Вы находитесь на странице: 1из 84

6/23/2003

AVO
• What is AVO ?
• Why does AVO happen?
• Can we predict an AVO effect?

Schlumberger Private
• What parameters effect AVO?
• Can AVO effects be measured?
• What can be displayed?
• What do the results mean?
• A summary of AVO

Schlumberger Private
• And some other bits
©WesternGeco

So loosely speaking, "AVO" is the data processing term by which we mean to


analyse the amplitudes in the seismic data in order to formulate some lithological
model.
We will try to answer these questions in our attempt to understand AVO

1
6/23/2003

Definition

The variation in the amplitude of a seismic reflection with


source-geophone distance. Depends on the velocity, density

Schlumberger Private
and Poisson ratio contrast. Used as a hydrocarbon
indicator for gas because a large change in Poisson’s ratio
(as may occur when the pore fluid is a gas) tends to produce
an increase in amplitude with offset

Schlumberger Private
©WesternGeco

From Sheriff's dictionary


For 'gas', you can read 'oil or gas'

It is now understood that the last part, concerning increasing amplitude with offset,
is not strictly true, some hydrocarbon traps can have deceasing amplitude with
offset

2
6/23/2003

A CMP Schematic

Schlumberger Private

Schlumberger Private
©WesternGeco

Conventional seismic acquisition normally implies grouping traces into CMP


gathers. Such gathers comprise many different source-receiver pairs.

3
6/23/2003

A CMP Seismic Event after NMO

Schlumberger Private
Amplitude changing
with offset

Schlumberger Private
©WesternGeco

This range of different source-receiver distances (offsets) enables examination of


the variation in the strength of a seismic reflection at different offsets.
(click) - This signal Amplitude Variation with Offset is what is meant by AVO
Amplitude can increase or decrease

4
6/23/2003

AVO
• What is AVO ?
• Why does AVO happen?
• Can we predict an AVO effect?

Schlumberger Private
• What parameters effect AVO?
• Can AVO effects be measured?
• What can be displayed?
• What do the results mean?
• A summary of AVO

Schlumberger Private
• And some other bits
©WesternGeco

The presentation follows the lines of answering questions

5
6/23/2003

Changes in Physical Properties

• What are the physical processes involved in the


reflection of a seismic wave from a boundary

Schlumberger Private
• Any reflection implies a change in either velocity
(compression/shear) and/or density
– i.e. a change in ACOUSTIC IMPEDANCE

Schlumberger Private
©WesternGeco

Let's examine the physical processes involved in the reflection of a seismic wave
from the surface separating two different rock layers.
Any reflection implies that there is a change in either velocity and/or density. And
there are the 2 kinds of velocity, compression (P) and shear (S).
Hence there is an acoustic impedance for p-waves AND an acoustic impedance for
s-waves!

6
6/23/2003

What is a reflection ?

• Seismic waves are reflected at a


Layer 1
boundary between different rocks

Schlumberger Private
• At normal incidence, incident reflected
the reflected wave 1 R
has an amplitude R : Z2 − Z1
R=
Z2 + Z1 Layer 2

transmitted
Zi = ρi × Vi T

= ( density) × ( wavespeed)

Schlumberger Private
• R = 0.1 is a BIG reflection !
©WesternGeco
pafc 02/02/97

When a P-wave is (normally) incident onto a boundary, a reflected and transmitted P-


wave is formed
The amount of energy that is reflected and the amount that is transmitted depends on
the CHANGE in rock properties across the boundary
No shear waves involved

Acoustic impedance Z is velocity * density

Reflection coefficient is (Z2 - Z1) / (Z2 + Z1)

7
6/23/2003

And now with an angle

NORMAL

Schlumberger Private
Down-going P-wave SHEAR Boundary

Schlumberger Private
©WesternGeco

We can think of the incident wave as having 2 components, normal and shear : the
normal producing p-waves and the shear, shear waves

8
6/23/2003

What happens at boundaries

Schlumberger Private
sin(θ ) sin(δ ) sin(θ ) sin(δ )
= = =
1 1 2 2

Schlumberger Private
VP 1 VS 1VP VS 2 2

©WesternGeco

However, when the P-wave strikes at an angle, something slightly different


happens. At every boundary, an incident 'P-wave' splits into a reflected P-wave,
reflected S-wave,transmitted P-wave and transmitted S-wave.

The ratio of angles to the relevant velocities obey Snell's law

Each of these waves will continue to reflect and refract at other interfaces, with the
P-waves generating more and more S-waves AND the S-waves reconverting back to
P-waves!

As P- & S-waves travel at different velocities, it's really quite surprising that seismic
sections contain any recognisable information at all!

Also notice that after the critical angle, there are no more transmitted/reflected P-
waves - these have been replace by a refraction BUT shear waves continue as
normal!

9
6/23/2003

How Much? – is this slide correct ?

• Rock properties determine amount, if any, of S-wave


energy produced

Schlumberger Private
– No S-waves : amplitude a function of 1/cos2θ –
amplitude increases with angle/offset
– S-waves being produced :
• normally : less amplitude increasing with angle/offset
• special conditions : amplitude DEcreases with angle/offset

Schlumberger Private
©WesternGeco

In the unlikely event of no shear waves being produced, there would still be an
AVO effect of amplitude increasing with offset
With shear waves being produced, the energy going into them is taken from the p-
wave energy thereby beginning to nullify the existing INCREASE in amplitude.
The boundary conditions define how the energy is split between P & S, reflected &
transmitted (this either means less reflected energy - dim spot, or less transmitted
energy - bright spot)

10
6/23/2003

AVA <- -> AVO

Schlumberger Private

Schlumberger Private
©WesternGeco

We have been discussing amplitude changing with angle, AVA, yet the talk is AVO
- what is the relationship.
A constant angle trace actually samples different offsets with change of time
Conversely, a constant offset trace samples different angles with change of time

11
6/23/2003

AVA <- -> AVO (contd)

• Change in Incident Angle -> Change in Amplitude


• Different Offsets, Same Time -> Different Incident
Angles -> Different Amplitudes

Schlumberger Private
• Same Offset, Different Times -> Different Incident
Angles -> Different Amplitudes

• Amplitude changes with Offset AND Time


• Temporal & Spatial change in velocity complicates the
proceedings!

Schlumberger Private
©WesternGeco

This energy loss at boundaries, due to conversion, will be dependant on the incident
angle of the P-wave, but there is a tentative relationship between incident angle and
offset

What is not covered here is how velocity, and how it is used ultimately determines
the angle of incidence, and hence any perceived AVO effect

12
6/23/2003

P-waves And S-waves

GECO LONGVA
OSLO

Water
Vp = 1500m/s reflected

Schlumberger Private
Vs = 0m/s P wave
incident
P wave

θ θ

Hard Sea-bed
Vp = 2500m/s
Vs = 1200m/s
transmitted
P wave
transmitted
S wave

Schlumberger Private
©WesternGeco

A reiteration that no shear waves are produced in water; hence at the water bottom no
shear waves are reflected back to the cable AND the transmitted shear wave will
eventually reflect from a deeper boundary with some of the energy as shear - when that
hits the water bottom from below, only p-wave energy gets transmitted into the water
and back up to the cable

13
6/23/2003

AVO
• What is AVO ?
• Why does AVO happen?
• Can we predict an AVO effect?

Schlumberger Private
• What parameters effect AVO?
• Can AVO effects be measured?
• What can be displayed?
• What do the results mean?
• A summary of AVO

Schlumberger Private
• And some other bits
©WesternGeco

The presentation follows the lines of answering questions

14
6/23/2003

AVO Prediction

• Given P-wave velocity, S-wave velocity and density of 2


bounding layers, we can calculate the amplitudes of the
different reflected and transmitted waves

Schlumberger Private

Schlumberger Private
©WesternGeco

15
6/23/2003

Elastic Waves at a Plane Interface


Continuous Discontinuous
Ray Paths Properties Properties
Vertical Horizontal
Displacements Stresses Stresses

Schlumberger Private
Pi Sr Pr Z1

σ zz
σ xz

1 X1 Torque

2 X2
σ x
σ xx

z
St Pt Z2

Snell’s Law X1 = X2 σ xz1 = σ xz2 σ xx1 ≠ σ xx2


Obeyed Y1 = Y2 σ =σ σ ≠σ

Schlumberger Private
yz1 yz2 yy1 yy2

Z1 = Z2 σ zz1 =σ zz2 σ xy1 ≠σ xy2

©WesternGeco

We must obey certain “guidelines”. The first is obeying Snell’s Law. The second is that the two layers cannot slide by each other (continuity of horizontal displacement)
and they cannot bounce up and down causing vertical separation (continuity of vertical displacement). Other properties shown are enough to define a series of equations
(Zoeppritz) that define the complex reflection coefficients of the media. The complex reflection coefficients define the amplitude and phase of reflected and refracted
rays.

References:
Aki and Richards, Quantitative Seismology, Theory and Methods. Volume 1.
Mallick, ,1993, A simple approximation to the P-wave reflection coefficient and its implication in the inversion of amplitude variations with offset: Geophysics 58, 4,
p544-552.
Shuey, 1985, A simplification of the Zoeppritz equations: Geophysics 54, 4, p609-614.
Fatti, et al., 1994, Detection of gas in sandstone reservoirs using AVO analysis: A 3-D seismic case history using the Geostack technique: Geophysics 59, 4, p1362-
1376.

16
6/23/2003

Zoeppritz Equations !

Schlumberger Private

Schlumberger Private
©WesternGeco

Which lead to this matrix.


We see all the coefficients are in terms of the p-wave/shear velocity/density values

17
6/23/2003

P-P and P-S Reflection Coefficients

Sv P
P

Schlumberger Private
θ 1 η1
α1 , β1 , ρ1

α 2 , β 2 , ρ2
η2 θ2
P
α = P-
P-wave velocity Sv

Schlumberger Private
β = S-
S-wave velocity
©WesternGeco ρ = bulk density

Let's revisit this picture and use alpha,beta and rho for the various parameters.
If we have an incident P wave (in from upper left) that impacts a surface (horizontal
blue line) across which there is a change in physical properties four different wave
types may be generated. They all obey Snell’s Law in their respective layers since
they all have the same ray parameter, p, defined by their respective sin/V ratio.
Two reflected waves (P and S) are formed and the S-wave is polarized so that
particle motion is in the plane of the figure (By definition, the is S vertical). We
also get two refracted rays (P and S).

18
6/23/2003

Another view of Zoeppritz Equations


Exact Formula (Aki & Richards, 1980; Mallick, 1993):
A + Bp + Cp − Dp 6 2 4
R PP = ,
E + Fp 2 + Gp 4 + Dp 6
sin θ1 sin η1 sin θ2 sin η2
p = = = = .

Schlumberger Private
α1 β1 α2 β2
 cos θ1 cos θ2   cos η1 cos η2 
A =  ρ2 − ρ1   ρ2 + ρ1 ,
 α1 α2  β1 β2 
 cos θ1 cos η1 cos θ2 cos η2  θ θ cos η1 cos η2
B = − 4 ∆ µ  ρ2 + ρ1  − ( ∆ ρ ) 2 + 4 ( ∆ µ ) 2 cos 1 cos 2 ,
 α1 β1 α2 β2  α1 α2 β1 β2
 cos θ1 cos η1 cos θ2 cos η2 
C = 4 ( ∆ µ ) 2  −  + 4 ∆ µ ∆ ρ,
 α1 β1 α2 β2 
D = 4( ∆ µ )2 ,
 cos θ1 cos θ2   cos η1 cos η2 
E =  ρ2 + ρ1   ρ2 + ρ1 ,
 α1 α2  β1 β2 
 cos θ1 cos η1 cos θ2 cos η2  2 cos θ1 cos θ2 cos η1 cos η2

Schlumberger Private

F = − 4 ∆ µ  ρ2 − ρ1  + ( ∆ ρ ) +4( ∆ µ )
2
,
 α1 β1 α2 β2  α1 α2 β1 β2

2  cos θ1 cos η1 cos θ2 cos η2 


G = 4 ( ∆ µ )  +  − 4 ∆ µ ∆ ρ ,
 α1 β1 α2 β2 
∆µ = µ2 − µ1 = ρ2 β22 − ρ1 β12 ; ∆ ρ = ρ2 − ρ1 .
©WesternGeco

The p-wave reflection coefficient of a boundary. This is nice, but to derive


something more accessible, and define seismically-derived attributes, we need to
“linearize” the equations. This is done by restating the problem so that we deal with
small changes in physical parameters across an interface.

19
6/23/2003

Zoeppritz

Schlumberger Private

Schlumberger Private
©WesternGeco

P-wave reflection amplitudes are shown as calculated by the Zoeppritz equations


THIS AVO effect is a direct hydrocarbon indicator (DHI)

20
6/23/2003

Approximations to Zoeppritz

Schlumberger Private
R(angle ) = R + G * (sin( angle ) )2 Beware : Angle < 30

Schlumberger Private
0

©WesternGeco

Shuey & others made approximations that lead to being able to plot amplitude
versus the square of the sine of the angle of incidence. Up to 30 degrees, a straight
line is formed.
R0 can be thought of as the 'intercept' reflection coefficient and G is the 'gradient'
This slope and intercept depend on the physical properties of the rocks surrounding
the reflecting interface.
Beware clients asking to go out further than 30 degrees and still doing this analysis -
it may be stretching the truth too much!

But how did we get to this equation and what do the terms really mean?

21
6/23/2003

How did we simplify?

Sv P
P

Schlumberger Private
θ 1 η1
α1 , β1 , ρ1

α 2 , β 2 , ρ2
η2 θ2
P
Sv

Schlumberger Private
θ = 1 ( θ 1 + θ 2 ), η = 1 (η 1 + η 2 )
©WesternGeco
2 2

Angles are averaged either side of the interface. (We assume that small differences
in the property changes makes this valid).

We will see that it is THIS theta that turns as being the angle plotted in previous
display

22
6/23/2003

And ….
• Use average properties and changes in properties
• Assume changes in properties are small compared to
average properties.

Schlumberger Private
α = 1 ( α1 + α2 ), β = 1 ( β1 + β2 ),
2 2
ρ = 1 ( ρ1 + ρ2 )
2

∆α = ( α 2 − α 1), ∆β = ( β 2 − β 1),
∆ρ = ( ρ2 − ρ 1), ∆µ =(µ 2 − µ 1 )

Schlumberger Private
©WesternGeco where µ is shear rigidity

The first step, following Shuey;and Aki and Richards, is to introduce average
properties and differences across layers. Here, (Greek) mu is rigidity that
determines shear-wave velocity.

Using these averages and differences in the ensuing calculations makes it very
difficult to associate seismic amplitudes with absolute values of rock properties
within the AVO process

23
6/23/2003

Approximation (P-P Reflection)

Bortfeld (1961):
β 2  ∆ρ 2∆β  2 1 ∆α
RPP ≈ R0 − 2 
 + 
 sin θ + tan2 θ
α  ρ
2
β  2 α
Shuey (1985):

Schlumberger Private
1 ∆α  2(1 −2σ ) 1 −2σ ∆σ  2 1 ∆α
Rpp ≈ R0 +  1 −  −2R0 + sin θ + tan 2θ
2 
2 α  1 −σ  1 −σ (1 −σ )  2 α
Mallick (1993):
∆µ 1 ∆α
Rpp ≈ R 0 − 2 sin 2
θ + tan2 θ
ρα 2
2 α

Schlumberger Private
σ, ∆σ: Average Poisson’s ratio and Poisson’s ratio contrast.
∆µ:
∆µ: Shear modulus (rigidity) contrast.
©WesternGeco

Based on Aki and Richards simplifications, everyone has their favorite version that
focuses on their property of interest. All these equations are equal. However, they
are not completely linearized (having the classic Intercept-Gradient form).

24
6/23/2003

Approximation (P-P Reflection)

General Form:
R PP ≈ R0 + G sin 2 θ + C tan 2 θ

Schlumberger Private
where
ρ2 α 2 − ρ1 α1 1  ∆α ∆ρ 
R0 = ≈  + 
ρ2 α 2 + ρ1 α1 2  α ρ 

is the normal incidence reflectivity


The approximation on the right is made to facilitate

Schlumberger Private
the formation of other attributes
©WesternGeco

Note the Ro is an approximation to the vertical-incidence reflection coefficient due


to an acoustic or fluid-fluid interface responding to a P-wave. This is the “3-term”
equation (having sine and tangent too). It is possible to derive AVO attributes,
similar to those described later, using these expanded equations. The good and bad
is that it does go out to higher angles, and is more sensitive to the higher-angle
amplitudes. Smith and Gidlow (1987) describe using these equations.

25
6/23/2003

AVO Attribute Inversion

from Mallick (1993),


∆µ 1 ∆α 2 θ

Schlumberger Private
Rpp ≈ R0 − 2 2 sin2θ + tan
ρα 2 α
The 30 degree
tan θ ≈ sin θ
2 2
by using
approximation !
this yields
 1 ∆α ∆µ 

Schlumberger Private
Rpp ≈ R0 +  − 2 2 sin2 θ
©WesternGeco 2 α ρα 

Arbitrarily taking Mallick’s equation (it makes no difference), we see that his is in
the “3-term” form. So, how do we squeeze out AVO attributes and rock properties?
First, simplify even more! Tangent is about equal to sine for small angles (up to 30
degrees). Combine the two terms to get the lower equation.

26
6/23/2003

Approximation (P-P Reflection)

The tangent equal sine approximation is made to


create the standard Intercept-Gradient Form:

Schlumberger Private
RPP ≈ R0 + G sin 2 θ

where G now includes the C term.

Schlumberger Private
©WesternGeco

This has typically been the linearized form used for deriving simplified attributes
from the seismic data. Recall that Ro is the approximation to vertical-incidence
reflection coefficient due to an acoustic or fluid-fluid interface for a P-wave.

27
6/23/2003

Approximations
Zoeppritz
Amplitude versus Angle Bortfeld
Bort (approx)
0.07

0.06

0.05

Schlumberger Private
0.04
Amplitude

0.03

0.02

0.01

0.00
0 10 20 30 40 50 60
-0.01

-0.02 tangent=sine
Degrees

Schlumberger Private
α1= 3000 β1= 1414 ρ1=2.40 α2= 3100 β2= 1500 ρ2=2.45
©WesternGeco

Here we analyze the accuracy of the Bortfeld, or “3-term” solution, compared to the
full Zoeppritz. Not bad! However, the tangent equal sine approximation sharply
limits the useable angle range. However, these are the angles we commonly find
available in most seismic data.

28
6/23/2003

Gulf Coast Sand - Zoeppritz/Shuey

Schlumberger Private

Schlumberger Private
©WesternGeco

For this rock type, we see an AVO effect and the Shuey approximation mirrors
Zoeppritz

29
6/23/2003

Dim Spot in Carbonate Rock

Schlumberger Private

Schlumberger Private
©WesternGeco

Here we find a difference so the Shuey approximation does not always work!

30
6/23/2003

An Example - CMP after NMO

Schlumberger Private
STACK Result

P-wave vel = 2440, S-wave vel = 1410

Schlumberger Private
P-wave vel = 3050, S-wave vel = 1960

©WesternGeco

Complete polarity reversal from near to far offset can happen under certain
circumstances! Here it is such that the CMP stack is ZERO !!

A change in P-wave/S-wave velocity as shown can cause something like this!

31
6/23/2003

AVO
• What is AVO ?
• Why does AVO happen?
• Can we predict an AVO effect?

Schlumberger Private
• What parameters effect AVO?
• Can AVO effects be measured?
• What can be displayed?
• What do the results mean?
• A summary of AVO

Schlumberger Private
• And some other bits
©WesternGeco

The presentation follows the lines of answering questions

32
6/23/2003

Changes in Physical Properties

Physical properties :
e.g. p-wave velocity

Schlumberger Private
s-wave velocity
density

Different Physical properties :


e.g. p-wave velocity
s-wave velocity

Schlumberger Private
density
©WesternGeco

Let's examine the physical processes involved in the reflection of a seismic wave
from the surface separating two different rock layers.
Any reflection implies that there is a change in either velocity and/or density. And
there are 2 kinds of velocity, compressional (p) and shear (s).
How do we measure these values?

33
6/23/2003

AVO Parameters

• P-Wave Velocity - Measured :


– Well logs

Schlumberger Private
– Seismic Velocity Analysis

• Density - Measured :
– Well logs
– Empirically from P-Wave Velocity

• S-Wave Velocity - Difficult to Measure


– P-Wave velocity/S-Wave velocity Ratio

Schlumberger Private
– Poisson's Ratio (Rock Property)

©WesternGeco

The main factors affecting AVO are changes in P-Wave velocity, S-Wave velocity
and/or density across a boundary.
The P-wave velocity can be measured from logs made within a well or failing that
from seismic velocity analysis (although the latter is rather coarse)
Density can also be measured within wells or failing that be empirically derived
from the P-wave velocity
S-Wave velocity is difficult (but not impossible) to measure within wells. Typically
it can be calculated from either knowing the Vp/Vs ratio or Poisson's Ratio of the
material.

34
6/23/2003

AVO
• What is AVO ?
• Why does AVO happen?
• Can we predict an AVO effect?

Schlumberger Private
• What parameters effect AVO?
• Can AVO effects be measured?
• What can be displayed?
• What do the results mean?
• A summary of AVO

Schlumberger Private
• And some other bits
©WesternGeco

The presentation follows the lines of answering questions

35
6/23/2003

Can we Measure AVO Effects?

A qualified YES !

Schlumberger Private

Schlumberger Private
©WesternGeco

36
6/23/2003

Zero Offset Reflectivity & Gradient

Schlumberger Private

Schlumberger Private
©WesternGeco

As from Shuey, plotting amplitude versus sine squared theta for a particular event
allows us to predict the amplitude at normal incidence (theta = zero) for that event
plus the gradient
Normal incidence only gives off P-waves so this intercept amplitude can be thought
of as being the P-wave only coefficient. Thus the section is sometimes called the P-
wave section.
It also turns out that normal incidence only occurs at zero offset, so the section can
also be called the (predicted) zero offset section
Only true up to 30 degrees !!

37
6/23/2003

Processing Pitfalls

• Sources
– depth / strength / timing / directivity
• Receivers

Schlumberger Private
– depth / sensitivity / directivity
• Noise
– random / coherent / multiples / scatter
• Timing
– tidal / NMO /inaccurate NMO/non-hyperbolic NMO /

Schlumberger Private
statics
©WesternGeco

Given we are going to analyse AMPLITUDES, sometimes on individual traces and


sometimes on partially stacked traces then our processing must result in :
NOISE-FREE data
and
AMPLITUDE PRESERVED data
and
PERFECTLY-CORRECTED data :

What affects this ? : all these ! and probably more!

38
6/23/2003

Processing Pitfalls (contd)

• Positioning
– navigational / geophysical

Schlumberger Private
– should data be migrated ?
• Amplitude Effects
– shallow anomalies / deep anomalies / spreading /
Q

Schlumberger Private
©WesternGeco

39
6/23/2003

AVO
• What is AVO ?
• Why does AVO happen?
• Can we predict an AVO effect?

Schlumberger Private
• What parameters effect AVO?
• Can AVO effects be measured?
• What can be displayed?
• What do the results mean?
• A summary of AVO

Schlumberger Private
• And some other bits
©WesternGeco

The presentation follows the lines of answering questions

40
6/23/2003

Display Options

• Any of the results of the AVO calculations


– Intercept (Zero Offset) & Gradient & Others

Schlumberger Private
• Crossplots
• ‘Near’, ‘Far’ & ‘Difference’ Stacks

Schlumberger Private
©WesternGeco

41
6/23/2003

Constant Angle

Schlumberger Private

Schlumberger Private
©WesternGeco

Generally to get the same angle with depth/time, you have to go out to longer
offsets

42
6/23/2003

AVO/AVA Analysis
What can be displayed

Schlumberger Private

Schlumberger Private
©WesternGeco

First, how do we derive the intercept and slope? Either by transforming the data
from AVO (amplitude versus offset) to AVA (amplitude versus angle) or simply ray
tracing and bookkeeping the angles, we analyze the moveout-corrected seismic data
at each sample. Effectively, ray-tracing is applied in both cases and needs to be
fairly sophisticated. Analysis QCs can include bar-graph plots of amplitudes
windowed around particular events.

43
6/23/2003

AVO Intercept and Gradient

Angle
R(θ) ≈ P + G sin2
R(θ
θ
Observed

Reflection Amplitude (R)

Schlumberger Private
Linear Fit

slope = G
P
intercept

sin2 θ
Intercept Gradient

Schlumberger Private
trace trace
©WesternGeco

Typically, each time sample is fit to an amplitude versus sin2 line to derive a
continuous intercept and gradient as seen here.
One can imagine how ACCURATE the NMO/MUTE/AMPLITUDE
PROCESSING must be for this to mean anything!

44
6/23/2003

An Example: Stacked Section

Schlumberger Private

Schlumberger Private
©WesternGeco
Bright Spot

Gulf of Mexico example.

45
6/23/2003

Example Angle Gathers


B A

Schlumberger Private

Schlumberger Private
©WesternGeco

CMP gathers from positions B & A respectively.


Both the height & thickness of the bar graphs at the bottom are proportional to the
relative RMS amplitude within window 1.2 to 1.3s.
The AVO response need not be caused solely by varying reflectivity with angle -
transmission effects, thin bed tuning, interference - can all play a part. The point is
that in the neighbourhood, it has ALREADY BEEN SHOWN that the increase in
amplitude with angle within the bright spot is consistent with the presence of gas at
this level.
At location B, amplitude variations are small

46
6/23/2003

Conventional & Intercept Stacks


AVO Intercept Stack : Zero Offset ?
B A

Schlumberger Private
Conventional Stack

Schlumberger Private
©WesternGeco

Here we see the conventional CMP-stacked section and the Intercept (zero offset) section. We see many differences showing that a simple summation of amplitudes as is
done in conventional stacking may not give us a 'true' zero offset section which is required for migration and any inversion.
However, the intercept is only as good as the prestack amplitudes - noisy samples will harmfully affect it, hence the requirement for good pre-processing

47
6/23/2003

Gradient Displays
Gradient superimposed on Intercept
B A

Schlumberger Private
Gradient *Intercept

Schlumberger Private
©WesternGeco

The gradient is overlaid on top of the P-wave stack. Blue is decrease in amplitude with offset (no interest in this example) whereas red is Increase in amplitude with
offset (possible interest). For these data we are more interested if any marked increase in AVO effect also corresponds to high reflectivity
So possibly a better display is to MULTIPLY the intercept by the gradient.
Depending on the lithology of the area, other displays, or combinations may highlight the characteristic required to find oil/gas

48
6/23/2003

AVO Attribute

 1 ∆α ∆µ 
Rpp ≈ R0 +  − 2 2 sin2 θ = P + G sin2 θ
2 α ρα 

Schlumberger Private
where we define:
1  ∆α ∆ρ 
P = R0 ≈  +  (AVO Intercept)
2  α ρ 
and
1 ∆α ∆µ

Schlumberger Private
G= −2 (AVO Gradient)
2 α
©WesternGeco
ρα2

Now, we can see that a straight-line fit of seismic amplitude versus sin2 of reflection
angle (determined via raytracing or other equations) gives us an intercept (θ = 0)
that correspond to the normal-incidence P-wave reflection coefficient! This is the
AVO Intercept or P-wave Reflectivity Section. The slope is G, but no big insights
there yet.

49
6/23/2003

More Attributes from G

µ = ρβ 2

Therefore using the total derivative,

Schlumberger Private
Big IF !
∆ µ = β ∆ ρ + 2 β ρ∆ β
2

Inserting this in G, and assuming α/β=2 gives

1  ∆α ∆ρ  ∆β
G =  −  −
 α ρ  β

Schlumberger Private
2
©WesternGeco

Intercept and gradient are interesting, but we need more insight. Now let’s pick
apart G. We do a bit of math to get the lower equation. Now we seem individual
terms for Vp (or α) , Vs (or β), and density! Smith and Gidlow follow other routes
where, instead of α/β = 2, they get rid of the density term by using Gardener’s
equation, ρ = K α 1/4. This yields ∆ρ/ρ = 1/4(∆α/α). Regardless, both of these
approaches linearize the gradient term for later recombination. The accuracy of
each method is another issue.

50
6/23/2003

S-wave Section

1  ∆α ∆ρ 
=
P R0 ≈  + 
2  α ρ 

Schlumberger Private
1  ∆α ∆ρ  ∆β
G = 
 −  −
2  α ρ  β

1 1  ∆β ∆ ρ  Pseudo S-wave

(P G ) = 
 β + 
ρ  Section

Schlumberger Private
2 2 
©WesternGeco

Combine the Intercept and Gradient terms and…

We get the Pseudo S-wave section. This may be used as input to inversion to S-
wave impedance.

51
6/23/2003

Poisson’s Ratio Section

1  ∆α ∆ρ 
=
P R0 ≈  + 
2  α ρ 

Schlumberger Private
1  ∆α ∆ρ  ∆β
G = 
 −  −
2  α ρ  β

4  Pseudo
( P +G ) = ∆ α − ∆ β  Poisson’s Ratio
3 α β

Schlumberger Private
 Contrast
©WesternGeco
Section

Combine the Intercept and Gradient terms differently and…

Poisson’s ratio contrast. Poisson’s ratio, σ, is:

σ = [(β/α)2 − .5]/[(β/α)2 − 1]

However, the Pseudo Poisson’s Ratio contrast is actually based on an the total
derivative of α/β. For small differences, the contrast in Poisson’s Ratio (∆σ/σ) is
close to the contrast in α/β (∆(α/β)/(α/β) derived from the average properties.

52
6/23/2003

Shear & Poisson’s Ratio Displays


Pseudo S-
S-wave Stack
B A

Schlumberger Private
Poisson Ratio Stack

Schlumberger Private
©WesternGeco

Unfortunately there is nothing in the seismic for us to 'measure' in order to determine some property of the S-wave; so normally something is 'assumed' in order to then
display the S-wave stack & Poisson's ratio stack. (and also other approximations are made!).
Here a Vp to Vs ratio of 2 has been assumed, but the timing of the S-wave stack is at P-wave velocity! It is calculated via subtracting the gradient from the zero offset (P -
G).
The Poisson’s ratio stack is rather a 'change in Poisson’s ratio stack and is obtained via addition of zero offset and gradient (P + G).
So, from the pre-stack AVO attributes (P & G) and some assumptions, we have measurements of Poisson’s ratio - this plus velocity and geologic knowledge may lead us
to oil/gas!

53
6/23/2003

All that glitters …...

Schlumberger Private

Schlumberger Private
©WesternGeco

At 1.25, we have the 'bright spot' on the p-wave (decrease in impedance) and
Poisson's ratio, but not on the s-wave - this combination of decrease in p-wave
impedance and little change in s-wave impedance has been associated with gas in
various other studies.
BUT Poisson's ratio has the same type of event at 1.3, but this time the p-wave and
s-wave show the same amount of increase in impedance - such behaviour, in this
area, is NOT suggestive of gas, and a wellog in fact confirms this.

54
6/23/2003

AVO ?

“AVO” should be expected - it is the ANOMALOUS


AVO that is of interest !

Schlumberger Private

Schlumberger Private
©WesternGeco

55
6/23/2003

AVO Cross Plot

Schlumberger Private
Gradient

Intercept

Schlumberger Private
©WesternGeco
Gradient
Intercept

Another plot gaining in popularity is the 'AVO Cross Plot'


The crossplot is formed from all the data points in the survey area associated with a
particular target. A background, linear trend (passing around and about 0,0) is to be
expected in many formations - outliers from this trend show the areas of interest that
deserve further study.
Intercept vs gradient crossplot constructed using SigmaView (by Richard Morton in
PH). The highlighted points (red) on the crossplot indicate points that are
anomalous due to fluid effects. On the section displays on the right these points can
be related to a shallow gas reservoir.

56
6/23/2003

AVO Stack

Schlumberger Private

Schlumberger Private
©WesternGeco

Another example
Please note that the plot is reversed to others that follow!

57
6/23/2003

Near Offset Stack

Schlumberger Private

Schlumberger Private
©WesternGeco

As said before, the intercept & gradient methodology relies on good S/N prestack
data; but we know that generally we rely on stack multiplicity to help us with this
enhancement. So in these situations, what can we do?
A common practice is still to improve the pre-stack data as much as possible, but
then recognising that the data is still too noisy for true AVO analysis, simply stack
'near' and 'far' offsets separately, where 'near' is defined by some angle that
generally bisects the angle determining the far offset mute.
Here we see the 'near' offsets from the previous section

58
6/23/2003

Far Offset Stack

Schlumberger Private

Schlumberger Private
©WesternGeco

And now the 'fars'

59
6/23/2003

Difference Stack

Schlumberger Private

Schlumberger Private
©WesternGeco

And the difference. One would hope that in a 'bland' area this difference would be
zero showing no AVO effects
We see some marked differences at 1.7 on the left hand side indicating an area
worthy of further investigation

60
6/23/2003

Enhanced Residual Gradient (ERG)


• Simple and quick AVO analysis product
• ERG=(F-N)*F (F-N)*N
• N=Amplitude envelope of near offset stack

Schlumberger Private
• F=Amplitude envelope of far offset stack
• Difference values are Enhanced by Far multiplication

Schlumberger Private
©WesternGeco

But only if amplitude increases with offset, if other way round then need (F-N)*N !!

61
6/23/2003

Poisson ratio vs ERG


AVO Attribute Sensitivity

1.5

1.4

1.3

Schlumberger Private
1.2
Normalised attribute

1.1

0.9

0.8

0.7

0.6

Schlumberger Private
0.5
2750 2850 2950 3050 3150 3250 3350
Time

poisson Erg
©WesternGeco

62
63
6/23/2003

Schlumberger Private
Schlumberger Private
ERG

©WesternGeco
6/23/2003

Foinaven Data Progression : Mid-Offset Stacks

W E

1993

Schlumberger Private
93 reprocessed 95

Schlumberger Private
©WesternGeco
GOC 95 towed

64
6/23/2003

AVO
• What is AVO ?
• Why does AVO happen?
• Can we predict an AVO effect?

Schlumberger Private
• What parameters effect AVO?
• Can AVO effects be measured?
• What can be displayed?
• What do the results mean?
• A summary of AVO

Schlumberger Private
• And some other bits
©WesternGeco

The presentation follows the lines of answering questions

65
6/23/2003

????

• Needs INTERPRETATION!
– Seismic by itself is difficult to interpret

Schlumberger Private
– Rock property change controls response, not
absolute values
• Needs CALIBRATION from well logs
– Model expected AVO

Schlumberger Private
©WesternGeco

Just creating the 'AVO sections' is not an end in itself, they have to be interpreted
and they have to be interpreted with the understanding that it is the CONTRAST of
the rock properties that causes AVO effects rather than the absolute values
So we need some 'ground truth' to help determine what these seismic amplitudes
really mean - well logs
One option is to model the AVO response using information derived from the wells
and compare it to the seismic - So ..

66
6/23/2003

Calibration Required

• Wells
– Logs : P-wave Velocity / Density

Schlumberger Private
– Petrophysical Analysis : S-wave Velocity
• Model expected AVO
– Match to Seismic
• assign specific lithological meaning to seismic amplitudes
– No Match to Seismic
• Rework Synthetic and/or Seismic

Schlumberger Private
©WesternGeco

From the various well logs we are likely to be able to determine P-wave velocity
and density. S-wave velocity could be more problematical - it can be recorded
using specialised sources and detectors, however, more likely is the need to do a
petrophysical analysis in order to determine it.
From these quantities, a (prestack) synthetic can be generated showing any AVO
response with offset/time. This can then be compared to the real seismic at that
position. A match would then allow specific lithologies (as determined from the
well information) to be assigned to seismic amplitudes, and then the seismic
amplitudes along the line,away from the well site can be assigned some meaning
If the synthetic does NOT match the seismic then either the well processing or the
seismic processing (or both!) needs to be redone

67
6/23/2003

North Sea 3D Example

Schlumberger Private

Schlumberger Private
Courtesy of Chevron
©WesternGeco

North Sea Interpretation


• P and S sonic logs are added to the PP and PS images to support the
interpretation.
• P sonic shows no Vp contrast in passing from the cap rock into the oil-filled
reservoir. Hence the PP seismic image does not see the top reservoir.
• S sonic shows a large Vs contrast at the top reservoir which is consistent
with the bright PS seismic events, clearly delineating the top reservoir.
• In both P and S sonic logs, the OWC is quite clearly seen and has an
expression on both types of sea-bed seismic.

68
6/23/2003

Flowchart

Seismic Data Wavelet Reflectivity Edited Logs

Schlumberger Private
Synthetic

Rework
Compare
Seismic
using ???

Good Fit ? Update Logs using ???

Schlumberger Private
©WesternGeco Final Model

The compare module might be a least squares approach


The update module might be a general linear inversion (GLI)
technique or even a genetic algorithm (GA) technique.
The idea is that often certain key logs are missing or (partly) unusable. However,
there may be a relationship between existing logs and the missing ones - certain
assumptions are made about this relationship and the 'update' module tries to home
in on the exact relationship - many iterations may take place before the criterion of
the synthetic being generated fits the data
In this (AVO) case - we might have CMP data and sonic / density logs but to model
the AVO effect, we might need shear velocity or Poisson's ratio - this is guessed at
or modelled from existing logs then iteratively altered to make the synthetic match
the data
Then we have the Poisson's ratio log and the fun can start!

69
6/23/2003

Possible Scenarios

• Increase in Amplitude with Offset


– Gas Sand Reservoir - Bright spot

Schlumberger Private
• Decrease in Amplitude with Offset
– Carbonate Reservoir - Dim Spot

Schlumberger Private
©WesternGeco

We can have increase and decrease in offset indicating hydrocarbons - depends on


the geology!

70
6/23/2003

AVO Anomaly Types for Sand


I
0.15
II
0.1 III
IV

Schlumberger Private
0.05
Reflectivity

0
0 500 1000 1500 2000 2500 3000

-0.05

-0.1

Schlumberger Private
-0.15
Offset
©WesternGeco

Shows typical AVO responses for the 4 AVO categories described in Castagna &
Swan, 1997.

71
6/23/2003

Cross Plot - Class Types for Sand

Schlumberger Private

Schlumberger Private
©WesternGeco

Returning to the crossplot, and in this case a sandstone formation,the different


'classes' fall into different quadrants

72
6/23/2003

Refl. Coefficient versus Angle of Incidence

Schlumberger Private

Schlumberger Private
©WesternGeco

The above graph describes the AVO conclusions of most importance:


---- If impedance and Poisson's ratio change in different directions, then the
reflection coefficients decrease with angle of incidence
---- If Poisson's ratio remains constant at the reflector, then the AVO response is
often almost flat.
---- If impedance and Poisson's ratio change in the same direction, then the
reflection coefficients increase with angle of incidence

73
6/23/2003

Saturation Changes

Schlumberger Private

Schlumberger Private
©WesternGeco

As we know, there are many factors that influence seismic amplitude, e.g. source
energy and spectrum, coupling, receiver response, spherical divergence, reflection
coefficients etc. It is because there are so many factors, that when analyzing
amplitudes, we take advantage of relative measurements, to look for redundancy or
changes in time and/or space, as in bright spot analysis. Anything that causes a large
change in impedance (velocity multiplied by density) can cause a large reflection.
Candidates include changes in lithology and changes in saturation. The bright spot
methodology, which is based on normal incidence reflectivity, NI, actually involves
three different reservoir scenarios. If rock properties are known, lithologic
identification can be made by identifying the clustering of NI reflectivities from
different lithologic boundaries, e.g. below:
•Dim spot scenario, i.e. a large positive amplitude reduces to a smaller positive
amplitude. Normally associated with large acoustic impedances. Inversion of the
stack section, i.e. calculation of the acoustic impedance (or velocity for an assumed
density), has been very successful to infer lithology.
•Phase reversal scenario, i.e. a small positive amplitude changes to a small negative
amplitude. Extremely difficult to interpret as shows a tendency to disappear on
conventional stacks. Geological faults are often mistakenly introduced.
Interpretation of phase reversal responses require additional tools for lithologic
identification.
•Bright spot scenario, i.e. a negative amplitude increases to a larger negative
amplitude. Most successful of the three scenarios for both interpreting lithology and
estimating sand thickness.

74
6/23/2003

AVO
• What is AVO ?
• Why does AVO happen?
• Can we predict an AVO effect?

Schlumberger Private
• What parameters effect AVO?
• Can AVO effects be measured?
• What can be displayed?
• What do the results mean?
• A summary of AVO

Schlumberger Private
• And some other bits
©WesternGeco

The presentation follows the lines of answering questions

75
6/23/2003

Conclusion

• AVO is a measurable physical effect that is due to


energy losses caused mainly by P-wave to S-wave
conversions at interfaces

Schlumberger Private
• For normal recording geometries, a plot of reflection
amplitude versus the square of the sine of the
reflection angle produces an approximate straight
line
• The values of the intercept & gradient of this line
provide valuable insight into the velocity, density and

Schlumberger Private
Poisson Ratio contrast across that interface

©WesternGeco

76
6/23/2003

Conclusion (contd)

• Model results, produced from well data, can be used


to calibrate these values and identify lithological
units within the seismic section

Schlumberger Private
• Careful processing of the seismic data, designed to
eliminate spurious amplitude effects due to the
acquisition, is necessary to reveal the sought after
effect

Schlumberger Private
©WesternGeco

77
6/23/2003

AVO
• What is AVO ?
• Why does AVO happen?
• Can we predict an AVO effect?

Schlumberger Private
• What parameters effect AVO?
• Can AVO effects be measured?
• What can be displayed?
• What do the results mean?
• A summary of AVO

Schlumberger Private
• And some other bits
©WesternGeco

The presentation follows the lines of answering questions

78
6/23/2003

AVO Attribute Inversion: More Attributes


µ Mu, provides information relating to the rock matrix.
λ Lambda, provides information regarding pore fluid.

Schlumberger Private
Generally, a gas-filled sand will be more compressible
(have lower lambda values) than surrounding wet
sands, while mu will remain largely unchanged.

Schlumberger Private
©WesternGeco

More attributes we can derive now that we have AI. We can also create the λ/µ or
“fluid stack”.

79
6/23/2003

AVO Attribute Inversion: More Attributes


By recognizing that:
(λ + 2µ) µ
α2 = β =2
ρ ρ

Schlumberger Private
AI(P) = αρ AI(S) = βρ
Combining acoustic impedance terms (Goodway, 1997) yields:
2 2 2
AI (P) - 2AI (S) = λρ AI (S) = µρ
Lamé parameter, λ,incompressibility, and µ, shear rigidity

Schlumberger Private
may be combined via the “Fluid stack” ratio λ/µ.
©WesternGeco

More attributes we can derive now that we have AI. We can also create the λ/µ or
“fluid stack”.

80
6/23/2003

AVO Attribute Inversion: Isolating Fluid Effects


Smith and Gidlow (1987) subtracted the theoretical brine-filled
response from the AVO intercept, Ro, to isolate fluid effects.
This is their “fluid factor” or Fatti’s (1994) “Geostack”.

∆α β ∆β

Schlumberger Private
∆F = − 1.16 α
α β
Actual intercept Mudrock prediction
β/α may be derived by combining P-wave stacking velocities
and the mudrock relation of Castagna (1985) where α = 1360 +

Schlumberger Private
1.16β. (∆α/α) and (∆β/β) are derived from intercept/ gradient-
like calculations as described earlier.
©WesternGeco

More attributes we can derive now that we have AI. We can also create the λ/µ or
“fluid stack”.

81
6/23/2003

Which Attribute ?
• R0 * G became very popular in the Gulf of Mexico for
highlighting class 3 gas sands
• Large +/- R0 and G will result in even larger product

Schlumberger Private
• This method will FAIL for the other classes
• R0 * sign G gets around some of the problem but still
does not work for class 2 gas sands
• Fluid Factor should be more robust

Schlumberger Private
©WesternGeco

82
6/23/2003

Rock Physics/AVO Summary


• AVO is often used when AVA is meant (as here!)
• AVO is a way to access shear properties
• Impedance controls normal incidence reflectivity

Schlumberger Private
• Poisson’s Ratio controls far offset reflectivity
• Bulk modulus is the key fluid indicator
• Density might be estimated from high angle data (50o)
• AVO modelling builds confidence in log-seismic
integration

Schlumberger Private
©WesternGeco

83
6/23/2003

Funtime!
• Some exercises

Schlumberger Private

Schlumberger Private
©WesternGeco

84

Вам также может понравиться