You are on page 1of 102

Landmark AVO Background A-1

!""#$%&'
AVO Background
The theory behind AVO analysis Ior hydrocarbons in sedimentary rocks is relatively
straightIorward. It depends upon the diIIerences in the response oI the P-wave velocity (V
p
) and
the S-wave velocity (V
s
) oI a reservoir rock to diIIerent Iluids in the pore spaces. P-waves are
sensitive to changes in pore Iluids. The introduction oI only a small amount oI gas into the pore
space oI a rock can reduce the P-wave velocity oI the rock signiIicantly. In contrast, S-waves do
not 'see the pore spaces oI a rock and have a velocity that depends mainly on the rock
Iramework.
Petrophysical interpretation oI seismic data has, to a large extent, traditionally been limited to one
parameter (such as the change in P-wave impedance or interval velocity). These single-parameter
methods, such as 'bright spot analysis, are very eIIective in some circumstances, but have a
limited range oI applicability.
AVO analysis has always had the potential to greatly improve and expand the applicability oI
petrophysical interpretation. This is because anomalously low V
p
/V
s
ratios (caused by
hydrocarbons) produce anomalous AVO responses. However, V
p
/V
s
ratios cannot be uniquely
inverted Irom AVO data alone. Thus, more intellectual eIIort is required to make AVO a valid
method Ior making inIerences concerning pore Iluid content.
This appendix details more oI the mathematics/rock physics that is touched upon in the body oI
the manual.
Topics covered in this appendix:
Where it all began
AVO empirical relationships
Where it is going...
!
Well Seismic Fusion Landmark
A-2 Where it all began Release 5000.0.0.1
Where it aII began
It all began around 1919 when proIessor Zoeppritz took Snell`s Law,
and applied it to the partitioning oI an incident seismic wave at a
reIlecting interIace in terms oI particle displacements and reIlection
coeIIicients.
Basic seismic reflection/transmission diagram
For the above diagram, Snell`s Law is shown below:

Snells Law
Zoeppritz required continuity oI displacement and stress across the
interIace as the seismic plane wave reIlects oII and transmits through a
welded contact between two elastic media. In other words, the media
on both sides oI the boundary cannot be ripped apart, and no slippage is
allowed along the interIace. These conditions are oIten reIerred to as
the kinematic boundary conditions. The stress across the boundary
must also be continuous, this is known as the dynamic boundary
condition. Zoeppritz assumed no absorption oI energy by the media.
And, like all elastic constants used in geophysics, Zoeppritz assumed
small deIormations where Hooke`s law holds and strain is proportional
to stress. This is a Iairly saIe assumption Ior low-amplitude seismic
waves. The Iull Zoeppritz equations are shown below:
V
p1
, V
s1
,
1
V
p2
, V
s2
,
2
A
o
= ncident P-wave
B
1
= Reflected S-wave
A
1
= Reflected P-wave
A
2
= Transmitted P-wave
B
2
= Transmitted S-wave

2
u
2
u
1
u
1

1
nterface
p
u
1
sin
V
P1
-------------
u
2
sin
V
P2
-------------

1
sin
V
S1
-------------

2
sin
V
S2
-------------
Landmark Well Seismic Fusion
Release 5000.0.0.1 Where it all began A-3
where
The full Zoepprit: Equations
For most interpreters, these equations are very complex and prone to
typos. That being said, these equations can be solved to give the
amplitudes oI the reIlected and transmitted P and S-waves given ,
V
p1
, V
s1
, , V
p2
, V
s2
, and . can be solved Ior as a Iunction oI
time and oIIset by iterative ray tracing through a P-wave interval
velocity Iield.
Although accurate, the Zoeppritz equations are too complex and,
thereIore, not very useIul Ior gaining insight into lithology and pore
Iluids by observing the shape oI the reIlection coeIIicient curve as a
Iunction oI oIIset or angle oI incidence. The Aki-Richards
approximation to the Zoeppritz equations provides more insight. This
approximation (Ior the reIlected P wave) is:
Note
The presentation oI the Aki-Richards approximation can take many Iorms, depending upon
what the author chooses to abbreviate and how the author chooses to expand the terms oI the
equation. This is also true with Shuey`s approximations shown on the Iollowing page. We have
chosen expansion in this documentation that we Ieel most readily reveals the inIormation
content contained in the amplitude behavior.
A
1
u
1
cos B
1

1
sin A
2
u
2
cos B
2

2
sin A
o
u
1
cos
A
1
u
1
sin B
1

1
cos A
2
u
2
sin B
2

2
cos A
o
u
1
sin
A
1
Z
1
2
1
cos B
1
W
1
2
1
sin A
2
Z
2
2
2
cos B
2
W
2
2
2
sin A
o
Z
1
2
1
cos
A
1

1
W
1
2u
1
sin B
1
W
1
2
1
cos A
2

2
W
2
2u
2
sin B
2
W
2
2
2
cos A
o

1
W
1
2u
1
sin

i
V
si
V
pi
-------- , Z
i

i
V
pi
, W
i

i
V
si
, i 1 2 ,
u
1

1

2
u
1
A
1
~ NI (1 - 4
|
o
---
\ .
| |
2
u
2
)
Ao
1 o ( )
2
-------------------- u
2

Ao
2o
------- u
2
tan 4
|
o
---
\ .
| |
2
u
2
sin
\ .
| |
sin sin
Well Seismic Fusion Landmark
A-4 Where it all began Release 5000.0.0.1
where
NI normal incident P-wave reIlection coeIIicient
average compressional velocity (V
p1
V
p2
)
average shear velocity (V
s1
V
s2
)
average density ( )
average Poisson's ratio ( )
average oI incident and transmission angles
lower medium's elastic property minus upper medium's
, Poisson's ratio
Here, it is assumed that the relative changes in the velocities and
density across the interIace are suIIiciently small to neglect second-
order terms and that does not approach the critical angle, , given
by 90 degrees in Snell`s Law, or
Shuey rewrote the Aki-Richards approximation in terms oI increasing
angle oI incidence:
A
1
~ near oIIsets
mid oIIsets
Iar oIIsets
u 0 ( )
1
2
-- -
Ao
o
-------
A

-------
\ .
| |

o
|

1

2
o o
1
o
2
u
Ao
o
1
1
2
---
V
p
V
s
------
\ .
| |
2

1
V
p
V
s
------
\ .
| |
2

---------------------------
u
1
u
c
u
2
u
c
V
p1
V
p2
---------
1
sin
1
2
---
Ao
o
-------
A

-------
\ .
| |
1
2
---
Ao
o
------- 4
|
o
---
\ .
| |
2
A|
|
------- 2
|
o
---
\ .
| |
2
A

-------
\ .
| |
u
2
sin
1
2
---
Ao
o
-------
\ .
| |
u
2
tan u
2
sin ( )
Landmark Well Seismic Fusion
Release 5000.0.0.1 Where it all began A-5
Shuevs equation or '3-term Shuev approximation above is what most
commercial AVO applications use Ior doing advanced AVO analysis. In
its simpliIied Iorm the 3-term approximation is represented by:
In this Iorm Shuey`s equation is quite understandable, and does not
require a computer to evaluate the magnitude oI the eIIect oI the
amplitude behavior. II your data are oIIset (i.e. angle oI incidence)
limited, the third term 'C can be removed and you are leIt with the
2-term straight line Shuev approximation.
A
1
( ) A B sin
2
or
Amplitude Intercept Gradient * sin
2
Amplitude |Iluid inIluence| |lithology inIluence (V
p
/V
s
)| * sin
2
Assuming V
s
/V
p
.5 and small angles, Shuey also helped explain
Koefoeds observation that reIlectivity is a Iunction oI normal incident
reIlectivity plus Poisson reIlectivity as below:
or
or
A
1
( ) A B sin
2
C (tan
2
- sin
2
) u u u u
Acoustic Impedance Poisson, V
s,
Rigidity
(Iluid eIIects)
Velocity Contrast
Near traces Far traces Very Iar traces
(25 - 45 degrees) (20 - 35 degrees) (0 - 20 degrees)
Porosity
Lithology (hard rock)
Fluid eIIect (soIt rock)
Fluid eIIect (hard rock)
Lithology (soIt rock)
Partial saturation (soIt rock)
Lithology (hard rock)
(can be used as density term
to discriminate Iizz gas)
u u
u
u
A
1
u
1
( )

2
o
2

1
o
1

2
o
2

1
o
1

-------------------------------
\ .
| |
u
1
2
cos
o
2
o
1

1
o
2
o
1

2
------------------
\ .
| |
2
----------------------------------- u
2
1
sin ~
A
1
u
1
( ) Normal-Incident ReIlectivity ( ) u
1
2
cos Poisson ReIlectivity ( ) u
2
1
sin ~
Well Seismic Fusion Landmark
A-6 Where it all began Release 5000.0.0.1
Koefoeds observation of reflectivitv contribution to amplitude
By assuming that V
s
/ V
p
(or equivalently .333) a Iurther
approximation can be written as:
The above relation is known as the Shuev-Hilterman approximation. It
clearly demonstrates that the elastic property most directly related to
the angular dependence oI the reIlection coeIIicient is Poisson's ratio,
. One oI the many advantages oI this simple relation Ior the P-wave
reIlection coeIIicient is the ability to easily determine the elastic
parameters Irom the shape oI the curve that describes the reIlection
coeIIicient as a Iunction oI oIIset (or angle oI incidence).
A
1
u
1
( ) A
NI
u
1
2
cos PR u
2
1
sin ~
.75
1.00
.25
30
o
60
o
90
o
0
o
.50
P
o
i
s
s
o
n

R
a
t
i
o
C
o
n
t
r
i
b
u
t
i
o
n
N
o
r
m
a
l
-

n
c
i
d
e
n
c
e
C
o
n
t
r
i
b
u
t
i
o
n
ncident Angle
A
(
u
)
o
A
1
NI u
2
2.25Ao u
2
sin cos
o
Landmark Well Seismic Fusion
Release 5000.0.0.1 Where it all began A-7
II you assume angles less than 30
o
, you can drop the cos
2
term and
rewrite the Shuey-Hilterman approximation in the Iorm y b mx:
The equation above shows amplitude to be a linear equation with
respect to . The gradient (or slope) oI the curve is given by m
(2.25 ) and the intercept is b (~NI).
Jarious P-wave reflection coefficients as a function of offset
One strategy Ior estimating the rock properties is to perIorm a linear
regression oI the amplitudes across a CMP gather at various times. This
will yield an estimate oI the normal incidence (NI) and the change in
Poisson's ratio (delta-sigma, ). Inversion oI the NI trace represents
acoustic impedance and the inversion oI the delta-sigma trace
represents Poisson's ratio.
u
A
1
b m u
2
sin ~
u
2
sin
Ao
.2
-.2
0.0 .25
A
1
Ao
Well Seismic Fusion Landmark
A-8 Where it all began Release 5000.0.0.1
The graphic below shows how the amplitudes oI CMP (measured along
a time, T
o
) are crossplotted to Iigure out NI and .
Linear regression of seismic amplitudes at time, T
o
Ao
s
I
o
p
e

~

b

A
o
A
1
Offset
T
i
m
e
T
o
Common Midpoint Gather (CMP)
Intercept
~ NI cos
2
u
With assumptions like V
p
/V
s

2, the 'b in the slope
'b tends toward 2.25. Ao
where:
Poisson`s
Ratio

o
Ao o
2
o
1

Landmark Well Seismic Fusion
Release 5000.0.0.1 Where it all began A-9
The chart below shows some oI the major events in AVO.
AJO Eras (adapted from Hilterman 2001)
Year Person Event
1800`s Poisson Theory - Poisson`s Ratio
1899 Knott Theory - Amplitude vs incident
angle
1919 Zoeppritz Theory - Amplitude vs incident
angle... Zoeppritz equations
1951 Gassmann Theory - Petrophysical link to
seismic data
1955 KoeIoed Application - Poisson`s Ratio
Irom RC( )
1961 BortIeld Theory - Linear approximation
equation
1960`s Oil Companies Bright Spot Discoveries
1976 Rosa Application - RC( ) elastic
inversion
1980 Aki and Richards Theory - 3-Term approximation
1982 Ostrander VeriIication oI AVO
1985 Shuey Theory - 2-Term and 3-Term
approximation
198? Castagna Theory - Mud Rock Line
198? Smith and Gidlow Application - Weighted Stacks
1990`s Vendors Application - AVO Crossplots
1997 Reger Theory - Anisotropic AVO
equation
2003 Landmark Graphics Well Seismic Fusion soItware
u
u
Well Seismic Fusion Landmark
A-10 AVO empirical relationships Release 5000.0.0.1
AVO empiricaI reIationships
Fluid substitution and relationships
Biot-Gassmann
The Well Seismic Fusion soItware uses the Gassmann equations to
calibrate the petrophysical properties to the seismic velocities,
especially Ior pore-Iluid predictions. V
p
and V
s
Ior a statistical
homogenous and isotropic rock are related to the bulk modulus and
shear modulus by the Biot-Gassmann equations:
The Gassmann model relates the bulk and shear moduli oI a porous
rock to the rock properties porosity, bulk modulus oI the solid and Iluid
phases, and the bulk and shear moduli oI the skeleton oI the rock:
where k
skel
and
skel
are the skeleton bulk and shear moduli, k
s
is the
bulk modulus oI the solid component, k
I
is the bulk modulus oI the
Iluid component, and is the porosity. The density is a volume
weighted average oI the solid and Iluid densities and :
In cases where the Iluid phase is a homogenous mix oI two phases
(hydrocarbons and brine), the eIIective value oI the Iluid bulk modulus
is given by Woods Equation (Wood, 1930):
V
p
k
4
3
---

--------------------- V
s
,

---
k k
skel
1
k
skel
k
s
--------------
\ .
| |
2
|
k
I
-----
1 | ( )
k
s
----------------
k
skel
k
2
s
--------------
------------------------------------------------- ,
skel

s

I

s
1 | ( )
I
|
1
k
I
-----
S
w
k
w
--------
1 S
w
( )
k
hc
---------------------
Landmark Well Seismic Fusion
Release 5000.0.0.1 AVO empirical relationships A-11
where S
w
is the water saturation (as a Iraction oI pore volume) and k
w

and k
hc
are the bulk moduli Ior the brine and hydrocarbon phases,
respectively. The Iluid phase density is given by:
where and are the densities oI the brine and the hydrocarbon
phases, respectively.
Other veIocity reIationships
Mudrock Line:
V
s
(km /sec) .8621 V
p
- 1.1724 (Castagna et al)
A mudrock is a brine-Iilled clastic silicate (sand-shale). This
equation is also appropriate to estimate the shear velocity oI shale.
Sandstone:
V
s
(km /sec) .8042 V
p
- .8559 (Castagna et al)
V
s
(km /sec) .7936 V
p
- .7868 (Han et al)
V
gas
(km /sec) -.070 V
2
brine
1.67 V
brine
- 1.74 (Biot-Gassmann)
Shale:
V
s
(km /sec) .7700 V
p
- .8674 (Castagna et al)
Limestone:
V
s
(km /sec) 1.017 V
p
- 0.055V
P
2
- 1.030 (Castagna et al)
Warning
Be very, very cautious when using any empirically derived equations. Empirically
derived equations have many assumptions and are usually biassed by the scientists
Iavorite geographical area.
For example, is the velocity oI a dry rock in a lab at very high Irequencies the same
as the velocity oI the same rock, wet at low Irequencies at Iormation depth? Do all
sands behave like GulI oI Mexico sands?

I

W
S
W

hc
1 S
W
( )

W

hc
Well Seismic Fusion Landmark
A-12 AVO empirical relationships Release 5000.0.0.1
Dolomites:
V
s
(km /sec) .5832 V
p
- .07776 (Castagna et al)
Empirical relations for densities (g/cm
3
)
1.741 V
p
.25
(km /sec) (Gardner, Gardner, Gregory 1974)
.23 V
p
.25
(It /sec) (Gardner, Gardner, Gregory 1974)
In 1993 Castagna extended Gardner`s work by developing velocity-
density transIorms Ior speciIic lithologies.
Notice how anhydrite is very heavy Ior its velocity, while salt is too
light Ior its velocity.
Warning
We cannot over-emphasize the importance oI recalibrating the V
p
to V
s
coeIIicients
Ior your local area when a measurement oI V
s
is available.
Lithology Density transform gm/cm
3
and ft/s
Sand
Shale
Limestone
Dolomite
Anhydrite
Salt relatively constant velocity and density

b
0.200V
P
0.261

b
0.204V
P
0.265

b
0.243V
P
0.225

b
0.226V
P
0.243

b
0.600V
P
0.160

Landmark Well Seismic Fusion


Release 5000.0.0.1 AVO empirical relationships A-13
Intercept and gradient relationship with near and far offset stacks
In many cases you will not have prestack data. However, iI you have
near and Iar oIIset stacks, you can still use much oI the power oI the
Well Seismic Fusion soItware. That is, you can turn near and Iar
amplitude maps into intercept and gradient maps. You can also turn
near and Iar oIIset stacks into intercept and gradient stacks.
Many interpreters believe this is the preIerred method to generate
intercept and gradient volumes because stacking is so powerIul at
removing noise.
There are Iour inputs:
Near amplitude map
Average angle oI near stack (e.g. 5 degrees)
Far amplitude map
Average angle oI Iar stack (e.g. 35 degrees)
Gradient AAmp / ASin
2
(u)
A
m
p
Sin
2
( )
Nears
(
n
)
Fars
(
f
)
Nears
( A
n
)
Fars
( A
f
)
-
0
+
0.4 0
Gradient
Intercept
A
m
p
Sin
2
( )
Nears
(
n
)
Fars
(
f
)
Nears
( A
n
)
Fars
( A
f
)
-
0
+
0.4 0
Gradient
Intercept
f
Gradient
A
n
A
I

u
2
n
sin u
2
I
sin
------------------------------------
Gradient A
n
1
u
2
n
sin u
2
I
sin
------------------------------------
\ .
|
| |
A
I
1
u
2
n
sin u
2
I
sin
------------------------------------
\ .
|
| |

Well Seismic Fusion Landmark
A-14 AVO empirical relationships Release 5000.0.0.1
Intercept A
n
- |Gradient X Sin
2
(u
n
)|
Once you have intercept and gradient volumes, you can rotate the data
to Iluid and lithology volumes as Iollows:
Rotation oI A (intercept) and B (gradient) axes counterclockwise by
degrees moves point X (a, b) to X (a
1
, b
1
).
a
1
a Cos() b Sin()
b
1
b Cos() - a Sin()
Intercept A
n
u
2
n
A
n
A
I

u
2
n
sin u
2
I
sin
------------------------------------
\ .
|
| |
sin
Intercept A
n
u
2
I
sin
u
2
n
sin u
2
I
sin
------------------------------------
\ .
|
| |
A
I
u
2
n
sin
u
2
n
sin u
2
I
sin
------------------------------------
\ .
|
| |

Landmark Well Seismic Fusion
Release 5000.0.0.1 Where it is going... A-15
Where it is going...
Elastic impedance
A normal incident (near oIIset or intercept) stack relates to changes in
acoustic impedance (AI, density * velocity, or ). These changes can
be tied to well logs using synthetic seismograms. However, Ior oIIset
seismic data, you must use 'Elastic Impedance to tie the well logs to
the seismic. That is, Elastic Impedance (EI) is a generalization oI
acoustic impedance Ior variable incidence angle. EI is a Iunction oI the
P-wave velocity, S-wave velocity, density and incidence angle. Like
Shuey`s approximation, EI is an approximation derived Irom a
linearization oI the Zoeppritz equations.
In the Well Seismic Fusion soItware, you use elastic impedance to
transIorm the P-wave and S-wave impedances so that they better
diIIerentiate between the Iluid and the rock matrix oI the porous
medium. This implies that the elastic impedance equation delivers
impedance values proportional to the bulk modulus and Lame`s
constant. Extended Elastic Impedance (EEI) 'extends the elastic
impedance to 90
o
.
To understand EI, you Iirst must understand reIlectivity. Reflectivitv is
the ratio oI the particle displacement amplitude oI a reIlected wave to
the displacement amplitude oI the incident wave. The relationship is
obtained by solving boundary condition equations that express the
continuity oI displacement and stress at the boundary. Mathematically
reIlectivity is given by:
Setting this equal to the expression Ior A
1
given by the Shuey equation
leads to:

where .
Note, Ior 90
o
, K 0.25, EI , which can easily be transIormed
into Poisson's ratio.
o
R u ( )
I t
i
( ) I t
i
1
( )
I t
i
( ) I t
i
1
( )
------------------------------------
1
2
---
AEI
EI
----------
1
2
---A EI ( ) ln
EI u ( ) o o
u
2
tan
|
8K u
2
sin

1 4K u
2
sin ( )
( )
K
|
2
o
2
------
u
o
2
|
2
------
Well Seismic Fusion Landmark
A-16 Where it is going... Release 5000.0.0.1
An alternative Iorm, without the term Iactored out:
The above EI equation will have variable dimensionality with angle oI
incidence. This is undesirable and can be overcome by normalizing the
equations with constants , , and .
where
This new Normali:ed Elastic Impedance Iunction will always predict
the AI log Ior 0 without any scaling (AI standard acoustic
impedance). This also allows Ior a direct comparison between elastic
impedance values across a range oI angles in a manner that is not
available in the un-normalized EI equation.
In practice, lithology and Iluid changes are more dramatic when plotted
in the elastic impedance domain then when plotted in the acoustic
impedance domain.
o
EI u ( ) o
1 u
2
tan ( )
|
8K u
2
sin

1 4K u
2
sin ( )

o
o
|
o

o
EI u ( ) o
o

o
o
o
o
------
\ .
| |
a
|
|
o
-----
\ .
| |
b

o
-----
\ .
| |
c
)
`

a 1 u
2
sin ( ) b , 8K u
2
sin ( ) c , 1 4K u
2
sin ( )
u o
Landmark Well Seismic Fusion
Release 5000.0.0.1 Where it is going... A-17
Extended elastic impedance
Extended Elastic Impedance (EEI) 'extends the normalized elastic
impedance to 90
o
or any combination oI intercept and gradient. EEI
replaces with and normalizes EI to one by dividing by
. With these substitutions, EEI ranges Irom a value oI A at
to B at , where A is the acoustic impedance (AI)
and B is the gradient impedance (GI).
where
EEI can be used to obtain the optimum constant angle projection oI a
seismic section in order to provide maximum discrimination between
Iluids or lithologies. The chart below maps between Theta and Chi.
0 0
29 13
37 20
57 35
90 45
no real equivalent 70
no real equivalent 90
no real equivalent -45
no real equivalent -51
Note
Keep in mind that the EEI Iunction is deIined as a Iunction oI the angle , not the reIlection
angle . The bulk modulus and Lame`s constant tend to lie within an area oI EEI space with
values oI Irom about 10
o
to 30
o
, while the shear modulus lies within a range oI Irom -30
o

to -90
o
. These are thereIore likely to be good starting points to look Ior optimum Iluid and
lithology impedance Iunctions.
u
2
sin _ tan
_ cos
_ 0
o
_ 90
o

EEI _ ( ) o
o

o
o
o
o
------
\ .
| |
p
|
|
o
-----
\ .
| |
q

o
-----
\ .
| |
r
)
`

p _ cos _ sin ( ) q , 8K _ sin ( ) r , _ cos 4K _ sin ( )


u _
_
u
_ _
Well Seismic Fusion Landmark
A-18 Where it is going... Release 5000.0.0.1
The EEI equivalent to is oI course acoustic impedance and at
EEI will have a reIlectivity corresponding to gradient
impedance. According to Whitcombe (2002), the EEI Iunction can be
used to represent nearly any elastic parameter by simply changing the
angle . The optimal value Ior is Iound by correlating the EEI
Iunction Irom to with the log curve you are
trying to represent with EEI. In Whitcombe`s paper (2002) he Iound the
Iollowing EEI relationships Ior a speciIic reservoir:
acoustic impedance
bulk modulus
Lame`s parameter
water saturation log
V
p
/V
s
ratio
No equivalent real values beyond
gamma ray log
gradient impedance
shear impedance
shear modulus
*These relationships assume the (EEI) 'K value is 0.25 and the 'I
value is 0.8.
_ 0
o

_ 90
o

_ _
_ 90
o
_ 90
o

o EEI _ 0
o
( ) EI u 0
o
( ) ~
k EEI _ 12.4
o
( ) EI u 28
o
( ) ~
EEI _ 19.8
o
( ) EI u 37
o
( ) ~
S
w
EEI _ 35
o
( ) EI u 57
o
( ) ~
o
|
---
\ .
| |
1.414
EEI _ 45
o
( ) EI u 90
o
( ) ~
u _ 45
o

GR EEI _ 70
o
( )
GI EEI _ 90
o
( )
| ( )
1.414
EEI _ -45
o
( )
EEI _ -51.3
o
( )
Landmark Well Seismic Fusion
Release 5000.0.0.1 Where it is going... A-19
Graphically, the elastic moduli project in AI/GI space with angles as
below.
EEI profections for some well known moduli (Whitcombe 2002)
Please be aware that the actual projection angle will depend on the
details oI the , , and relationships. The above relationships
assume the (EEI) 'K value is 0.25 and the 'I value is 0.8.
These relationships are taken Irom Whitcombe (Geophysics, Vol. 67
2002), 'Various authors have noted the diIIiculty oI extracting the C
term Irom real seismic data. Shuey (1985) examined the ratio CA and
noted that this parameter tended to lie between 0 and 1. We deIine CA
as I. The value oI I 0.8 describes rocks that Iollow Gardner et al.`s
(1974) relationship. Rather than determine C directly, we replace it
with I A and select an I value appropriate to the rock properties in the
area.
_
Acoustic mpedance
G
r
a
d
i
e
n
t

m
p
e
d
a
n
c
e
SI -45
o
k 12.4
o
AI 0
o
o/| 45
o
19.8
o
-51.3
o
S
w
35
o
gamma ray 70
o
GI 90
o
o | o
Well Seismic Fusion Landmark
A-20 Where it is going... Release 5000.0.0.1
When parameterizing elastic impedance within the Well Seismic
Fusion soItware you should use the average oI the velocities and
densities in the sand combined with the average velocities and densities
in the shale. Do not use the average over the entire log curve because
this may introduce a sampling bias iI there is a much greater amount oI
shale than sand in the well.
LMR analysis
Intercept, gradient and impedances are mixed combinations oI bulk
modulus, shear modulus, and density. With lambda-rho () and mu-
rho () analysis (Goodway 1997), you can actually separate
(incompressibility) and (rigidity). Many interpreters Iind it easier to
understand the connection oI reservoir properties to Iundamental rock
properties (such as compressibility, rigidity, and density) than it is to
understand their connection to traditional seismic attributes, like
amplitude and velocity. Geophysicists think oI displacement and stress
being derived Irom a plane wave solution oI the acoustic wave
equation:
Many purists consider density and the moduli to be the only
Iundamental properties oI a rock because the underlying physics in the
wave equation:
does not involve seismic velocities, but instead the ratio oI density ()
to modulus (M). The purists do not consider velocity to be a
Iundamental property, because the velocity is a Iunction oI the moduli.
The mathematics oI the Lame moduli are shown below:
lambda Lame`s lambda constant incompressibility axial
resistance to axial compression
Note
A modulus is simply the coeIIicient oI a certain kind oI strain produced by a certain type oI
stress (i.e. stress modulus * strain).
u Ae
ie t
x
V
----
\ .
| |

d
2
u
dx
2
--------

M
-----
d
2
u
dt
2
--------
\ .
|
| |

Landmark Well Seismic Fusion


Release 5000.0.0.1 Where it is going... A-21
mu rigidity resistance to shearing
Lame Impedance
Lame Impedance
Compressional ReIlectivity
Shear ReIlectivity
AVO - Fatti Equation
where background scalar
Compressional Velocity V
p
2

----------------
k
4
3
---

---------------------
Shear Velocity V
s

---
Acoustic Impedance I
p
V
p
2
Shear Impedance I
s
V
s

I
p
2
2I
s
2

I
s
2

R
p
AI
p
2I
p
--------
R
s
AI
s
2I
s
--------
A u ( ) 1 u
2
tan ( )R
p
8,
2
u
2
R
s
sin ~
,
V
s
V
p
------
Well Seismic Fusion Landmark
A-22 Where it is going... Release 5000.0.0.1
DiIIerent types oI lithologies tend to separate along orthogonal
boundaries in a crossplot oI versus . Consequently, these two
parameters may be considered independent and can be interpreted
separately. Furthermore, the separation is oI much greater magnitude
than the separation oI V
p
/V
s
and Poisson`s ratio.
LMR Space
Some interpreters skip the crossplotting oI versus because
various Iluids and lithologies separate along roughly orthogonal
boundaries. This means the Iluids and lithologies are nearly
independent and, thereIore, can be interpreted separately.
60
Gas

-10
-10 60
P
o
r
o
s
i
t
y
S
a
n
d
S
h
a
l
e
C
a
r
b
o
n
a
t
e
s
Landmark Well Seismic Fusion
Release 5000.0.0.1 Where it is going... A-23
This is a signiIicant improvement over previous crossplotting
techniques Ior identiIying gas accumulations, where both parameters
had to be interpreted simultaneously. For example, zones oI low
incompressibility () suggest the presence oI gas or coal. When you
add rigidity () inIormation, gas sands can be more-clearly
distinguished Irom coals because sand has high rigidity and coal has
low rigidity. A chart oI this methodology is shown below.
Chart for interpretation of and sections based on the
crossplot in Goodwav (1997).
Both rigidity and Lame`s modulus can be obtained Irom prestack
seismic data using the method oI Goodway (1997). Once you have
these volumes, you can interpret Ior potential sand zones; then,
interpret the potential sand zones on the section Ior gas. This type oI
analysis enables gas-sand anomalies to be interpreted with much less
ambiguity. That is, you can separate them Irom shale, coal, and tight
sands.
Warning
Depth and pressure aIIect the moduli.
high
coal

high low
low
gas-sand
shale
sand
cemented-sand
carbonates
Well Seismic Fusion Landmark
A-24 Where it is going... Release 5000.0.0.1
Typically a Lambda-rho stack gives inIormation about pore Iluid. This
stack distinguishes sands, gas-sands, shales, coals, and carbonates.
Sands containing gas are more compressive than wet sands, so they
have low incompressibility values. Sands containing gas are also less
dense than sands containing water. These two low values enhance each
other to produce very low values oI lambda ( ). In many regions
analysis gives improved sensitivity over traditional AVO - Poisson
type analysis (Goodway et al., 1997).
Incompressibilitv based on lithologv (Goodwav 1997)
The Lambda/mu stack is the most rock property sensitive moduli. This
stack is good Ior distinguishing between gas sands and shales.
The Mu-rho stack gives inIormation about lithology, such as
distinguishing rock matrix type: sand, shales, coals, and carbonates.
The Mu-rho stack is unaIIected by pore Iluid.
Many interpreters preIer this method over elastic impedance because
they Ieel R
p
and R
s
are more physical/understandable parameters than
elastic impedance. Critics oI lambda/mu/rho argue that in noisy data
gives inIerior results because it squares the data. For example, iI
you square the impedance, you also square the noise in the impedance.
also has some problems because oI its isotropic simpliIication oI
the wave equation.
Proponents oI claim that it is not about transIorm, it is about the
mapping oI porosity and Iluid, and does a good job in many Iields.

coals
gas-sands
shales
wet-sands
carbonates
very low incompressibility
high incompressibility
shale
gas-sands
high ratio
low ratio
coal
shale
least rigid
sands
carbonates
more rigid

Landmark Well Seismic Fusion


Release 5000.0.0.1 Where it is going... A-25
Azimuthal AVO Analysis in Anisotropic Media
Alas, the time has come Ior geophysicists to acknowledge that the earth
is not isotropic. All seismic algorithms must begin accounting Ior
anisotropy. For example, migrations and velocity analysis should
incorporate anisotropy parameters.
(Thomsen 2002)
V
PNMO
normal moveout velocity Ior P-waves
V
P0
P-velocity, a rock property
near-vertical P-wave polar anisotropy parameter
Synthetics Irom deviated wells need to incorporate anisotropy
parameters.
Anisotropv effect on compressional plane wave (Thomsen 2002)
V
P
P-velocity, a rock property
Polar angle oI incidence oI a wave
OI course AVO analysis needs to account Ior anisotropy.
JTI reflection coefficients (Rger 2002)
Finally, AVO analysis must account Ior HTI. Please reIer to Rger,
(2002) Ior more inIormation on HTI.
V
PNMO
V
P0
1 o ( )
o
V
p
u ( ) V
p0
1 o u
2
u
2
cos sin c u
4
sin | | ~
u
R
P
VTI
u ( )
1
2
---
AZ
Z
-------
1
2
---
AV
P0
V
P0
-------------
2V
S0
V
P0
------------
\ .
|
| |
2
AG
G
-------- Ao
\ .
|
| |
u
2
sin
1
2
-- -
AV
P0
V
P0
------------- Ac
\ .
| |
u
2
u
2
tan sin
R
SV
VTI
u ( )
1
2
------
AZ
S
Z
S
----------
7
2
-- -
AV
S0
V
S0
------------- 2
A

-------
1
2
-- -
V
P0
V
S0
---------
\ .
|
| |
2
Ac Ao ( )
\ .
|
| |
u
2
sin
1
2
---
AV
S0
V
S0
------------- u
2
u
2
tan sin
R
SH
VTI
u ( )
1
2
---
AZ
S
Z
S
----------
1
2
---
AV
S0
V
S0
------------- A
\ .
| |
u
2
tan
Well Seismic Fusion Landmark
A-26 What Data does Well Seismic Fusion software use in AVO Analysis? Release 5000.0.0.1
What Data does WeII Seismic Fusion software use in
AVO AnaIysis?
Data the Well Seismic Fusion software reads
Data Type Data Read From Data Format
Log Curves OpenWorks database OpenWorks
Synthetics (oIIset and
normal incidence)
OpenWorks database OpenWorks
OIIset and Corridor Stack
VSPs
OpenWorks database OpenWorks
3D VSPs dir.dat .3dv, .bri, .cmp
Velocity Data -
TD Curves
OpenWorks database OpenWorks
Velocity Data - ProMAX
Tables
OWPROJDATA~/OWProject~/FusionData/
tabledata *
ProMAX
Velocity Data - TDQ &
DepthTeam Express
Models
OWPROJDATA~/OWProject~/I3DV* or
OWPROJDATA~/OWProject~/
VDSMODEL*
I3DV TDQ
VDS DepthTeam Express
Well Picks OpenWorks database OpenWorks
Wavelets OpenWorks database OpenWorks
Stacked Seismic dir.dat, GeoProbe shortcut, or GeoProbe shared
memory
.3dv, .bri, .cmp, .vol, or .shm
Prestack Seismic Pointed to via OpenWorks, typically in dir.dat/
Iusionseismic
ProMAX, SEG-Y or
JavaSeis
Prestack smartMap and
smartFramework Iiles
File next to Prestack data (iI writable), otherwise in
dir.dat/Iusionseismic
smartMap Fusion/binary
smartFramework XML
Crossplots OWPROJDATA~/OWproject~/
FusionData/crossplot *
Fusion/XML
Polygons OWPROJDATA~/OWproject~/
FusionData/crossplot *
Fusion/XML
Prestack Horizons dir.dat/Iusionhorizons ProMAX
Stacked Horizons Pointed to via dir.dat Iile SeisWorks
Sessions and Colormaps DeIault save location is |UserHome|/Landmark/
DecisionSpace/Sessions, Colorbars are in
|UserHome|/.lgcpalettes
Fusion/XML
Landmark Well Seismic Fusion
Release 5000.0.0.1 What Data does Well Seismic Fusion software use in AVO Analysis? A-27
Parameter Files:
(Fluid Substitution)
(Porosity Substitution)
(Multi-Well setup)
(Serialized Properties)
OWPROJDATA~/OWProject~/Fusion/
userid*
(Can browse to any location Ior Iluid or porosity
parameter Iiles.)
Fusion/XML
Error Logs |UserHome|/run/DecisionSpace5000.0#.err ASCII text
Scratch Data OWPROJDATA~/OWproject~/
FusionData/scratch*
ProMAX
Data Type Data Read From Data Format
Well Seismic Fusion Landmark
A-28 What Data does Well Seismic Fusion software use in AVO Analysis? Release 5000.0.0.1
*To Iind the location oI OWPROJDATA, view $OWHOME/
conI/owdir.dat, or run the command 'lgcgetowdir.
Landmark Well Seismic Fusion
Release 5000.0.0.1 What Data does Well Seismic Fusion software use in AVO Analysis? A-29
Data the Well Seismic Fusion software creates
Data Type Data Saved To Data Format
Log Curves Creates in memory
Saves to OpenWorks database
Exports to LogM and LAS
OpenWorks, LogM, or LAS
Synthetics (oIIset and
normal incidence)
Saves to OpenWorks database OpenWorks
Velocity Data -
TD Curves
Creates in memory
Saves to OpenWorks
OpenWorks
Wavelets Creates in memory
Saves to OpenWorks
OpenWorks
Stacked Seismic Creates directly to location speciIied in dir.dat,
GeoProbe shortcut, or GeoProbe shared memory
Can export to SEG-Y
.3dv, .bri, .cmp, .vol, .shm,
or .sgy
Prestack Seismic Creates directly to dir.dat/Iusionseismic
Note: II the Iusionseismic directory is not
speciIied in the dir.dat, then prestack seismic data
created in a Well Seismic Fusion session will be
written to the OTHERFILES Iile system.
Can export to SEG-Y
8-bit, 16-bit, or 32-bit
ProMAX
32-bit SEG-Y
8-bit, 16-bit, 32-bit Iloating,
or SeisPEG compressed
Prestack smartMap and
smartFramework Iiles
File next to Prestack data (iI writable), otherwise to
dir.dat/Iusionseismic
smartMap Fusion/binary
smartFramework XML
Crossplots Creates in memory
Saves to OWPROJDATA~/OWproject~/
FusionData/crossplot
Fusion/XML
Polygons Creates in memory
Saves to OWPROJDATA~/OWproject~/
FusionData/crossplot
Fusion/XML
Stacked Horizons Creates directly to SWProj pointed to via dir.dat SeisWorks
Prestack Horizons Creates directly to dir.dat/Iusionhorizon.
Note: II the Iusionhorizon directory is not speciIied
in the dir.dat, then prestack horizons created in a
Well Seismic Fusion session will be written to the
OTHERFILES Iile system.
ProMAX
Sessions and Colormaps DeIault save location is |UserHome|/Landmark/
DecisionSpace/Sessions, Colorbars are in
|UserHome|/.lgcpalettes
(Can browse and save to any location.)
Fusion/XML
Well Seismic Fusion Landmark
A-30 What Data does Well Seismic Fusion software use in AVO Analysis? Release 5000.0.0.1
From an input/output perspective, here is the data Ilow oI Fusion.
Parameter Files:
(Fluid Substitution)
(Porosity Substitution)
(Multi-Well setup)
(Serialized Properties)
DeIault save location is
OWPROJDATA~/OWproject~/Fusion/
userid
(Can browse and save Iluid and porosity Iiles to any
location.)
Fusion/XML
Error Logs |UserHome|/run/DecisionSpace5000.0#.err ASCII text
Scratch Data OWPROJDATA~/OWproject~/
FusionData/tabledata/scratch
ProMAX
Data Type Data Saved To Data Format
WeII Seismic Fusion
AVO WorkIlows
Iog
curves
& picks
stacked
seismic
prestack
seismic
horizons
3D veIocity
modeIs
T-D tabIes
waveIets
1D & 2D synthetics
offset VSPs
crosspIots
poIygons
sessions
coIormaps
parameters
Iog
curves
horizons
crosspIots
poIygons
sessions
coIormaps
parameters
.smartMaps
.ScannedData
waveIets
1D & 2D synthetics
stacked
seismic
prestack
seismic
Seisworks
GeoProbe
SeisWorks
Seisworks
GeoProbe
OpenWorks
OpenWorks
Fusion
T-D
TabIes
OpenWorks
Fusion
Fusion
ProMAX
SEG-Y
OpenWorks
ProMAX
SeisWorks
OpenWorks
OpenWorks
OpenWorks
Landmark Seismic Velocity and Wavelet Background B-1
!""#$%&'
Seismic Velocity and Wavelet
Background
This appendix details more oI the seismic velocity background that is touched upon in the body oI
the manual.
Topics covered in this appendix:
Seismic velocity attributes
Seismic velocity accuracy
Seismic wavelet extraction
(
Well Seismic Fusion Landmark
B-2 Seismic velocity background Release 5000.0.0.1
Seismic veIocity background
Seismic velocity attributes
The initial seismic velocity derived Irom surIace acquisition is the
'NMO (normal moveout) or 'stacking velocity. NMO is the
hyperbolic correction made to seismic traces to make them appear as iI
they were recorded at zero oIIset. The NMO correction is displayed
below. Once all the trace times are corrected to 'zero-oIIset, the traces
are 'stacked together to get a more robust statistical solution. The
NMO velocities are initially used to create the unmigrated or stacked
image. Dix (1954) showed how interval velocities might be determined
Irom NMO velocities.
Landmark Well Seismic Fusion
Release 5000.0.0.1 Seismic velocity background B-3
FieId Recording Picture
*
Sounds
Like OiI
to Me!
Raw Seismic DispIay
NMO = time
to move
refIector up
to time t0
t0
t
i
m
e
Offset=0
Offset
Well Seismic Fusion Landmark
B-4 Seismic velocity background Release 5000.0.0.1
The Iollowing diagram will help deIine the parameters Ior calculating
geophysical velocities.
V
i
Layer Interval Velocity
Z
i
Depth to base oI Layer i
AZ
i
Layer Thickness
T
i
One Way travel time Irom surIace to base oI layer
AT
i
One Way Layer transit times
Surface
Z1, T1
Z2, T2
V1
V2
AT1
AT2
AZ1
AZ2
Z3, T3
V3
AT3 AZ3
Z0, T0
I Iike
Iayercake
geoIogy!
Landmark Well Seismic Fusion
Release 5000.0.0.1 Seismic velocity background B-5
Based on the previous deIinitions, the Iollowing geophysical velocities
can be deIined:
The NMO (V
nmo
) or Stacking Velocity can also be deIined:

IntervalVelocity
AZ
i
AT
i
--------- V
i

RMSVelocity
EAT
i
V
i
2

EAT
i
------------------------- V
rms
i

AverageVelocity
EAZ
i
EAT
i
-------------
EAT
i
V
i

EAT
i
------------------------ V
ave

T
x
2
T
0
2
X
2
V
nmo
2
-------------------
T0
Tx
x
ReIlector
Well Seismic Fusion Landmark
B-6 Seismic velocity background Release 5000.0.0.1
V
nmo
is the velocity which, when substituted into the normal moveout
equation, best Iits normal moveout times to a hyperbola. See the Iigure
below.
SIow VeIocity
hyperboIa
Fast VeIocity
hyperboIa
X
X
X
X
IntervaI
VeIocity
VeIocity SembIance
DispIay with Picks
Landmark Well Seismic Fusion
Release 5000.0.0.1 Seismic velocity background B-7
Applying the NMO velocities Ilattens the seismic data to make it
appear as iI all the traces were recorded at zero oIIset.
For a single Horizontal reIlector:
Data FIattened
on NMO curves
V
int
V
ave
V
rms
V
nmo

Well Seismic Fusion Landmark
B-8 Seismic velocity background Release 5000.0.0.1
When analyzing velocities, you may receive a Iile with stacking
velocity Iunctions in it Irom the people who processed the data.
DepthTeam Express uses the Dix equation to convert stacking
velocities to interval velocities. BeIore executing the conversion Irom
V
nmo
to V
int
you need to be aware where the assumptions about
stacking velocities break down.
For a single dipping reIlector with interval velocity V
int
:
As shown above, the stacking velocity is not equal to the true rock
velocity. To compensate Ior the cosine component oI V
nmo
a process
called Dip Moveout (DMO) is applied to the data. II you have stacking
velocities computed aIter the application oI DMO, these velocities will
better reIlect the real rock velocities.
Another potential pitIall is pictured below:
This situation cannot be resolved without DMO*. In seismic
processing, only one velocity can be applied.
* Special thanks to Dave Hale Ior his award winning work on DMO.
u
V
int
V
int
V
ave
V
rms
V
nmo
u cos
V5000It/s
CMP
T0 for Flat event = T0 for dipping event
V
nmo
for Flat Event = 5000 ft/s
60
o
V
nmo
for dipping event = (5000 ft/s) / (cos 60
o
) = 10000 ft/s
T0
T
0
Landmark Well Seismic Fusion
Release 5000.0.0.1 Seismic velocity background B-9
For multiple dipping reIlectors accurate velocity calculations get much
more complicated.
In complex geology, NMO tends to be hyperbolic over short oIIsets.
However, over large oIIsets, reIlections on CMP gathers may not be
hyperbolic.
Another seismic velocity problem interpreters need to be aware oI is
diagramed on the next page.
What's
going on
down there?
Well Seismic Fusion Landmark
B-10 Seismic velocity background Release 5000.0.0.1
NMO/stacking velocities are oIten picked in such a way that they give
unreasonable interval velocities. Consider the velocity picks below.
Stacking velocities causing unreasonable interval velocitv
approximations
Landmark Well Seismic Fusion
Release 5000.0.0.1 Seismic velocity background B-11
The CMP gather looks like the display on page B-7, but the semblance
display has more detailed picks. The additional picks have added more
detail to the interval velocities. Although the V
nmo
velocities look
reasonable, the interval velocities calculated via the Dix equation are
unreasonable.
Because the Dix equation is so sensitive to changes in the stacking
velocities, we recommend that you smooth your stacking velocities
beIore converting them to interval velocities. Multiples and noisy
seismic gathers also contribute to errors in the stacking velocities,
which cause additional errors in the Dix interval velocity calculation.
One oI the beauties oI Landmark`s DepthTeam Express is that you can
visually QC your interval velocities beIore you run the depth
conversion.
The Iollowing table deIines the meaning oI 'Seismic Stacking
Velocity (V
s
) when reIerenced to the CMP domain:
Seismic Stacking VeIocity
UnIortunately, non-hyperbolic moveout is the case Ior many geologic
settings.
Model Travel Time
Equation
Velocity Interpretation
Single Flat Layer Hyperbolic Formation or Interval Velocity
V
s
V
int
Single Dipping Layer Hyperbolic V
s
V
int
/
Flat Multi-Layered Earth Hyperbolic V
s
V
RMS
root-mean-square
velocity
UniIormly Dipping Layered Earth Hyperbolic V
s
V
RMS
/
Arbitrarily Dipping Layered Earth Hyperbolic V
s
I(V
k
,
k
, d
k
)
Ray Tracing Problem
Layered Earth with Arbitrary Curved
InterIaces
Non-Hyperbolic V
s
? Seismic Velocity no
longer a useIul concept.
V
i
T
i
V
2
rms
i
T
i 1
V
2
rms
i 1

T
i
T
i 1

--------------------------------------------------------------------------
u cos
u cos
u
Well Seismic Fusion Landmark
B-12 Seismic velocity background Release 5000.0.0.1
Non-HyperboIic Moveout
A major deIiciency oI the NMO equation is that it assumes straight ray
paths with no lateral velocity variation. This is one, among many, oI the
reasons Ior pre-stack depth migration. You should also be aware that
the amount oI NMO to move 't to 't
o
increases with time and thus
magniIies the NMO velocity error with time. A poor signal-to-noise
ratio also increases the seismic velocity error.
VeIocity AnomaIy
*
Who is
making aII
that noise?
True Seismic Ray Path (ray tracing)
NMO Assumed Straight Ray Path t
2
=t
o
2
+(
x
/
v
)
2
Landmark Well Seismic Fusion
Release 5000.0.0.1 Seismic velocity background B-13
There are several heuristic explanations oI the petrophysical eIIects oI
velocity in Chapter 1. Pay particular attention to the Pressure, (P)
section.
Seismic velocity accuracy
Uses of Seismic VeIocity - V
s
For applications such as Well Seismic Fusion, you can see that you
need a very accurate velocity Iield
Uses Required
Accuracy
Velocity Type
Signal Enhancement: NMO and
Stack
2 - 10 V
s
Structure: Migration Depth
Conversion
1 - 5 V
RMS
, V
AVG,
, V
int
Stratigraphy: Stratigraphic
Correlation or Lithologic Variation
(Sand/Shale Ratio)
1 - 2 V
int
Physical Properties: Porosity,
Density, Fluid Content (Water/Oil/
Gas)
1 - 2 V
int
Well Seismic Fusion Landmark
B-14 Seismic velocity background Release 5000.0.0.1
TypicaI Depth Error
II 5 to 10 percent depth error is acceptable, you may be able to obtain
this goal with a DepthTeam Express technique. This is signiIicantly
Iaster than using the more accurate methods Iound in DepthTeam
Explorer and DepthTeam Extreme.
For example, let us say that your prospect is an anticline at an
approximate depth oI 3000 meters, and an estimated structural relieI oI
500 meters. II the objective oI your depth conversion project is to
determine that the structure is present in depth, a DepthTeam Express
workIlow could be used to achieve this objective. Since the depth relieI
oI the prospect is 17 percent (500 / 3000), an accuracy oI /- 5 to 10
is suIIicient. However, iI the estimated structural relieI is only 30
meters, the depth relieI oI the prospect would be one percent (30 /
3000), requiring a more accurate solution than the /- 5 to 10
accuracy that DepthTeam Express can provide.
Large, thick targets can typically allow greater depth uncertainty than
small thin ones. However, allowable depth uncertainty, is always tied to
drilling economics, so it is helpIul to have some idea oI the minimum
allowable Iield size. The closer you are to this economic limit, the
greater is your need Ior an accurate depth conversion solution.
Method Typical Depth Error
Vertical depth conversion/Dix velocities -5 to 10 (Dix velocities, typically
too Iast)
Map Migration/CMP Coherency Inversion /-0 to 3
Geostatistical Pseudo-interval velocity
integration
/-0 to 1
Landmark Well Seismic Fusion
Release 5000.0.0.1 Seismic wavelet extraction B-15
Seismic waveIet extraction
The seismic wavelet is a seldom known commodity. Despite the best
eIIorts oI modern seismic data processing, the phase and the Irequency
content oI the 'best wavelet Ior a synthetic is still in doubt. Wavelet
extraction algorithms are used to extract Irequency and phase
inIormation Irom the seismic data. When you apply the 'extracted
wavelet to the reIlection coeIIicients, you get a more accurate/realistic
synthetic seismogram.
The wavelet extracted in Fusion will also help you globally scale your
seismic to the well logs. The scaling will be 'relative because the
practical problems in recovering absolute reIlection amplitudes seem
more severe than the problems oI recovering the relative variation oI
reIlection amplitude with oIIset. In general, reIlection coeIIicients are
between -1.0 and 1.0, while the seismic reIlectivity range is usually
between -0.3 and 0.3. Hence, the reIlection amplitudes have been
scaled by the incident-wave amplitude. We believe it is appropriate to
use the inIormation content oI the relative reIlection coeIIicient curve
R( )/R
o
in your Fusion analysis.
On real data, this 'wavelet matching step is essential because Ilat-
lying, thick constant-velocity beds that produce zero-phase reIlections
are seldom observed.
There are many tools you can use to derive a wavelet. For example, you
can use LogM`s WavX
TM
or SynTool`s SeisWell
TM
routine to extract/
generate wavelets and store them to the OpenWorks database. Once the
wavelets are in the OpenWorks database, they are available Ior use by
the Fusion algorithms.
You can also use Fusion itselI to extract a representative wavelet Irom
one oI your prestack gathers or Irom a stacked trace. Fusion extracts a
wavelet by comparing a synthetic reIlectivity series with the equivalent
real seismic. ReIer to WorkIlow 7.1: Wavelet estimation in Chapter 7
on page 3 Ior more inIormation on extracting a wavelet.
The shape/size/phase oI the extracted wavelet should look 'stable and
well balanced. II the wavelet lobes are not balanced, you will get a
disproportionate amount oI geologic mixing either above or below
reIlecting boundaries.
u
Well Seismic Fusion Landmark
B-16 Seismic wavelet extraction Release 5000.0.0.1
Landmark Fluid Substitution Background C-1
!""#$%&'
Fluid Substitution Background
This appendix details more oI the Iluid substitution background that is touched upon in the body
oI the manual.
Topics covered in this appendix:
Summary oI Methods
Fluid substitution using Gassmann`s equation
Estimation oI shear-wave velocity
Estimation oI P-wave velocity
Porosity substitution
Normalization oI mineral volumes
Shaly rocks
User interIace
Recommended workIlow
)
Well Seismic Fusion Landmark
C-2 Summary of Methods Release 5000.0.0.1
Summary of Methods
Fluid substitution
1. Gassmann`s equation
Vp-Vs estimation
1. Wet-rock relationships
Castagna mudrock line
V
s
C
0
C
1
V
p
C
2
V
p
2
(Greenberg-Castagna)
V
s
2
B
0
B
1
V
p
2
2. Dry-rock relationship
Spencer`s Iormula Ior dry-rock Poisson`s ratio

Porosity substitution
Use the relation that expresses dry-rock bulk modulus in terms oI pore
bulk modulus ( ), grain bulk modulus (K
GR
) and porosity ( ):

Key assumptions:
is constant under a small change in porosity
Dry-rock Poisson`s ratio is constant under a small change in
porosity
Note
Other than the Castagna mudrock line, the above relations can be used to estimate
Vp or Vs Ior any in-situ Iluid.
K
|
|
1
K
DRY
--------------
|
K
|
------
1
K
GR
-----------
K
|
Landmark Well Seismic Fusion
Release 5000.0.0.1 Summary of Methods C-3
Fluid properties
1. Fluid component properties (density and bulk modulus oI brine, oil,
and gas components)
Batzle-Wang(1992), or select input curves or enter constants
2. Fluid mixing bulk modulus
Small scale mixing: Wood`s equation
Patchy saturation: Voigt average
Brie`s method
3. Fluid mixing density
Weighted-by-volume average oI brine and hydrocarbon
densities
Grain properties
1. Grain bulk modulus
Voight-Reuss-Hill average oI component mineral bulk moduli



where vol
i
is the normalized volume oI i
th
mineral component
oI the grain, and Km
i
is the bulk modulus oI i
th
mineral
component oI the grain.


|
|
.
|

\
|
+ =
HC
W
BRINE
W
FLUID
K
S
K
S
K
1
1
HC W BRINE W FLUID
K S K S K + = ) 1 (
( )
HC
e
W HC BRINE FLUID
K S K K K + =
HC w BRINE w FLUID
S S + = ) 1 (
K
GR
1
2
--- vol
i
Km
i

(
vol
i
Km
i
----------
.
|

\
|
1
.
|

Well Seismic Fusion Landmark
C-4 Summary of Methods Release 5000.0.0.1
2. Grain density
Weighted-by-volume average oI component mineral densities
Landmark Well Seismic Fusion
Release 5000.0.0.1 Algorithms C-5
AIgorithms
Fluid substitution using Gassmann's equation
Gassmann's equation
Gassmann`s equation relates the bulk modulus (or incompressibility) oI
the rock with Iluid-Iilled pore space to the bulk modulus oI the dry rock
(where the pore space is devoid oI all Iluids), the bulk modulus oI the
solid grains, and the bulk modulus oI the Iluid:
(1)
where
K
SYS
bulk modulus oI the Iluid Iilled rock (rock system)
K
DRY
bulk modulus oI the dry rock Irame
K
GR
bulk modulus oI the solid grains
K
FL
bulk modulus oI the Iluid
porosity
The bulk modulus oI the rock is calculated Irom density and velocities
as
(2)
where
density oI rock
J
p
compressional wave velocity oI rock (Irom sonic log)
J
s
shear wave velocity oI rock (typically estimated Irom V
p
-V
s

trends)
FL GR
FL
DRY GR
DRY
SYS GR
SYS
K K
K
K K
K
K K
K

=
|
1
|
|
.
|

\
|
=
2 2
3
4
S P SYS
J J K

Well Seismic Fusion Landmark


C-6 Algorithms Release 5000.0.0.1
The grain bulk modulus (K
GR
) is calculated as a Voight-Reuss-Hill
average:
where J
i
and K
i
are the normalized volume and bulk modulus,
respectively, oI the i
th
mineral component oI the grain.
The Iluid bulk modulus (K
FL
) is calculated using one oI the mixing
Iormulas (see Summary oI Methods above). The bulk modulus oI the
brine and hydrocarbon components oI the Iluid are calculated using the
Batzle-Wang Iormulas. Alternatively, the properties oI the Iluid
components can be read Irom input curves or speciIied as constants.
The calculated quantities (K
SYS
, K
GR
, and K
FL
) and the porosity are
then used to solve Gassmann`s equation Ior the dry-rock bulk modulus
(K
DRY
).
FIuid substitution
Fluid substitution involves two applications oI Gassmann`s equation
the Iirst to go Irom the in-situ saturation state to the dry-rock state, and
the second to go Irom the dry-rock state to the new saturation state.
The steps oI Iluid substitution are as Iollows:
1. Calculate the in-situ rock bulk modulus (K
SYS1
), the grain bulk
modulus (K
GR
), and in-situ Iluid bulk modulus (K
FL1
).
2. Using these and porosity, solve Gassmann`s equation Ior the dry-rock
bulk modulus (K
DRY
).
3. Calculate the bulk modulus oI the substitute Iluid (K
FL2
).
4. Using K
DRY
, K
GR
, K
FL2
, and porosity, solve Gassmann`s equation Ior the
rock bulk modulus (K
SYS2
) at the new saturation state.
5. Calculate the density at the new saturation state:

( )
|
|
.
|

\
|
|
|
.
|

\
|
+ + + + + + + =

...) /( 1 ...
2
1
3
3
2
2
1
1
3 3 2 2 1 1 ) (
K
J
K
J
K
J
K J K J K J K
H R J GR

2

1

Iluid2

Iluid1
( ) |
Landmark Well Seismic Fusion
Release 5000.0.0.1 Algorithms C-7
6. Calculate the new shear velocity. The shear modulus oI the rock ( )
is assumed to be constant under a change oI pore Iluid,
The new shear velocity is

7. Calculate the new P-wave velocity (J
P
) using K
SYS2
, , J
S2
, and
solving equation (2).
VaIidation
Fluid substitution can give a physically unreasonable result. This can
occur because the input porosity is underestimated (Ior example, by
using eIIective porosity instead oI total porosity), the in-situ Iluid is not
well known, the grain composition is not well known, the input
velocity and density logs are bad, or Ior other reasons.
At each depth sample processed in Iluid substitution, a test is
perIormed to determine whether the input data and parameters are
reasonable (selI-consistent as a set), such that new curve data
calculated by Iluid substitution will be physically valid. When the test
Iails, an error Ilag is set and all output values at that depth are set to
null.
The deIault test is to check that the rock bulk modulus at in-situ
saturation (K
SYS1
) Ialls between the Reuss and Voigt limits calculated at
the in-situ saturation.
This is equivalent to checking that the drv rock bulk modulus is greater
than or equal to zero and less than or equal to K
JOIGTDRY
(1- )K
GR
.

2
2 2
2 2
1 1 S DRY S DRY S
J J J = = =
2
1
1 2

S S
J J =

2
( ) | |
1 1
2
1
2
1 1 1
1
1
1 )
3
4
(
1
1
FL GR JOIGHT S P SYS
FL GR
REUSS
K K K J J K
K K
K | |
| |
+ = s
(

= s
(

|
|
.
|

\
|
+

=
|
Well Seismic Fusion Landmark
C-8 Algorithms Release 5000.0.0.1
An alternate test that users can choose checks that the dry-rock
Poisson`s ratio Ialls between user-speciIied minimum and maximum
values. The dry-rock Poisson`s ratio is calculated at the intermediate
stage oI Iluid substitution.
Robust estimation
The purpose oI the robust estimation option is to give Iluid substitution
a second chance at depth samples where the validation test Iails. This
can help to eliminate null values in the output curves.
A general implementation oI robust estimation might allow users to set
up a prioritized set oI rules that the program would Iollow to modiIy
various input data within expected uncertainties, when Iluid
substitution Iails and then perIorm the calculations again using the
modiIied data. However, given the various possible errors in input data
(errors in log data and uncertainties in grain and Iluid parameters) that
can cause Iluid substitution to Iail, it is probably diIIicult to come up
with an objective set oI rules that would generally work.
Well Seismic Fusion`s implementation oI the robust estimation option
is very simple. When a test value (either rock bulk modulus or dry-rock
Poisson`s ratio) at a particular depth Ialls outside oI the limits, the value
is moved to the nearest limit. A new input value Ior either V
s
or V
p
is
recalculated using the modiIied intermediate value and the Iluid
substitution calculation is rerun. The deIault is to recalculate V
s
.
Note
The most common problem in Iluid substitution is that K
SYS
Ialls below the Reuss
limit or, equivalently, that K
DRY
is negative.
Note
Spencer et al (1994) Iound that the lower limit oI the dry-rock Poisson`s ratio Ior
sand/shale sequences is approximately 0.1.

o
2 6
2 3
+

=
DRY
DRY
DRY
K
K
Landmark Well Seismic Fusion
Release 5000.0.0.1 Algorithms C-9
Recalculation oI J
s
or J
p
uses a modiIied value oI the in-situ rock bulk
modulus:
or
where K
SYSMOD
is the new value oI the rock bulk modulus at in-situ
saturation.
Recalculation oI J
S
or J
P
using a modiIied value oI the dry-rock
Poisson`s ratio involves the Iollowing steps:
1. Calculate the dry-rock bulk modulus using the new dry-rock
Poisson`s ratio.
2. Solve Gassmann`s equation to Iind a new value Ior the rock bulk
modulus at in-situ saturation.
3. Calculate a new value Ior input J
s
or J
p
value as above.
The robust estimation option should be used with care. It is strongly
recommended that Iluid substitution be initially run without using the
robust estimation option. The error checking provides a good way to
determine the consistency oI the input data (log data and grain and Iluid
parameters). The input data should be inspected at depths where the
computation Iails (null values in the output curves). Obvious errors in
the input data should be corrected beIore rerunning Iluid substitution
(again without robust estimation) to check the eIIect oI the edits.
Correcting errors in the input data will give better values oI the test
parameter (bulk modulus or dry-rock Poisson`s ratio) than simply
moving it to the nearest limit, and this will lead to better output values.
Fluid substitution will succeed at all depths Ior any given substitute
Iluid, once a consistent set oI input data is obtained.
Note
The actual input curve is not modiIied. The recalculated value is written only to the
robust estimation (input data) section oI the ASCII data report Ior comparison to the
original input value.
) (
4
3
2 2

MOD SYS
P S
K
J J =

MOD SYS
S P
K
J J
2 2
3
4
+ =
Well Seismic Fusion Landmark
C-10 Algorithms Release 5000.0.0.1
Estimation of shear-wave velocity
Fluid substitution requires shear-wave velocity (J
S
) at in-situ saturation
as input. This is oIten unavailable and must be estimated.
SpeciaI case (wet rock, "wet" method)
When the rock is 100 water saturated (and reIerence brine parameters
have not been speciIied), shear velocity can be calculated directly Irom
P-wave velocity by applying one oI the Iollowing 'wet rock V
p
-V
s

relationships:
1. Castagna mudrock line
J
P
(km/s) 1.36 1.16J
S
2. Greenberg-Castagna
J
S
C
0
C
1
V
P
C
2
J
P
2

3. Velocity-squared relation
J
S
2
B
0
B
1
J
P
2
The Castagna mudrock line method Ior estimating J
S
or J
P
requires only the
other velocity as the input. DiIIerent constants Ior intercept and slope can be
entered iI these have been calibrated locally.
The coeIIicients in the Greenberg-Castagna and 'velocity-squared methods
are mineral-speciIic (that is, a separate set oI coeIIicients is speciIied Ior each
mineral component oI the rock grains). Relative mineral volumes are also
required. The procedure Ior estimating a velocity Ior a multi-mineral rock
using either oI these two methods is shown below.
1. Compute a value oI J
S
Ior each mineral type using the input J
P
and
the speciIied coeIIicients Ior that mineral.
2. Use the computed mineral-speciIic values oI J
S
and the relative
mineral volumes to compute the eIIective J
S
as


Note
These steps reIer to estimating V
S
, but the same steps apply when estimating V
P
.
V
s
1
2
--- vol
i
V
s

vol
i
V
s
---------

\ .
| |

Landmark Well Seismic Fusion
Release 5000.0.0.1 Algorithms C-11
where vol
i
and Js
i
are the normalized volume and mineral-speciIic J
S
,
respectively, oI the i
th
mineral component oI the grain. This is
analogous to computing a Voigt-Reuss-Hill average oI the elastic
moduli. ReIer to Greenberg and Castagna (1992) Ior a justiIication oI
this approach.
GeneraI case
All cases oI J
S
estimation (other than the special case deIined above)
require an internal application oI Gassmann`s equation to go Irom the
in-situ saturation state to the reIerence saturation state Ior which the
selected estimation method is valid. These cases include:
In-situ saturation not 100 wet
In-situ saturation 100 wet, but the 'dry rock method Ior
calculating J
S
has been selected
Estimation of in-situ shear velocity using a ~wet rock V
P
-V
S

relationship
The Iact that Gassmann`s equation requires J
S
as input, but an
estimation oI J
S
, in general, requires an application oI Gassmann`s
equation, leads to a circular dependence. Problems such as this can be
handled either by requesting additional inIormation, such as the dry-
rock Poisson`s ratio (which may be diIIicult Ior the user to provide), or
by solving the problem iteratively in an attempt to converge to a
solution. The estimation oI J
S
in Well Seismic Fusion uses the iterative
approach as described below.
The general estimation oI shear velocity using a 'wet rock J
P
-J
S

empirical relationship (either the Greenberg-Castagna quadratic
equation or the linear equation in velocity squared) proceeds as
Iollows:
1. At the current depth sample, make an initial guess at Jpwet (the P-
wave velocity oI the rock with the in-situ pore Iluid replaced by
100 brine). This initial guess by the program is midway between
a lower bound oI 0.9 * Jpinsitu and an upper bound oI 8000 m/s.
Note
The actual value oI Vpwet is expected to be Iaster than Vpinsitu (where the in-situ Iluid
includes hydrocarbons), but a Iactor oI 0.9 is used in the initial lower bound to help stabilize
convergence to a solution.
Well Seismic Fusion Landmark
C-12 Algorithms Release 5000.0.0.1
2. Use the Jpwet guess to estimate Jswet. This is just a direct
application oI the selected 'wet rock J
S
-J
P
relationship as
described in the Special Case section above.
3. Calculate the bulk density oI the wet rock:

where is rock bulk density at in-situ saturation, is
density oI brine, and is density oI the in-situ pore Iluid.
This calculation only needs to be done once at each depth sample,
not on each iteration at that sample.
4. Using Jpwet (guess), Jswet (estimated Irom guess), densitvwet
(calculated), grain bulk modulus, brine bulk modulus, and in-situ
bulk modulus at the current depth, perIorm a Iluid substitution to
replace the (reIerence state) brine with the in-situ Iluid. This gives
estimated values Ior Jpinsitu and Jsinsitu.
5. Check the estimated Jpinsitu against the known Jpinsitu. II the
estimated value is within 5 m/s oI the known value, then the value oI Jp
wet that was used as input to the J
S
-J
P
wet estimation in Step 2 is
assumed to be accurate enough, and the estimated Jsinsitu is accepted.
The iteration stops at this point.
6. II Jpinsitu estimated by Gassmann`s equation is not within 5 m/s oI the
known Jpinsitu, then adjust the upper or lower bounds oI Jpwet. II the
estimated value is too low, set the lower bound to be equal to the previous
midpoint. II the estimated value is too high, set the upper bound to be
equal to the previous midpoint. Take the midpoint oI the new range as the
next guess at Jpwet and go back to Step 2.
This is a bisection routine, since the range oI possible Jpwet values is
reduced by one halI aIter each iteration. Convergence to a solution
typically occurs within 5 or 6 iterations. II convergence does not occur
within 20 iterations, Jsinsitu is set to null at that depth and the process
moves to the next depth sample.
Note
Brine properties at the wet reIerence state are assumed to be those oI the brine component oI
the in-situ Iluid.

WET

1

BRINE

FLUID1
( )|

1

BRINE

FLUID1
Landmark Well Seismic Fusion
Release 5000.0.0.1 Algorithms C-13
Estimation of in-situ shear velocity a ~dry rock relationship
The estimation oI J
S
with 'dry rock as the reIerence state uses the
same iterative bisection routine as described above, with the Iollowing
diIIerences:
The reIerence Iluid state is dry instead oI wet, so the program needs
to Iind a value Ior Jpdrv, which is the velocity oI the rock with all
in-situ Iluid removed.
In Step 2 oI the iteration, the value oI Jsdrv is calculated using
Spencer`s Iormula. See below for details.
In Step 3, densitvdrv is calculated.
In Step 4, the internal Iluid substitution goes Irom dry-rock (no
Iluid) to in-situ saturation.
Estimation of *+,%-. using Spencer`s formula
1. Calculate the solid grain Poisson`s ratio:
Calculate the shear modulus oI each mineral (grain component)
Irom the bulk modulus and Poisson`s ratio, e.g. Ior shale
Calculate the Voight-Reuss-Hill average oI the bulk moduli oI
the grain components (same way that K
GR
is calculated Ior use
in Gassmann`s equation)
) 2 2 /( ) 6 3 ( * + =
SH SH SH SH
K o o
...

+ + + =
LS LS SS SS SH SH JOIGHT GR
K vol K vol K vol K
...) /( 1

+ + + =
LS
LS
SS
SS
SH
SH
REUSS GR
K
vol
K
vol
K
vol
K
) (
2
1
REUSS GR JOIGHT GR H R J GR
K K K + =

Well Seismic Fusion Landmark
C-14 Algorithms Release 5000.0.0.1
Calculate the Voight-Reuss-Hill average oI the shear moduli oI
the grain components:
Calculate the grain Poisson`s ratio
2. Calculate the Irame (dry-rock) P-wave modulus:
P
A

DRY
Vp
DRY
2
3. Use Spencer`s Iormulas to calculate the dry-rock Poisson`s ratio:
II P
A
~ 15 GPa (consolidated sandstone)
II (unconsolidated sand)
Note
DeIault values Ior the empirical constants are 0.205 and b0.04 when P
A
is in units
oI gigaPascals (GPa).
...

+ + + =
LS LS SS SS SH SH JOIGHT GR
vol vol vol
...) /( 1

+ + + =
LS
LS
SS
SS
SH
SH
REUSS GR
vol vol vol

) (
2
1
REUSS GR JOIGHT GR H R J GR
+ =

H R J GR H R J GR
H R J GR H R J GR
GR
K
K


+

=


2 6
2 3

o
049 . 0 647 . 0 + =
GR DRY
o o
P
A
15GPa s
049 . 0 647 . 0 )) * exp( 1 exp( * ) 049 . 0 647 . 0 (
0
+ + =
GR A GR DRY
P b o o o o
o
0
Landmark Well Seismic Fusion
Release 5000.0.0.1 Algorithms C-15
4. Calculate Js
DRY
using the standard transIorm:
2 / 1
1
5 . 0
|
|
.
|

\
|

=
DRY
DRY
DRY DRY
Jp Js
o
o
Well Seismic Fusion Landmark
C-16 Algorithms Release 5000.0.0.1
Estimation of P-wave velocity
An option is provided in Well Seismic Fusion Iluid substitution to
estimate P-wave velocity Irom shear-wave velocity.
SpeciaI case
Conditions Ior this case are the same as Ior the special case Ior
estimating shear velocity Irom P-wave velocity: The rock is 100
water saturated, reIerence brine parameters have not been speciIied,
and one oI the 'wet rock methods has been selected. The relations are
the inverse oI those Ior estimating shear velocity Irom P-wave velocity:
1.
2.
The empirical coeIIicients Ior the diIIerent constituent minerals oI the
grain are linearly mixed to obtain average values to use in the
equations.
GeneraI case
The estimation oI Jpinsitu using either the wet-rock methods or the
dry-rock method (Spencer`s Iormula) is much more direct than the
estimation oI in-situ shear velocity. This is because the shear modulus
oI the rock is assumed to be constant under a change oI Iluid, that is
The steps in the estimation oI Jpinsitu are:
1. Calculate the rock bulk density at the reIerence state (either wet or
dry).
2. Use the above assumption to calculate J
S
at the reIerence state.
3. Use the selected J
S
-J
P
estimation method to calculate J
P
at the
reIerence state.
For the wet-rock methods, this is an application oI the
appropriate equation shown above under Special case.
( )
2
0 2
2
1 1
2
4
C
C J C C C
J
S
P
+ +
=
1
0
2
B
B J
J
S
P

=
2 2 2
DRY DRY WET WET INSITU INSITU
Js Js Js = = =
Landmark Well Seismic Fusion
Release 5000.0.0.1 Algorithms C-17
The dry-rock method Iirst calculates the dry-rock Poisson`s
ratio using Spencer`s Iormula Ior consolidated sandstone



Jpdrv is then calculated as



4. Apply Gassmann`s equation to transIorm back to in-situ saturation,
calculating Jpinsitu.
Porosity substitution
Porosity substitution as implemented in Well Seismic Fusion may be
suitable Ior relatively clean rocks where the average pore aspect ratio is
not expected to change when porosity is increased or decreased by a
small amount (Ior example, by 5 to 10 pu). This assumption is implicit
in the use oI the Iormula



where (the bulk modulus oI the pore spaces) is assumed to be
constant under a change oI porosity.
The computations Ior porosity substitution use the Iollowing steps:
1. Solve Gassmann`s equation to Iind the dry-rock bulk modulus
(K
DRY
) at the current porosity.
2. Use the Iollowing Iormula to compute the dry-rock bulk modulus at
the new porosity:


3. Compute the dry-rock density, shear-wave velocity, and P-wave
velocity at the new porosity. This requires either additional
Note
The dry-rock P-wave modulus cannot be calculated without Iirst knowing Vpdry, and
thereIore Spencer`s Iormula Ior unconsolidated sands and sandstones cannot be used here.
o
DRY
0.647o
GR
0.049
Vp
DRY
Vs
DRY
1 o
DRY

0.5 o
DRY

---------------------------
1
K
DRY
--------------
|
K
|
------
1
K
GR
-----------
K
|
1
K
DRYnew|
-------------------------
|
new
|
-----------
1
K
DRY
-------------- 1
|
new
|
-----------
\ .
| |
1
K
GR
-----------
Well Seismic Fusion Landmark
C-18 Algorithms Release 5000.0.0.1
inIormation or an additional assumption.
4. PerIorm a Iluid substitution Irom the dry rock at the new porosity to
in-situ saturations.
Fluid properties
The density and bulk modulus oI the hydrocarbon and brine
components oI the pore Iluid are calculated using Iormulas published
by Batzle and Wang (1992). Alternatively, users can either enter
constants or select log curves Ior the density and bulk modulus oI the
brine and hydrocarbon (oil or gas) components.
Details oI interest in the application oI the Batzle-Wang Iormulas in
Well Seismic Fusion:
Brine density and bulk modulus calculations do not take into
account any dissolved gas. Dissolved gas has a negligible eIIect
on brine properties
Live oil density (Batzle-Wang equation 24) is adjusted only
once (using Batzle-Wang equation 18), not twice as suggested
in the original paper. This Iollows the example calculation
given in The Rock Physics Handbook (p. 219) and has been
conIirmed by Mike Batzle (personal communication with
Reginald Beardsley).
Note
In the current implementation oI Well Seismic Fusion, the assumption is made that the dry-
rock Poisson`s ratio remains constant under a small change in porosity.
Note
The Batzle and Wang (1992) Iormulas should not be used to compute the properties oI
condensate or to compute the properties oI oil at high pressure and temperature.
Landmark Well Seismic Fusion
Release 5000.0.0.1 Algorithms C-19
Normalization of mineral volumes
When the rock type is shaly sand or oI mixed lithology, mineral
volumes are normalized by the program at each depth sample.
Normalized volumes are used to calculate eIIective grain bulk modulus
Ior use in Gassmann`s equation, average coeIIicients Ior estimating V
S

or V
P
(wet methods), and grain bulk density Ior estimating porosity or
rock bulk density.
The input mineral volumes, whether entered as constants or read Irom
volume curves, may represent either the volume Iraction oI minerals in
the solid grains (excluding pore space) or the volume Iraction oI
minerals relative to the whole rock volume (including porosity), as long
as all speciIied volumes are oI the same type. Normalized mineral
volumes represent the volume Iractions oI the solid grain.
Users are given two options Ior normalizing mineral volumes:
1. Preserve relative volumes
2. Apply linear scaling to volume oI shale beIore normalizing
Preserve reIative voIumes
The simplest normalization option is 'Preserve relative volumes. With
this option, the volumes oI all solid grain components are scaled to add
up to 1, while keeping the same proportions relative to each other.
Example:
vol. shale 8
vol. sandstone 20
vol. limestone 35
vol. anhydrite 30
other volumes 0
sum 0.08 0.20 0.35 0.30 0.93
normalized vol. shale 0.08 / 0.93 0.086 (8.6)
normalized vol. sandstone 0.20 / 0.93 0.215 (21.5)
normalized vol. limestone 0.35 / 0.93 0.376 (37.6)
normalized vol. anhydrite 0.30 / 0.93 0.323 (32.3)
other volumes 0
Well Seismic Fusion Landmark
C-20 Algorithms Release 5000.0.0.1
AppIy Iinear scaIing to voIume of shaIe before normaIizing
This option allows the shale volume to be adjusted prior to
normalization. During normalization, the adjusted volume oI shale is
held constant, while the other volumes are scaled proportionately
(relative to each other) to Iill the remaining volume oI the grain.
Users may want to use this option to adjust a volume oI shale curve that
has been calculated in the usual way so that values range Irom 0 in
clean Iormations to 1 in clay-rich shales, to make it more representative
oI the clay in the rock. For example, in a particular area a shale volume
oI 0.65 might be used as a threshold value that separates grain-
supported rocks Irom clay-matrix-supported rocks. (Keith Katahara,
private communication.)
The steps involved in this normalization method are:
1. Find the relative proportion oI the solid grain occupied by shale.
This step does the same calculations as the 'Preserve relative
volumes option. It is necessary to do this Iirst step to ensure that
the volume oI shale is the proportion oI shale in the solid grain, and
not the proportion oI shale in the whole rock.
Example:
Input volumes:
vol. shale 30
vol. sandstone 40
vol. limestone 5
other volumes 0
Using the calculation method shown in 'Preserve relative
volumes, this step gives:
vol. shale in grain 0.3 / (0.3 0.4 0.05) 0.4(40)
Landmark Well Seismic Fusion
Release 5000.0.0.1 Algorithms C-21
2. Apply linear scaling to the volume oI shale.
Linear scaling uses a 'Shale threshold value. This is entered as a
decimal number between 0.0 and 1.0 and represents the value oI
shale volume that will be scaled up to 1.0 (or 100). All values
greater than the threshold value will be set to 1.0 and all values less
than the threshold value will be linearly scaled between 0.0 and 1.0.
Example: (using relative vol. oI shale in solid grain calculated
above)
vol. shale 0.4 (40)
shale threshold 0.8
scaled volume oI shale 0.4 / 0.8 0.5 (50)
3. Normalize the other mineral volumes
Volumes oI the 'clean grain components are scaled
proportionately to Iill the volume oI grain not occupied by the
scaled volume oI shale.
Example: (using scaled vol. oI shale calculated above and the other
input volumes)
sum oI clean components 0.40 0.05 0.45
volume oI grain not occupied by shale 1.0 0.5 0.5
normalized vol. sandstone (0.40 / 0.45) * 0.5 0.444(44.4)
normalized vol. limestone (0.05 / 0.45) * 0.5 0.056(5.6)
Shaly rocks
Apart Irom the special volume normalization option Ior shale discussed
above, Well Seismic Fusion Iluid substitution treats shale as just
another component oI the grain. Within this general model oI the rock,
users can handle the shale component diIIerently themselves by the
type oI input data values they provide. The main thing is to provide a
consistent set oI values Ior porosity, water saturation, and clay or shale
bulk modulus. Possible approaches to handling shale are:
Note
When the speciIied shale threshold is 1.0, the values oI shale volume are not
changed prior to normalization. The normalized mineral volumes in this case are
identical to those obtained using 'preserve relative volumes.
Well Seismic Fusion Landmark
C-22 Algorithms Release 5000.0.0.1
1. Input the rock porosity as total porosity, including the non-eIIective
microporosity in the shale. Input water saturation as total water
saturation, including all bound and interstitial water trapped in the
shale. For shale bulk modulus, use a relatively high value oI bulk
modulus Ior dry clay. A high value oI porosity can help keep the
dry rock bulk modulus Irom going negative (Keith Katathara,
private communication).
2. Input the rock porosity and water saturation to include interparticle
water in shales, but to exclude water bound to clay surIaces. For
shale bulk modulus, enter a value Ior surIace-wetted clay.
3. Input eIIective porosity and eIIective water saturation. Input a low
value Ior shale bulk modulus that is representative oI the shale
'grain including shale microporosity. It may be diIIicult to come
up with an appropriate value oI shale bulk modulus when handled
this way.
Because oI the diIIiculty in accurately characterizing the shale component in
shaly rocks (including clay bulk modulus, eIIective vs. total porosity, and
eIIective vs. total Iluid saturations), Well Seismic Fusion Iluid substitution
(including estimation oI shear velocity in hydrocarbon zones) works best
when applied to relatively clean reservoir rock, such as a blocky sandstone.
Calculations oIten break down in the transition zones Irom clean reservoir to
shale, especially when eIIective porosity is used. Some Ieatures speciIically
related to shaly rock are described below.
ControI curve
An optional control curve can be selected when speciIying the
computational depth range. The control curve, along with a cutoII value
and relational condition (greater than, less than, equals), are used as
discriminators to restrict computations to sub-zones where control
curve values satisIy the speciIied condition. This is normally used to
restrict computations to clean zones when you want to perIorm
computations over a depth range that includes multiple reservoir zones
within interbedded shales.
Landmark Well Seismic Fusion
Release 5000.0.0.1 Algorithms C-23
Linear scaIing of shaIe voIume
This option is only available in the Fluid Substitution Wizard (Expert
Mode). This option in mineral normalization lets you adjust the volume
oI shale by linear scaling prior to normalization oI other mineral
volumes. The linear scaling can be used to adjust a shale volume curve
to make it more representative oI clay in the rock. For example, in a
particular area a shale volume oI 0.65 might be used as a threshold
value that separates grain-supported rocks Irom clay-matrix-supported
rocks (Keith Katahara, private communication).
The steps involved in mineral normalization using this option are:
1. II mineral volumes have been entered as Iractional volumes relative
to the entire rock (including porosity), then convert the volume oI
shale to a Iractional volume oI the solid grain.
2. Apply linear scaling to the shale volume. This uses a speciIied
'threshold value between 0.0 an 1.0 that represents the value oI
the shale volume that will be scaled up to 1.0. All values equal to or
greater than the threshold value are set to 1.0. All values less than
the threshold value are linearly scaled between 0.0 and 1.0.
3. Normalize other mineral volumes by scaling them proportionately
to Iill the grain volume not occupied by the scaled volume oI shale.
Substitute into effective porosity onIy
Gassmann`s equation requires porosity, grain bulk modulus, and Iluid
bulk modulus. II input porosity is total porosity (including porosity
trapped in shale particles), then the mineral volume and bulk modulus
entered Ior the shale (clay) component oI the grain should be Ior dry
clay, and the Iluid saturations required to compute the bulk modulus
and density oI the in-situ Iluid should be total saturations (that is, water
saturation includes brine trapped in the shale particles).
For Iluid substitution to be as accurate as possible, Iluid saturations
used to compute the bulk modulus and density oI the substitute Iluids
should also be total saturations. However, in shaly rocks, where a
varying amount oI the total porosity is tied up in the (varying) shale
portion oI the rock, it is diIIicult Ior users to provide this inIormation. It
is more convenient to be able to speciIy the substitute Iluid saturation
as a constant value relative to eIIective porosity and let the program
compute the varying total saturations that this implies, given either
eIIective porosity as an additional input or using the shale volume and
an additional approximate value Ior porosity in shale.
Well Seismic Fusion Landmark
C-24 Algorithms Release 5000.0.0.1
An option to speciIy substitute saturations relative to eIIective porosity
is available in the Fluid Substitution Wizard (Expert Mode). The
conversion oI eIIective saturations to total saturations Iollows these
steps:
1. II 'relative porosity in shale is provided as the additional
inIormation (instead oI eIIective porosity), then compute eIIective
porosity as



where is the relative porosity in shale.
2. Compute total saturations as





where and reIer the saturations oI hydrocarbons and
water, respectively.
Correction for mud filtrate invasion
The correction Ior mud Iiltrate invasion, which is an option in the Fluid
Substitution Wizard (Expert Mode) and in the V
S
and V
P
Estimation
wizards, is perIormed in two steps:
1. Given two sets oI invaded-zone saturations (one oI which the
density log 'sees and the other oI which the sonic log 'sees
because oI diIIerent depths oI investigation), adjust the density log
values to be relative to the same invaded-zone saturation state as Ior
the sonic log



where is a measured density log value, is the density
corrected to the saturation state that the sonic log sees, and the Iluid
densities and are calculated as a weighted average
oI the component Iluid densities, given the appropriate saturations
oI water (S
w
), hydrocarbon (S
hc
), and mud Iiltrate (S
mf
) using the
Iollowing expression


|
eII
|
total
Vshale |
shale
( )
|
shale
Shc
total
|
eII
|
total
------------ 1 Sw
eII
( )
Sw
total
1 Shc
total

Shc Sw

2

1

Iluid2

Iluid1
( )|

1

2

Iluid1

Iluid2

Iluid
S
w

brine
S
hc

hc
S
mI

mI

Landmark Well Seismic Fusion
Release 5000.0.0.1 Algorithms C-25
2. PerIorm a Iluid substitution Irom the mud-Iiltrate-invaded Iluid (at
the sonic log depth oI investigation) to the uninvaded reservoir
Iluid.
References
Batzle, M.L., and Wang,Z.J., 1992, Seismic properties oI pore Iluids:
Geophysics, 57, 1396-1408.
Brie, F. Pampuri, A. Marsala, and O. Meazza, 1995, SPE 30595, SPE
Annual Technical ConIerence, Houston.
Greenberg, M.L., and Castagna,J.P., 1992, Shear-wave velocity
estimation in porous rocks: theoretical Iormulation, preliminary
veriIication and applications, Geophysical Prospecting, 40, pp. 195-
209.
Mavko, G., Mukerji, T., and Dvorkin, J., 1998, The Rock Physics
Handbook, Tools Ior Seismic Analysis in Porous Media, Cambridge
University Press
Spencer, J. W., Cates, M.E., and Thompson, D.D., 1994, Frame moduli
oI unconsolidated sands and sandstones, Geophysics, 59, 1352-1361.
Well Seismic Fusion Landmark
C-26 User nterface Release 5000.0.0.1
User Interface
User interface
The Iluid substitution interIace uses a wizard-type dialog to help guide
users through process selection, data input, and selection oI output.
Some general Ieatures oI this are:
Inputs that are not required Ior a particular process are disabled.
Input data can generally be provided either by selecting a log
curve or by entering a constant value.
Calculated data can be used to update existing curves or to
create new ones.
All inputs and selections are dynamically saved and reapplied
as deIaults when the Iluid substitution dialog is opened again
during the current Well Seismic Fusion session.
Any number oI parameter templates can be saved by the user. A
saved template can be loaded to initialize dialog settings Ior any
well.
A typical run through the Basic Fluid Substitution wizard is shown on
the Iollowing pages. ReIer to the online documentation Ior detailed
descriptions oI the dialog controls.
Note
The Basic Fluid Substitution wizard is generally adequate Ior most oI the work your
workIlows. A Fluid Substitution wizard (Expert Mode) is also available, which can be used
when you want to perIorm any oI the Iollowing tasks:
1- You need to correct Ior mud Iiltrate invasion
2- The in-situ Iluid is a mixture oI brine, oil, and gas
3- You want to perIorm oil and gas substitutions using a number oI water saturations
4- You want to perIorm a Iluid substitution to a Iluid that is a mixture oI brine, oil, and gas
Landmark Well Seismic Fusion
Release 5000.0.0.1 Recommended workflow C-27
Recommended workfIow
Fluid substitution(s) on a well with one or more hydrocarbon zones will
generally require at least three runs through the Fluid Substitution
Wizard (assuming a measured shear sonic log is unavailable), and may
require several more iI data problems are encountered.
Objectives in managing log curves created through this process should
be to avoid creating short logs that only cover the hydrocarbon zones,
since these are not useIul Ior either vertical or AVO synthetics, and to
use meaningIul names so that a set oI curves can be easily recognized
(e.g. 'DT gas, 'DTS gas, 'RHOB gas, 'PR gas).
The example workIlow detailed below assumes a gas zone in a well
where a shear sonic is unavailable. The three runs correspond to the
example runs shown above under User InterIace.
1. Correct well logs Ior drilling Iluid invasion in reservoir zones and
make other edits as necessary. Evaluate logs to get estimates oI
water saturation and pore Iluid, lithology (volume curves oI grain
components), and porosity. The preconditioning and estimation oI
input data is a critical step. Except Ior simple cases, this will
generally need to be done in a petrophysical package such as
PetroWorks.
2. Run 1. Estimate a background Vs (shear sonic) log and Poisson's
ratio log over the entire length oI the input sonic log by assuming
that in-situ Iluid everywhere is 100 wet. This assumption will
give incorrect estimations oI Vs and Poisson`s ratio in hydrocarbon
zones, but these zones are corrected in Run 2. Name the new curves
appropriately (e.g. 'ShSonic gas and 'PR gas iI the main zone oI
interest has gas in situ).
3. Run 2. In hydrocarbon zones, run the Vs estimation with the
correct Iluid parameters. Update the background Vs and Poisson's
ratio logs created previously.
4. Inspect the results. A good way to Iind bad data points is to look Ior
NULL values (-9999.25) in the ASCII data report, which can be
selected Ior output in the last step oI the Fluid Substitution Wizard.
II necessary, edit the input logs (sonic, density, porosity, mineral
volumes) and other parameters and re-run the Vs estimation.
Because oI the diIIiculty in estimating shale parameters, it may be
necessary to directly edit gaps in output curves that correspond to
transitions Irom relatively clean reservoir rock to shale.
Well Seismic Fusion Landmark
C-28 Recommended workflow Release 5000.0.0.1
5. Make copies oI the in-situ sonic, shear sonic (Vs) and Poisson's
ratio logs. The copied logs will be updated with the results oI Iluid
substitution (Run 3), so name them accordingly (e.g. 'Sonic oil,
'ShSonic oil, 'Density oil, 'PR oil).
6. Run 3. Run Iluid substitution over the hydrocarbon zone oI
interest. Update the copies oI the in-situ sonic, density, shear sonic,
and Poisson's ratio logs made above.
Warning
Save the well Irequently. There is no Undo aIter logs are updated in Iluid
substitution except to use the Reload Iunction, which restores all logs back to the
state they were in at the time oI the last Save.
Landmark Log Processing D-1
!""#$%&'
Log Processing
This appendix details more oI the log processing background that is touched upon in the body oI
the manual.
Topics covered in this appendix:
Summary oI log processing to block log curves
Methodology Ior picking blocking boundaries
Current user interIace
Backus averaging
Log Scales in Fusion
/
Well Seismic Fusion Landmark
D-2 Log processing in Fusion Release 5000.0.0.1
Log processing in Fusion
Summary of log processing to block log curves
General characteristics
Variable thickness layers, with layer boundaries set at signiIicant log
changes. The same set oI blocking boundaries is applied to all logs
selected Ior blocking, so that layer interIaces are aligned across the set
oI blocked logs.
Parameterization
Start depth and end depth oI the overall interval to apply blocking to.
Automatic picking oI blocking boundaries is controlled by the
minimum layer thickness, minimum Iractional log change (to be
considered signiIicant), and a prioritized list oI control logs (one or
more). Control log(s) will normally be in the set oI logs selected Ior
blocking, but other logs may also be selected. For example, it might by
useIul to select a previously blocked log as a control curve in order to
obtain the same blocking boundaries.
Averaging within blocks
General multi-log blocking calculates the arithmetic average oI raw log
values between the blocking boundaries.
'Multi-log blocking with Backus averaging calculates the arithmetic
average oI density log values within each block and the harmonic
averages oI shear modulus and P-wave modulus values, with averaged
values oI Vp and Vs extracted Irom these.
Landmark Well Seismic Fusion
Release 5000.0.0.1 Log processing in Fusion D-3
Algorithm for picking blocking boundaries
The algorithm Ior picking blocking boundaries in Well Seismic Fusion
is adapted Irom the ProMAX method oI multi-log blocking using a
median Iilter plus 'leading-trailing windows. (See details below.) The
ProMAX method has been modiIied to remove directionality Irom the
result. (Directionality is where diIIerent blocking boundaries might be
picked depending on whether a control log is scanned top-to-bottom or
bottom-to-top.) The method has also been modiIied to allow Ior the use
oI multiple control logs.
Methodology for picking blocking boundaries
For each control log, in order oI priority, do the Iollowing steps:
1. Apply a median Iilter to keep the stronger changes in log values,
while removing spikes. This helps to ensure that a blocking
boundary is not picked on an insigniIicant thin bed located near a
major log change. See example shown below. The length oI the
median Iilter is equal to the user-speciIied minimum layer
thickness, to the nearest number oI odd samples.
Note
The internally generated median-Iiltered version oI the control log is only used to
pick blocking boundaries. Averaging oI log values between blocking boundaries is
done using the unIiltered curve data.
Well Seismic Fusion Landmark
D-4 Log processing in Fusion Release 5000.0.0.1
2. Pick 'candidate blocking boundaries.
The Iirst evaluation depth is located a 'minimum layer
thickness below the user-speciIied start depth. This is because
blocking boundaries are always located at the start and end
depths and no other blocking boundaries can be within the
'minimum layer thickness oI these.
Average the median-Iiltered log values over a trailing window
above the current depth sample and a leading window below the
current depth sample. Both windows have a length equal to the
minimum layer thickness.
Calculate the relative diIIerence between the leading and
trailing averages.
Fractionalchange 2 * (leadingavg - trailingavg) / (leadingavg
trailingavg)
This is actually the 'diIIerence in averages divided by the
'average oI the averages and ensures that, given two averages
(leading and trailing), the same Iractional change is calculated
regardless oI which average is greatest.
II the Iractional change is less than the user-speciIied minimum
change, then advance both leading and trailing windows by one
depth sample and evaluate the relative change between averages
at the new depth.
II the Iractional change is greater than the user-speciIied
minimum change, then Iind the steepest slope in log values
within the leading window and set a candidate blocking
boundary at that depth. Advance the leading and trailing
windows so that the next evaluation depth is one sample below
the depth oI the new candidate boundary.
Continue down the log until the evaluation depth is within a
'minimum layer thickness oI the end depth.
Landmark Well Seismic Fusion
Release 5000.0.0.1 Log processing in Fusion D-5
An example oI picking a blocking boundary is shown below.
3. Attempt to add candidate boundaries, in order of strength, to a Iinal
'key list oI blocking boundary depths that will be used to block
the logs. The key list is a composite list oI blocking boundaries, all
separated by at least the 'minimum layer thickness, picked on the
control log(s).
The 'key list initially holds the start and end depths oI the
overall blocking interval.
Starting with the strongest 'candidate blocking boundary (i.e.
largest Iractional change between 'leading and 'trailing
windows averages) on the current control log, test whether or
not the depth oI the boundary can be inserted into the 'key list.
II the depth is at least the 'minimum layer thickness away
Irom the neighboring depths already in the 'key list, then add
the new depth. Otherwise, skip it and move to the next strongest
blocking boundary and try inserting its depth.
Continue until the end oI the list oI candidate boundaries is
reached or until there is no more room to insert additional
depths into the key list while honoring the minimum layer
thickness.
Well Seismic Fusion Landmark
D-6 Log processing in Fusion Release 5000.0.0.1
An example oI a set oI blocking boundaries picked on two control
curves is illustrated below. This example assumes that the input logs Ior
automatically picking the blocking boundaries are also the logs that the
blocking is applied to, which is the usual case. The combined set oI
blocking boundaries (shown on the right) is used to block all logs
selected Ior blocking.
Landmark Well Seismic Fusion
Release 5000.0.0.1 Log processing in Fusion D-7
An example oI a set oI blocking boundaries picked on three control
curves is illustrated below. The combined set oI blocking boundaries
(shown on the right) is used to block all logs selected Ior blocking.
Note that the blocky-looking highest priority control curve on the leIt is
not the result oI this process, but is a selected control curve that may
have been previously blocked or may be a Ilag curve that represents
lithology types.
Well Seismic Fusion Landmark
D-8 Log processing in Fusion Release 5000.0.0.1
Current user interface
GeneraI muIti-Iog bIocking:
Multiple curves can be selected Irom the pop-up curve selector
Ior the curves to be blocked.
Control curves are used to determine the blocking boundaries.
The priority oI the control curves is determined by the order oI
display in the Control Curve list, with the curve at the top oI the
list having the highest priority. Select the Iirst control curve
using either Insert or Append. Then add additional control
curves. Use Insert to add a new control curve above the initial
control curve, or Append to add another control curve below
the initial control curve.
Landmark Well Seismic Fusion
Release 5000.0.0.1 Log processing in Fusion D-9
MuIti-Iog bIocking with Backus averaging:
Users are required to pick a Vp (or sonic), a Vs (or shear sonic),
and a density log. Otherwise, the controls are the same as Ior
general multi-log blocking.
This may need to be tied in with synthetics and AVO so that
optimum blocking parameters can be determined interactively
by comparing synthetics generated using blocked logs against
synthetics generated using unblocked logs.
Backus averaging
Blocking boundaries Ior Backus averaging are picked using the same
method as Ior general multi-curve blocking. The diIIerence between
Backus averaging (blocked logs) and general multi-log blocking is in
the way the 'average values are calculated Ior each layer.
Following the equations Ior Backus averaging in The Rock Physics
Handbook (p. 144),
Well Seismic Fusion Landmark
D-10 Log processing in Fusion Release 5000.0.0.1
Vertical P-wave velocity:
where 'C is the harmonic average oI P-wave modulus :
(N is the number oI samples being averaged.)
Shear wave velocity:
where 'D is the harmonic average oI shear modulus :
The blocked value oI density is calculated as an arithmetic average
oI density log values.
The algorithm proceeds as Iollows:
1. At each depth sample within the current 'block, compute the P-
wave modulus and the shear modulus using unIiltered values Irom
the selected input logs (density, V
p
, V
s
).
2. Calculate the arithmetic average oI the unIiltered density values over the
interval oI the 'block.
3. Calculate the harmonic average oI the P-wave modulus and the harmonic
average oI the shear wave modulus over the interval oI the 'block.
4. Use the arithmetic average oI density and the harmonic averages oI P-
wave modulus and shear wave modulus to solve Ior the Backus-averaged
values oI V
p
and V
s
.
/
,
C J
v P
=
V
P
2
C
1
1
N
----*
1

1
V
P1
2
---------------
1

2
V
P2
2
--------------- ...
1

N
V
PN
2
-----------------
\ .
|
| |
---------------------------------------------------------------------------------------
/
,
D J
v SJ
=
V
S
2
D
1
1
N
----*
1

1
V
S1
2
---------------
1

2
V
S2
2
--------------- ...
1

N
V
SN
2
-----------------
\ .
|
| |
---------------------------------------------------------------------------------------

block
1
N
----
1

2

3
...
N
( )
Landmark Well Seismic Fusion
Release 5000.0.0.1 Graphical log scaling in Fusion D-11
GraphicaI Iog scaIing in Fusion
Log Scales in Fusion
Fusion Iinds a unique log scale Ior each existing or computed curve it
displays. The implementation Ior Iinding log scales uses three methods
in the order oI priority shown below. II a log scale cannot be Iound
using one method, then the next method is used. II all three methods
Iail to return a log scale, the Well Seismic Viewer uses min and max
values. This should only occur when no scales have been deIined Ior
the curve name or curve unit and all values in the log curve are null.
Method 1 - Curve Name
This method is intended to provide log scales Ior 'unitless curves such
as Poisson's ratio and Vp/Vs. However, it can be used to set Iixed log
scales Ior any curve by name. Curve names and associated leIt/right
scale values are stored in the Ilat Iile FusionCurveNameLogScale.dat
located in the $FUSIONHOME/etc directory. The current content oI
this Iile is shown below:
# Log scales for unitless curves
# curve name | left scale | right scale
PR | 0 | 0.5 |
Poisson | 0 | 0.5 |
Poissons | 0 | 0.5 |
Poisson's | 0 | 0.5 |
Vp/Vs | 1.5 | 3.0 |
VpVs | 1.5 | 3.0 |
Vp-Vs | 1.5 | 3.0 |
Vp_Vs | 1.5 | 3.0 |
Multiple rows are used to handle potential variations in spelling. The
search order Ior Iinding a curve name (and associated scale) in this Iile
is:
1. Look Ior an exact match to the curve name
Well Seismic Fusion Landmark
D-12 Graphical log scaling in Fusion Release 5000.0.0.1
2. Look Ior a partial match. II the curve name contains a name Irom
this Iile as a substring surrounded by non-alpha characters, it is
considered a match.
Example:
PR | 0.1 | 0.5 deIined in the Iile
These curve names will match: LFPPR, PR1, PR NEW,
LPFPREDITED
These names will not match: PRESSURE, PRNEW
Landmark Well Seismic Fusion
Release 5000.0.0.1 Graphical log scaling in Fusion D-13
Method 2 - Curve Unit
This method looks Ior log scales based on curve unit abbreviation (e.g.
m/s, It/s). The method allows multiple sets oI log scales to be deIined
Ior any given unit. The method Iinds the log scales that Iit the data best.
The unit strings and associated leIt/right scales are stored in the
$FUSIONHOME/etc/FusionUnitLogScale.dat Iile. Some scale
deIinitions Irom the deIault version oI this Iile are shown below:
# Log scales for curve units
# ds Unit abbrev | left log scale | right log
scale | optional comment
m/s | 2400 | 5000 | Hoot Vp
m/s | 2000 | 7000
m/s | 1200 | 2400 | Hoot Vs
ft/s | 8000 | 12000 | Hoot Vp
ft/s | 6000 | 22000
ft/s | 2500 | 8000 | Hoot Vs
us/m | 300 | 100 |
us/m | 500 | 100 |
us/m | 500 | 300 |
us/ft | 140 | 40 |
us/ft | 240 | 40 |
us/ft | 280 | 80 |
g/cm3 | 2.0 | 3.0 |
g/cm3 | 1.8 | 2.8 |
g/cm3 | 0.0 | 0.3 | fluid density - gas
g/cm3 | 0.6 | 0.9 | fluid density - oil
g/cm3 | 0.0 | 1.1 | fluid density - brine and
brine/hc mix
kg/m3 | 2000 | 3000 |
kg/m3 | 1800 | 2800 |
kg/m3 | 0 | 300 | fluid density - gas
kg/m3 | 600 | 900 | fluid density - oil
kg/m3 | 0 | 1100 | fluid density - brine and
brine/hc mix
Well Seismic Fusion Landmark
D-14 Graphical log scaling in Fusion Release 5000.0.0.1
This is intended to be a general method that does not need to know
about the 'Ilavor oI a log curve oI a given unit, such as compressional
velocity vs. shear velocity. As long as appropriate log scales have been
deIined, the two types oI velocity logs should Iall into distinct (but
consistent) log scale ranges. The multiple log scale ranges can also be
used to handle data Irom diIIerent geologic areas. For example, sonic
log scales Ior the hardrock clastic/carbonate Western Canadian basin
vs. scales more appropriate Ior the slower sand/shale sequences in the
GOM.
The algorithm Ior Iinding the best log scale uses the Iollowing steps:
1. Find all oI the log scales (Ior the given unit) that hold at least 98
oI the curve data. This allows Ior up to 2 bad non-null log values.
(Null values are accounted Ior and ignored in the calculation oI the
percent oI values within the log scale range.)
2. OI the log scales that hold at least 98 oI the curve values, Iind the
one that has the smallest range (max log scale value - min log scale
value). This is the keeper.
3. II no log scales hold at least 98 oI the curve values, then Iind the
one that holds the highest percentage. II the best log scale holds at
least 80 oI the curve values, then use it. The number to use Ior
this is debatable. Ninety percent might be better. In any case, it's
useIul to allow Ior some oII-scale values to make it more likely that
raw and edited versions oI a log will have the same log scale.
4. II no log scales hold at least 80 oI the curve values, then return
null. This drops the log scale Iinding down to the third method -
calculate rounded log scales based on min and max curve values.
Landmark Well Seismic Fusion
Release 5000.0.0.1 Graphical log scaling in Fusion D-15
Examples:
LFPRHOBLOG (Fusion Hoot well log)
Only one log scale (1.8 to 2.8) holds more than 98 oI the curve
values, so this is the one that is used. II this log scale was not
included in the Ilat Iile, then the standard log scale (2.0 to 3.0)
would be taken as acceptable.
LFPVPLOG
Two log scales hold all oI the curve data. OI these, the scale with
the smallest range is 2400 to 5000, so this is the one that is used.
Log scales for ~gm/c3 Log scale range Percent of values in range
2.0 to 3.0 1.0 96.3
1.8 to 2.8 1.0 99.8
1.5 to 2.5 1.0 93.3
Log scales for ~m/s Log scale range Percent of values in range
2000 to 7000 5000 100
2400 to 5000 2600 100
1200 to 2400 1200 0
Well Seismic Fusion Landmark
D-16 Graphical log scaling in Fusion Release 5000.0.0.1
Method 3 - Rounded ScaIes (easiIy divisibIe)
For any magnitude oI log values, this method Iinds a log scale that
satisIies these conditions:
The scale Iully brackets the data (including any bad non-null
values). Currently looks at min and max log values only, excluding
the null value.
The log scale range can be divided into 10 increments where each
increment is a power-oI-ten multiple oI a number Iound in a set oI
hardcoded 'base' increment values.
The leIt and right log scale values are evenly divisible by the same
increment.
The set oI 'base log scale increments currently used in Fusion is 10,
20, 50. This means that any log scale Iound by this method can be
divided into 10 intervals where the interval range is in the set 0.001,
0.002, 0.005, 0.01, 0.02, 0.05, 0.1, 0.2, 0.5, 1.0, 2.0, 5.0, 10, 20, 50,
100, 200, 500, 1000, 2000, 5000, . The smallest log scale increment
(one tenth oI the smallest log scale range) is 0.001 because the viewer
only displays log scales to 2 decimal places.
Examples:
The set oI 'base log scale increments to use is subjective. This method
is borrowed Irom LogM, which uses a much larger set: 10, 20, 25, 30,
40, 50, 60, 75, 80. Although a larger set allows Ior tighter bracketing
oI curve values, it is more diIIicult to estimate log values visually when
the log scale increment (on 10 divisions oI the log scale) is 0.3 or 7.5,
Ior example.
Min, max curve values Log scale Log scale increment (10
divisions)
(-0.0043, -0.0004) (-0.01, 0.0) 0.001
(651.013, 651.026) (651.01, 651.03) 0.002
(0.101, 0.412) (0.1, 0.6) 0.05
(121.5, 251.3) (100, 300) 20
(-29.3, 19.2) (-30, 20) 5
(-639, 43) (-700, 300) 100
(2.7513E12, 2.7581E12) (2.75E12, 2.76E12) 1000000 (e.g. brine bulk
modulus in Pa)
Landmark Well Seismic Fusion
Release 5000.0.0.1 Graphical log scaling in Fusion D-17
Discussion
This is a 'Iirst release solution to the problem oI Iinding consistent log
scales Ior all curves displayed in Fusion, whether computed or not. It is
expected that this will be changed again aIter Iirst release when there
will be more time to investigate the 'ultimate solution that should
include using log scale inIormation stored in the OpenWorks Curve
Dictionary.
Some things to be aware oI:
Log scales are Iound each time a curve is displayed by the Well
Seismic Viewer. This includes edited versions oI curves, where the
new log scale may be diIIerent than the log scale set Ior the original
curve (or Ior the edited curve created by a previous Apply) iI the
edit changes the range oI curve values so that it doesn't Iit the same
scale. This behavior depends on how tightly the log scales Iit the
data (should be 'loose enough to allow Ior reasonable edits) and
whether or not the log scales deIined Ior a unit are distinct enough.
For example, the log scales deIined Ior 'g/cm3 in the test version
oI the FusionUnitLogScale.dat Iile are poor because there are two
scales - 1.8 to 2.8 and 2.0 to 3.0 - that are nearly identical. Ordinary
edits can easily cause the log scales to Ilip Irom one to the other.
When really bad edits are made to a log, then it's useIul to see the
scales change, since this can Ilag the bad edit to the user. II the
scales don't change, then bad edits just 'max out at either the leIt
or right side oI the track and there's no indication oI how bad it is.
The curve names method (method #1) can be used to set Iixed
scales based on curve name iI scales are jumping around too much
on the edited versions oI logs. However, it's probably better to
adjust the scales deIined Ior units (method #2) than to add a lot oI
curve names.
Units added to the FusionUnitLogScale.dat Iile must match a unit
abbreviation in the list oI units Iound in units.dat in inIra/units,
otherwise they will be ignored. The optional comment is ignored
when scales are read Irom the FusionUnitLogScale.dat Iile.
The rounded scales method (method #3) was included as a Iall-back
method because it was available Irom LogM. It can use some minor
adjustments so that curve data is centered better within the log
scales. It can be removed iI min/max scales are preIerred as a Iall-
back method.
Well Seismic Fusion Landmark
D-18 Graphical log scaling in Fusion Release 5000.0.0.1
Landmark Well Seismic Fusion
Release 5000.0.0.1 ndex - 1
A
Acoustic impedance 0-25, A-15
active well 1-7
amplitude balancing 5-29
Angle gathers 5-8
anisotropy 0-18
attribute
AVO 5-9, 6-30
gradient 6-16
horizon 4-38
intercept 6-15
LMR 4-53
seismic 6-11
sensitivity 6-30
unconsolidated 0-41
velocity B-2
volumes 6-11
AVO
deIinition 0-3
Iluid stack 6-13
gradient 0-3
intercept 0-3
inversion 6-45
lithology stack 6-13
B
Backus averaging 2-11, D-9
bad logs 2-4
Biot-Gassmann 0-36, A-10
bright-spot 0-8
Bulk density 0-20
Bulk modulus 0-36
C
carbonate rocks 0-42, 2-21, 2-29
cemented rocks 0-15
Check shot correction 3-5
Check shot surveys 3-5
Check shots 3-5
Compressional velocity
See P-wave velocity 0-4
consolidated and unconsolidated rocks 0-41
crossplot
AI vs GI 0-48
AI vs Poisson`s ratio 4-39
amplitude vs angle 0-46
amplitude vs oIIset 4-11
background trend 6-7
Iluid direction 4-10
horizon vs horizon 4-38
intercept vs gradient 0-47, 4-18
inversion 6-45
LMR A-22
log vs log 4-3
NMO sensitivity 5-32
polygon classiIication 6-40
seismic background 6-3
seismic vs seismic 4-22
variance 4-34
Vp vs Vs 0-13
wavelet phase sensitivity 5-28
D
Data Format A-26, A-29
Data Save A-29
Data Type A-26, A-29
Datum
reIerence 3-6
seismic 3-6
SeisWorks 3-6
Decimating 5-6
DecisionSpace 0-52
Density
bulk 0-20
empirical relations A-12
Depth error B-14
divergence 5-29
DMO B-8
dynamic boundary condition A-2
E
EEI
See Extended elastic impedance A-17
eIIective medium theory 2-11
Elastic impedance 0-26, A-15
elastic properties 0-4
Extended elastic impedance
deIinition A-17
relationships A-18
extracted wavelet B-15
Index
Well Seismic Fusion Landmark
- 2 ndex Release 5000.0.0.1
F
Iluid Iactor 6-17
Iluid rotation angle 6-14
Iluid stack 6-13, 6-49
Fluid substitution
Robust estimation C-8
Shaly rocks C-21
using Gassmann`s equation C-5
workIlow 2-3, 2-18
Formation velocity 0-15
Irequency matching 5-25
Fresnel zone 0-25
Iundamental elastic constants 4-53
Iundamental rock properties
Iundamental elastic constants 0-35
G
gas eIIect 0-9
Gassmann`s equation C-5
gather Ilattening 5-20
geopressure 0-40
Geostack technique 6-4
gradient
gradient 6-16
See AVO gradient 0-3
Gradient impedance A-17
H
hard rock 0-41
Hooke`s law 0-35
I
Incompressibility (bulk modulus) 0-36
intercept
intercept 6-16
See AVO intercept 0-3
inverse Q-Iilter 5-25
inversion A-7
AVO 6-45
seismic 6-45
iso-angle mutes 5-7
K
kinematic boundary conditions A-2
KoeIoed`s observation A-5
L
Lambda/mu stack A-24
Lambda-rho stack A-24
Lame constant 0-38
LGCDATAHOME directory 1-5
lithology Iamilies 4-5
lithology Iamily 2-5
lithology stack 6-13, 6-49
lithology substitution 2-29
log blocking
See also Backus averaging 2-11
log display scaling D-11
M
matching 5-1
modeling
Iluid substitution 2-3, 2-18
oIIset synthetic seismograms 9-5
porosity substitution 2-29
zero oIIset synthetic seismograms 4-22, 4-35, 4-38
modulus 0-35
Mudrock line 0-10
Mu-rho stack A-24
mute 5-7
N
NMO B-2
NMO stretch 5-25
Non-Hyperbolic Moveout B-12
Normalized elastic impedance A-16
O
oIIset synthetic seismograms 9-5
oIIset/angle stacks 6-36, 6-40
overpressured 0-15
P
partial stacking 5-6
phase matching 5-25
Picking blocking boundaries D-3
Poisson reIlectivity 6-11
Poisson`s ratio 0-28
Polarity convention 3-9
Porosity 0-26
porosity substitution 2-29
Pressure 0-39
Pseudo-Poisson`s Ratio ReIlectivity 6-11
P-wave velocity
Well Seismic Fusion Landmark
- 3 ndex Release 5000.0.0.1
deIinition 0-5
Estimation C-16
reIlectivity 6-4
Q
Q-Iilter 5-25
R
reIerence datum 3-6
reIlection coeIIicient 0-25
ReIlectivity A-15
Rigidity (shear modulus) 0-34
RNMO
residual moveout 5-20
S
Saved session directory 1-5
seismic inversion 6-45
Seismic processing
amplitude balancing 5-29
Decimating 5-6
Irequency matching 5-25
gather Ilattening 5-20
iso-angle mutes 5-7
oIIset/angle stacks 6-36, 6-40
partial stacking 5-6
phase matching 5-25
super-gathers 5-7
weighted stacks 6-34
Seismic velocity B-2
SeisWorks Datum 3-6
Shear modulus 0-34
Shear-wave velocity
deIinition 0-10
Dry rock C-13
Estimation C-10
reIlectivity 6-4
Spencer`s Iormula C-13
Wet rock C-11
Shuey`s equation
2-term Shuey approximation A-5
3-term Shuey approximation A-5
Shuey-Hilterman approximation A-6
soIt rock 0-41
Spencer`s Iormula
See Also Shear-wave velocity C-13
spherical divergence 5-29
Stacking velocity B-11
stretch and squeeze your curves 3-3
super-gathers 5-7
synthetic seismograms
oIIset 9-5
zero oIIset 4-22, 4-35, 4-38
T
Temperature 0-41
time-depth corrections 3-4
tuning equation 3-29
U
USERPROFILE directory 1-5
V
Velocity
accuracy B-13
empirical relations A-10
Iormation 0-15
Irequency 0-16
seismic B-2
stacking velocity B-11
virgin zone 2-1
Vshale substitution 2-29
VSP`s 3-5
W
wavelet extraction B-15
weighted stacks 6-34
well calibration 3-2
Well Seismic Fusion 0-1
Wood`s equation A-10
Wyllie time-average 0-26
Wyllie`s equation 0-21
Z
zero oIIset synthetic seismograms 4-22, 4-35, 4-38
Zoeppritz equations 0-3, A-3
Well Seismic Fusion Landmark
- 4 ndex Release 5000.0.0.1
Landmark Notes N - 1
!""#$%&'
Notes
0
Well Seismic Fusion Landmark
N - 2 Notes Well Seismic Fusion 5000.0.0.1
Landmark Well Seismic Fusion
Well Seismic Fusion 5000.0.0.1 Notes N - 3
Well Seismic Fusion Landmark
N - 4 Notes Well Seismic Fusion 5000.0.0.1
Landmark Well Seismic Fusion
Well Seismic Fusion 5000.0.0.1 Notes N - 5
Well Seismic Fusion Landmark
N - 6 Notes Well Seismic Fusion 5000.0.0.1

You might also like