Professional Documents
Culture Documents
(12) United
Yang et al.
States Patent (10 ) Patent No.: US 10,571,586 B2
(45 ) Date of Patent: Feb. 25 , 2020
(54 ) FALSE IMAGE REMOVAL IN REVERSE (56 ) References Cited
TIME MIGRATION
U.S. PATENT DOCUMENTS
( 71) Applicant: Saudi Arabian Oil Company , Dhahran 8,116,168 B1 2/2012 Luo et al .
(SA ) 2012/0051176 A1 3/2012 Liu
2014/0328140 A1 11/2014 Khalil et al.
( 72 ) Inventors: Jiarui Yang, Beijing ( CN ); Tong Wang 2015/0124559 Al 5/2015 Cha et al.
Fei, Dhahran (SA ); Yi Luo , Dhahran 2016/0341837 A1 11/2016 Jing et al.
(SA )
OTHER PUBLICATIONS
(73 ) Assignee: Saudi Arabian Oil Company, Dhahran
(SA ) Wang et al., RTM Artifact Removal Via a Fast Algorithm for
Wavefield Decomposition, 2016 SEG International Exposition and
( * ) Notice : Subject to any disclaimer , the term of this 86th Annual Meeting, pp . 4690-4694 ( Year: 2016 ).*
patent is extended or adjusted under 35 (Continued )
U.S.C. 154(b ) by 79 days.
Primary Examiner Toah M Le
(21 ) Appl. No .: 15 /701,038 (74 ) Attorney, Agent, or Firm — Fish & Richardson P.C.
(22 ) Filed : Sep. 11 , 2017 (57) ABSTRACT
(65) Prior Publication Data The present disclosure describes methods and systems,
US 2019/0079206 A1 Mar. 14 , 2019
including computer- implemented methods, computer pro
gram products, and computer systems, for removing false
(51) Int. Cl. image in reverse time migration (RTM ). One computer
GOIV 1/36
implemented method includes dividing volume of a source
( 2006.01) wavefield and a receiver wavefield into a plurality of win
GOIV 1/32 ( 2006.01 ) dows, calculating a down-going component of the source
GOIV 1/28 (2006.01 ) wavefield and an up -going component of the receiver wave
(52 ) U.S. Ci. field for each window in the plurality of windows and at a
CPC GOIV 1/36 (2013.01); GOIV 1/282 particular value of time, calculating a down - going compo
(2013.01 ); GOIV 1/32 (2013.01); GOIV nent of the source wavefield at the particular value of time
2210/44 (2013.01); GOLV 2210/48 (2013.01); by summing the down -going component of the source
GOIV 2210/679 ( 2013.01 ) wavefield in each window of the plurality of windows, and
(58 ) Field of Classification Search calculating an up -going component of the receiver wavefield
CPC .... GO1V 1/36 ; GOIV 1 /366-368 ; GOIV 1/32; at the particular value of time by summing the up - going
GO1V 1/325 ; GOIV 1/282 ; GOIV component of the receiver wavefield in each window of the
2210/679; GO1V 2210/48; GOLV plurality of windows.
2210/44 ; GOIV 2210/40
See application file for complete search history . 16 Claims, 5 Drawing Sheets
100
Divide vuitunc ota uveocid into oxizple overlappog winds. Wadut a wodicom ,
the background velocity for RIM Smumately omgevous.
Fra auch wichw.da kuwai spatial Funier transion to $ (7,4;), S'li .t .Risti ),
arid '( ,t:), and a $ (7 .),STEP 7.1i ) and (h.t;), respectively
leke«up the it. >> port for: (k.r )und $ 8.4.).and
isht ou!ks < part for Rík,e ) and (2,1 ). That is :
{t ${ ?..:)0 ., K ,: 50
>0 8.6. ) SIR .t:AKSO
..>
RelE.Eja {PIRMO
0,4,30 O.K. > O
SOD all tie visdon - lesalis of solar;) and R (1.4 ) vid ispering is oi eripuing area .
the goong 14 vidoxo - going zaretied source and on -going for reciver.
US 10,571,586 B2
Page 2
( 56 ) References Cited
OTHER PUBLICATIONS
Fei et al., “ An Endemic Problem in Reverse - Time Migration ,” 84th
Annual InternationalMeeting, SEG , Expanded Abstracts, Published
in 2014 , pp . 3811-3815 .
Fei et al., “ Removing False Images in Reverse- Time Migration : the
Concept of De-Primary,” Geophysics, 80, No. 6 , Nov.-Dec . 2015,
pp . S237 -S244.
Fletcher, “ Suppressing Unwanted Internal Reflections in Prestack
Reverse - Time Migration ,” Geophysics, 71, No. 6 , Nov.-Dec . 2006 ,
pp . E79- E82.
Liu et al., “ An Effective Imaging Condition for Reverse - Time
Migration using Wavefield Decomposition ,” Geophysics, vol. 76 ,
No. 1, Jan. 2011, pp . S29 -S39 .
Liu et al., “ Reverse - Time Migration using One -Way Wavefield
Imaging Condition ,” 77th Annual InternationalMeeting , SEG Expanded
Abstracts, published in 2007 , pp . 2170-2174 .
Hu et al., “ Angle gathers from reverse time migration using analytic
wavefield propagation and decomposition in the time domain ,"
Geophysics vol. 81, No. 1, Jan. 1, 2016 , 9 pages.
Kim et al., “ 3D Least- Squares Reverse Time Migration Using
Wavefield Decomposition via Hilbert transform ,” 79th EAGE Con
ference & Exhibition , Jun . 12 , 2017, 5 pages .
International Search Report and Written Opinion issued in Interna
tional Application No. PCT/US2018 /049834 dated Dec. 20 , 2018 ,
17 pages.
* cited by examiner
U.S. Patent Feb. 25 , 2020 Sheet 1 of 5 US 10,571,586 B2
1007
Divide volume of a wavefield into multiple overlapping windows. Within a window ,
the background velocity for RTM is approximately homogenous ,
For each window .do local spatial Fourier transform to $ (x ,t ) . S '(Ft;), R ( 7.ti ),
and R'(& tj). and get $ (k , ti ). S'(k.t;). R (k.tz ), and R (k ,ti),respectively.
Take out theky >> O part for $ (k ,t;) and S (K,tj),and
take out k , < part for å (k ,t?) and ? (k.t;). That is :
$ (k , ti ).kz > 0
Sock,ti)= {$(6,41 0.ky 50
$'(k,t ),ky > 0
$\.(8.4) --{$( ,
Ru(E,tr)= {R(E,t) in < 0 Z R_(k,te)= {RP((Ret;0),,Kk ,, 20<
Calculate # 5 : (K ,t1) and HR- (k.ti ) is the average velocity
inside the window , A = k + k + k? is the norm of the wavenumbur vector, é is
a user given small quantity to avoid zero divisor and ; is the inaginary unit .
FIG . 1
U.S. Patent Feb. 25 , 2020 Sheet 2 of 5 US 10,571,586 B2
2007
205
when
tan mer
152
5 225
Output lau ( 8 )
15 235
FIG . 2
U.S. Patent Feb. 25 , 2020 Sheet 3 of 5 US 10,571,586 B2
305
310
FIG . 3
2,
U.S. Patent Feb. 25 , 2020 Sheet 4 of 5 US 10,571,586 B2
4002 S 405
Divide volume of source and receiver wavefields into a plurality ofwindows
For each window in the plurality of windows and at a particular value of ime
Calculate a positive component of Fourier transformation of the source wavefield , and
calculate a negative componentofFourier transformation ofthe receiver wavefield
Calculate a down-going component ofthe source wavefield at the particular value oftime by
summing the down - going componentof the source wavefield in each window of the plurality 5 430
ofwindows, and calculate an up -going componentofthe receiver wavefield at the particular
value oftime by summing the up-going component ofthe receiverwavefield in each window
of the plurality ofwindows
FIG .
U.S. Patent Feb. 25 , 2020 Sheet 5 of 5 US 10,571,586 B2
5002
Network 530
Interface
503
OROKOK
Processor
Memory Database
Oooooor
Wavefields 518
VVV Time-derivative of
Wavefields 520
Service Layer Down- going and Up
going Wavefields 522
Computer 502
FIG . 5
US 10,571,586 B2
1 2
FALSE IMAGE REMOVAL IN REVERSE A first aspect, combinable with the general implementa
TIME MIGRATION tion , comprising , calculating a positive component of Fou
rier transformation of time-derivative of the source wave
TECHNICAL FIELD field and calculating a negative component of Fourier
5 transformation of time- derivative of the receiver wavefield .
This disclosure relates to seismic data processing and , A second aspect, combinable with any of the previous
more specifically , to false image removal in reverse time aspects , wherein the plurality of windows is a plurality of
migration . overlapping windows.
A third aspect, combinable with any of the previous
BACKGROUND aspects, wherein summing the down- going component of
10
the source wavefield and summing the up - going component
High resolution depth images of earth subsurface layers of the receiver wavefield in each window comprises sum
are important for reservoir exploration , delineation , and ming the down -going component of the source wavefield
development. Wave equation based seismic depth migration and summing the up - going component of the receiver wave
techniques, such as reverse time migration (RTM ), are 15 field in each window with tapering in overlapping area .
suitable techniques for seismic depth imaging in the oil aspectsA fourth aspect, combinable with any of the previous
industry, especially in complex environments such as subsalt Fourier, transformation
wherein the Fourier transformation is 1D spatial
.
exploration . However, imaging noise exists in RTM due to
unwanted cross - correlation of source and receiver wave 20 aspects, comprising , calculating with
A fifth aspect , combinable
an
any of the previous
image term based on the
fields at nonreflecting points along a raypath . down -going component of the source wavefield and the
SUMMARY up- going component of the receiver wavefield at the par
ticular value of time.
The details of one or more implementations of the subject
The present
including disclosure
computer describesmethods
- implemented methods, computer
and systems
pro, 25 drawings
matter of this specification are set forth in the accompanying
gram products , and computer systems for removing false and advantages and the description below . Other features, aspects,
of the subject matter will become apparent
image in reverse time migration (RTM ). One computer from the description , the drawings, and the claims.
implemented method includes dividing volume of a source
wavefield and a receiver wavefield into a plurality of win- 30 BRIEF DESCRIPTION OF THE DRAWINGS
dows, calculating a down- going component of the source
wavefield and an up -going component of the receiver wave FIG . 1 illustrates an example method for calculating a
field for each window in the plurality of windows and at a down-going component of a source wavefield and an up
particular value of time, the down- going component of the going component ofa receiver wavefield , according to some
source wavefield being a real component of inverse Fourier 35 implementations.
transformation of a sum of a positive component of Fourier FIG . 2 illustrates an example method for calculating an
transformation of the source wavefield and Hilbert transfor image based on a down -going component of a source
mation of the positive component of the Fourier transfor wavefield and an up - going component of a receiver wave
mation of the source wavefield , the up - going component of field , according to some implementations.
the receiver wavefield being a real component of inverse 40 FIG . 3 illustrates example images before and after remov
Fourier transformation of a sum of a negative component of ing false image in reverse time migration (RTM ), according
Fourier transformation of the receiver wavefield and Hilbert to some implementations.
transformation of the negative component of the Fourier FIG . 4 is a flowchart illustrating an example method for
transformation of the receiver wavefield , calculating a calculating a down- going component of a source wavefield
down-going component of the source wavefield at the par- 45 and an up -going component of a receiverwavefield, accord
ticular value of time by summing the down -going compo ing to some implementations.
nent of the source wavefield in each window of the plurality FIG . 5 is a block diagram illustrating an example com
of windows, and calculating an up- going component of the puter system used to provide computational functionalities
receiver wavefield at the particular value of time by sum associated with described algorithms, methods, functions ,
ming the up -going component of the receiver wavefield in 50 processes, flows, and procedures as described in the instant
each window of the plurality of windows. disclosure, according to some implementations .
Other implementations of this aspect include correspond Like reference numbers and designations in the various
ing computer systems, apparatuses , and computer programs drawings indicate like elements.
recorded on one or more computer storage devices, each
configured to perform the actions of the methods. A system 55 DETAILED DESCRIPTION
of one or more computers can be configured to perform
particular operations or actions by virtue of having software , The following detailed description describes removing
firmware , hardware , or a combination of software , firmware , false image introduced by reverse time migration (RTM )
or hardware installed on the system that in operation causes image condition applied on source and receiver wavefields
the system to perform the actions. One or more computer 60
programs can be configured to perform particular operations
with erroneous propagation direction combinations and is
presented to enable a person skilled in the art to make and
or actions by virtue of including instructions that, when use the disclosed subjectmatter in the contextof one or more
executed by data processing apparatus, cause the apparatus particular implementations . Various modifications, altera
to perform the actions. tions, and permutations of the disclosed implementations
The foregoing and other implementations can each , 65 can be made and will be readily apparent to those skilled in
optionally , include one or more of the following features , the art , and the general principles defined may be applied to
alone or in combination : other implementations and applications without departing
US 10,571,586 B2
3 4
from the scope of the disclosure. Thus, the present disclosure to the ?( k . ),where o denotes frequency and k = k ,k , k )
is not intended to be limited to the described or illustrated denotes a wavenumber vector. The down -going component
implementations, but is to be accorded the widest scope in (kw ) domain is the part with ok >0 , while the up - going
consistent with the principles and features disclosed .
High resolution depth images of earth subsurface layers 5 component in (kw ) domain is the part with wk_<0 .How
are important for successful reservoir exploration , delinea ever, in practical implementation of RTM , the wavefield is
tion , and development. Wave equation based seismic depth typically calculated by the finite -difference method , and due
migration techniques, such as RTM , are suitable techniques to the large data amount, only a few time-slices of the
for seismic depth imaging in the oil industry , especially in wavefield can be stored . Therefore, Fourier transform of the
complex environments such as subsalt exploration . How- 10 wavefield from time t to frequency w , or Hilbert transform
ever, imaging noise exists in RTM due to unwanted cross to pick the positive /negative w part cannot be performed .
correlation of source and receiver wavefields , at nonreflect At a high level, the described approach provides an
ing points along a raypath . approximation method to split a wavefield into down- going
In RTM , the image IX ) is obtained by applying a and up -going components. The described approximation
conventional image condition : 15 method splits a wavefield into down-going and up -going
components with , for example, only the knowledge of the
1(A ) Tmavs (7,7R (2,1)dt (1) current time-slice and the current time-derivative slice of the
wavefield . The current time- slice and the current time
where S (x ,t) is a source wavefield, R (x ,t) is a receiver derivative slice of the wavefield can be acquired from the
wavefield , x = (x,y,z ) represents a position vector, and t 20 finite -difference forward modeling of the wavefield in RTM .
represents time and ranges between 0 and Tmax (e.g. , the Based on this splitting, the image Idulx ) can be calculated
maximum recording time).In general, the image I( X ) can be and the false image produced by wrong component combi
decomposed , according to Liu et al., 2007, into four com nations (e.g., Idd X ), Iu ( x ), Iud( X )) can be eliminated from
ponents :
25 I( x ) to enhance the final image ofRTM . As a result, the final
I(* )=Idu(* )+ Idd )+ Imu(m )+ Iud * ) ( 2 ) image of RMT can be used , for example, to better estimate
where Idu is the cross correlation of the down -going source ration properties of earth subsurface layers for oil and gas explo
.
and up -going receiver wavefields, Ida is the cross correlation FIG . 1 illustrates an example method 100 for calculating
offieldsthe, Iuudown -going
is the crosssource and down
correlation of the-going receiver
up - going sourcewave
and 30 up a down-going component of a source wavefield and an
up -going receiver wavefields, and Iud is the cross correlation some -going component of a receiver wavefield, according to
of the up -going source and down -going receiver wavefields. implementations . For example , based on RTM forward
In addition , the four components are expressed as : modeling, a time-slice of the source wavefield S ( x ,t), atttig
35 is S( x ,t;), and a time-slice of the receiver wavefield R (x ,t),
laul * )=6.7ImerSax ,t)R , (* ,t)dt (3)
at t =t ;, is R ( x ,t :) . In addition , a time-derivative slice of the
Idx ) =lTmars . ) , t)dt (4 ) source wavefield S ( x ,t) and a time- derivative slice of the
luula )= . Tmers, (5,1 )R ( ,t)dt (5 ) 40
receiver wavefield R (x ,t), at t=t ;, can be obtained as:
Iud(x )=S.Tmars ( 1,1) R2( ,t)dt (6 )
S' , 1) = dS dt(x , t)
where S ( x ,t) and R ( X. ) are the down - going components
of the source and receiver wavefields, respectively , and Su (8 )
(x ,t) and R (X ,t)are theup -going components of the source
45
R ' (x , t;) dRX
dt
, 1)
and receiver wavefields, respectively.
The low -frequency imaging noises in RTM can be sup With a time-slice and a time-derivative slice of a wavefield ,
pressed by eliminating the Idd( x ) and Iud(x ) components the wavefield can be split into down-going and up -going
from the image I(X ). To calculate the remaining image Idu 50 components by the workflow shown in FIG . 1.
(X ) + Iudx ), a Hilbert transform may be applied to the As illustrated in FIG . 1, the example method 100 starts
wavefields along depth coordinate z , and the image integra with dividing the volume of a wavefield (e.g., the source
tion - combinations can be calculated afterwards. The Iudl x ) wavefield and/or the receiver wavefield ) into multiple over
component can also form a false image and needs to be 55 in a 3D space and. The
lapping windows window , for example, can be defined
provides a confined 3D sub -volume. In
eliminated . As a result , only the Idul x ) component is kept in each dimension in the 3D space , the window is half
the image I(x ) in order to remove false image in RTM . To overlapped
size can be ,
with an adjacentwindow . In general, the window
for example, one - tenth of its dimension size . In
keep the Idul x ) component only, at least one of the source the present disclosure , the velocity of media within a win
wavefield or receiver wavefield is split into down-going and 60 dow is assumed to be approximately homogeneous, and the
up -going components . Therefore , Hilbert transform previ dispersion relation of the wavefield for all wavenumber
ously described may not produce the image Idulx'). vector k is:
One way to split a wavefield into down- going and up
going components is to perform the split in a frequency 65 0 = + C KI= +Wk + 2+ kz? (9 )
wavenumber domain . For example , the source wavefield where V is the average velocity of media inside the window ,
S ( x ,t) can be transformed (e.g., via 2D Fourier transform ) and w is the circular frequency .
US 10,571,586 B2
5 6
For each window in the multiple overlapping windows,
local spatial Fourier transform (e.g., 1D Fourier transform ) ( 18 )
is applied to the time- slice of the source wavefield ( i.e., AS R. ) S '.. )
VIKI+ )
+
(13) 25
R?_{k,t?)= {? (k ,tj), k? < 0
0,k, = 0
By the conjugate properties of Fourier transform , the down
going source wavefield for the particular window can be
obtained :
If K corresponds to , for example , only two circular 30 Sp (2,4 )= Re( IFFT ($ D :(2,1:))) (21)
frequencies, one positive and one negative as shown in where IFFT( ) is the inverse spatial Fourier transform from
Equation (9 ), ? ( K , to ), in Equation (10 ) can be decomposed k Similarly
to x , and, Re ( ) is a function that returns real part.
into two terms as: from Equations ( 16 ) and (19 ), the following
35
( i.e. , the component with k <0 and a < 0 ) can be obtained :
?+ [k, ti ) = tw= V ?k +k +k } (k ,t;) + + w =-V k?+k%+k (14)
Ru -l , t;) = R - R, 1 ) + HR_R,1 ) = 2 "fw-V/23-43.4?)(2,1) (22)
Faqi Liu , Guanquan Zhang, Scott A.Morton and Jacques P. and wherein Ê ( k , t;) is the Fourier transformation of
Leveille, 2011, An effective imaging condition for the receiver wavefield ;
reverse -time migration using wavefield decomposition : calculating , b the one or more hardware processors ,
Geophysics, vol. 76 , no . 1, S29S39 . 25
Faqi Liu , Guanquan Zhang, Scott A. Morton , Jacques P.
Leveille , 2007 , Reverse-time migration using one-way = R'k ,t ),kz < 0
wavefield imaging condition: 77th Annual International 0 , k , 20
Meeting, SEG Expanded Abstracts .
Tong W. Fei, Yi Luo and Fuhao Qin , 2014 , An endemic 30 a negative component of Fourier transformation of
problem in reverse -time migration : 84th Annual Interna time- derivative of the receiver wavefield , wherein
tional Meeting, SEG Expanded Abstracts .
Ñ '(K , t?) is the Fourier transformation of time-deriva
tive of the receiver wavefield :
The invention claimed is : 35
1. A method comprising: calculating , by the one or more hardware processors,
dividing, by one or more hardware processors, volume of HK_ ( K , t:) , Hilbert transformation of the negative
a source wavefield S( x , t) and a receiver wavefield component of the Fourier transformation of the
R ( x , t) into a plurality of windows, wherein X = (x , y, receiver wavefield ; and
z ) represents a position vector and t represents time; 40 calculating , by the one or more hardware processors,
for each window in the plurality of windows and at a Rulx , t;), an up -going component of the receiver
particular value of time tti: wavefield in the particular window as a real compo
calculating, by the one or more hardware processors , nent of inverse Fourier transformation of a sum of
? ( K , t;), a positive component of Fourier transfor 45 the R_ (K , t ) and the R_ ( K , t );
mation of the source wavefield , wherein K = k , k ,, calculating , by the one or more hardware processors,
k ) represents a wavenumber vector, wherein Salâ , t? ), a down-going component of the source
wavefield at the particular value of time by summing
the down -going component of the source wavefield
in each window of the plurality of windows;
-{$s[8,0k0,,kzti).,sok:>0,
SL( , tz)
50
calculating , by the one or more hardware processors,
Ry( , tz), an up -going component of the receiver
wavefield at the particular value of timeby summing
the up -going component of the receiverwavefield in
and wherein ? ( k , t?) is the Fourier transformation of the 55 each window of the plurality ofwindows; and
source wavefield ; generating, by the one or more hardware processors, an
calculating, by the one or more hardware processors ,
image based on the calculated Sdx , t ) and R , (x , t;)
to remove false image in reverse time migration
(RTM ).
S..(nt)={$ ,1),6:30
S'k , ti ),kz > 0
0, k , = 0
60
2. The method of claim 1 , wherein
$ 2.v)={5&.1,k:>0
0 , k , so
i(V |K | + E) j(V 1K | + €)
and wherein V is an average velocity in the particular
a positive component of Fourier transformation of 55 window , Ikl=VkZ2 +k,2+k_2 is norm of the wavenumber
time- derivative of the source wavefield , wherein vector, E is a user specified number , and j is an imaginary
$'( K ,t,) is the Fourier transformation of time-deriva unit.
tive of the source wavefield :
calculating HR (K , t;), Hilbert transformation of the 60
10. The device of claim 9, wherein S ( x , t;)=Re(IFFT
positive component of the Fourier transformation (S. (K , t:)+ HS ( K ,t;))) and Ru(x , t;)=Re(IFFT?_(K , t;) +
of the source wavefield ; HR_( K , t;))).
calculating Sp ( x , t;), a down-going component of 11. The device of claim 8 , wherein the plurality of
the source wavefield in the particular window as a windows is a plurality of overlapping windows.
real component of inverse Fourier transformation 65 12. The device of claim 11 , wherein summing the down
going component of the source wavefield and summing the
of a sum of the ? .(K , t;) and the AS .(K , t;); up -going component of the receiver wavefield in each
US 10,571,586 B2
19 20
window comprises summing Sp ( x , t) and R , ( x ,t ) in each component of inverse Fourier transformation of a
window w with tapering in overlapping area . sum of the S.(k , t ) and the HS .(K , tj);
13. The device of claim 8 , wherein the Fourier transfor
mation is 1D spatial Fourier transformation . calculating (K , t ), a negative component of Fourier
14. The device of claim 8 , the operations further com 5 transformation of the receiver wavefield , wherein
prising :
calculating an image term Idu ( x ) based on theSX, t;)
and the R ( x , t;). RR, 1 ) = R'k0 ,,kt,),=k?0 < 0
15. A non - transitory computer-readable medium storing 10
instructions executable by a computer system to perform
operations comprising :
dividing volume of a source wavefield S( x , t) and a and wherein  ' k , t?) is the Fourier transformation of
the receiver wavefield ;
receiver wavefield R (x , t) into a plurality of windows, 15 calculating
wherein x =(x, y, z) represents a position vector and t
represents time;
for each window in the plurality of windows and at a
particular value of time tti: RO<CE,i)={R}(k,t)i,k:<0
O , kz 20
>