Sunteți pe pagina 1din 13

DOI: 10.

1590/2317-4889201820170072

ARTICLE

Processing of large offset


data: experimental seismic line
from Tenerife Field, Colombia
Francisco Gamboa Ortega1, Amin Bassrei1*, Ellen de Nazaré Souza Gomes2,
Michelângelo Gomes da Silva1, Andrei Gomes de Oliveira2

ABSTRACT: Exploration seismology provides the main source of information about the Earth’s subsurface, which in many cases can be presented as a
simple model of horizontal or near-horizontal layers. After the seismic acquisition step, conventional seismic processing of reflection data provides an image
of the subsurface by using information about the reflections of these layers. The traveltime from a source to different receivers is adjusted using a hyperbolic
function. This expression is used in the case involving an isotropic medium, which is a simplification of nature, whereas geologically complex media are
generally anisotropic. A subsurface model that more closely resembles reality is the vertical transverse isotropy, which defines two parameters that are required
to correct the traveltimes: the NMO velocity and the anellipticity parameter. In this paper, we reviewed the literature and methodology for velocity analysis
of seismic data acquired from anisotropic media. A model with horizontal layers and anisotropic behavior was developed and evaluated. The anisotropic
velocity was compared to the isotropic velocity, and the results were analyzed. Finally, the methodology was applied to real seismic data, i.e. an experimental
landline from Tenerife Field, Colombia. The results show the importance of the anellipticity parameter in models with anisotropic layers.
KEYWORDS: non-hyperbolic velocity analysis; anisotropy; anellipticity parameter; experimental seismic line.

INTRODUCTION Several research studies demonstrated that the pres-


ence of a simple seismic anisotropy in a model, such as
the vertical transverse isotropy (VTI) model, produces sig-
The main objective of exploration seismology is to obtain nificant distortions in conventional seismic data analysis.
subsurface images that may indicate possible hydrocarbon For instance, the normal moveout (NMO) velocity is not
reservoirs after proper interpretation. In the seismic inter- equal to the root mean square (RMS) velocity, both for
pretation step, the images obtained from seismic reflec- small and large offsets.
tion data should be faithful to subsurface characteristics. This type of medium produces a non-hyperbolic trav-
However, the seismic interpretation of geologically-com- eltime curve, which is manifested by significantly large off-
plex media images is usually complicated. One factor that sets for PP-waves, i.e. a P-wave reflected as a P-wave. As to
contributes to this difficulty is the type of processing per- the PS-wave, i.e. a P-wave that converts to an S-wave in the
formed on data. In many situations, the geological medium reflection, this behavior is observed in both small and large
is not isotropic but anisotropic, that is, a given physical offsets (Alkhalifah 1997). A possible solution would be to
property varies with the direction. Accurate modeling of remove overcorrected traces and to stack all the others. Herein,
anisotropy features is frequently ignored in seismic data the images could not provide the complete information.
processing, especially since geologically-complex media In an anisotropic medium, the mathematical repre-
in the quasi-static regime behave similarly to anisotropic sentation of the source-reflector-receiver traveltime can be
media (Helbig 1994). expressed by a shifted hyperbola (Castle 1994). Such hyperbola

1
Instituto de Geociências & Instituto Nacional de Ciência e Tecnologia de Geofísica de Petróleo, Universidade Federal da Bahia, Salvador (BA), Brazil.
E-mails: franciscogamboa12@gmail.com, bassrei@ufba.br, mgs@cpgg.ufba.br
Faculdade de Geofísica, Universidade Federal do Pará, Belém (PA), Brazil. E-mails: ellensgufpa@gmail.com, andrei.jval@gmail.com
2

*Corresponding author
Manuscript ID: 20170072. Received on: 05/21/2017. Approved on: 11/13/2017.

147
Brazilian Journal of Geology, 48(1): 147-159, March 2018
n

2 2
∑v
i =1
2
nmo i t
V rms =V nmo = n
∑t
i =1
i

Seismic processing of large offset data


vnmo = vi 1 + 2 δ i
⎧ n 4 ⎫
vnmo ti [1 + 8η i ] ⎪
1 ⎪∑
η = ⎨ i =1 nt 2correction 2 ‒1 x⎬2
may be usedeff to do 8⎪ the NMO = t + ⎪2 with knowledge The medium is ideal, without energy loss and is also considered
of three parameters: ⎩
Vnmo 4
∑i =1
x
ti 0 Vnmo
⎭ non-dispersive. The mathematical representation of the trav-
N
■■ the zero offset source-reflector-receiver
t = τvelocity
x2
τ 02 +V 2 ; V 2 = i =1
∑ vi titraveltime t0;
2 eltime from the source to the receiver in a single-layer model
■■ the NMO s+ νnmo rms is given by Equation 1:
N
■■ time-weighted moment of the velocity
t ∑ t i
distribution (µ).
τ0 = 0 i =1
2
S x2 2ηx 4 t 2 = t 2 + x (1)
Alkhalifah and Tsvankint 2(1995) 2 showed that three
τ = τ ( S ‒ 1) x = t 0 +
Vnmo 2
‒ 2
x
⎡ 2 2the+ (1 + 2 η)x 2 ⎤⎦ Vnmo
0 2
parameters s are 0necessary for VTI media to Vperform
nmo ⎣t 0 Vnmo
ν = SVrms
2 2 N
time processing:
µ4 source-reflector-receiver
µ4
δ << 1 ∑ vi2 ti
■■ the zero2 offset traveltime t ; Where: 2
Vrms = N i =1
xS = 2 = 4 ε << 1 γ 0

t x2 ■=the
t 02 +NMO
Vnmo 2
µvelocity
2 V rms V nmo
;
ε ‒δ
x is the offset; ∑ ti
N µ = ∑
■■ the anisotropy ∆τparameter
V j η . = t is the traveltime
i =1
for zero offset;
k k
1+ 2 δ 0
x2 2ηx 4 x; and
∑ vi ti ∑ ∆τk
2j
n
t
txx = t 0 + 2 ‒ 2 the
2 is the 2 traveltime at offset
Vrms 2 Fomel
= i =1N S(2004) = 1 + 8and ηeff Aleixo and2Schleicher ∑
(2010) 2
vnmo textended
i Vnmo is theVNMO nmo Vvelocity. ⎡ 2 2In+conventional
nmo ⎣t 0 Vnmo (1 + 2 η)x 2 ⎤⎦ velocity analysis,

2 i =1
this parameterization ti by determining Vrms = Van = n 2
nmoapproximation that the
δ << NMO 1 velocity is denoted as the RMS velocity.
xt
was closer kto2 =exact 1data.
i =1 ω v
‒2 2x 4
2 2
k x2 t x
2
= t 2
0 + ∑i =1 2
i
ε << 1 γ
x z vt 2 of ωpaper 2ηx was not to showV
t x2 =Thet 02 +objective ‒ this
x = t0 + 2
vnmo+ 2=2ηv)2ix 21⎤+ 2Nδ i 2
the
nmo accuracy In a model with N flat layers, composed of an isotro-
ε ‒δ
of the traveltime Vnmo 2
VVnmo
2
⎡⎣t 02VV
approximations
2
2 nmo+ (1
ωfound kx ⎦ in∑ literature, but η = medium,
pic 2 the RMS velocity is calculated as Equation 2:
nmo ω nmo vi ti
k = N ‒ ⎧ ⎫ 1 2+ 2 δx
ti [1 +real i]
n 2
to
δ compare
<< 1 z the NMO v V ∑ correction
nmo i ω
2
v ti ‒ in
2 2
1nmo⎪rms∑
2Vsynthetic
V ηk=xvnmoNand
2 2 i
4 = 1
8ηdata t
⎪ x 0 Vnmo
= t + 2 n
by <<
ε
using 1 two techniques:
V 2
=
γ rmst obs ‒ tNapp
i =1 one that depends
η eff = ⎨ i =1

on t the
in
NMO ‒ 1 ⎬ N2
∑ 2
vnmo ti
8 ⎪ ⎪
∑i ⎩ 2 nmox∑ ∑
2 i =1
velocity estimate Error =in addition t × to
100 the anelasticity i
V 2 ti
=
4 1
parameter V rms = V nmo v =2
t
ε ‒δ i =1 The two ⎭ 2ηx2
i i n

4
η ,=and another that uses t obsi =1only the NMO 2 velocity. V = i =1 ti (2)
t = t + ‒ rms
1+ 2 δ Δ f 2 Δt 2 x 2 x 0 4

N

2η x 2V 2
∑ ti ⎦
η
2 2 2 2 i =1
techniques are nbased V t V + (1 + 2 ) x
=t x 2= ton nmo+the classical
2 t = τ
‒ 2 +equations τ 022+ 2
x ofnmo Alkhalifah nmo ⎣ 0 nmo
vnmo = vi i =11 + 2 δ i
f∑
0
and Tsvankin vnmott0i and
(1995), Vnmo of Castle Vnmos ⎡⎣(1994) tδ02V<< nmo1+ νfor(1 +the 2 η)NMO x 2 ⎤⎦
2
⎧ xn 4 2ηx 4⎫
‒ ti 2[1 + 82 η i ]2 ⎪
2 2 i =1
V rms = Vnmo Δ
correction =t nmo t app1‒ (1994)
.δn=Castle
<< t0 equation
t 0 << has the character- t x2 = t 02 +⎪ ∑ v
τ0 = ε 1 γ
∑ Vnmo ⎡⎣t 0 Vnmo + (1 + 2 η)x 2 ⎤⎦
2
istic of using a displacement ti factor S, which may or may not Where: 1 Vinmo nmo
S η = ⎨ =1
‒ 1⎬
iε =1 << 1 γ ε ‒δ eff n
τ s = τ 0 η( S.=When ‒ 1) this parame- ⎪
18 interval 4velocity of the⎪ i-th layer and
depend on the anisotropy parameter
vnmo = vi 1 + 2 δ i ε ‒ δ 1+ 2 δ
vδi is<< the

Vnmo ∑ ti

ter depends on η ,=the equationν 2of=Ursin SVrms 2 and Stovas (2006) t is
εi << 1 the vertical γ 2 traveltime
i =1 of the i-th layer. The RMS veloc-
⎧ 1 + 2 δ ⎫
n

ti [1The i]

n 2
vnmo ti
1in⎪a∑
was implemented. 4
vnmo + 8ηefficiency n⎪
of these approaches was ity by Equation x 2 refers to a set of N layers from the top of
2 µ4 2 µ4 t = τsε+‒ δτ 02 + 2
‒∑ η ν
2
tested
η eff = ⎨ i synthetic
=1 experiment S =v with t
1⎬ nmo2 i rms 4 nmo V =a = V
five-layer = i =1
model. the = first layer (i = 1) to the bottom of the last one (i = N).
i =1 µ2
n
1+ 2 δ
8 ⎪ V
This articleVisnmoorganized

4
2 n

rms = V 2
=
ti nmo as follows. ⎪
⎭ t ∑ ∆τkVk j
n
V
First, the description
rms
∑ t i of
τ0 = 0
t n
the two methods developed i =1
µ ∑
by Alkalifah = i and Tsvankin
vnmo = vi 1 2 δ i +
i =1
(1995) A S more ∑
realistic
2
vnmo trepresentation
i of the traveltime equa-
and Castle (1994), x 2
t = τ + τ 02 + v2 nmo = vi 1 + 2 δ i
which corrects
j
i= 1

traveltime ∆τ k curves in aniso- V
tion
τ rms
2
= =
τ 0 (
V
implies
2
S
nmo ‒ = 1 )
i = 1
knowledge
n of the anisotropic medium. For

s
isνreviewed. Then,S =these ⎧ n 4 ⎫ results, tithe appropriate resolution in the differ-
ηeff methodologies v ti [1 + 8η i ] ν ⎪= SVrms
⎫∑1 nmo
tropics media, 1 + 8two are better with
1 ⎪in
2 2
⎧ n real i =1
applied
τ0 =
t 0 to synthetic data and ⎪ ∑
1Colombia.
v t [
4 data acquired
1 +
nmo i ω eff i v kx⎪
η8 η =] 2 ⎨ 2 i =the Tenerife
n
ent
v‒ 1⎬events,µ
= v 1 µ
it
+ is important to take the medium’s anisotropy
2 δ
Field, SMagdalenaηValley, ⎨ i =1 kz The = experimental1 ‒ ‒821⎪⎬ 2D Vnmo ∑
seismic
4
ti Snmo
into =⎪ account.
4 i
= 44 i
eff = ω
nv ⎩ ⎭ µ 2
For a⎪V
⎧ nrms 4 ⎫
τ s =isτ 09(km
line S ‒in 1)length and⎪presents 8 Vnmolarge
4
∑ ti offsets2 of⎪ up to 52km. i = 1 2

medium vjnmo ti [1with i]
+ 8ηVTI anisotropy,
⎪ the conventional
2 The methodologies
ν = SVrms 2

(Alkalifah iV
and =1 nmo
kz = t = τ +2 τ‒2 +2Tsvankin ω 1995, ⎭
x ω
2
Castlek x
2
method
ηµeff== ∑ 1 ∆τ
⎨ hasi k=V
1 several
k limitations, ‒1 ⎬ including the fact that the
x NMO∑
n
1994) were validated by applying 2
x each method v V
s 0ω ν‒22V 2 ηk
nmo to synthetic nmo
j 8 ⎪ ∆τk Vnmo
velocity is ∑
4 ⎪
notti equal to the RMS velocity, whether
µ4 µ4 t = τs + τ 02 + 2 ⎩ ⎭
and
S = real =data. The results show ν that the t use t‒ of
t conventional S
in = 1 +
small 8 η or large i =1
offsets. An anisotropic medium produces
=τ0 = media×may
obs 0 app eff
seismic µ2 processing
2
Vrms 4
techniques Error
for isotropic 100result a non-hyperbolic x 2 traveltime curve, which is more signifi-
t tSobs t = τsω + τ 02 +v 2 kx22
in ∑ ∆τkVk results.S
µj inconsistent
=
j τ = 0
0
Δf τΔs =t nmoτ 0 ( S ‒ 1)
kz = manifested
cantly 1 ‒ ν2 in large offsets in the case of PP waves.
v ω
∑ ∆τk τs = τ 0 ( S ‒ 1) f ν 2t= SVrms2 = Regarding
τ =
t 0 PS waves, this behavior is observed in all offsets,
VSnmotheyωare smallωor klarge
2 2 2
S = 1 + 8 ηeff ν 2 = SVrms 2
0
k
0
whether = ‒ x (Alkalifah 1997).
Δ t nmo = t app ‒4 t 0 µ4 µ
( )
z
TRAVELTIME CALCULATION S= 2 = 4 τ s =To τ 0vclassify
S ‒V1nmo
2
an offsetω 2 ‒by 2Vits 2
ηkx we must consider the asso-
nmosize,
ω 2 2
v kx µ4 µ µ2 Vrms
kz = 1 ‒ BY S ALKHALIFAH
= 2 = 44 AND ciation
ν 2 = SVbetween 2t
‒ t appthe offset x and the depth z. If x/z>1.5, the
v ω2
TSVANKIN’S µ 2 V rms METHOD µj =
∑ ∆τ V
k k
j
Error =
offset is considered
µ4
rmsobs

µt4obs
×large
100 related to a non-hyperbolic travel-

kz =
Vnmo ω 2
µ‒j = 2
∑ kx k ω ∆τ 2 2 j
k V ∑ ∆τk S = curve.
time
Δ f µ22 ΔV
= 4If this ratio is lower than 1.5, the offset will be
t nmo
Thev conventional
Vnmo 2
‒∑2V∆τnmo
ω velocity 2
ηkx
k analyses = 1 + 8 ηtypically
Smethods used small = (Alkalifah rms
1997).
eff
f ∑ ∆τt 0kVk j
for seismict obsdata ‒ t appS = 1
processing + 8 η assume an ideal reflector, i.e. a homo- µj = The standard NMO equation used by the industry only
ω v 2 kx2

eff
Error =
geneous reflector × 100 k = 1 ‒ Δ t = t∆τ ‒ t
t obs withωflat interfaces and z constant thickness. considers the k first two terms of Taylor’s series expansion,
nmo app 0
v2k 2 v ω2
kz = 1 ‒ 2x S = 1 + 8 ηeff
Δf Δ t nmo v ω Vnmo ω 2 ω 2 kx2
= k = ‒ ω v2k 2
f t0 Vnmo ω 2 ω kx v Vnmo ω 2 148
z 2 2 2
‒ 2Vnmo2
kzη=kx 1 ‒ 2x
Δ t nmo = t app ‒ t 0kz = v V 2 ‒Brazilian ω 2 ‒ 2Vnmo 2Journal of Geology, 48(1): 147-159,
ηtkx ‒ t
v ωMarch 2018
nmo obs app
Error = × 100 V ω2 ω 2 kx2
t obs ‒ t app t obs kz = nmo ‒ 2
Error = × 100 v Vnmo ω ‒ 2Vnmo 2 2
ηkx
t obs Δf Δ t nmo
rms N 2 x 0
x Vnmo 2
t x2 = t 02 +∑22ti
x
t x2 = t 02 + iV=1nmo
N

NV x
22
2η x 4 ∑ vi2 ti
t x = t0∑
2 2 nmo2
+ v2i ti ‒ 2 Vrms = 2 i =1
N
i =1Vnmo Vnmo ⎡⎣t 02Vnmo 2
+ (1 + 2 η)Nx 2t⎤⎦
rms = ∑N ∑
2
Francisco GamboaVOrtega
2
v
eti al. t i i
Vδrms 2
<<= 1i =1∑ N ti
i =1

x2 2ηx 4
ε << 1∑ ti γ
i =1
2 t x24 = t 02 + 2 ‒ 2
i =1 x 2ηx Vnmo Vnmo ⎡⎣t 0 Vnmo + (1 + 2 η)x 2 ⎤⎦2 2
t x2 = tε02 ‒+δ 22 ‒ 2 4
which results in Equation 1. This treatment is suitable only η2 == 2t 2 + Vxnmo ‒ Vnmo ⎡⎣t 02Vnmo
twhere:
2 x
2η +δ(1 << + 21η)x 2 ⎤⎦
x
x 10+ 2Vδnmo 2
V 2
⎡ t 2 2
V + (1 + 2 η)x 2 ⎤⎦
nmo ⎣ 0 nmo
2 2
for small offsets. For a VTI medium, we need to use more t = t + vδi is<< the 1 interval n velocity; and
2
x 2 2 x 2 x 0 ε << 1 γ

2
δVis<<
2
t x2 = t 02 of
terms + the 2 t xseries = t 0 +to achieve 2an appropriate correction
nmo the 11 i-th layer vThomsen’s
nmo t i parameter. In a multi-layer with
Vnmo V 1995).2
x N ε 2 << 2 γ i = 1 ε ‒ δ
V = Vnmo = nthe value of η in
(Alkhalifah N
& Tsvankin t x2 nmo = t 02 + 2
V ∑ VTI
ε rms<< vi2εanisotropy,
t1i ‒ δ γ
∑ ti Equation 7 (Alkhalifah
= Equation 3 is denoted by
1+ 2 δ

N
∑ i =η
2
The Equation
v t 3 adds 2
vi ti N a third nmo term in Taylor’s series 2
Vrms = Neff ε ‒ δ 1 =and expressed as 1997):
i i
1 + 2 δ i =1 x2 n 2
Vexpansion:
rms
2
= i =1
N Vrms = N 2 i =1
∑ vi ti2 2 2 2 x x
2
t x =t x t = t 0 +2 2
2 2 η
v∑
= ti= v 1 + 2nδ
1 + 2i δ
t x
2
= t 0
2
+ 2 ∑ vnmo ti
0 + V2 nmo
∑ t i
t x2 ∑
Vrmst2i = xN
2 i =12
Vnmo V 2
2 nmox
inmo
=1
⎧2 n= i4=1v 2 2η
∑ n
i 2
v nmo i 4 t V

2
rms = V nmo = i =1
n
i =1 =i =1t 0 +
V ∑ 2
4 i xt 2 t 2
xN = Nt 0 + 2 2 V x =V
t x = t 0 + 2 2 1‒2⎪ ∑
2 2
rms nmo v∑ t nmo [1 t +
i
x8η
] ⎪2 ∑ i i i =1 ti
N
v 2
t ∑

n

t x = t 0 + t 2 2=ttx‒2 =+ t202x + 2 t xN2 ‒= t 02 +2 22 2 2∑


4 2 V 2 2
x2 2 2 x 2 22ηnmo x x
⎨Vnmo ⎣t 0∑
2 2 i nmo =12 2i 2 i
2 2 i = 21 2η x v
(3) tv
i ii i t nmo V V =
ηrmseff nmo= nmo V i =
= 1 ⎡ ntVx nmo t=i t+0 (12
+ + 22‒ηV1)x⎬2 = ⎤⎦ i =1
(7)
Vx nmo
x 2
0 V 2 ⎡V
Vnmo ⎣2 0nmo
t
t V
=
2
∑ t
nmo 2 Vinmo
+
+v (1 2x
t i ⎡trms
+ V V 2
⎣ ‒
η
2V
0 Vrms
nmo )= x 2 i=
nmo ⎦+N(1N

=1 i =1 N
+∑ 2
42
2 ηv)xt ⎤⎦ δ << 1
2η x 8 ⎪ V ∑ t ti= tt + nmo nmo = vi N 1t + 2 δ i
i4=12 n 2 Vx2 vrms
∑2 0i 2 ⎭ ∑ ⎪
2 2 2∑ ⎡⎣t=02∑ tVi i =2t1i + (1 + 2 η)x 2 ⎤ vnmo = vi ⎩t 12 += 2t 2δi =+i1 ix=1 N Vnmo
nmo
tδx2 =<< t 021+ 2 N Vrms =
nmo 2 x i =01 N2 i i x
i⎧
i
x2 2⎫ x2
Vnmo δ <<
∑=includes
1
vi ti ∑2
N Vnmo v V
2 t i i =1
∑ i t i
V
nmo rms i =1 i =1
nmo N ε << 1vnmo = vγi ⎧1n+ 2 δ i2 V2 2 ∑
⎦ x 0 v 2
t
=1 n
⎪∑ t
v 42 = t 2 +
x t [
0 1 + 8 η
2i ]
2ηx 4⎪⎬
t x = t 0
2
+ 2
2 2⎫ Vnmo Vnmo
i i 2 nmo i

N2 2 22 2 x x∑ i
2 1 xi =1
N
2 2
t x = t0 + 2
x ε2 << Equation 1N V 22 =εγ3i<< 1 1 δ << Vthree
γ
rms i1=1
= unknowns:
xt = +t + i‒=14 ‒
the
2 2 zero t offset2ηx2η 4 4
t x = τ +
ε ‒ δ s 1⎧ i 0=1 4 ν 2 2 2 i =1 ⎪ τ∑ 2 v 4x V
+ t [1
nmo i =1
=+ 8∑ η ]v tη
⎫ ⎪ = t= + ⎨ 2 ‒ ‒ 1

rms N i i eff

x 0
⎡⎣tt 02Vbehav- ⎪ + 2 η)x 2v⎤⎦2 t
[1nv+=i t8i η∑ ]1example
n nmo i i n
t2x tt=xthe N
8 ⎪Vnmo 2 N 2 N
vti0t, i the Nvelocity Vtnmo 2 2 2 η .2= 2 η We ⎪2⎨)⎤x∑ ⎤⎦ nmo∑ VV nmo ∑ ∑ nmo + (1

2
traveltime rms 2= , tand 0 γ+ i 2
0 new 0 2 parameter
=42)xηmade 2 v a simulation tVi rms t‒i 1⎪⎬ that explores 4
nmo the
2
i vi t i
Vnmo 2 ε ‒ i =1δ ∑ <<2 1 xx 2η x xV2nmo ⎡⎣t 0 V⎡⎣tnmo
2
ε 2t δ
‒ ε
t1x = t 0 + 32is reduced ‒ 2 nmo V V V V V 1 + + (1 +2 2η+ eff2x+η
(1
δ 1 ⎦
8⎨⎪Vi =12 =V i4=when
iN
⎩ i =i1=1 1 shows ⎭
i i
N V ηrms= =thatN if η =0,
∑ ⎬⎪<<is1present.
nmo

nmo nmo 0 nmo
Note = i =Equation i i = 1
t 2
to
2= tEquation
2
20 + ‒ 1. η ior of= t 0traveltime 1
t anisotropy
i =

t 1δ V 2Figure
= V 2
= i =1
1 2+ 2 δxt 1 + 22 δ ε2 ‒Vnmo δ δVxNnmo ⎡ 2

(1 +2η2xηV)x2 ⎦ ⎡t 2V 2τ0 +=(18+⎪2⎩ η2 )x 2 ⎤4 Ni =1 i =1x 2 ⎪⎭
∑ t 0 +∑ 12⎣t 0 V nmo +V
2 2 x 4 eff n i
vt i22 t=i The <<δ << 1the 2 rms
S∑ vnmo tVi computed
nmo n i
3 2as 2η
rms
2x
4 N rms N
parameter i n x η =can t x n=bet 02related
‒ 11 +2 2v2δ2 2 tVi2=nmo
+∑ 2η x‒4to
2 vi t i 2
well-known
nmo nmo ⎣ 0 nmo the traveltime

⎦nmot x∑ ∑ = ti i0 + by22Equation
2 tt 2
ε
t ⎭ ‒=<< τ 1
+ τ + γ
axfunction ∑ tof the ∑ ti
2 ∑ nmo
x it=12 2= t 2 + 2 ⎡ (Thomsen ⎤⎦ ==τ from i
0 v δ2t i<<
2
2 η=)xVoffset ( )
4

i =1 2 2 2
Vrms = N Thomsen’s Vxnmoanisotropic parameters ε 1<<ε , <<
1δ Vand nmo 1⎣tγ02V
1
<< γnmo + (1 V + rms 2
τ
2 i =1
S ‒ 1a 2 2i i==11 2 Vx
horizontal
x reflector V s2
at ⎡⎣at 0depth
2
V0 2η 2
ν +2x(1 + 2
of i =η
1.582 1 )x ⎤ 2
km⎦
i =1
V V2η V ⎡ =
rmst V n + (1 + 2 η)x ⎦ ⎤ t =2 τs + 2 τ20t + 222t x =x 2t 0 + V 2 ‒ Vε2 ‒2ηδ⎡tx24V 2 2 + (12+ 2 ηx)2x 2 ⎤ 2
nmo nmo nmo

4 nmo
N2 x =1 x⎣ 0 nmoN 2
nmo i s 0 n 2
tt2irms= =t 02V+Parameters ∑
2
x4

nmo
V 2
nmo 2=
i =21 nmo
x2η
<< ∑ 2 i= txν +
2
2Vrms ‒ n=2V ε nmo =1 i satisfy
2 δ2 << n 1γ the vfollowing
t2εi‒εtδi<< andnmoη =0.026.
1986).
i =1 x ∑ v
Vδ << 1∑ t Vnmo
and
ti 4V⎡⎣t 02εV‒nmo +2 (1t =+η2i ==η1ηi =)1x= ⎤⎦
εnmo ‒ δ1 conditions: γ t with
ν = =
τ V
SV
+ nmo
=3.893
t
τ + δ (km/s)<< 1 ‒ 2 τ0 =
= nmo t 0 ⎣ The 0 nmo t reflection
= t
1 +2 2 +δ (1 + 2 η)x 2 ⎤ V 2 V 2 ⎡t 2VV2 2 + (1V
+ before ‒⎦ t = t 2
+ ‒
∑ ν 2 δV<<
x 0 x 0 x 0
i =1 nmo
i i
rms = V
δ the
s
NMO t
rms
i =10
correction
2
is shownVnmo ⎡⎣in
t 02V light blue circles. ⎦ The trav- nmo ⎣ 0 nmo
V
x 22
= 2η x ε <<nmo1i 1 γ + 1 2 + δ 2 δ +µ δ i µ4 ε << 1
nmo 1 S nmo nmo
xηt2i = ε ‒ δn n 2ηxv4nmo = veltime γ nmo
Siτ0=1= t 0S422curve,
n
1i =1 η = γ
0 n
‒ 1,2 Nε t << 12++η2)xδt x22ε⎤=‒tδ02 +∑ ∑
rms
t x2 = t 02 + δ2 << τγs = τ 0 ( S is‒ 1in )
i =1
∑ = δ 4<<after 2
‒1 +22 δ v22 vt22 t 1 ε << NMO correction, δ blue v<<
nmo1 tusing theδ << 1
Vvnmo ⎡ + δ ∑ ∑ τ 1
2 2
= V v 1
i i =1 ⎣ 0 nmo t i V
v 2 = + v (1 + 1 2 δ ⎦ 2
⎡ η
=
0⎧ ⎤ 2 µ V ε ‒
⎫ δ i
η ( )
i = 1
ε << 1 η n=2 γε ‒ δ V 2
v0nmo tεi [ << +j 81η i ] 1ε+γ⎪‒2δδ ν = SVrms ε << 1
n
nmo nmo nmo i
= V t V + (1 + 2 ) x η V 2
= V 2
= i =1
=tV= V = = n n v 2 t 1 ⎪s ∑∑
rms
i i n
2 2 2nmo 2 2 nmoi =1⎣i 0=1 nmo
nmo nmo
i i
correct
τ = ⎦ τ
S 4 value
S ‒ 1 of =0.026.
= The curve in black represents the
vi1x +4⎫1∑ γ ε << 1
2 2
⎧x 4 v⎧nmon =2η 2+Vδ2rms vδVnmo n rms nmo n
γ
δ << 1 in2 which
t x = tε0 1‒+⎪δ∑ 2 both v 1 ‒ +
terms 2t δ [ 1 zero
+
2
⎪ ∑
8 η in
v ] 4 an
2 δ t <<[
isotropic
V22nmo ⎧= n n V2 ⎤2v 2 = tV 2 ∑=⎫∑
⎪ 1 i =+ 118 η ]n
irms i nmo⎫nmo
medium. ⎪ The value∑ t η
nmo i = τwrong
µ
ν ⎨
sj =
2= i =1τ (value
= SV S∆τ2 ‒ kV1 )kof η =0.5 = ‒ 1 ⎬and the curve in red ∑ t
represents
i the
tii =1 i
∑rmsnmorms n ε ‒ δ 1 + 2 δ µ µ i =1 ε ‒ δ ε ‒δ
iV
2 1rms 2=
0 n
x∑rms
nmo i
ηη eff=γ1=will
eff
S∑
nmo i i
⎡⎣t 0i n=V1xzero 8value ⎪ ⎪
nmottii i =[1 ηii ]nmoi =1⎪i =we
=21 24 η)paper, ∆τ
ε << 1 of +82⎨⎪δbe
Vinmo Vnmo 2 ‒+1 (1⎬ +this 24η
∑ v 242 t=i 44 η eff= is equal to η =
ηconsidered
t
eff 2=
= tn 2⎨
+ ∑ vnmo
nmo
2 ε⎪ t
1 24=⎪wave
in
∑<< n 2v1 ∑ ‒⎦γ+1nmo ⎬8because 1 n are

ν⎩2 = µSV of
V =0.
µ
=k tiNotice that,
1 + 2 δ ⎭
for
v
=example,
n nmo
µ
= v 1
V + 2 δ 1 + 2 δ 1 +2δ
x 2 80 ⎪
2∑
= = v=n v⎪1 +12+δ2in ti =ηeffµi =441 there ∑
1
=1V ⎨⎭iV
i
S ==1 +48because

2 2 i = 1 2
ε ‒ δ assuming a2 two-dimensional 4
i =1vpropagation ‒i 1δ⎬i the zx above Vrms =is Vonly =one nmo v 2 it
layer.
=VnVnmo ηtieffi2V rms i
δ << 1 ⎩Vrms = =n2rmsVxnmo =1nmo
vtnmo µ

nnmo
ε i δnmo
µ2 =2 Vrms 2 2V = V v 2= ti =1 ∑
n nmo i
η= i =1t⎩ ‒
∑ ⎭ti 2 n n

i i
vtnmo ti ⎧∆τkVvk4 t 1 + 8η ∑ v⎫nmo
nmo
∑ ∑
n i =1 j
nmo
8η21⎪,=+isi 2=1defined ⎪
µj i==1 ⎪ ∑ nmo2 i [ 2 i ] i =1 ⎪2
n
i nmo ∑
2 4 4 2 2
= t + S = 2
t v
1 + 2 δplane. γi =12 nmo ∑
A fourth parameter,
v iN
= v δ V 4 as t in Equation 4, x
2 << 1 2 i2 t i i =⎧ 1 i =⎧v 4 14+⎭2 δ t = τ + ⎫ ⎫τ µ+ω V 2 v j kx 2
[ ]
n
i vt [1 ]
x 0 n rms nmo
Vεrms 2⎩ 1 + 2 δ vi nmo 0 2∑
2 2 4 nmo i i

n
n= Vnmo 2= x n iv
V
=1i x tnmo =1⎪ ∑ ⎪ ∑
v t + i1
8 η
+ 8 η s
⎪ k ⎪ = ν ∆τ
21V ‒ V
rms
= V = i =1
∑ t1i∑ ∆τ
⎨ i =1 V = V = V ⎬ = V 2
= i =1
⎧ = vi4 1 + 2 δ1i n1i =⎨1 2i =⎧1 ⎫n vnmo = vi 1t+η2effδii==1 8 ⎪ ‒ 1n
i n
t =∑ τsε+v‒2 δτt 0 +t =V2τ2s + nmo nmo
i i i i
µ = ⎫v ∆τ V ω k k

=ν2 v∑
z
t=i +i =2τ10NδNv+nmo η efftxiη2[=eff1 +=8∑ ⎨η i v] t⎪ n4 n ‒8t1η⎬‒ 1j⎬ ∑
2 rms 2 nmo n k rms nmo rms nmo n

i ⎪v∑ ∑ ∑ ∑ ∑
j
[ ]
n
⎪ ti
2 η =i =21
Vrms = Vnmo = t x1 +=n 2t 0δ+ 2 η2 t =∑
nmo i
v (4)
2 nmo x i =i 1
rms 1
∑ 1 ν 2i2 tiv2nmo
τ + i =21τ 0 +⎧ n22 4 8 ⎪i =1 ⎪1‒V1⎬i4=V1 t⎫ tτ0 2= ⎪ j ⎪ Vnmo ω vnmo = vωi⎧i =k11nx + 2 δ i
8 ⎪ ∑ v t 1 + µ = ⎪ ∆τk k
k 2 i S = 1 + 8 ηeff V 4
∑ t
∑∑i=1 inx S⎭kSz=⎭‒=11⎬+∑
2 nmo i 0
t 0v 1 +V2 δ 2⎧Vt 0rmsneff ==4 i =s1tNi⎨Vrms =⎪Vνnmovn η=⎫efft⎩ =[1n⎩+ 8⎨ηnmo
4 nmo i i 2 2
⎩ nmo
⎫i =1 i i =⎭
2i ] i =1 2i ⎪
1 i=
vτnmo = =∑ t τ n= x 2∑ vnmo t i [1 + 8V
i =18 ⎪
1 ∑
η i ]i =1∑⎪ti t⎪ =
nmo
t +
8 η∆τ eff k 2 ‒ 2
v = v 1 + 2 δ⎪ ∑ v 4
t [
ω 1 + 8 η v]
v
2 2
k ⎪= v 1 + 2 δ v = v 1 + 2 δ

4 nmo i
4 t i⎪ x ‒ i ω ‒12⎧ 2x ⎫ i
τ s 2=ti τ20 S( S= 1‒+1⎪ω Vi =1 kη4kx=
nmo
8
1v⎪nmo Si =t12i ηx∑
2 nmo i i
Si t=1x =i Ntn0 +∑ nmo ∑
8vηeff nmoVnmo
i i

⎭) 0.75 v kxηη=⎧eff0.000 z t [1 +18‒ 2 ] x‒


0 2 2 n
teff20⎩ t=i ⎨nmo 0 4⎬
n 1 ⎪ ∑t =
2 2 V1 = ⎨ inmo 2 1⎬
0 nmo inmo i i
⎧2η eff =vi4V i =1⎬ 2ηix n=1 ⎩ ⎭ V vnmo n τ η+i ω ⎪ ⎧ n
i =1 n‒ ⎫8 ⎪ τ
2 2
2 2 2⎨ 2 ⎪ 2i =1nmox
⎪S∑
τ i+

1(nmo
‒=1vτ)nmo
i =1 Ns 8n⎪ 0 ( VV4nmo
=t1tnmo = t + =
[10+S8‒η1i ]S2) v 2 approximation ‒ 1
2 δ∑
x x 8 4⎪
v ⎫
]n1 ti ⎪ ων2‒ 1⎪⎭⎬⎪ 1 ⎪ ω∑2 kvnmo ⎧ ⎫
⎩⎪ti [1 + 8nmo 2 t i [1 + 8 ⎪ ] vnmo
n
τ+srms=2represents
δ=τ2i0V V+τnmo t =
+τt0i+anellipticity t +2 2 k2z = ⎨ i =1 V η i∑
2
x ti iτanisotropic ⎪V 2t 2=t⎡tτ=2sV 2s +τ 1t obs‒ ‒2 t app
∑ η i∑
4
vnmo = vi 1V
s 0
and the
i= 0 2
of 2
4+2
4 2
x⎩vi ⎣ 01⎪+nmo v= 1 ‒ v ωk2x1ηη=⎪eff×0.026
4 4
2 η2)xx ⎤ 0ν V =vnmo ⎪t
2 nmo ∑ rms ω
2 N 2
2= SVError
=i 1 xν ν x 2⎭
+i2η0 (1
η
introduced
⎧ νn eff =4 SV
2 V= rms =⎨ 2 by
i =1
N
∑ ν 2 ∑
Alkhalifah
= v ⎩t2 t2
SV =
iin⎫ = τ 2 tτ = + + and
τ
x‒nmo
τ 2t⎬
i =(210S ‒
1 i =
Tsvankin
+ ‒
nmo
1 ) ⎭
(1995). τ This param-
τ 2 ∑ ⎦ v i tnmo
i k z = η 1 t
eff ω = ⎨ i
100
= 1 8
k V 4 V
= t ∑
i =nmo
‒ t 1 ⎬ η ‒= ⎨ i = 1 x
η =
1
⎨ i = 1
‒ 1 ⎬
+x∑ ⎪ νVnmo 2 tnt = 2s +
+ (1V0+rms+2because
2
2[⎪
t28called 1t i] =ην)2xi =21SN⎤⎦⎫N= µ µ4vV ω⎩24kx22 nτ0tzx=2 Si =v1 ⎪Vnmo
0

1 ⎪∑
s
2 eff ⎭ 2
i x 0
i nmo ∑
⎧τ ⎡⎣=0xt 022tV04 nmo
s 0 2
rms 2
η=V δi4=1<<⎪rms 1ti2 2 V 2 obs
ω2 8 ⎪
nmo i
ω8 ⎪‒ 2Vnmo
eff n
8⎪ ⎪4
2vnmo 1= 8ianellipticity,
[ ] ∑ 4ηk
2
⎭ 1τ0 0⎪+0∑ 2Snmo i τx202∑ nmo ∑ i
eter t = is t V + and nmo 2 it is τ usually
2= v positive
t 1 + 8 η v
⎪=1t µ Δz Vf V Δ
42
t = V V t V
=40vi=⎩ntoften = τrms
nmo
x µ
i rms
1=µ2i+4=τ12+2δεi>tµ‒<< 1ν µ xt=t4 SV k = t nmo ‒

x
s+
22 t = τ2 + 2nmo +=1ηki 22
η eff = ⎨ iit vS=1nmo N
i =2⎬
i 2 i 4
ω ⎩ ω k ⎭ ⎩ ⎩ ⎭nm
4Ss∑
=is very V = iδτ0in 4τ1 =<<+
1= seismic 12 γ 2η
0 data. 4 ⎨ S ν t0
i
2
Vrms ‒=i1=⎬1 N kz = i 2 rms = v 1.25 V ω s ‒ 2 V 2 τ s x=
i ν x τ ( St obs ‒ 1‒) t app i = 1
2Vn xi =12 µ 2 0⎪ Vν 4µ η eff
i nmo
=x
i = 1 nmo ‒
8⎪ nmo
τ S0 (‒4SS1‒) 1with )
nmo
µ ⎪τ s τ=0 0(τV τ τ
2
Srms 4 =2t⎫µof = f t
τs + ∑ 2 ∑
24
tVx nmo ⎧0 +t2rms x 4τ8 s =
n
⎪∑
j 2 t
η =ω
= 2 + 2 0 η
2 + Error
k 2 =2
0
× 100
2= N tcan 42 ‒2 εthree
⎤⎦2 ∑ t2i these ti ∆τkVkv t obs 0V s‒ 2V
xτ t4t+appτ τ=2 t+0 x nmo ν 2νx= SVrms
i ε[τV 8τδ1η⎡µτti ]02types
0.500 2 2
⎩ t = We =1τ∑ i 2define
2 1⎭<< +Snmo2‒= wavefronts t obst = τ + τ 2 + xt = τ + τ 2 + x
∑ i⎪
∆τ 0v
V
+ V vttjnmo ν η 2xt=
∑12 =1τxi0tk=01=2 µk+ = ∑1s‒+∆τ2kδV0 k2 ‒ 1S⎬ rmsν ν= SV = ⎣ ( 2jV
S= ‒ V⎪
1
0γ 4
)
+ 2η (1 x + ⎩4 2 η)x 2nmo 2 µ⎭jxi=
= τrms0 (i =S1t‒x 1=) t 0 + 2 ‒2∑ 2∆τk⎡ 2 t obs2 t‒ t app 4 tS0 2ν⎤2
Δ t = t 2η
= ‒‒nmo
Time (s)

Error = 1.5 × 100


parameters:
2 i i 0 nmo 2 2
τ s SV
= 2 = 1 nmo app s 0 0 0 s 0 ν2 s 0 ν2
tηµV 2 2==22t xi =⎨
jeff=xt= + 2 n2 ε 2‒ δ2 2 2 j V Vnmo ⎣=t 0 Vnmo +obs (1x+τ0×2=100 η)x ⎦ Δ f µ4 Δ tµnmo

rms
2x rmsδ t 0 << 1
8 ⎪ ∆τ
N2 Sj V4nmo η ν∑ = V
= SV ∑ ⎡
τ ⎣=aτcircular t ∆τ V ⎪ k (2Sk ‒ 1µ
V + (1 )wavefront,
+ 2 η ) 2x ⎤
2
2 nmox Error
= t 0S += 1 +2 8 ηΔ

t τ = τS0 ( S2‒ 1S)2 = 22 = x 4
S∑
t = τs + τ■ ■ isotropic Vnmot medium ∆τwith µ x µ4 ⎦µrms4 int x which = 4

⎩∑
0τ0+= 2
n 2
‒efff 2 Δ t t
+ t=S τ=24ν+=4= =SV
2
t=
1=1 21S = 1(2γ+i8=1ηeff)µ24ν 2 ∑
2 2
τ s =i τV0nmo ‒ 1µt1ij +=k2srms
nmo 0 nmo
δt∑
0 k
S = 1εν+=S8<< =⎭0vτnmo sS t 0τ2tV
⎡⎣1.75 2 0 s
+ 2 η)x2 ‒⎤⎦1t x f=µt 02 + V t 0rms τ0 = 0 τ0 = 0
2 i 02 2ν 2 4 4 δ << 1
obs
δNη=0; Vnmo V nmoS+ τ(1 2 = τ (S
<<
ieff
V2rms2 S==Vnmo2 = = SV
µ ∆τ µ µ V V ω Δff =v kΔx ttnmo
nmo= 0nmo
) 2 S S
τ 0∑
n k
2 2 µrms rms µ
2 2
ν s = SV Vtnmo ‒j t

i = 1

2i =(ε )
ν‒v‒2xiδ1=2twith Δ t nmo
4 rms
4∑
0
t τ s22■=medium i2SV
2 =∆τ V
S v t << ( 4) rms µ4µj = N 2 s 0 ( ) s 2 0 ( S ‒ 1)
4n 4 4 rms
τ0 = 0 ■ 2 rmselliptical S = 1 + 2η
8 η x
anisotropy, S in = = ε δ <<1 k = γ 1 ‒
4=i =1t i µ4 ∑
t
∑ which = ; and 1 τ τ τ τ τ τ
0
x
1v k‒x ω2γ µ2 v2 kτVµrms = S ‒ 1 app 0
= S ‒ 1 =
0∑
η=τ=sω t=+0ε +12N<< 2
‒ t00 ν =µSV
nmo i j j 2 2 k k
S V tkt= τ‒02 2+
2 2 eff 0 ∆τµk∆τ 2 V 4
V
2 kk V
z
v Δf t nmoω=2 tt0app

s

x2 =
ν = SV1rmsS+ 2ε=tnmo
rms
■ medium 1 V k µ
with ν =2 Vnon-elliptical
δ‒ωzδ42 =Vnmo 4 ⎣t 0∑
2µ = 2VS 2= x=
1 ⎡ ‒ Vnmo 2 + i(1 = µ
anisotropy.
=1 +22 j η µ
= 42 )x ⎦
= 2

k rms
ε ‒δ S 2= = 4 ∆τvi kti 2 2x
τ s = τ 0 ( S ‒ 1) vη ∑ µkV22 k vSVktrms ∑ ∑ =kεj << 1 V Δ γ t nmo =2 t app ν ‒=tSV ν = SVtrms x = t 0 +ν 2= SVrms
z rms nmo j 2 2 2 2

n
ω∆τ
j
µ224 Vµrms
=
µt40 1 i =11µ+4µ222vδnmo V=vjrms2vzi t 1∑
i 2 vµ 4= ω
k = 2 22τ
+ 2s δ=i τi0=1(2S ‒j µ1j )=
1 ‒ ∑ x ∆τ k∆τ ∆τ kηkV ω 2.25 0ωrms
2 2µ
k
S = 2 = 4Srms V
4
2
=
= 1 + 8 ηNeff
i =1
V
S1=2k+k182+η8eff ηeff∑ ∆τ η1=+ 2ε kδ‒z δ= nv 2 V22 ‒xω
n

i nmo x
τδ==<< ∑ 2 2µ
4 2V 2µ µ∑ ti 2S =2 µ4 = µ4 2 NS =2 µ4 x= µ4
nmo
2 S V2nmo= 24ω3 can Vnmo nmojωi ω kvx ∆τ 4k ∆τ Sω=kω V 42ηk∆τ VkV j
∑ 2
ν 2 = SVrms k2z0 =VEquation µ S
2 x 2 k ∑ = ‒
∆τv
be
V n ⎧generalized
v µ n 2η =
1 ν x
+ ‒
2
=
2 δ SV
[nmo
for 2 a x case with multiple

]ω1ηk+ωx⎪8vk2ηkxeffvx22 k2tox2 RMS
k
δ ∑ t = t nmo + S = ‒
2µ =4
η == 0.026
nmo x
rmsk
ω t
rms ∑ iµV
= t +
∑ V2.5 µ2 Vrms∑ v k
i = 1 2 2
tηnmo
2
t xε =<< = V = = 1 + 2 v t v t

2
1 ‒2 2x µ2 V
‒γ 2ω Snmo n= 1 ‒v2⎪+nmo 8V
t2 1+ µj nmo v∑ ∑
2 4x
=the 2 k 2 k j 2 eff ηk 2i rms 2
2
txiVω ηSnmo 2 2 x 0 i = 1 0
V4
µτ s4 = τ00vwith 2V
4 2
rms z
2V
2 iV vnmoω 1∆τ ‒+η2k)8isVxω
2 j
∆τ∆τ kV
j 2
( ) ∑ i= kkkz =
rms
µ4 layers, nmo condition ⎡ that 2equivalent

nmo i
i =1 nt obs ‒ t app nmo
i2
nmo rms
δ=kVVk t2η eff=∑
S ‒ 1
S= 2 = 4 ∑
V j V ∆τ k 1 t
t V
= 2nmo inmo + (1 + 2 ‒ 2 µ = x 2
2η i =1 4
x
‒2∆τ ==i 1=k1⎣tzni ⎨ωS⎧=tvS1 +=vv842µV 2z2 k
ηkk4efftnnmo = µ= ⎦21 ‒1 ‒ ⎫22Vrms = Vnmo = = n v 2 tN × 100 ∆τ
∑ x = t 00 + 22 ‒ 4 2∑
v2 ω Vrms V= j N N ∆τ V j
∑v 2t2k k
=1z +ω84 vη‒ 12⎬⎪Vωnmo ∑ µvj i2=ti S = 1k +k8∑
εVrms nmo
Error
4nmo2xi [ v 4 i ]ω
nmo 0 i nmo
= 1 n j
-8 -6V η-4eff∆τkt-2 6 ∆τ
⎪ωηvk2x 2kx2
j 2
µ2 Vvelocity µηrmsj2 = 2t obs ‒ nmo
(Equation app i5): ‒⎪ 1∑
2
δ =<<
νError = SV 1
=∑ S = 1 + 8 η ×k 100 8

= ⎪ ‒ µ ∑ ∑ t tnmo ∑ i

∑ V µ2 jV =
nmo ⎣t 0ω
⎡ 2V2nmo
k 82 k
2 +∑ µ(1j =+t∑ i 2 η)i x i ⎤
‒ ∑ i =1 i =1 ∑ ∆τk
obs app
1 + 2 ∆τ
δ Error eff = 1t V × 100 t V k = 1 ‒ V = V 2
= i = 1 obs
V nmoω k ⎦
vj nmo = vi t obs1 +n 2ηδeffi i==2⎩1vt2obs⎨ tωobsω‒i =1t appnvV Vj nmov⎭‒ω12⎬ω 2 ωrms2 ωΔ
rms i
z
S =∆τ 1 +k 8 ηeff k2 =2 x (km) ∆τV
∑ ∆τSkεV=k<<
k i i = 1 nmo 2 i rms
8xk⎪z == V∑
2 z2
knmo ω
2 nmo 2
kfx2 kx V 2 i =1 n 2
Δrmst nmo = i =1N ω vδz k<<
nmo k x 2
= N
+ 2=2δ2 i 2∑
1 µ+418 ηeff µ4 γω 2 Error × x100 ⎪ t
⎫z ∑
2rms
µj = v k k14 ‒∆τ k
= =V ‒ ‒ x 1
v V 2
ω 2
‒ 2 V η k
∑f11v4+1nmo ∑ S =t 1 + 8kηz eff= 1‒ 2 2 2 S = 1 + 8 η2eff nmo S =xx1 + 8tηeff
t 02 + 2 ∑
2
z kt i2 k 2 2 2 v 2
∑ ∆τSVΔk=2f µε==22v‒Vnmo=Δδk2Vtz=nmorms⎧⎪=4=vv∑ δt=ii1tV ⎩2 ηµ]xj v2= ⎭ nmoω2=2ω‒vi 2‒V1fnmo
i
Δ ‒2Δ ωt2nmo obs i =1ω v ωvV kVx nmo kxηxkti0=x1 1.
ηkFigure nmo
[ vωtime vω
n 2
V 22Vω kx t x =offset ‒i 2
∑ nmo nmoω
i
2i vi =2v = kτ t ω 2+nmo
nmo
τ 8 ⎪
(5) ‒ ∆τ nmo
i =1 Reflection k = 1 ‒ as << a function of the
η = ω 1 k t = nmo z +i
s Δf v + i
Δ νV‒nmo ω k = η ‒ ⎧ + 2t tδapp ⎫ ε 1 t ‒ γ t 2V i =1
ωV ⎡2t 02V
[ ] 2 ω ω1.582
2 2 2 n 2 2 2 2
S = 1 + 8 ηeff kz =η eff =∆τ f t V t k
‒ 2 V k vnmo = ωvΔi t nmo 124 V=from z
vreflector
kx ω ω42 at =a depth v km kx 2 nmo v⎣ k
1∑
2
⎨ j⎧f2 xnmo ⎪ ∑‒
2⎬ ⎫ω
0
rms nmo
1δ‒0kV x n t 2 2 z
t obs tnmo obstω
2
txV
kapp 2
η ‒x8at 0ηhorizontal Error
obs
ω
app
of × 100 nmo

i [k 1 +=8η ]ωError
i = 1 n
1 ‒
v ‒ 2 2
v 2 i1 k
+ V 2
k 2
V∑ ∑ x 28 η x ⎪ 2η x x
4 02 = S = nmo x
+ 2 1 + k = 1 ‒ k = 1 ‒ k = 1 ‒
t v=t4=tnmoi ω ε ‒ δ t obs
2
τtstf+
nmo app
µΔj t=2nmov2 = kt8app⎪=1‒ω Δ⎪tnt0nmo ⎪kxError ⎪eff 2 = ‒= 2 ×1100 nmo
t ‒‒z τtt0 +t 0ivνt2app
nmo x
k n 2
×2 100 nmo i
2
t x = t 0V+ =3.893 2 i
‒⎬ k⎫zkm/s = 2 2 ‒ 2 v δ << tω 2 2
x1= t 0 +v 2 ‒ ω 2
2 0∑
z z 2 z
2 Vω
2
v kx η∑ i=V Vnmo 2 ω 2ηVeffnmo =ηkx⎨ i =⎧1 n vwith n [1 nmo
2‒ 1v 2 and η2=0.026.
ω = + 2ωη)x‒The ω
22Vkreflection prior
2 2i 20obs2 ‒
]
2 22
t obs‒t obs x ηkx
∑ to ⎡ v0nmo ⎤
2 V
k ⎨
ω 2⎬ t V V
2τv0v=
=11nmo
∆τ ‒ η V t V + (1 V
+8∑
z= inmo 4
⎩ Error V
app
= ω ‒ 2 ‒

V 1 × η100 k 8 ⎪×1100 ⎪ t + 8 k ⎪ = ⎣ 2 is shown 1 ‒ + 2 δ ⎦ 2 blue circles.
V⎪‒nmo
ω tiΔi4nmo =ω ω nmo vx kx = 2
Δ2 fω 2 k 2Δ
nmo nmo
∑ t light
obs app nmo nmo
kz = 1 ‒v 2V=nmo eff 1 nk 4nmo the
ti NMO i correction nmo nmo
ωx2in‒nmo
nmo
= ‒tΔtf0⎪1Δ‒ftError
i
⎪in
=1δ ‒ t 0t ∑ =k ttapp
η eff =⎩ ⎨ i =V1 nmo Vnmoε << ω12 Vγnmo ω 2 kx2ω 2
nmoS
⎬ ω
z
v V Skz=nmo 2ω + 8vηi2eff 1=as
=1Where, i2 iEquation
2V
nmo xt6:
2i z obs obs ‒ app 2 nmo nmo t obs
Δ tΔ t δ << k = ⎭ 1time v t V‒nmo
‒ t app =the 2Vnmo
k n ηk
= δ <<
k ‒ =
rms = V τ 1 1
τ20224τ=+k2s 2=tobs2Δτ 0f (SSapp 1)t ηk=⎫x v ηkx ∑ vnmo tvi Vεnmo
⎩ n xω t = 2 nmo t ⎭ = ω × 100
= The reflection curve after NMO correction
2 x is
0 ‒‒ 2iError =2 ω 2 f‒ 2V 2 t0
2 obs
v ⎧ nVnmo V=‒1nmo 8⎪ i = 1 n
⎪ Vnmo
vError ×nmo100
∑ ‒ δ ω 2 ‒v 2Vnmo
nmo z z z
Δ
2 2 2
V ηkx
Vnmo ωt2=ωτs + ∑
4
Vnmo ti in blue
Error
⎪ ∑ vω v k 2xtν [
i1+ 28 2=η i ]
× 100 nmo

f t obsf Δ f2t 0 t 0Δ t 2 2 ⎩ 2presented
x ε << 1 γ ⎭ usingt obsthe tV ‒ t2app
Δ tcorrect
= V = 2 value
t = ‒ i =1 of η =0.026.
t = ε << 1 γ
nmo

kνx2 0 (Δ fkt‒0z =1=)⎬ ΔΔt nmo


kz = kz =2 t‒1=1t⎨τobs ‒ ‒ t ν
nmo x
τ=
i
SV
= xt τ S ‒ 1 V ω = t
nmo
= ω τ k + τ 2
+ i = 1 Error = obs
rms nmo nmoapp × 100 1 + 2 δ
2 i =1 2 2 s f rms Δ t nmo t obs ‒ trepresents 0n
t t app t obs ‒ t app
=f 2t =app‒t app ‒ t2t0‒ t 0 s 2 0 ν 2The ti = obs ‒ of
i = 1 obs
2Vτnmo
nmo
t nmo x curve in black the wrong value

app
ηVeffnmo=v= ω appt t obs
2
v Error t s ‒+ω 0+ ×ηn100 2 ε ‒ δError =Δ f Δ × 100 Error Error ε ×‒100 δ=n 2
τ0= =v8i ⎪0Δ1 +ft obs2=Vδ i4Δνt2µ v Vnmo ω 0‒ 2Vnmo ηkx 2 ηx=0.5, nmo
vnmo µ ⎪ (6) =
= t∑2 tnmo
Δ t = t ‒ t = η = ∑
2
nmo
4= SV f app 0t t‒=tt0 τs + τ 0 + 2 and the curve in
Δf f t obs Δtt0nmo red represents the value of t obs =0. vnmottobsi
k4x2 ⎫⎭ 0 t Δ‒ttnmo = t app
4 i = 1
t obs ‒ t app V Δ⎧S⎩tfnt 0ω 2Snmo =iω 2rms ν 1+ 2 δ = 21 + 2 δ
Error = kΔz f=τ =τnmo nmo 4 ‒ µ 2 µ VΔ
τ0 = 0 + 2Δδ if VΔrms
2 1t)i [12+ 84η i ] 2Error
2
× 100 0i =1
µ v = v 1 = V = i =1

t obs s =1v0τ⎪0Δ
= ∑ (=St1VSnmov‒nmo
nmox= tω
2S = ‒
rms
= nmonmo
t 4⎪
j 4ηkx
= t=app 0
obst
‒ app
× 100 S t Δ f
∑ = vΔnmo
n Δ t
f nmo = ttapp
2 t nmo 0 ‒ t0
nmo i t nmoΔ f
= ∑
Δn t nmo
n


nmo
‒2t20V =

2
f t t t t vnmo
ηt eff=ν=τ2s + ⎨ τ0 2+ app ∆τ
µ V ⎬ τ = (
0
149 ) ⎧ ⎫
Vrms = Vnmo = i =1 n nmo app 0 1 ⎪ ∑ vnmo ti [1 + 8η iV ] rms ⎪= Vnmo i==1 i =1 n
k V
n
i 2
0 (S
k‒rms 1 τ = τ S ‒ 1 f Δ t i
= t ‒ t f t f t
ν4‒j2 =1)n
i
µ obs 4
Δ t nmo τ8=s ⎪=tt obs
Δf Δ t nmo = SVτrms 2 0
0 S 2 2 0 0 2 2 0
0nmo ∑ ∑
app ‒V
0‒ ttapp
ti∑∆τ ⎪fj Brazilian Δ t nmo Journal2of Geology,
s
= ∆τ k kΔ V 48(1): 147-159, Δ t March = t 2018 ‒ t i =1 Δ t = t ‒‒t 01Δ⎬ t nmo + = t app ‒ tt0
0 ( S ‒ 1)

Error = × 100 ν ∑ η nmo = vi 1 2 δ i ∑
2
τ s = τrms = v
tν0 ⎩4= =SV

t obsµ 4 µij=1=
⎭ = = SV ti
nmo app 0 nmo app
f t0 2 k eff n
⎪ ⎪
η ∑ 8 i
τ = S
Δ t nmo = t appΔ0‒f t 0 SµΔ22 tµ2 Vxrms
= rms S42
= 1 + 8 eff ∆τ f t 0 2µ µ
i =1 Vnmo ∑ ti
4
⎧ i =1
S =ν 42= =SVrms44 vnmo = vi 1 + 2 δ i ⎭ v 4 +t [1 + 8η
2 n
k

1 ⎪=∑
µ
t = τs=+S = τ 0nmo + = 2 jS44 =ω1 + 8 ηeffv 2Δktx2nmo = t app ‒ t 0 vnmo vi nmo 1 i 2δi
4 i =1
τfs = τ 0 (∑ Stµ‒∆τ 21 ν
) V
k kV = 1 ‒ µ 2 µ V rms µ x 2 η = ⎨ i =1
µ = 0 2 z k rms
2 4 4
⎧ n ⎫ 2 eff ⎧ n n
V rms 2= =ε2V‒nmo δN 2=2 Vn nmo
= i =1 n ⎡⎣t 0 Vnmo + (1 + 2 η)x ⎤⎦
VV
∑vt =V1Vnmo
2 2 2 2 2
N
ηvnmo 2=Vrms

rmsnmo
i i
∑ Ntδ2i =<<
v t =
i1 2+ 2 δ
v x t+2nmo 2 δ
∑ t
i =1 ∑ i
2 i i i
V = i =1N t
= i =1N∑ x v tt01+⎧ n2
2 i =1 i
2
rms
Vrms ⎫4
∑ i nmo i η i ]2ηx⎪
[
i i V 2 4n
ti V rms ∑xnmo
2
= i =12
v
t ε t =<< =t 2
1 v + ⎪ ∑ x
1
nmov
+ 2∑
γ
nmo
‒ δ tv 2 1 i =1
+
t 8
2 i =1 22δ i ⎡ 2 2 ‒ 1⎬
i =1
i =η
N2i v 0 Ni =
nmo ⎨2Vv2i=
i =1 1 +i
eff4 t== Vnmo
V1∑ ∑ n ⎣t 0 Vnmo + (1 + 2 η)x ⎤
2
⎧δ vnmo ni t i V ⎦
x 2 processing of 2ηlarge xirms 22 iε 8‒ ⎪ ⎪ ⎫
t[1ixt+ti 81η+i ]8η ⎭⎪ ⎫
n nmo
i 4∑
Seismic offset data
=‒+11i 2=⎩1⎪Nδ∑ ⎧vVnmo ∑
4n 4
t x = t 0 + 2 ‒ 2 2 2 2 2 Vηxδrms
4
==
[ ]
2 2
t2η

1
t =
⎡ t +
Vnmo Vnmox ⎣t20 V0 nmo2 η+2(1x+=2 η⎨2)xi1=1⎦2 i2=1 2ηx << 1 2 ⎪ nmo
⎤ v ⎪
+n (1 + 2 η)‒x12⎬⎤⎦‒ 1⎬
2 i = 1 4
‒∑
nmo i i
t x = t 0 V+vεnmo eff η Vnmo
<< 8 =⎪ 12t+⎡⎣i⎨t 02γiV=nδ12nmo ⎪2
δ << 1
δt x2<< = t102 +tV=
nmoV
x nmo 2 2=eff1
τ + i ⎩2Vτnmo
v i = 1 82 +⎪ ∑
0 2 i⎣
V x
⎡t 0nmo
ν
2 4 22
ivnmo
V ∑
Vnmo 4t i+
=1 ∑ i
ti (1 n
t+ 2 η)4⎭x ⎤⎦⎪
2 s
2 =V⎧ = ε ‒ δ x ⎩ =12 i nmo
2η x ⎫ ⎭
2 vnmo2t i 2[1 + 8 ]
n
Assuming a constant value of ε =0.2 << 1 and varying γ δ << 1where tηV2rms== t 2 +nmo∑ 4‒ n i =1
η
ε << 1 N x nmoγ1t 00+12⎪V =21+ ∑
⎪ + 2 η)x 2 ⎤
xti2 ⎡⎣t 0 Vnmo9:
x Vnmo
+γ τ⎨δτinmo
2 i 2
τtη0=eff= is2tτ=expressed + (1⎬ ⎦
in the range from -0.2 to 0.2 in Equation ε ‒ δ4, the values 2as Equation ‒ 1
η= ε ε ‒∑
<< 1
δ vi tSi 8 ⎪s + ν0 +4 2ν 2 s= 0 τ n i
2 = 1 n

will fall within the range from 0.5 to 10.+ The 2 δ value ηV =0= 2 = i =v1 = v 1 + 2∑ δ << 1 v ∑iti t
ε ‒τNsnmo δ2= τt00 (iS⎩2 ‒ 1)V δnmo nmo

ηrms1=+ 2 δV
i
corresponds to the conventional or isotropic NMO 2 cor- τε0rms= =n 1Vnmo t 0 n2= γi =1 n i =1 (9)
∑ <<
n
∑ v t 1 + 2 δ t τ
i 0S 2 x = ⎧
∑ ∑ [ ] ⎫
τ∑
rection. This NMO correction is accurate 2 only for the
nmo i ν t
2
21 = SV
= t 2v
+
2
⎪ S t v 4 2
x t t 1 + 8 η ⎪
2
Vrms = Vnmo = i =1 n 2 2 tη
i =
τx= ε0+1‒n(nmo rms
δτ‒20=2i1+ 2
1t) ν i2=1 2ηx 4
nmo i i i
condition x/z≤1.5 as shown in Figure 1. These dataVare 2 = V2nmo and ==2τi=τs=10is
effx µ
s=
∑=n the ⎨SτVviµ (4intersection
S i‒ 1) n time‒ of 1⎬ the Dix equation hyperbola

nmo
2δ1 + 2 δ4
0nmo
2 21 + 2
rms
ti tVx 2= t=0 V +Svnmo 2=
s
4
‒ 8 == 1⎪ ⎪2
from Thomsen (1986) and are specific to the Mesaverde rms ν nmo µ ∑
asymptotes
τV0 nmo
= =
=ν20i∑
= t viN
22 i t 2 422 V2 i 2
SV
=V ⎩SV i v with
nV ⎡ n nmothe
rmst i t 0 Vnmo
ntirms⎣
∑+time t i(1 + 2 axis ⎤
η)x⎭(Equation 10):
⎫⎦
i =1 rms
2 µS i =∑ 1 µi j∑
=1⎧nmo 2i =1
4 vnmo t i
sandstone. vnmo = vi 1 + 2 δ i δ =<<vi1 V V +2=2=∑ = ∆τµ⎪i2N=4∑ 1kVkvnmo t [1 + 8η i ] ⎪
δ2V ==141)iµ
4 4
vnmo µSτ1jrms = 1i =nmo x=124n i
0 (∑
rms
=Sτµ=τ2= ⎨S2τVi‒=t2rms ‒ 1⎬ (10)
η i=]1⎧vi n2⎪1 4 γ2∑
x2 ⎧ t η = ⎫ + +
8δµ⎪i2i=2∆τ
+ V∑
ti [1 +εvnmo
n seff 4
∑ 8<< 1 k j νrms
2 2
t x = t0 + 2 vnmo4 s 0 i 2 tn
i ⎫ ⎪
Vnmo
η

1 i =1
⎨ ε
1 ⎪‒‒S∑ ν⎧= 1v=nnmo
µ δ ⎬ +t∑ SV 84tηirms ∆τ
⎩ ∑[1xk+V2∆τ8k VηVnmo ii]=1j ∑⎪i ⎫
4
t 4⎭

µ2[+14k 22+‒δ822iη i‒2] 1⎬0 2⎪ 2


TRAVELTIME CALCULATION = 1 = 2η x

eff
j In Equation 10, t is the zero-offset double vertical time
η == ⎨1vtτx⎪nmo =tjv04∑
i = 1
N eff
8⎪ 4 η eff
n i 2=
0⎪
1
=µµ 0
2
v=+i tn∆τ i1
k k


THROUGH vi2 ti CASTLE’S METHOD ⎩
Vnmo ∑ ηti 18=+⎪2and S ⎨δ=⎭i =V
τ

v S2is
nmo
4 =an
∑V∑
τ
nmo v ∆τ
auxiliary
t k x kVnmo ‒ ⎡⎣t10⎬Vnmo +expressed
⎪variable (1 + 2 η)x 2as⎤⎦ Equation 11:
k S = 1 µ +
⎩8τz⎪s =Sτs =20 1(+S4ieff=281‒0rms 8 η ⎧
1 ‒ 2 4 x

veffi)jnmo
i =1 eff t = + V n n+

t4ν2i 2ti [1 + 8⎪η i ] ⎪
2 i =1 nmo i
Vrms = N
∑ ∑ η1tω
n
δ << v∑ 1 Vnmo ⎪v
∑ ην2 2 ====Vxt∑
ti x 2
⎩ i21=1∆τ ⎭
⎨2 i k=V
nmo
ω0 rms 2 2 ‒ 1⎬ (11)
i1=1k2 2
Castle (1994) proposed an approach t = τthat τ 02 + 2 three
requires v 2k 2 2
s+ τkµτε0z20jeff+== <<SV
2
ν tV = rmsτs += Vknmo 128 ⎪n21ω‒ ωγ 2xv4‒ knx ω kx ⎪

i =1
k 2 nmo ∑
nmo
k ν = ‒
t = τs + S =τ 02µ+εzvS4v‒∑
x ∆τ 1 V t
⎩ tµiV4ω
parameters: x 2
2ηx 4 z
nmo ωω i =1 ‒ 2Vnmo ηk
2 2 i 2
⎭x
t x2 ■=zero-offset
t 02 + 2 ‒source-reflector-receiver t0 ν=iη=S2δveff1 ‒S 41factor
■ Vnmo Vnmo ⎡⎣t 0 Vnmo + (1 + 2 η)0x ⎤⎦S
2 2 2 τ =2traveltime t 0
; t Sη
where
τ0v = =t v k 1=+12+δ2tVδobsnmo
0 τ =
s µV == 1 +τ 28
(
the
0nmo Vrms ω )
‒ t app2ω‒ 2
2
2 2
is expressed
ω 2 2
k
ω
x 2 2
kx
as Equation 12:
z2 k = i 2 2 2j2x
2
■■ NMO velocity (Vnmo)
τ0 = tν=n=τsz∑
nmoS 0 iError ‒‒ 22V ηk2
τ s = τ 0 ( S ‒ 1) ω
SV+=v ∆ττ 0VV+vnmokn x2 2×ω100 2
δ ■ << 1 1 ‒vkt obsk∑ νVvnmo ω ‒nmo 2Vnmox ηkx
■ τ s = τ 0 (SSkµ ⎧‒zj 1= =) 4 rms 2 nmo t i ⎫
2
weighted moment (µ).
ε << 1 ν = SVrms
2 2 ⎪ ∑
(⎨ΔτSrms=if==‒1µ=1=20)V=∑
2 µ vtv4nmo t t
=nmoit∆τ
Δ
[
µ 1 ‒ +ω t 8 η ] ⎪
γ ν 2ητ= SVτ 02V
2
s =
1 SError obs=4kt app i t
2obs ‒
i =
2 n ×
1
(12)
eff = rmsS0=21Error V 2η V
nmo
=nrms xt4ω app
‒100 1×⎬ω100 2 2
kx
obs t∑
t = + S t8 + t
ε ‒ δ 1 describes a symmetric hyperbola
Equation µ4 µwith respect 8 ⎪ f 2 t ⎪

nmo
ν µ= SVrms
2
k 2
=
µz 4= τVvnmo
x 0 4 eff
tj2i iobs ‒ i

s t4f ∑
η= S = 2 = 44 St tV
0 2
0=(Δ )2 2=1 ω ‒ ⎭ 2Vnmo ηkx
2 2
to the 1 +time2 δ axis, whose asymptotes intersect µ2 at theVrmscoordinate S = 42 µ= µ ⎩ΔτΔ ω ∆τ ‒iknmoVV
=11
nmo
‒knmo
v tk0x
=µ Δ = f Δ t
S µ=2 42 vV vf =x∑
‒+ 22tnmo δ2 i µ , which is already called time-weighted
nmo app
k=nmo j == 4 v 21 1
N
system origin(x=0, n
2 t=0). Figure 2 illustrates ∆τ
∑ nmocurve
v t ∑ thisV geometry,
j
t=∑ µ
τ +2∆τSkV
νz2rms
Error
f 2=The
V
τ=01k2+
4SV
rms ⎧
j ∑ =tt∆τ
i2value
rmsobs 0 v k‒ ω
it iapp j
of
× 100 ⎫
µj = k k
ΔV 2t 0 t [1 +distribution, 8ωη2i ]kx2 ⎪ is expressed by Equation 13:
n
in2which2 the solid i
represents the actual time curve µj of = s moment +=8νj =2ηiof =1
the 0
‒v 4velocity
Vrms = Vnmo = i =1 n ∑ ∆τk ∑
trms µΔ V t ⎪ ∑
teffapp µ = t4ω t ‒ t
µj =∑
nmo N obs
S ∆τ
kηΔz kf= =2 Δ⎨∑
= V 4
1
nmo = =14t i2 2 ‒
nmo i
t 0 ∆τ
nmo app 0
a reflector and the dashed curve represents the time com- 2‒ 1⎬

k k
ti eff µω 2v V
tii=nmo vnmo kx ωn 2 ‒ 2Vnmo
S =τ01=+ 8 η∑
S = 1 + 8 ηeff ∆τ = 8 ⎪ 1V
rms
2
η⎪kx
j 2nmo ∑ t i
puted using Equation 3. Asymptotes of the Dix or NMO k = 1 ‒ 2V
4
S eff zf 2 ∑
i =1 k
v ∆τ
⎩ tt 0kVxω 2η 4
⎭x (13)
equations
vnmo = vi 1are + 2represented
δi by dashed straight ω curves. 2 2
v kx S = 1 + 8µηteff=2= 2t +obs ‒k t app‒ i =1
s ω 0 (Error Δ vt k)x=∑
2
kz = 1‒ 2 τ = τ S j ‒x 1 = t app
0 ‒ 2t 02 V× 2100⎡ω + (1 + 2 η)x 2 ⎤⎦
2
A more kz =2 1k‒2 =nmoV 2nmo2 ∆τtknmo
Vω nmo ⎣t 0 V
22 2
knmo
⎧ n exact mathematical ⎫ expression v for the ω NMO ω v2 k 2 obs x ‒
x
ti [1 + 8η i by ] Malovichko ω τ τ
1 ⎪∑
4
vnmoprovided ⎪ νk ==2SV2 rms
v t z=
S 1=‒1 +s 8v21ηΔefft0 Vnmo + x + 2
ν 2 ω ‒ 2Vnmo ηkx
2 2
correction

was
⎬ V(1978) ωand 2 used z
ω kvx Δδf 2<< = ω nmo 2 2
η =
byeff Castle
i =1
‒ 1
asn in Equation 8: kz =
nmo
‒ 2 nmo V µ µ
ω ω 2kx
8 ⎪ (1994),
4 2 where: tω0 2‒1ttobs0 v‒2tkapp
⎪ 2
ωkzS=‒= 2VV =τηfεk=x442<< γω

Vnmo 4
∑ t i

v Vnmo
kz =µv2 nmo
2nmok Error
VΔz0V Visnmothe
k trms
ω =
S 2=‒ω
1 ‒
tinterval
‒app
2
‒ 22Vnmo
‒ω
t obs 2 t0
x 22 2
×kηx100
velocity kx of the k-th layer; and
i =1
t obs ‒ t app nmovj ε δ ω ηk
∑ τ = tτkapp10the
2
vt∆τη V=isV ‒ 2 V
+(ΔS×2t100 ‒δ1)2 time ofx2 the
vertical k-th layer.
Error = × 100µj = = obsΔs k‒f V
nmo nmo
x2 Error ω ω kx2
Error ∑
t = τs + τ 0 + 2 2
(8) t k t = ‒= t
nmo nmo

=νz∆τ ft obs
obs
ν 2
=k SV
obs app 2
v rmst 0and
n
nmo∑
×V100 2
vωnmo2 2 ‒ 2V 2 ηk
ti(2006)
Δf Δ t nmo S = 1 + 8 η Ursin
t Stovas nmo observed
x that the η factor
t0 = Δf Δ t Δnmo
eff
V 2µ
trms
nmo4==Vttnmo
obs 2µ ‒ t i =1
4‒=t 0
τ0 = f t0 = introduced
SΔ=t 22 =2 obs 4 app app by Alkhalifah and Tsvankin (1995) and the S
f Δ f ω= tError
n
=k V
S
Δ t nmo = t app ‒ t 0 k f= factor 1 t‒ of
0
nmoµ
v 2 x ∑ × t100
thet obsshifted hyperbola approximation are related
rms i
τ s = τ 0 ( S ‒ 1) Δ tznmo v= tby app 0‒ ∑

Equation
2
0 ∆τ V 14:
j i =1

Δ t = µ Δ v
t f
= ‒ = t Δv t k k
1 + 2 δ
ω ∑ ∆τ ω 2 k 2
nmo
nmo=2 0 i
ν 2 = SVrms 2 nmo
V
j app i
kz = nmo f 2 ‒t⎧0 kn2 4 x2 ⎫
i [1 η+kx8η i ] ⎪(14)
µ µ v SΔ=t1V+nmo 8 ηeff⎪ω∑‒vt nmo 2Vtnmo
S = 42 = 44 nmo =1t appi =‒ 0
tηobseff‒ω=t app8 ⎨ v 2 kx2 4 n ‒ 1⎬
1
µ2 Vrms
⎪ ⎪ the displaced hyper-
nmo ∑ i
Error = kz =According ×‒ 1002Vto
µj =
∑ ∆τkVk j
t v
1

bola approach can be used to describe the traveltime
obs ω
this
i = 1
t relation,

∑ ∆τk Δf Δ t nmo V media. ω2 x 2 ω 2 kx2
S = 1 + 8 ηeff = in kzt VTI = τsnmo + τ 02 +2 ‒2 2
f t 0 Figure v 3Villustrates nmo
ν ω ‒the 2Vnmo 2
ηkx
geometry of Equation 8, which
ω 2 2
v kx Δ t =
nmo describes t ‒ t 00 tshifted t symmetric hyperbola (dashed curve) regard-
kz = 1‒ 2 Error τ0 = = obs ‒ app × 100
app

v ω ing theStimet axis, whose asymptotes intersect at point (x=0,


τ 0 ( S ‒31also ) shows the time curve of an actual reflec-
obs
Vnmo ω 2 ω 2 kx2 t=τ s).=Figure
kFigure= 2. Geometry ‒ related to the Dix equation, in Δ f Δ t nmo
torν (solid 2 = curve) and the asymptotes of the shifted hyper-
2
f = SVtrms
z
whichv theVsolid
2
nmo ωcurve
2
nmo ηkx
2
‒ 2Vrepresents the real reflection
bola equation 0 (dashed straight curves). Comparing Figures
time of ta reflector, ‒ t the dashed curve represents the Δ t µ=4 t µ‒4t
Error from =
obs app
× 100 2 Sand =
nmo 3, we =app observe 0 that the NMO equation with a shifted
time the Dix equation (Dix 1955), and the dashed µ22 Vrms 4
t
lines are asymptotes to the Dix equation (modified offset hyperbola is a better approximation to data than the

obs
j
Δf
from Δ t nmo
Castle 1994). conventional ∆τ V Dix equation.
µj = k k

f
=
t0 ∑ ∆τk
Δ t nmo = t app ‒ t 0 S = 1 + 8 ηeff
150
Brazilian Journal of Geology, 48(1): ω v kx 2018 2 2
k 147-159,
= 1 ‒ March z
v ω2
Vnmo ω 2 ω 2 kx2
kz = ‒ 2
v V 2
ω ‒ 2V 2 ηk
Francisco Gamboa Ortega et al.

SIMULATIONS
WITH SYNTHETIC DATA

We used a five-layer model, shown in Figure 4, which


has VTI-based anisotropy in the second, third, and fourth
layers. Table 1 shows the elastic and anisotropic parameters.
The offset depth ratio was greater than 1.5, with maximum
2 2 x2
offset of 9 km and maximum depth of 5 km.t xThe = t 0parame-
+ 2
ters were chosen to be like the real data used in this work. Vnmo
N
Figure 5 shows a Common Midpoint (CMP) with ∑ max-
vi2 ti
imum fold, where four PP-wave reflections areVeasily 2
rms =
i identi-
=1
N
fied and each reflection represents an interface. We also added ∑ ti
noise in such a way that S/N =10. The ray tracing modeling i =1
was
made with the seismic processing package SU –t Seismic
2 2 x2
Unix 2η
Figure 3. Geometry related to the shifted hyperbola x = t 0 + ‒ 2
equation, in which the solid curve represents the (Stockwell Jr. 1997, Cohen & Stockwell 2010). Figure 6 shows Vnmo Vnmo ⎡⎣t 0 Vnmo +
2 2 2

real reflection time of a reflector, the dashed curve the velocity analysis using the hyperbolic approximation
δ << 1 for the
represents the time from the Malovichko equation CMP displayed in Figure 5, as well as the NMO-corrected CMP.
(1978), and the dashed lines are asymptotes to the ε << 1 γ
shifted hyperbola equation. Notice that in relation For large offsets in a VTI medium, we can use Equation 3
ε ‒δ
to Figure 2, the two reflection times curves are much that has three unknown parameters: t0, Vnmo, and ηeff=. The search
1+ 2 δ
closer. Modified from Castle (1994). for these parameters can be separated into two steps. In the n

2 2
∑i =1
2
vnmo ti
Vrms = Vnmo = n
∑ ti i =1

A B vnmo = vi 1 + 2 δ i
⎧ n 4
v t [1 + 8η i ]
1 ⎪ ∑ nmo i
η eff = ⎨ i =1 n

8⎪

4
Vnmo ∑
i =1
t i

x2
t = τs + τ 02 +
ν2
t0
τ0 =
S
τ s = τ 0 ( S ‒ 1)
ν 2 = SVrms
2

µ µ
Figure 4. Layered model used to generate the synthetic seismogram: (A) P-wave vertical velocity; (B)S anisotropic
= 42 = 44
µ2 Vrms
parameter. The first and last layers are isotropic, whereas the three intermediate layers are anisotropic.
The parameters of each layer are shown in Table 1.
µj =
∑ ∆τ V k k
j

∑ ∆τ k

Table 1. Elastic and anisotropic parameters of the synthetic model. S = 1 + 8 ηeff

Layer ρ(g/cm3) Vp(m/s) ω v2k 2


ε δ η
kzVs=(m/s) 1 ‒ 2x
v ω
1 2.16 0 0 0 2000 1400
Vnmo ω 2 ω 2 kx2
kz = ‒
2 2.23 0.081 -0.178 0.402 2409 1324v 2
Vnmo ω 2 ‒ 2Vnmo
2

3 2.29 0.218 0.028 0.182 2757 1509 t obs ‒ t app


Error = × 100
t obs
4 2.38 0.512 0.242 0.180 3308 2260
Δf Δ t nmo
=
5 2.57 0 0 0 4450 f2300 t 0
Δ t nmo = t app ‒ t 0

151
Brazilian Journal of Geology, 48(1): 147-159, March 2018
x2
t x2 = t 02 + 2
Vnmo
N

2
∑vt
i =1
2
i i

Seismic processing of large V


offset rms =
data N
∑t i =1
i

x2 2ηx 4
t x2 = t 02 + 2 ‒ 2
first step, we need toVobtain nmo Vthenmo t⎣
⎡tand
2 2
V a +velocity x 2 ⎤⎦from a
(1 + 2 η)file
0 0 nmo
x (m)
0 1000 2000 3000 conventionalδvelocity << 1 analysis, which is still effective for rel-
0 atively small offsets. In the second step, the velocities of the
ε << 1 γ
first step are used and they are kept constant. Subsequently,
ε ‒δ
the value of ηeff=is estimated by a coherence analysis, as shown
1+ 2 δ
0.5 in Figure 7A. We use the parameters n
obtained in the previous
steps to apply the NMO correction
2 2
∑i =1
v 2
t
nmo i through Equation 3, which
provides anVapproximation
rms = Vnmo = for
n the traveltimes. This correc-
1.0 tion results in the horizontal ∑ t
curves
i =1
i observed in Figure 7B.
In the sequence, we applied Castle’s (1994) method for
vnmo = vi 1 + 2 δ i
NMO correction, which uses Malovichko’s (1978) approach
Time (s)

given by Equation 8.⎧⎪ The analysis⎫⎪(Figure 6A) pro-


nmo t i [1 + 8 η i ]
n

1.5
vides the RMS 2 2 1 x i =1
∑2 v velocity
4

t xη=eff t 0velocity
=+ ⎨2 that is nused in ‒the 1⎬ Dix formula to
obtain the interval 8velocity. V⎪nmo Then, ⎪ velocity is used
N ⎩
Vnmo ∑ ti
4 the interval

in Equation 13 for∑calculating 2
vi ti the µ, which in its turn is used
i = 1

2.0 in EquationVrms 12
2 to idetermine
=1 x 2the value of parameter S. Finally,
t = =τs + N τ 02 + 2
with the value of S, ∑ we tobtain
i
ν the seismic section of Figure 8A,
in which we observe t 0 i =1that for offsets above42,000 m there is a
τ0 = 2 x 2 x the conventional
2ηand
2.5 significant difference
t x = t 0 S+ between
2
‒ 2this result
2
⎡ + 2 η)xthe ⎤⎦ NMO
0 ( S ‒ 1)
NMO correction τ s = τseen V V
in Figure 6B.
nmo nmo ⎣ 0 We t 2 2
V +
nmoalso(1applied 2

correction using
δν 2<< =Ursin
1SV 2 and Stovas (2006) traveltime approxi-
rms
mation, where two input data are necessary: the estimation of
Figure 5. CMP gather for the model in Figure 4. The ε <<µ14 µ4γ
traveltimes were obtained using Seismic Unix (SU) the RMS velocity,
S = 2like = the standard Castle’s method, and the
εµ‒2 δ Vrms 4
ray tracing for a homogeneous anisotropic model. parameter ηeff=, which is obtained from the coherence analysis
1 +∑ 2 δ∆τkVk j
µj = n

2
∑2
∆τk∑ vnmo 2
ti
i =1
VSrms= 1=+V8nmoηeff= n

ω
∑ ti
v 2 ki =x21
kz = 1‒
vnmo = v i 1 + 2ωδ2i
A Velocity
Velocity (m/s) (m/s) B x (m) x (m) 2 2
Vnmo⎧ n ω4 2 ω k ⎫
t ‒[2000 η i ] x2 3000
0 z 1v⎨ ∑
2000 2000 3000 3000 4000 4000 0 0 k1000 = 1000
⎪ v2000 nmo 1 +283000 ⎪
2 i
0 0 0
η eff = i =1 Vnmo ω ‒ 2V‒nmo1⎬ηkx
n
8 ⎪ t ‒Vt4 ⎪
app∑ t i
Error =⎩ ⎭
obs nmo
i =×1 100
t obs 2
0.5 0.5 0.5 0.5 x
t =Δ τfs + Δτ 02t nmo + 2
0.4 0.4 = ν
f t0
t0
1.0 1.0 1.0 1.0 τ0Δ=t nmo = t app ‒ t 0
0.3 0.3 S
τ s = τ 0 ( S ‒ 1)
Time (s)

Time (s)

Time (s)

Time (s)

0.2 0.2 1.5 1.5 ν = SVrms


2 2
1.5 1.5
µ4 µ
S= = 44
0.1 0.1 µ22 Vrms
2.0 2.0 2.0 2.0
µj =
∑ ∆τ V k k
j

∑ ∆τ k
S = 1 + 8 ηeff
2.5 2.5 2.5 2.5
ω v2k 2
kz = 1 ‒ 2x
v ω
(a) (a) (b) (b)
Vnmo ω 2 ω 2 kx2
k = ‒
Figure 6. (A) Velocity spectra using SU in the CMP gather displayed zin Figure 5.2 (B) same gather as in Figure 5
v Vnmo ω 2 ‒ 2Vnmo 2
ηkx
after the isotropic moveout correction.
t obs ‒ t app
Error =
× 100
152 t obs
Δ fMarchΔ t2018
Brazilian Journal of Geology, 48(1): 147-159, nmo
=
f t0
Δ t nmo = t app ‒ t 0
2
∑vt
i =1
i i
Vrms = N
∑t
i =1
i

x2
Francisco Gamboa Ortega et al. t x2 = t 02 + ‒ 2
Vnmo ⎡⎣t 02V
2
Vnmo
δ << 1
ε << 1 γ
(Figure 7A). The result using Ursin and Stovas (2006) equa- We also applied migration to these synthetic data.
ε ‒δ
tion is the seismic section shown in Figure 8B. Comparing The NMO correction, either by the estimation of η eff=or by
1+ 2 δ
Figures 8A and 8B, the Castle’s method was slightly better. the calculation of S, reduces the NMO stretching for large n
This is true mainly for the second interface. offsets. This process is performed in the time domain, but
2 2

i =1
2
vnmo ti
Vrms = Vnmo = n
∑ ti i =1
B
A vnmo = vi 1 + 2 δ i
⎧ n 4
vnmo ti [1 + 8
1 ⎪∑
η eff = ⎨ i =1 n
8⎪

4
Vnmo ∑
i =1
ti

x2
t = τs + τ 02 +
ν2
t0
τ0 =
S
τ s = τ 0 ( S ‒ 1)
ν 2 = SVrms
2

µ µ
S = 42 = 44
µ2 Vrms

µj =
∑ ∆τ V k k
j

∑ ∆τ k

(a) (a) (b) (b) S = 1 + 8 ηeff

Figure 7. (A) Anisotropic parameter analysis using Focus of the CMP gather in Figure 5. (B) NMO-corrected ω v k 2 2
kz =CMP1 ‒ 2x
of the CMP gather in Figure 5 using the Alkhalifah and Tsvankin (1995) traveltime approximation. v ω
Vnmo ω 2 ω
kz = ‒ 2
v Vnmo ω ‒
2

A B t obs ‒ t app
Error = × 100
t obs
Δf Δ t nmo
=
f t0
Δ t nmo = t app ‒ t 0

Figure 8. (A) NMO-corrected CMP of the CMP gather in Figure 5 using Castle (1994) traveltime approximation. (B)
Same gather after moveout correction using the Ursin and Stovas (2006) traveltime approximation.

153
Brazilian Journal of Geology, 48(1): 147-159, March 2018
⎣ ⎦ SS = µ4 = µ4 n
δ << 1 τ s = τ 0 ( Sµ‒221) Vrms 4
∑ 2
vnmo ti
ν = SVrmsrms ∑nmo
2 2 i =1
ε << 1 γ 2 V 2 = V ∆τkVk n = j

ε ‒δ µ4
µj =
µ∑ 4 ∆τk
∑ ti
η= Seismic processingS =of 2large = 4offset data
i =1
1+ 2 δ µv2Snmo= 1V=+rms 8viηeff1 + 2 δ i
n
∑ 2
vnmo ti
µ =
∑ ∆τkωVk ⎧j n v 24kx2
kz = 1 ⎪1∑ ‒ vnmo t [1 + 8η i ] ⎪

η∑ =v k ⎨ ω
2 2
Vrms = Vnmo = i =1 n j
∆τ 2 i
i =1
‒ 1⎬
in an analogue ∑ way. ti In the depth domain, the phase-shiftS = 1 + 8effη 8 ⎪ n
ti ω kx ⎪
kz = nmo2⎩ 2 2 nmo‒∑
2 4
i =1 effV ωV 2 2
migration with anisotropic parameters implies in the col- i =12 ⎭ (16)
vnmo = vi 1 + 2 δ i ω v kx Vnmo2 ω ‒ 2Vnmo 2
ηkx
lapse of diffractions associated with large offsets, even for akz = 1‒ x
⎧ n model. ⎫ of the seismic section in vt = τs + ωt2τobs02 +‒ t app2
flat-layered
⎪ ∑ vnmo ti [1 + 8η i ] ⎪
4 The diffractions
Error = ν × 100
1 ω 2 kx2 frequency, v is the vertical velocity
nmo t ω is tthe
2
Figure 9A,
η eff = ⎨ i =1obtainedn after an‒ 1isotropic ⎬ pre-stack migration Vwhere obs angular
k = τΔv0 f= 0 2 ‒ 2

with 108 shots ⎪ using the model in Figure
nmo ∑ t i
z
spaced
V 4 500 m apart of the Vmedium,
Δ t nmoω ‒and 2Vnmo k2 x ηiskxthe wavenumber in the x direc-
⎩ ⎭ S nmo
=
4A, could be avoided in two ways: tion. For the migration experiments, the SU – Seismic Unix
τ sf =t obsτ 0 ‒( Sttapp0‒ 1)
i =1

■■ decreasing 2
xthe distance between shots in the isotropicErrorpackage= was ×used.
100
t = τs + τ 0 + 2 νΔ2 t=nmoSV 2t app ‒ t 0
t=obsrms
2
migration;νor
■■ usingt an anisotropic migration, as shown in Figure 9B, Δ f =S =ΔThe t nmo
µ µ
4 anisotropic
= 44 migration was more effective than the
τ0 = 0 f t
µ 2
V
inSwhich we can see that the diffractions were collapsed. isotropic. 0 2 We rms can also observe the influence of large anisot-

τ s = τ 0 ( S ‒ 1) Δ t nmoropy
µj =
= t app ∑ k k j on the migration. For instance, in Figure 9A,

parameters
∆τ t 0 V
In such
ν 2 = SVrms 2 case, the anisotropic migration was implemented the amount ∑ ∆τkof noise in the second interface is a result of the
for the same 10 shots spaced apart by 500 m. We used the large anisotropy
S = 1 + 8 ηeff parameters from the second layer.
µ4 µ4
depth-domain
S = 2 = 4 phase-shift migration with interpolation for To check the efficiency of the two methods used in
bothµexamples
2 Vrms (Gazdag 1978, Gazdag & Squazzero 1984). this
kz = paper,
ω v 2 kx2
1 ‒we calculated the percentage error between the
For
µj =

the
j
∆τkVk case, the wavenumber in the depth direc-
isotropic observed v traveltimes ω2 and the times calculated by the approx-
tion is ∑ as seen
∆τk in Equation 15: imation, V using ω 2
the ω kx in Equation 17:
estimator 2 2
kz = nmo ‒ 2
S = 1 + 8 ηeff v Vnmo ω ‒ 2Vnmo2 2
ηkx
ω v2k 2 t obs ‒ t app
kz = 1 ‒ 2x (15) Error = × 100 (17)
v ω t obs
V ω2 ω 2 kx2 Δf Δ t nmo
kz = nmo ‒ 2 =
2
ω ‒ 2case
and forv the Vanisotropic
nmo
2
Vnmo ηkin
as x Equation 16 (Alkhalifah where tobs tis0 the observed traveltime and tapp is the traveltime
f
2000): t obs ‒ t app Δ t nmo = t appby‒ Equations
computed t0 1, 3, and 8.
Error = × 100
t obs
Δf Δ t nmo
=
f t0 xx(m)
(m) xx(m)
(m)
A B
Δ t nmo = t app
00 ‒ t 0 1000
1000 20002000 30003000 4000
4000 5000
5000 00 1000
1000 2000
2000 3000
3000 4000
4000 5000
5000
00 00

500
500 500
500

1000
1000 1000
1000
Depth (m)
(m)

Depth (m)
(m)
Depth

Depth

1500
1500 1500
1500

2000
2000 2000
2000

2500
2500 2500
2500

(a)
(a) (b)
(b)
Figure 9. (A) Isotropic phase shift depth migration of the CMP stack using the P-wave vertical velocities from
Figure 4A. (B) Anisotropic phase shift depth migration using the same true vertical velocities and the anisotropy
parameter from Figure 4B.

154
Brazilian Journal of Geology, 48(1): 147-159, March 2018
= τV ( S ‒ 1) τ s = τ 0 ( S ‒ 1)
2 rms

µ =


τ ∆τ
∆τ V s
k
0 j
k
j

µ = ν = SV k k
∑j
∑ ∆τ
j
∆τ
2
k
k
2
rms ν 2 = SVrms
2

SS == 11 ++ 88S ηη=effeff µ4 = µ4 µ
S = 42 = 44
µ
etµal. µ2 Vrms
2 4
Francisco Gamboa Ortega 2V
ω
ω
2v 2
v 2 k 2 rms
k
kkzz == v 11 ‒‒∑ ∑ ∆τkVk j
x
x j

v µj = ω ω∆τ 2 V
2 k k
µj =
Vnmo ω ω∑ 2 ∆τk
2 ω 2 2
ω 2 kkxx2
∑ ∆τk
kkz == Vnmo S = 1 + 822ηeff‒ 22 ‒ S = 1 + 8 ηeff
For the multilayer model shown in Figure 4, a com- z The
vv frequency Vnmo ω ωdistortion
‒‒ 22V
Vnmo 2 η
2
orkxNMO stretching observed
V nmo nmo ηkx
parison between the presented approximations and the for large offsets ω in v 2 2
Figure
k 6B can be ω
expressed by v 2 2
Equations
k
Error kz tt=obs
obs ‒ ‒ tt app
1 ‒ × 100
app x
kz = 1 ‒ 2x
exact traveltimes from the interfaces is in Figure 10. 18
Error and==19 (Yilmaz
vt ×ω1987):
100 2
v ω
We also computed the error given by Equation 17 for each t obs
obs

Δ ff k Δ Δ t nmo Vnmo ω 2
ω kx2 2
V ω2 ω 2 kx2
interface. For the first interface, all approximations have Δ == z =t nmo ‒ 2 kz = nmo
2 (18)
‒ 2
v Vnmo ω ‒ 2Vnmo ηkx v Vnmo ω ‒ 2Vnmo
2 2 2
ηkx
the same behavior, that is, the errors were the same (see ff tt 00
Figure 10A), because the first layer is isotropic. Alkhalifah Δ
Δ tt nmo
nmo = = tt app
app ‒
t t
‒= tt 00 obs ‒ app × 100 (19) t t
obs ‒ app
Error Error = × 100
and Tsvankin’s (1995) approximation presents a smaller t obs t obs
error than Castle’s (1994) for the second and third inter- where: Δ f is the Δ t nmofrequency variation,Δ f is the Δ t nmo
predominant
= =
faces, as seen in Figures 10B and 10C, respectively. The frequency, f and t is
0t 0
the zero-offset traveltime.
f Int 0conventional
opposite occurs for the fourth interface, in which Castle’s processing, Δ t nmothe= stretching
t app ‒ t 0 effect is removed Δ t nmo =byt app
muting
‒ t 0 a por-
result is better. tion of the affected part.
(a)(a)
(a) (b)
(b)(b)
A x/zx/z
x/z B x/zx/z
x/z
2
2 2 4
4 4 6 6 6 8 8 8 0
0 0 1
1 1 2
2 2 3 3
3 4 4
4
0.014
0.014
0.014 14
14 14
Dix
DixDix(1955)
(1955)
(1955) Dix (1955)
DixDix (1955)
(1955)
Alk-Tsv
Alk-Tsv
Alk-Tsv (1995)
(1995)
(1995) Alk-Tsv (1995)
Alk-Tsv
Alk-Tsv (1995)
(1995)
0.012 S=1+8η 12
12 12 S=1+8η
0.012
0.012 S=1+8η
S=1+8η eff eff S=1+8η
S=1+8η eff eff
S=μ /V 4 eff
4 4 S=μ /V 4 eff
4 4
S=μS=μ
4 /V /V
rms
4 4rms rms S=μS=μ
4/V /V
rms
4 4rms rms
10
10 10
0.01
0.01
0.01
(%)

(%)
Error (%)

Error (%)
Error (%)

Error (%)

8 8
8
0.008
0.008
0.008
Error

Error

6 6
6
0.006
0.006
0.006
4 4
4
0.004
0.004
0.004 2 2
2

0.002
0.002
0.002 0 0
0
1 1
1 2
2 2 3 3 3 4
4 4 0 0
0 1 1
1 2 2
2 3 3
3 4 4
4
x (km)
x (km)
x (km) x (km)
x (km)
x (km)
(c)
(c)(c) (d)(d)
(d)
x/z
x/zx/z x/z
x/zx/z
C 0 0 0.6
0.60.6 1.3
1.31.3 2 2 2 2.62.6 D 0
0 2.6 0 0 0.57 0.571.14
0.57 1.14 1.141.71 1.712.29
1.71 2.29
2.29
10 10
10 8
8 8
Dix (1955)
DixDix (1955)
(1955) DixDix
Dix (1955)
(1955)
(1955)
9 9Alk-Tsv
9 (1995)
Alk-Tsv
Alk-Tsv (1995)
(1995) Alk-Tsv
Alk-Tsv(1995)
(1995)
S=1+8η 7 7Alk-Tsv
7 (1995)
S=1+8η
8 8 S=1+8η
S=1+8η 4 eff
eff eff S=1+8η
S=1+8η 4 eff
eff eff
8 S=μ 4/V
S=μS=μ
4 /V4rms
/V44rms
rms 6
6 6 S=μ 4/V
S=μS=μ
4 /V4rms
/V44rms
rms
7
7 7
(%)

(%)
Error (%)

Error (%)
Error (%)

Error (%)

6 6 5
5 5
6 x2
5 5
5 4
4 4
t x2 = t 02 + 2
Vnmo
Error

Error

4 4
4 3 N

3 3
3
3 3
2
∑vt
i =1
2
i i
2
2 2 V rms = N
2 2
2
1 1 1
1 1
∑t i =1
i
1
0 0 0 x2 2ηx 4
0 0 0 t x2 = t 02 + ‒ 2
3Vnmo3 ⎡⎣t 0 V4
4nmo +4(1 + 2 η)x 2 ⎤⎦
0 0 1 1 2 2 3 3 4 4 0 1 2 2
0 1 2 3 4 0 0 1 1 2 V2nmo 3 2 2

x (km)
x (km)
x (km) δ <<x
x1 (km)
x (km)
(km)

ε << 1 γ
Figure 10. Relative error (%) of VTI traveltimes for the model displayed in Figure 4A comparing the approximations
ε ‒δ
of Dix (1955), Alkhalifah and Tsvankin (1995) and Castle (1994) by using S=1+8 η eff=and Castle (1994) using S=µ4/Vrms
4
:
(A) first interface; (B) second interface; (C) third interface; (D) fourth interface. 1 + 2 δ
n

2 2
∑vi =1
2
nmo i t
V =V =
155 rms nmo n

Brazilian Journal of Geology, 48(1): 147-159, March 2018 ∑t


i =1
i

vnmo = vi 1 + 2 δ i
⎧ n 4 ⎫
v t [1 + 8η i ]
1 ⎪ ∑ nmo i ⎪
ω v2k 2
kz = 1 ‒ 2x
v ω
Vnmo ω 2 ω 2 kx2
kz = ‒
v Seismic
Vnmo2
ωprocessing
2 2
‒ 2Vnmo ηkxof large offset data
t obs ‒ t app
Error = × 100
t obs
Δf
The use of traveltime approximation Δ t nmo equation, with and 30% of plane ground. The elevation along the seismic
=
f improves
more terms in the velocity analysis, t0 the distortion line varied from 78 to 153 m. Three wells were drilled in the
problem, because the value of Δ t nmo decreases
= t app ‒ t 0 in Equation field (Tenerife 1, 2 and 3) reaching rocks from Tertiary and
18. This observation can be seen in Figures 7B, 8A and 8B; Cretaceous ages. There is oil production in the first two wells.
and no muting was applied in these figures. The real data (a 2D multi-component land seismic line)
In conclusion for this application, the compensation of were from an exploration campaign conducted by Ecopetrol
anisotropy effects allows visualizing the reflector in all data, in 2010. Data were collected through continuous/sequential
including large offsets. It is important to notice that in the survey, using the Common Depth Point (CDP) technique
conventional seismic data processing, sometimes the reflec-
tors are edited and silenced, to not show the stretching effect.

APPLICATION TO REAL DATA

The proposed methodology was also used on real data


to detect anisotropy. We used a 2D experimental land seis-
mic line with maximum offset of 9,000 m, acquired in the
Tenerife Field, Colombia. This field is in the Valle Médio del
Magdalena Basin (VMMB), around 260 km distant from
Bogota (Fig. 11). The basin is bordered to the North by the
Bucaramanga Fault, inclined towards the East, with a homo-
clinal trend. It has an area of approximately
​​ 28,300 km2,
and the sediments can reach an 8,500 m thickness.
The Tenerife Field is represented by a polygon in Figure 12
and is structurally limited by a system of reverse faults. The seis- Figure 12. Satellite image showing the area of Tenerife
mic line, indicated by the green line in the same figure, was Field and the location of the experimental seismic line
acquired in an area made up of 70% of mountainous terrain (source: Ecopetrol).

Figure 11. On the left, map showing the location of VMMB (Valle Médio del Magdalena Basin) in red, and on the
right, the Tenerife Field (source: Google Maps and Ecopetrol).

156
Brazilian Journal of Geology, 48(1): 147-159, March 2018
x2
t x2 = t 02 + 2
Vnmo
Francisco Gamboa Ortega et al. N

2
∑vt
i =1
2
i i
V rms = N
∑ ti
and an asymmetrical array. The total length of the line was Figure 13 shows the result of conventional
i =1
NMO cor-
9 km and there were 900 channels (accelerometers) per shot. rection using the Dix equation.
2 2 One x 2 should pay 2η x4
atten-
t x = t0 + 2 ‒ 2
The same processing flow used for the synthetic data Vnmo above
tion to the stretching effects for offsets ⎡⎣t 02Vnmo
Vnmo 300 m 2
+ (1 + 2 η)x 2 ⎤⎦
from
was applied to the Tenerife line. The results are in Figures receivers 1 to 47. Figure 14δ shows
<< 1 the result of NMO cor-
13, 14 and 15. Data were sorted in CDP families with rection with Castle’s method, and Figure 15 presents the
ε << 1 γ
maximum coverage, and the NMO correction was applied result of NMO correction using Alkhalifah and Tsvankin’s
ε ‒δ
with zero stretch. method with a constant η eff= equal to 0.1.
1+ 2 δ
n

2 2
∑v i =1
2
nmo i t
V rms =V nmo = n
∑t
i =1
i

vnmo = vi 1 + 2 δ i
⎧ n 4 ⎫
⎪ ∑ vnmo ti [1 + 8η i ] ⎪
1
η eff = ⎨ i =1 n
‒ 1⎬
8⎪ ⎪

4
Vnmo ∑
i =1
t i

x2
t = τs + τ 02 +
ν2
t0
τ0 =
S
τ s = τ 0 ( S ‒ 1)
ν 2 = SVrms
2

µ µ
S = 42 = 44
µ2 Vrms

µj =
∑ ∆τ V k k
j

∑ ∆τ k
S = 1 + 8 ηeff
Figure 13. Seismic section using hyperbolic NMO correction with Dix (1955) equation.
ω v2k 2
kz = 1 ‒ 2x
v ω
Vnmo ω 2 ω 2 kx2
kz = ‒
2
v Vnmo ω 2 ‒ 2Vnmo
2
ηkx
t obs ‒ t app
Error = × 100
t obs
Δf Δ t nmo
=
f t0
Δ t nmo = t app ‒ t 0

Figure 14. Seismic section using Castle (1994) NMO correction.

157
Brazilian Journal of Geology, 48(1): 147-159, March 2018
Seismic processing of large offset data

x2
t x2 = t 02 + 2
Vnmo
x2
N
t x2 = t 02 +
2
∑vt
i =1
2
i i
2
Vnmo
Vrms = N
∑vt
N
∑t
2
i i i
2 i =1
i =1 V rms = N
x2
t x2 = t 02 + 2 ‒ 2
2ηx 4 ∑t i
Vnmo Vnmo ⎡⎣t 02Vnmo + (1 + 2 η)x 2 ⎤⎦
2 i =1

x2
δ << 1 t x2 = t 02 + ‒ 2
Vnmo ⎡⎣t 02V
2
Vnmo
ε << 1 γ
δ << 1
ε ‒δ
η= ε << 1 γ
1+ 2 δ
n ε ‒δ
2 section using Alkhalifah and Tsvankin (1995) NMO correction with a constant value for η . =
Figure 15. Seismic
2 2
∑i =1
vnmo ti 1+ 2 δ
Vrms = Vnmo = n 2 n
x
∑ t i
x 2 t x
2
= t 2
0 +
Vnmo2 2 2

i =1
2
vnmo ti
i =1 2 2
t x = t0 + 2 V rms = V nmo = n
vnmo = vi 1 + 2 δ i was smaller for weak anisotropy,
N
Vnmo that is, for

N
the 2 third and
vi ti ∑i =1
ti
⎧ n 4 CONCLUSIONS ⎫ fourth interfaces, when∑ using vi2 ti Alkhalifah = i =and Tsvankin’s
vnmo ti [1 + 8η i ] ⎪
2 1
Vrms
1 ⎪∑
N vnmo = vi 1 + 2 δ i
η eff = ⎨ i =1 ‒ 1 ⎬ method. In addition, Vrms =the error
2 i =1 was also smaller ∑ ti for stron-
using ⎧⎪ ∑ v4nmo
N
ti [1 + 8η
n

n
NMO 8 ⎪correction produces⎪ distortions in frequency ger anisotropy, i.e. for the second
ti interface, i =1
when

4 4
Vnmo t x 2
1 2η x
for shallow ⎩ events as i
i =1 well as for ⎭ large offsets. However, Castle’s method. The results indicatei =1 2that 2NMO correction ⎨
t = t 0 + 2 4 ‒ 2eff 2 2 η = i =1
n
x 2 of realx data, 2η ximages ⎡⎣t 0 V ⎪ + (1V+42 η)xt 2 ⎤⎦
8 nmo
a portion of these x 2 distortions is removed in the conven- was successful, and t 2 in the
= t 2 case
+ ‒ the Vnmo Vnmopro-

2 η)x 2 ⎤⎦ ⎩
2
t = τs hyperbolic
+ τ 02 + t2 2processing, x x 0
Vnmo2
Vnmo 2
⎡⎣tquality
2 2
nmo i
0V1nmo + for (1 +shallow
2
tional ν x = t 0 + 2 thus deleting information vided by Castle’s method had a better δ <<
i =1

from large Vnmo x2


t 0 offsets. events. On the other δ << hand,1 the results ε
using
<< 1 Alkhalifah γ t =
and
τs + τ 02 + 2
τ0An N
= anisotropic medium2 produces a non-hyperbolic ν
S ∑ vi ti Tsvankin’s equation on
ε << 1
real data were consistent, despite the
γ ε ‒δ
reflection from
τ s = τ 0 ( S ‒ 1) Vrms = Na 2given i =1interface, which is manifested fact of keeping a constant value for η = . This suggests that ta
ε ‒δ eff
1+ 2 δ τ0 = 0
more significantly
ν = SVrms
2 2
for large ∑ ti offsets. Castle’s method is variable value for η eff=would improve the reflector imaging.
1+ 2 δ n S
a non-hyperbolic approach i =1
for NMO correction that n ∑ v τ
2
nmo
s =t i τ 0 ( S ‒ 1)
µ4 depend µ4 2on any x2 2ηx 4 but just
does not
S = 2 = 4 t x = t0 + 2 ‒ 2
2 anisotropy parameter ∑ v 2 2
V t
rms
nmo i
= V 2
nmo = i =1
n ν = SV
2 2

µ2 VrmsS, which inVnmo


a parameter turnVdepends ⎡t 2 2
nmo ⎣ 0 nmoV +
on (1 +
the 2 η)x 2
inter-⎤
⎦ ACKNOWLEDGMENTS
V 2
rms = V 2
nmo = i =1
n ∑ t i
µ
rms

µ
val time
µj =
∑ ∆τkVand k
j
δinterval
<< 1 velocity V k of the k-th layer. ∑
i =1v
ti
= v 1 + 2 δ
i =1
S = 42 = 44
µ V
A second ∑non-hyperbolic
∆τk ε << 1 approach was presented by
γ
F. Gamboa thanks Coordenaçãonmode Aperfeiçoamento
vnmo = vi 1 + 2 δ i
i


i


2 rms

a η ] k k⎫⎪
j
∆τ V
[
n
Alkhalifah
S = 1 + 8 ηeffand Tsvankin, ε ‒δ
which makes use of an anisot- de Pessoal de Nível Superior (CAPES) for
scholarship. The authors⎧⎪ thank
⎪ ∑ receiving
v 4
1 1 ⎫ the real ∑ ∆τk
nmo t µ
i j 1 = + 8 i
η[1eff+ =8η i ]⎨ i =for
n
ropy parameter2 η2 eff=.
ω v k 1 + 2 δ 1 1 ∑ 4
vnmo tiEcopetrol ⎪
8 ⎪ Científico n
4 S = 1 +e8 η
‒ 1⎬

kz In = this1paper
‒ 2x we applied the nmo ∑ i
methods of Castle and data, Conselho Nacional
η eff = ⎨dei =Desenvolvimento ‒ 1 ⎬ V t eff
n
⎩ ⎭
Alkhalifah and Tsvankin for ∑
ω n
v NMO 2
vnmo ticorrection, both in Tecnológico (CNPq), and PETROBRAS8 ⎪ Vnmo ∑ ti for supportingω
4 ⎪ i = 1
2 22 2 i =1 ⎩ ⎭ 2 k =
v 2 kx2
synthetic V and ωrealVrms 2
data. = VThe ωnmokx=real data were from an exper- the project INCT – Petroleum Geophysics.2 We i =1 x would z 1‒ 2
2 t = τs + τ0 + 2 ω
n
kz = nmo ‒ 2 v
imental v Vnmo ω ‒ 2Vnmo ηk∑
seismic 2 line from the
2 t
Tenerife i Field, Colombia. also like to thank Financiadora
t = τs + τ 02 + 2
x de Estudos e ν Projetos
ix=1 Vnmo ω 2 ω2
For the synthetic example, the simulations indicated that (FINEP) for supporting the ν
project Research t Network k = in ‒
t obs ‒ t appvnmo× 100 = vi 1 + 2 δ i τ0 = 0 z
S and Paradigm v Vnmo ω ‒ 2V
2 2
Error
the = traveltime
relative error (between the observed travel- Geophysical Exploration, t 0 and Landmark
t ⎧ n 4 by the approximation) ⎫ τ0 =
vnmo ti [1 + 8η i ] ⎪ S licenses.τ s = τ 0 ( S ‒ 1)
1 ⎪∑
time and the obs traveltime provided for the academic software t obs ‒ t app
Δf Δ t nmo Error = × 100
= η eff = ⎨ i =1
n
‒ 1⎬ τ s = τ 0 ( S ‒ 1) ν = SV 2 2 t obs
f t0 8⎪ ⎪ rms

Δ t nmo = t app ‒ t 0 ⎩
Vnmo4
∑ ti
⎭ ν 2 = SVrms 2
µ µ Δf
=
Δ t nmo
i =1 S = 42 = 44
µ µ µ2 Vrms f t0
x2 S = 42 = 44
t = τs + τ 0 + 2 2
ν
158 µ2 Vrms ∑ ∆τkVk j Δ tnmo = t app ‒ t0
µj =
t
Brazilian Journal of Geology, 48(1): 147-159, March 2018 ∆τ V j
µj =
∑ k k ∑ ∆τk
τ0 = 0
S ∑ ∆τk S = 1 + 8 ηeff
τ s = τ 0 ( S ‒ 1) S = 1 + 8 ηeff
ω v 2 kx2
Francisco Gamboa Ortega et al.

REFERENCES

Aleixo R. & Schleicher J. 2010. Traveltime approximations for Gazdag J. 1978. Wave-equation migration by phase-shift method.
q-P waves in vertical transversely isotropy media. Geophysical Geophysics, 43:1342-1351.
Prospecting, 58(2):191-201.
Gazdag J. & Squazzero P. 1984. Migration of seismic data by phase
Alkhalifah T. & Tsvankin I. 1995. Velocity analysis for transversely shift plus interpolation. Geophysics, 49(2):124-131.
isotropic media. Geophysics, 60(5):1550-1566.
Helbig K. 1994. Foundation of Anisotropy for Exploration Seismics.
Alkhalifah T. 1997. Velocity analysis using nonhyperbolic moveout Pergamon, Oxford.
in transversely isotropic media. Geophysics, 62(6):1839-1854. Malovichko A.A. 1978. A new representation of the traveltime curve of
Alkhalifah T. 2000. An acoustic wave equation for anisotropic media. reflected waves in horizontally layered media. Applied Geophysics, 91:47-53.
Geophysics, 65(4):1239-1250. Stockwell Junior J.W. 1997. Free software in education: A case study
Castle R.J. 1994. A theory of normal moveout. Geophysics, 59(6):983-999. of CWP/SU: Seismic Un*x. The Leading Edge, 16(7):1045-1049.

Cohen J.K. & Stockwell J.W. 2010. CWP/SU: Seismic Un*x Release Thomsen L. 1986. Weak elastic anisotropy. Geophysics, 51(10):1954-1966.
No. 42: an open source software package for seismic research and Tsvankin I. & Thomsen L. 1994. Nonhyperbolic reflection moveout
processing. Center for Wave Phenomena, Colorado School of Mines. in anisotropic media. Geophysics, 59(8):1290-1304.
Dix C.H. 1955. Seismic velocities from surface measurements. Ursin B. & Stovas A. 2006. Traveltime approximations for a layered
Geophysics, 20(1):68-86. transversely isotropic medium. Geophysics, 71(2):D23-D33.
Fomel S. 2004. On anelliptic approximations for qP velocities in VTI Yilmaz O. 1987. Seismic Data Processing. Society of Exploration
media. Geophysical Prospecting, 52(3):247-259. Geophysicists, Tulsa, Oklahoma.

© 2018 Sociedade Brasileira de Geologia


This is an open access article distributed under the terms of the Creative Commons license.

159
Brazilian Journal of Geology, 48(1): 147-159, March 2018

S-ar putea să vă placă și