Processing of Reflection Data

Improve the signal to noise ratio: e.g. by measuring of

several channels and stacking of the data (white noise
is suppressed).
SEISMIC Obtain a higher resolution by adapting the waveform
DATA of the signals.

PROCESSING Isolate the wanted signals (isolate reflections from

multiples and surface waves).
(PART I) Obtain information about the subsurface (velocities,
reflectivity etc.).

Obtain a realistic image by geometrical correction

(Conversion from travel time into depth and correction
1 from dips and diffractions). 2

Seismic Processing Seismic Processing Systems

Usually geared to a particular type of application
Mostly CMP reflection processing;
Land or marine, 2D or 3D.

Commercial:
ProMAX (Landmark);
Omega (Western Geophysical, marine);
Focus (Paradigm);
Amoco and almost every other company
have their own
Vista (Seismic Image Soft)

Universities:
Stanford Exploration Project;
Seismic UNIX (Colorado School of Mines);
FreeUSP (Amoco);
3 SIOSEIS (Scripts, marine); 4
Seismic CMP Processing Sequence
Processing 1. Demultiplex, Vibroseis Correlation, Gain Recovery.
Conversion from file formats produced by field data loggers into
Systems processing:
SEG-Y, SEG-2.
ProMax, Focus, Omega, SU, Vista, etc., internal formats.

2. Field Geometry.
Assignment of source-receiver coordinates, offsets, etc. in
the trace headers.

3. Edit.
Removal of bad traces (noisy channels, poorly planted
geophones, channels contaminated by power line noise,
etc.).
Printing/storage 5 6

CMP Processing Sequence (Continued) CMP Processing Sequence (Continued)

4. First Arrival Picking. 7. Top, Bottom, and Surgical Mute.
May be semi-automatic or manual. Eliminates (sets amplitude=0) the time intervals where strong
Required for generation of refraction static, models and for non-reflection energy is present: First arrivals, ground roll,
designing the mutes. airwave.

5. Elevation Static. 8. Refraction Static.

Based on geometry information, compensates the travel-time Compensates geometrical spreading.
variations caused by variations in source/receiver elevations. Based on a simple heuristic relation.
Transforms the records as if recorded at a common
horizontal datum surface. 9. Trace Balance.
Equalizes the variations in amplitudes caused by differences
6. Refraction Static. in coupling.
Builds a model for the shallow, low-velocity subsurface. In true-amplitude processing, replaced with surface-
Compensates travel-time variations caused by the shallow consistent deconvolution.
7 8
velocities.
CMP Processing Sequence (Continued)
10. Deconvolution or Wavelet Processing.
Compresses the wavelet in time, attenuates reverberations.
Converts the wavelet to zero-phase for viewing.
11. Gather, CMP Sort.
12. Moveout (Radon, -p, f-k) Filtering.
Attenuates multiples, ground roll.
13. Velocity analysis.
For each of the CMP gathers, determines the optimal
stacking velocity.
9 10
CMP Processing Sequence (Continued)
14. Dip Moveout (DMO) correction.
Transforms the records so that the subsequent NMO + stack
work well even in the presence of dipping reflectors.
15. Normal Moveout (NMO) correction.
Removes the effects of source-receiver separation from
reflection records.
Transforms the records as if recorded at normal incidence.
16. Residual statics.
Removes the remaining small travel time variations caused
by inaccurate statics or velocity model.
Steps 13-16 above are usually iterated 3-5 times to produce accurate
velocity and residual statics models.
Success of velocity analysis depends on the quality of DMO/NMO and
residual statics, and vice versa.

CMP Processing Sequence (continued) Initial Process

17. CMP Stack.
Schematic arrangement of a seismic field tape:
Produces a zero-offset section.
Utilizes CMP redundancy to increase the Signal/Noise ratio.
Can employ various normalization ideas, e.g., diversity
stack.

18. Migration.
Transforms the zero-offset time section into a depth image.
Establishes correct extents and dips of the reflectors.

Unspecified blocks contain mandatory or optional

19. Frequency Filtering and Display.
information.
Attenuates noise.
Provides best display for interpretation.

11 12
Demultiplexing (Sorting of The Data) Demultiplexing (Sorting of The Data)
The principle of multiplexing is already discussed in the section To deal with seismic data and their processing it is helpful to know
measurement system. It is used when the capacities of the AD converter are something about the data formats which are used to store these data.
not sufficient to digitize and save all channels at the same time. This is
common for older measurement systems or for measurements with a large
time window and a lot of channels per shot. The separate values of all Almost every program and every large oil company has developed their
channels are sorted by samples and not by channels. own format. In the course of time several standards have been
developed, to make an exchange possible between the different
It is difficult to programs. The most important standards are:
process the data SEG-Y
in this form. It is
SEG-D
more convenient
and illustrative SEG-2
when the data is
sorted by traces. The other Formats SEG-D and SEG-2 are often used for the storage of
raw data. They are suitable for multiplex data and make it possible to
save traces with different lengths. e.g. the SEG2 format saves each shot
13 14
separately.

Structure of The SEG-Y Format Field Geometry

SEG stays for Society of Exploration Geophysicists. This is the most
important society of geophysicists and seismologists in the oil Assignment of source-receiver coordinates, offsets, etc.
industry.
in the trace headers:
The different formats are very similar. The most important standard
today is the SEGY format. Every file consist of several parts. The Determine Source and receiver position for
standard describes which information is put where in the file :
measured data.
Calculate CMP position.

15 16
Trace Edit Editing

Often, signal traces must be examined visually to detect

and correct or reject those errorneously recorded or
exceptionally noisy. For this purpose a compressed plot
is made off all records. Example of the former are blank
or dead traces, reversed polarity, signal clipping,
incorrect trace sequence, and errorneous recording time.

17 18

Editing Trace Mute

Muting avoids noise wave contamination of reflections on

summed (stacked trace).

19 20
Static Correction Static Correction
Static corrections are applied to seismic data to compensate for the effects of
variations in elevation, weathering thickness, weathering velocity, or
Static Correction : The whole trace is corrected with the same
reference to a datum. The objective is to determine the reflection arrival time
time shift.
which would have been observed if all measurements had been made on a
(usually) flat plane with no weathering or low velocity layer present. These
Objectives of static corrections :
corrections are based on up-hole data, refraction first-breaks, or event
Adjust the seismic traces in such a way that the sources
smoothing.
and receivers are present at one horizontal level. To
achieve this, the travel times of the separate traces are
corrected.

Dynamic Correction : Different time windows in the trace

are corrected differently. This results
in stretching and compression of the
events (e.g. NMO Correction).
21 22

Reference Diagram for Land Static

Datuming
Corrections

23 24
Methods for Static Correction Topographic Correction
Vertical aligning of the different elevations of sources
Topographic Correction (elevation statics).
and receivers.

Up-hole-correction using shots in borehole. Shot-Static

= (Elevation of source - Elevation of reference level) /
Refraction statics : Corrections for weathered layer Velocity.
Delay-Time.
Generalized reciprocal method (GRM). Receiver-Static
Diminishing residual matrices (DRM). = (Elevation of receiver - elevation of the reference
level) / Velocity.

Correction time for a trace

25
= Shot-Static + Receiver-Static. 26

Topographic Correction Uphole-Static

Subdivision of time shift for source and receiver.
All traces with equal source are corrected for the time
shift of the specific source.
All traces with equal receiver are corrected for the time
shift of the specific receiver.
Correction for the weathered (low velocity) layer (area
above water layer where pores are filled with air rather
than water).
When a shot is fired, also the travel time to the surface
is recorded and from this travel time, the velocity of the
weathered layer can be estimated.

The statics correction is the sum of the corrections for

appropriate source and receiver.

27 28
Refraction Statics

Correction for the weathered layer.
Using the first breaks of a certain shot (refracted
energy) a model can be constructed for the weathered
layer (velocities and depth).
Methods to determine the velocity and depth of the
weathered layer using refractions.
Delay-Time.
GRM (generalized reciprocal method).
DRM (diminishing residual matrices).

When the distance between the receiver is too large,

sometimes supplementary refraction measurements are
carried out.

29 30

Stack of Line without The Application of

NO STATICS
Statics
WITH STATICS

32
Example Where ProMax Routine Made a
Static Solution FACTORS WHICH AFFECT AMPLITUDE

33

Amplitude Correction Loss of Amplitude

The amplitude of a seismic signal decays with Loss of amplitude due to:
increasing travel time.
Reflection and transmission at an interface.
To obtain a realistic image, this decay must be
compensated for. Geometrical spreading.

In general, it is difficult to describe this amplitude decay Absorption.

analytically, so an approximation is usually made.
Receiver response.

Measurement system
35 36
Problem for Data Processing OBJECTIVES OF GAIN
STRUCTURAL PROCESSING
Individual large amplitudes dominate the processing. (AGC or TRACEWISE BALANCE)

Best continuty
Reflections are difficult to recognize.
Events visable at all times

Strong amplitude contrasts influence the digital filtering

(especially for large travel-times). STRATIGRAPHIC PROCESSING

(RELATIVE AMPLTUDES)

Bright spots visible

Amplitudes proportional to reflection

coefficients
37
GAIN 6

AMPLITUDE VARIATIONS Methods to Preserve Amplitude Information

SOLUTIONS: Trace equalization.

1. Tracewise balance
All times AGC (Automatic gain control).
All traces

Correction for the spherical divergence.

2. Relative amplitude
Systematic
Programmable gain functions
Output proportional to reflection
Coefficients

GAIN FUNCTIONS VARY WITH: TIME, RANGE, LATERALLY

40
GAIN 10
GAIN FUNCTION APPLICATION GAIN FUNCTION APPLICATION

GAIN 11 GAIN 13

EXAMPLE OF LATERAL VARIATION Trace Equalization

The simplest method is the normalization of the different

traces. All absolute values of a trace are summed and
compared with a reference value. A scaling factor is
determined from the difference between the summation
and the reference value, which is used to multiply all data
with.

44
GAIN 14
Automatic Gain Control (AGC) Automatic Gain Control (AGC)

Normalization of amplitude for a certain time sample in a Automatic gain controls scales by increasing the amplitudes in
certain time window (not for the whole trace). segments of the trace where the amplitudes are low.

Advantage:
All traces are more equal which is needed for further
processing.
(Stacking: summation of different traces)
Amplification of Amplitudes for larger travel times.
It does this by using a moving time window of around a second and
dividing the amplitude of the center point by the root mean-squared
(rms) amplitude of the window.
Some reflectors have stronger amplitudes than others and this may
obscure smaller, but important, reflections.

Disadvantage:
Automatic Gain Control (AGC) boosts the amplitudes so that they
Shadow effect. are all of similar size.
Can lead to amplification of noise.
No physical base for amplification.
45 46

Automatic Gain Control (AGC) Correction for Spherical Divergence

Homogeneous space :

1
A( t ) = G ( t ) = v .t
r
Layered space :

1
G (t ) = [vrms (t tw ) / v (0) ] [ttw / ttw (0) ]
2
A (t ) =
[vrms (ttw )]2 ttw

47 48
Correction for Spherical Divergence Programmable Gain Control

Advantage:
Physical base for amplitude correction.
Relative Amplitude difference remains equal.

Disadvantages:
Velocity function not know beforehand.
Noise sources can still remain dominant.

49 50

Programmed Gain Curve Applying Amplitude Corrections

Gain is time-variant scaling based on function, g (t). Base on some

Calculation of decay of amplitude and determine a criteria, this function is defined at the time samples (shown by solid
Gain function circles) that are usually at the center of specified time gates along the
traces as indicated by 1,2,3 and 4. Gain application simply involves
51 multiplying g (t) by the input trace amplitudes. 52
Four Different PGC Functions Corrected Field Records From a Land Survey

Scale factors are indicated by the circled numbers at the Restored amplitudes at late times by correcting for geometrical
times of application spreading (unfortunately ambient noise also has been strengthened
53 (Yilmaz, 1987). 54

Before AGC After AGC

55 56
Trace Balancing
All traces are normalized using a certain amplitude:
RMS
Median value
Maximum Value

Advantage:
All traces are more equal which is needed for further
processing (Stacking: summation of different traces).

Disadvantage:
No physical base for amplification. Common-shit gathers after trace balancing. Balancing is time-
No equalization of losses with time. invariant scaling of amplitudes to a common rms level for all traces.
Large value in a trace can dominate. 57 58

SPLIT-
SPLIT-SPREAD FIELD RECORD
Filtering

59
 TYPES OF NOISE

62

CAUSES OF POOR SIGNAL-

SIGNAL-TO-
TO-NOISE
NON-
NON-OPTIMAL FIELD PROCEDURES

STRONG COHERENT NOISE *

SCATTERING OR ABSORPTION

- VULCANICS

- FRACTURED ZONE

NEAR-
NEAR-SURFACE PROBLEMS

- POOR COUPLING

- ABSORPTION

IMPROPER PROCESSING (STACK)

 TYPES OF FILTERS

Instrument Compensation

Bandpass Filters

Deconvolution

Waveshaping

Spatial (Mixing)

K-F Filters (Velocity, Wedge, Pie Slice)

CONSTANT FREQUENCY BANDWIDTH

NARROWING THE PASSBAND VARYING ENDPOINTS
 STEEPENING THE SLOPES LENGTHENING THE OPERATOR

FOURIER TRANSFORM
RULE OF THUMB
FOR FILTER SPECIFICATION...
 FILTER SCAN WITH NORMALIZING AFTER FILTER

No Filter 4-8-16-24 6-12-24-36 8-16-32-48 10-20-40-60 14-28-56-84 18-16-72-108

74

F-K Filter F-K Filter

X
-K +K

F
T
REJECT or PASS
75 76
Contoh Aplikasi F-K Filter pada data VSP
Filter TEST?

77 78

Convolution
Convolution is a mathematical operation defining the change of
shape of a waveform resulting from its passage through a filter.
Deconvolution

The asterix denotes the convolution operator. In seismic, we

obtain a response for a certain model by convolving the seismic
signal of the source with the reflectivity function.
79 80
Convolution Example of a Convolution
Mathematically the convolution is defined as follows:

where k = 0 ... m+n; gi = (i=0 ... m) and fj = (j= 0 ... n).

The convolution can also be performed in the Fourier

domain:
Convolution in time domain = Multiplication (of the
Amplitude and spectrum and addition of the Phase
spectrum) in Fourier domain. 81 82

Example of a Convolution Convolutional Model of a Seismogram

Convolution of the reflectivity function with the signal of the source

returns the seismic trace.
83 84
Cross-Correlation Cross-Correlation Function
The cross-correlation function is a measure of the similarity between two
data sets. One dataset is displaced varying amounts relative to the other and
corresponding values of the two sets are multiplied together and the
products summed to give the value of cross-correlation.

The cross-correlation is defined as:

xy ( )= xi + . yi
i
where xi: (i=0 ... n); yi: (i= 0 ... n); xy( ): (-m < < +m) with m = max. displacements.

Similar to the convolution, the cross-correlation can also performed in the

Fourier domain :
Cross-correlation = Multiplication of Amplitudes and Subtraction of Phase
spectrum.
85 86

Auto-Correlation Auto-Correlation of Two Identical Waveforms

The Auto-correlation is a Cross-correlation of a function

with itself. It is mathematically defined as:

xy ( )= xi + .xi
i

where xi = (i=0 ... n); xx( ) = (-m < < +m)

and
m = max. displacement.

87 88
 Autocorrelation Functions Contain
Auto-Correlation: Multiples
Reverberations

A gradually decaying function indicative of short-

period reverberation.
A function with separate side lobes indicative of long-
89 90
period reverberations: multiples.

Normalization of Correlation Deconvolution

Auto-correlation : Convolution was the forward operation :

Source (t) * reflectivity (t) = signal (t)

Deconvolution is the reverse operation:

Reflectivity (t) = signal (t) *-1 source (t)
Cross-correlation :

91 92
 WAVELET CONTRIBUTORS
CONVOLUTIONAL MODEL

95
WAVELET CONTRIBUTORS
IDEAL DECONVOLUTION

97 98

PULSE COMPRESSION SUMMARY OF PURPOSES

99 100
Optimum Wiener Filters for The
Deconvolution
Deconvolution of Seismic Data
Objectives:

ii (0) ii (1) ao id (0)

= Compress the wavelet
ii (1) ii (0) a1 id (1) into a sharp minimum- or
zero-phase shape.
Autocorrelation Cross-Correlation
Inverse filter Broaden (flatten) its
spectrum.
i = input
d = desired

101 102

Spiking and Predictive Deconvolution SUPPRESSION OF COHERENT NOISE

AFTER PREDICTIVE DECONVOLUTION
Basic Idea
INPUT OUTPUT

103
 A CMP stack (a) with no deconvolution, (b) with spiking deconvolution before
stack, (c) with signature processing (minimum-phase conversion of the
measured signature) followed by spiking deconvolution, (d) with signature
processing only (conversion of the signature to a spike).
105 106

107 108
 109
CMP Sort
Sorting from shot gathers to common-midpoint coordinates.
(CMP) gathers.
In a CMP gather, the reflections all come from the
same point for flat layer.
We can also make common-receiver gathers.
A common-depth-point (CDP) is similar to a CMP, but
adjusts for the dip of the layers.
111
110
CMP SORTING
Seismic data acquisition with multifold coverage is done in shot-receiver (s,g)
The figure is a schematic depiction of the recording geometry and ray path
associated with flat reflector. Seismic Data processing, on the other hand,
conventionally is done in midpoint-offset (y,h) coordinates. The required
coordinate transformation is achieved by sorting the data into CMP gathers.
Based on the field geometry information, each individual trace is assigned to
the midpoint between the shot and receiver location associated with that
trace. Those trace with the same midpoint location are grouped together,
making up a CMP gather.
Albeit incorrectly, the term common depth point (CDP) and Common
Midpoint (CMP) often are used interchangeably. 112
Common Midpoint Sorting Common Midpoint Sorting

Raypaths associated with

Common Shot Gather

Raypaths associated with

Common Mid-Point Gather

Raypaths associated with

Figure depicts the geometry of a CMP gather and ray paths associated with Common Offset Gather
a flat reflector. Note that CDP gather is equivalent to a CMP gather only
when reflectors are horizontal and velocities do not vary horizontally.
However, when there are dipping reflectors in the subsurface, these two
Raypaths associated with
gathers are not equivalent and only the term CMP gather should be used.
Selected CMP gathers obtained from sorting the deconvolved shot gathers. Common Receiver position
113 114
Gather

CMP Gather Shot Gathers and Midpoints

115 116
CMP Gathers Now Fold of Coverage for CMP Stacking
For most recording geometries, the fold of coverage nf
for CMP stacking is given by :

ng g : receiver group
g
nf = s : shot interval
2 s ng : The number of recording channels

117 118

NMO Correction of a CMP gather Stacking

CMP Stack a CMP gather
 Example of stacking to
attenuate multiple

STACK
SIGNAL-TO-NOISE
SIGNAL-TO-NOISE
AMPLITUDE
AMPLITUDE IMPROVEMENT
IMPROVEMENT

where
where N
N == number
number of
of unique
unique
range
range traces
traces stacked
stacked

Stack section Velocity Analysis

In addition to providing an improved signal-to-noise ratio,

multifold coverage with nonzero-offset recording yields
velocity information about the subsurface. Velocity
analysis is performed on selected CMP gathers or groups
of gathers. The output from one type of velocity analysis
is a table of numbers as function of velocity versus two-
way zero-offset time (velocity spectrum). These numbers
represent some measure of signal coherency along the
hyperbolic trajectories governed by velocity, offset, and
travel time.

124
 Velocity Analysis Determine The Velocity

The aim of the velocity analysis is to find the velocity, that

flattens a reflection hyperbola, which returns the best result There are different ways to determine the velocity:
when stacking is applied. This velocity is not always the real (t2-x2)-Analysis.
RMS velocity. Constant velocity panels (CVP).
Constant velocity stacks (CVS).
Therefore, a distinction is made between:
Analysis of velocity spectra.
Vstack : The velocity that returns the best stacking result.
Vrms : The actual RMS-velocity of a layer. For all methods, selected CMP gathers are used.

For a horizontal layer and small offsets, both velocities are

similar. When the reflectors are dipping then vstack is not equal
to the actual velocity, but equal to the velocity that results in a
similar reflection hyperbola.
125 126

(t2-x2)-Analysis

The (t2-x2) -Analysis is based on the fact, that the Moveout-

expression for the square of t and x result in a linear event.
When different values for x and t are plotted, the slope can be
used to determine v2, the square root returns the proper
velocity.
Example of a t2-x2-Analysis :

Examining the normal moveout equation, it is possible to analyze

127 NMO velocities by plotting reflections in T2 X2 space. 128
 Options in The ProMax Velocity Analysis
CVP - Constant Velocity Panels
Routine

The NMO-correction is applied for a CMP using different

constant velocities. The results of the different velocities
are compared and the velocity that results in a flattening of
the hyperbolas is the velocity for a certain reflector.

129 130

CVS - Constant Velocity Stacks Stacking Velocity

Similar to the CVP-method the data is NMO-corrected.
This is carried out for several CMP gathers and the
NMO-corrected data is stacked and displayed as a
panel for each different stacking velocity. Stacking
velocities are picked directly from the constant velocity
stack panel by choosing the velocity that yields the
best stack response at a selected event.

CVP and CVS both have the disadvantage that the

velocity is approximated as good as the distance
between two test velocities. Both methods can be
used for quality control and for analysis of noisy data. 131 132
Concept of Constant Velocity Stack as An Aid
Velocity-Spectrum
to Stacking Velocity Estimation
The velocity spectrum is obtained when the stacking results
for a range of velocities are plotted in a panel for each
velocity side by side on a plane of velocity versus two-way
travel-time. This can be plotted as traces or as iso-
amplitudes. This method is commonly used by interactive
software to determine the velocities.

133 134

Methods to Calculate The Velocity-Spectrum Velocity Spectrum

Amplitude of stacking
Normalized amplitude of stacking
Semblance n
Amplitude of Stacking : s t = wi ,t
i =1

| st |
Normalized Amplitude of stacking : nst = n
wi ,t |
i =1
st2
1
Semblance : Semblancet = t
n wi2,t
t i

Wi,t value for i-th trace, time t Mapping of the offset axis to the velocity axis
135 136
Demonstration of The Velocity Spectra Constant Velocity Stack (CVS)

One Method to Determine Stacking Velocity is to Use A Constant

137 Velocity Stack (CVS) for Several CDP Gathers. 138

Constant Velocity Stack (CVS) Constant Velocity Stack (CVS)

Same CVS Panel of Traces as Before Switching to Variable Same as Previous Color Panels with Velocity Range Now
Density Color for The Traces to Utilize Dynamic Range. 139 Halved to Better Pick Correct Velocities. 140
Constant Velocity Moveout (CVP) Corrections Constant Velocity Moveout (CVP) Corrections
(1) (2)

Velocity ft/s
141 142

Influence of Missing Long Offset Traces on Influence of Missing Long Offset Traces and
Velocity Spectra Statics on Velocity Spectra

143 144
Semblance-Analysis

145 146

Error for High Velocities and Large Travel

Times

147 148
 Residual Static 2. Why are residual static needed?
CMP GATHER
1. What are residual static?

Static
Corrections applied to seismic data to compensate
for the effects of variations in:
- elevation,
- weathering thickness,
- weathering velocity
(near-surface velocity anomalies) or
- reference to a datum.

CMP GATHERS BEFORE AFTER RESIDUAL STATICS

STACK before & after residual statics

After residual statics

Before residual statics
STAT 5
STACK before & after residual statics

THE END OF
SEISMIC DATA PROCESSING
PART I

153 154