sand in the
gather display. Which
phase gives better result ?
Refer to Figure E11 to
answer your question.

Processing & Pitfall of AVO Analysis


50
By : Sigit Sukmono
Figure E11. A-B cross-plot of (a) original real gather, (b) 90o phase-rotated super gather, and (c-i) A-B cross-plot for 21 phase-
rotated and NMO-corrected gathers in 30o steps (Srivastava.et al, 2006).
Figure E12. Logs and
litho column for well B
(Srivastava.et al, 2006).

Question :
By referring to Figures
E12, 13 and 14, can you
identify gas sand
intervals ?
Figure E13. Logs, synthetic zero- and minimum-phase gathers along with original real gathers at well B (Srivastava.et al, 2006). .
Question : By referring to Figures E12, 13 and 14, can you identify gas sand intervals ?

Processing & Pitfall of AVO Analysis


53
By : Sigit Sukmono
Figure E14. Panel showing AVO attribute A∗B and cross-plot generated from zero-phase, synthetic NMO-corrected
gathers at well B (Srivastava.et al, 2006). Question : By referring to Figures E12, 13 and 14, can you identify gas
sand intervals ?

Processing & Pitfall of AVO Analysis


54
By : Sigit Sukmono
DT

Figure E15. Logs and litho column


of well C.

Question :
Using the log data, can you
identify gas sand intervals ?
Figure E16. Seismic section with inserted P-wave curve at well C. Question : Based on the log response in Figure E15, can
you identify gas sand intervals ?

Processing & Pitfall of AVO Analysis


56
By : Sigit Sukmono
Exercise II

AVO Analysis for Fluid Discrimination

(This exercise is taken from : K..T. Young & Tatham R.H., 2007, Fluid discrimination of post-stack “bright spots” in the Columbus
basin, offshore Trinidad, the Leading Edge, 1508-1515).

In this Exercise students learn how to analyze the log data, AVO cross-plot, modeling and analysis for fluid discrimination.

Please answer the questions given in each figure.

Background

Bright spot analysis of stacked seismic data has been used as an effective method of hydrocarbon exploration in many areas around the
world, including Trinidad. The unconsol­idated Tertiary sediments of the Columbus Basin, offshore Trinidad, create geologic conditions
that are particularly favorable to the existence and detection of seismic ''bright spots" associated with gas deposits. This success of the
bright spot technique as a direct hydrocarbon indicator is based on the premise that gas-bearing sands will exhibit seismic-reflection
amplitudes that are anomalously higher than surrounding reflections. There are, however, several other combinations of geological
conditions that can give rise to large amplitudes such as high porosity, clean sands, tuning effects, over pressured shales or anisotropy.

In this exercise AVO LMR technique is used to minimize the ambiguity of amplitude anomaly for fluid discrimination. The results used
also for identifying potential development area.

Processing & Pitfall of AVO Analysis


57
By : Sigit Sukmono
Figure II.1. Data preconditioning using super gathers and trim statics. Figure II.2. Comparison of prestack synthetic seismogram
Note the reduced noise in the second panel and the straighter events in modeling related with POSITIVE gas interval to observed CMP
the third panel. gather with NMO corrections applied.

Processing & Pitfall of AVO Analysis


58
By : Sigit Sukmono
Figure II.3. A post-stack amplitude map of the
study area and associated offset displays of
several amplitude anomalies.

Question
Based on modeling results in Fig.II.2, Identify
amplitude anomaly area associated with positive
class III AVO anomaly.

Processing & Pitfall of AVO Analysis


59
By : Sigit Sukmono
Figure II.4. (a) Vp versus Vs crossplot with data points colored by lithologies. The gas-saturated sediments are highlighted by the pink polygon
and have a smaller Vp and Vp/Vs ratio than brine-saturated sediments. (b) Plot of density and true vertical depth. Shales generally have higher
densities than sands, and gas-saturated sediments have significantly lower densities than brine-saturated sediments and from a separate trend.

Processing & Pitfall of AVO Analysis


60
By : Sigit Sukmono
B

Figure II.5. An illustration showing bright spots


anomalies of A-B-C around a discovery well in
Field-1.
3500 ms
Question
Using results in Figures II.6-8 identify the gas
bearing sands

Processing & Pitfall of AVO Analysis


61
By : Sigit Sukmono
Figure II.6. λρ section for Field 1 with horizons A, B and C Figure II.7. µρ (matrix indicator) section for Field 1.
identified. Question. Identify the gas bearing sands Question. Identify lithology of A-B-C horizons

Processing & Pitfall of AVO Analysis


62
By : Sigit Sukmono
Figure II.8. Narrowing the range of displayed λρ and µρ values to determine the truly anomalous low and high zones
respectively in Field 1

Processing & Pitfall of AVO Analysis


63
By : Sigit Sukmono
Figure II.9. Field-1 map view of λρ (left) and µρ (right) extracted along horizon A. Question. Identify the gas
bearing sands area

Processing & Pitfall of AVO Analysis


64
By : Sigit Sukmono
Figure II.10. Cross-plot of λρ and µρ seismic volumes for zone Figure II.11. Cross-section of field 1 with LMR cross-plot attribute.
analysis. The data points on the cross-plots are from the same data
volumes in Figures II.6-7. .

Processing & Pitfall of AVO Analysis


65
By : Sigit Sukmono
Figure II. 12. LMR cross-plot attribute extracted
along three target horizons in Field 3. Question.
Identify distribution of sands and the gas bearing
sands area

Processing & Pitfall of AVO Analysis


66
By : Sigit Sukmono
Figure II.13. Summary of the ranges of λρ values in GPa*g/cc for Figure II.14. Summary of the ranges of λρ values in GPa*g/cc for each
each field. Gas sands are in red bars, brine sands are in blue, and field. Gas sands are represented are in red bars, brine sands are in blue
shales are in gray bars. Note the separation of the gas-and brine- and shales are in grey bars. Note the separation of the sandy and shaly
saturated lithologies and that λρ is independent of depth. Note that lithologies regardless of fluid saturant and the prevailing depth trend on
all nine of the gas-saturated zones are correctly discriminated from µρ. It is also clear that the separation of lithological properties is not as
the nonproductive zones. pronounced as it is with fluid properties.

Processing & Pitfall of AVO Analysis


67
By : Sigit Sukmono
Exercise III

AVO Analysis for Stratigraphic Play Exploration

(This exercise is taken from : May, J.A.. et al. , 2007, Amplitude analysis in a sequence stratigraphic framework : exploration successes
and pitfalls in a subgorge play, Sacramento Basin, California,, the Leading Edge, 1516-1526).

In this Exercise students learn how to apply AVO for analyzing stratigraphic play in field explorationn.

Please answer the questions given in each figure.

Background

The Sacramento Basin is part of the Great Valley, a prolific hydrocarbon province that is the remnant of a late Mesozoic-early Cenozoic
forearc basin in California. A series of buried submarine canyons extend seaward from the eastern margin of the forearc. These "gorges"
formed during multiple episodes of relative sea-level fall during the Tertiary, truncating Late Cretaceous through Eocene marine and
nonmarine sandstones. Mud­stones dominate the canyon fills, creating lateral and top seals for numerous gas reservoirs (Figure III.1).

In this exercise AVO technique is used to minimize the ambiguity of amplitude anomaly in stratigraphic prospect analysis.

Processing & Pitfall of AVO Analysis


68
By : Sigit Sukmono
Figure III.1. (a) Base map for
the Sacramento Basin, showing
locations of buried Tertiary
submarine canyons and the
proprietary 3D seismic survey
extending across the Magenos
Gorge. (b) Stratigraphic section.
The succession of Tertiary
submarine canyons is shown.

Processing & Pitfall of AVO Analysis


69
By : Sigit Sukmono
Figure III.2. Time structure map for
sequence boundary at the base of the
Meganos Gorge. Subgorge gas fields and
prospects (A1, A2, A3, B, and C) are
labeled. The structural pattern is indicated
by a gradational color bar (from the
shallowest portion at 1080 ms in yellow to
the deepest areas at 1866 ms in dark
purple). The main canyon thalweg trends
NE-SW, with a smaller tributary entering
from the north.

Processing & Pitfall of AVO Analysis


70
By : Sigit Sukmono
Figure III-3. Arbitrary seismic line extending
from King Island gas field to Prospect A1
(location shown on the inset map). In this
variable-density display, the seismic troughs are
presented in red, grading though white at the
zero crossing, with the peaks in blue. The
strongest trough amplitudes are highlighted in
yellow and the strongest peak amplitudes are in
cyan. Note the amplitude anomaly at King
Island Field and analogous signature at
Prospect A1. The green line indicates the
erosional base of the Meganos Gorge. Seismic
character for the mudstone-dominated canyon
fill is discontinuous and chaotic; seismic
reflections of the Mokulemne River section
outside the canyon are laterally continuous.

Processing & Pitfall of AVO Analysis


71
By : Sigit Sukmono
Figure III.4. Velocity field developed from well logs in the project area. Velocity-versus-depth is plotted, and
demonstrates different trends for gas-charged sandstones, water-wet sandstones, and shales. Note that the gas-
charged sandstones display much slower velocities than the other two lithologies down to about 9500 ft.

Processing & Pitfall of AVO Analysis


72
By : Sigit Sukmono
Figure III.5. Velocity effects due to gas.
(a) Well log for the Quintana 1 Moresco
et al., King Island Field. Vertical scale in
ft. Gas in the reservoir sandstone creates
a large decrease in acoustic travel time
(sonic velocity). The well has produced
almost 5 bcf of gas from 55' of net-pay.
(b) Arbitrary seismic line across King
Island Field. This is a variable-density
display, with the seismic-wavelet peaks
in black and grading to the seismic
troughs in light gray. The color bar
emphasizes the strongest troughs in red
to yellow and the strongest peaks in blue
to cyan. The gamma-ray and resistivity
curves are presented for the productive
well, the Quintana 1 Moresco et al. The
sequence boundary at the base of the
Meganos submarine canyon is shown in
green.

Processing & Pitfall of AVO Analysis


73
By : Sigit Sukmono
Figure III.6. Well logs for
various lithologies that
display amplitude
anomalies. Vertical scales
are in feet. Both lignites (a)
and low-velocity shales (b)
produce large decrease in
the sonic and density
readings. In contrast,
conglomerates (c) and
carbonate-cemented
sandstones “bones” (d)
yield large increase in the
sonic and density values.

Processing & Pitfall of AVO Analysis


74
By : Sigit Sukmono
Figure III.8. AVO modeling. (a) P-wave
velocity, S-wave velocity, Poisson's ratio,
and P-wave/S­wave ratio are shown by the
blue curves for the Coastal 1-32 Live Oak
well. For modeling, the data are blocked,
indicated in red. The seismic wavelet at the
well location is displayed in black and the
synthetic seismogram is in blue. In this
well, a wet sandstone overlies a bone,
producing a strong peak amplitude on top of
a strong trough. The amplitudes decrease
with increasing offset. (b) A gas-charged
sandstone is substituted for the wet
sandstone, and the concomitant changes in
P-wave velocity, S-wave velocity, Poisson's
ratio, and P-wave/S-wave ratio are shown.
The result is a strong trough over a peak,
with the trough amplitude increasing with
farther offsets. (c) Removing the bone, but
leaving the wet sandstone, yields a much
weaker peak amplitude, which decreases in
amplitude with offset.

Processing & Pitfall of AVO Analysis


75
By : Sigit Sukmono
Figure III.9. AVO results for
King Island Field. The negative
(trough) amplitude anomaly
just below 1300 ms coincides
with the gas reservoir. The
amplitude of the trough
increases with increasing offset
for both the (a) scaled raw
gathers and (b) supergathers.
The color bar illustrates the
range in amplitudes for the
peaks (in cyan and blue) and
troughs (in yellow and red).
The data are presented for
seismic line 1603, with
crossline numbers 293 to 296
shown along the top for each
offset gather. (c) Cross plot of
the AVO gradient and P-wave
intercept. Most of the data
points along the amplitude
anomaly at -1300-1310 ms
(shaded area) fall in the
southwest quadrant, typical of a
Class III AVO response.

Processing & Pitfall of AVO Analysis


76
By : Sigit Sukmono
Figure III.10. AVO results for lithologies that
produce amplitude anomalies, but are not
gas-charged sandstones. (a) Gathers and (b)
supergathers for a lignite in the Domengine
section at -1260 ms on seismic line 1517.
The amplitude of the trough decreases with
increasing offset. (c) Gathers and (d)
supergathers for a presumed wet and/or
carbonate-cemented Winters sandstone at
2185 ms on seismic line 1841. The amplitude
of the peak and the underlying trough both
decrease with increasing offset.

Processing & Pitfall of AVO Analysis


77
By : Sigit Sukmono
Figure III.11. Seismic images for gorge-margin
prospects A1, A2, and A3. (a) Perspective view
with amplitude draped on structure. Three separate
anomalies occur along a ridge that plunges to the
southwest. The ridge is a topographic remnant
produced by erosion along the canyon margin. Two
nearby dry holes are shown by the green wellbores.
(b) Arbitrary seismic line down plunge along the
ridge, showing the separate amplitude anomalies
for prospects A1, A2, and A3.

Processing & Pitfall of AVO Analysis


78
By : Sigit Sukmono
Figure III.12. AVO
response for Prospect A1. A
negative (trough) amplitude
anomaly at 1400 ms
coincides with the
presumed gas reservoir. The
amplitude of the trough
increases with an increase
in offset for both the (a)
scaled raw gathers and (b)
supergathers. The color bar
shows the range in
amplitudes for the peaks
and troughs. These data are
presented for seismic line
1403, with crossline
numbers 206 to 209 shown
along the top for each offset
gather. (c) Cross-plot of the
AVO gradient and P-wave
intercept. Most of the data
points along the amplitude
anomaly (∼1400 ms) fall in
the southwest quadrant,
typical of a Class III AVO
gas response.

Processing & Pitfall of AVO Analysis


79
By : Sigit Sukmono
Figure III. 13. Seismic images for gorge-bottom prospects B and C. (a) Arbitrary seismic line extending from King Island
Field through prospects Band C (location shown on the inset map). This variable-density display uses the same color bar
as described for Figure 5. Note the negative (trough) amplitude anomaly that corresponds to the gas reservoir at King
Island Field and the analogous signatures at the prospect locations. The sequence boundary at the base of the Meganos
Gorge is shown in green. (b) Perspective view of King Island Field and prospects Band C. The trough amplitudes are
draped on structure. The base of the Meganos Gorge is shown as a translucent red surface.
Processing & Pitfall of AVO Analysis
80
By : Sigit Sukmono
Figure III. 14. Montage for Prospect B. The amplitude map (a) portrays a negative (trough) anomaly that trends east-
west. The total area covered by the anomaly is about 130 acres; the strongest (eastern) portion covers an area of -80
acres. The horizontal scale (1 mile) is shown below the figure. A time slice (b) at 1704 ms cuts the shallowest portion of
the anomaly. Two arbitrary seismic lines provide strike (c) and dip (d) views of the anomaly beneath the Meganos Gorge
sequence boundary. Locations of the lines are shown in (a).

Processing & Pitfall of AVO Analysis


81
By : Sigit Sukmono
Figure III. 15. Montage for Prospect C. The amplitude map (a) portrays a negative (trough) anomaly that trends east-west..
The total area covered by the anomaly is about 80 acres; the strongest (eastern) portion covers an area of ~47 acres. The
horizontal scale (1 mile) is shown below the figure. A time slice (b) at 1640 ms cuts the shallowest portion of the anomaly.
Two arbitrary seismic lines provide strike (c) and dip (d) views of the anomaly beneath the Meganos Gorge sequence
boundary. Locations of the lines are shown in (a).

Processing & Pitfall of AVO Analysis


82
By : Sigit Sukmono
Bibliography
1. Allen J.L, and Peddy, C.P., 1993. , Amplitude Variation with offset: Gulf coast case studies, Geophys. Dev. Series, Vol 4, SEG.
2. Berkhout, A.J., 1985, Seismic resolution: Geophysical Press.
3. Brown, R.L., 1992, Estimation of AVO in the presence of noise using prestack migration; 62nd Ann. Internat. Mtg., Soc. Expl.
Geophys., Expanded Abstracts, 885-888.
4. Canning, A., 1993, Organization and imaging of three-dimensional seismic reflection data before stack: Ph.D.thesis, Rice
University.
5. Chiburis, E.F., 1984, Analysis of amplitude versus offset to detect gas/oil contacts in the Arabian Gulf: Presented at the 54th Ann.
Internat. Mtg. Soc. Expl. Geophys., Atlanta.
6. —1987, Studies of Amplitude versus Offset in Saudi Arabia: 57th Ann. Internat. Mtg., Soc. Expl. Geophys, Expanded Abstracts,
1091-1094.
7. De Beukelaar, P., Ricarte, P., and Richard, V., 1991, AVO processing and Analysis for a 2-D heterogeneous reservoir: 61st Ann.
Internat. Mtg., Soc. Expl. Geophys., Expanded Abstracts Volume II, 1091-1094.
8. Dey-Sarkar, S. K., Svatek, S. A., and Crewe, 0. W ., 1986, Prestack analysis: data processing: 56th Ann. Internat. Mtg., Soc. Expl.
Geophys., Expanded Abstracts, 339-341.
9. Duren, R. E., 1991, Seismic range equation: Geophysics, 56, 1015-1026.
10. Gassaway, G. S., Brown, R. S., and Bennett, L. E., 1986, Ptifalls in seismic amplitude analysis: case histories: 56th Ann. Internat,
Mtg. Soc. Expl, Geophys., Expanded Abstracts, 332-334.
11. Keho, T.H., Smith, S. W., Abarr, G. J., and Hargrove, K. L" 1992, An example of uncollapsed migration followed by AVO: Joint
SEG/EAEG Summer Research Workshop, Expanded Abstracts, 314-315.
12. Lanfranchi, P ".1992, How good is good pre-processing?: Joint SEG/EAEG Summer Research Workshop, Expanded Abstracts, 317 -
327.
13. Lee, S., Ho, M. S., and Purnell, G. W ., 1991, Preprocessing for amplitude-versus-angle (AVA) analysis of land seismic data: 61st
Ann. 1nternat. Mtg., Soc. Expl. Geophys., Expanded Abstracts, 1070-1071.

Processing & Pitfall of AVO Analysis


83
By : Sigit Sukmono
14. Lindsey,J. P., 1990, The Fresnel zone and its interpretive significance, The Leading Edge, reprinted in Seismic Interpretation
Series 2, 37-43.
15. May, J.A.. et al. , 2007, Amplitude analysis in a sequence stratigraphic framework : exploration successes and pitfalls in a
subgorge play, Sacramento Basin, California,, the Leading Edge, 1516-1526
16. Martinez, R. D., 1992, Constrained seismic data processing sequence for AVO analysis including a case history: 62nd Ann.
Internat. Mtg., Sac. Expl. Geophys., Expanded Abstracts, 835-838.
17. Mazotti, A., and Mirri, S., 1991, An experience in seismic amplitude processing: First Break, 9, 65-73.
18. Resnick, J. R., Ng, P ., and Lamer, K., 1987, Amplitude versus offset analysis in the presence of dip: 57th Ann. Internat. Mtg.,
Soc. Expl. Geophys., Expanded Abstracts, 617 -620.
19. Ross, C. P., and Beale, P. L., 1992, Offset balancing: Joint SEG/EAEG SummerResearch Workshop, Technical Program and
Abstracts, 340- 349.
20. Srivastava, A.K., Singh, V., Tiwary, D.N., and Rangachari, V., 2006, Impact of phase variations on quantitative AVO analysis :
21. An example from Krishna Godavari Basin, India , The Leading Edge, 1344 – 1358.
22. Taner, M. T., and Koehler, F., 1981, Surface consistent corrections: Geophysics, 46, 17-22.
23. Treadgold, G. E., Ritchie, K., and Dey-Sarkar, S. K., 1990, AVO: an example of processing pitfalls: 6Oth Ann. Internat. Mtg.,
Sac. Expl. Geophys., Expanded Abstracts, 1487 -1490.
24. Walden,A. T., 1991, Making AVO sections more robust: Geophys.Prosp., 39, 915-942.
25. Young, K.T. & Tatham R.H., 2007, Fluid discrimination of post-stack “bright spots” in the Columbus basin, offshore Trinidad,
the Leading Edge, 1508-1515
26. Yu, G., 1985, Offset-amplitude variation and controlled-amplitude processing: Geophysics, 50, 2697-2608.

Processing & Pitfall of AVO Analysis


84
By : Sigit Sukmono
Neural Networks Seismic Multi-Attribute
Analysis for Lithology & Porosity Mapping

Neural Networks Seismic Multi-Attributes for Lithology & Porosity Mapping


By : Sigit Sukmono 1
1. Methodology

Seismic multi-attribute analysis is a broad term that encompasses all


geostatistical methods using more than one attribute to predict some physical
property of the Earth. Schultz et al. (1994) identified three major subcategories
of geostatistical multi-attribute analysis techniques:
1) The extension of cokriging to include more than one secondary attribute to
predict the primary parameter.
2) Methods that use the covariance matrix to predict the parameter from a
linearly weighted sum of input attributes.
3) Methods that use artificial neural networks (ANNs) or nonlinear optimization
techniques to combine attributes into an estimate of the desired parameter.

The multi-attribute analysis technique discussed here falls under the second
category. A complete discussion on the procedure of multi-attribute analysis to
predict log properties is provided by Russell et al. (1997) paper as enclosed in
the appendix. In general, the multi-attribute algorithm will look for a function to
convert m different attributes into the desired property. This may be written:

P (x,y,z) = F [A1(x,y,z), … , Am (x,y,z)] (1)

where P = the log property as a function of coordinates x,y,z; F = the


functional relationship; and Ai = the m attributes.

In the simplest possible case, it would be a linearly weighted sum:

P = w0+ w1 A1 + ... wm Am (2)

where wi (i = 0, ..., m) = the weights.

The number and type of attributes are determined geophysically and


statistically. First, the algorithm will investigate the applicability of internal
seismic attributes (such as amplitude envelope, amplitude weighted cosine
phase, amplitude weighted frequency, etc.). These internal attributes are used
by themselves or combined with other attributes.

Followings are examples on the application of multi-attributes techniques for


lithology and porosity mapping.

Neural Networks Seismic Multi-Attributes for Lithology & Porosity Mapping


By : Sigit Sukmono 2
2. Case Study-1 : Application of multi-attribute analysis in mapping
lithology and porosity in the Pematang-Sihapas groups of Central
Sumatra Basin, Indonesia
2.1. Introduction

The Central Sumatra back-arc basin in the central portion of Sumatra Island,
Indonesia (Figure 1), is one the country’s main hydrocarbon-producing basins.
Two of the largest oil fields in Indonesia, Minas and Duri, are in this basin.

Figure 2 shows that the oldest unit in the basin, the Pematang Group, was
deposited more than 25.5 million years ago. This group has three formations:
(oldest to youngest) the Lower Red-Bed, the Brown Shale, and the Upper
Red-Bed. The Lower Red-Bed formation consists of claystone, siltstone,
fanglomerate and conglomerate deposited in fluvial, lacustrine and deltaic
environment. The Brown Shale Formation, with an average thickness of
around 600 ft, consists of brown shale deposited in a lacustrine environment.
The Upper Red-Bed Formation consists of sandstone, conglomerate and red-
green shale deposited under lacustrine and fluvial environments.

Deposited unconformably above the Pematang Group from 16.5 to 25.5


million years ago is the Sihapas Group. The main formations, from oldest to
youngest, are Menggala, Bangko, Bekasap, and Duri. Menggala Formation
has an average thickness of 1800 ft and consists of conglomeratic-to fine-
grain sandstone deposited under a fluvial channel-to-braided stream
environment. This formation changes lat­erally and vertically to the west into a
marine shale of the Bangko Formation, and to the east into the Bekasap
For­mation. Bangko has an average thickness of 300 ft and consists of a silty-
to-carbonaceous shale deposited under shelf to delta-plain environments.
Bekasap has an average thick­ness of 1300 ft and mainly consists of
glauconitic sandstone with an intercalation of shale, limestone and coals,
deposited under estuarine, intertidal, and inner-outer neritic envi­ronments.
Duri Formation has an average thickness of 900 ft and mainly consists of fine-
to-coarse grained sandstone deposited under barrier bar and delta front
environments.

Deposited unconformably above the Sihapas Group from 13 to 21 million


years ago is Telisa Formation which has an average thickness of 1600 ft and
consists mainly of shale with calcareous silt intercalations.

In the Central Sumatra Basin, the Brown Shale Formation of the Pematang
Group is the main source rock. The Menggala, Bekasap, and Duri formations
are the main reservoirs, and the Bangko and Telisa formations are the main
sealing rocks. Therefore, the ability to identify and to map the shales and
sands in the Pematang and Sihapas groups is very important in hydrocarbon
exploration.

Neural Networks Seismic Multi-Attributes for Lithology & Porosity Mapping


By : Sigit Sukmono 3
The fluvial-lacustrine-deltaic depositional environment of Pematang and
Sihapas causes the sand distribution to be very discontinuous in relatively short
distances (Figure 3). The discontinuous and onlapping nature of the sands
makes well-to-well correlation on seismic very difficult. Consequently, in this
area, seismic data and the acoustic impedance inversion derived from the
seismic data are often interpreted incorrectly due to overlapping impedance
values of the sands and shales.

2.2. Crossplot analysis of log properties.

To determine the best seismic method for identifying and mapping sand and
shales in the Pematang and Sihapas groups, Sukmono (2000) did the cross-
plot analysis of log properties.

Density, sonic, and gamma-ray logs have been collected from 432 wells in the
Central Sumatra Basin. The density and sonic logs are used to compute AI, and
the gamma-ray data to differentiate sands and shales. Figure 4a shows a
typical crossplot for younger formations (e.g., Telisa and Duri) where sands
have lower porosity and, thus, lower AI than shales. For this case, AI seismic
inversion is recommended for mapping sands and shales.

The sands and shales in older formations (e.g., Bekasap, Menggala, and
Pematang) typically have overlapping porosity and AI values (Figure 4b). In this
case, multi-attribute analysis is applied to generate pseudo gamma-ray and
pseudoporosity logs which are subsequently used in the creation of the sand,
shale, and porosity maps.

2.3. Multi-attribute analysis

The multi-attribute algorithm computes a training data set using available


original logs. This will produce two outcomes :(1) a possible statistical
combination of the internal attributes and (2) the training error, which measures
the match between the predicted and the original log.

The algorithm checks the validation error of every attribute combination by


excluding one well in the training process. The target log is predicted using the
attribute set in the training process. This predicted log is then compared to the
original log and the validation error computed.

Figures 5 and 6 illustrate the process of finding the best combination of internal
attributes for predicting the gamma-ray and porosity logs. The analysis table
shows the type of attributes and the related training and validation errors. The
optimal number of attributes is shown by the minimum validation error value. In
this case, the optimal number was seven for gamma ray and four for porosity.

Neural Networks Seismic Multi-Attributes for Lithology & Porosity Mapping


By : Sigit Sukmono 4
Figure 7 compares the original and predicted gamma-ray and porosity logs, and
the average prediction errors for the seven wells used in the training process.
For example, during the gamma-ray log prediction, the correlation coefficients
and the prediction errors are 0.79 and 18.29 (which means the error is around
10% from maximum measured gamma ray of 175). For porosity, the correlation
coefficients and the prediction errors are 0.93 and 1.5%. These coefficients and
error levels were considered very good, so the entire seismic cube was
processed using these parameters.

Figure 8 shows the resulting pseudo gamma-ray and porosity sections


generated from the related seismic section. The original logs are inserted.
There is a good match between the generated pseudolog sections and the real
logs. To illustrate the benefit of using a multi-attribute technique for guiding well-
to-well sand body correlations, Figure 9 compares a normal ampli­tude section
(top) and a pseudo gamma-ray sec­tion (bottom). In the normal amplitude
section, because the impedance values are not entirely diagnostic of sands or
shales, individual sands are not visible. In the pseudo gamma-ray section,
however, the sand and shales identified in the wells can be tied fairly
consistently. The sands are associated with red (low gamma ray) and shales
with green (high gamma ray).

2.4. Validation of results using infill well data.

For additional verification of the validity of the tech­nique, Sukmono et al. (2003)
made several com­parisons of the predicted sand distribution with the actual
distribution determined from recently acquired infill well data in Minas Field. The
upper panel of Figure 10 shows the actual drilling results and the lower panel
shows the pseudo gamma-ray section resulting from multi-attribute analysis.
The red wireline logs on the left on each well in the upper panel are the actual
gamma-ray logs; sand bodies associate with lower gamma ray (highlighted by
blue circles). Red and green in the pseudo gamma-ray section (lower panel)
associate with low and high or sand and shale, respectively. The infill data show
that the sand thickness varies abruptly over a relatively short distance, but the
variation can still be modeled by the pseudo gamma-ray section derived from
multi-attribute analysis; the average error between the predicted gamma-ray
and the infill wells is 10%.

Neural Networks Seismic Multi-Attributes for Lithology & Porosity Mapping


By : Sigit Sukmono 5
Figure 1. Location of Central Sumatra back-arc basin (Sukmono, 2007).

Figure 2. Regional Stratigraphic Column of Central Sumatra back-


arc basin (Sukmono, 2007) .
Neural Networks Seismic Multi-Attributes for Lithology & Porosity Mapping
By : Sigit Sukmono 6
Figure 4. Illustration of the discontinuous lateral distribution of the Pematang and
Sihapas groups due to fluvial-lacustrine-deltaic depositional environments. The
distance between the two wells in the figure is approximately 500 m (Sukmono,
2007) .

Figure 3. Log property crossplots of (a) younger formations (Duri and Bekasap),
show that AI of the sands is higher than the shales and (b) older formations
(Bekasap, Menggala, and Pematang) show that AI of sands and shales overlaps
(Sukmono, 2007) .
Figure 5. Illustration of the process of finding the best combination of
attributes for creating pseudo gamma-ray logs. Analysis table (top)
shows the types of attributes and related training and validation errors.

Figure 6. Illustration of the process of finding the best combination of


attributes for creating pseudoporosity logs. Analysis table (top) shows
the types Neural Networks Seismic
of attributes Multi-Attributes
and related for Lithology
training & Porosity Mapping
and validation errors
(Sukmono,By 2007)
: Sigit.Sukmono 8
Figure 7. (a) Comparison between original logs (black) and the pseudo
gamma-ray logs (red). Also shown is the average prediction error for the seven
wells used in the training. The data were analyzed only in the yellow-shaded
windows. (b) Comparison between original logs (black) and the pseudo-porosity
logs (red). Also shown is the average prediction error for the seven wells used
in the training. The data were analyzed only in the yellow-shaded windows
(Sukmono, 2007) .
Neural Networks Seismic Multi-Attributes for Lithology & Porosity Mapping
By : Sigit Sukmono 9
Figure 8. Example of (a) pseudo gamma-ray, and (b) porosity sections
generated from seismic using the multi-attribute process. For comparison,
the real gamma-ray and porosity logs are inserted in the sections.
sizes certain frequencies which increase the resolution and thus the ability of
seismic data to detect the lithology (Sukmono, 2007).

Neural Networks Seismic Multi-Attributes for Lithology & Porosity Mapping


By : Sigit Sukmono 10
Figure 9. Comparison between a normal ampli­tude section (top) and a
pseudo-gamma ray sec­tion (bottom) that illus­trates the improvement in well-
to-well sand correla­tions. In the amplitude section, the impedances are not
completely diag­nostic of sands and shale, thus masking the individ­ual sand
units. Using the pseudosection derived from the multiattribute approach, the
well-to-well sand correlation is easier to see because the sands consistently
relate to low gamma-ray (red) and the shales relate to high gamma-ray (green)
(Sukmono, 2007) .

Neural Networks Seismic Multi-Attributes for Lithology & Porosity Mapping


By : Sigit Sukmono 11
Figure 10. Comparison of multi-attribute predicted interwell sand body (white
circles in lower panel) correlations with actual sand body (blue circle in upper
panel) correlation from new infill well data. Analysis target interval is yellow. The
multi-attribute product shown in lower panel is pseudo-gamma ray section where
red is associated with low gamma-ray sands and green indicates high gamma-ray
shales. The infill wells are 8D-77-EP1 and 8D-78-EP1 (yellow boxes), the existing
wells are 8D-68, 8D-69, 8D-78 and 8D-79 (white boxes). The average distance
between the wells is 100 m. The pseudo-gamma ray section (lower panel) can
predict well the lateral discontinuity of the sands even though it occurs in very short
distance (Sukmono, 2007) .
3. Case Study-2 : Porosity and lithologic estimation using rock physics
and multi-attribute transforms in Balcon Field, Colombia

3.1. Background.

Balcon Field was discovered in 1988. The sandstone reservoir is hydrostatically


pressured without any gas cap and is buried at depths between 9000 and 10
000 feet. The reservoir is highly compacted and cemented with porosities
ranging from 10 to 15% and permeabilities between 30 and 400 md. The
characteristics of fluids are: GOR of 450 scf/bbl and 32° API of oil gravity.

The OOIP has been calculated several times with values that range from 88.3
to 52 MBLS (Amaya et al., 1999 and Taylor, 1992). This fact can reflect the
uncertainty that still exists with the static geologic model due to two main
factors: (1) the structure and (2) the spatial distribution of reservoir properties,
such as porosity and permeability.

In Balcon Field, the spatial distribution of reservoir prop­erties, which depends


on processes related to deposition and diagenesis, has been estimated using
well information, without consideration of the 3D seismic data. Only probabilistic
models have been used to determine the sand body distribution from well
logging, core data, and production well test data (Taylor, 1992). Some variations
in reservoir properties are observed between wells due to stratigraphic changes
(e.g., Balcon-4 and Colombina-1) (Figure 11). Such variations control the spatial
distribution of reservoir properties such as porosity and permeability which play
an important role in the management of this field because they determine the
storage capacity and the movement of the fluids into the reservoir.

The field’s 3D seismic data have previously only been used to define the
structural model and never have been used to predict petrophysical properties
at the reservoir level. The quality of the seismic data in most of the field is
negatively affected by the subsurface structure and surface topography, but the
seismic data are of good quality for the central part of the survey. This facilitates
the extraction of reservoir infor­mation from seismic data and allows reduction
of the uncer­tainty of reservoir properties at the interwell scale.

Seismic attributes such as acoustic impedance are often used to map lateral
variation of reservoir rock properties (e.g., porosity, lithology, fluids, etc). What
we need is to understand the relation between these properties and seis­mic
acoustic properties (i.e., velocity, acoustic impedance). We attempt not only to
take advantage of possible deter­ministic physical links between seismic elastic
properties and rock properties, but also to improve the predictive ability of such
properties derived from well logs through multi-transform techniques, which
combine acoustic impedance with other seismic attributes using multilinear
transforms and neural network inversion (NNI).

Neural Networks Seismic Multi-Attributes for Lithology & Porosity Mapping


By : Sigit Sukmono 13
3.2. Location and geologic setting.

The study area is located in the western Neiva subbasin within the Upper
Magdalena Valley, an elongated intermontaine basin, bounded by the Central
Cordillera to the west and the Eastern Cordillera to the east in Colombia, South
America (Figure 12). The present configuration of these mountains and the
subsurface struc­ture is the result of multiple phases of tectonic events
over­printed by the Tertiary Andean deformation style, which is dominantly
compressional and dip slip (Dengo and Convey, 1993). Basin fill is
predominantly siliciclastics, mostly marine Cretaceous and continental Tertiary
deposits (Figure 13). The Cretaceous sedimentation in the Neiva sub-basin,
started in the Middle Cretaceous (Aptian-Albian) with the deposition of the
Caballos Formation, a sequence of fluvio-deltaic sands and conglomerates,
which unconformably overlay the eroded Triassic/Jurassic basement (Taylor,
1992).

The stratigraphy and sedimentology of the Caballos Formation in the Neiva


Basin have been thoroughly studied by Sneider (1988), Sanchez and Robles
(1987), and Taylor (1992). From these studies, the Caballos Formation is
divided into three major units (Figure 14):
1) The lower unit, with 250 ft average thickness, is inter­preted as braided river
deposits and is composed of flu-vial-channel sandstones and conglomerates,
crevasse splay sandstones, and floodplain shales. The thickness of the
sandstones varies between 5 to 15 ft with porosities between 1 and 8%.
2) The middle unit, composed of about 100 ft of marine bay shale, is very
continuous through the field and acts as a seal between the lower and upper
unit.
3) The upper unit consists of approximately 200 ft of quartz sandstones
intercalated with siltstones and shales. This unit contains the main reservoir
formed by massive clean and well-sorted sandstones intercalated with finely
laminated shales. The total bed thickness of the reservoir ranges between 10
and 100 ft and porosities vary from 8 to 17%.

The main reservoir development is contained in the Upper Caballos


sandstones, interpreted as distributary channels. These sandstones are well
interconnected in most of the area of the field and are present in all of the wells.
Changes in thickness and lithology of this unit are inter­preted from the gamma-
ray curve response (Figure 4), which is a good indicator of clay content and the
reservoir qual­ity in the field. Based on the correlation between gamma ray and
core data we can see that readings in gamma ray above 75 API units indicate
high clay content (shale) and low reservoir quality. Gamma-ray measurements
below 50 API units indicate low clay content (clean sandstones) and good
reservoir quality in terms of porosity and permeabil­ity. As mentioned before,
changes in reservoir properties in the field are related to depositional process.
For example, clean sandstones are associated with distributary channel
sandstones while shales are associated with deposits of mud deltaic plain.

Neural Networks Seismic Multi-Attributes for Lithology & Porosity Mapping


By : Sigit Sukmono 14
3.3. Data and methodology.

The study area, which covers about 25 km2 in the central part of Balcon Field,
was selected for three main reasons: (1) the good quality of the surface seismic
(Figure 15) (2) a better definition of structural features, which is simpler than the
rest of the area, and (3) the well control represented by 11 wells, most of them
with a com­plete set of well logs. To transform seismic data into petrophysical
properties such as porosity, lithology, and clay content, we try to understand the
effect of each one of these variables on elastic wave properties. Rock physics
plays an important role in helping to establish the systematic relationships. Once
these relations are understood, seismic attributes, derived by seismic inver­sion
techniques, can be used to describe reservoir properties based on physical links
to seismic wave propagation.

Although rock-physics relationships provide the link between seismic attributes


and log properties, sometimes such links can be difficult to derive theoretically
from first principles (Schultz et al., 1994). In such cases multi-attribute
transforms and NNI can help to overcome the difficulty of having no linear
relation between acoustic impedance derived by seismic inversion and any rock
property derived from logs (Russell et al., 2001). This approach can be used
either to find statistical relations between seismic data and log properties or can
be combined with acoustic impedance to apply multi-attribute transforms and
NNI. In this study these methods were applied to improve the prediction of
reservoir properties in the field. One of the main advantages of this approach is
that it offers the possibility to invert seismic data for other parameters than
acoustic impedance, for example the gamma-ray response, which is a good
indica­tor of lithology and reservoir quality in the field. The key aspect of this
method is the selection of seismic attributes to be used either in the multi-
attribute linear analysis or in the neural network training.

By applying the method of stepwise regression it is possible to obtain the set of


attributes that are optimal in the sense that the mean-squared prediction error
between the actual target logs and the predicted target logs is minimized
(Hampson et al., 2001). The best attribute (A1) is the one with the lowest
prediction error. The best pair of attributes (A1, A2) assumes that the first
member (A1) is the best attribute. For the best third attribute the method
assumes that the first pair of attributes are the first two members A1 and A2. This
process is repeated as long as desired and is usually guided by cross-validation.
For a nonlinear rela­tionship between the attributes and the selected log
prop­erty we apply NNI. If too many attributes are employed in NNI, overtraining
can result in low training error but high validation error; thus, it is desirable to
minimize the num­ber of attributes used.

In this case, a combination of seismic attributes that include acoustic impedance


inversion are input into multi­attribute transforms in an effort to describe the
spatial distribution of the reservoir
Neural Networks Seismic Multi-Attributes for Lithology & Porosity Mapping
By : Sigit Sukmono 15
3.4. Rock physics trends and cross-plot analysis.

Crossplot of velocity versus porosity in selected wells is done to see how the
well-log data fall within the framework of Han’s reference trends. Figure 16
shows the crossplot of velocity versus density-porosity for Balcon wells 2, 4,
and 6, individually and combined. From this crossplot we can see that there is a
good cor­relation between Han’s data and the Balcon data. The cleaner the
rock, the better match between Han’s model and the Balcon data. Most of the
points that fall outside Han’s model correspond to rocks with porosities between
0 and 5%. These rocks exhibit the highest scattering in the crossplots and are
more difficult to predict from seismic data.

However, two main trends can be distinguished, one for rocks with velocities
higher than about 4.5 km/s and other for rocks with velocities lower than 4.5
km/s. The first group of these rocks corresponds to dirty sandstones and
conglomerates that are partly represented by Han’s database when the clay
content is between 18 and 50%. On the other hand the second group is
normally associated with siltstones or shales and the data points are not
covered by Han’s reference frame­work, which only include sandstones in the
analysis. The high scattering of these rocks can likely be related to the high clay
content, which affects not only the stiffness of the rock but also the pore
geometry and the aspect ratio. The latter factor along with high clay
compressibility are responsible for a considerable reduction of the velocity in the
shales of the Caballos Formation in Balcon Field.

The main reservoir in the field is associated with rocks with porosities that are
higher than 10%. These rocks fall between Han’s zero-clay reference line and
the reference line for clay contents between 12 and 17%. These rocks are
com­posed of clean sandstones that have velocities ranging from
4.0 to 5.0 km/s.

To determine the velocity-porosity and velocity-shale content dependence, we


plot velocity-porosity color-coded by gamma ray for the three Balcon wells 2, 4,
and 6 (Figure 17). For a better illustration of low porosity points (i.e., for shales)
the density-porosity for this example is calculated using limestone as the
density of the matrix (2.71 g/cc), which explains why the porosity looks
overestimated. We can see good correlation between velocity and gamma ray,
and velocity and porosity. Moreover, porosity correlates with gamma ray—the
smaller gamma ray the larger porosity. Green color corresponds to low shale
content while pur­ple color indicates high shale content.

Neural Networks Seismic Multi-Attributes for Lithology & Porosity Mapping


By : Sigit Sukmono 16
The acoustic impedance versus porosity crossplots shows three main trends:
one for high impedance and low porosity, another for low impedance and high
porosity, and the third for low impedance and low porosity (Figure 18).

If we plotted P impedance-porosity with gamma ray for the well Balcon-4, we


found the same relation as for velocity. P-impedance increases with the
decreasing porosity and decreases with the increasing shale content, inferred
from gamma ray (Figure 18). P-impedance could be used to estimate porosity if
shale content is known using linear regression.

As mentioned before, the Caballos Formation is divided into three major units,
which can be subdivided into eight stratigraphic units from top to bottom:
KCUA1, KCUC1, KCUC2, KCUE, KCM, KCLA, KCLB, and KCLC (Figure 19).
In this section we plotted velocity versus density-porosity, velocity versus core
porosity, and gamma ray for Balcon wells 2, 4, and 6 for each stratigraphic unit
of the Caballos Formation.

The main reservoir is associated with the units KCUC1 and KCUC2, separated
from each other by a thin and discontinuous shale layer. In the crossplots, this
unit is yellow color-coded and corresponds approximately with Han’s line of
clay, between 0 and 11% (Figure 19). The impedance of this unit is slightly
lower than the sandstones of the lower Caballos (KCLB&C), but slightly higher
than the shales of the middle Caballos and some fine-grained sandstones and
siltstones from the unit KCUA1 and KCUE (Figure 20).

The prediction of lithology and porosity from impedance derived from seismic
inversion is feasible due to the fact that this attribute is a good discriminator of
lithology plus the fact that there is a good relationship between velocity and
poros­ity for the upper sandstones of the Caballos Formation, particularly for
the units KCUC1 and KCUC2 (Calderon, 2003). Also the good seismic quality in
the central part of the survey and the good well control (seven wells) facilitates
the prediction of rock properties from seismic data.

The final objective is to predict well log properties such as porosity and lithology
from seismic attributes like acoustic impedance, which bear a theoretical
relationship to log properties under some ideal conditions. Although acoustic
impedance provides valuable information of reservoir properties in Balcon Field,
there is a need to use additional information to improve the predictive power
and to overcome limitations of the seismic data like bandwidth, tuning and
interference, noise, and nonuniqueness. In this study, the acoustic impedance is
considered the most important attribute to make quantitative predictions of
reservoir prop­erties in the field, such as porosity and lithology. However, when
this attribute is combined with other attributes the prediction ability from seismic
attributes is improved.

Neural Networks Seismic Multi-Attributes for Lithology & Porosity Mapping


By : Sigit Sukmono 17
3.5. Porosity and lithology prediction.

To determine layer properties rather than interface properties, the seismic


amplitudes are inverted into acoustic impedance. Model-based inversion,
which is a robust inversion method that accounts for the well data and seismic
interpretation, can provide significant information of the spatial distribution of
these properties in terms of changes in acoustic impedance.

Figure 21 shows an arbitrary line that crosses along the wells Balcon-7 and 4
to indicate the seismic character of the Caballos Formation. The gamma-ray
curve is displayed as reference to visualize the difference in the frequency
content and the vertical resolution between well logs and seis­mic data in the
field. For this area, the vertical resolution at reservoir level is about 100 ft.
Although the total thickness of the Caballos Formation is about 600 ft, our
analysis is mainly focused in the upper section of the Caballos Formation
(about 100 ft thick), which contains the main producer sandstones in the field
(units KCUC1 and 2). Two reflectors of good definition and continuity across
the field define the top and the base of the main producer zone. These
horizons correspond to the top of the Caballos Formation and the top of the
unit KCUE (Figure 21). Seismically, the top of these units represent high
contrast in velocity and acoustic impedance that facilitates the lithol­ogy
prediction from seismic data. These units, which can be grouped and
considered a single unit for seismic purposes, are embedded into thick layers
of siltstones and shale of gen­erally low velocity and acoustic impedance.
The lateral amplitude changes of these reflections may be related to porosity
and lithologic changes seen in the wells.

The seismic volume was converted into acoustic impedance using information
of the seven wells and the interpre­tation of horizons associated with the top of
the unit KCUE, the top of the Caballos Formation, and the top of the Tetuan
marker, which is also a good seismic reflector. The good correlation of the
synthetic seismograms with the seismic data in most of the selected wells
allows one to have high confi­dence in the quality of the inverted impedance
volume. The quality of the seismic-well tie is measured by the correlation
coefficient, which is calculated in a window of about 500 ms, including
Caballos and Villeta formations. The best correlation was obtained in the well
Balcon-4 (Figure 12), which is located in an area of good seismic quality. The
wavelet used to invert the volume was obtained from this well. Before the
inversion, the seismic amplitudes were balanced with a 1000 millisecond
operator length. Figure 13 shows the impedance derived from seismic
inver­sion for the arbitrary line.

Neural Networks Seismic Multi-Attributes for Lithology & Porosity Mapping


By : Sigit Sukmono 18
Visual correlation of the gamma-ray curve for the wells Balcon-7 and 4 with the
impedance attribute indicate that the Upper Caballos sandstones (KCUC1 and
2), with porosities generally higher than 8%, normally have impedance values
between 36 000 and 40 000 ft g/s cc. Dirty sandstones of the Lower Caballos,
with porosities normally lower than about 8%, have impedances higher than 40
000 ft g/s cc, and shales and siltstones of the Middle Caballos with porosi­ties
generally lower than 5%, have impedances lower than 36 000 ft g/s cc. This
interpretation is consistent with the crossplots of impedance versus porosity and
impedance versus gamma ray (Figures 19 and 20).

A data slice between the top and the base of the Upper Caballos is also
extracted to determine the spatial distribution of this attribute in the study area
(Figure 24). The impedance map revealed distinct patterns that may be related
to changes in porosity and/or lithology. From this attribute map, we can see that
the impedance varies from 30 000 to 40 000 ft g/s cc. According to the
crossplots, imped­ance greater than 35 000 ft g/s cc is more likely related to
sandstones, while impedance lower than this value is likely related to shales and
siltstones. Porosities in sandstones are normally higher than 8%, while for shales
and siltstones the porosity is less than 5%. By visual correlation between
gamma-ray curve and impedance in Figure 23 we can see that low impedances
(less than 36 000 ft g/s cc) are more likely related to shales, while high
impedances (higher than 36 000 ft g/s cc) are more likely related to sandstones.
By using additional information that comes from other seismic attributes, it is
possible to improve the prediction of well log properties such as porosity and
lithology in Balcon Field.

The relationship between porosity and impedance derived from seismic inversion
for the selected wells in the Caballos Formation is shown in Figure 25. Although
there is high scatter of the data points in the crossplot, we can infer an inverse
relation between poros­ity and impedance given by the slope of the regression
line. Thus, porosity decreases when the impedance increases. The correlation in
this crossplot and the ability of the imped­ance attribute to predict porosity can
be improved if this attribute is combined with other attributes, to account for
lithological variations.

In Figure 26, the validation error curve indicates that only two attributes are
statistically significant. The lower curve (black) shows the prediction error when
all wells are used in the analysis. The upper curve (red) shows the validation
error. These two attributes are, in order, 1/(impedance) and a band pass filter
15/20–25/30 likely related to clay content. According to the validation error curve,
the error increases after the second attribute.

Neural Networks Seismic Multi-Attributes for Lithology & Porosity Mapping


By : Sigit Sukmono 19
These two attributes are the most significant to reduce the scattering observed in
Figure 25. The first attribute 1/impedance is related to porosity and the filter
attribute help to reduce the scattering due to clay content.

We generate volumes of porosity using the most signif­icant attributes and


applying multi-attributes and NNI. Figures 27 and 28 presents the data slices for
a window defined by the two seismic horizons (top and base of the upper
Caballos sandstones), resulting from the weighted sum of the first two attributes
(1/(impedance) and filter 15/20–25/30) using multi-attribute analysis and NNI. In
general, there is similarity between these two porosity maps, which confirms that
the key point is the selection of attrib­utes rather than the procedure used to
predict porosity. The main difference between these two maps is in the area
close to the well Colombina-1 where the porosity derived from neural networks is
more consistent with the well data.

In general there is a good correlation between porosity calculated from seismic


attributes and well logs which are plotted in both maps. However, if the attribute
of time is included as a third attribute in the NNI, meaningful results are obtained.

Figure 29 shows the porosity map obtained with the first two significant attributes
plus time, which rep­resent the influence of the structure and depth in the
reser­voir properties. This attribute needs to be considered because the big
differences in structural elevations of the reservoir in Balcon Field, which in some
cases are about 900 ft (e.g., between Balcon-1 and 4). We can see that most of
the wells located in low structural areas have less porosity than the wells located
in high structural areas. Another way to compare the results of both procedures
is by comparing the original curve and the modeled curve derived from NNI.
Using time as the third attribute, the pre­diction power is improved, especially for
the Upper Caballos. Figure 30 shows an example for the Balcon-4 well.

The gamma-ray curve is a good indicator of lithology in the field. Thus, low
gamma-ray readings correspond to sand­stones and high gamma-ray readings
correspond to shales and siltstones. The objective is to predict gamma ray from
seismic attributes with physical and geologic meaning like acoustic impedance.
However, the relation observed in the crossplot between gamma ray and
impedance derived from inversion is not clear and high scattering is present in
the crossplot, for the selected wells (Figure 31). Therefore, we need to combine
impedance, which is the most significant attribute in the field, with other attributes
using the same methodology that for prediction of porosity. We decided to apply
NNI to predict gamma-ray curve from seismic data because a nonlinear relation
is observed in the crossplot.

Neural Networks Seismic Multi-Attributes for Lithology & Porosity Mapping


By : Sigit Sukmono 20
After doing several tests to select the best match between observed curve and
predicted curve, we found that the best attributes to predict gamma-ray curve are:
(impedance)2, time and filter 15/20–25/30. The result of this inversion dis­played in
the arbitrary line (Figure 32) shows that there is good agreement between gamma ray
derived from seismic attributes via NNI and the actual gamma ray of the selected
wells. Each one of the stratigraphic units of the Caballos is well defined by the
predicted gamma ray. Low gamma-ray values indicate low clay content and are
associated with sandstones. On the contrary high values of gamma ray indi­cate high
clay content and are related to shales or siltstones.

Finally, if we compare the slice at the Upper Caballos sandstones and the amplitude
map at reservoir level with the amplitude map extracted along the horizon of interest,
we find strong similarity between these two maps (Figures 33 and 34). This similarity
allows one to postulate that the seismic amplitude in Balcon Field is strongly affected
by changes in lithology in the Upper Caballos. Therefore, changes in amplitude may
reflect lithologic changes in the Units KCUC1 and 2, for example between Balcon-4
and Colombina-1 wells (Figure 11). High amplitude and low gamma ray are good
indicators of the presence of sandstones in the Upper Caballos, while low amplitudes
and high gamma ray val­ues are good indicators of the presence of shales and
silt­stones in the Upper Caballos.

Neural Networks Seismic Multi-Attributes for Lithology & Porosity Mapping


By : Sigit Sukmono 21
Figure 11. Gamma-ray correlation for wells Balcon-4 and Colombina-1.

Figure 12. Location of Balcon Field.

Neural Networks Seismic Multi-Attributes for Lithology & Porosity Mapping


By : Sigit Sukmono 22
Figure 13. Cretaceous section in the Neiva subbasin.

Figure 14. Units and depositional model for Caballos Formation (modified
from Amaya et al., 1999).

Neural Networks Seismic Multi-Attributes for Lithology & Porosity Mapping


By : Sigit Sukmono 23
Figure15. Seismic cross-section showing the main structural features in
Balcon Field.

Figure 16. Velocity versus density-porosity for Balcon wells 2, 4,


and 6. Large square symbols are for Han’s database.
Neural Networks Seismic Multi-Attributes for Lithology & Porosity Mapping
By : Sigit Sukmono 24
Figure 17. Combined data from well logs from Balcon-2, 4, and 6.
Velocity versus density-porosity, color-coded by gamma ray.

Figure 18. P-impedance versus density-porosity crossplot for Balcon-4.

Neural Networks Seismic Multi-Attributes for Lithology & Porosity Mapping


By : Sigit Sukmono 25
Figure 19. Velocity versus density-porosity and impedance versus den­sity
porosity for Balcon-4.

Figure 20. Velocity versus gamma ray and impedance versus


gamma ray for Balcon-4.

Neural Networks Seismic Multi-Attributes for Lithology & Porosity Mapping


By : Sigit Sukmono 26
Figure 21. Cross-section that contains Balcon-7 and Balcon-4 wells.

Figure 22. Well-seismic tie at Balcon-4.

Neural Networks Seismic Multi-Attributes for Lithology & Porosity Mapping


By : Sigit Sukmono 27
Figure 23. Arbitrary line showing the P-Impedance attribute derived from
model-based inversion

Figure 24. P-impedance slice of the Upper Caballos


Neural Networks Seismic Multi-Attributes for Lithology & Porosity Mapping
By : Sigit Sukmono 28
Figure 25. Crossplot between density-porosity and P-
Impedance seismic inversion using points for the Caballos
Formation

Figure 26. The avaerage error for the best five attributes found by
multilinear regression

Neural Networks Seismic Multi-Attributes for Lithology & Porosity Mapping


By : Sigit Sukmono 29
Figure 27. Data slice of porosity of the upper Caballos generated by
multi-attribute analysis

Figure 28. Data slice of porosity of the upper Caballos generated by NNI
Neural Networks Seismic Multi-Attributes for Lithology & Porosity Mapping
By : Sigit Sukmono 30
Figure 29. Data slice of porosity of the upper Caballos resulting
from the weighted sum of three attributes (1/impedance and
filter 15/20-25/30 and time) using NNI

Figure 30. Comparison of validation results for Balcon-4 using NNI


(a) Two attributes and (b) adding time as a third attribute

Neural Networks Seismic Multi-Attributes for Lithology & Porosity Mapping


By : Sigit Sukmono 31
Figure 21. Crossplot between gamma ray and P-impedance
seismic inversion for the Caballos Formation

Figure 22. Gamma ray from NNI using (impedance)2, time and filter 15/20-25/30

Neural Networks Seismic Multi-Attributes for Lithology & Porosity Mapping


By : Sigit Sukmono 32
Figure 23. Data slice of gamma ray of the Upper Caballos sandstones

Figure 24. Extracted amplitude map, centered at the horizon


of the base of the Upper Caballos sandstones

Neural Networks Seismic Multi-Attributes for Lithology & Porosity Mapping


By : Sigit Sukmono 33
References

This chapter is compiled from the following references :

1. Calderon, J.E. and Castagna, J., 2006, Porosity and lithologic estimation using
rock physics and multi-attribute transforms in Balcon Field, Colombia, The
Leading Edge, 142-150.

2. Sukmono, S., 2007, Application of multi-attribute analysis in mapping lithology


and porosity in the Pematang-Sihapas groups of Central Sumatra Basin,
Indonesia, The Leading Edge, 126-131.

Neural Networks Seismic Multi-Attributes for Lithology & Porosity Mapping


By : Sigit Sukmono 34
Neural Networks Seismic Multi-Attributes for Lithology & Porosity Mapping
By : Sigit Sukmono 35
Neural Networks Seismic Multi-Attributes for Lithology & Porosity Mapping
By : Sigit Sukmono 36
Neural Networks Seismic Multi-Attributes for Lithology & Porosity Mapping
By : Sigit Sukmono 37
Neural Networks Seismic Multi-Attributes for Lithology & Porosity Mapping
By : Sigit Sukmono 38
Neural Networks Seismic Multi-Attributes for Lithology & Porosity Mapping
By : Sigit Sukmono 39

You might also like