Migration
What is Migration ?
Migration is a process which attempts to correct the
distortions of the geological structure inherent in the
seismic section
Migration re-distributes energy in the seismic section
to better image the true geological structures
Why Migrate ?
Rearrange seismic data so that reflection events
may be displayed at their true position in both space
and time.
laterally in up-dip direction
upward in time
Collapse diffractions back to their point of origin
Improve lateral resolution - collapse Fresnel zone
To obtain more accurate velocity information (when performed
pre-stack)
For more accurate depth section
3
Apparent Dip
Wrong
Correct
Wrong
Correct
A zero offset stack section gives a false picture of dipping reflectors as events A
and B are plotted at trace positions A and B respectively.
The apparent dip of an event on a zero offset stack section
is less than the true dip of the event.
4
Recap
Depth section
SR
Time section
SR
SR
SR
x
after migration
how the event appears
before migration
depth
true subsurface position
of the event
time
This is a constant velocity case
> Migration moves events updip
> Migration steepens events
> Migration shortens events
5
Diffractions
When wave energy strikes discontinuities in the subsurface, diffraction occurs
and the seismic waves are bent around the discontinuity.
This happens at layer pinchouts, faults etc.
layer pinch-out
The termination point of the
pinchout acts like a point source
and seismic energy is reflected
out in all directions.
How the diffraction
appears on a seismic
cross-section.
6
Data Example
Diffractions
Diffracted energy appears
very similar to that reflected
from an anticline.
Diffracted energy IS useful
Velocity Errors
Diffractions not collapsed
Turned into a smile
Undermigrated
Overmigrated
Velocities too low
Velocities too high
8
Important Structural Features
Anticline
1
7
x
Geology
z
1
7
x
Stacked (structure appears
too broad)
T
Anticline - Data Example
STACK
Geology
MIGRATION
10
Important Structural Features
Syncline
Left, below, shows true zero-offset ray paths for coincident sources
and receivers at surface points A through G to a synclinal reflector.
The curvature of the reflecting horizon is such that there are multiple
perpendicular ray paths from surface positions B through F. On the
right is shown how a syncline will appear on a stacked section
Distance
Distance
A
Depth
Time
Geology
Stack
Syncline - Data Example
STACK
(showing
bow-tie
effect)
Geology
MIGRATION
12
Important Structural
Features
Fault
Geology
Stack
Fault - Stacked Data Example
Importance of Velocity
A local high- Anticline appears wider on the stacked section. The
apparent width depends on the dips on each flank - the steeper
the dips the wider the structure on the stacked section.
migrated section
stacked section
Anticline
If our velocities are incorrect, the final migrated structure may be
narrower or wider than the true structure - this could lead to the
wrong estimate of any oil or gas reserves under this high.
Importance of Velocity
A local depression - Syncline appears narrower on the stacked section
Syncline
stacked section
migrated section
If the velocities are too high, the syncline will be wider than it should
be after migration. Velocities that are too low will under migrate
the structure, leaving it too narrow.
Importance of velocity
A perfectly horizontal event with a fault plane.
If the velocity field is incorrect, the fault plane will not move
to the correct place and the diffraction curves from the corners
of the structure will not focus correctly
Migration Velocity Field
Modifications must be made to the stacking velocity field so that it can be
used for Migration.
Smoothing
Time migration algorithms cannot handle strong lateral velocity variations
Filtering in time / horizon
consistent manner
Stacking velocity field
Smoothed velocity field
18
Migration Velocity Field
Scaling
Trial migrations are conducted using varying percentages of the stacking velocity
field
80% of stacking
velocities
100% of stacking
velocities
120% of stacking
velocities
Conversion to interval velocities
Stacking velocities must be converted to interval velocities to be of use to
most migration algorithms. These conversions are done internally.
19
A strange structure
If we had an inverted perfectly parabolic event in the ground, with its centre
on our CDP, we would see only one point on our stacked section
Parabolic smile
A glitch or a spike
(stacked section)
(migrated section)
If we have a single point of energy in our stacked section that
does not correspond to a real event (e.g a spike), this will
migrate into a parabolic smile on our migrated section.
Recap
Why do we need to migrate seismic data ?
re-arrange events to true subsurface positions
dipping events will shorten, steepen and move updip
diffraction energy will be collapsed
amount of under and over migration will depend on
the velocity
21
Types of Migration
Principles
Techniques / Algorithms
Time or Depth?
Pre-Stack or Post Stack?
2D or 3D?
Important Principles
Exploding Reflectors
Each point on a reflector can be considered as a secondary source of energy.
If we could measure the shape of the wave front at t=0 (i.e. at the reflector surface),
since no propagation has occurred at this time then the wave front shape must be the
same as the reflector shape that generated the wave front. This is the IMAGING principle.
surface
Point Reflector
(exploding reflector)
Based on Huygens Principle
23
Important
Principles
Downward Continuation
The imaging is achieved by extrapolating the wave-field
back in depth, from the surface to the exploding reflector
where, t =0
Downward continuation is performed at regular depth
or time intervals until all reflectors are imaged.
Migration Techniques
Most modern migration techniques are based on
the scalar wave equation, which relates the spatial
and time dependence of a disturbance which can
propagate as a wave*.
* R.E.Sheriff
In rectangular co-ordinates x, y and z
2
2
2
1
2
2
2
2
x
y
z
V
Where represents wave displacement (pressure, rotation etc)
and V the velocity of the wave
Migration Techniques
Three main categories
(most of these use the principle of downward continuation
and exploding reflector model)
Finite-Difference Methods- based on differential
solutions to the scalar wave equation
Integral Methods- integral solution to the scalar wave
equation. Based on summing the seismic amplitudes
along a diffraction hyperbola whose curvature is
governed by the medium velocity
Transform Methods- involving co-ordinate
transformation into mainly F-K domain
Migration Techniques
Finite-Difference Methodsgenerally used when vertical velocity variations exist with gentle
lateral velocity variations and dips of the events are moderate.
Integral Methods-. Used mainly when data exhibits high
dips. Lateral velocity variations can still be a problem. Kirchhoff
migration is an example. Uses RMS velocities instead of interval
velocities.
Transform Methods- usually implemented in F-K domain.
Examples are: Phase-shift and Stolt migrations. These are
relatively cheap to run.
Important Migration Algorithms
Name
Finite-Difference
Kirchhoff
Method
Downward continuation
migrates in steps from receivers
downwards
based on diffraction
summation. Uses Huygens
Comments
can only handle dips up to 60 degrees
can handle minor lateral velocity variations
can handle low S/N
can handle dips up to 90 degrees
cannot handle strong lateral velocity
variations
F-K
(Stolt, Phase-Shift)
F-X
(Omega-X)
principle
converts to frequencywave-number domain before
migrating. Stolt can be used
to perform residual migration
poor when low S/N
can handle dips up to 90 degrees
generally poor at handling lateral velocity
variations.
poor when low S/N
type of Finite-Difference
migration in the frequency
domain
can handle dips up to 90 degrees
can handle moderate velocity variations.
each frequency can be processed separately.
poor when low S/N
Time or Depth Migration ?
Ideally we want a depth section from a stacked section. The best
way to achieve this is to do Depth Migration. However, time
migration is considered adequate for most surveys. Depth
Migration is too time consuming and expensive to be widely used
interpreters prefer to evaluate the velocity of migrated section by
comparing with un-migrated data. More common is to migrate with
varying percentages and then compare the results.
Stacking velocities are accurate for stacking and may not be true
velocities. Much work is required to build a velocity model for
depth migration.
However,
when lateral variations in velocity are severe, Depth Migration is
needed as the ray-path bending takes place due to strong velocity
contrasts. In such cases, Time Migration is not adequate.
Time or Depth Migration
Model
Time or Depth Migration ?
Time or Depth Migration ?
Time or Depth Migration ?
Model
Pre Stack Migration
Pre-Stack Migration is performed on unstacked data
It can be performed in both Time and Depth domains
Ability to correct for ray path distortions and uncommon
reflection points in areas of complex structure and dip
Better quality Migrated stack result
Retention of AVO and phase change information
Lower S/N ratio
Cost
34
Pre-stack or Post-stack Migration
?
Post-stack depth migration
Pre-stack depth migration
Problems with Post Stack Time Migration
Time migration followed by depth conversion (as shown earlier)
will not produce accurate imaging in the presence of steep
dips and severe lateral velocity variations
R
E
A
S
O
N
Poor quality imaging from time migration
R
E
A
S
O
N
Time migration equations describe wavefield
propogation through a horizontally layered medium
SOLUTION ?
36
Dip-Moveout Correction (DMO)
The CMP stack is an approximation to the zero-offset
case - based on NMO correction 2
t X= t0 + X / V
But for dipping layers, correction should be 2
t X = t 0 + X cos q / V
Or
t X = t 0 + X / V - X sin q / V
NMO term
DMO term
37
DMO - A Pre-Stack Partial Migration
DMO + Post Stack Migration = Full Pre Stack Migration
does both
corrects for reflector
dispersal
repositions reflection points
to true subsurface location
DMO allows post-stack time migration to become equivalent
to pre-stack time migration.
38
Migration of 3D Data
2D Migration
3D Migration
hyperboloid
hyperbola
Only energy reflected in the plane
of the section is correctly imaged
uses energy from both
in and out of the plane
of the section
3D Migration will have a higher resolution as it can move energy
from outside the plane back to its correct position
39
One pass versus Two pass
Migration of 3D data can be performed in one or two passes
One Pass
Two Pass
3D stack data in
3D stack data in
migrate inlines and
xlines simultaneously
migrate in inline
direction
3D migrated data
migrate in xline
direction
3D migrated data
40
One pass versus Two pass
Two Pass
One Pass
overmigration
increase in accuracy and cost
When limited velocity variation
Two pass is a good match for One pass 3D migration
41
One pass versus Two pass
Two Pass
One Pass
Simple to implement with
existing algorithms
variation in velocities more
accurately handled
efficient use of resources
both time and depth migration
are possible
allows for use of two
different algorithms
increased resources
allows extra QC step
restricted to time migration
tends to overmigrate for realistic
velocity models
use of omega-x migration
algorithm - each frequency
can be processed separately
42
Which Migration To Use?
Type
Pluses
Minuses
Pre
stack
Post
stack
Time
Migrated data is used to pick
velocity analysis.
Higher cost than post stack.
Low S/N.
High S/N ratio.
Lower cost than pre-stack
Assumptions in stack process
breakdown when dip and velocity
variation.
Good result if velocity and dip
variation not too complex - at an
affordable price
Algorithms do not take account of ray
bending - poor when large dip and
velocity variations.
Depth
Algorithms take account of ray
bending.
Requires very accurate velocity-depth
model. Time and cost increase.
2D
Two pass on 3D data allows for
use of different algorithms, extra
QC.
Only uses energy from plane of
section
3D
Uses energy from in and out of
plane of section.
Resource/cost issues
43
Which Migration To Use?
What does the choice of migration type depend on ?
type of data
2D or 3D
cost
pre or post stack, algorithm
data characteristics
S/N ratio, maximum dip, lateral velocity variation, algorithm
client objectives
zone of interest, resources, own ideas
44
Factors Affecting Migration
Some additional factors that affect migration quality
Noise - mainly coherent noise
Spatial Sampling - in the presence of steep dips, aliasing
can be a problem
Migration Aperture - related to horizontal displacement
of the reflection point
Amplitude Anomalies - spikes, noise bursts, truncated traces etc.
Example of aliasing
What velocities do we use for migration ?
Post-stack time migration
stacking velocity field may be smoothed and scaled
Pre-stack time migration
partial migration based on constant velocity (approx. 2000-3000m/s)
performed on common-offset gathers. The data is then
demigrated and velocities a re picked
Depth migration
requires highly accurate velocity-depth model
47
Summary
48
