Vous êtes sur la page 1sur 84

SEISMIC DATA

PROCESSING

 Demultiplexing
 Trace header generation <…observers’ data
 Spherical divergence correction
 Deconvolution before stack
 Band pass filter
 Trace normalization
SEISMIC DATA
PROCESSING (contd)

 Velocity Analysis
 Normal Move Out Correction
 CMP Stack
 Residual statics estimation & application
 Dip Move Out Correction
 Velocity analysis
SEISMIC DATA
PROCESSING (contd)
 DMO stack
 Random noise attenuation
 Decon after stack
 Time Variant Filter
 Migration
 Scaling
Depth & Time
receivers
Shot surface
0

Arrival
Layer-1
time
Layer-2

Layer-3
Demultiplexing
 Required if the Seismic Data is recorded in
multiplexed format
 Conversion of scan sequential mode to trace

sequential mode.
 Essentially a Matrix transposition (rows to
columns and vice versa)
Multiplexing

Ch-1
Ch-48 Ch-2

Ch-20 Ch-3

Ch-12 Ch-4
Ch-5

SCAN-1,ch-1 SCAN-1,ch-24 SCAN-1,ch-48 SCAN-2,ch-1

SCAN-1 SCAN-2 SCAN-47 SCAN-48


SCAN SEQUENTIAL DATA
Demultiplexing
Amp

SCAN-1,ch-1 SCAN-1,ch-24 SCAN-1,ch-36 SCAN-1,ch-48

SCAN-2,ch-1 SCAN-2,ch-24 SCAN-2,ch-36 SCAN-2,ch-48

SCAN-3,ch-1 SCAN-3,ch-24 SCAN-3,ch-36 SCAN-3,ch-48

SCAN-4,ch-1 SCAN-4,ch-24 SCAN-4,ch-36 SCAN-4,ch-48

SC-1200,ch-1 SC-1200,ch-24 SC-1200,ch-36 SC-1200,ch-48

SC-2500,ch-48 SC-2500,ch-24 SC-2500,ch-36 SC-2500,ch-48

ROWS  Scan sequential  Multiplexing


COLUMNS  Trace sequential  Demultiplexing
REF
LEC
TIO
NS
GRO
UND
ROL
L

REF
LEC
TIO
NS
GROUND
ROLL

REF
LEC
TIO
NS
GRO
UND
ROL
L
TYPES OF NOISE
 Coherent Noise (Ground roll, backscatter, multiples)
Methods of reduction :
- Multi channel filtration in t-x,
f-k, Radon domains
- Model based (WEMA,SRME)
 Random Noise
Methods of reduction : CMP stacking,
Predictive decon in f-x domain
NOISE SECTION

First Arrivals

Reflections

Groundroll
First Arrivals

Reflections

Groundroll
Trace Headers Generation
 Generation of addresses to the traces
 Geographical positioning
 Facilitates for the unique identification
 Sorting with respect to a common group

(common shot, common receiver, common


midpoint & common offset)
GATHERS OF DIFFERENT TYPES

receivers
Shot

Common shot gather


GATHERS OF DIFFERENT TYPES

receiver
Shots

Common Receiver gather


GATHERS OF DIFFERENT TYPES

receiver
Shots
Mid Point

Common Mid Point gather


Spherical Divergence
Correction
 Seismic Amplitude decays as a function of
time due to spherical spherical spreading
and inelastic attenuation.
 Compensation is done using a gain function
that is inverse of the decay curve.
 Objective is to see that nearly same amount
of energy is reaching at every layer of the
subsurface.
Amplitude
Amplitude decay

Time 

Recovery
Amplitude
function

Decay curve

Time 
Deconvolution Before Stack
 Earth acts as a high cut filter. Loss of high
frequencies result in loss of resolution.
 The High frequencies that are poorly
represented in the input can be brought on
par with those of better represented.
 Achieved through an inverse filter
application.
Amplitude
Deconvolution

Amplitude

Input

Frequency 
output

Inverse opr Frequency 

Frequency 
Band Pass Filter
 Generallydeconvolution before stack
enhances frequencies.

 Tolimit the frequencies to the seismic


range a band pass filter is conventionally
applied
(8-70 Hz)
Trace Normalization

 The amplitude values are scaled by a scalar


estimated in the user defined time window
to bring them down to observable range.

 Relative amplitude variation is preserved.


Static Corrections
 The elevation differences among the traces of a cmp
gather cause delays.
 The Low Velocity Layer(LVL) near the surface also
introduces delays in the observed travel times.
 The data has to be corrected to a reference surface
(Datum) removing these differences.
 These corrections are static; they don’t change with
time; hence the name ‘Static correction’.
 NMO correction is ‘Dynamic’; it is a function of
time (To), source to receiver offset, and Velocity.
Static corrections

Surface

DATUM

LVL

Reflector
Velocity Analysis

 Estimation of Velocity that yields best


alignment of nonzero offset travel time to
its zero offset time.
 Based on Hyperbolic assumption.
 Results in the best stacking of traces of a
common mid point gather.
Stack Power as a
Function of Velocity
And Time
Normal Move Out Correction
 Non zero offset data is characterized by a travel time
increase with increase in offset distance from the
source to the reflector.
 Non zero offset to zero offset conversion is achieved
through a correction called as NMO (nomal move out)
correction. The NMO equation for a flat layer case is :

Ti**2 = To**2 + Xi**2/V**2, where


Ti = Travel time at offset distance Xi
To = Zero offset travel time
V = NMO velocity or stacking velocity at time To.
Ti – To = DT nmo for offset distance Xi.
 NMO correction is ‘Dynamic’; it is a function of time
(To), source to receiver offset, and Velocity.
Offset 

Time

Before NMO Cor. After NMO Cor.


Stacking
 Each common mid point gather after
normal move out correction is summed
together to yield a stacked trace.
 Stacking enhances the in-phase components
and reduces the random noise.
 Stacking yields Zero offset section (in the
absence of dipping layers in the subsurface)
Brute STACK
STACK after
Dip filtering
Crooked (slalom) Profile & Mid
Point distribution
STACK after Crooked
Profile adjusting
STACK after Velocity
Analysis and NMO
Residual Static Corrections
 Field static corrections are computed using the
velocity of LVL and are based on the ray paths.
 Field statics alone, can not correctly account strong
near surface velocity irregularities.
 Residual static corrections are estimated on the filed
statics applied and NMO corrected gathers in a
surface consistent approach; that is time shifts are
only dependant on the source receiver locations , but
not on the ray paths from shots to receivers.
 Velocity Analysis and stacking performed after
accounting residual static corrections yield
improved resolution.
STACK after Residual
Statics application
Dip Move Out Correction
 NMO ensures non zero offset to zero offset conversion in the
absence of dipping layers. In the presence of contrasting dips, the
estimated velocity will be :
V* = V/Cosine(Alfa), where
V* = dip corrupted velocity,
V = actual velocity and
Alfa = dip angle (measured wrt horizontal).
Ti**2 = To**2 + Xi**2/V**2 – (Xi*Sin (Alfa))**2 /V**2
 The term (Xi*Sin (Alfa))**2 /V**2 is the Dip Move Out term.
 This additional correction promises non zero offset to zero offset
conversion.
Dip Move Out & Migration

Pm
Pd
Pn
(Brian Russel, 1998)
NMO STACK
DMO STACK
NMO STACK DMO STACK
Random Noise Attenuation

 Seismic noise can be either random or coherent.

 Random noise is random. Its estimation is done in


Frequency – space (FX) domain. The predictable
nature of the Sinusoidal signals offer their removal
through a deconvolution. Total field minus the
signal field gives the random noise field.
Coherent Noise Attenuation

 Coherent noise attenuation is achieved through multi


channel (TX, FK, Tau-P domain) filters.
The characteristics of coherent noise are velocity,
frequency, etc.

 Random noise is random. Its estimation is done in


Frequency – space (FX) domain. The predictable nature
of the Sinusoidal signals offer their removal through a
deconvolution. Total field minus the signal field gives
the random noise field.
Signal Enhancement

 Coherent signal can be searched in adjoining traces


in a specified narrow range of dips and can be gathered.

 Multi channel input data facilitates such processes.

 For structural interpretations this is permitted, but not


always a recommended practice for stratigraphic
interpretations.
Deconvolution After Stack
 NMO correction and Stacking also act as a
high cut filter. Loss of high frequencies
result in loss of resolution.
 The deconvolution employed at the post
stack stage is similar to that in the pre –
stack stage, but the parameters are such that
the decon action is milder.
DECON AFTER STACK
Time Variant Filter
 Earth consists of sets of layers (strata) that
are distinctly characterized by certain band
of frequencies.
 It is often advantageous to view the seismic
section (which is a cross section of the earth)
in tune with the characteristic band of
frequencies.
 Different parts of the seismic section can be
subjected different sets of band pass filters in
a time variant manner following tests.
TIME VARIANT FILTER
Migration
 When the subsurface consists of dipping layers, the
Zero offset section does not represent the cross
section of the earth because the reflected energies
are placed at apparent spatial locations.
 Moving the reflection energies from the apparent
locations to the true locations is achieved through
‘Migration’. The spatial velocity distribution of the
velocity is used here for the identification of these
true points in the subsurface.
 Migration improves the spatial disposition of the
reflecting layers and hence achieves ‘Imaging’.
Migration
A B

P
P’

True dip
Q’
Q
Apparent dip

Migration Equation
Tan(app. Dip) = Sin(true dip)
Migration

Syncline
(after migration)
Bow Tie
(before migration)
POST STACK MIGRATION

During migration, trace energy is smeared along a surface of all


possible reflector positions. This means a given wavelet of certain
period (frequency) is placed along a circle of radius equals to the
two way time with the observed CMP location as center. Now
problem lies in the way these smeared energies are placed on the
CMP traces.
CMP
MIGRATION & WAVELET DISTORTION

CMP

Energy placed Energy placed


Along the cmp axis Along the radius
MIGRATION & WAVELET DISTORTION

Higher the Dip, larger will be wavelet stretch.


High frequencies suffer more at higher dips.

CMP Dip=90

Dip=45

Dip=0
MIGRATION & SPATIAL ALIASING

CMP
MIGRATION & SPATIAL ALIASING

CMP
MIGRATION & SPATIAL ALIASING

CMP

Fine output sampling


unique dips
Fine input sampling
unique dips
MIGRATION & SPATIAL ALIASING

CMP

Coarse output sampling


Non unique dips Coarse input sampling
Non unique dips
TEMPORAL ALIASING

Fine input sampling


Proper restoration

Coarse input sampling


Improper restoration
Loss of resolution