J. Ind. Geophys.

Union ( November
Comparing 2018
pre- and ) stack seismic inversion methods - a case study from Scotian Shelf, Canada
post-
v.22, no.6, pp: 585-597

Comparing pre- and post- stack seismic inversion methods -

a case study from Scotian Shelf, Canada
S.P. Maurya* and N.P. Singh
1
Department of Geophysics, Institute of Science, Banaras Hindu University Varanasi-221005, U.P., India
*Corresponding author: spm.bhu@gmail.com

ABSTRACT
In the present study, pre- and post-stack seismic inversions are performed to estimate subsurface petrophysical
properties from the seismic reflection data, obtained over the Penobscot field, Canada. Such methods
are routinely used to estimate subsurface property from seismic reflection data but the problem arises in
deciding which method will provide more efficient and detailed information of the subsurface structures.
Simple procedure is used here at the every step of seismic inversion methods, differences have been taken
between pre- and post- stack seismic inversion. It is broadly applied in two steps; first, they are applied to
the composite trace near to well location and inverted to obtain petrophysical parameters, which are then
compared with the well log data. The analysis shows that both curves (inverted and well log) are in good
agreement and display high correlation. The average correlation of post stack inversion (0.92), is higher
than the pre-stack seismic inversion method (0.89). In the second step, entire seismic section is inverted
and P- impedance, S- impedance, density, P- wave velocity, S- wave velocity and VP/VS ratio are estimated.
The post stack inversion methods provide only acoustic impedance which is transformed into the other
petrophysical parameters, using relationship derived from the well log data. These parameters are helpful
in interpreting seismic sections. The analysis shows that both (pre- and post- stack) methods produces
subsurface property in similar way, with high resolution from post-stack inversion, compared to the pre-
stack inversion. Although the pre-stack inversion method produces more petrophysical parameters that are
sensitive towards fluid and rock property, but the resolution is relatively poorer compared to the post-stack
inversion methods. The interpreted inverted section indicates that the studied area does not have any major
hydrocarbon accumulation.

Keywords: Pre- and Post- stack seismic inversion, Well log analysis, Amplitude spectrum, Petrophysical
properties, Scotian shelf Canada

INTRODUCTION
Seismic inversion is a technique that estimates subsurface
petrophysical properties from seismic reflection data with
the integration of well log data. These petrophysical
parameters help to interpret acquired seismic section (Krebs
et al., 2009). In hydrocarbon industry, seismic inversion
techniques are used as a tool to locate hydrocarbon-bearing
strata at subsurface depths (Morozov and Ma, 2009;
Lindseth, 1979).The physical parameters that are of interest
to a modeler performing inversions are impedance (Z),
P-wave (VP) and S-wave (VS) velocity, apart from density.
Lame parameters that are more sensitive towards fluid
and saturation in rocks (Clochard et al., 2009) can also
be derived from inverted impedance. Similarly, the other
petrophysical parameters like porosity, sand/shale ratio
and gas saturation too can be estimated with the help of
inverted volumes (Goodway, 2001). A reliable estimate
of the reservoir properties is critical in decision-making
process, especially during developmental phase (Pendrel,
2006).
Basically, the seismic inversion techniques can be
divided into two broad categories, post-stack and pre-stack
inversion. The first approach is the most commonly used,
where the effect of the wavelet is removed from the seismic
data and high-resolution images of the subsurface are
produced (Chen and Sidney, 1997). The second approach
relies mainly on model building from well log, seismic
and geological data (Downton, 2005). This method also
generates high-resolution images of the subsurface
The advantages of post-stack inversion are many.
For example, (i) stratigraphic interpretation is easier on
impedance data than seismic data, (ii) the reduction
of wavelet effects, side lobes and tuning enhances the
resolution of subsurface layers, (iii) the acoustic impedance
can be directly computed and compared to well log
measurements, that serve as a link to reservoir properties,
(iv) porosity can be related to the acoustic impedance using
geostatistical methods, these impedance volumes can be
transformed into the porosity volumes, (v) The acoustic
impedance can be utilized to locate individual reservoir
regions, (vi) it takes very less time than pre-stack inversion,
and (vii) it does not give shear wave information to
discriminate the fluid effects (Russell, 1999; Morozov and
Ma, 2009; Maurya and Singh, 2015; Maurya et al., 2018a).
Similarly, the elastic properties of the subsurface
such as S- wave velocity of the subsurface layers, which
are sensitive to the fluid saturation can be obtained from

585
S.P. Maurya and N.P. Singh
Figure 1. Flowchart of seismic inversion methods.
pre-stack inversion (Moncayo et al., 2012). This method
transforms seismic reflection data into P- impedance,
S-impedance, density and Vp/Vs ratio sections, through
the integration of well log data and horizon information
from seismic reflection data. P- impedance and VP/VS ratio
are considered reliable, depending on the target depth and
acquisition configuration, and thus can be used to predict
reservoir properties away from well locations (Carrazzone et
al., 1996). Pre-stack seismic inversion also provides several
benefits that include (i) P- impedance, S- impedance and
density which is layer property, whereas seismic data is an
interface property, (ii) enhancement of the resolution of
sub-surface layers due to reduction of wavelet effects, tuning
and side lobes, (iii) acoustic impedance can be directly
compared to well log measurements, which in turn are
linked to the reservoir properties, and (iv) compared with
other inversion techniques (e.g. post-stack inversion), the
data offers additional information to distinguish between
lithology and fluid effects (Gholami, 2016).
Further, wavelet is very crucial part of seismic inversion
methods. During seismic inversion, one integrates seismic
and well log data together, but both are in different domain
(one is in time and other is in depth) and hence needed to
convert in same domain. For that reason, wavelet is extracted
from the seismic section and the process used includes (i)
extraction of the analysis window in seismic section, (ii)
applying Taper at the start and end of the seismic window,
(iii) calculate the autocorrelation of the data window, (iv)
586
calculate the amplitude spectrum of the autocorrelation, (v)
calculate the square root of the autocorrelation spectrum.
This process approximates the amplitude spectrum of the
wavelet, (vi) add phase of the wavelet, and then finally, the
inverse Fast Fourier Transform gives the desired wavelet
(Vardy et al., 2018, Maurya et al., 2018b).
In the present study, comparison of pre- and post-
stack seismic inversion methods have been performed and
their relative performance is estimated in each intermediate
steps. For the analysis, pre-stack gather from the Penobscot
region, is used along with one well (L-30). The inversion
is first tested for composite trace near to well location
and then performed to the entire gathers. A variety of
attributes i.e. P- impedance, S- impedance, P- wave, S- wave
velocity, density and VP/VS ratio is estimated in inter-
well region from pre-stack inversion methods, while only
P- impedance is estimated from the post-stack inversion
methods. Thereafter, the P-impedance from the post-
stack inversion method is transformed into S- impedance,
P- wave, S- wave velocity, density and VP/VS ratio, using
relationship derived from well log data. Thereafter, point by
point comparison is performed between attributes derived
from pre- and post-stack seismic inversion methods. The
inversion is performed using Hampson Russell software
(10.2) and the analysis of the inversion results are carried
out using Matlab programming language (Matlab, 2015b).
Figure 1 shows flowchart of seismic inversion methods
used in this study.
 Comparing pre- and post- stack seismic inversion methods - a case study from Scotian Shelf, Canada

Figure 2. Map showing Penobscot study area highlighted by red rectangle.

STUDY AREA which include graben-bounded faults, channels, mega-scale

de-watering faults and reefs (Kidston et al., 2007).
The present study area Penobscot lies in the Scotian shelf
(Figure 2) in offshore Nova Scotia, Canada. The survey METHODOLOGY
contains several structural and stratigraphic features. Two
major faults within the survey area that form an echelon Pre-stack inversion
pattern (Campbell et al., 2015), can be seen in the seismic
sections. This fault mostly displaces the interpreted The pre-stack inversion technique utilizes pre-stack CMP
surfaces horizons. The seismic attributes, particularly gather to estimate P- impedance (ZP), S- impedance (ZS),
multi-trace similarity, volumetric curvature or volumetric density (ρ) and VP/VS ratio. The P- and S- impedance are
amplitude highlight these features more clearly (Cummings related to each other in wet clastic rocks and also according
and Arnott, 2005). The measured seismic signal is poor to Castagna equation (Eq. 1), the P- wave velocity is related
below 3.0 seconds in length (about 5km). Two wells, L-30 to the S- wave velocity (Castagna et al., 1985) in following
and B-41, were drilled in the area penetrating Misaine way.
and Baccaro members of the Jurassic Abenaki Formation 𝑉𝑉𝑉𝑉𝑝𝑝𝑝𝑝 = 1.16𝑉𝑉𝑉𝑉𝑠𝑠𝑠𝑠 + 1360 (1) (1)
(Kidston et al., 2005). The data used in this study contains Further, according
𝜌𝜌𝜌𝜌 = 0.23𝑉𝑉𝑉𝑉 to0.25 the Gardner equation (Eq. 2), the (2)
inline 1161-1200 and cross-line 1000-1481 with one P-wave velocity is related to the density (Gardner et al.
𝑉𝑉𝑉𝑉
𝑝𝑝𝑝𝑝 =𝑆𝑆𝑆𝑆1.16𝑉𝑉𝑉𝑉
ln(𝑍𝑍𝑍𝑍 𝑠𝑠𝑠𝑠 + 1360
) = 𝑘𝑘𝑘𝑘ln(𝑍𝑍𝑍𝑍𝑃𝑃𝑃𝑃 ) + 𝑘𝑘𝑘𝑘𝐶𝐶𝐶𝐶 + ∆𝐿𝐿𝐿𝐿𝑆𝑆𝑆𝑆
(1)
(3)
well (L-30). Several oil and gas fields are located in this 1974) as
Late Jurassic to Cretaceous intervals. The Lower Logan 0.25
𝑉𝑉𝑉𝑉𝑝𝑝𝑝𝑝 = 1.16𝑉𝑉𝑉𝑉𝜌𝜌𝜌𝜌
ln⁡ = 0.23𝑉𝑉𝑉𝑉
(𝜌𝜌𝜌𝜌)
𝑠𝑠𝑠𝑠 + 1360
= 𝑚𝑚𝑚𝑚ln(𝑍𝑍𝑍𝑍𝑃𝑃𝑃𝑃 ) + 𝑚𝑚𝑚𝑚 𝐶𝐶𝐶𝐶 + ∆𝐿𝐿𝐿𝐿𝐷𝐷𝐷𝐷 (2)
(1)
(2) (4)
Canyon sands in L-30 were considered oil bearing (minor Therefore,ln(𝑍𝑍𝑍𝑍 pre-stack seismic inversion uses these
∆𝑉𝑉𝑉𝑉𝑃𝑃𝑃𝑃 𝑃𝑃𝑃𝑃 ) + 𝑘𝑘𝑘𝑘
0.25 𝑆𝑆𝑆𝑆 ) = 𝑘𝑘𝑘𝑘ln(𝑍𝑍𝑍𝑍
∆𝑉𝑉𝑉𝑉𝑆𝑆𝑆𝑆+ ∆𝐿𝐿𝐿𝐿𝑆𝑆𝑆𝑆 ∆𝜌𝜌𝜌𝜌 (3)
accumulation) (Smith and Gidlow 1987). 𝜌𝜌𝜌𝜌 = 0.23𝑉𝑉𝑉𝑉 𝑅𝑅𝑅𝑅(𝜃𝜃𝜃𝜃)
relationships = 𝑐𝑐𝑐𝑐1 following
to derive + 𝑐𝑐𝑐𝑐2 𝐶𝐶𝐶𝐶equations.
+ 𝑐𝑐𝑐𝑐3 (2) (5)
ln⁡ 𝑉𝑉𝑉𝑉
(𝜌𝜌𝜌𝜌) = 𝑚𝑚𝑚𝑚ln(𝑍𝑍𝑍𝑍 𝑉𝑉𝑉𝑉 𝜌𝜌𝜌𝜌
𝑃𝑃𝑃𝑃 ) + 𝑚𝑚𝑚𝑚𝑆𝑆𝑆𝑆 + ∆𝐿𝐿𝐿𝐿 (4)
The seismic and well data have been downloaded 𝑃𝑃𝑃𝑃 𝐶𝐶𝐶𝐶 𝐷𝐷𝐷𝐷
ln(𝑍𝑍𝑍𝑍𝑆𝑆𝑆𝑆 ) = 𝑘𝑘𝑘𝑘ln(𝑍𝑍𝑍𝑍𝑃𝑃𝑃𝑃 ) + 𝑘𝑘𝑘𝑘𝐶𝐶𝐶𝐶 + ∆𝐿𝐿𝐿𝐿 𝑆𝑆𝑆𝑆 (3)
from the OpendTect seismic data portal. The time to ∆𝑉𝑉𝑉𝑉𝑃𝑃𝑃𝑃 1 ∆𝑉𝑉𝑉𝑉𝑆𝑆𝑆𝑆 ∆𝜌𝜌𝜌𝜌 𝑉𝑉𝑉𝑉𝑆𝑆𝑆𝑆 2 (3)2
depth models have been taken from the CNSOPB report ln⁡ where
𝑅𝑅𝑅𝑅(𝜃𝜃𝜃𝜃)𝑃𝑃𝑃𝑃=
(𝜌𝜌𝜌𝜌) = 𝑚𝑚𝑚𝑚ln(𝑍𝑍𝑍𝑍 )+𝑐𝑐𝑐𝑐 =
1𝑐𝑐𝑐𝑐1𝑚𝑚𝑚𝑚𝐶𝐶𝐶𝐶 + + ∆𝐿𝐿𝐿𝐿2𝑐𝑐𝑐𝑐𝐷𝐷𝐷𝐷𝜃𝜃𝜃𝜃2 , 𝑐𝑐𝑐𝑐
+ 𝑐𝑐𝑐𝑐23 = 0.5 − � � 𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠
(4) 𝑠𝑠𝑠𝑠 𝜃𝜃𝜃𝜃 and 𝑐𝑐𝑐𝑐(5)
(4) 3 = 4�
𝑉𝑉𝑉𝑉2𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑠𝑠𝑠𝑠
𝑃𝑃𝑃𝑃 𝑉𝑉𝑉𝑉𝑆𝑆𝑆𝑆 𝜌𝜌𝜌𝜌 𝑉𝑉𝑉𝑉𝑃𝑃𝑃𝑃
(Kidston et al., 2005). The survey contains a limited range where coefficients ∆𝑉𝑉𝑉𝑉𝑃𝑃𝑃𝑃 (𝑘, ∆𝑉𝑉𝑉𝑉 𝑘𝑐,𝑆𝑆𝑆𝑆𝑚 𝑎𝑛𝑑 ∆𝜌𝜌𝜌𝜌 𝑚𝑐) are calculated using well
of pre-stack data. The data covers many geological features, 𝑅𝑅𝑅𝑅(𝜃𝜃𝜃𝜃)
log cross 𝑐𝑐𝑐𝑐1𝑇𝑇𝑇𝑇(𝜃𝜃𝜃𝜃)
= plot and +=DL 𝑐𝑐𝑐𝑐2𝐶𝐶𝐶𝐶S1and
𝑊𝑊𝑊𝑊(𝜃𝜃𝜃𝜃)+DL 1𝑐𝑐𝑐𝑐∗3D𝐷𝐷𝐷𝐷𝐿𝐿𝐿𝐿
represent 𝑃𝑃𝑃𝑃 + 𝐶𝐶𝐶𝐶2 𝑊𝑊𝑊𝑊(𝜃𝜃𝜃𝜃) the deviation
∗ 𝐷𝐷𝐷𝐷𝐿𝐿𝐿𝐿𝑉𝑉𝑉𝑉𝑆𝑆𝑆𝑆 +2from
𝑆𝑆𝑆𝑆
𝐶𝐶𝐶𝐶3(5) 𝑊𝑊𝑊𝑊(𝜃𝜃𝜃𝜃) ∗ 𝐷𝐷𝐷𝐷𝐿𝐿𝐿𝐿𝐷𝐷𝐷𝐷
where 𝑉𝑉𝑉𝑉𝑃𝑃𝑃𝑃 𝑐𝑐𝑐𝑐1 =𝑉𝑉𝑉𝑉𝑆𝑆𝑆𝑆 𝜌𝜌𝜌𝜌 , 𝑐𝑐𝑐𝑐2 = 0.5 − � � 𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠2 𝜃𝜃𝜃𝜃 and 𝑐𝑐𝑐𝑐3 = 4 �
𝐿𝐿𝐿𝐿𝑃𝑃𝑃𝑃 = ln(𝑍𝑍𝑍𝑍𝑃𝑃𝑃𝑃 ) ,2𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑠𝑠𝑠𝑠 𝐿𝐿𝐿𝐿𝑆𝑆𝑆𝑆 = 2 𝜃𝜃𝜃𝜃 ln(𝑍𝑍𝑍𝑍 ) , and 𝐿𝐿𝐿𝐿 = ln(𝜌𝜌𝜌𝜌). 𝑉𝑉𝑉𝑉𝑃𝑃𝑃𝑃
𝑆𝑆𝑆𝑆 𝐷𝐷𝐷𝐷
1 𝑉𝑉𝑉𝑉𝑆𝑆𝑆𝑆 2 2
𝑉𝑉𝑉𝑉𝑆𝑆𝑆𝑆 2
where 𝑐𝑐𝑐𝑐1 =𝑇𝑇𝑇𝑇(𝜃𝜃𝜃𝜃) = 12 𝐶𝐶𝐶𝐶1,𝑊𝑊𝑊𝑊(𝜃𝜃𝜃𝜃) 1 𝑐𝑐𝑐𝑐2∗ = 𝐷𝐷𝐷𝐷𝐿𝐿𝐿𝐿0.5 𝑃𝑃𝑃𝑃 +−𝐶𝐶𝐶𝐶� 2 𝑊𝑊𝑊𝑊(𝜃𝜃𝜃𝜃) 1
� 𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠∗𝑠𝑠𝑠𝑠𝐷𝐷𝐷𝐷𝐿𝐿𝐿𝐿
𝜃𝜃𝜃𝜃 𝑆𝑆𝑆𝑆and + 𝐶𝐶𝐶𝐶𝑐𝑐𝑐𝑐3 3𝑊𝑊𝑊𝑊(𝜃𝜃𝜃𝜃)
= 4 �∗ 𝐷𝐷𝐷𝐷𝐿𝐿𝐿𝐿 � 𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠2 𝜃𝜃𝜃𝜃
= 𝑐𝑐𝑐𝑐𝜃𝜃𝜃𝜃1 + 𝐾𝐾𝐾𝐾𝑐𝑐𝑐𝑐2 + 𝑚𝑚𝑚𝑚𝑐𝑐𝑐𝑐3 , 𝑉𝑉𝑉𝑉𝑃𝑃𝑃𝑃 𝐶𝐶𝐶𝐶2 = 𝑐𝑐𝑐𝑐2 , 587
𝐶𝐶𝐶𝐶12𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑠𝑠𝑠𝑠 𝑊𝑊𝑊𝑊(𝜃𝜃𝜃𝜃) 𝑉𝑉𝑉𝑉𝑃𝑃𝑃𝑃 𝐷𝐷𝐷𝐷
2 2 2
𝐿𝐿𝐿𝐿𝑃𝑃𝑃𝑃 = ln(𝑍𝑍𝑍𝑍 𝑃𝑃𝑃𝑃 ) , 𝐿𝐿𝐿𝐿𝑆𝑆𝑆𝑆 = ln(𝑍𝑍𝑍𝑍𝑆𝑆𝑆𝑆 ) , and 𝐿𝐿𝐿𝐿𝐷𝐷𝐷𝐷 = ln(𝜌𝜌𝜌𝜌).
𝑇𝑇𝑇𝑇(𝜃𝜃𝜃𝜃) = 𝐶𝐶𝐶𝐶1 𝑊𝑊𝑊𝑊(𝜃𝜃𝜃𝜃)𝑇𝑇𝑇𝑇(𝜃𝜃𝜃𝜃1 )∗ 𝐷𝐷𝐷𝐷𝐿𝐿𝐿𝐿𝑃𝑃𝑃𝑃 𝐶𝐶𝐶𝐶+1 (𝜃𝜃𝜃𝜃 𝐶𝐶𝐶𝐶2 𝑊𝑊𝑊𝑊(𝜃𝜃𝜃𝜃)
)𝑊𝑊𝑊𝑊(𝜃𝜃𝜃𝜃 ∗)𝐷𝐷𝐷𝐷 𝐷𝐷𝐷𝐷𝐿𝐿𝐿𝐿𝑆𝑆𝑆𝑆 + ⋯ 𝐶𝐶𝐶𝐶3 𝑊𝑊𝑊𝑊(𝜃𝜃𝜃𝜃)
𝐶𝐶𝐶𝐶 ∗ 𝐷𝐷𝐷𝐷𝐿𝐿𝐿𝐿𝐷𝐷𝐷𝐷 (6)
3 (𝜃𝜃𝜃𝜃3 )𝑊𝑊𝑊𝑊(𝜃𝜃𝜃𝜃3 )𝐷𝐷𝐷𝐷
1 1 1 1
1
𝐿𝐿𝐿𝐿𝑃𝑃𝑃𝑃 = ln(𝑍𝑍𝑍𝑍𝑃𝑃𝑃𝑃�𝐶𝐶𝐶𝐶)1, =⋮𝐿𝐿𝐿𝐿𝑆𝑆𝑆𝑆 = �𝑐𝑐𝑐𝑐1=ln(𝑍𝑍𝑍𝑍
+� 𝑆𝑆𝑆𝑆𝐾𝐾𝐾𝐾𝑐𝑐𝑐𝑐 ) ,2and +⋮ 𝑚𝑚𝑚𝑚𝑐𝑐𝑐𝑐 𝐿𝐿𝐿𝐿𝐷𝐷𝐷𝐷3= , ln(𝜌𝜌𝜌𝜌). ⋱2 = 𝑐𝑐𝑐𝑐2 , ⋮ 𝑊𝑊𝑊𝑊(𝜃𝜃𝜃𝜃)�
𝐶𝐶𝐶𝐶
𝑇𝑇𝑇𝑇(𝜃𝜃𝜃𝜃𝑁𝑁𝑁𝑁2) 𝐶𝐶𝐶𝐶21 (𝜃𝜃𝜃𝜃𝑁𝑁𝑁𝑁 )𝑊𝑊𝑊𝑊(𝜃𝜃𝜃𝜃𝑁𝑁𝑁𝑁 )𝐷𝐷𝐷𝐷 ⋯ 𝐶𝐶𝐶𝐶32(𝜃𝜃𝜃𝜃𝑁𝑁𝑁𝑁 )𝑊𝑊𝑊𝑊(𝜃𝜃𝜃𝜃𝑁𝑁𝑁𝑁 )𝐷𝐷𝐷𝐷
 1 2 3
𝑉𝑉𝑉𝑉𝑃𝑃𝑃𝑃 𝑉𝑉𝑉𝑉𝑆𝑆𝑆𝑆 𝜌𝜌𝜌𝜌
1 𝑉𝑉𝑉𝑉𝑆𝑆𝑆𝑆 2 𝑉𝑉𝑉𝑉𝑆𝑆𝑆𝑆 2
where 𝑐𝑐𝑐𝑐1 = , 𝑐𝑐𝑐𝑐2 = 0.5 − � � 𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠2 𝜃𝜃𝜃𝜃 and 𝑐𝑐𝑐𝑐3 = 4 � � 𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠2 𝜃𝜃𝜃𝜃
2𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑠𝑠𝑠𝑠 2 𝜃𝜃𝜃𝜃 𝑉𝑉𝑉𝑉𝑃𝑃𝑃𝑃 𝑉𝑉𝑉𝑉𝑃𝑃𝑃𝑃
𝑇𝑇𝑇𝑇(𝜃𝜃𝜃𝜃) = 𝐶𝐶𝐶𝐶1 𝑊𝑊𝑊𝑊(𝜃𝜃𝜃𝜃) ∗ 𝐷𝐷𝐷𝐷𝐿𝐿𝐿𝐿𝑃𝑃𝑃𝑃 + 𝐶𝐶𝐶𝐶2 𝑊𝑊𝑊𝑊(𝜃𝜃𝜃𝜃) ∗ 𝐷𝐷𝐷𝐷𝐿𝐿𝐿𝐿𝑆𝑆𝑆𝑆 + 𝐶𝐶𝐶𝐶3 𝑊𝑊𝑊𝑊(𝜃𝜃𝜃𝜃) ∗ 𝐷𝐷𝐷𝐷𝐿𝐿𝐿𝐿𝐷𝐷𝐷𝐷 (6)
𝑉𝑉𝑉𝑉 = 1.16𝑉𝑉𝑉𝑉 + 1360
𝑝𝑝𝑝𝑝 (1)
S.P. Maurya𝑠𝑠𝑠𝑠 and N.P. Singh 𝐿𝐿𝐿𝐿𝑃𝑃𝑃𝑃 = ln(𝑍𝑍𝑍𝑍𝑃𝑃𝑃𝑃 ) , 𝐿𝐿𝐿𝐿𝑆𝑆𝑆𝑆 = ln(𝑍𝑍𝑍𝑍𝑆𝑆𝑆𝑆 ) , and 𝐿𝐿𝐿𝐿𝐷𝐷𝐷𝐷 = ln(𝜌𝜌𝜌𝜌).
0.25
𝜌𝜌𝜌𝜌 =
𝑉𝑉𝑉𝑉 0.23𝑉𝑉𝑉𝑉 (2)
𝑝𝑝𝑝𝑝 = 1.16𝑉𝑉𝑉𝑉𝑠𝑠𝑠𝑠 + 1360 (1) 1 1 1
𝐶𝐶𝐶𝐶1 = 𝑐𝑐𝑐𝑐 + 𝐾𝐾𝐾𝐾𝑐𝑐𝑐𝑐2 + 𝑚𝑚𝑚𝑚𝑐𝑐𝑐𝑐3 , 𝐶𝐶𝐶𝐶2 = 𝑐𝑐𝑐𝑐2 , 𝑊𝑊𝑊𝑊(𝜃𝜃𝜃𝜃)
ln(𝑍𝑍𝑍𝑍
𝜌𝜌𝜌𝜌 𝑆𝑆𝑆𝑆 ) = 𝑘𝑘𝑘𝑘ln(𝑍𝑍𝑍𝑍𝑃𝑃𝑃𝑃 ) + 𝑘𝑘𝑘𝑘𝐶𝐶𝐶𝐶
= 0.23𝑉𝑉𝑉𝑉 0.25 + ∆𝐿𝐿𝐿𝐿𝑆𝑆𝑆𝑆 (2) (3) 2 1 𝑉𝑉𝑉𝑉𝑝𝑝𝑝𝑝 2= 1.16𝑉𝑉𝑉𝑉 𝑠𝑠𝑠𝑠 + 1360 2 (1)
the best fitted line. The Aki-Richard’s equation is redefined is the acoustic impedance of first (top) layer and ZN is the
ln(𝑍𝑍𝑍𝑍
ln⁡ (𝜌𝜌𝜌𝜌) ) = 𝑘𝑘𝑘𝑘ln(𝑍𝑍𝑍𝑍 ) + 𝑘𝑘𝑘𝑘 + ∆𝐿𝐿𝐿𝐿 (3) 𝑇𝑇𝑇𝑇(𝜃𝜃𝜃𝜃 ) 𝑉𝑉𝑉𝑉 𝐶𝐶𝐶𝐶
= (𝜃𝜃𝜃𝜃
1.16𝑉𝑉𝑉𝑉 )𝑊𝑊𝑊𝑊(𝜃𝜃𝜃𝜃 + )𝐷𝐷𝐷𝐷
1360 ⋯ 𝐶𝐶𝐶𝐶 (𝜃𝜃𝜃𝜃 )𝑊𝑊𝑊𝑊(𝜃𝜃𝜃𝜃 )𝐷𝐷𝐷𝐷 (1)
Fatti=as𝑚𝑚𝑚𝑚ln(𝑍𝑍𝑍𝑍 𝑃𝑃𝑃𝑃 ) + 𝑚𝑚𝑚𝑚(Fatti 𝐶𝐶𝐶𝐶 + ∆𝐿𝐿𝐿𝐿 (4) 1 𝑝𝑝𝑝𝑝 1 1 0.25
𝑠𝑠𝑠𝑠 1 3 3 3
𝑆𝑆𝑆𝑆 𝑃𝑃𝑃𝑃 𝐶𝐶𝐶𝐶 𝑆𝑆𝑆𝑆 th 𝜌𝜌𝜌𝜌 = 0.23𝑉𝑉𝑉𝑉 (2)
by followings et𝐷𝐷𝐷𝐷 al., 1994). AI
� of ⋮ N� =layer, �𝑝𝑝𝑝𝑝 = 1.16𝑉𝑉𝑉𝑉 ri is⋮ the reflection ⋱ coefficient ⋮ at� ith interface. (7)
ln⁡ (𝜌𝜌𝜌𝜌) = 𝑚𝑚𝑚𝑚ln(𝑍𝑍𝑍𝑍𝑃𝑃𝑃𝑃 ) + 𝑚𝑚𝑚𝑚𝐶𝐶𝐶𝐶 + ∆𝐿𝐿𝐿𝐿𝐷𝐷𝐷𝐷 (4) 𝑉𝑉𝑉𝑉 0.25 + 1360 (2)(1)
) 𝜌𝜌𝜌𝜌 = (𝜃𝜃𝜃𝜃 0.23𝑉𝑉𝑉𝑉 𝑠𝑠𝑠𝑠
𝑉𝑉𝑉𝑉𝑝𝑝𝑝𝑝 = 1.16𝑉𝑉𝑉𝑉 𝑉𝑉𝑉𝑉𝑝𝑝𝑝𝑝∆𝑉𝑉𝑉𝑉 𝑠𝑠𝑠𝑠=+1.16𝑉𝑉𝑉𝑉
∆𝑉𝑉𝑉𝑉 1360 𝑃𝑃𝑃𝑃 𝑠𝑠𝑠𝑠 + 1360 ∆𝑉𝑉𝑉𝑉𝑆𝑆𝑆𝑆 ∆𝜌𝜌𝜌𝜌 (1) (1) This 𝑇𝑇𝑇𝑇(𝜃𝜃𝜃𝜃𝑁𝑁𝑁𝑁equation ln(𝑍𝑍𝑍𝑍 𝐶𝐶𝐶𝐶 1 𝑆𝑆𝑆𝑆 )is valid
𝑁𝑁𝑁𝑁=)𝑊𝑊𝑊𝑊(𝜃𝜃𝜃𝜃 𝑘𝑘𝑘𝑘ln(𝑍𝑍𝑍𝑍 𝑁𝑁𝑁𝑁for 𝑃𝑃𝑃𝑃 ) +
)𝐷𝐷𝐷𝐷 most 𝑘𝑘𝑘𝑘𝐶𝐶𝐶𝐶 +𝐶𝐶𝐶𝐶
⋯ of∆𝐿𝐿𝐿𝐿 the
3 (𝜃𝜃𝜃𝜃 practical
𝑆𝑆𝑆𝑆𝑁𝑁𝑁𝑁 )𝑊𝑊𝑊𝑊(𝜃𝜃𝜃𝜃 𝑁𝑁𝑁𝑁 )𝐷𝐷𝐷𝐷 cases where (3)
𝑅𝑅𝑅𝑅(𝜃𝜃𝜃𝜃) = 𝑐𝑐𝑐𝑐1𝑃𝑃𝑃𝑃 +∆𝑉𝑉𝑉𝑉𝑆𝑆𝑆𝑆𝑐𝑐𝑐𝑐 ∆𝜌𝜌𝜌𝜌+ 𝑐𝑐𝑐𝑐3 (5)
(5) 𝑉𝑉𝑉𝑉𝑝𝑝𝑝𝑝 ==0.23𝑉𝑉𝑉𝑉
𝜌𝜌𝜌𝜌 1.16𝑉𝑉𝑉𝑉0.25 𝑠𝑠𝑠𝑠 +and 1360 (2)(1)
𝑅𝑅𝑅𝑅(𝜃𝜃𝜃𝜃) = 𝑐𝑐𝑐𝑐1 0.25 +𝑃𝑃𝑃𝑃𝑐𝑐𝑐𝑐2 0.252+ 𝑐𝑐𝑐𝑐𝑉𝑉𝑉𝑉
𝑉𝑉𝑉𝑉 𝜌𝜌𝜌𝜌 (5) rj≤|0.3| [𝐿𝐿𝐿𝐿𝑃𝑃𝑃𝑃 Δ𝐿𝐿𝐿𝐿𝑆𝑆𝑆𝑆(Berteussen
ln(𝑍𝑍𝑍𝑍
ln⁡Δ𝐿𝐿𝐿𝐿(𝜌𝜌𝜌𝜌) 𝑆𝑆𝑆𝑆]) = = 𝑚𝑚𝑚𝑚ln(𝑍𝑍𝑍𝑍 𝑘𝑘𝑘𝑘ln(𝑍𝑍𝑍𝑍 𝑃𝑃𝑃𝑃 ) +Ursin,
+ 𝑘𝑘𝑘𝑘𝑚𝑚𝑚𝑚𝐶𝐶𝐶𝐶 ++∆𝐿𝐿𝐿𝐿 1983). (3)
= [𝑙𝑙𝑙𝑙𝑠𝑠𝑠𝑠(𝑍𝑍𝑍𝑍 𝑃𝑃𝑃𝑃 ) 0 0)], ∆𝐿𝐿𝐿𝐿 (4)
3𝑆𝑆𝑆𝑆 𝑆𝑆𝑆𝑆
𝜌𝜌𝜌𝜌 = 0.23𝑉𝑉𝑉𝑉 𝜌𝜌𝜌𝜌 𝑉𝑉𝑉𝑉=𝑃𝑃𝑃𝑃 0.23𝑉𝑉𝑉𝑉 𝑉𝑉𝑉𝑉𝑆𝑆𝑆𝑆 𝜌𝜌𝜌𝜌 (2) (2) 𝐷𝐷𝐷𝐷
0.25
𝑃𝑃𝑃𝑃0 𝐶𝐶𝐶𝐶 𝐷𝐷𝐷𝐷
2 2 The acoustic
𝜌𝜌𝜌𝜌
ln(𝑍𝑍𝑍𝑍
ln⁡ = (𝜌𝜌𝜌𝜌) 0.23𝑉𝑉𝑉𝑉
𝑆𝑆𝑆𝑆 ) = = 𝑚𝑚𝑚𝑚ln(𝑍𝑍𝑍𝑍 impedance
𝑘𝑘𝑘𝑘ln(𝑍𝑍𝑍𝑍 𝑃𝑃𝑃𝑃 )) +
+ 𝑘𝑘𝑘𝑘
𝑚𝑚𝑚𝑚 𝐶𝐶𝐶𝐶 + model
+ ∆𝐿𝐿𝐿𝐿
∆𝐿𝐿𝐿𝐿 𝑆𝑆𝑆𝑆 of low frequency (2)is (3)
(4)
ln(𝑍𝑍𝑍𝑍𝑆𝑆𝑆𝑆 )𝑐𝑐𝑐𝑐=ln(𝑍𝑍𝑍𝑍 𝑘𝑘𝑘𝑘ln(𝑍𝑍𝑍𝑍 1 1+ 𝑘𝑘𝑘𝑘𝐶𝐶𝐶𝐶 + 𝑉𝑉𝑉𝑉 2
𝑉𝑉𝑉𝑉𝑆𝑆𝑆𝑆 𝑉𝑉𝑉𝑉 2
𝑉𝑉𝑉𝑉𝑆𝑆𝑆𝑆 𝑆𝑆𝑆𝑆(𝑡𝑡𝑡𝑡) = 𝑊𝑊𝑊𝑊(𝑡𝑡𝑡𝑡) ∗ 𝑟𝑟𝑟𝑟(𝑡𝑡𝑡𝑡) + ∆𝑉𝑉𝑉𝑉 𝑠𝑠𝑠𝑠(𝑡𝑡𝑡𝑡) 𝑃𝑃𝑃𝑃 ∆𝑉𝑉𝑉𝑉 𝐶𝐶𝐶𝐶 ∆𝜌𝜌𝜌𝜌
𝐷𝐷𝐷𝐷 (8)
𝑆𝑆𝑆𝑆 ) 𝑃𝑃𝑃𝑃=) 𝑘𝑘𝑘𝑘ln(𝑍𝑍𝑍𝑍 ∆𝐿𝐿𝐿𝐿𝑆𝑆𝑆𝑆𝑘𝑘𝑘𝑘𝐶𝐶𝐶𝐶−+� ∆𝐿𝐿𝐿𝐿 (3) 𝑃𝑃𝑃𝑃 𝑆𝑆𝑆𝑆
𝑐𝑐𝑐𝑐𝑃𝑃𝑃𝑃,2 )=+0.5 � 𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑐𝑐𝑐𝑐2 𝜃𝜃𝜃𝜃(3) 𝑐𝑐𝑐𝑐3 over the entire seismic (5)
𝑆𝑆𝑆𝑆 𝑆𝑆𝑆𝑆
where
where1 𝑐𝑐𝑐𝑐=1 = , 𝑐𝑐𝑐𝑐 = 0.5�𝑆𝑆𝑆𝑆 𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠− 𝑠𝑠𝑠𝑠2 𝜃𝜃𝜃𝜃� and � 𝑐𝑐𝑐𝑐3𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠=𝑠𝑠𝑠𝑠42�𝜃𝜃𝜃𝜃 and = 4 � � obtained
𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠 𝑠𝑠𝑠𝑠 2
𝜃𝜃𝜃𝜃 by𝑅𝑅𝑅𝑅(𝜃𝜃𝜃𝜃)
ln(𝑍𝑍𝑍𝑍
ln⁡ estimating
(𝜌𝜌𝜌𝜌) ) ==𝑐𝑐𝑐𝑐𝑘𝑘𝑘𝑘ln(𝑍𝑍𝑍𝑍
𝑆𝑆𝑆𝑆 = 𝑚𝑚𝑚𝑚ln(𝑍𝑍𝑍𝑍
1 +these
) 𝑐𝑐𝑐𝑐+2 𝑘𝑘𝑘𝑘 values
𝑚𝑚𝑚𝑚 + +∆𝐿𝐿𝐿𝐿 (3)
2𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑠𝑠𝑠𝑠2𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑠𝑠𝑠𝑠 2 𝜃𝜃𝜃𝜃 2 𝜃𝜃𝜃𝜃 2 𝑉𝑉𝑉𝑉𝑃𝑃𝑃𝑃
𝑉𝑉𝑉𝑉𝑃𝑃𝑃𝑃 𝑉𝑉𝑉𝑉𝑃𝑃𝑃𝑃 3
𝑉𝑉𝑉𝑉𝑃𝑃𝑃𝑃 section ∆𝑉𝑉𝑉𝑉
𝑉𝑉𝑉𝑉𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃 𝑃𝑃𝑃𝑃 𝑉𝑉𝑉𝑉𝐶𝐶𝐶𝐶𝑆𝑆𝑆𝑆𝐶𝐶𝐶𝐶𝑆𝑆𝑆𝑆 + ∆𝐿𝐿𝐿𝐿𝑆𝑆𝑆𝑆𝐷𝐷𝐷𝐷∆𝜌𝜌𝜌𝜌
∆𝑉𝑉𝑉𝑉 𝜌𝜌𝜌𝜌 (4)
ln⁡ 𝑍𝑍𝑍𝑍 = 𝑉𝑉𝑉𝑉𝜌𝜌𝜌𝜌 using 𝑅𝑅𝑅𝑅(𝜃𝜃𝜃𝜃)kriging = 𝑐𝑐𝑐𝑐1 interpolation + 𝑐𝑐𝑐𝑐2 + 𝑐𝑐𝑐𝑐3 techniques (9) at the wells (5)
𝑉𝑉𝑉𝑉 𝑝𝑝𝑝𝑝 (𝜌𝜌𝜌𝜌)
= 1.16𝑉𝑉𝑉𝑉 =ln⁡ 𝑚𝑚𝑚𝑚ln(𝑍𝑍𝑍𝑍 𝑠𝑠𝑠𝑠(𝜌𝜌𝜌𝜌) + 1360 = 𝑃𝑃𝑃𝑃 )𝑚𝑚𝑚𝑚ln(𝑍𝑍𝑍𝑍 + 𝑚𝑚𝑚𝑚𝐶𝐶𝐶𝐶 𝑃𝑃𝑃𝑃+) + ∆𝐿𝐿𝐿𝐿𝑉𝑉𝑉𝑉𝑚𝑚𝑚𝑚 𝑝𝑝𝑝𝑝 =
𝐷𝐷𝐷𝐷 + ∆𝐿𝐿𝐿𝐿
𝐶𝐶𝐶𝐶 1.16𝑉𝑉𝑉𝑉 𝑠𝑠𝑠𝑠 +𝐷𝐷𝐷𝐷 1360 (1) (4) (4)
(1)
𝑉𝑉𝑉𝑉 𝑉𝑉𝑉𝑉 𝜌𝜌𝜌𝜌
𝑇𝑇𝑇𝑇(𝜃𝜃𝜃𝜃) The = 𝐶𝐶𝐶𝐶1 𝑊𝑊𝑊𝑊(𝜃𝜃𝜃𝜃) above ∗ 𝐷𝐷𝐷𝐷𝐿𝐿𝐿𝐿
equation 𝑃𝑃𝑃𝑃 + 𝐶𝐶𝐶𝐶 2 𝑊𝑊𝑊𝑊(𝜃𝜃𝜃𝜃) estimate ∗ 𝐷𝐷𝐷𝐷𝐿𝐿𝐿𝐿 +
𝑆𝑆𝑆𝑆 0.25reflection 𝐶𝐶𝐶𝐶 3 𝑊𝑊𝑊𝑊(𝜃𝜃𝜃𝜃) ∗ 𝐷𝐷𝐷𝐷𝐿𝐿𝐿𝐿 coefficient
𝐷𝐷𝐷𝐷 (6)
as(2)a ln⁡ (𝜌𝜌𝜌𝜌) = 𝑚𝑚𝑚𝑚ln(𝑍𝑍𝑍𝑍 ∆𝑉𝑉𝑉𝑉 𝑃𝑃𝑃𝑃 𝑃𝑃𝑃𝑃1
𝑃𝑃𝑃𝑃 ) + ∆𝑉𝑉𝑉𝑉
𝑚𝑚𝑚𝑚 𝑆𝑆𝑆𝑆
𝐶𝐶𝐶𝐶𝑆𝑆𝑆𝑆 + ∆𝐿𝐿𝐿𝐿 ∆𝜌𝜌𝜌𝜌
𝐷𝐷𝐷𝐷 2
𝑉𝑉𝑉𝑉𝑆𝑆𝑆𝑆 Generally, (4) 𝑉𝑉𝑉𝑉𝑆𝑆𝑆𝑆 2
𝑇𝑇𝑇𝑇(𝜃𝜃𝜃𝜃) = 𝐶𝐶𝐶𝐶 0.25 ∆𝑉𝑉𝑉𝑉 𝑊𝑊𝑊𝑊(𝜃𝜃𝜃𝜃) ∗∆𝑉𝑉𝑉𝑉
∆𝑉𝑉𝑉𝑉 𝐷𝐷𝐷𝐷𝐿𝐿𝐿𝐿 𝑆𝑆𝑆𝑆 𝑃𝑃𝑃𝑃 +∆𝑉𝑉𝑉𝑉 𝐶𝐶𝐶𝐶 2𝑆𝑆𝑆𝑆𝑊𝑊𝑊𝑊(𝜃𝜃𝜃𝜃)∆𝜌𝜌𝜌𝜌
𝜌𝜌𝜌𝜌 = 0.23𝑉𝑉𝑉𝑉
∆𝜌𝜌𝜌𝜌 ∗ 𝐷𝐷𝐷𝐷𝐿𝐿𝐿𝐿𝑆𝑆𝑆𝑆 + 𝐶𝐶𝐶𝐶 3 𝑊𝑊𝑊𝑊(𝜃𝜃𝜃𝜃) ∗ 𝐷𝐷𝐷𝐷𝐿𝐿𝐿𝐿 (Maurya (6) −
𝑍𝑍𝑍𝑍𝑠𝑠𝑠𝑠+1 𝑅𝑅𝑅𝑅(𝜃𝜃𝜃𝜃) and
where 𝑍𝑍𝑍𝑍𝑠𝑠𝑠𝑠 = Singh𝑐𝑐𝑐𝑐 𝑐𝑐𝑐𝑐1= 2017; + 𝑐𝑐𝑐𝑐2 , Schuster, +𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐3 = 0.52017). − � � 𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠 𝑠𝑠𝑠𝑠 2
𝜃𝜃𝜃𝜃 and (5)
𝑐𝑐𝑐𝑐 = 4 � � 𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠2 𝜃𝜃𝜃𝜃
𝜌𝜌𝜌𝜌 = 0.23𝑉𝑉𝑉𝑉 1 𝑃𝑃𝑃𝑃 𝑃𝑃𝑃𝑃 (2) 𝐷𝐷𝐷𝐷 𝑟𝑟𝑟𝑟 = 1 𝑉𝑉𝑉𝑉 𝑉𝑉𝑉𝑉 2 𝜌𝜌𝜌𝜌 (10) 2 3
𝑉𝑉𝑉𝑉𝑃𝑃𝑃𝑃𝑆𝑆𝑆𝑆 2
𝑅𝑅𝑅𝑅(𝜃𝜃𝜃𝜃)
𝐿𝐿𝐿𝐿
function 𝑃𝑃𝑃𝑃 = ln(𝑍𝑍𝑍𝑍 = 𝑅𝑅𝑅𝑅(𝜃𝜃𝜃𝜃) 𝑐𝑐𝑐𝑐𝑃𝑃𝑃𝑃of
1) , incident 𝐿𝐿𝐿𝐿𝑆𝑆𝑆𝑆==
𝑉𝑉𝑉𝑉𝑃𝑃𝑃𝑃
+𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐ln(𝑍𝑍𝑍𝑍 12 𝑉𝑉𝑉𝑉
) ,+
𝑆𝑆𝑆𝑆angle +
and 𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐23𝐿𝐿𝐿𝐿q. 𝐷𝐷𝐷𝐷 =
ln(𝑍𝑍𝑍𝑍
𝜌𝜌𝜌𝜌 For
𝑆𝑆𝑆𝑆 )+ ln(𝜌𝜌𝜌𝜌). 𝑐𝑐𝑐𝑐a3 given
= 𝑘𝑘𝑘𝑘ln(𝑍𝑍𝑍𝑍 𝑃𝑃𝑃𝑃 ) + 𝑘𝑘𝑘𝑘𝐶𝐶𝐶𝐶 + angle ∆𝐿𝐿𝐿𝐿𝑆𝑆𝑆𝑆 T (q),(5) we can(5) (3) 𝑠𝑠𝑠𝑠
acoustic 𝑍𝑍𝑍𝑍 impedance
+ 𝑍𝑍𝑍𝑍 ∆𝑉𝑉𝑉𝑉 𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃 1 𝜃𝜃𝜃𝜃 ∆𝑉𝑉𝑉𝑉
2𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑠𝑠𝑠𝑠
is not
2
recorded 𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆 ∆𝜌𝜌𝜌𝜌
during 𝑉𝑉𝑉𝑉𝑃𝑃𝑃𝑃𝑆𝑆𝑆𝑆 acquisition
the (5)
𝑉𝑉𝑉𝑉 𝑉𝑉𝑉𝑉 𝜌𝜌𝜌𝜌 𝑠𝑠𝑠𝑠+1 𝑅𝑅𝑅𝑅(𝜃𝜃𝜃𝜃)
where 𝑠𝑠𝑠𝑠 = 𝑐𝑐𝑐𝑐 𝑐𝑐𝑐𝑐 = + 𝑐𝑐𝑐𝑐 , + 𝑐𝑐𝑐𝑐
𝑐𝑐𝑐𝑐 = 0.5 − � � 𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠 𝑠𝑠𝑠𝑠 2
𝜃𝜃𝜃𝜃 and 𝑐𝑐𝑐𝑐 = 4 � � 𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠2 𝜃𝜃𝜃𝜃
1𝑃𝑃𝑃𝑃 ) , 𝑃𝑃𝑃𝑃 𝐿𝐿𝐿𝐿 ) (𝜌𝜌𝜌𝜌)
𝑆𝑆𝑆𝑆
𝐿𝐿𝐿𝐿𝑃𝑃𝑃𝑃 =𝑆𝑆𝑆𝑆1)ln(𝑍𝑍𝑍𝑍
ln(𝑍𝑍𝑍𝑍
therefore = 𝑘𝑘𝑘𝑘ln(𝑍𝑍𝑍𝑍 derive ) a𝑆𝑆𝑆𝑆+relationships = 𝑘𝑘𝑘𝑘𝑃𝑃𝑃𝑃𝐶𝐶𝐶𝐶 ln(𝑍𝑍𝑍𝑍 + ∆𝐿𝐿𝐿𝐿𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆ln⁡ 𝑆𝑆𝑆𝑆, and
1= 𝑚𝑚𝑚𝑚ln(𝑍𝑍𝑍𝑍 between 𝐿𝐿𝐿𝐿𝐷𝐷𝐷𝐷𝑃𝑃𝑃𝑃 ) = + 𝑚𝑚𝑚𝑚ln(𝜌𝜌𝜌𝜌).
𝐶𝐶𝐶𝐶 + ∆𝐿𝐿𝐿𝐿𝐷𝐷𝐷𝐷
the P- impedance, (3) (4) 1 1
𝑉𝑉𝑉𝑉2𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑠𝑠𝑠𝑠 1 ∗2 𝜃𝜃𝜃𝜃
2
𝑉𝑉𝑉𝑉𝑆𝑆𝑆𝑆 + 𝐶𝐶𝐶𝐶2can 3
𝜌𝜌𝜌𝜌 be∗estimated 𝑉𝑉𝑉𝑉𝑃𝑃𝑃𝑃𝑆𝑆𝑆𝑆+2𝐶𝐶𝐶𝐶 𝑊𝑊𝑊𝑊(𝜃𝜃𝜃𝜃) 3
𝑉𝑉𝑉𝑉𝑃𝑃𝑃𝑃𝑆𝑆𝑆𝑆 2 (6)
𝐶𝐶𝐶𝐶1 = 𝑐𝑐𝑐𝑐1 + 𝐾𝐾𝐾𝐾𝑐𝑐𝑐𝑐2 + 𝑚𝑚𝑚𝑚𝑐𝑐𝑐𝑐3 , 𝐶𝐶𝐶𝐶2 = 𝑐𝑐𝑐𝑐2 , ∆𝑉𝑉𝑉𝑉𝑃𝑃𝑃𝑃 𝑊𝑊𝑊𝑊(𝜃𝜃𝜃𝜃) 2 𝑆𝑆𝑆𝑆 2 2 2of well log𝑇𝑇𝑇𝑇(𝜃𝜃𝜃𝜃) data.
where 𝑐𝑐𝑐𝑐1 = =
𝑁𝑁𝑁𝑁 These
𝐶𝐶𝐶𝐶 1
𝑃𝑃𝑃𝑃 parameters
𝑊𝑊𝑊𝑊(𝜃𝜃𝜃𝜃) 𝐷𝐷𝐷𝐷𝐿𝐿𝐿𝐿
, 𝑃𝑃𝑃𝑃 𝑊𝑊𝑊𝑊(𝜃𝜃𝜃𝜃) 𝐷𝐷𝐷𝐷𝐿𝐿𝐿𝐿
𝑐𝑐𝑐𝑐2 = 0.5 − � �2 𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠 𝜃𝜃𝜃𝜃 and 𝑐𝑐𝑐𝑐3 = 4 � �2 𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠2 𝜃𝜃𝜃𝜃
2 𝑆𝑆𝑆𝑆 3 directly
2 ∗ 𝐷𝐷𝐷𝐷𝐿𝐿𝐿𝐿 𝐷𝐷𝐷𝐷
2 2 1 1 𝑅𝑅𝑅𝑅(𝜃𝜃𝜃𝜃) 2= 𝑐𝑐𝑐𝑐1 𝑉𝑉𝑉𝑉𝑆𝑆𝑆𝑆𝑐𝑐𝑐𝑐2 + 𝑐𝑐𝑐𝑐32𝑉𝑉𝑉𝑉𝑆𝑆𝑆𝑆
∆𝑉𝑉𝑉𝑉 ∆𝜌𝜌𝜌𝜌 𝑉𝑉𝑉𝑉𝑆𝑆𝑆𝑆(5) 𝑉𝑉𝑉𝑉
S-impedance
ln⁡ (𝜌𝜌𝜌𝜌) = 1𝑐𝑐𝑐𝑐1where = 1𝑐𝑐𝑐𝑐𝑃𝑃𝑃𝑃and
𝑚𝑚𝑚𝑚ln(𝑍𝑍𝑍𝑍 )2= + 𝑚𝑚𝑚𝑚 density
, 𝐶𝐶𝐶𝐶 + 𝑐𝑐𝑐𝑐∆𝐿𝐿𝐿𝐿 by 0.5modifying
+
�1𝑉𝑉𝑉𝑉𝑆𝑆𝑆𝑆− 𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠�Fatti’s 𝑐𝑐𝑐𝑐equation
2(4) 𝑠𝑠𝑠𝑠�2 𝜃𝜃𝜃𝜃𝑆𝑆𝑆𝑆 � from 2 the sonic and 2𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑠𝑠𝑠𝑠
density 1 ∗=𝜃𝜃𝜃𝜃
2
log. The 𝐷𝐷𝐷𝐷𝐿𝐿𝐿𝐿𝑉𝑉𝑉𝑉 𝑉𝑉𝑉𝑉
where = − � 0.5 𝜃𝜃𝜃𝜃 𝐿𝐿𝐿𝐿𝐷𝐷𝐷𝐷objective 3function
𝑉𝑉𝑉𝑉𝑆𝑆𝑆𝑆𝑃𝑃𝑃𝑃𝑆𝑆𝑆𝑆+ 𝐶𝐶𝐶𝐶(11) 𝑉𝑉𝑉𝑉𝑃𝑃𝑃𝑃𝑆𝑆𝑆𝑆
, 𝑐𝑐𝑐𝑐 𝑉𝑉𝑉𝑉𝑃𝑃𝑃𝑃= 𝜌𝜌𝜌𝜌 �and 𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠 𝑠𝑠𝑠𝑠 𝜃𝜃𝜃𝜃 =and 4 � 𝑐𝑐𝑐𝑐3 �= 𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠 4 𝑍𝑍𝑍𝑍
𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠 𝑠𝑠𝑠𝑠= 𝜃𝜃𝜃𝜃𝑍𝑍𝑍𝑍 𝑇𝑇𝑇𝑇(𝜃𝜃𝜃𝜃)
𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑝𝑝𝑝𝑝 �� = 𝐶𝐶𝐶𝐶𝑠𝑠𝑠𝑠 �
𝑟𝑟𝑟𝑟 𝑊𝑊𝑊𝑊(𝜃𝜃𝜃𝜃) 𝐷𝐷𝐷𝐷𝐿𝐿𝐿𝐿 + )𝐶𝐶𝐶𝐶,2and 𝑊𝑊𝑊𝑊(𝜃𝜃𝜃𝜃) ∗= 𝑊𝑊𝑊𝑊(𝜃𝜃𝜃𝜃) ∗ 𝐷𝐷𝐷𝐷𝐿𝐿𝐿𝐿𝐷𝐷𝐷𝐷 (6)
𝐶𝐶𝐶𝐶𝑇𝑇𝑇𝑇(𝜃𝜃𝜃𝜃 2 𝐷𝐷𝐷𝐷 3 𝑁𝑁𝑁𝑁 1 𝐿𝐿𝐿𝐿 = ln(𝑍𝑍𝑍𝑍 1 ) , 𝐿𝐿𝐿𝐿 ln(𝑍𝑍𝑍𝑍 𝑃𝑃𝑃𝑃
− �ln(𝜌𝜌𝜌𝜌).
1 = 𝑐𝑐𝑐𝑐al. 1𝐶𝐶𝐶𝐶+ 𝐾𝐾𝐾𝐾𝑐𝑐𝑐𝑐𝜃𝜃𝜃𝜃2to + )𝐷𝐷𝐷𝐷𝑚𝑚𝑚𝑚𝑐𝑐𝑐𝑐 𝜃𝜃𝜃𝜃3 , 𝐶𝐶𝐶𝐶following 3 (𝜃𝜃𝜃𝜃𝐶𝐶𝐶𝐶 2 = )𝐷𝐷𝐷𝐷𝑐𝑐𝑐𝑐2 , 𝑉𝑉𝑉𝑉𝑃𝑃𝑃𝑃 𝑊𝑊𝑊𝑊(𝜃𝜃𝜃𝜃)
1 2
(Fatti 1)
et 12𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑠𝑠𝑠𝑠
(𝜃𝜃𝜃𝜃1 )𝑊𝑊𝑊𝑊(𝜃𝜃𝜃𝜃
1994) 12𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑠𝑠𝑠𝑠
obtain
2⋯ 3 )𝑊𝑊𝑊𝑊(𝜃𝜃𝜃𝜃 𝑉𝑉𝑉𝑉
1 𝑃𝑃𝑃𝑃3 equation: 𝑉𝑉𝑉𝑉𝑆𝑆𝑆𝑆 2
𝑉𝑉𝑉𝑉𝑃𝑃𝑃𝑃 𝑉𝑉𝑉𝑉𝑆𝑆𝑆𝑆 𝑉𝑉𝑉𝑉
2
𝑃𝑃𝑃𝑃 2 where 𝑃𝑃𝑃𝑃 𝑐𝑐𝑐𝑐1 = 𝑃𝑃𝑃𝑃 𝑆𝑆𝑆𝑆 , 𝑆𝑆𝑆𝑆 𝑐𝑐𝑐𝑐2 = 0.5 � 𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠2 𝜃𝜃𝜃𝜃 and 𝑐𝑐𝑐𝑐3 = 4 � � 𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠2 𝜃𝜃𝜃𝜃
� ⋮ �2= � ∆𝑉𝑉𝑉𝑉𝑃𝑃𝑃𝑃2 ⋮ ∆𝑉𝑉𝑉𝑉𝑆𝑆𝑆𝑆 ⋱ where ∆𝜌𝜌𝜌𝜌 𝑐𝑐𝑐𝑐1 = ⋮2𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑠𝑠𝑠𝑠2 𝜃𝜃𝜃𝜃2, � 𝑐𝑐𝑐𝑐2 = 0.5 − �𝑉𝑉𝑉𝑉 � (7) 𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠 𝜃𝜃𝜃𝜃 and 𝑐𝑐𝑐𝑐3 = 4 � � 𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠 which needs to minimize is:
2𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑠𝑠𝑠𝑠 2 𝜃𝜃𝜃𝜃 𝑉𝑉𝑉𝑉𝑆𝑆𝑆𝑆𝑃𝑃𝑃𝑃+ 𝐶𝐶𝐶𝐶3 𝑊𝑊𝑊𝑊(𝜃𝜃𝜃𝜃) ∗ 𝐷𝐷𝐷𝐷𝐿𝐿𝐿𝐿𝐷𝐷𝐷𝐷 𝑉𝑉𝑉𝑉𝑃𝑃𝑃𝑃
2
𝜃𝜃𝜃𝜃 𝑇𝑇𝑇𝑇(𝜃𝜃𝜃𝜃)
𝐿𝐿𝐿𝐿 = =
𝑠𝑠𝑠𝑠=1
ln(𝑍𝑍𝑍𝑍 𝐶𝐶𝐶𝐶 1 𝑊𝑊𝑊𝑊(𝜃𝜃𝜃𝜃)
) , 𝐿𝐿𝐿𝐿 ∗= 𝐷𝐷𝐷𝐷𝐿𝐿𝐿𝐿
ln(𝑍𝑍𝑍𝑍 𝑃𝑃𝑃𝑃 + 𝐶𝐶𝐶𝐶
) ,2 𝑊𝑊𝑊𝑊(𝜃𝜃𝜃𝜃)
and 𝐿𝐿𝐿𝐿 ∗ = 𝐷𝐷𝐷𝐷𝐿𝐿𝐿𝐿
ln(𝜌𝜌𝜌𝜌). (6)
𝑅𝑅𝑅𝑅(𝜃𝜃𝜃𝜃) 𝑐𝑐𝑐𝑐1𝐶𝐶𝐶𝐶𝑊𝑊𝑊𝑊(𝜃𝜃𝜃𝜃) 𝑃𝑃𝑃𝑃 (5) 𝑃𝑃𝑃𝑃 1 𝑃𝑃𝑃𝑃 1 𝑆𝑆𝑆𝑆 𝑆𝑆𝑆𝑆 𝐷𝐷𝐷𝐷 1
=𝑁𝑁𝑁𝑁+ 𝐶𝐶𝐶𝐶∗𝑐𝑐𝑐𝑐1𝐷𝐷𝐷𝐷𝐿𝐿𝐿𝐿 )𝐷𝐷𝐷𝐷++∗𝐶𝐶𝐶𝐶⋯ 𝑐𝑐𝑐𝑐𝐷𝐷𝐷𝐷𝐿𝐿𝐿𝐿 𝑉𝑉𝑉𝑉𝑃𝑃𝑃𝑃
𝑇𝑇𝑇𝑇(𝜃𝜃𝜃𝜃)
𝑇𝑇𝑇𝑇(𝜃𝜃𝜃𝜃 = 𝐶𝐶𝐶𝐶
𝑁𝑁𝑁𝑁 ) ) 𝑇𝑇𝑇𝑇(𝜃𝜃𝜃𝜃) 1𝑉𝑉𝑉𝑉(𝜃𝜃𝜃𝜃 )𝑊𝑊𝑊𝑊(𝜃𝜃𝜃𝜃 2𝑊𝑊𝑊𝑊(𝜃𝜃𝜃𝜃)
𝑁𝑁𝑁𝑁𝑉𝑉𝑉𝑉𝑃𝑃𝑃𝑃 23𝑊𝑊𝑊𝑊(𝜃𝜃𝜃𝜃) 𝐶𝐶𝐶𝐶3+ (𝜃𝜃𝜃𝜃 ∗)𝑊𝑊𝑊𝑊(𝜃𝜃𝜃𝜃
𝐶𝐶𝐶𝐶 𝐷𝐷𝐷𝐷𝐿𝐿𝐿𝐿
𝑊𝑊𝑊𝑊(𝜃𝜃𝜃𝜃) 𝑆𝑆𝑆𝑆 𝑁𝑁𝑁𝑁+)𝐷𝐷𝐷𝐷 ∗𝐶𝐶𝐶𝐶3𝐷𝐷𝐷𝐷𝐿𝐿𝐿𝐿 𝑊𝑊𝑊𝑊(𝜃𝜃𝜃𝜃)
𝑆𝑆𝑆𝑆 + 𝐶𝐶𝐶𝐶 ∗3𝐷𝐷𝐷𝐷𝐿𝐿𝐿𝐿
𝑊𝑊𝑊𝑊(𝜃𝜃𝜃𝜃) 𝐷𝐷𝐷𝐷 ∗ 𝐷𝐷𝐷𝐷𝐿𝐿𝐿𝐿 (6) (6) (6) 𝐶𝐶𝐶𝐶𝑀𝑀𝑀𝑀 = 𝑐𝑐𝑐𝑐 + 𝐾𝐾𝐾𝐾𝑐𝑐𝑐𝑐 + 𝑚𝑚𝑚𝑚𝑐𝑐𝑐𝑐 , 𝐶𝐶𝐶𝐶 = 𝑐𝑐𝑐𝑐 , 𝑊𝑊𝑊𝑊(𝜃𝜃𝜃𝜃)
𝑇𝑇𝑇𝑇(𝜃𝜃𝜃𝜃 1 𝑃𝑃𝑃𝑃 𝐶𝐶𝐶𝐶 1 1 (𝜃𝜃𝜃𝜃 )𝑊𝑊𝑊𝑊(𝜃𝜃𝜃𝜃 𝑆𝑆𝑆𝑆 1 )𝐷𝐷𝐷𝐷 𝜌𝜌𝜌𝜌
𝑇𝑇𝑇𝑇(𝜃𝜃𝜃𝜃)
𝑃𝑃𝑃𝑃
= ⋯ 𝑁𝑁𝑁𝑁 2
𝐶𝐶𝐶𝐶1 𝑊𝑊𝑊𝑊(𝜃𝜃𝜃𝜃) 𝐶𝐶𝐶𝐶
∗ (𝜃𝜃𝜃𝜃
𝐷𝐷𝐷𝐷𝐿𝐿𝐿𝐿
3 𝑃𝑃𝑃𝑃 3 2 + )𝑊𝑊𝑊𝑊(𝜃𝜃𝜃𝜃
𝐶𝐶𝐶𝐶 𝑊𝑊𝑊𝑊(𝜃𝜃𝜃𝜃) ∗ )𝐷𝐷𝐷𝐷
𝐷𝐷𝐷𝐷𝐿𝐿𝐿𝐿
3 𝑆𝑆𝑆𝑆 3 + 𝐶𝐶𝐶𝐶 𝑊𝑊𝑊𝑊(𝜃𝜃𝜃𝜃) ∗
𝐷𝐷𝐷𝐷
𝐷𝐷𝐷𝐷𝐿𝐿𝐿𝐿𝐷𝐷𝐷𝐷 (6) 𝑇𝑇𝑇𝑇(𝜃𝜃𝜃𝜃)
1
𝐿𝐿𝐿𝐿𝑃𝑃𝑃𝑃 = ln(𝑍𝑍𝑍𝑍 = 1 𝐶𝐶𝐶𝐶
2 11𝑃𝑃𝑃𝑃 ) ,1
1 𝑊𝑊𝑊𝑊(𝜃𝜃𝜃𝜃)
21 𝐿𝐿𝐿𝐿𝑆𝑆𝑆𝑆 = ln(𝑍𝑍𝑍𝑍 ∗
2 𝐷𝐷𝐷𝐷𝐿𝐿𝐿𝐿 +
𝑃𝑃𝑃𝑃 𝑆𝑆𝑆𝑆 ) ,2and
3 𝐶𝐶𝐶𝐶 𝑊𝑊𝑊𝑊(𝜃𝜃𝜃𝜃) 2 ∗
𝐿𝐿𝐿𝐿𝐷𝐷𝐷𝐷 =1 𝐷𝐷𝐷𝐷𝐿𝐿𝐿𝐿 2
2 ln(𝜌𝜌𝜌𝜌).
𝑆𝑆𝑆𝑆 + 𝐶𝐶𝐶𝐶3 𝑊𝑊𝑊𝑊(𝜃𝜃𝜃𝜃) ∗ 𝐷𝐷𝐷𝐷𝐿𝐿𝐿𝐿 𝐷𝐷𝐷𝐷 (6)
(𝑆𝑆𝑆𝑆 2−+𝑊𝑊𝑊𝑊𝑚𝑚𝑚𝑚𝑐𝑐𝑐𝑐 2
𝐿𝐿𝐿𝐿
� 𝑃𝑃𝑃𝑃 [𝐿𝐿𝐿𝐿 =⋮𝑃𝑃𝑃𝑃ln(𝑍𝑍𝑍𝑍
Δ𝐿𝐿𝐿𝐿�𝑆𝑆𝑆𝑆𝐿𝐿𝐿𝐿Δ𝐿𝐿𝐿𝐿 𝑃𝑃𝑃𝑃 )= 𝐷𝐷𝐷𝐷,� 𝐿𝐿𝐿𝐿=𝑆𝑆𝑆𝑆 [𝑙𝑙𝑙𝑙 =𝑃𝑃𝑃𝑃𝑠𝑠𝑠𝑠)(𝑍𝑍𝑍𝑍 ln(𝑍𝑍𝑍𝑍 ⋮ 0𝑆𝑆𝑆𝑆 )= , and 𝑃𝑃𝑃𝑃 = 𝐿𝐿𝐿𝐿 𝐷𝐷𝐷𝐷)⋱ 𝑃𝑃𝑃𝑃,= , ln(𝜌𝜌𝜌𝜌). ⋮ 𝑒𝑒𝑒𝑒(𝑟𝑟𝑟𝑟) = 𝛼𝛼𝛼𝛼 ��𝑟𝑟𝑟𝑟 𝐶𝐶𝐶𝐶1 =𝑗𝑗𝑗𝑗 � +𝑐𝑐𝑐𝑐1 +∥ 𝐾𝐾𝐾𝐾𝑐𝑐𝑐𝑐 ∗ 3𝑟𝑟𝑟𝑟) , ∥ 𝐶𝐶𝐶𝐶2 = 𝑐𝑐𝑐𝑐2 , (12)(12)
𝑊𝑊𝑊𝑊(𝜃𝜃𝜃𝜃)
Where 𝑃𝑃𝑃𝑃= ] ln(𝑍𝑍𝑍𝑍
1 , 𝑃𝑃𝑃𝑃0 𝐿𝐿𝐿𝐿 𝑆𝑆𝑆𝑆 0)], ln(𝑍𝑍𝑍𝑍
𝐿𝐿𝐿𝐿 ln(𝑍𝑍𝑍𝑍
𝑆𝑆𝑆𝑆 )and 𝐿𝐿𝐿𝐿𝑆𝑆𝑆𝑆 =𝐿𝐿𝐿𝐿𝐷𝐷𝐷𝐷
𝑉𝑉𝑉𝑉𝑆𝑆𝑆𝑆 2 =
ln(𝑍𝑍𝑍𝑍 𝑆𝑆𝑆𝑆 ) ,ln(𝜌𝜌𝜌𝜌).
and The
𝐿𝐿𝐿𝐿𝐷𝐷𝐷𝐷 = ln(𝜌𝜌𝜌𝜌). � other 𝑉𝑉𝑉𝑉𝑆𝑆𝑆𝑆 2 (7) 2 2𝑃𝑃𝑃𝑃 ) 𝐶𝐶𝐶𝐶 2𝜎𝜎𝜎𝜎1𝐿𝐿𝐿𝐿(𝜃𝜃𝜃𝜃 2(𝜃𝜃𝜃𝜃
2 2 𝐿𝐿𝐿𝐿𝑇𝑇𝑇𝑇(𝜃𝜃𝜃𝜃
𝑃𝑃𝑃𝑃 =1ln(𝑍𝑍𝑍𝑍 )1 ,1 𝑆𝑆𝑆𝑆 1= )𝑊𝑊𝑊𝑊(𝜃𝜃𝜃𝜃
ln(𝑍𝑍𝑍𝑍1 𝑆𝑆𝑆𝑆)𝐷𝐷𝐷𝐷 ) , and ⋯ 𝐿𝐿𝐿𝐿𝐷𝐷𝐷𝐷 𝐶𝐶𝐶𝐶=31 ln(𝜌𝜌𝜌𝜌).3 )𝑊𝑊𝑊𝑊(𝜃𝜃𝜃𝜃3 )𝐷𝐷𝐷𝐷
where
constants
𝑆𝑆𝑆𝑆(𝑡𝑡𝑡𝑡)𝑇𝑇𝑇𝑇(𝜃𝜃𝜃𝜃=𝑁𝑁𝑁𝑁1𝑊𝑊𝑊𝑊(𝑡𝑡𝑡𝑡)𝑐𝑐𝑐𝑐)1 =values 𝐶𝐶𝐶𝐶11 (𝜃𝜃𝜃𝜃
∗1𝑟𝑟𝑟𝑟(𝑡𝑡𝑡𝑡) +𝑁𝑁𝑁𝑁 are , like
)𝑊𝑊𝑊𝑊(𝜃𝜃𝜃𝜃
𝑠𝑠𝑠𝑠(𝑡𝑡𝑡𝑡)
1 𝑐𝑐𝑐𝑐𝑁𝑁𝑁𝑁2 )𝐷𝐷𝐷𝐷 𝐶𝐶𝐶𝐶= 0.5 1 −1�
𝑐𝑐𝑐𝑐⋯ +1 𝐾𝐾𝐾𝐾𝑐𝑐𝑐𝑐𝐶𝐶𝐶𝐶 + �(𝜃𝜃𝜃𝜃
𝑚𝑚𝑚𝑚𝑐𝑐𝑐𝑐𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠
31 , 𝑠𝑠𝑠𝑠
)𝑊𝑊𝑊𝑊(𝜃𝜃𝜃𝜃𝜃𝜃𝜃𝜃𝐶𝐶𝐶𝐶(8) and 1 𝑐𝑐𝑐𝑐 = 4 �
2 = 𝑁𝑁𝑁𝑁𝑐𝑐𝑐𝑐)𝐷𝐷𝐷𝐷 , 3 𝑊𝑊𝑊𝑊(𝜃𝜃𝜃𝜃) is � 𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠 𝜃𝜃𝜃𝜃
𝑗𝑗𝑗𝑗 =1
𝐶𝐶𝐶𝐶 1 = 𝑐𝑐𝑐𝑐1=+� 𝐾𝐾𝐾𝐾𝑐𝑐𝑐𝑐2 +⋮ 𝑚𝑚𝑚𝑚𝑐𝑐𝑐𝑐3 , 𝐶𝐶𝐶𝐶 2 = 𝑐𝑐𝑐𝑐2 , ⋮ 𝑊𝑊𝑊𝑊(𝜃𝜃𝜃𝜃)�
2𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑠𝑠𝑠𝑠 2 𝜃𝜃𝜃𝜃 1 =
2 1 2 𝑉𝑉𝑉𝑉2𝑃𝑃𝑃𝑃3 𝑁𝑁𝑁𝑁 2 2 𝑉𝑉𝑉𝑉 � 𝑇𝑇𝑇𝑇(𝜃𝜃𝜃𝜃 ⋮ 1 )2 �
1 optimization 𝐶𝐶𝐶𝐶
2
1 (𝜃𝜃𝜃𝜃 )𝑊𝑊𝑊𝑊(𝜃𝜃𝜃𝜃 )𝐷𝐷𝐷𝐷 ⋯⋱ 𝐶𝐶𝐶𝐶 2 (𝜃𝜃𝜃𝜃
1 for 3 )𝑊𝑊𝑊𝑊(𝜃𝜃𝜃𝜃3 )𝐷𝐷𝐷𝐷
(7)
𝐶𝐶𝐶𝐶1 = 𝑐𝑐𝑐𝑐𝐶𝐶𝐶𝐶11+= 𝐾𝐾𝐾𝐾𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐12 + 𝑚𝑚𝑚𝑚𝑐𝑐𝑐𝑐 𝐾𝐾𝐾𝐾𝑐𝑐𝑐𝑐 3 ,2 + 𝑚𝑚𝑚𝑚𝑐𝑐𝑐𝑐 𝐶𝐶𝐶𝐶23= , 𝐶𝐶𝐶𝐶 𝑐𝑐𝑐𝑐(𝜃𝜃𝜃𝜃𝐶𝐶𝐶𝐶22,)𝑊𝑊𝑊𝑊(𝜃𝜃𝜃𝜃 = )𝐷𝐷𝐷𝐷𝑊𝑊𝑊𝑊(𝜃𝜃𝜃𝜃) 𝑐𝑐𝑐𝑐2⋯, 𝐶𝐶𝐶𝐶 (𝜃𝜃𝜃𝜃𝑊𝑊𝑊𝑊(𝜃𝜃𝜃𝜃) 𝑃𝑃𝑃𝑃
The least squares
1 1 1 3
the
𝑍𝑍𝑍𝑍 = 𝑉𝑉𝑉𝑉𝜌𝜌𝜌𝜌 wavelet
[𝐿𝐿𝐿𝐿 2 Δ𝐿𝐿𝐿𝐿 Δ𝐿𝐿𝐿𝐿 2at 2 angle ] = 2 [𝑙𝑙𝑙𝑙θ, 𝑠𝑠𝑠𝑠 (𝑍𝑍𝑍𝑍 D 𝑇𝑇𝑇𝑇(𝜃𝜃𝜃𝜃 is0 1) the 0)], 2differentiation
1 1 (9) 12 3 derivative
3 )𝑊𝑊𝑊𝑊(𝜃𝜃𝜃𝜃3 )𝐷𝐷𝐷𝐷 𝜆𝜆𝜆𝜆 + 𝐶𝐶𝐶𝐶
� 𝑇𝑇𝑇𝑇(𝜃𝜃𝜃𝜃
12𝜇𝜇𝜇𝜇 =⋮ 𝑁𝑁𝑁𝑁))�𝑐𝑐𝑐𝑐1=
𝑇𝑇𝑇𝑇(𝜃𝜃𝜃𝜃 +�𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶1𝐾𝐾𝐾𝐾𝑐𝑐𝑐𝑐 (𝜃𝜃𝜃𝜃𝑁𝑁𝑁𝑁2)𝑊𝑊𝑊𝑊(𝜃𝜃𝜃𝜃
(𝜃𝜃𝜃𝜃 +⋮ 𝑚𝑚𝑚𝑚𝑐𝑐𝑐𝑐𝑁𝑁𝑁𝑁is
)𝑊𝑊𝑊𝑊(𝜃𝜃𝜃𝜃 , performed
)𝐷𝐷𝐷𝐷
3)𝐷𝐷𝐷𝐷 ⋯
⋯𝐶𝐶𝐶𝐶
⋱ 2 =𝐶𝐶𝐶𝐶 3 (𝜃𝜃𝜃𝜃
𝐶𝐶𝐶𝐶 𝑐𝑐𝑐𝑐2𝑁𝑁𝑁𝑁,)𝑊𝑊𝑊𝑊(𝜃𝜃𝜃𝜃
(𝜃𝜃𝜃𝜃
minimizing
⋮
)𝑊𝑊𝑊𝑊(𝜃𝜃𝜃𝜃
)𝐷𝐷𝐷𝐷�
𝑊𝑊𝑊𝑊(𝜃𝜃𝜃𝜃)
𝑁𝑁𝑁𝑁)𝐷𝐷𝐷𝐷 (7)
𝑇𝑇𝑇𝑇(𝜃𝜃𝜃𝜃) =𝑃𝑃𝑃𝑃 𝐶𝐶𝐶𝐶1 𝑊𝑊𝑊𝑊(𝜃𝜃𝜃𝜃) 𝑆𝑆𝑆𝑆 𝐷𝐷𝐷𝐷 ∗ 𝐷𝐷𝐷𝐷𝐿𝐿𝐿𝐿𝑃𝑃𝑃𝑃 + 𝐶𝐶𝐶𝐶2 𝑊𝑊𝑊𝑊(𝜃𝜃𝜃𝜃) 𝑃𝑃𝑃𝑃0
� ⋮ �= ∗ �𝐷𝐷𝐷𝐷𝐿𝐿𝐿𝐿𝑆𝑆𝑆𝑆 +⋮ 𝐶𝐶𝐶𝐶3 𝑊𝑊𝑊𝑊(𝜃𝜃𝜃𝜃) ⋱ ∗ 𝐷𝐷𝐷𝐷𝐿𝐿𝐿𝐿⋮𝐷𝐷𝐷𝐷 � (6)(7) 𝑉𝑉𝑉𝑉𝑃𝑃𝑃𝑃 =difference
the � 12
between 21 the 1
real 1
and the modeled 32(13) 3
reflectivity3
operator. ) −𝑇𝑇𝑇𝑇(𝜃𝜃𝜃𝜃
𝑇𝑇𝑇𝑇(𝜃𝜃𝜃𝜃𝑍𝑍𝑍𝑍1𝑠𝑠𝑠𝑠+1 If𝑍𝑍𝑍𝑍𝑠𝑠𝑠𝑠𝐶𝐶𝐶𝐶in the above
11)(𝜃𝜃𝜃𝜃1 )𝑊𝑊𝑊𝑊(𝜃𝜃𝜃𝜃 𝐶𝐶𝐶𝐶1 (𝜃𝜃𝜃𝜃 1 )𝐷𝐷𝐷𝐷 1 )𝑊𝑊𝑊𝑊(𝜃𝜃𝜃𝜃
equation ⋯𝑁𝑁𝑁𝑁1))𝐷𝐷𝐷𝐷𝐶𝐶𝐶𝐶3𝐶𝐶𝐶𝐶(𝜃𝜃𝜃𝜃
𝑇𝑇𝑇𝑇(𝜃𝜃𝜃𝜃 θ=0
1 (𝜃𝜃𝜃𝜃 )𝑊𝑊𝑊𝑊(𝜃𝜃𝜃𝜃
⋯𝑁𝑁𝑁𝑁3)𝑊𝑊𝑊𝑊(𝜃𝜃𝜃𝜃 𝐶𝐶𝐶𝐶𝑁𝑁𝑁𝑁then
3)𝐷𝐷𝐷𝐷 3 )𝐷𝐷𝐷𝐷
(𝜃𝜃𝜃𝜃 3⋯)𝑊𝑊𝑊𝑊(𝜃𝜃𝜃𝜃
this equation
3 )𝐷𝐷𝐷𝐷 𝑁𝑁𝑁𝑁 )𝐷𝐷𝐷𝐷
𝐶𝐶𝐶𝐶3 (𝜃𝜃𝜃𝜃𝑁𝑁𝑁𝑁 )𝑊𝑊𝑊𝑊(𝜃𝜃𝜃𝜃 𝜌𝜌𝜌𝜌
� 𝑇𝑇𝑇𝑇(𝜃𝜃𝜃𝜃 [𝐿𝐿𝐿𝐿
⋮ 𝑁𝑁𝑁𝑁 )Δ𝐿𝐿𝐿𝐿 � = Δ𝐿𝐿𝐿𝐿
� 𝐶𝐶𝐶𝐶1 (𝜃𝜃𝜃𝜃 ] =
𝑁𝑁𝑁𝑁 )𝑊𝑊𝑊𝑊(𝜃𝜃𝜃𝜃
⋮[𝑙𝑙𝑙𝑙 𝑠𝑠𝑠𝑠 (𝑍𝑍𝑍𝑍 𝑁𝑁𝑁𝑁 )𝐷𝐷𝐷𝐷 0 0)], ⋯⋱ 𝐶𝐶𝐶𝐶3 (𝜃𝜃𝜃𝜃 𝑁𝑁𝑁𝑁 )𝑊𝑊𝑊𝑊(𝜃𝜃𝜃𝜃⋮ 𝑁𝑁𝑁𝑁 )𝐷𝐷𝐷𝐷 � (7)
𝑠𝑠𝑠𝑠 ==
𝑟𝑟𝑟𝑟𝐿𝐿𝐿𝐿
�𝑆𝑆𝑆𝑆(𝑡𝑡𝑡𝑡) = 𝑊𝑊𝑊𝑊(𝑡𝑡𝑡𝑡) �)𝑠𝑠𝑠𝑠⋮,zero-offset 𝐿𝐿𝐿𝐿�∗ 𝑟𝑟𝑟𝑟(𝑡𝑡𝑡𝑡) +𝑆𝑆𝑆𝑆 )𝑠𝑠𝑠𝑠(𝑡𝑡𝑡𝑡) (10) (8) 𝑇𝑇𝑇𝑇(𝜃𝜃𝜃𝜃1is) achieved 𝑃𝑃𝑃𝑃 𝑆𝑆𝑆𝑆 𝐶𝐶𝐶𝐶1 (𝜃𝜃𝜃𝜃1 )𝑊𝑊𝑊𝑊(𝜃𝜃𝜃𝜃
𝐷𝐷𝐷𝐷
1 )𝐷𝐷𝐷𝐷
𝑃𝑃𝑃𝑃0 ⋯ the 𝐶𝐶𝐶𝐶3 (𝜃𝜃𝜃𝜃3 )𝑊𝑊𝑊𝑊(𝜃𝜃𝜃𝜃between 3 )𝐷𝐷𝐷𝐷
reduces 𝑃𝑃𝑃𝑃 ⋮ 𝑍𝑍𝑍𝑍ln(𝑍𝑍𝑍𝑍 � �= + to 𝑃𝑃𝑃𝑃𝑍𝑍𝑍𝑍 𝑆𝑆𝑆𝑆 = = �⋮ln(𝑍𝑍𝑍𝑍 , ⋮and
(model-based) ⋱[𝐿𝐿𝐿𝐿𝐿𝐿𝐿𝐿 𝐷𝐷𝐷𝐷 𝑆𝑆𝑆𝑆 = Δ𝐿𝐿𝐿𝐿𝐷𝐷𝐷𝐷ln(𝜌𝜌𝜌𝜌). ]⋱= [𝑙𝑙𝑙𝑙⋮𝑠𝑠𝑠𝑠inversion (𝑍𝑍𝑍𝑍𝑃𝑃𝑃𝑃0 0 0)], � ⋮ methods � (7) (7) section. This 𝑇𝑇𝑇𝑇(𝜃𝜃𝜃𝜃 by analyzing 𝐶𝐶𝐶𝐶3 (𝜃𝜃𝜃𝜃misfit
𝑁𝑁𝑁𝑁𝑃𝑃𝑃𝑃)Δ𝐿𝐿𝐿𝐿 𝐶𝐶𝐶𝐶1𝐷𝐷𝐷𝐷(𝜃𝜃𝜃𝜃 )𝑊𝑊𝑊𝑊(𝜃𝜃𝜃𝜃 𝑁𝑁𝑁𝑁 )𝐷𝐷𝐷𝐷 ⋯ 𝑁𝑁𝑁𝑁 )𝑊𝑊𝑊𝑊(𝜃𝜃𝜃𝜃 𝑁𝑁𝑁𝑁 )𝐷𝐷𝐷𝐷 �
𝑃𝑃𝑃𝑃 Δ𝐿𝐿𝐿𝐿
𝑠𝑠𝑠𝑠+1
�
𝑆𝑆𝑆𝑆(𝑡𝑡𝑡𝑡) = [𝐿𝐿𝐿𝐿
⋮ � =
𝑊𝑊𝑊𝑊(𝑡𝑡𝑡𝑡) 𝑆𝑆𝑆𝑆 Δ𝐿𝐿𝐿𝐿
� ]
∗ 𝑟𝑟𝑟𝑟(𝑡𝑡𝑡𝑡) =
𝑁𝑁𝑁𝑁 ⋮[𝑙𝑙𝑙𝑙
+real 𝑠𝑠𝑠𝑠 (𝑍𝑍𝑍𝑍 0 0)],
⋱
𝑠𝑠𝑠𝑠(𝑡𝑡𝑡𝑡)trace and modifying the (8)
𝑃𝑃𝑃𝑃0 ⋮ (7)
𝑇𝑇𝑇𝑇(𝜃𝜃𝜃𝜃𝑁𝑁𝑁𝑁1) 𝑇𝑇𝑇𝑇(𝜃𝜃𝜃𝜃 𝑝𝑝𝑝𝑝 𝐶𝐶𝐶𝐶 𝑁𝑁𝑁𝑁(𝜃𝜃𝜃𝜃 𝑁𝑁𝑁𝑁 )𝑊𝑊𝑊𝑊(𝜃𝜃𝜃𝜃 )𝐷𝐷𝐷𝐷 ⋯=𝑁𝑁𝑁𝑁 )𝐷𝐷𝐷𝐷 𝐶𝐶𝐶𝐶 31(𝜃𝜃𝜃𝜃 𝑁𝑁𝑁𝑁 )𝑊𝑊𝑊𝑊(𝜃𝜃𝜃𝜃 )𝐷𝐷𝐷𝐷
𝑉𝑉𝑉𝑉𝜌𝜌𝜌𝜌𝑐𝑐𝑐𝑐 𝑉𝑉𝑉𝑉 1= )1.16𝑉𝑉𝑉𝑉 𝐶𝐶𝐶𝐶𝑠𝑠𝑠𝑠 1+ (𝜃𝜃𝜃𝜃 1360 )𝑊𝑊𝑊𝑊(𝜃𝜃𝜃𝜃
(9) (9) in(1)
∗⋯ +𝐶𝐶𝐶𝐶𝑠𝑠𝑠𝑠(𝑡𝑡𝑡𝑡)
3 (𝜃𝜃𝜃𝜃 )𝑊𝑊𝑊𝑊(𝜃𝜃𝜃𝜃𝑁𝑁𝑁𝑁 )𝐷𝐷𝐷𝐷 the synthetic trace and the
𝑆𝑆𝑆𝑆(𝑡𝑡𝑡𝑡) 𝑊𝑊𝑊𝑊(𝑡𝑡𝑡𝑡) 𝑟𝑟𝑟𝑟(𝑡𝑡𝑡𝑡) (8)
(Shu-jin,
𝑍𝑍𝑍𝑍 = 2007). 1
𝑁𝑁𝑁𝑁 The equation
𝑁𝑁𝑁𝑁 𝑁𝑁𝑁𝑁 (6) can be 𝑁𝑁𝑁𝑁implemented
𝑁𝑁𝑁𝑁 𝜇𝜇𝜇𝜇 𝑇𝑇𝑇𝑇(𝜃𝜃𝜃𝜃 [𝐿𝐿𝐿𝐿 ) Δ𝐿𝐿𝐿𝐿 Δ𝐿𝐿𝐿𝐿 𝐶𝐶𝐶𝐶 (𝜃𝜃𝜃𝜃
] = )𝑊𝑊𝑊𝑊(𝜃𝜃𝜃𝜃
[𝑙𝑙𝑙𝑙 𝑠𝑠𝑠𝑠 (𝑍𝑍𝑍𝑍 )𝐷𝐷𝐷𝐷 0 ⋯
0)], 𝐶𝐶𝐶𝐶 (𝜃𝜃𝜃𝜃 )𝑊𝑊𝑊𝑊(𝜃𝜃𝜃𝜃 )𝐷𝐷𝐷𝐷
𝐶𝐶𝐶𝐶 = + 𝐾𝐾𝐾𝐾𝑐𝑐𝑐𝑐 + 𝑚𝑚𝑚𝑚𝑐𝑐𝑐𝑐 , 𝑍𝑍𝑍𝑍 = 𝐶𝐶𝐶𝐶 𝑉𝑉𝑉𝑉𝜌𝜌𝜌𝜌 = 𝑐𝑐𝑐𝑐 , 𝑊𝑊𝑊𝑊(𝜃𝜃𝜃𝜃) 𝑉𝑉𝑉𝑉𝑆𝑆𝑆𝑆 = �size𝑆𝑆𝑆𝑆(𝑡𝑡𝑡𝑡) = 𝑊𝑊𝑊𝑊(𝑡𝑡𝑡𝑡) ∗ 𝑟𝑟𝑟𝑟(𝑡𝑡𝑡𝑡) + 𝑠𝑠𝑠𝑠(𝑡𝑡𝑡𝑡)
𝑁𝑁𝑁𝑁 𝑃𝑃𝑃𝑃 𝑆𝑆𝑆𝑆 1𝐷𝐷𝐷𝐷 𝑁𝑁𝑁𝑁 𝑁𝑁𝑁𝑁 𝑃𝑃𝑃𝑃0 3
(14)error (Ferguson𝑁𝑁𝑁𝑁 𝑁𝑁𝑁𝑁 (8)
𝑍𝑍𝑍𝑍𝑁𝑁𝑁𝑁1 =[𝐿𝐿𝐿𝐿𝑍𝑍𝑍𝑍21Δ𝐿𝐿𝐿𝐿 block 𝜌𝜌𝜌𝜌 𝑍𝑍𝑍𝑍and = 𝑉𝑉𝑉𝑉𝜌𝜌𝜌𝜌 the amplitude to reduce the (9)
1 �� 𝑟𝑟𝑟𝑟𝑠𝑠𝑠𝑠 2
matrix 𝑃𝑃𝑃𝑃 form
𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑝𝑝𝑝𝑝 Δ𝐿𝐿𝐿𝐿 2𝐷𝐷𝐷𝐷𝑃𝑃𝑃𝑃as
[𝐿𝐿𝐿𝐿 ]Δ𝐿𝐿𝐿𝐿= �following. [𝑙𝑙𝑙𝑙0.25
𝑆𝑆𝑆𝑆 Δ𝐿𝐿𝐿𝐿𝑠𝑠𝑠𝑠
(𝑍𝑍𝑍𝑍3 = 0[𝑙𝑙𝑙𝑙0)],
𝐷𝐷𝐷𝐷 ] 𝑃𝑃𝑃𝑃0 𝑠𝑠𝑠𝑠(𝑍𝑍𝑍𝑍
2
𝑃𝑃𝑃𝑃0−0 2𝑍𝑍𝑍𝑍 0)], 2 (11)
𝑍𝑍𝑍𝑍𝑠𝑠𝑠𝑠+1𝑆𝑆𝑆𝑆𝜌𝜌𝜌𝜌−=𝑠𝑠𝑠𝑠=1 𝑍𝑍𝑍𝑍0.23𝑉𝑉𝑉𝑉𝑠𝑠𝑠𝑠 𝑟𝑟𝑟𝑟𝑠𝑠𝑠𝑠 =
𝑍𝑍𝑍𝑍𝑠𝑠𝑠𝑠+1 𝑠𝑠𝑠𝑠
(2)
(10)
and Margrave,𝑍𝑍𝑍𝑍1996). 𝑆𝑆𝑆𝑆(𝑡𝑡𝑡𝑡)
𝑍𝑍𝑍𝑍 = [𝐿𝐿𝐿𝐿𝑉𝑉𝑉𝑉𝜌𝜌𝜌𝜌
= 𝑃𝑃𝑃𝑃 Δ𝐿𝐿𝐿𝐿𝑊𝑊𝑊𝑊(𝑡𝑡𝑡𝑡) 𝑆𝑆𝑆𝑆 Δ𝐿𝐿𝐿𝐿𝐷𝐷𝐷𝐷∗]𝑟𝑟𝑟𝑟(𝑡𝑡𝑡𝑡) = [𝑙𝑙𝑙𝑙 + 𝑠𝑠𝑠𝑠(𝑍𝑍𝑍𝑍 𝑠𝑠𝑠𝑠(𝑡𝑡𝑡𝑡)𝑃𝑃𝑃𝑃0 0 0)], (9) (8)
𝑟𝑟𝑟𝑟𝑠𝑠𝑠𝑠𝑇𝑇𝑇𝑇(𝜃𝜃𝜃𝜃
𝑆𝑆𝑆𝑆(𝑡𝑡𝑡𝑡) = 1=) 𝑊𝑊𝑊𝑊(𝑡𝑡𝑡𝑡) 𝐶𝐶𝐶𝐶= ∗𝑆𝑆𝑆𝑆(𝜃𝜃𝜃𝜃 1 )𝑊𝑊𝑊𝑊(𝜃𝜃𝜃𝜃 )𝐷𝐷𝐷𝐷 +⋯𝑠𝑠𝑠𝑠(𝑡𝑡𝑡𝑡) 𝑍𝑍𝑍𝑍𝑠𝑠𝑠𝑠+1 𝐶𝐶𝐶𝐶 +3𝑍𝑍𝑍𝑍(𝜃𝜃𝜃𝜃 𝑠𝑠𝑠𝑠 3 )𝑊𝑊𝑊𝑊(𝜃𝜃𝜃𝜃3 )𝐷𝐷𝐷𝐷 (10) (8)
(8) 𝑠𝑠𝑠𝑠+1 − 𝑍𝑍𝑍𝑍𝑠𝑠𝑠𝑠
𝑍𝑍𝑍𝑍 𝑆𝑆𝑆𝑆(𝑡𝑡𝑡𝑡)
𝑀𝑀𝑀𝑀ln(𝑍𝑍𝑍𝑍
+
1
𝑍𝑍𝑍𝑍 )𝑟𝑟𝑟𝑟(𝑡𝑡𝑡𝑡)
𝑊𝑊𝑊𝑊(𝑡𝑡𝑡𝑡)
= 𝑘𝑘𝑘𝑘ln(𝑍𝑍𝑍𝑍 + ∗𝑠𝑠𝑠𝑠(𝑡𝑡𝑡𝑡) 1𝑟𝑟𝑟𝑟(𝑡𝑡𝑡𝑡)
𝑃𝑃𝑃𝑃 ) + 𝑘𝑘𝑘𝑘𝐶𝐶𝐶𝐶 + ∆𝐿𝐿𝐿𝐿𝑆𝑆𝑆𝑆𝑁𝑁𝑁𝑁 (3) 𝑍𝑍𝑍𝑍 2
= (𝜌𝜌𝜌𝜌𝑉𝑉𝑉𝑉 𝑟𝑟𝑟𝑟)𝑠𝑠𝑠𝑠2 == = 𝜇𝜇𝜇𝜇𝜌𝜌𝜌𝜌 (15) (10) (8)
� ⋮ �=� 1 ⋮1 ⋱ ⋮ � 𝑆𝑆𝑆𝑆(𝑡𝑡𝑡𝑡)
𝑍𝑍𝑍𝑍 = 𝑉𝑉𝑉𝑉𝜌𝜌𝜌𝜌 𝑊𝑊𝑊𝑊(𝑡𝑡𝑡𝑡) ∗ 𝑟𝑟𝑟𝑟(𝑡𝑡𝑡𝑡) + 𝑠𝑠𝑠𝑠(𝑡𝑡𝑡𝑡) (9)
(7)(7)
𝑠𝑠𝑠𝑠+1 𝑠𝑠𝑠𝑠 𝑆𝑆𝑆𝑆 𝑆𝑆𝑆𝑆 𝑍𝑍𝑍𝑍𝑠𝑠𝑠𝑠+1 − + 𝑍𝑍𝑍𝑍𝑠𝑠𝑠𝑠
𝑒𝑒𝑒𝑒(𝑟𝑟𝑟𝑟) == 𝛼𝛼𝛼𝛼 ��𝑟𝑟𝑟𝑟𝑗𝑗𝑗𝑗 � + ∥ (𝑆𝑆𝑆𝑆 − 𝑊𝑊𝑊𝑊 ∗𝑍𝑍𝑍𝑍𝑁𝑁𝑁𝑁𝑟𝑟𝑟𝑟)= ∥𝑍𝑍𝑍𝑍21 𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑝𝑝𝑝𝑝 �� 𝑟𝑟𝑟𝑟𝑠𝑠𝑠𝑠 � (12)
𝑍𝑍𝑍𝑍𝑇𝑇𝑇𝑇(𝜃𝜃𝜃𝜃 𝑉𝑉𝑉𝑉𝜌𝜌𝜌𝜌
𝑁𝑁𝑁𝑁 ) 𝑍𝑍𝑍𝑍 ln⁡=𝐶𝐶𝐶𝐶(𝜌𝜌𝜌𝜌) 𝑉𝑉𝑉𝑉𝜌𝜌𝜌𝜌 (𝜃𝜃𝜃𝜃= 𝑚𝑚𝑚𝑚ln(𝑍𝑍𝑍𝑍
𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁)𝑊𝑊𝑊𝑊(𝜃𝜃𝜃𝜃
2 𝜎𝜎𝜎𝜎 𝑁𝑁𝑁𝑁 )𝐷𝐷𝐷𝐷 𝑃𝑃𝑃𝑃 ) + ⋯ 𝑚𝑚𝑚𝑚𝐶𝐶𝐶𝐶 + 𝐶𝐶𝐶𝐶3∆𝐿𝐿𝐿𝐿 (𝜃𝜃𝜃𝜃𝑠𝑠𝑠𝑠=1𝑁𝑁𝑁𝑁 𝐷𝐷𝐷𝐷 )𝑊𝑊𝑊𝑊(𝜃𝜃𝜃𝜃𝑁𝑁𝑁𝑁 )𝐷𝐷𝐷𝐷
(9) (9) (11)(4) LMR
𝑟𝑟𝑟𝑟𝑠𝑠𝑠𝑠 =2 (10)
𝑗𝑗𝑗𝑗 =1 1 𝜆𝜆𝜆𝜆𝜌𝜌𝜌𝜌 = 𝑍𝑍𝑍𝑍𝑃𝑃𝑃𝑃Transform 2
−𝑍𝑍𝑍𝑍2𝑍𝑍𝑍𝑍 =𝑆𝑆𝑆𝑆𝑉𝑉𝑉𝑉𝜌𝜌𝜌𝜌 𝑍𝑍𝑍𝑍𝑠𝑠𝑠𝑠+1 − + 𝑍𝑍𝑍𝑍𝑠𝑠𝑠𝑠 𝑁𝑁𝑁𝑁 (16) (9)
𝑍𝑍𝑍𝑍𝑠𝑠𝑠𝑠+1 − 𝑟𝑟𝑟𝑟 = (10)
Δ𝐿𝐿𝐿𝐿𝑍𝑍𝑍𝑍𝐷𝐷𝐷𝐷𝑍𝑍𝑍𝑍
𝑀𝑀𝑀𝑀

𝑟𝑟𝑟𝑟𝑍𝑍𝑍𝑍𝑠𝑠𝑠𝑠 𝑁𝑁𝑁𝑁=[𝐿𝐿𝐿𝐿
= 𝑍𝑍𝑍𝑍𝑟𝑟𝑟𝑟1𝑆𝑆𝑆𝑆𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑝𝑝𝑝𝑝
𝑃𝑃𝑃𝑃 Δ𝐿𝐿𝐿𝐿
Solve 𝑠𝑠𝑠𝑠 �� ]𝑠𝑠𝑠𝑠+1 =−[𝑙𝑙𝑙𝑙𝑟𝑟𝑟𝑟𝑍𝑍𝑍𝑍∆𝑉𝑉𝑉𝑉 𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠�(𝑍𝑍𝑍𝑍 𝑃𝑃𝑃𝑃 𝑃𝑃𝑃𝑃0 0𝑒𝑒𝑒𝑒(𝑟𝑟𝑟𝑟) ∆𝑉𝑉𝑉𝑉
0)], ∆𝜌𝜌𝜌𝜌1 1 (𝑆𝑆𝑆𝑆 − 𝑊𝑊𝑊𝑊 ∗ 𝑟𝑟𝑟𝑟) ∥2 methods, (11)(12) 𝑍𝑍𝑍𝑍𝑠𝑠𝑠𝑠𝑁𝑁𝑁𝑁 =−𝑍𝑍𝑍𝑍
𝜆𝜆𝜆𝜆 + =the
𝑅𝑅𝑅𝑅(𝜃𝜃𝜃𝜃)
𝑠𝑠𝑠𝑠 2𝜇𝜇𝜇𝜇 =above 𝑐𝑐𝑐𝑐1 𝑠𝑠𝑠𝑠 matrix + 𝑐𝑐𝑐𝑐2 by =𝑆𝑆𝑆𝑆 𝛼𝛼𝛼𝛼 matrix
+��𝑟𝑟𝑟𝑟 𝑐𝑐𝑐𝑐 𝑗𝑗𝑗𝑗 � + 2 ∥ inversion 𝜎𝜎𝜎𝜎 (10) (10) (5) 𝑍𝑍𝑍𝑍𝑆𝑆𝑆𝑆 = 0.741𝑍𝑍𝑍𝑍 𝑃𝑃𝑃𝑃
𝑍𝑍𝑍𝑍𝑍𝑍𝑍𝑍
1900
𝑠𝑠𝑠𝑠+1
𝑠𝑠𝑠𝑠+11 𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑝𝑝𝑝𝑝 + 𝑍𝑍𝑍𝑍
− 𝑍𝑍𝑍𝑍�� 𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠 𝑁𝑁𝑁𝑁 𝑟𝑟𝑟𝑟𝑠𝑠𝑠𝑠 � (13) (11)
� 𝑍𝑍𝑍𝑍
𝑉𝑉𝑉𝑉𝑃𝑃𝑃𝑃 = initialize 𝑠𝑠𝑠𝑠+1 + 𝑍𝑍𝑍𝑍 𝑍𝑍𝑍𝑍 + 𝑍𝑍𝑍𝑍 𝑉𝑉𝑉𝑉 𝑉𝑉𝑉𝑉 𝑗𝑗𝑗𝑗 =1 3
𝜌𝜌𝜌𝜌 (13) The LMR 𝑟𝑟𝑟𝑟 =
𝑍𝑍𝑍𝑍𝑠𝑠𝑠𝑠𝑁𝑁𝑁𝑁 (Lambda-mu-rho) method was originally (10)
and
𝑆𝑆𝑆𝑆(𝑡𝑡𝑡𝑡) = 𝑊𝑊𝑊𝑊(𝑡𝑡𝑡𝑡)
𝑠𝑠𝑠𝑠+1
𝜌𝜌𝜌𝜌 𝑉𝑉𝑉𝑉 𝑉𝑉𝑉𝑉𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝 =
𝑠𝑠𝑠𝑠
=∗ 1.16𝑉𝑉𝑉𝑉 the
1.16𝑉𝑉𝑉𝑉𝑠𝑠𝑠𝑠=1 +
𝑟𝑟𝑟𝑟(𝑡𝑡𝑡𝑡) solution 𝑠𝑠𝑠𝑠 𝑃𝑃𝑃𝑃
+𝑠𝑠𝑠𝑠(𝑡𝑡𝑡𝑡) 1360 to𝑆𝑆𝑆𝑆 𝜆𝜆𝜆𝜆[L PΔLSΔLD] = [ln(ZP0 (8)0 0)],(1) (1) 𝜌𝜌𝜌𝜌 = 0.000113𝑍𝑍𝑍𝑍 =𝑍𝑍𝑍𝑍𝑍𝑍𝑍𝑍 + 1 𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑝𝑝𝑝𝑝 + 𝑍𝑍𝑍𝑍��
1.57 𝑠𝑠𝑠𝑠=1 𝑁𝑁𝑁𝑁 𝑟𝑟𝑟𝑟𝑠𝑠𝑠𝑠 � (14) (11)
𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠 + 1360 𝑃𝑃𝑃𝑃 𝑠𝑠𝑠𝑠+1 𝑠𝑠𝑠𝑠
where ZP0 𝑝𝑝𝑝𝑝is 𝑉𝑉𝑉𝑉 =
𝑀𝑀𝑀𝑀 the initial 1.16𝑉𝑉𝑉𝑉
𝑁𝑁𝑁𝑁 + 𝑁𝑁𝑁𝑁 1360 impedance + 2𝜇𝜇𝜇𝜇
model. 𝑉𝑉𝑉𝑉𝑆𝑆𝑆𝑆 2 (1) 𝑉𝑉𝑉𝑉𝑆𝑆𝑆𝑆proposed 2 by𝑍𝑍𝑍𝑍𝑁𝑁𝑁𝑁𝑃𝑃𝑃𝑃Goodway
=+ 𝑍𝑍𝑍𝑍889 𝑀𝑀𝑀𝑀 et 𝑠𝑠𝑠𝑠=1al., (1997). LMR uses the following
1 𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑝𝑝𝑝𝑝 �� 𝑁𝑁𝑁𝑁 𝑟𝑟𝑟𝑟𝑠𝑠𝑠𝑠1� 1 (15) (11)
𝑠𝑠𝑠𝑠
1 11𝑟𝑟𝑟𝑟 � , 𝜌𝜌𝜌𝜌 𝑐𝑐𝑐𝑐2 = 0.5 𝑉𝑉𝑉𝑉 = 0.296𝑍𝑍𝑍𝑍
𝑉𝑉𝑉𝑉𝑃𝑃𝑃𝑃 = � (13)
𝑍𝑍𝑍𝑍 ==𝑉𝑉𝑉𝑉𝜌𝜌𝜌𝜌𝑍𝑍𝑍𝑍 𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑝𝑝𝑝𝑝 𝜌𝜌𝜌𝜌
where =�� 0.23𝑉𝑉𝑉𝑉 =0.25 (9) 𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠2 𝜃𝜃𝜃𝜃 and (2) � 𝑃𝑃𝑃𝑃 𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠2 𝜃𝜃𝜃𝜃 𝑒𝑒𝑒𝑒(𝑟𝑟𝑟𝑟)
𝑍𝑍𝑍𝑍
𝑒𝑒𝑒𝑒(𝑟𝑟𝑟𝑟)𝑁𝑁𝑁𝑁
= 𝜇𝜇𝜇𝜇 1𝑍𝑍𝑍𝑍𝛼𝛼𝛼𝛼 𝑁𝑁𝑁𝑁𝜌𝜌𝜌𝜌
𝜌𝜌𝜌𝜌 ��𝑟𝑟𝑟𝑟 = =
= 𝑍𝑍𝑍𝑍1𝑐𝑐𝑐𝑐𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑝𝑝𝑝𝑝
0.23𝑉𝑉𝑉𝑉
0.23𝑉𝑉𝑉𝑉 �
𝑟𝑟𝑟𝑟1𝑠𝑠𝑠𝑠 �
+
0.25
0.25 �� ∥
2𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑠𝑠𝑠𝑠 𝑠𝑠𝑠𝑠2(𝑆𝑆𝑆𝑆
𝜃𝜃𝜃𝜃 − 𝑊𝑊𝑊𝑊 ∗ 𝑟𝑟𝑟𝑟) ∥ 2 − � � (11)
𝑉𝑉𝑉𝑉 (2)
(2)(11) (12)
𝑐𝑐𝑐𝑐3 = 4 � relationships
𝑉𝑉𝑉𝑉 𝑉𝑉𝑉𝑉 = 0.26𝑍𝑍𝑍𝑍 − =
between
428 𝛼𝛼𝛼𝛼 ��𝑟𝑟𝑟𝑟 𝑀𝑀𝑀𝑀 𝑗𝑗𝑗𝑗𝑠𝑠𝑠𝑠=1 V �P+ , V S ∥, r (𝑆𝑆𝑆𝑆 and − the 𝑊𝑊𝑊𝑊 ∗ Lame 𝑟𝑟𝑟𝑟) ∥2 parameters (16) (11) - (12)
𝑉𝑉𝑉𝑉𝑆𝑆𝑆𝑆 = �𝑍𝑍𝑍𝑍 𝑗𝑗𝑗𝑗 (14) 𝑃𝑃𝑃𝑃 𝑃𝑃𝑃𝑃 𝑆𝑆𝑆𝑆 𝑍𝑍𝑍𝑍𝑃𝑃𝑃𝑃𝑁𝑁𝑁𝑁 = 𝑍𝑍𝑍𝑍1 𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑝𝑝𝑝𝑝 �� 𝑟𝑟𝑟𝑟𝑠𝑠𝑠𝑠1 2� 𝜎𝜎𝜎𝜎 1
Post-stack 𝜌𝜌𝜌𝜌
𝑠𝑠𝑠𝑠+1 ln(𝑍𝑍𝑍𝑍 − 𝑍𝑍𝑍𝑍𝑠𝑠𝑠𝑠 𝑠𝑠𝑠𝑠=1
inversion ) = 𝑘𝑘𝑘𝑘ln(𝑍𝑍𝑍𝑍𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃 )) + 2 𝑠𝑠𝑠𝑠=1 𝜎𝜎𝜎𝜎 +𝑉𝑉𝑉𝑉𝑆𝑆𝑆𝑆 𝑘𝑘𝑘𝑘
𝑘𝑘𝑘𝑘=𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶�𝜌𝜌𝜌𝜌+
𝜇𝜇𝜇𝜇
+ ∆𝐿𝐿𝐿𝐿 ∆𝐿𝐿𝐿𝐿𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆 (3) Lambda (l) and Mu (m): 𝑒𝑒𝑒𝑒(𝑟𝑟𝑟𝑟) = 𝛼𝛼𝛼𝛼 ��𝑟𝑟𝑟𝑟
𝑗𝑗𝑗𝑗 =1
𝑀𝑀𝑀𝑀 � + ∥ (𝑆𝑆𝑆𝑆 − 𝑊𝑊𝑊𝑊 ∗ 𝑟𝑟𝑟𝑟) ∥ 2
(12)
𝑟𝑟𝑟𝑟𝑠𝑠𝑠𝑠 = ln(𝑍𝑍𝑍𝑍
ln(𝑍𝑍𝑍𝑍 𝑗𝑗𝑗𝑗 =1 𝑆𝑆𝑆𝑆 ) = 𝑘𝑘𝑘𝑘ln(𝑍𝑍𝑍𝑍
𝑆𝑆𝑆𝑆
∗ 𝐷𝐷𝐷𝐷𝐿𝐿𝐿𝐿𝑆𝑆𝑆𝑆(10) + 𝐶𝐶𝐶𝐶3 𝑊𝑊𝑊𝑊(𝜃𝜃𝜃𝜃) ∗ 𝐷𝐷𝐷𝐷𝐿𝐿𝐿𝐿(3)
(14) 𝑗𝑗𝑗𝑗𝑠𝑠𝑠𝑠=1
2 𝜎𝜎𝜎𝜎
1 1
𝑍𝑍𝑍𝑍𝑠𝑠𝑠𝑠+1 𝑇𝑇𝑇𝑇(𝜃𝜃𝜃𝜃) +𝑀𝑀𝑀𝑀 𝑍𝑍𝑍𝑍𝑠𝑠𝑠𝑠 𝑆𝑆𝑆𝑆 )==𝐶𝐶𝐶𝐶𝑀𝑀𝑀𝑀𝑘𝑘𝑘𝑘ln(𝑍𝑍𝑍𝑍 1 𝑊𝑊𝑊𝑊(𝜃𝜃𝜃𝜃) 𝑃𝑃𝑃𝑃 )∗+ 𝐷𝐷𝐷𝐷𝐿𝐿𝐿𝐿 𝑘𝑘𝑘𝑘𝐶𝐶𝐶𝐶𝑃𝑃𝑃𝑃 + + ∆𝐿𝐿𝐿𝐿 𝐶𝐶𝐶𝐶2 𝑊𝑊𝑊𝑊(𝜃𝜃𝜃𝜃) 𝑆𝑆𝑆𝑆 (3)
𝐷𝐷𝐷𝐷 (6)
𝑒𝑒𝑒𝑒(𝑟𝑟𝑟𝑟) = 𝛼𝛼𝛼𝛼 ��𝑟𝑟𝑟𝑟 𝑗𝑗𝑗𝑗 =1 �+ ∥ (𝑆𝑆𝑆𝑆 − 𝑊𝑊𝑊𝑊 ∗ 𝑟𝑟𝑟𝑟) ∥ 2
(12)
)2ln⁡ =(𝜌𝜌𝜌𝜌) 𝜇𝜇𝜇𝜇𝜌𝜌𝜌𝜌 = 1 1 𝑃𝑃𝑃𝑃 )1+
𝑚𝑚𝑚𝑚ln(𝑍𝑍𝑍𝑍 𝑍𝑍𝑍𝑍 2 𝑚𝑚𝑚𝑚 =1(𝜌𝜌𝜌𝜌𝑉𝑉𝑉𝑉+ 𝑆𝑆𝑆𝑆 ) ∆𝐿𝐿𝐿𝐿 (4) 𝜆𝜆𝜆𝜆 +𝑀𝑀𝑀𝑀2𝜇𝜇𝜇𝜇 𝑗𝑗𝑗𝑗
2
𝑍𝑍𝑍𝑍 𝑆𝑆𝑆𝑆 = (𝜌𝜌𝜌𝜌𝑉𝑉𝑉𝑉 ln⁡ (𝜌𝜌𝜌𝜌) = 𝑚𝑚𝑚𝑚ln(𝑍𝑍𝑍𝑍 𝑃𝑃𝑃𝑃 ) −+ 𝑚𝑚𝑚𝑚 + 𝐷𝐷𝐷𝐷 (15) 2
−∆𝐿𝐿𝐿𝐿 (4) 2 𝜎𝜎𝜎𝜎
2
= 2𝜇𝜇𝜇𝜇𝜌𝜌𝜌𝜌 (15)
𝑒𝑒𝑒𝑒(𝑟𝑟𝑟𝑟) = 𝑆𝑆𝑆𝑆𝛼𝛼𝛼𝛼𝑒𝑒𝑒𝑒(𝑟𝑟𝑟𝑟) ∥ 𝑗𝑗𝑗𝑗 �𝐿𝐿𝐿𝐿(𝑆𝑆𝑆𝑆 𝑚𝑚𝑚𝑚𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶∗(𝑆𝑆𝑆𝑆 1 1
𝜆𝜆𝜆𝜆 ��𝑟𝑟𝑟𝑟 ln⁡
+𝑃𝑃𝑃𝑃 (𝜌𝜌𝜌𝜌)
𝐿𝐿𝐿𝐿 =𝑗𝑗𝑗𝑗 𝑁𝑁𝑁𝑁�ln(𝑍𝑍𝑍𝑍 =+��𝑟𝑟𝑟𝑟
𝛼𝛼𝛼𝛼 𝑃𝑃𝑃𝑃 ) ,𝜎𝜎𝜎𝜎
𝑚𝑚𝑚𝑚ln(𝑍𝑍𝑍𝑍 + 𝑃𝑃𝑃𝑃𝑆𝑆𝑆𝑆)of = +∥𝑆𝑆𝑆𝑆𝑊𝑊𝑊𝑊 ln(𝑍𝑍𝑍𝑍 +𝑟𝑟𝑟𝑟)
𝑆𝑆𝑆𝑆 ) ,∥2and
∆𝐿𝐿𝐿𝐿 𝑊𝑊𝑊𝑊 𝐷𝐷𝐷𝐷 ∗ 𝐿𝐿𝐿𝐿 𝑟𝑟𝑟𝑟) ∥= ln(𝜌𝜌𝜌𝜌). (12) (12) (4)
𝐷𝐷𝐷𝐷
There 2 are 2many 2types
2𝜇𝜇𝜇𝜇 2 𝜆𝜆𝜆𝜆𝜌𝜌𝜌𝜌 post = 𝜎𝜎𝜎𝜎 𝑍𝑍𝑍𝑍𝑃𝑃𝑃𝑃2 − 2𝑍𝑍𝑍𝑍 stack 𝐷𝐷𝐷𝐷seismic inversion (16)
𝑉𝑉𝑉𝑉𝑃𝑃𝑃𝑃 = =
𝑒𝑒𝑒𝑒(𝑟𝑟𝑟𝑟) � 𝛼𝛼𝛼𝛼 ��𝑟𝑟𝑟𝑟 𝑗𝑗𝑗𝑗 =1
𝜆𝜆𝜆𝜆 +𝜌𝜌𝜌𝜌2𝜇𝜇𝜇𝜇 𝑗𝑗𝑗𝑗 � + ∥ (𝑆𝑆𝑆𝑆 − 𝑊𝑊𝑊𝑊 ∗ 𝑟𝑟𝑟𝑟) ∥ 2 (13)(13) (12)
𝑉𝑉𝑉𝑉𝑁𝑁𝑁𝑁𝑃𝑃𝑃𝑃 ==
𝜆𝜆𝜆𝜆𝜌𝜌𝜌𝜌 =𝑍𝑍𝑍𝑍𝑃𝑃𝑃𝑃� 𝑍𝑍𝑍𝑍− 2𝑍𝑍𝑍𝑍 � ∆𝑉𝑉𝑉𝑉 ∆𝑉𝑉𝑉𝑉𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃 ∆𝑉𝑉𝑉𝑉𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆
∆𝑉𝑉𝑉𝑉 ∆𝜌𝜌𝜌𝜌
∆𝜌𝜌𝜌𝜌 (16) (13) 𝑉𝑉𝑉𝑉𝑃𝑃𝑃𝑃 = � 𝑗𝑗𝑗𝑗 =1 2 𝜎𝜎𝜎𝜎
𝑆𝑆𝑆𝑆
𝑍𝑍𝑍𝑍 𝑗𝑗𝑗𝑗 =1
1 𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑝𝑝𝑝𝑝 𝜌𝜌𝜌𝜌�� 1 𝑟𝑟𝑟𝑟𝑗𝑗𝑗𝑗𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑠𝑠𝑠𝑠=1 (11) based (13)
𝑆𝑆𝑆𝑆
algorithm 𝑅𝑅𝑅𝑅(𝜃𝜃𝜃𝜃)
𝑅𝑅𝑅𝑅(𝜃𝜃𝜃𝜃) available =
= 1 ∆𝑉𝑉𝑉𝑉 but 1𝑃𝑃𝑃𝑃 + +in𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑍𝑍𝑍𝑍22𝑆𝑆𝑆𝑆the ∆𝑉𝑉𝑉𝑉
= 0.741𝑍𝑍𝑍𝑍 + 𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑃𝑃𝑃𝑃 33− ∆𝜌𝜌𝜌𝜌
𝑆𝑆𝑆𝑆 present
+ 1900 study, 1 Model (5)
(5) (13)
𝑍𝑍𝑍𝑍𝑆𝑆𝑆𝑆 = 0.741𝑍𝑍𝑍𝑍𝐶𝐶𝐶𝐶 𝑅𝑅𝑅𝑅(𝜃𝜃𝜃𝜃) 1 − = 1900
𝑠𝑠𝑠𝑠=1 = 𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐
1
1
1 + 𝑉𝑉𝑉𝑉
𝑉𝑉𝑉𝑉 𝑃𝑃𝑃𝑃 +
𝐾𝐾𝐾𝐾𝑐𝑐𝑐𝑐 2 𝑐𝑐𝑐𝑐
+2
𝜌𝜌𝜌𝜌 = 𝑉𝑉𝑉𝑉
𝑚𝑚𝑚𝑚𝑐𝑐𝑐𝑐
𝑉𝑉𝑉𝑉 𝑆𝑆𝑆𝑆
0.000113𝑍𝑍𝑍𝑍 3+ , 𝑐𝑐𝑐𝑐 3 𝜌𝜌𝜌𝜌
𝐶𝐶𝐶𝐶
𝜌𝜌𝜌𝜌
+ 2
1.57 = 𝑐𝑐𝑐𝑐 2 ,
(13) 𝑊𝑊𝑊𝑊(𝜃𝜃𝜃𝜃) (14) (5) 𝜆𝜆𝜆𝜆 +𝜌𝜌𝜌𝜌2𝜇𝜇𝜇𝜇
inversion 𝜆𝜆𝜆𝜆 + (MBI)
𝑃𝑃𝑃𝑃
2𝜇𝜇𝜇𝜇 𝜆𝜆𝜆𝜆 +algorithm 2 2𝜇𝜇𝜇𝜇 𝑃𝑃𝑃𝑃 𝑉𝑉𝑉𝑉 2 𝑃𝑃𝑃𝑃 𝑉𝑉𝑉𝑉 𝑆𝑆𝑆𝑆
is𝑆𝑆𝑆𝑆0.296𝑍𝑍𝑍𝑍 used
𝑃𝑃𝑃𝑃
𝜌𝜌𝜌𝜌
for the analysis. 2 MBI (15) and 𝑉𝑉𝑉𝑉𝑃𝑃𝑃𝑃 = �𝜇𝜇𝜇𝜇 (13)
𝜌𝜌𝜌𝜌
𝑉𝑉𝑉𝑉𝑃𝑃𝑃𝑃==0.000113𝑍𝑍𝑍𝑍
𝑉𝑉𝑉𝑉𝑃𝑃𝑃𝑃 = 𝑃𝑃𝑃𝑃 + 889 (14) 𝜆𝜆𝜆𝜆 +𝜌𝜌𝜌𝜌2𝜇𝜇𝜇𝜇
� 𝑉𝑉𝑉𝑉𝑃𝑃𝑃𝑃 𝑀𝑀𝑀𝑀𝑇𝑇𝑇𝑇(𝜃𝜃𝜃𝜃 = 𝑃𝑃𝑃𝑃�+) 1.57 𝐶𝐶𝐶𝐶11 (𝜃𝜃𝜃𝜃 11 )𝑊𝑊𝑊𝑊(𝜃𝜃𝜃𝜃
1 (13) 2
2 (13)
𝑉𝑉𝑉𝑉3𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆)𝑊𝑊𝑊𝑊(𝜃𝜃𝜃𝜃
𝑉𝑉𝑉𝑉 2 𝑉𝑉𝑉𝑉𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆 222
𝑉𝑉𝑉𝑉 𝑉𝑉𝑉𝑉𝑆𝑆𝑆𝑆 = � (14)
is
𝑉𝑉𝑉𝑉𝑃𝑃𝑃𝑃 based = 0.296𝑍𝑍𝑍𝑍 𝜇𝜇𝜇𝜇 on 𝜌𝜌𝜌𝜌
where + the 8891 𝑐𝑐𝑐𝑐convolution 𝜌𝜌𝜌𝜌1 = 1
𝑉𝑉𝑉𝑉𝑆𝑆𝑆𝑆 = 0.26𝑍𝑍𝑍𝑍
, theory, 1 )𝐷𝐷𝐷𝐷 𝑃𝑃𝑃𝑃 − 428
𝑐𝑐𝑐𝑐 = ⋯ which
0.5 𝐶𝐶𝐶𝐶
− 3 (𝜃𝜃𝜃𝜃
� states
𝑉𝑉𝑉𝑉 (15) � 𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠 that
𝑠𝑠𝑠𝑠
3 2)𝐷𝐷𝐷𝐷 the (16)
2 𝜃𝜃𝜃𝜃 and 𝑐𝑐𝑐𝑐 = 4 � 𝑉𝑉𝑉𝑉 � 𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠 𝑠𝑠𝑠𝑠 2
2 𝜃𝜃𝜃𝜃 𝑉𝑉𝑉𝑉 = � 𝜇𝜇𝜇𝜇
𝜌𝜌𝜌𝜌 (13)
𝑒𝑒𝑒𝑒(𝑟𝑟𝑟𝑟) =�𝛼𝛼𝛼𝛼 ��𝑟𝑟𝑟𝑟 where 𝑃𝑃𝑃𝑃
⋮𝑗𝑗𝑗𝑗 �can + 𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐11 = ∥� 2𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑠𝑠𝑠𝑠 (𝑆𝑆𝑆𝑆 − 𝜃𝜃𝜃𝜃𝑊𝑊𝑊𝑊,, ∗ 𝑟𝑟𝑟𝑟)𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐from ∥2222= =⋱0.5 −
− ��convolution
𝑆𝑆𝑆𝑆 � 𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠(12)
�⋮ 𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠2 𝜃𝜃𝜃𝜃 𝜃𝜃𝜃𝜃 �and 𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐33 = 4 �𝑉𝑉𝑉𝑉𝑆𝑆𝑆𝑆 �� 𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠 𝑠𝑠𝑠𝑠 𝜃𝜃𝜃𝜃 𝑃𝑃𝑃𝑃
𝑉𝑉𝑉𝑉𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆 ==0.26𝑍𝑍𝑍𝑍
seismic
𝑉𝑉𝑉𝑉 trace where
� − 428 �1= 2 =be 2𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑠𝑠𝑠𝑠
𝜎𝜎𝜎𝜎 generated
2
2 𝜃𝜃𝜃𝜃⋮
2 𝜃𝜃𝜃𝜃 0.5 the 𝑉𝑉𝑉𝑉(14)
𝑉𝑉𝑉𝑉 𝑃𝑃𝑃𝑃
(16) and of 3 = 4 �𝑉𝑉𝑉𝑉 𝑃𝑃𝑃𝑃 (7)𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠2 𝜃𝜃𝜃𝜃 𝑉𝑉𝑉𝑉𝑆𝑆𝑆𝑆 = � 𝜌𝜌𝜌𝜌 (14) (14)
𝜌𝜌𝜌𝜌 𝑗𝑗𝑗𝑗 =1
𝑃𝑃𝑃𝑃 2𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑠𝑠𝑠𝑠 𝑉𝑉𝑉𝑉 𝑃𝑃𝑃𝑃 𝑉𝑉𝑉𝑉𝑃𝑃𝑃𝑃 𝜌𝜌𝜌𝜌
𝜇𝜇𝜇𝜇
𝑇𝑇𝑇𝑇(𝜃𝜃𝜃𝜃𝑁𝑁𝑁𝑁 ) 𝐶𝐶𝐶𝐶1 (𝜃𝜃𝜃𝜃𝑁𝑁𝑁𝑁 )𝑊𝑊𝑊𝑊(𝜃𝜃𝜃𝜃𝑁𝑁𝑁𝑁 )𝐷𝐷𝐷𝐷 ⋯ 𝐶𝐶𝐶𝐶3 (𝜃𝜃𝜃𝜃𝑁𝑁𝑁𝑁𝑃𝑃𝑃𝑃)𝑊𝑊𝑊𝑊(𝜃𝜃𝜃𝜃𝑁𝑁𝑁𝑁 )𝐷𝐷𝐷𝐷 𝑃𝑃𝑃𝑃
𝑉𝑉𝑉𝑉𝑆𝑆𝑆𝑆2 = =� (𝜌𝜌𝜌𝜌𝑉𝑉𝑉𝑉𝑆𝑆𝑆𝑆 )2 = 𝜇𝜇𝜇𝜇𝜌𝜌𝜌𝜌 (14)
wavelet 𝜇𝜇𝜇𝜇 with 𝑇𝑇𝑇𝑇(𝜃𝜃𝜃𝜃)the = 𝐶𝐶𝐶𝐶
𝜇𝜇𝜇𝜇= 𝐶𝐶𝐶𝐶earth`s 𝑊𝑊𝑊𝑊(𝜃𝜃𝜃𝜃) ∗∗∗reflectivity
𝐷𝐷𝐷𝐷𝐿𝐿𝐿𝐿𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃 + + 𝐶𝐶𝐶𝐶 𝐶𝐶𝐶𝐶22 𝑊𝑊𝑊𝑊(𝜃𝜃𝜃𝜃)𝑊𝑊𝑊𝑊(𝜃𝜃𝜃𝜃) and(14) ∗∗∗ addition
𝐷𝐷𝐷𝐷𝐿𝐿𝐿𝐿𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆 + + 𝐶𝐶𝐶𝐶 𝐶𝐶𝐶𝐶33 𝑊𝑊𝑊𝑊(𝜃𝜃𝜃𝜃)
𝑊𝑊𝑊𝑊(𝜃𝜃𝜃𝜃)
of noise ∗∗∗ 𝐷𝐷𝐷𝐷𝐿𝐿𝐿𝐿
𝐷𝐷𝐷𝐷𝐿𝐿𝐿𝐿𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷 (6) 𝑍𝑍𝑍𝑍 (15)
𝑉𝑉𝑉𝑉𝑆𝑆𝑆𝑆 = � 𝑉𝑉𝑉𝑉𝑆𝑆𝑆𝑆𝑇𝑇𝑇𝑇(𝜃𝜃𝜃𝜃) 𝑇𝑇𝑇𝑇(𝜃𝜃𝜃𝜃)= � = 𝐶𝐶𝐶𝐶 1 𝑊𝑊𝑊𝑊(𝜃𝜃𝜃𝜃)
1
𝑊𝑊𝑊𝑊(𝜃𝜃𝜃𝜃) 𝐷𝐷𝐷𝐷𝐿𝐿𝐿𝐿
𝐷𝐷𝐷𝐷𝐿𝐿𝐿𝐿 + 𝐶𝐶𝐶𝐶 𝑊𝑊𝑊𝑊(𝜃𝜃𝜃𝜃) 𝐷𝐷𝐷𝐷𝐿𝐿𝐿𝐿
𝐷𝐷𝐷𝐷𝐿𝐿𝐿𝐿 + 𝐶𝐶𝐶𝐶
(14) 𝑊𝑊𝑊𝑊(𝜃𝜃𝜃𝜃) 𝐷𝐷𝐷𝐷𝐿𝐿𝐿𝐿 (6)
(6) 𝜌𝜌𝜌𝜌 2
𝜇𝜇𝜇𝜇
(Mallick, 2 𝜌𝜌𝜌𝜌
𝜆𝜆𝜆𝜆 +𝐿𝐿𝐿𝐿𝑆𝑆𝑆𝑆1995), 2𝜇𝜇𝜇𝜇 =2 [𝐿𝐿𝐿𝐿 Δ𝐿𝐿𝐿𝐿
𝜌𝜌𝜌𝜌ln(𝑍𝑍𝑍𝑍
1Δ𝐿𝐿𝐿𝐿 ] = [𝑙𝑙𝑙𝑙𝑠𝑠𝑠𝑠 𝑃𝑃𝑃𝑃(𝑍𝑍𝑍𝑍 20 0)], 𝑆𝑆𝑆𝑆 3 𝐷𝐷𝐷𝐷
Therefore, 𝑍𝑍𝑍𝑍
𝑉𝑉𝑉𝑉𝑆𝑆𝑆𝑆 =
2
=
=� (𝜌𝜌𝜌𝜌𝑉𝑉𝑉𝑉 ) = 𝜇𝜇𝜇𝜇𝜌𝜌𝜌𝜌 (14) (15)
𝑍𝑍𝑍𝑍𝑆𝑆𝑆𝑆 ==�(𝜌𝜌𝜌𝜌𝑉𝑉𝑉𝑉
𝑉𝑉𝑉𝑉 𝐿𝐿𝐿𝐿 𝑃𝑃𝑃𝑃) = =
𝑃𝑃𝑃𝑃 𝜇𝜇𝜇𝜇𝜌𝜌𝜌𝜌𝑆𝑆𝑆𝑆 ) 𝐷𝐷𝐷𝐷
ln(𝑍𝑍𝑍𝑍 𝑃𝑃𝑃𝑃 ) ,, 𝐿𝐿𝐿𝐿 𝐿𝐿𝐿𝐿 𝑆𝑆𝑆𝑆 = = ln(𝑍𝑍𝑍𝑍
ln(𝑍𝑍𝑍𝑍
𝑃𝑃𝑃𝑃0
𝑆𝑆𝑆𝑆 ) )) ,,, and and 𝐿𝐿𝐿𝐿
𝐿𝐿𝐿𝐿 𝐷𝐷𝐷𝐷 = = (13) ln(𝜌𝜌𝜌𝜌).
ln(𝜌𝜌𝜌𝜌). (15) 𝜆𝜆𝜆𝜆𝜌𝜌𝜌𝜌 2
𝑍𝑍𝑍𝑍𝑃𝑃𝑃𝑃𝜌𝜌𝜌𝜌 − 𝑆𝑆𝑆𝑆 2𝑍𝑍𝑍𝑍 2
𝑆𝑆𝑆𝑆 (16)
𝑃𝑃𝑃𝑃 𝐿𝐿𝐿𝐿
𝜌𝜌𝜌𝜌
𝑃𝑃𝑃𝑃
𝑃𝑃𝑃𝑃 = ln(𝑍𝑍𝑍𝑍 𝑃𝑃𝑃𝑃
𝑃𝑃𝑃𝑃 ) , 𝐿𝐿𝐿𝐿 𝑆𝑆𝑆𝑆
𝑆𝑆𝑆𝑆 = ln(𝑍𝑍𝑍𝑍 𝑆𝑆𝑆𝑆
𝑆𝑆𝑆𝑆 and 𝐿𝐿𝐿𝐿 𝐷𝐷𝐷𝐷
𝐷𝐷𝐷𝐷 = ln(𝜌𝜌𝜌𝜌). 𝑍𝑍𝑍𝑍 = 2 (𝜌𝜌𝜌𝜌𝑉𝑉𝑉𝑉
2 ) 2 2
= 𝜇𝜇𝜇𝜇𝜌𝜌𝜌𝜌 (15)(15)
𝑆𝑆𝑆𝑆(𝑡𝑡𝑡𝑡) =21 𝑊𝑊𝑊𝑊(𝑡𝑡𝑡𝑡) ∗ 𝑟𝑟𝑟𝑟(𝑡𝑡𝑡𝑡) + 𝑠𝑠𝑠𝑠(𝑡𝑡𝑡𝑡)
1𝜇𝜇𝜇𝜇𝜌𝜌𝜌𝜌 1(15) (16) (8) 𝜆𝜆𝜆𝜆𝜌𝜌𝜌𝜌
𝑍𝑍𝑍𝑍𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆 = = 0.741𝑍𝑍𝑍𝑍 𝑍𝑍𝑍𝑍𝑃𝑃𝑃𝑃 − 𝑆𝑆𝑆𝑆 2𝑍𝑍𝑍𝑍 (16) (13)
𝑍𝑍𝑍𝑍 𝑍𝑍𝑍𝑍𝑃𝑃𝑃𝑃2𝑍𝑍𝑍𝑍𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆2− (8) 𝑃𝑃𝑃𝑃 − 1900
2 𝑆𝑆𝑆𝑆
𝜆𝜆𝜆𝜆𝜌𝜌𝜌𝜌 𝑆𝑆𝑆𝑆 = =(𝜌𝜌𝜌𝜌𝑉𝑉𝑉𝑉 )2=2𝑍𝑍𝑍𝑍 = (𝜌𝜌𝜌𝜌𝑉𝑉𝑉𝑉 1
𝜇𝜇𝜇𝜇𝜌𝜌𝜌𝜌
1 ) 2 1
= 1 1
1 (15)
𝐶𝐶𝐶𝐶11 =
𝐶𝐶𝐶𝐶 = 𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐11 + 𝑆𝑆𝑆𝑆 𝑆𝑆𝑆𝑆 + 𝐾𝐾𝐾𝐾𝑐𝑐𝑐𝑐 𝐾𝐾𝐾𝐾𝑐𝑐𝑐𝑐22 + + 𝑚𝑚𝑚𝑚𝑐𝑐𝑐𝑐 𝑚𝑚𝑚𝑚𝑐𝑐𝑐𝑐33 ,, 𝐶𝐶𝐶𝐶22 =
𝐶𝐶𝐶𝐶 = 𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐22 ,, 𝑊𝑊𝑊𝑊(𝜃𝜃𝜃𝜃)
𝑊𝑊𝑊𝑊(𝜃𝜃𝜃𝜃) 𝑍𝑍𝑍𝑍𝑆𝑆𝑆𝑆==
𝜆𝜆𝜆𝜆𝜌𝜌𝜌𝜌
2
=0.000113𝑍𝑍𝑍𝑍 (𝜌𝜌𝜌𝜌𝑉𝑉𝑉𝑉
𝑍𝑍𝑍𝑍𝑃𝑃𝑃𝑃2 − 𝑆𝑆𝑆𝑆 )2𝑍𝑍𝑍𝑍
2
=𝑆𝑆𝑆𝑆2−𝜇𝜇𝜇𝜇𝜌𝜌𝜌𝜌 (15)
(16)
Where 2S(t), 𝑍𝑍𝑍𝑍1==
𝐶𝐶𝐶𝐶 W(t),2𝑉𝑉𝑉𝑉𝜌𝜌𝜌𝜌 2
2 𝑐𝑐𝑐𝑐 1 r(t)+ 2
2 and 𝐾𝐾𝐾𝐾𝑐𝑐𝑐𝑐 2 +n(t) 𝑚𝑚𝑚𝑚𝑐𝑐𝑐𝑐 are3 , synthetic 𝐶𝐶𝐶𝐶 2 = 2
2 𝑐𝑐𝑐𝑐 trace,
2 , 𝑊𝑊𝑊𝑊(𝜃𝜃𝜃𝜃)
wavelet, (9) and 𝜌𝜌𝜌𝜌
𝑍𝑍𝑍𝑍 𝑆𝑆𝑆𝑆 0.741𝑍𝑍𝑍𝑍 𝑃𝑃𝑃𝑃 𝑃𝑃𝑃𝑃 +
1900 1.57 (14)
(13)
𝜆𝜆𝜆𝜆𝜌𝜌𝜌𝜌
𝑍𝑍𝑍𝑍𝑆𝑆𝑆𝑆 = =
= 𝑍𝑍𝑍𝑍0.741𝑍𝑍𝑍𝑍 𝜇𝜇𝜇𝜇𝜆𝜆𝜆𝜆𝜌𝜌𝜌𝜌
𝑃𝑃𝑃𝑃 − 2𝑍𝑍𝑍𝑍 = 𝑆𝑆𝑆𝑆𝑍𝑍𝑍𝑍 𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃)−
22
−1900 2𝑍𝑍𝑍𝑍𝑆𝑆𝑆𝑆22 2(16) (16) (13)
𝑉𝑉𝑉𝑉
earth �reflectivity 𝑇𝑇𝑇𝑇(𝜃𝜃𝜃𝜃
𝑇𝑇𝑇𝑇(𝜃𝜃𝜃𝜃1 ))𝑠𝑠𝑠𝑠+1 𝐶𝐶𝐶𝐶 𝑍𝑍𝑍𝑍 and − 𝐶𝐶𝐶𝐶
𝑍𝑍𝑍𝑍 noise
𝐶𝐶𝐶𝐶1𝑠𝑠𝑠𝑠 (𝜃𝜃𝜃𝜃(𝜃𝜃𝜃𝜃
(𝜃𝜃𝜃𝜃1 )𝑊𝑊𝑊𝑊(𝜃𝜃𝜃𝜃 )𝑊𝑊𝑊𝑊(𝜃𝜃𝜃𝜃 component
)𝑊𝑊𝑊𝑊(𝜃𝜃𝜃𝜃1 )𝐷𝐷𝐷𝐷 )𝐷𝐷𝐷𝐷
)𝐷𝐷𝐷𝐷 ⋯ ⋯
⋯ 𝐶𝐶𝐶𝐶 (14) 𝐶𝐶𝐶𝐶
respectively.
𝐶𝐶𝐶𝐶3 (𝜃𝜃𝜃𝜃(𝜃𝜃𝜃𝜃 )𝑊𝑊𝑊𝑊(𝜃𝜃𝜃𝜃
(𝜃𝜃𝜃𝜃3 )𝑊𝑊𝑊𝑊(𝜃𝜃𝜃𝜃3 )𝐷𝐷𝐷𝐷 )𝐷𝐷𝐷𝐷
)𝐷𝐷𝐷𝐷 If the 𝑉𝑉𝑉𝑉
𝜌𝜌𝜌𝜌
𝜆𝜆𝜆𝜆𝜌𝜌𝜌𝜌
𝑍𝑍𝑍𝑍 𝑃𝑃𝑃𝑃 = =
= 0.296𝑍𝑍𝑍𝑍
0.000113𝑍𝑍𝑍𝑍
𝑍𝑍𝑍𝑍
0.741𝑍𝑍𝑍𝑍
2
− 2𝑍𝑍𝑍𝑍𝑃𝑃𝑃𝑃
2+ 889
− 𝑃𝑃𝑃𝑃 +
1900 1.57 (16) (16) (15)
(14)
(13)
𝑆𝑆𝑆𝑆
𝜌𝜌𝜌𝜌 𝑟𝑟𝑟𝑟𝑠𝑠𝑠𝑠𝑇𝑇𝑇𝑇(𝜃𝜃𝜃𝜃 1 1 1 1 3
3 3 )𝑊𝑊𝑊𝑊(𝜃𝜃𝜃𝜃
3 3 𝑆𝑆𝑆𝑆 𝑃𝑃𝑃𝑃 𝑃𝑃𝑃𝑃 𝑆𝑆𝑆𝑆
𝜌𝜌𝜌𝜌𝑆𝑆𝑆𝑆 =
𝑍𝑍𝑍𝑍
noise
= 0.000113𝑍𝑍𝑍𝑍
0.741𝑍𝑍𝑍𝑍
in 𝑍𝑍𝑍𝑍the � ==
𝑆𝑆𝑆𝑆� ⋮
𝑃𝑃𝑃𝑃data − 1 1900
⋮0.741𝑍𝑍𝑍𝑍 𝑍𝑍𝑍𝑍 �𝑃𝑃𝑃𝑃is
� =+
= + �1.57
uncorrelated
𝑃𝑃𝑃𝑃� −1 1900
𝑍𝑍𝑍𝑍
1 ⋮
⋮
1
with ⋱
⋱the seismic ⋮ signal,
⋮
3(13) � the(14)
(10)
� (13) (7)
(7) 𝑉𝑉𝑉𝑉
𝜌𝜌𝜌𝜌 = == 0.26𝑍𝑍𝑍𝑍
0.296𝑍𝑍𝑍𝑍
0.000113𝑍𝑍𝑍𝑍 − + 428 889
+ 1.57 (16)
(15)
(14)
𝑃𝑃𝑃𝑃 − 1900 the above equations express (13)
𝑆𝑆𝑆𝑆 𝑃𝑃𝑃𝑃
𝜌𝜌𝜌𝜌 �𝑇𝑇𝑇𝑇(𝜃𝜃𝜃𝜃⋮ )𝑠𝑠𝑠𝑠+1 � =1.57 �𝐶𝐶𝐶𝐶 𝑠𝑠𝑠𝑠 (𝜃𝜃𝜃𝜃 )𝑊𝑊𝑊𝑊(𝜃𝜃𝜃𝜃 ⋮ ⋱ ⋮ (14) � Through 𝑍𝑍𝑍𝑍seismic
(7) 𝑆𝑆𝑆𝑆 = 0.741𝑍𝑍𝑍𝑍
𝑃𝑃𝑃𝑃
inversion, 𝑃𝑃𝑃𝑃 𝑃𝑃𝑃𝑃
𝑉𝑉𝑉𝑉𝑃𝑃𝑃𝑃2==
trace
𝑍𝑍𝑍𝑍 =0.000113𝑍𝑍𝑍𝑍
can 0.296𝑍𝑍𝑍𝑍
(𝜌𝜌𝜌𝜌𝑉𝑉𝑉𝑉 𝜌𝜌𝜌𝜌 )= be 2 0.000113𝑍𝑍𝑍𝑍
𝑇𝑇𝑇𝑇(𝜃𝜃𝜃𝜃
𝑇𝑇𝑇𝑇(𝜃𝜃𝜃𝜃 = 𝑃𝑃𝑃𝑃𝑁𝑁𝑁𝑁
solved 𝑃𝑃𝑃𝑃)
𝑁𝑁𝑁𝑁𝜇𝜇𝜇𝜇𝜌𝜌𝜌𝜌++889
) for 𝐶𝐶𝐶𝐶11𝑃𝑃𝑃𝑃(𝜃𝜃𝜃𝜃
𝐶𝐶𝐶𝐶 (𝜃𝜃𝜃𝜃 +𝑁𝑁𝑁𝑁
the 𝑁𝑁𝑁𝑁1.57 )𝑊𝑊𝑊𝑊(𝜃𝜃𝜃𝜃
)𝑊𝑊𝑊𝑊(𝜃𝜃𝜃𝜃earth 𝑁𝑁𝑁𝑁 )𝐷𝐷𝐷𝐷
)𝐷𝐷𝐷𝐷
𝑁𝑁𝑁𝑁 )𝐷𝐷𝐷𝐷 reflectivity
⋯
⋯ 𝐶𝐶𝐶𝐶
⋯ 𝐶𝐶𝐶𝐶33 (15)
𝐶𝐶𝐶𝐶
(𝜃𝜃𝜃𝜃
(𝜃𝜃𝜃𝜃function.
(𝜃𝜃𝜃𝜃 𝑁𝑁𝑁𝑁 )𝑊𝑊𝑊𝑊(𝜃𝜃𝜃𝜃
)𝑊𝑊𝑊𝑊(𝜃𝜃𝜃𝜃𝑁𝑁𝑁𝑁
𝑁𝑁𝑁𝑁 )𝑊𝑊𝑊𝑊(𝜃𝜃𝜃𝜃 )𝐷𝐷𝐷𝐷This(15)
)𝐷𝐷𝐷𝐷
𝑁𝑁𝑁𝑁 )𝐷𝐷𝐷𝐷
(14) 𝑉𝑉𝑉𝑉 == 0.26𝑍𝑍𝑍𝑍
0.296𝑍𝑍𝑍𝑍 − + 428 889 (16)
(15)
(14)
𝑆𝑆𝑆𝑆 𝑆𝑆𝑆𝑆 𝑁𝑁𝑁𝑁 1 𝑁𝑁𝑁𝑁 𝑁𝑁𝑁𝑁 𝑁𝑁𝑁𝑁 3 𝑁𝑁𝑁𝑁 𝑁𝑁𝑁𝑁 P-impedance 𝜌𝜌𝜌𝜌 = 0.000113𝑍𝑍𝑍𝑍
𝑆𝑆𝑆𝑆
𝑃𝑃𝑃𝑃 and S-impedance 𝑃𝑃𝑃𝑃 𝑃𝑃𝑃𝑃 𝑃𝑃𝑃𝑃 + 1.57 in rock properties. It is
𝑉𝑉𝑉𝑉 =
𝑉𝑉𝑉𝑉𝑆𝑆𝑆𝑆a=non-linear 0.296𝑍𝑍𝑍𝑍
0.26𝑍𝑍𝑍𝑍 𝑉𝑉𝑉𝑉 = +
0.296𝑍𝑍𝑍𝑍 889 + 889 (15) (16) (15)
2−𝑃𝑃𝑃𝑃
[𝐿𝐿𝐿𝐿 428
Δ𝐿𝐿𝐿𝐿 Δ𝐿𝐿𝐿𝐿
𝑃𝑃𝑃𝑃 𝐷𝐷𝐷𝐷 ]] = = [𝑙𝑙𝑙𝑙𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠(𝑍𝑍𝑍𝑍can 0 be 0)],solved iteratively as 𝑉𝑉𝑉𝑉𝑆𝑆𝑆𝑆𝑃𝑃𝑃𝑃 = 0.26𝑍𝑍𝑍𝑍 𝑃𝑃𝑃𝑃 𝑃𝑃𝑃𝑃−+ 428 (16)
𝑃𝑃𝑃𝑃 𝑃𝑃𝑃𝑃
is 𝑃𝑃𝑃𝑃 𝑃𝑃𝑃𝑃[𝐿𝐿𝐿𝐿
𝑍𝑍𝑍𝑍𝑃𝑃𝑃𝑃2 𝑉𝑉𝑉𝑉−𝑍𝑍𝑍𝑍𝑃𝑃𝑃𝑃2𝑍𝑍𝑍𝑍 𝑁𝑁𝑁𝑁 = [𝐿𝐿𝐿𝐿 equation
𝑃𝑃𝑃𝑃𝑍𝑍𝑍𝑍Δ𝐿𝐿𝐿𝐿 𝑆𝑆𝑆𝑆 Δ𝐿𝐿𝐿𝐿��
1 𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑝𝑝𝑝𝑝
𝑆𝑆𝑆𝑆 which 𝑟𝑟𝑟𝑟𝑠𝑠𝑠𝑠 � [𝑙𝑙𝑙𝑙
[𝑙𝑙𝑙𝑙𝑠𝑠𝑠𝑠(𝑍𝑍𝑍𝑍 (𝑍𝑍𝑍𝑍𝑃𝑃𝑃𝑃0 𝑃𝑃𝑃𝑃0 0 0)], (11) claimed that 𝑉𝑉𝑉𝑉 =lr 0.296𝑍𝑍𝑍𝑍 and mr 889 help to discriminate fluid effects (15)
𝜆𝜆𝜆𝜆𝜌𝜌𝜌𝜌 = 0.26𝑍𝑍𝑍𝑍 𝑃𝑃𝑃𝑃 Δ𝐿𝐿𝐿𝐿𝑆𝑆𝑆𝑆 Δ𝐿𝐿𝐿𝐿 (16)
𝐷𝐷𝐷𝐷 ] =
𝑉𝑉𝑉𝑉𝑆𝑆𝑆𝑆 = 𝑃𝑃𝑃𝑃0 0 0)], (16)
𝑆𝑆𝑆𝑆 =−
𝑆𝑆𝑆𝑆 428
0.26𝑍𝑍𝑍𝑍 (16)
𝑃𝑃𝑃𝑃 − 428
𝐷𝐷𝐷𝐷
follows (Latimer 𝑆𝑆𝑆𝑆(𝑡𝑡𝑡𝑡) =
𝑆𝑆𝑆𝑆(𝑡𝑡𝑡𝑡) = 𝑊𝑊𝑊𝑊(𝑡𝑡𝑡𝑡) et
𝑊𝑊𝑊𝑊(𝑡𝑡𝑡𝑡) ∗∗ 𝑟𝑟𝑟𝑟(𝑡𝑡𝑡𝑡) al., 𝑠𝑠𝑠𝑠=1 2000):
𝑟𝑟𝑟𝑟(𝑡𝑡𝑡𝑡) + + 𝑠𝑠𝑠𝑠(𝑡𝑡𝑡𝑡) 𝑠𝑠𝑠𝑠(𝑡𝑡𝑡𝑡) (8)
(8) from lithology 𝑉𝑉𝑉𝑉 𝑆𝑆𝑆𝑆 = 0.26𝑍𝑍𝑍𝑍
effects − 428
𝑃𝑃𝑃𝑃 (Srivastava and Maultzsch 2018). (16)
𝑍𝑍𝑍𝑍𝑆𝑆𝑆𝑆 = 0.741𝑍𝑍𝑍𝑍 𝑆𝑆𝑆𝑆(𝑡𝑡𝑡𝑡) 𝑃𝑃𝑃𝑃 − =1900 𝑊𝑊𝑊𝑊(𝑡𝑡𝑡𝑡) 𝑀𝑀𝑀𝑀 ∗ 𝑟𝑟𝑟𝑟(𝑡𝑡𝑡𝑡) + 𝑠𝑠𝑠𝑠(𝑡𝑡𝑡𝑡)
(13) (8)
𝑍𝑍𝑍𝑍 = 𝑉𝑉𝑉𝑉𝜌𝜌𝜌𝜌 1 1 (14) (9) (9)
𝜌𝜌𝜌𝜌 = 0.000113𝑍𝑍𝑍𝑍 𝑍𝑍𝑍𝑍
𝑍𝑍𝑍𝑍 =
𝑒𝑒𝑒𝑒(𝑟𝑟𝑟𝑟) = 𝑉𝑉𝑉𝑉𝜌𝜌𝜌𝜌 𝑃𝑃𝑃𝑃 +
𝑉𝑉𝑉𝑉𝜌𝜌𝜌𝜌
= 𝛼𝛼𝛼𝛼 1.57 ��𝑟𝑟𝑟𝑟𝑗𝑗𝑗𝑗 � + ∥ (𝑆𝑆𝑆𝑆 − 𝑊𝑊𝑊𝑊 ∗ 𝑟𝑟𝑟𝑟) ∥2 (9)
(9) (12)
𝑉𝑉𝑉𝑉𝑃𝑃𝑃𝑃 = 0.296𝑍𝑍𝑍𝑍𝑃𝑃𝑃𝑃 +𝑍𝑍𝑍𝑍 𝑍𝑍𝑍𝑍𝑠𝑠𝑠𝑠+1889−=1𝑍𝑍𝑍𝑍 𝑍𝑍𝑍𝑍𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠 2 𝜎𝜎𝜎𝜎 (15) RESULTS AND DISCUSSIONS
𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠 = 𝑍𝑍𝑍𝑍 𝑠𝑠𝑠𝑠+1 𝑗𝑗𝑗𝑗− − 𝑍𝑍𝑍𝑍 (10)
𝑉𝑉𝑉𝑉𝑆𝑆𝑆𝑆 = 0.26𝑍𝑍𝑍𝑍𝑟𝑟𝑟𝑟𝑠𝑠𝑠𝑠𝑃𝑃𝑃𝑃 = = − 428 𝑠𝑠𝑠𝑠+1
𝑍𝑍𝑍𝑍𝑠𝑠𝑠𝑠+1 + 𝑍𝑍𝑍𝑍
𝑠𝑠𝑠𝑠
𝑍𝑍𝑍𝑍𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠 (16) (10)
(10)
(10)
𝑍𝑍𝑍𝑍
𝑍𝑍𝑍𝑍𝑠𝑠𝑠𝑠+1 +
𝜆𝜆𝜆𝜆 ++2𝜇𝜇𝜇𝜇
𝑠𝑠𝑠𝑠+1 𝑍𝑍𝑍𝑍𝑠𝑠𝑠𝑠
𝑁𝑁𝑁𝑁
Before moving towards seismic inversion methods, the
𝑉𝑉𝑉𝑉𝑃𝑃𝑃𝑃 = � 𝑁𝑁𝑁𝑁 (13)
𝑍𝑍𝑍𝑍𝑁𝑁𝑁𝑁 = 𝑍𝑍𝑍𝑍 𝑍𝑍𝑍𝑍11 𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑝𝑝𝑝𝑝 𝜌𝜌𝜌𝜌
𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑝𝑝𝑝𝑝 ��
𝑁𝑁𝑁𝑁
�� 𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠 � � (11) seismic reflection data needs to be pre-conditioned to
𝑍𝑍𝑍𝑍
𝑍𝑍𝑍𝑍𝑁𝑁𝑁𝑁 = (11)
𝑁𝑁𝑁𝑁 = 𝑍𝑍𝑍𝑍1 𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑝𝑝𝑝𝑝 �� 𝑟𝑟𝑟𝑟𝑠𝑠𝑠𝑠 � (11) (11) increase the signal to noise ratio. There are five major steps
𝑠𝑠𝑠𝑠=1
𝑠𝑠𝑠𝑠=1
𝑠𝑠𝑠𝑠=1
𝜇𝜇𝜇𝜇 𝑀𝑀𝑀𝑀 𝑀𝑀𝑀𝑀
that are used to improve seismic signal i.e. Bandpass filtering,
𝑉𝑉𝑉𝑉𝑆𝑆𝑆𝑆 = � 𝑀𝑀𝑀𝑀 1 1
1 1 (14)
These
𝑒𝑒𝑒𝑒(𝑟𝑟𝑟𝑟)
𝑒𝑒𝑒𝑒(𝑟𝑟𝑟𝑟) equations
=
= 𝜌𝜌𝜌𝜌 𝛼𝛼𝛼𝛼
𝛼𝛼𝛼𝛼 ��𝑟𝑟𝑟𝑟
��𝑟𝑟𝑟𝑟 are 1used
1
𝑗𝑗𝑗𝑗 � + ∥ (𝑆𝑆𝑆𝑆 − 𝑊𝑊𝑊𝑊
� + ∥ in
(𝑆𝑆𝑆𝑆 − practice
𝑊𝑊𝑊𝑊 ∗∗ 𝑟𝑟𝑟𝑟)
𝑟𝑟𝑟𝑟) ∥∥222 for recursive (12)
(12) Muting, Super gather, Parabolic radon transform and Trim
𝑒𝑒𝑒𝑒(𝑟𝑟𝑟𝑟) = 𝛼𝛼𝛼𝛼 ��𝑟𝑟𝑟𝑟𝑗𝑗𝑗𝑗 � + 2 𝑗𝑗𝑗𝑗 𝜎𝜎𝜎𝜎 (𝑆𝑆𝑆𝑆 − 𝑊𝑊𝑊𝑊 ∗ 𝑟𝑟𝑟𝑟) ∥
2 ∥ 𝜎𝜎𝜎𝜎 (12)
inversion with the 𝑗𝑗𝑗𝑗𝑗𝑗𝑗𝑗aim =1
=1 of transforming
2 𝜎𝜎𝜎𝜎 reflectivity function statics. Figure 3 displays output of pre-conditioning, where
𝑍𝑍𝑍𝑍𝑆𝑆𝑆𝑆2 = (𝜌𝜌𝜌𝜌𝑉𝑉𝑉𝑉𝑆𝑆𝑆𝑆𝑗𝑗𝑗𝑗)=1 2
= 𝜇𝜇𝜇𝜇𝜌𝜌𝜌𝜌 (15)
into acoustic impedance (Berteussen and Ursin, 1983). Z1 Figure 3a shows raw seismic gather at inline 1161, Figure 3b
𝜆𝜆𝜆𝜆𝜌𝜌𝜌𝜌 =� 𝜆𝜆𝜆𝜆
2𝜆𝜆𝜆𝜆 +
+
𝑍𝑍𝑍𝑍𝑃𝑃𝑃𝑃𝜆𝜆𝜆𝜆 − 2𝜇𝜇𝜇𝜇
2𝜇𝜇𝜇𝜇
2𝑍𝑍𝑍𝑍 2
(16)
𝑉𝑉𝑉𝑉 = + 2𝜇𝜇𝜇𝜇 𝑆𝑆𝑆𝑆 (13)
𝑉𝑉𝑉𝑉 =�
𝑉𝑉𝑉𝑉𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃 = � 𝜌𝜌𝜌𝜌 𝜌𝜌𝜌𝜌 (13)
(13)
𝑍𝑍𝑍𝑍𝑆𝑆𝑆𝑆 = 0.741𝑍𝑍𝑍𝑍 𝜌𝜌𝜌𝜌 𝑃𝑃𝑃𝑃 − 1900 (13)
588
𝜌𝜌𝜌𝜌 = 0.000113𝑍𝑍𝑍𝑍 𝑃𝑃𝑃𝑃 + 1.57 (14)
𝜇𝜇𝜇𝜇
𝜇𝜇𝜇𝜇
𝑉𝑉𝑉𝑉𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑃𝑃𝑃𝑃 =
𝑉𝑉𝑉𝑉 =
=� 𝜇𝜇𝜇𝜇
0.296𝑍𝑍𝑍𝑍 + 889 (14)
(14) (15)
𝑉𝑉𝑉𝑉𝑆𝑆𝑆𝑆 = � �𝜌𝜌𝜌𝜌
𝜌𝜌𝜌𝜌
𝑃𝑃𝑃𝑃
(14)
𝑉𝑉𝑉𝑉𝑆𝑆𝑆𝑆 = 0.26𝑍𝑍𝑍𝑍 𝜌𝜌𝜌𝜌 𝑃𝑃𝑃𝑃 − 428 (16)
𝑍𝑍𝑍𝑍𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆222 = (𝜌𝜌𝜌𝜌𝑉𝑉𝑉𝑉𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆 ))222 = 𝜇𝜇𝜇𝜇𝜌𝜌𝜌𝜌 (15)
𝑍𝑍𝑍𝑍
𝑍𝑍𝑍𝑍 = = (𝜌𝜌𝜌𝜌𝑉𝑉𝑉𝑉
(𝜌𝜌𝜌𝜌𝑉𝑉𝑉𝑉 ) = = 𝜇𝜇𝜇𝜇𝜌𝜌𝜌𝜌
𝜇𝜇𝜇𝜇𝜌𝜌𝜌𝜌 (15)
(15)
 Comparing pre- and post- stack seismic inversion methods - a case study from Scotian Shelf, Canada

Figure 3. a) Penobscot pre-stack seismic gather, b) Trim statics gather and c) CDP stack section.

Figure 4. Well log crossplots. (a) ZP versus ZS, (b) ZP versus density, (c) VP versus ZP, (d) VS versus ZP, (e) lr versus ZP and (f)
mr versus ZP.

illustrate trim static gather generated after pre-conditioning,
and Figure 3c shows stack section. Pre-stack trim static
gather is used by pre-stack inversion methods, whereas stack
section is used by post-stack seismic inversion methods.
Seismic pre-stack inversion methods estimate P-
impedance (ZP), S- impedance (ZS), Density (ρ), VP/VS
ratio, lambdarho (λρ) and murho (μρ), whereas post-stack
inversion methods estimate P- impedance only. To compare
other petrophysical parameters, the post-stack impedance
is transformed into S- impedance, density, VP/VS ratio,
lambdarho (λρ) and murho (μρ) sections, using relationship
derived from well log data. Figure 4 depicts crossplot of well
log curves to derive relationship between them. Figure 4a
shows crossplot of ZP versus ZS, figure 4b depicts crossplot
of ZP versus density, and Figures 4c-f depict crossplots of
VP, VS, lr and mr with respect to ZP respectively.

589
S.P.
𝑉𝑉𝑉𝑉𝑝𝑝𝑝𝑝 =Maurya
1.16𝑉𝑉𝑉𝑉𝑠𝑠𝑠𝑠 + and
1360N.P. Singh (1)
𝜌𝜌𝜌𝜌 = 0.23𝑉𝑉𝑉𝑉 0.25 (2)
ln(𝑍𝑍𝑍𝑍𝑆𝑆𝑆𝑆 ) = 𝑘𝑘𝑘𝑘ln(𝑍𝑍𝑍𝑍𝑃𝑃𝑃𝑃 ) + 𝑘𝑘𝑘𝑘𝐶𝐶𝐶𝐶 + ∆𝐿𝐿𝐿𝐿𝑆𝑆𝑆𝑆 (3)
(𝜌𝜌𝜌𝜌) = 𝑚𝑚𝑚𝑚ln(𝑍𝑍𝑍𝑍𝑃𝑃𝑃𝑃 ) + 𝑚𝑚𝑚𝑚𝐶𝐶𝐶𝐶 + ∆𝐿𝐿𝐿𝐿𝐷𝐷𝐷𝐷
ln⁡ (4)
∆𝑉𝑉𝑉𝑉𝑃𝑃𝑃𝑃 ∆𝑉𝑉𝑉𝑉𝑆𝑆𝑆𝑆 ∆𝜌𝜌𝜌𝜌
𝑅𝑅𝑅𝑅(𝜃𝜃𝜃𝜃) = 𝑐𝑐𝑐𝑐1 + 𝑐𝑐𝑐𝑐2 + 𝑐𝑐𝑐𝑐3 (5)
𝑉𝑉𝑉𝑉𝑃𝑃𝑃𝑃 𝑉𝑉𝑉𝑉𝑆𝑆𝑆𝑆 𝜌𝜌𝜌𝜌
1 𝑉𝑉𝑉𝑉𝑆𝑆𝑆𝑆 2 𝑉𝑉𝑉𝑉𝑆𝑆𝑆𝑆 2
where 𝑐𝑐𝑐𝑐1 = , 𝑐𝑐𝑐𝑐2 = 0.5 − � � 𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠2 𝜃𝜃𝜃𝜃 and 𝑐𝑐𝑐𝑐3 = 4 � � 𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠2 𝜃𝜃𝜃𝜃
2𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑠𝑠𝑠𝑠 2 𝜃𝜃𝜃𝜃 𝑉𝑉𝑉𝑉𝑃𝑃𝑃𝑃 𝑉𝑉𝑉𝑉𝑃𝑃𝑃𝑃
𝑇𝑇𝑇𝑇(𝜃𝜃𝜃𝜃) = 𝐶𝐶𝐶𝐶1 𝑊𝑊𝑊𝑊(𝜃𝜃𝜃𝜃) ∗ 𝐷𝐷𝐷𝐷𝐿𝐿𝐿𝐿𝑃𝑃𝑃𝑃 + 𝐶𝐶𝐶𝐶2 𝑊𝑊𝑊𝑊(𝜃𝜃𝜃𝜃) ∗ 𝐷𝐷𝐷𝐷𝐿𝐿𝐿𝐿𝑆𝑆𝑆𝑆 + 𝐶𝐶𝐶𝐶3 𝑊𝑊𝑊𝑊(𝜃𝜃𝜃𝜃) ∗ 𝐷𝐷𝐷𝐷𝐿𝐿𝐿𝐿𝐷𝐷𝐷𝐷 (6)
𝐿𝐿𝐿𝐿𝑃𝑃𝑃𝑃 = ln(𝑍𝑍𝑍𝑍𝑃𝑃𝑃𝑃 ) , 𝐿𝐿𝐿𝐿𝑆𝑆𝑆𝑆 = ln(𝑍𝑍𝑍𝑍𝑆𝑆𝑆𝑆 ) , and 𝐿𝐿𝐿𝐿𝐷𝐷𝐷𝐷 = ln(𝜌𝜌𝜌𝜌).
1 1 1
𝐶𝐶𝐶𝐶1 = 𝑐𝑐𝑐𝑐 + 𝐾𝐾𝐾𝐾𝑐𝑐𝑐𝑐 + 𝑚𝑚𝑚𝑚𝑐𝑐𝑐𝑐3 , 𝐶𝐶𝐶𝐶2 = 𝑐𝑐𝑐𝑐 , 𝑊𝑊𝑊𝑊(𝜃𝜃𝜃𝜃)
2 1 2 2 2 2
𝑇𝑇𝑇𝑇(𝜃𝜃𝜃𝜃1 ) 𝐶𝐶𝐶𝐶1 (𝜃𝜃𝜃𝜃1 )𝑊𝑊𝑊𝑊(𝜃𝜃𝜃𝜃1 )𝐷𝐷𝐷𝐷 ⋯ 𝐶𝐶𝐶𝐶3 (𝜃𝜃𝜃𝜃3 )𝑊𝑊𝑊𝑊(𝜃𝜃𝜃𝜃3 )𝐷𝐷𝐷𝐷
� ⋮ �=� ⋮ ⋱ ⋮ � (7)
𝑇𝑇𝑇𝑇(𝜃𝜃𝜃𝜃𝑁𝑁𝑁𝑁 ) 𝐶𝐶𝐶𝐶1 (𝜃𝜃𝜃𝜃𝑁𝑁𝑁𝑁 )𝑊𝑊𝑊𝑊(𝜃𝜃𝜃𝜃𝑁𝑁𝑁𝑁 )𝐷𝐷𝐷𝐷 ⋯ 𝐶𝐶𝐶𝐶3 (𝜃𝜃𝜃𝜃𝑁𝑁𝑁𝑁 )𝑊𝑊𝑊𝑊(𝜃𝜃𝜃𝜃𝑁𝑁𝑁𝑁 )𝐷𝐷𝐷𝐷
[𝐿𝐿𝐿𝐿𝑃𝑃𝑃𝑃 Δ𝐿𝐿𝐿𝐿𝑆𝑆𝑆𝑆 Δ𝐿𝐿𝐿𝐿𝐷𝐷𝐷𝐷 ] = [𝑙𝑙𝑙𝑙𝑠𝑠𝑠𝑠(𝑍𝑍𝑍𝑍𝑃𝑃𝑃𝑃0 0 0)],
𝑆𝑆𝑆𝑆(𝑡𝑡𝑡𝑡) = 𝑊𝑊𝑊𝑊(𝑡𝑡𝑡𝑡) ∗ 𝑟𝑟𝑟𝑟(𝑡𝑡𝑡𝑡) + 𝑠𝑠𝑠𝑠(𝑡𝑡𝑡𝑡) (8)
𝑍𝑍𝑍𝑍 = 𝑉𝑉𝑉𝑉𝜌𝜌𝜌𝜌 (9)
𝑍𝑍𝑍𝑍𝑠𝑠𝑠𝑠+1 − 𝑍𝑍𝑍𝑍𝑠𝑠𝑠𝑠
𝑟𝑟𝑟𝑟𝑠𝑠𝑠𝑠 = (10)
𝑍𝑍𝑍𝑍𝑠𝑠𝑠𝑠+1 + 𝑍𝑍𝑍𝑍𝑠𝑠𝑠𝑠
𝑁𝑁𝑁𝑁

𝑍𝑍𝑍𝑍𝑁𝑁𝑁𝑁 = 𝑍𝑍𝑍𝑍1 𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑝𝑝𝑝𝑝 �� 𝑟𝑟𝑟𝑟𝑠𝑠𝑠𝑠 � (11)

𝑠𝑠𝑠𝑠=1
𝑀𝑀𝑀𝑀
1 1
𝑒𝑒𝑒𝑒(𝑟𝑟𝑟𝑟) = 𝛼𝛼𝛼𝛼 ��𝑟𝑟𝑟𝑟𝑗𝑗𝑗𝑗 � + ∥ (𝑆𝑆𝑆𝑆 − 𝑊𝑊𝑊𝑊 ∗ 𝑟𝑟𝑟𝑟) ∥2 (12)
2 𝜎𝜎𝜎𝜎
𝑗𝑗𝑗𝑗 =1

𝜆𝜆𝜆𝜆 + 2𝜇𝜇𝜇𝜇
𝑉𝑉𝑉𝑉𝑃𝑃𝑃𝑃 = � (13)
𝜌𝜌𝜌𝜌

𝜇𝜇𝜇𝜇
Figure
𝑉𝑉𝑉𝑉𝑆𝑆𝑆𝑆 = � 5. Seismic inversion analysis plot(14)
for well L-30 where (a) shows comparison of P-impedance, (b) S-impedance, (c) density,
𝜌𝜌𝜌𝜌
and (d) VP/VS ratio.
𝑍𝑍𝑍𝑍𝑆𝑆𝑆𝑆2 = (𝜌𝜌𝜌𝜌𝑉𝑉𝑉𝑉𝑆𝑆𝑆𝑆 )2 = 𝜇𝜇𝜇𝜇𝜌𝜌𝜌𝜌
A best fit straight line gives the relationship among
𝜆𝜆𝜆𝜆𝜌𝜌𝜌𝜌 = 𝑍𝑍𝑍𝑍𝑃𝑃𝑃𝑃2 − 2𝑍𝑍𝑍𝑍𝑆𝑆𝑆𝑆2 parameters, as given below(16)
petrophysical
𝑍𝑍𝑍𝑍𝑆𝑆𝑆𝑆 = 0.741𝑍𝑍𝑍𝑍𝑃𝑃𝑃𝑃 − 1900
𝜌𝜌𝜌𝜌 = 0.000113𝑍𝑍𝑍𝑍𝑃𝑃𝑃𝑃 + 1.57
𝑉𝑉𝑉𝑉𝑃𝑃𝑃𝑃 = 0.296𝑍𝑍𝑍𝑍𝑃𝑃𝑃𝑃 + 889
𝑉𝑉𝑉𝑉𝑆𝑆𝑆𝑆 = 0.26𝑍𝑍𝑍𝑍𝑃𝑃𝑃𝑃 − 428
𝜆𝜆𝜆𝜆𝜌𝜌𝜌𝜌 = 4.3 × 103 𝑍𝑍𝑍𝑍𝑃𝑃𝑃𝑃 − 2.8 × 106
𝜇𝜇𝜇𝜇𝜌𝜌𝜌𝜌 = −0.00023𝑍𝑍𝑍𝑍𝑃𝑃𝑃𝑃 + 3.76
Using these relationships (Eq. 13-18), inverted P-
impedance from the post stack inversion is transformed
into the S- impedance, density, VP/VS ratio, lambdarho (λρ)
and murho (μρ) sections.
The inversion methods has been applied in two
steps, firstly, one composite trace near to well location
(L-30, inline 1177, xline 1153) is extracted and inverted
for petrophysical parameters to cross verify the results.
Figure 5 depicts comparison of inverted results with well
log. Figure 5a shows comparison of P-impedance, Figures
5b, 5c and 5d show comparison of S-impedance, density
and VP/VS from inverted results and well log curves. From
Figure 5, it is noticed that the inverted curves from pre- and
post- stack inversion methods agree well with the well log
curves. The only difference noticed here is that the post-
(15)
stack curves are more close to the well log compared with
the pre-stack results. The error analysis among all the
(13) inverted parameters are performed and described in table 1.
(14) Figure 6 depicts crossplot of inverted petrophysical
(15) parameters with the well log petrophysical parameters
(16) for the Quality check (QC) of the inverted results. Figure
(17) 6a depicts crossplot of inverted and original (well log) P-
(18) impedance, while Figures 6b, 6c and 6d depict crossplot
of inverted and original S- impedance, density and VP/
VS ratio respectively. In Figure 6, the blue circle indicates
inversion results from pre-stack inversion methods while
the red circle indicates post-stack inversion results. The
distribution of scatter points from the best fit line indicates
that the inverted results are very close to the original
values as the point’s lies very close to the best fit line for
both the cases.
After getting satisfactory results from the composite
trace, both inversion methods (pre- and post- stack) are
applied to the seismic volume to estimate rock properties
of the subsurface. Figure 7 depicts cross-section of inverted
P- impedance for inline 1161. Figure 7a shows impedance
cross-section generated using pre-stack inversion algorithm,
Figure 7b depicts cross-section of impedance generated

590
 Comparing pre- and post- stack seismic inversion methods - a case study from Scotian Shelf, Canada
Figure 6. Crossplot of inverted and original, (a) P-impedance, (b) S-impedance, (c) density, and (d) VP/VS ratio.
using post-stack seismic inversion method and Figure 7c
depicts difference between pre- and post-stack impedance
sections. From Figure 7, it can be noticed that both the
inversion methods (pre- and post-stack) give similar type of
results with least differences between them. The relatively
large differences (300-800 m/s*g/cc) are observed at later
time which may be due to presence of high impedance
contrast in this zone.
Figure 8 shows cross-section of inverted S- impedance
at inline 1161. Figure 8a depicts cross-section of S-
impedance generated using pre-stack inversion technique,
Figure 8b shows cross-section of S-impedance generated
using post-stack inversion and Figure 8c shows difference
between them. From Figure 8, it is noticed that both the
inversion methods give high resolution images of the
subsurface in similar way with least differences between
them. The relatively large differences (200-500 m/s*g/cc)
are observed because of S- impedance is inverted from pre-
stack inversion directly, whereas the S-impedance is derived
from post-stack P-impedance using well log relation which
holds good hold for trace near to well location and shows
small deviation far traces away from the borehole.
Figure 9 shows cross-section of inverted density at
inline 1161 only for simplicity. Figure 9a depicts cross-
section of density generated using pre-stack inversion
algorithm, Figure 9b depicts cross-section of density
generated using post-stack seismic inversion method and
Figure 9c shows density differences between pre - and post-
stack inverted results. It is noticed from Figure 9 that both
the inversion methods again give similar type of results of
the subsurface, with very small differences between them.
The relatively large differences (0.1-0.3 g/cc) are observed in
middle time interval that may be due to transformation of
P- impedance into density section from post-stack inversion
while pre-stack inversion methods estimate density sections
directly although the differences are negligible.
Figure 10 shows cross-section of inverted lr at inline
1161, wherein Figure (a) shows inverted lr generated using
pre-stack inversion, while Figure (b) to that generated using
post-stack seismic inversion utilizing relationship from
well log data. Similarly, Figure (c) reveals the differences
between pre-stack and post-stack generated lr section. It
can be noticed from these figures that both the inversion
methods (pre- and post-stack) give similar type of results
with very small differences between them. The relatively
large differences (1-5GPa*g/cc) are observed in middle
time interval that may be due to the transformation of
P-impedance into lr section from post-stack inversion
591
S.P. Maurya and N.P. Singh

Figure 7. Cross-section of inverted P- impedance estimated using (a) pre-stack, (b) post-stack inversion, and (c) difference
between them.

Figure 8. Cross-section of inverted S-impedance estimated using a) pre-stack, b) post-stack inversion and (c) difference between
them.

while pre-stack inversion methods estimate lr sections
directly.
Figure 11 shows cross-section of inverted mr at inline
1161 for simplicity, although the inversion is performed
for entire volume (inline 1161-1200, cross-line 1000-1481).
Figure (a) shows cross-section of mr generated using pre-
stack inversion algorithm while Figure (b) depicts cross-
section of mr generated using post-stack seismic impedance
inversion using well log relation. Similarly, Figure (c)
shows differences of mr between pre- and post-stack results.
Similar types of results are found from both the inversion
methods with very small difference between them. The
relatively large differences (1-5 GPa*g/cc) are observed in
the middle time interval.
Table 1 illustrates quantitative differences between
pre- and post-stack inverted results. Column 1 of table
1 depicts parameters to be compared, column 2 and 3
depict correlation and RMS error for pre-stack inversion,
while column 4 and column 5 illustrates correlation and
RMS error for post-stack seismic inversion methods. The
quantitative values show that the inverted results from
the post stack inversion are more efficient and close to

592
 Comparing pre- and post- stack seismic inversion methods - a case study from Scotian Shelf, Canada

Figure 9. Cross-section of inverted density estimated using (a) pre-stack, (b) post-stack inversion and (c) difference between them.

Figure 10. Cross-section of inverted λρ estimated using (a) pre-stack and (b) post-stack inversion. Figure (c) shows the difference
between the two.

the original results compared to the pre-stack inversion
methods. The interpretation of inverted section shows
continuous variation of murho parameters in the subsurface
and it indicates non-availability of reservoir.
Further, the porosity is predicted in inter-well region
by using acoustic impedance generated from pre- and
post- stack seismic inversion methods. Generally, this
property is predicted using neural network, or single
attribute regression or multi attribute regression etc. but
these tools required more than one well in the area, hence
cannot be used in the present study as the area does have
only one well (L-30). An alternative way has been used in
the present study to estimate porosity section from the
impedance section, i.e. to use relationship between porosity
and impedance derived from the well log data. This type
of relationship holds good for traces near to well location,
but shows small deviation and quantitative interpretation
cannot be performed accurately. Initially, well log porosity
and impedance is cross plotted (Figure 12) and a best fit
straight line give relationship between these two quantities.
The estimated relation between density porosity (ϕ) and
impedance (ZP) can be given as

593
S.P. Maurya and N.P. Singh

Figure 11. Cross-section of inverted μρ estimated using (a) pre-stack, (b) post-stack inversion, and (c) difference between the two.

Figure 12. Crossplot of well log porosity and well log impedance.

Table 1. Comparison of inversion results.

S.No. Parameters Pre-stack Post-stack

Correlation RMS Error Correlation RMS Error
1. P-impedance 0.99 0.06 0.94 0.11
2. S-impedance 0.92 0.15 0.89 0.18
3. Density 0.93 0.13 0.90 0.16
4. Vp/Vs 0.89 0.22 0.88 0.30
5. λρ 0.91 0.19 0.87 0.32
6. μρ 0.90 0.18 0.88 0.29

(19) inverted impedance, while Figure (b) shows porosity section

Using equation 19, acoustic impedances estimated from
pre- and post - stack inversion methods are transformed
into porosity section. Figure 13 depicts cross-section of
estimated porosity from acoustic impedance. Figure (a)
shows porosity section generated from pre-stack seismic
generated from post-stack inverted impedance. Similarly,
Figure (c) shows difference between them. As the impedance
and porosity is closely related to each other, hence the
cross-section of porosity shows very high resolution of
subsurface porosity. The Figures demonstrate that almost

594
 Comparing pre- and post- stack seismic inversion methods - a case study from Scotian Shelf, Canada
Figure 13. Cross-section of inverted μρ estimated using (a) pre-stack, (b) post-stack inversion and (c) difference between them.
both methods estimate porosity in similar way with only
3 very small patches of large difference seen which is
highlighted in the Figure 13c. The porosity section shows
continuous variation of porosity and does not indicate any
high porosity patches and confirm non-existence of any
major reservoir in this zone.
From the analysis it is concluded that the pre - and
post - stack inversion methods provides similar results with
higher resolution from post-stack methods, however this
method only estimate P-impedance. On the other hand pre-
stack inversion methods estimate variety of attributes i.e.
P- impedance, S- impedance, density, VP/VS ratio, lambdarho
and murho. Relationships are derived from well log data
and used to transform P- impedance from post stack into
attributes and compared with pre-stack. It is noticed that
the derived ZS, Density, VP/VS ratio, lambdarho and murho
are following trend from well log data nicely. The difference
among these petrophysical parameters demonstrate that
both methods produces similar results, with small deviation,
particularly where larger values are present. These differences
also arise due to use of equations from well log data which
is best suited for the traces which are very close to the well
log data and shows deviation away from the boreholes. The
interpretation of inverted results indicates non-existence
of any major reservoir. Due to presence of reservoir, low
impedance and corresponding high porosity patches should
appear but in the present study continuous variation of
derived parameters have been noticed.
CONCLUSIONS
The present study utilizes two types of seismic inversion,
namely pre-stack and post-stack, to estimate subsurface
petrophysical parameters, which are helpful in interpreting
seismic sections. The relative performances of these
two techniques are discussed in each steps of seismic
inversion. The methods first applied to the composite
trace near to well location and inverted for petrophysical
parameters i.e. P- impedance, S- impedance, density, P-
wave, S- wave velocity, VP/VS ratio, lambdarho and murho
sections. The pre-stack inversion methods estimate these
parameters directly from seismic data, while post-stack
inversion methods estimate only P-impedance section. This
P- impedance section is transformed into S- impedance,
density, P- wave, S- wave, VP/VS ratio, lambdarho and
murho sections using relationship derived from well log
data. The analysis for composite trace shows that both
inverted curves follow well log data nicely and depict good
performance of the algorithm. The error analysis shows
that the inverted curves from post-stack inversion have
higher correlation and low RMS error, compared with the
pre-stack inversion methods. Thereafter, the methods are
applied to entire seismic section to estimate subsurface
petrophysical parameters. The inverted sections from
both the methods shows very high resolution image of
the subsurface. The resolution of post-stack is slightly
greater than pre-stack methods. The difference among these
petrophysical parameters demonstrate that both methods
produces similar results with small deviation particularly
where larger values are present. These differences may also
arise due to use of equations from well log data which is
best appropriate for the traces which are very close to the
well log data and shows deviation away from the boreholes.
The interpretation of inverted results also indicates non-
existence of any major reservoir.
ACKNOWLEDGEMENTS
One of the author (S.P. Maurya) is indebted to Science
and Engineering Research Board, Department of Science
595
S.P. Maurya and N.P. Singh
and Technology, New Delhi for financial supports in form
of research project (Grant no PDF/2016/000888) under
National Post-doctoral Fellowship scheme. Authors also
acknowledge the CGG Veritas for providing seismic and
well log data of Blackfoot field, Alberta, Canada.
Compliance with Ethical Standards
The authors declare that they have no conflict of interest
and adhere to copyright norms.
REFERENCES
Berteussen, K. and Ursin, B., 1983. Approximate computation
of the acoustic impedance from seismic data. Geophysics,
48(10), 1351-1358.
Campbell, D.C., Shimeld, J., Deptuck, M.E. and Mosher, D.C., 2015.
Seismic stratigraphic framework and depositional history of a
large upper Cretaceous and Cenozoic depocenter of southwest
Nova Scotia Canada. Mar. Petrol. Geol., 65, 22-42.
Carrazzone, J.J., Chang, D., Lewis, C., Shah, P.M. and Wang,
D.Y., 1996. Method for deriving reservoir lithology and
fluid content from pre-stack inversion of seismic data. US
Patent, 5, 583, 825.
Castagna, J.P., Batzle, M.L. and Eastwood, R.L., 1985. Relationships
between compressional-wave and shear-wave velocities in
clastic silicate rocks. Geophysics, 50(4), 571-581.
Chen, Q. and Sidney, S., 1997. Seismic attribute technology for
reservoir forecasting and monitoring. The Leading Edge,
16(5), 445-448.
Clochard, V., Delépine, N., Labat, K. and Ricarte, P., 2009.
Post-stack versus pre-stack stratigraphic inversion for
CO 2 monitoring purposes: A case study for the saline
aquifer of the Sleipner field. SEG Annual Meeting, Society
of Exploration Geophysicists, pp. 2417-2421, https://doi.
org/10.1190/1.3255345.
Cummings, D.I. and Arnott, R.W.C., 2005. Growth-faulted shelf-
margin deltas: a new (but old) play type offshore Nova
Scotia. B. Can. Petrol. Geol., 53(3), 211-236.
Downton, J.E., 2005. Seismic parameter estimation from AVO
inversion. PhD Thesis, University of Calgary, Department
of Geology and Geophysics.
Fatti, J.L., Smith, G.C., Vail, P.J., Strauss, P.J. and Levitt, P.R., 1994.
Detection of gas in sandstone reservoirs using avo analysis:
A 3-d seismic case history using the geo-stack technique.
Geophysics, 59(9), 1362-1376.
Ferguson, R.J. and Margrave, G.F., 1996. A simple algorithm
for band-limited impedance inversion. CREWES research
report, 8(21), 1-10.
Gardner, G., Gardner, L. and Gregory, A., 1974. Formation velocity
and density-the diagnostic basics for stratigraphic traps.
Geophysics, 39(6), 770-780.
596
Gholami, A., 2016. A fast automatic multichannel blind
seismic inversion for high-resolution impedance recovery.
Geophysics, 81(5), V357-V364.
Goodway, B., 2001. AVO and Lamé constants for rock parameterization
and fluid detection. CSEG Recorder, 26(6), 39-60.
Goodway, B., Chen, T. and Downton, J., 1997. Improved AVO
Fluid detection and Lithology discrimination using Lamé
petrophysical parameters; “λρ” μρ λμ Fluid Stack” from
P and S inversions. SEG Annual Meeting, Society of
Exploration Geophysicists, pp.183-186.
Kidston, A.G., Brown, D.E., Smith, B.M. and Altheim, B., 2005.
The Upper Jurassic Abenaki Formation, Offshore Nova
Scotia: A Seismic and Geologic Perspective. Canada-Nova
Scotia Offshore Petroleum Board, Halifax, Nova Scotia, p.168.
Kidston, A.G., Smith, B.M., Brown, D.E., Makrides, C. and
Altheim, B., 2007. Nova Scotia Deepwater post-drill analysis
1982-2004. Unpublished Report to the Canada-Nova Scotia
Offshore Petroleum Board, p.181.
Krebs, J.R., Anderson, J.E., Hinkley, D., Neelamani, R., Lee, S.,
Baumstein, A. and Lacasse, M.D., 2009. Fast full-wave field
seismic inversion using encoded sources. Geophysics, 74(6),
WCC177-WCC188.
Latimer, R.B., Davidson, R. and Van Riel, P., 2000. An interpreter’s
guide to understanding and working with seismic-derived
acoustic impedance data. The leading edge, 19(3), 242-256.
Lindseth, R.O., 1979. Synthetic sonic logs-a process for
stratigraphic interpretation. Geophysics, 44(1), 3-26.
Mallick, S., 1995. Model-based inversion of amplitude-variations-
with-offset data using a genetic algorithm. Geophysics,
60(4), 939-954.
Matlab and Statistics Toolbox Release, 2015b. The Math Works
Inc. Natick Massachusetts United States.
Maurya, S. and Singh, K.H., 2015, LP and ML sparse spike
inversion for reservoir characterization- a case study from
Blackfoot area Alberta Canada. 77th EAGE Conference and
Exhibition, 1st – 4th June, Madrid, Spain.
Maurya, S.P. and Singh, K.H., 2017. Band limited impedance
inversion of blackfoot field, Alberta, Canada. J. Geophysics,
38(1), 57-61.
Maurya, S.P., Singh, K.H. and Singh, N.P., 2018b. Qualitative
and quantitative comparison of geostatistical techniques of
porosity prediction from the seismic and logging data: a case
study from the Blackfoot Field Alberta Canada. Mar. Geophys.
Res., 1-21, https://doi.org/10.1007/s11001-018-9355-6
Maurya, S.P., Singh, K.H., Kumar, A. and Singh, N.P., 2018a.
Reservoir characterization using Post-Stack seismic inversion
techniques based on real coded genetic algorithm. J.
Geophysics, 39(2), 95-103.
Moncayo, E., Tchegliakova, N. and Montes, L., 2012. Pre-stack
seismic inversion based on a genetic algorithm: A case
from the Llanos Basin (Colombia) in the absence of well
information. CT and F-CienciaTecnologíay Futuro, 4(5), 5-20.
 Comparing pre- and post- stack seismic inversion methods - a case study from Scotian Shelf, Canada

Morozov, I.B. and Ma, J., 2009. Accurate post-stack acoustic
impedance inversion by well-log calibration. Geophysics,
74(5), R59-R67.
Pendrel, J., 2006. Seismic inversion - a critical tool in reservoir
characterization. Scandinavian oil-gas magazine, 5(6), 19-22.
Russell, B., 1999. Comparison of post-stack seismic inversion
methods. SEG Technical Program Expanded Abstracts, p.10.
Schuster, G.T., 2017. Seismic inversion. Society of Exploration
Geophysicists. p1-353, https://doi.org/10.1190/1.9781560803423.
Shu-jin, Y., 2007. Progress of pre-stack inversion and application
in exploration of the lithological reservoirs. Prog. in geophy.,
3, p.032.
Smith, G. and Gidlow, P., 1987.Weighted stacking for rock property
estimation and detection of gas. Geophys. Prospect., 35(9),
993-1014.
Srivastava, R. and Maultzsch, S., 2018. Integration of results
from two seismic inversion methods – A Case Study. In
80th EAGE Conference and Exhibition, 11 th-14th June,
Copenhagen, Denmark.
Vardy, M.E., Clare, M.A., Vanneste, M., Forsberg, C.F. and Dix,
J.K., 2018. Seismic inversion for site characterization: When,
Where and Why Should We Use It? In Offshore Technology
Conference, 30th April – 3rd May, NRG Park, Houston, Texas,
USA, https://doi.org/10.4043/28730-MS.

Received on: 23.7.18; Revised on: 21.9.18; Accepted on: 28.9.18

597