Академический Документы
Профессиональный Документы
Культура Документы
Course Contents
1 - Introduction 2 - Basic Rock Physics 3 - Basic AVO theory 4 - GeoCluster AVO modules 5 - Preparing data for AVO
Overview 2D Land example 3D Land example 3D Marine example
Course Aims
Understand .
the basics of AVO theory parameters of GeoCluster batch modules
DINAT MUTAN ANGLE AMPVO
AVO - Introduction
AVO - Introduction
Reflected Amplitudes
are determined by the Reflection Coefficients being depend upon the rocks Physical Properties
velocity density
AVO - Introduction
AVO analysis - provides circumstantial evidence for the possible presence of hydrocarbons
X X X
gas
Basic Assumptions .
the Earth acts as a discretely layered medium
hydrocarbons change the rock properties amplitude changes across a CMP gather represent true variations of the reflection coefficient with incidence angle
Historically a very basic AVO analysis called Bright Spot Analysis was used in the 1960s and 1970s
this analysis was mainly post-stack and often by chance but very worthwhile: located many gas rich HC accumulations
Since the mid 1980s several additional types of AVO analysis have been developed
this analysis is mainly pre-stack
HOWEVER .
can be an extremely valuable method has been used to reduce the number of dry wells being drilled
Cefoga DPST07 - overview slide 9 June 02
Geoscience Training Centre
Review of
Basic Rock Properties
Isotropy and Homogeneity Porosity and Permiability Density
Anisotropy
Anisotropic media the physical property of the rock changes according to the direction in which it is measured
Homogeneity
Homogeneous media has physical properties which are the same everywhere within the material.
Porosity
pore volume per unit material high porosity = 35%, low = 10 %
Permeability
the ease with which a fluid can travel through the pore spaces
X Y
m = V
Cefoga
Geoscience Training Centre
January 2002
Matrix
Cement
Wyllies Equation.
= m (1 ) + f
January 2002
water gas
oil
In reality the fluids will be mixed into an emulsion containing different percentages of the different phases
clasts, grains
Matrix
Cement
Therefore:
f = (1 S )l + S g
January 2002
Elastic Moduli
Stress - the Force per unit area AND Strain - the degree of
deformation
Bulk Modulus
V = (V - V2)
F
F
Examples. Limestone Granite Sandstone 3.7 5.7 2.7 3.3 ~1.25
V PH = K V
K = Bulk Modulus PH = Hydrostatic Stress (acts equally in all directions) V/V = Volumetric Strain
Cefoga
January 2002
BH 070797
Shear Modulus
Y Ps = X
PS = Shear Stress
F
Examples Limestone Granite Sandstone 2.1 3.0 1.5 2.4 ~0.6
Mu,
= Shear modulus
POISSONS RATIO
F
R
R l
R R = l L
= Poissons Ratio
Final volume
Later, we see can also be expressed in terms of the velocities of body waves (Vp and Vs)
Cefoga DPST07 - Part 2 Page 15 January 2002
Geoscience Training Centre
BH 070797
Velocity = VP
Dilation Compression
SV WAVE
Wavelength
Direction of propagation
Velocity = VS
Double amplitude
VERTICAL MOVEMENT
Double amplitude
VVertical
VVertical
Isotropic
VVertical = VHorizontalY = VHorizontalX
Anisotropic
VVertical = VHorizontalY = VHorizontalX
Cefoga
January 2002
K + Vp =
4 3
1/2
Vs =
1/ 2
January 2002
BH 070797
100% Gas
Water saturation
January 2002
100% Gas
Water saturation
100% Water
K + Vp =
4 3
1/2
.we expect that the introduction of gas will create. A reduction in bulk modulus K which results in a reduction in VP. A reduction in density which results in an increase in VP.
VP
The net result is that the progressive introduction of gas theoretically causes an initial sharp reduction in VP after which there is a small increase.
Water saturation
100% Water
Cefoga
100% Gas
January 2002
VS =
1/2
.we expect that the introduction of gas will create... No change in overall shear modulus as fluids can not be sheared. A reduction in density which results in an increase in VS.
VS
The net result is that theoretically the progressive introduction of gas causes a small increase in VS.
Water saturation
100% Water
100% Gas
Nature of saturating fluid (gas, liquid) Saturation (oil, water) Water content Density
Unfortunately, until holes are drilled we usually do not know the values of these parameters.
Cefoga DPST07 - Part 2 Page 24 January 2002
Geoscience Training Centre
BH 070797
VP VS
+ Cefoga
January 2002
BH 070797
Major Controls
Cementation If no cementation
Grain shape Degree of grain sorting (by size) Overburden pressure
Modest Controls
Rock Type Clay content
Minor Controls
Saturant
Conclusion: The effects on Shear wave velocity caused by hydrocarbons being present in the saturant are relatively minor ones!
Cefoga DPST07 - Part 2 Page 26 January 2002
Geoscience Training Centre
BH 070797
1.0
Depth (kms)
Gas Sand
Oil Sand
Brine Sand
Below about 2.5 kms depth the curves tend to converge - the implication is that velocity
effects due to the presence of hydrocarbons will be difficult to see in the deeper parts of a section.
2.0
3.0
4.0
Why should the curves converge i.e. why does the influence of hydrocarbons become diminished?
1000 2000 3000
5.0
Velocity
Cefoga DPST07 - Part 2 Page 27 January 2002
4000
BH 070797
DEPTH OF BURIAL
Leads to an increase in velocity Beyond a certain depth the porosity is so low that the fluid composition has little influence on the average velocity. Small separation between gas and water sands - DHI/AVO anomaly less likely How does this effect correlate with the depth of real oil fields?
2.0
January 2002
BH 070797
Depth (kms)
1.0
Gas Sand
Oil Sand
2.0
3.0
24% of fields
Percentage of fields
Cefoga
January 2002
BH 070797
V1 V2 V1 V2 V1 V2 V1 Z V1 V2
Proportion r1
Proportion r2
1 r1 r2 = + V A V1 V2
It is a method of computing average velocity based on the velocities of the different layers and their relative preponderance.
Cefoga
January 2002
BH 070797
Matrix
Cement
1 1 1 = (1 ) + V Vm Vf
is porosity
It is often stated that porosity is the most important factor determining the velocity of sedimentary rocks
Cefoga DPST07 - Part 2 Page 31 January 2002
Geoscience Training Centre
BH 070797
Porosity = 25%
Fluid Vel (Water) = 1480m/s
V = 3328m/s
Cefoga
What predictions does Wyllies equation give for V when hydrocarbons are introduced?
Geoscience Training Centre
January 2002
BH 070797
2.5
2.0
1.5
This equation gives a Vp curve for gas sands with a continuing reduction in velocity!
0.8
1.0 Water
January 2002
BH 070797
Deep Gas Sand Wyllie prediction Measured For shallow sands, experiments show a more dramatic dip in velocity as soon as gas is introduced than predicted by Wyllie time averaging.
0 Oil or Gas
0.2
0.4
0.6
0.8
1.0 Water
Water saturation
K b + 4 / 3 b + F Vp =
1/ 2
b VS =
1/2
K b = bulk modulus(average)
K + 4 3 Vp =
1/2
January 2002
BH 070797
KS = bulk modulus(solid)
K f = bulk modulus(fluid)
K b = bulk modulus(average)
= porosity
Note:
This becomes zero if Kb/Ks = 1 i.e. the medium is solid with no fluids. This fluid factor is NOT the same as that we will see later in GeoCluster module AMPVO.
Cefoga DPST07 - Part 2 Page 36 January 2002
Geoscience Training Centre
BH 070797
K b + 4 / 3b + F Vp =
VS =
1/2
1/ 2
Vp = f( K, , , )
Vs = f( , )
Therefore given . Vp, density , porosity , fluid fill ..it is possible to generate models using new porosities or fluid fills. e.g. What happens if interstitial water is replaced by gas or oil?
Cefoga DPST07 - Part 2 Page 37 January 2002
Geoscience Training Centre
BH 070797
Biot-Gassman also has limitations in its applicability. It works well when the following assumptions are met.
The porous rock framework (skeleton) is macroscopically isotropic and homogeneous The skeleton, grains, fluids and saturated rock obey Hookes law.
No fluid enters or leaves any volume of the system and no cavitation occurs.
Biot Gassman is mathematically complex and also falls down when applied to small grained clastic rocks (e.g. mudstones)
Low permiability
2.4
Domenico Effect
2.1
VP
1.8
The Vp curve models the sharp change at small %gas better than the Wyllie Formula.
1.5
VS
0.8
January 2002
BH 070797
Vs VP
VP = VS
4 + 3
Where.. = bulk modulus or incompressibility = shear modulus or rigidity
VP VS
The Vp/Vs ratio is potentially an important diagnostic tool in seismic lithological determination.
Water saturation
100% Gas
The initial drop in the VP/VS 100% Water ratio is an HCI indicator.
Cefoga
January 2002
BH 070797
Coal
2.5 3.0 3.5
Vp/Vs
6 4 2
. .. . . . . . .. . . . . . . . . . . .. .. .. . . . . . . . . . . . . .. .. . . .. .... . . . . . .. . .. . . . . . . . . . . . . Can be used to predict S wave . . . . .... . . ... . . . . . . . . . . .... velocities where no such .. ...... . . . .. . ..... data has been recorded.
0 1 2 3
Vs (km/sec)
Cefoga
January 2002
BH 070797
6 4 2
Shale (or mud-rock) line for the basin under study Any rocks showing a variation from this line may be an indication of the presence of hydrocarbons.
.... . . . .. .....
0 1 2 3
Vs (km/sec)
Cefoga
January 2002
BH 070797
POISSONS RATIO
We saw earlier that Poissons Ratio can be defined in terms of its change in lateral dimension relative to its change in vertical dimension
In fact Poissons Ratio can also be expressed purely in terms of Vp and Vs.
1 Vp 1 2 Vs = 2 Vp 1 Vs
OR
VS = VP
1 2 1
POISSONS RATIO
Plotting Poissons Ratio for various ratios of Vp and Vs.
-0.1 -0.2 1 3 5 7
A small change in the Vp/Vs ratio in the range of 1.5 to 2 generates a large change in
January 2002
BH 070797
POISSONS RATIO
Poissons Ratio
0.4
0.3
99% 96%
0.2
0.1
January 2002
BH 070797
0.4
With the introduction of a little gas Vp drops rapidly but Vs 0.2 doesnt Therefore VP/VS also drops rapidly. As is closely related to VP/VS then shows a 0 similar drop.
0.3
0.1
0 100% Gas
0.2
0.8
January 2002
BH 070797
0.50
Vp
0.25
Vs
0 0.5 1
0.00 0 0.5 1
Gas
Water saturation
Water
Gas
Low observed for gas filled reservoir Forms basis of the AVO method for direct hydrocarbon detection (DCI).
Poissons ratio
Velocity
Vp Vs
0.50
0.25
Water
Water
Oil has less effect on rock properties than gas Nevertheless AVO method useful for certain oil cases
Useful in cases of live oil (oil containing dissolved gas), but not so useful in cases of dead oil (oil without gas)
Cefoga DPST07 - Part 2 Page 50 January 2002
Geoscience Training Centre
BH 070797
P-wave Velocity and Lithology Can P-wave velocity be used as a diagnostic tool for indicating lithology?
A plot of P-wave Velocity against occurrence for commonly found rocks..
Occurrence (normalised)
Conclusion is that this diagnostic may be of some limited use. There is a large degree of overlap between the different types of rock.
2.0
4.0
6.0
7.0
Poissons Ratio vs. P-wave Velocity Can Poissons Ratio be used as a diagnostic tool for indicating lithology?
A plot of Poissons Ratio against P-wave Velocity for commonly found sedimentary rocks..
Poissons Ratio 0.5
Shale
0.4
0.3
Water Sand
Limestone/ Dolomite
Conclusion is that, again the diagnostic may be of some limited use. There is still a large degree of overlap between the different types and classes of rock.
P-wave Velocity (kms/s)
0.2
7.0
If well data available can perform Vp/Vs, and Poissons ratio studies. Poissons ratio is the key factor controlling AVO
Contents
Reflections at normal and non-normal incidence Zoeppritz equations and their approximations AVO classes - Rutherford and Williams
Rock properties
Determines
P & S velocities
Determines
AVO
Geoscience Training Centre
Incident P V1 . 1 = Z1 V2 . 2 = Z2
Reflected P
Transmitted P
V2 2 V1 1 R0 = V2 2 + V1 1
Z 2 Z1 Z 2 + Z1
V
Sandstone on Limestone Limestone on sandstone Soft ocean bottom Hard ocean bottom Base of weathering Shale over water sand Shale over gas sand Gas sand over water sand 2.0 3.0 1.5 1.5 0.5 2.4 2.4 2.2
V
3.0 2.0 1.5 3.0 2.0 2.5 2.2 2.5
Z1/Z2
0.67 1.5 0.5 0.2 0.19 0.96 1.39 0.69
R0
0.2 - 0.2 0.33 0.67 0.68 0.02 - 0.16 0.18
Mode conversion occurs at non-normal incidence. Conversion of P-wave energy to S-wave energy.
Reflected S Incident P Reflected P V1p V1s 1
V2p V2s 2
Transmitted P Transmitted S
Cefoga
January 2002
Incident P
Reflected P Reflected S
Incident S
Reflected P Reflected S
Cefoga
January 2002
A B
-Sin 1 -Cos 1
Complex - yield little physical insight into what is -d22 -1 -d22 happening Sin1 to amplitudes Cos22 Sin2 2 D 1 d11 1 d1 Even more powerful digital computers needed B before using routinely in exploration applications A S
1
=
Cos22 C Sin21 -Cos 1
1
1
R (A)
P
11d1 22d2
P S
Model B
Vp = 3048 m/s Vs =1524m/s = 2.20 g/cc
Model C
Vp = 4877 m/s Vs= 2438 m/s = 2.40 g/cc
R0=0.63
R0=0.41
Vp = 6096 m/s Vs = 3048 m/s = 2.65 g/cc
R0=0.16
Zoeppritz equations show how the reflection coefficient changes with incidence angle..
Model C
P-wave reflection coefficient versus incidence angle for 3 different interface models
Cefoga
January 2002
Allow Poissons Ratio to Vary across the Interface Koefoed (1955) made models by varying Poissons Ratio ().
Reflection coefficient
= 0.25 = 0.40
Change in of lower rock
= 0.25 = 0.15
Knott Zoepritz
Bortfield
Aki / Richards
Shuey
Approximation adopted by CGG
Cefoga DPST07 - Part 3 Page 12 January 2002
Geoscience Training Centre
Where
R() = P wave reflection coefficient VP = Change in P wave velocity (VP2 VP1) VP = Average P wave velocity (VP2 + VP1)/2 VS = Change in S wave velocity (VS2 VS1) VS = Average S wave velocity (VS2 + VS1)/2 = Change in density (2 1) = Average density (2 + 1)/2
DPST07 - Part 3 Page 13 January 2002
( + )
2
In practice is approximated by .
Cefoga
Geoscience Training Centre
DPST07 - Part 3 Page 14 January 2002
Cefoga
Geoscience Training Centre
Dominates a near trace stack which can be considered to image P wave impedance contrasts.
Dominates a far trace stack which can be considered to image Poissons ratio contrasts.
Typically, as seismic surveys only used to involve incidence angles to about 300, this term may be dropped.
Although offset is limited by acquisition, mute and NMO stretch effects the third term will be required in future!
Cefoga DPST07 - Part 3 Page 15 January 2002
Geoscience Training Centre
Important Approximations to Zoeppritz Comparison of Zoeppritz with approximations for a simple gas sand model.
Amplitude
0.5
0.4
0.3
0.2
0.1
Approximations give
Shuey = 2% error at 300 Aki & Richards = 5% error at 400
0 10 20 30 40
Cefoga
0.0 50
January 2002
R ( )
9 2 = R 0 + R 0 sin 4
9 R( ) = R0 + R0 sin 2 4
R ()
R( ) = R0 + Gsin
2
Where, Gradient
Intercept = R0
9 G = R0 4
Sin 2
Note that the horizontal axis is in terms of incidence angle - not offset x !
Cefoga DPST07 - Part 3 Page 18 January 2002
Geoscience Training Centre
R G= sin2
R
Theoretical curve
sin2
R0
January 2002
AI
+ve
+ve
+ve
-ve
Time
Increasing positive +
Time
AI = acoustic impedance = Poissons ratio R0 Offset Polarity reversal
Cefoga
R0
Offset -
Increasing negative
January 2002
BH 070797
AI
-ve
+ve
-ve
-ve
Time
Increasing positive +
Time
AI = acoustic impedance = Poissons ratio
Offset R0
Increasing negative
R0 Cefoga
January 2002
BH 070797
In this paper a simple earth model was proposed to represent a typical potential hydrocarbon trap a gas filled sandstone layer sandwiched between two impervious shales.
Shale The consequences of changing the Acoustic Impedance of the gas sand relative to that of the encasing shales is considered
Gas sand
Shale
Class 1
Significantly higher impedance gas sand encased within lower impedance shales
R() + R0
Cefoga
Geoscience Training Centre
BH 070797
Class 2
Gas sand impedance very similar to surrounding shale, either slightly above or below..
Class 3
Significantly lower impedance gas sand encased within higher impedance shales
R() +
high impedance shale low impedance gas sand high impedance shale
R0
-
Cefoga
Geoscience Training Centre
R0 +
Upper interface...
(degrees)
low impedance shale high impedance gas sand low impedance shale
Class 1
Lower interface...
(degrees)
R0 R0 R0
+ + Upper interface...
shale
(degrees)
Class 2
Lower interface...
(degrees)
In fact the Zoeppritz equations show that the exact solution is not a true mirror image. The Shuey approximation does not however recognise this.
Class 3
R0 R0
+ + -
Upper interface...
(degrees)
high impedance shale low impedance gas sand high impedance shale
Cefoga
Lower interface...
(degrees)
January 2002
BH 070797
Class 1
Class 2
Class 3
class 1
class 2
5 -1 10 15 20 25 30 35
Angle of incidence
class 1
class 3
-2
class 2 class 3
Cefoga DPST07 - Part 3 Page 28 January 2002
Geoscience Training Centre
Near
Far
Porosity 0%
Peak, fairly constant w.r.t offset
Near
Far
Porosity 20%
Trough, reducing w.r.t offset
Near
Far
Porosity 0%
Peak, reducing w.r.t offset
Near
Far
Porosity 20%
Trough, slightly reducing w.r.t offset
Near
Far
Near
Far
Anhydrite Porosity 0%
Peak, fairly constant w.r.t offset
Porosity 20%
Trough, reducing w.r.t offset
Porous Dolomite
Near
Far
Porosity 0%
Peak, reducing w.r.t offset
Near
Far
Porosity 20%
Trough, increasing w.r.t offset
AVO Pitfalls
The AVO effect may not show true variation of reflection coefficient with incidence angle
e.g. processing artefacts, noise interference, tuning effects (lower frequency on far traces)
Amin direction
Amax direction
0
X
Amplitude measured for a 2D survey: A(x,0)
This implies that wide azimuthally acquired data should be analysed in terms of elliptical variation of amplitude! For details see the CEFOGA course DPST22 Anisotropy
Cefoga DPST07 - Part 3 Page 32 January 2002
Geoscience Training Centre
Summary
Reflection amplitudes at non-normal incidence governed by Zoeppritz equations These require approximations to be useful Shuey approximation (as used in Geocluster) AVO response falls into four classes There are several AVO pitfalls waiting to trap the unwary!
AVO Products
It follows from the previous section that all reflecting interfaces have an AVO response. Therefore it is the changes, the anomalies, in this response which we are seeking..
X X X
gas
R sin2 R sin2 R sin2
The basic AVO attributes extracted from the data are R0 and G. It is possible to combine theses attributes in various ways in order to enhance the anomalous areas
Cefoga DPST07 - Part 4 Page 2 January 2002
Geoscience Training Centre
R( ) R0 + Gsin
2
Where. R() = reflection coefficient at any
incidence angle R0 = zero offset reflection coefficient
R()
Gradient
Intercept
9 G = R0 4
G = gradient =
a complicated combination of density contrast and reflection coefficients and is related to the change in Poissons ratio, = (1 2)
R0 Sin2
= angle of incidence
Cefoga DPST07 - Part 4 Page 4 January 2002
Geoscience Training Centre
AVO In Practice
1 2 3
sample time
R()
Fit a linear regression line to create a single R0 and G.
R0
* * *
* ** * *G
Sin 2
R0 Plot
R0
Stack Amplitudes
R0
G plot
Changes in Vp/VS produce changes in the Gradient. The presence of gas in porous rocks affects the Vp/VS. It follows that Gradient could be a good indicator of gas reservoirs.
Cefoga DPST07 - Part 4 Page 7 January 2002
Geoscience Training Centre
G plot
However Gradient based on amplitude may be somewhat unstable and susceptible to small variations in phase and residual velocity errors!
G (envelope) computation
In order to resolve the potential problems with Gradient based on amplitude it can be based on trace envelope.
Differences between Gradient based on Amplitude and Gradient based on Envelope.
Envelope Gradient not effected by alternating sign of the seismic wavelet BUT: Can result in loss of resolution.
G (amplitude)
G (envelope)
The anomaly in the G (envelope) stands out better from the background but the display is lower frequency.
Cefoga DPST07 - Part 4 Page 10 January 2002
Geoscience Training Centre
ANGLE Stacks
CMP gather
Angle stacks.
Angle stacks are computed for defined angle ranges.
time
Angle ranges
STACKS
The concept is to produce a plot of the value of the HCI at its correct position in t.w.t. time and space.
Cefoga DPST07 - Part 4 Page 12 January 2002
Geoscience Training Centre
There are several HCIs in common use today, including AVO Response Indicator
G (amplitude) x sign(R0) All are used for identifying CLASS III type anomalies.
R0 < 0
G<0
R0
+
Lower interface...
Increasing negative
Class 3
Increasing positive
(degrees)
R0 > 0
G>0
R0
-
Both top and bottom reflections produce an increase in amplitude with offset
Cefoga
January 2002
AVO Response Indicator This is the product of G (amplitude) and the Sign of R0
G contains information about how R() changes with offset. R0 denotes the starting point for variation in the reflection coefficient.
+R0 -R0
-
+ +
+G
Increasing positive amplitude
(degrees)
+R0 -R0
-
Cefoga
-G
+G
-G
January 2002
AVO Response Indicator example The AVO Response Indicator on a Class 3 anomaly is characterised by a red doublet when displayed with a standard colour palette
Colour palette. Positive Zero Negative The red doublet shows a layer where both the top and bottom reflections have an overall increase in amplitude magnitude with increasing offset.
Product = G (amplitude) x R0
+
+R0 -R0
-
+ +
+G
Increasing positive amplitude
(degrees)
+R0 -R0
-
-G
+G
-G
Now the larger the initial values of R0 and/or G the higher the product. The result is a HCI that shows any Class 3 anomaly as a strong red doublet.
Other Class anomalies and non-hydrocarbon related interfaces should not show this response.
Cefoga DPST07 - Part 4 Page 18 January 2002
Geoscience Training Centre
-R01
This is likely to be the situation where there is no signal present, only noise. In this case, the product of R0 and G produces a negative value.
In summary, the presence of noise will bias the product towards negative values.
Cefoga DPST07 - Part 4 Page 19 January 2002
Geoscience Training Centre
Unbiased AVO Product Indicator This is the product of G (amplitude) and Stack.
The product is unbiased and is therefore an alternative and better HCI then the AVO Product Indicator...
..... . . . . . . . . .
. . . . . . . . . . .. . . . .... . .. .. .
(degrees)
Predicted amplitudes are summed (stacked) Stack amplitude normalised by the stack fold.
i.e. Method uses the amplitudes from the best fit line - not the raw data amplitudes themselves.
Cefoga DPST07 - Part 4 Page 20 January 2002
Geoscience Training Centre
Using G (envelope) Any of the foregoing HCI can be computed using, instead of the Gradient based on Amplitude the Gradient based on the Envelope.
Fluid Factor
The Fluid Factor HCI is used to highlight the presence of gas, regardless of the class of AVO anomaly.
The exact meaning of the Fluid Factor is still being debated. This is because there are several ways in which it can be derived e.g. either theoretically or empirically.
This should not be confused with the fluid factor in the Bio-Gassmann equations.
The underlying rational is to find a regional trend which, when subtracted from the input data zeroes all values except potential Hydrocarbon zones
Input data Determine Background trend Background Trend Subtract Background from input Output = Difference = Fluid Factor Section
Fluid Factor
Recalling Castagnas Mud rock line
Vp (km/sec) Shale (or mud-rock) line - Water Filled
6 4 2
. . . ... .... . . . . . Vp (km/sec) . . ... . . . .. .. . .. . . . . . . . . . . . 6 . ..... ... . .. .... .. . .. . . . . .. . . . . . . . . . . .... . . . . . . . . .... . ... .... .. . . . .. .. ... .. . . . Vp = 1.16Vs + 1.36 (km/sec) 4
0 1 2 3
Vs (km/sec)
Thus if the background trend is assumed to be the mud rock line, when it is subtracted.. Water filled sands should disappear Gas filled sands will remain as an anomaly.
ud-
k ro c
line
Ga
s nd a sS
Low porosity rocks
Dr
0
s nd a yS
1
The Fluid Factor can be considered to be the difference between the background trend and the residual values.
However there are other rocks (e.g. Carbonates) that will not lie along the mud rock line!
Cefoga DPST07 - Part 4 Page 23 January 2002
Geoscience Training Centre
Smith and Gidlow did some hard sums and came up with a theoretical Fluid Factor given by
VS VS VP FF = 1.16 VP V P VS
The presence of a coefficient of 1.16 is a clue to the involvement of Castagnas mud rock line for the Vp/Vs relationship during the derivation of this equation.
In fact this Fluid Factor turns out to not very useful when applied to real rocks. It is for this reason that various empirical estimated methods have been developed.
Cefoga DPST07 - Part 4 Page 24 January 2002
Geoscience Training Centre
This coincides with Castagnas Mudrock line where Vp = 3238m/s and Vs = 1619m/s. This corresponds to the zone of high porosity sandstones
line k c - ro Mud
nds a S ter Wa
High porosity rocks
1 Cefoga
3 Vs (km/sec)
January 2002
1 VP 1 VP VS 1 2 + + sin R( ) 2 VP 2 VP 2 VS
The normal incidence reflection coefficient is given by
1 VP for P waves + R0 = 2 VP
Also
R0 S
1 VS for S waves = + 2 VS
VP
Smith and Gidlows theoretical value of the Fluid Factor can be modified to become
FF 1.252 R0 + 0.58G
Therefore a Fluid Factor can be obtained by summing the scaled AVO attributes R0 and G. The result of summing normalised values of R0 and G produces what is called the fluid factor section.
Cefoga DPST07 - Part 4 Page 27 January 2002
Geoscience Training Centre
0.0
-1.0 Any points not lying on the trend are indicative of a HCI anomaly.
-2.0
-0.2
-0.1
0.0
0.1
0.2
Normalised Intercept R0
Normalised gradient
0.5
0.0
Class IV
Normalised intercept
Cefoga DPST07 - Part 4 Page 29 January 2002
Geoscience Training Centre
AVO Anomaly Classes on the Cross-plot Summarising the anomaly classes. Class I Class IV AI Z
AI Z
Amplitude decreases with offset
R0, G Cross-plot
1.0 0.5
G 0.0
Class III
AI Z
Amplitude increases with offset = Bright Spot
Class II
AI Z
-0.5 0.0 0.5 1.0
R0
0.0
-1.0
-2.0
Anomalous points which do not have plot along the background trend.
-0.2
-0.1
0.0
0.1
0.2
Intercept
Cefoga
January 2002
R0
Ro and G Normalisation
To find the scaling factors for R0 and G there are three possible options:
Theoretical
As noted, not really applicable to real data.
Empirical
Choose scaling visually Apply to the anomaly
FF = aR0 + bG
It is possible to combine the scaling coefficients a and b into one coefficient as.
FF = R0 cos + G sin
is determined empirically by
scanning with several values
Select the value of where the energy in the panel is the weakest Apply that value to the region of the anomaly.
Cefoga DPST07 - Part 4 Page 35 January 2002
Geoscience Training Centre
FF= C(x,t)R0 + G
Where C is the variable, in time and space, scaling factor.
In practice the result is a stack, with amplitudes weighted by offset - hence far-offset stack method.
Cefoga DPST07 - Part 4 Page 36 January 2002
Geoscience Training Centre
350 550
< 350
The 3 term Shuey approximation now becomes preferable. It can be expressed in the form
R( ) aZ P + bZ S + c
= change in density
Cefoga
Conclusions
DINAT - incidence angle computation ANGLE - computation of angle stacks and gathers MUTAN - muting according to angle value (level 8100) AMPVO
Intercept (R0) and Gradient (G) outputs QC of R0 and G computation HCI indicators
DINAT
Vint(1)
Vint(2)
Vint(3)
no SMTHANG
SMTHANG
Parameter SMTHVEL smooths velocities prior to interval velocity and angle computation
DINAT display
MUTAN result
ANGLE (1)
Computation of angle stacks and /or gathers Angle computation method same as DINAT
straight ray ray bending (SMTHANG , SMTHVEL)
R S R
ANGLE (2)
Angle stacks
CMP
offset 5 degrees
STACK
time 25 degrees
ANGLE (3)
5 deg
10 deg
15 deg
20 deg
AMPVO
- Ro
can be written as
R () = Rp + G sin
2
W here... R ( ) = change of reflection coeff with at angle ( ) Rp = zero (normal) incidence P wave reflection coeff G = gradient term depending upon change in Poisson' s ratio
Cefoga DPST07 - Part 4 Page 14 January 2002
Geoscience Training Centre
AMPVO - In Practice
* * * * *
* * * * *
Second fit
sin2
Cefoga DPST07 - Part 4 Page 16 January 2002
Geoscience Training Centre
*
A
* * * * * *
* *
sin2
Cefoga DPST07 - Part 4 Page 17 January 2002
Geoscience Training Centre
*
A
* * * * * *
* *
*
LORENTZ
ANDREWS
sin2
Cefoga DPST07 - Part 4 Page 18 January 2002
Geoscience Training Centre
Tri-median fitting
find median of each group
* * * * * * * *
sin2
divide into 3 groups
Cefoga DPST07 - Part 4 Page 19 January 2002
Geoscience Training Centre
Assumes events originate from single reflectors Assumes seismic data are convolution of zero offset data with AVO response
R ()
*
Ro
gradient G= 9/4
- Ro
Sin 2
Perform AVO analysis in rolling time window (WLEN) centred on current sample time Estimate wavelet from near mid or full stacks Use samples of estimated wavelet and actual wavelet to extract the AVO response along the offsets
WLEN
full
0 deg
near
10 deg
20 deg
far
30 deg
WLEN
too large, then possibly more than one event - small gradient values too small, full benefits in improvement of S/N not achieved
WLEN
WLEN
AMPVO - R0
Main output
R0 computed from regression analysis
stack
R0
AMPVO - Ga
stack
Gradient (Ga)
AMPVO - Ge Computation
AMPVO - Ge and Ga
Example of Ga and Ge
Ga
Ge
Can we believe our estimated AVO attributes? Do we believe our AVO models assumptions?
Confidence in:
estimated AVO attributes
attribute error estimate correlation coefficient
Coefficient = 0.94
Coefficient = 0.32
Comparison with Ga
gradient
Gradient
Correlation coefficient
Cefoga DPST07 - Part 4 Page 30 January 2002
Geoscience Training Centre
residual
Z=+ve x x x x x
Z=+/-0 x x x xx x x
x x x
run
Cefoga DPST07 - Part 4 Page 31 January 2002
Geoscience Training Centre
*
A
* * * * * *
* *
*
predicted
observed
Stack of residual gathers should be white noise! Data which stacks up may be due to Residual NMO, Statics, Multiples etc.
Residual stack
Residual gathers
Cefoga
January 2002
AMPVO - HCIs
Three HCIs are in common use today - all may be output from AMPVO
1) AVO Response Indicator 2) AVO Product Indicator 3) Fluid Factor Indicator
class 1
1
class 2
5 -1 10 15 20 25 30
Angle of incidence
35 class 1
class 3
-2
class 2
class 3
AI
shale
class 1
shale shale
sand
Stack
polarity reversal
class 2
sand shale
Near zero impedance contrast between sand and shale Zero synthetic gives very poor tie to stack Classic bright spot DHI Easiest to detect using AVO attributes (e.g. Ro * G)
shale
class 3
sand shale
(Ga)* sign(R0 ) - AVO Response Indicator (Ga)* (R0 ) - product (Ga* STACK) - unbiased version of product
All used for identifying CLASS III anomalies:
SHALE
R0 < 0, G < 0
R -R0 sin2
GAS SAND
R0 > 0, G > 0
R +R0 sin2
Both top and bottom reflections produce increase in amplitude with offset
SHALE
Cefoga DPST07 - Part 4 Page 36 January 2002
Geoscience Training Centre
6 4 2
Vs (km/sec)
R0 and G Crossplot
0.0
-1.0 anomalies
-2.0
intercept
-0.2
DPST07 - Part 4 Page 39
-0.1
January 2002
0.0
0.1
Cefoga
0.2
gradient
0.0
Class IV
-0.5
Class III
-1.0 -1.0 -0.5 0.0
Class II
0.5 1.0
intercept
Cefoga DPST07 - Part 4 Page 40 January 2002
Geoscience Training Centre
Usually R0 and G are different orders of magnitude Need to find scaling factors
R0
Ga
Three options:
Theoretical (not applicable to real data) Empirical
choose scaling visually apply to the anomaly
*
A
* * * * * *
* *
*
gradient
Amplification Factor =
stack
sin2
Cefoga DPST07 - Part 4 Page 46 January 2002
Geoscience Training Centre
*
A
st
xt , x pred
* * * * *
sin 2 (x) sin 2 (s)
* *
sin2
Cefoga
Geoscience Training Centre
January 2002
In practice amplification factor is derived from measured amplitudes within a time window
Analysis window
Ro and G are then derived from the amplification factor and stack amplitudes
Amplification factor
gradient
intercept
Conclusions
DINAT - incidence angle computation ANGLE - computation of angle stacks and gathers MUTAN - muting according to angle value (level 8100) AMPVO
Intercept (R0) and Gradient (G) outputs QC of R0 and G computation HCI indicators
DINAT - incidence angle computation ANGLE - computation of angle stacks and gathers MUTAN - muting according to angle (level 8100) AMPVO
Main and auxiliary (HCI) outputs QC of R0 and G computation
DINAT
The approximation is made that Vrms equals Vstacking in the input velocity library.
t0/2
(one way)
tX/2
tX/2
Vavg
Method assumes that the event of interest is a flat interface with only one layer above it.
sin =
X 2 V avg tX 2
X V avg t X
Where.
X2 tX = t + 2 Vavg
2 0
Cefoga
January 2002
Vint(1)
t0/2
H1
Assumes that the input velocity picks are on geological interfaces. It builds up a flat-layered horizon model from the picks. The ray paths are allowed to bend across these interfaces but are otherwise straight.
2 int, n
2 rms , n o , n
2 rms , n 1 o , n 1
V
j =1 n
2 int, j
t j
j
t o ,n t o ,n 1
t
j =1
Assuming that Vrms can be approximated by the stacking velocity, the hyperbolic NMO equation is t2(x)
= t02 + x2/V2rms.
dt ( x ) x = 2 dx Vrms t ( x )
dt(x)/dx is also the slope of the time-distance curve. It can be shown (!) that for every layer n (for example, see Yilmaz 1987, pages 429-431), dt(x)/dx is given by
dt ( x ) sin( ) = dx Vint, n
sin( ) =
2 Vrms
Vint x
2 x 2 t0 + 2 Vrms
Note that as the interval velocity function is not smooth it will cause discontinuities in the computed angles across the layer interfaces (when Vint jumps).
As the Incidence angles are computed from the Interval velocities - blocky interval velocities can cause corresponding blockiness in the angle computation Sometimes the blocky effect is large enough to produce unwanted artefacts in the AVO attribute sections. To correct this problem the incidence angles can be smoothed in DINAT in either of two ways - Directly or Indirectly.
Cefoga DPST07 - Part 4 Page 9 January 2002
Geoscience Training Centre
First calculates the angles, then smooths the sin()'s over time (at constant offset) with a simple sinc function. The length of the sinc filter can be changed using the SMTHANG parameter recommended value is 64
samples (i.e.256 ms @4 ms)
no SMTHANG
Cefoga DPST07 - Part 4 Page 10 January 2002
SMTHANG
The method then follows the raybending approximation and calculates the interval velocity and sin().
As the input Vrms are smoothed so are the computed interval velocities and consequently so are the calculated angles i.e. the smoothing of input velocity has indirectly smoothed the incidence angles.
Cefoga DPST07 - Part 4 Page 11 January 2002
Geoscience Training Centre
Direct smoothing, SMTHANG, is likely to produce better results if the velocity picks are on real, geological horizons.
Indirect smoothing, SMTHVEL, allows the degree of polynomial fit to be changed (namely linear, quadratic or cubic). Because it applies a sinc function it cannot work on the very ends of the trace Note that the indirect smoothing will not, in general, preserve the details of the velocity function.
MUTAN result
Key parameters defining mute limits
MUTAN basically computes angles using the identical methods as described for module DINAT.
ANGLE (1)
Ray bending
SMTHANG , SMTHVEL
S R
ANGLE (2)
Angle stacks.
To compute angle stacks the user supplies a set of angles, plus a parameter RANGE CMP gather
offset Angle ranges to be stacked For an angle of 5 deg and a value of RANGE = 1 the program creates a corridor stack 4-6 degrees.
STACKS
5 degrees
15 degrees The angle is written in WORD 7, allowing sorting into angle gathers
time
25 degrees
5 deg
10 deg
15 deg
20 deg
AMPVO
Compute linear regression to create a single R0 and G for each sample time
Sin 2
January 2002
A
Second fit
* * *
* * * * * *
sin2
AMPVO - Fitting Strategies for the Regression L1 norms L1 first option - Minimises the absolute value difference between model and
observed values.
*
A
* * * * * *
* *
sin2
Cefoga DPST07 - Part 4 Page 20 January 2002
Geoscience Training Centre
*
A
Initial fit
* * * * * *
* *
LORENTZ
Second fit
sin2
Cefoga DPST07 - Part 4 Page 21 January 2002
ANDREWS
* * * * * * * * *
* *
*
2) find median of each group
sin2
3) fit using least squares regression
Cefoga DPST07 - Part 4 Page 22 January 2002
Geoscience Training Centre
**
* *
*
sin2
** * * *
sin2
To minimise this effect instead of using single samples measurements using a wavelet of defined length can be used.
Cefoga DPST07 - Part 4 Page 23 January 2002
Geoscience Training Centre
R ()
Ro
January 2002
The window is centred on the current sample time. Should be about the same length as the dominant wavelength
If WLEN too large, then possibly more than one event - small gradient values
WLEN
WLEN
10 deg
full
20 deg
far
30 deg 30 deg
The stack with maximum energy is used as the characteristic wavelet shape.
Advantages: The inclusion of more data and hence noise influence in the analysis results. Disadvantages: A poor gradient result if the actual wavelet is inconsistent
Cefoga DPST07 - Part 4 Page 26 January 2002
Geoscience Training Centre
R0
Gradient (Ga)
AMPVO - Ge and Ga
Ga
May be weighted by a correlation coefficient (see later)
Ge Notice how the Ge stands out much better from the background
Cefoga
January 2002
* * * * * * * * * * * * * * * *
Can we believe
G? * G? * G?
Sin 2
Cefoga DPST07 - Part 4 Page 29 January 2002
Geoscience Training Centre
A
* * * * * * * * * *
* * ** * * *
A
G
* * * * * *
Sin 2
* * * * * * *
* * *
G+1
* *
G + 10
*
Sin 2
Suggested in XDOC that error estimates are plotted as a grey scale section.
Cefoga
For derivation of the error see Advanced Technical Description or Draper & Smith 1981
January 2002
AMPVO QC Tools - Correlation Coefficient (1) Correlation Coefficient (goodness of regression fit)
Is a measure of how well the data points line up in a straight line...
Coefficient = 0.94 Coefficient = 0.32
Gradient Ga Correlation coefficient CC Red and Blue represent high coefficients Green represents low coefficient
In Geocluster CC varies between +10000 and -10000
Cefoga DPST07 - Part 4 Page 33 January 2002
Geoscience Training Centre
?
*
* *
* * * * * *
The runs statistic therefore gives a measure of how well a straight line fit represents a realistic model.
12
* * *
4
* * *
6
* *
8
* * * * *
11
13
7
Number of runs =13 Number of points = 18
In some cases comparing the number of runs to the total number of points is
regarded as the statistic the smaller the ratio the better!
Number of runs =13 Number of points = 18
However, in this example that particular statistic would give the same result
* *
* * *
* * * 4
* * *
* * * 8
* * *
1 2
*
5 6
9 10
*
11
* 12
13
*
Z=
If result is.
u 0.5
where. =
=
2n1n2 +1 n1 + n2
+1
Large -ve Z
means too few runs.
and.
2
2 n 12 n 2 ( 2 n1 n 2 n1 n 2 )
(n1 + n 2 ) (n1 + n 2 1 )
2
Near zero Z:
means a near correct number of runs - the model line is appropriate for the line.
Large +ve Z
means too many runs.
Where.. u is the number of observed runs is the number of expected runs n1 is the number of positive residual points n2 is the number of negative residual points
Run statistic value of Z is multiplied by 10000 in Geocluster.
*
predicted
* * * *
Residual
At each sample time the resulting values for the offset range should, on average, be random
Residual A
* * *
observed
* *
* *
January 2002
Residual gathers Data which stacks up may be due to Residual NMO, Statics, Multiples etc.
Cefoga DPST07 - Part 4 Page 38 January 2002
Residual stack
AMPVO - HCIs
Several of the HCIs in common use today may be output from AMPVO AVO Response Indicator - (Ga) x sign(R0 ) AVO Product Indicator - (Ga) x (R0 ) Unbiased version of product - (Ga x STACK)
All used for identifying CLASS III anomalies:
Pass Pass 1 1
Output 00and OutputR R andG Gplus plusa afew fewQC QCoptions options
E.g. E.g.Regression Regressionplots, plots,residual residualCDP CDPgathers gathers
Pass Pass 2 2
Output Outputall allAVO AVOdisplays displays Perform Performan anin-depth in-depthQC QCanalysis analysis
Pass Pass 3 3
Final Finalpass passto totweak tweakinput inputparameters parameters
* * ** * * *
* *
R0
S
Cefoga DPST07 - Part 4 Page 42 January 2002
Geoscience Training Centre
TWIN = 70ms
* * * * * *
sin2 ( x) sin2 ( s)
st xt , x pred
* *
From Shuey ,
pred 2 xt , x = R0,t + Gt sin x
sin2
Cefoga
January 2002
R0 and G are then derived from the Amplification factor and stack amplitudes.
Amplification factor
gradient
intercept
THOLD= 0.2
The approach for TAVOF analysis involves the creation of 3 data sets.
Zero offset projection section
Derived from the amplification factor.
The production of these robust sections reduces the need for in house storage of large data sets.
Conclusions
DINAT - incidence angle computation ANGLE - computation of angle stacks and gathers MUTAN - muting according to angle value (level 8100) AMPVO
Intercept (R0) and Gradient (G) outputs QC of R0 and G computation HCI indicators