Petroleum Geology & Geophysics
Common Midpoint Velocity Flattened
Gather Semblance Gather
FWSchroeder 1
Outline
• Introduction to Processing
• Basics of a Stacked Seismic Section
• Pre-Stack Processing Steps
• Seismic Migration
FWSchroeder 2
Processing Objectives
Obtain an accurate, interpretable subsurface image
Seismic data acquisition and
seismic data processing work
together to produce the best
earth image.
Ideally, processed seismic data
should represent the true earth
response. In practice, the
processed data will only be an
approximation.
Subsurface ‘Image’
The image to the right shows a
well processed, interpretable
subsurface image.
FWSchroeder 3
Seismic Processing - Basics
Data Processing
Stream
Field Record
(marine)
Subsurface ‘Image’
FWSchroeder 4
Processing Challenge
Estimate and remove the effects of non-geologic
signals, without impacting the amplitude and
phase of the primary reflections.
If we understand how waves propagate, can we
develop and apply corrections to generate a
more accurate image of the subsurface?
FWSchroeder 5
Elements of Good Processing
1. Accurate Geometry
reflection travel times free
from distortion for true trap
definition
2. Compact Wavelets
correct reflection amplitude Good Processing
and pulse shape for reservoir
lithology and fluid property
prediction
3. Reduced Noise
good signal to noise for
mapping reflections of
interest
Great Processing
FWSchroeder 6
Generic Processing Flow
Processing generally proceeds in several stages
• Signal – steps to:
Signal Processing
• Control and shape the wavelet.
• Preserve the relationship of amplitude to the RCs.
• Improve the signal-to-noise ratio (SNR).
• Imaging
Imaging – steps to:
• Focus the subsurface image.
• Correct for positioning errors.
• Signal Processing– steps involving:
SignalProcessing
• Application of pre- and post-stack enhancements.
FWSchroeder 7
Generic Processing Flow (post-stack migration)
Front End /Geometry
Amplitude/Phase Control Signal
Processing
Noise Mitigation
* Stack
Velocity Analysis
Imaging
Migration
Post-Migration Processing
Signal
Processing
Post-Stack Processing
FWSchroeder 8
Generic Processing Flow (pre-stack migration)
Front End /Geometry
Amplitude/Phase Control Signal
Processing
Noise Mitigation
Velocity Analysis
Imaging
Migration
Post-Migration Processing
Signal
* Stack
Processing
Post-Stack Processing
FWSchroeder 9
Developing a Stacked Section
• To establish some basics, we’ll begin our discussion
with the process called stacking.
• Other processing steps precede stacking, which we
will cover later in the lecture.
• A review of stacking lays a good foundation for our
other discussions about seismic processing.
• Stacking = summing appropriate traces to enhance
the geologic signal and reduce the random noise.
FWSchroeder 10
Repeated Measurements
The seismic method is plagued with noise – both
random noise and coherent (systematic) noise.
To reduce random noise, we get repeated
measurements for each individual subsurface point.
We discussed fold and how we
get repeated measurements
in the acquisition lecture.
We can get four (4) independent
measures of the geology in the
red box by using a series of
different shot points and
receiver locations
FWSchroeder 11
A Marine Shot Record
Shot Record
Source Receivers
Direct Arrival Offset (Distance)
S1
R1 R2 R3 R4 R5 R1 R2 R3 R4 R5
1
Direct Arrival
Two-Way Travel Time
2
Reflections
3
A shot record shows all the traces (with
different offsets) for a single shot Reflection 3
This is what we collect in the field during
acquisition
FWSchroeder 12
A CMP Gather
For Point A
Sources Receivers
A CMP Gather
S5 S4 S3 S2 S1 R1 R2 R3 R4 R5
1
CMP = common mid point
2
In processing, we sort the shot-receiver pairs
so that data from the same ‘bounce’ point 3
is captured.
This gives us a CMP gather.
Offset Distance
FWSchroeder 13
CMP Gather
CMP Gather
Travel times differ with offset since
the path for a near offset trace
is less than the path for a far
offset trace.
With the correct velocity, we can
correct for the difference in
travel time for each trace.
The curvature of this hyperbola is a
function of the average velocity
down to the depth of the reflector
Offset Distance
FWSchroeder 14
With Correct Velocity, the Gather is Flattened
CMP Gather
Velocity
Too Slow
Curves
Down
Velocity
Correct
Flat
Velocity
Too Fast Curves
Up
Offset Distance
FWSchroeder 15
A Stacked Trace
CMP Gather Moveout Corrected Stacked
Midpoint Gather Trace We stack several
offset traces
(# traces = fold)
The geologic
‘signal’ will be
additive
The random
‘noise’ will cancel
Stacking greatly
improves SNR
(signal-to-noise)
Offset Distance
10 Fold
FWSchroeder 16
How the Full Stack is Created
S5 S4 S3 S2 S1 R1 R2 R3 R4 R5
A full stack trace is composed of many
traces recorded with different reflection
Depth
angles but with a common subsurface
reflection point.
This gives us
CMP Gather NMO Correction Mute Applied ONE trace
Stacked
Section
FWSchroeder 17
Full Stack – Characteristics
• Summation (stacking) decreases random noise and
increases signal.
• Often the best dataset for structural interpretation.
• Cleanest dataset to interpret in areas with low SNR.
• Summation (stacking) of all angles (near to far)
angles will decrease the overall bandwidth.
• May not be the best dataset for stratigraphic detail.
• May not be the best dataset to investigate DHIs.
DHI = Direct Hydrocarbon Indicator
FWSchroeder 18
Partial Stacks
• Traces within a specific angle range of a CMP gather are
summed to make angle stacks
– e.g. as near, mid, and fars
A
25°
20°
Note that with increasing depth, the emergent
angle is limited by the offset of the acquisition
• Nature of seismic acquisition results in offset limited data
FWSchroeder 19
Angle Stacks – How They are Created
Raypaths – Identical Angle Contours of Reflection Angle
80
575
655
60 80
50
10
75
70 80
45 65
35 0 75
20 4
15
55 60 70
30
1000 65
50
25 70
5
60
45
Constant Angle
55
2000
40
60 65
Contours
35
50
70
3000
55
30
Two way Time (ms)
10
20
45 65
15
25 60
40 50
4000 35
45
30
5
40
5000 25 35
20
Nears 10
15
Mids Fars 30
6000
25
20
Far
A velocity model must 7000 Near
stack
be available to create stack 15
5
mutes
these contours mutes
8000
0 1000 2000 3000 4000 5000
Offset (m)
FWSchroeder 20
Near and Far Angle Stacks
NMO corrected CMP Gather
Near Angle Far Mute Far Angle
Mute Mute
Near Mute
FWSchroeder 21
Pre-Stack Processing
• There are a number of processing steps that are
done before stacking.
• These are done to:
– Correct for amplitude loss with depth (time).
– Correct for elevation changes for land acquisition.
– Minimize the effect of the wavelet (deconvolution).
– Determine the optimum velocity function for stacking.
FWSchroeder 22
Amplitude Loss with Time
Spherical Divergence
Energy varies as 1/radius2
• Loss of energy through or 1/time2
time/depth/distance
traveled.
• Energy decays by the
inverse square of the
distance traveled.
Since we know how signal strength
is lost by geometric spreading
from the surface source location,
we are able to apply a correction
factor.
FWSchroeder 23
Amplitude Loss with Time
Transmission Loss
Energy is partitioned at each
acoustic interface.
N=1
• At a reflecting interface, normal
incidence energy is converted
N=2
into two waves:
– reflected P waves, and N=3
– transmitted P waves.
N=4
• The incident energy at layer N
has been reduced by all of the
overlying reflected energy.
2 2
Amplitude varies as T
Total = R N (1 - R N - 1 ) (1 - R N - 2 ) … (1 - R 21 )
FWSchroeder 24
Amplitude Loss with Time
Absorption Attenuation
Q = 50, F = 5 Hz
• Seismic energy is converted to
heat.
• This is frequency dependent.
• The higher the frequency, the
greater the loss of signal.
• Absorption is referred to as Q.
Attenuation
• Q varies by rock type. Q = 50, F = 50 Hz
Amplitude varies as Ao e-X/Q
FWSchroeder 25
FWSchroeder 26
FWSchroeder 27
Gain Compensation
Automatic Gain Control Programmed Gain Control
Original Original
Trace Trace
15 sample
AGC applied PGC
applied
Gain Gain
Function Function
Boosting signal strength to a Controlling signal strength to
preset level over a specified preserve proportionality with
time interval bedding reflectivity
FWSchroeder 28
Elevation Correction
Statics
• Source and receivers may be Shot Layout
Variable Surface Elevations
located at different surface Source
Receivers
elevations. R
R R
R
R R
R Receivers
R
• Travel time to any reflector (a
R
flat one in the figure) will
vary as the length of the
travel path changes.
• This effect must be removed Gathers are not flat after Moveout
from the data if the traces Correction due to Elevation Artifacts
are to be successfully Time
Origin
flattened and then stacked.
Distorted Hyperbolas
FWSchroeder 29
Elevation Correction
Seismic Reference Datum
• After applying this time shift, Shot Layout
Variable Surface Elevations
the data appears as if it has Source
all been recorded on the Receivers
R
R
R
R R
R Receivers
R
seismic reference datum. Seismic Reference Datum
• This is referred to as a static
shift.
• All the samples for each
individual trace are moved Shot Record:
by the same amount. Corrected Moveout Curve
Time
Origin
‘Clean’ Hyperbolas
FWSchroeder 30
Static Correction
A time
correction was
applied to
datum traces
to a common
reference
elevation.
Distorted Hyperbolas ‘Cleaner’ Hyperbolas
FWSchroeder 31
Datum Plain
FWSchroeder 32
FWSchroeder 33
FWSchroeder 34
FWSchroeder 35
FWSchroeder 36
FWSchroeder 37
FWSchroeder 38
Trace Muting
Removing
shallow
portions of
traces that are
dominated by
direct arrivals
and refractions.
FWSchroeder 39
Seismic Trace
FWSchroeder 40
FWSchroeder 41
FWSchroeder 42
Removing the Wavelet Shape
Deconvolution
Wavelet shaping to improve recognition and resolution of
reflected events.
Ringing
FWSchroeder 43
Surface-related multip
Multiple Removal reflection at
Multiples: Seismic energy that
Surface-related multip
has been reflected more than
once.
Surface-related multiples have at least one
reflection at
reflection at the surface
Water Second-order
Bottom
First-order Multiple
Surface-related
Multiple removal steps are often
applied several times, especially S R S R
in areas with strong multiples. Source-side
pegleg
lated multiples First-order have
Second-order
A “peel the onion” approach.
First-order Second-order First-order
Surface-related
SubsurfaceSecond-order
Multiple
eflection Surface-related at the surf
Receiver-side
Surface-related
pegleg
Pre-Stack and Post-Stack S R Surface-related
S R
migrationS steps.R S R
Source-side
Source-side pegleg
pegleg
Second-order Receiver-side
First-order
Receiver-side Peg-leg Multiple
pegleg
-related pegleg Surfa
FWSchroeder
Surface-related 44
Velocity Analysis
Reflections on a Velocity Semblance
CMP Stack Diagram
Used to determine the
velocity function that best
flattens the gathers
Two-Way Time
Values that
indicate the
degree of
‘flatness’
Velocity
FWSchroeder 45
Velocity Segregation of Multiples
Semblance Gather
Muted
Energy from Energy from
Multiples Multiples
FWSchroeder 46
Normal Moveout (NMO) Corrected Gather
About
400 fold
Energy from
Multiples • What happens to the energy
from multiples when traces
are stacked?
• What is the difference in
multiple removal for near vs.
far angle stacks?
FWSchroeder 47
Normal Moveout (NMO) Corrected Gather
About
400 fold
Energy from
Multiples
The gently dipping The strongly dipping multiple
multiple will not be will be canceled by destructive
canceled during stacking interference during stacking
FWSchroeder 48
Multiple Removal
Approach:
• Marine: technique is determined by water depth
Deep water – velocity separation (Radon)
Shallow Water - deconvolution
• Land: uses velocity separation (Radon)
and/or deconvolution
Radon In “velocity space” the
multiples can be
recognized and filtered
out
FWSchroeder 49
Radon Applied to Remove Multiples
Before Radon After Radon
Conflicting dip due to Multiples effectively
primaries and multiples removed with Radon
FWSchroeder 50
Pre-Stack Processing
• There are other advanced processing steps that can
be done before stack.
• These additional pre-stack processes are beyond the
scope of this class.
• They are quite expensive is terms of money, time and
resources.
• We will move on to the principle imaging technique –
migration.
FWSchroeder 51
Non-Layer Cake Geology
• Thus far my diagrams, by intention, have had nearly
horizontal layering.
• Things get more complicated with dipping strata.
• The ‘bounce’ point is not simply the midpoint.
Midpoint
Angle of Incidence = Angle of Reflection
FWSchroeder 52
Positioning Problems
Energy
Source
True Path
Posted
Position
0.4 s -
The reflection is displayed
The seismic ray hits an inclined beneath the source-
surface at 90º and reflects back receiver midpoint
FWSchroeder 53
Where Would the Reflection Lie?
1 2 3 4 5 6
90º
FWSchroeder 54
Where Would the Reflection Lie?
1 2 3 4 5 6
Compass
Where would the reflection lie?
FWSchroeder 55
Where Would the Reflection Lie?
1 2 3 4 5 6
Where would the reflection lie?
FWSchroeder 56
Where Would the Reflection Lie?
1 2 3 4 5 6
The reflection is positioned downdip
and has less dip than the interface
FWSchroeder 57
Imaging – Correcting the Position
• Unless we do something, we get a distorted image of
the subsurface.
• Dipping reflectors will be out of position.
• Fortunately, our data processors have a solution.
• They can correct this mispositioning problem through
FWSchroeder 58
Unmigrated Impulse Response
Migrating the Stacked Image
Time
This energy (peak/ trough) The shape of this arc is
could have come from controlled by the
anywhere along this arc. velocity field.
Unmigrated Image
FWSchroeder 59
Sweeping Arcs for Three Wavelets
Migrating the Stacked Image
Time
Destructive
Interference
Constructive
We get constructive Interference
interference where the
true reflector is located.
Unmigrated Image
FWSchroeder 60
When Many Arcs are Swept
Migrating the Stacked Image
Time
Poststack Migration
Sweeps Out Arcs
(Kirchoff Summation)
FWSchroeder 61
Seismic Migration
Migrating the Stacked Image
Time
Unmigrated
Image
In Red
Reflection
Surface
Migrated
Time Display Image
Location
In Blue
FWSchroeder 62
Seismic Migration
Red lines mark the unmigrated
position of two reflections
FWSchroeder 63
Seismic Migration
Migration noise
Red lines mark the unmigrated
position of two reflectors
Edge effect from
the model
Migrated Model
FWSchroeder 64
Stacked – Unmigrated Line
Reflections are not properly located in the subsurface
It is difficult to interpret the structure
FWSchroeder 65
Stacked – Migrated Line
Reflections are mare accurately located in the subsurface
Now a thrust fault can be interpreted
FWSchroeder 66
Migration Options
• There are many methods/algorithms for seismic
migration.
• Each has advantages and disadvantages as we go
from simple, quick methods to more rigorous,
expensive methods.
• “You get what you are willing to pay for”
Post-Stack Pre-Stack Pre-Stack
Time Time Depth
$ $$$$ $$$$ $$$$
Good to Poor Great to Fair Excellent to
Imaging Imaging Good Imaging
FWSchroeder 67
Migration Types
Which type of migration to use is based on velocity variability,
geologic complexity, timing, budget, and business need
Highly Variable
Prestack Depth
With multi-path
Velocity Variation
Beam WEM RTM
Prestack Depth
Kirchhoff
Prestack Time
Kirchhoff WEM = Wave Equation Migration
RTM = Reverse Time Migration
Post-stack Time
Very smooth
Layer Highly
cake
Geologic Complexity Structured
FWSchroeder 68
Migration Velocity Considerations
Smooth => much longer than spread length Time migration
Moderate => close to seismic spread length
Depth migration
Extreme => less than a seismic spread length
Seismic Spread
Length
Sloping Water Bottom
Simple Salt (smooth to moderate)
(smooth - moderate)
Faults
(moderate to extreme)
Complex Salt
(moderate-extreme)
FWSchroeder 69
When is Time Migration Preferred?
• Turn around time is critical.
• No velocity control (wells, VSPs or check shots)
• Business phase
– Exploration – mostly uses time migration
• Discovery wells usually aim for the middle of the target
• “If it’s so small we can’t hit it with this data, it’s too small”
– Development – may be able to get by
• Depth migration is better
• One side-track will pay for a depth migration project
– Production – depth migration should be used
in almost every case
FWSchroeder 70
When is Depth Migration a Must?
• Positioning Accuracy is Paramount, e.g. in the
Production stage
• Strong Lateral Velocity Changes, e.g. irregular
salt bodies
• Severe Anisotropy (>5%) – velocity varies by
direction, e.g. fractured reservoirs
Always remember to budget enough
Time and Money!
FWSchroeder 71
Land vs. Marine Seismic Data
Marine Data Land Data
• Same wavelet for each shot • Variable wavelet
• Simple geometry • Complex geometry
• Long lines • Short lines
• Large volume of data • Small volume
• Only need horizontal locations • Need vertical locations
• No statics • Static problems
FWSchroeder 72
FWSchroeder 73