You are on page 1of 17

Stack

CMP gather
NMO
corrections STACK

Objectives
 improve signal-to-noise ratio;
 attenuate multiples;
 attenuate ground roll.

Procedure
 velocity analysis (stack velocity);
 NMO corrections (Normal Moveout);
 CMP stack.

NMO corrections

1
CMP CMP gather
Stack section
CMP 656

NMO STACK

CMP number

time
[ms]

Fold 100%

(Courtier and Mendenhall , 1967)

2
Fold 1200%

(Courtier and Mendenhall , 1967)

Horizontal interface (CMP gather)


S 0 x (offset) R

From Pythagorean theorem:


h V
v2  t 2  x2  4  h2
Reflector

x2 4  h2
two - way time  t 2   2
v2 v
2h
S’ minimum time  t0 
CMP v
x x
Two-way traveltime of a
hyperbola asymptotes  
t0 v
reflection assuming
horizontal interface and
homogeneous medium
hyperbola

3
Dipping interface (CMP gather)

S 0 x R
For the Cosine Rule:

v 2t 2  x 2  4h 2  4 xh cos(90   )
hh 
h’ V
x
Since h'  h  sin  and cos(90    )  sin 
2
x x
 v t  x  4( h ' sin  )  4 x sin  ( h' sin  )
2 2 2 2
M’
2 2
v 2t 2  x 2  4h '2  x 2 sin 2   4 h ' x sin   4 h' x sin   2 x 2 sin 2 
v 2 t 2  x 2 cos 2   4 h'2
x 2 4  h '2
two - way time  t 2   2
CMP va2 v
x
Two-way v
traveltime of a va  apparent velocity  v
reflection
t0 cos 
assuming dipping 2  h'
interface and hyperbola
minimum time  t0 
homogeneous v
medium x
t hyperbola asymptotes  
va

Two-way traveltime in a layered structure (CMP gather)


x (offset)
S R
CMP
1 1 e1 V1 V V V
1 sin   sin  ............ sin   cost
1 2 n

2 e2  V2 1 2 n
2
3 e3  V3
3

n en  Vn
n
CMP Vn 1
x
Two-way traveltime of a x
t
reflection assuming a Vmax
layered structure with
t0
horizontal interfaces Observed curve t(x)

Hyperbola
t
x2
t (x) is not a hyperbola anymore but it t 2  t0 2  2
is nearly hyperbolic : V NMO

4
Which velocity is VNMO?
By differentiating the hyperbolic equation :
2 x dx 2 x dx
2t  2  VNMO 
V dt t dt
NMO

dx
dx V1dt dx V
dt V1dt  dx sin 1   1
V1 dt sin 1
1
x x  2 ei tan i
At short offsets  tan i  sin i  x  2 ei sin i
If  i is the two-way vertical traveltime through the ith layer
2ei  Vi i  x   Vi i sin i
By multiplying and sin i sin 1
x   Vi  i x Vi  i
2 2
dividing for Vi Vi V1

Vi  i
2
2
Substituting the results in the VNMO expression: V
NMO 
t
Vi  i  V 2
2
2
At short offsets  t  t0 VNMO 
 i
RMS

Average velocities (in a layered structure)


n n
2 ei V  i i
Vm (n)  n
1
 Vm (n)  1
n
simple mean
1
i  1
i

V 
2
i i
VRMS (n)  1
n
root mean square
 1
i

For the 1st reflector  Vm  VRMS  VNMO


For the following reflectors  Vm  VRMS and VRMS  VNMO (for x << z)

5
DIX equation
It is the inverse relation of VRMS expression.
According to VRMS expression:
2 n n
VRMS (n)   i   Vi  i
2

1 1
From top to bottom :
VRMS (1)  V1 ;  1  t1
2 2
VRMS (2)  t 2  VRMS (1)  t1  V22  2 ;  2  t 2  t1
2 2
VRMS (2)t 2  VRMS (1)  t1
 V2
2

t 2  t1 DIX equation
2 2
VRMS (3)t3  VRMS (1)  t1  V22 (t 2  t1 )  V32 (t3  t 2 )
2 2 2 2
VRMS (3)t3  VRMS (2)  t 2 V (n)t n  VRMS (n  1)  t n 1
 V3 …….. RMS  Vn
2 2

t3  t 2 t n  t n 1

NMO corrections (or dynamic corrections)


x2
t x  t0 
2
2
VNMO
x2
NMO  t x  t0  t0   t0
2
The correction consists of : 2
VNMO

To apply NMO corrections we need VNMO (to) Velocity analysis

If x << z :
 
x2  x2 
NMO  t0 1   t 0  t 0 1   t0
 2V 2 

2 2
2
VNMO t0  NMO 0 
t

x2
NMO  2 Parabolic approximation
2VNMOt0

6
Velocity analysis

Common methods for velocity analysis:

- Constant Velocity Gathers (CVG)

- Constant Velocity Stacks (CVS)

- Velocity spectra

CMP gather (synthetic data)


Original Right correction Overcorrection Undercorrection
data (VNMO=Vo) (VNMO<Vo) (VNMO>Vo)

(Yilmaz, 2001)

7
Constant Velocity Gathers (CVG)
Original After constant
data Velocity corrections

(Yilmaz, 2001)

Constant Velocity Gathers (CVG)


After constant
Velocity corrections

(Yilmaz, 2001)

8
Constant Velocity Stacks (CVS)

(Yilmaz, 2001)

Velocity spectra
At any time, a semblance function is applied to CMP data after
NMO corrections with a set of different velocities.
Possible semblance functions: semblance , sum , cross-correlation.

2
Semblance M  si,j = i-th sample of the j-th trace
 si , j  of a CMP gather
1  j 1 
CSMB  0  C SMB  1
M M 2
 si, jj 1
max. semblance

As an example, for M = 2:
si2,1  si2, 2  2 si ,1si , 2

2si2,1  si2, 2 
CSMB

4 si2,1
if si,1 = si,2 (max. semblance)  CSMB  1
4 si2,1

9
Actually, semblance is calculated as an average in a short time window.

2
N
M 
  si , j 
CSMB 
i 1  j 1  Note Semblance is sensitive either to amplitude
N M or waveform differences.
M  2
s
i, j
i 1 j 1

Normalized Crosscorrelation
 N 
2 M   si , j  si ,k  Note Normalized
 
C NXC   j ,k ( k  j )  N
M M  1 1
i 1
N 
crosscorrelation is

  i , j  si ,k
exclusively sensitive to
s 2
 2
 waveform differences.
 i 1 i 1 

Simple sum
N M
CSUM   i  si , j
1 j 1

Velocity spectrum
Synthetic data

(Yilmaz, 2001)

10
Real data Velocity spectrum

(Yilmaz, 2001)

Sensitivity Vs. Spread Length

(Yilmaz, 2001)

11
Velocity spectra with noisy data

(Yilmaz, 2001)

Velocity spectra with noisy data

(Yilmaz, 2001)

12
CSMB (average over 2 CMPs) CSMB (average over 8 CMPs)

(Yilmaz, 2001)

Quality control of stack velocity


Interval velocities area checked for their geological plausibility

Interval velocities from Dix equation

VNMO from velocity analysis

13
Stretching

NMO corrections are dynamic, thus they generate


wave distortion (stretching).

T’
T

(a) (b)

t t

Before NMO After NMO

S1 S2 S3 S1 S2 S3
0 R1 R2 R3 x 0 R1 R2 R3 x

100 100

200 200

300 300
ms ms
t a) t b)
Deformation of reflected pulses by NMO correction.
(a) Pulses before correction, and (b) After correction.

14
Muting

Trace portions where stretching overcomes an acceptable


percentage, i.e., lower times and longer offsets, are muted so that
they do not contribute to the construction of the stack trace.

Objectives :

- prevent stack degradation caused by NMO stretching.

- remove direct and refracted events from stack section.

- prevent stacking of badly aligned events (at x>>z


reflections are not hyperbolic anymore).

Before NMO After NMO After muting

(Yilmaz, 2001)

15
Multiple attenuation

Surface

Zm

Assuming tp = tm

Comparison of primary and multiple Normal Moveout

NMO

NMO p

NMOm

16
Example of data affected by multiples corrected with Vp
Before NMO Velocity Spectrum After Nmo and muting

(Yilmaz, 2001)

17

You might also like