AAPG

26 EXPLORER
Geophysical Corner April, 1998 Figure 1 – A shot
record showing
source, receivers,

Normal Moveout rays, reflection points

and midpoints. The
increase of travel time
with offset (source-

An Effect, Process receiver distance) is

the normal moveout
effect.

(Editor’s note: The Geophysical seismology in the 1910s. In those

Corner is a regular column in the days a seismic shot consisted of a
EXPLORER and is produced by the source (dynamite) sending waves into
AAPG Geophysical Integration the earth to bounce around and return
Committee. This month’s column is the to a few geophones. The data were
first of a two-part series on “Moveout: preserved as wiggly lines (traces) on
What Is It, and What’s Normal About a rotating drum of paper, or as dark
It?”) lines on a photographic record.
The human eye is wonderfully
By CHRISTOPHER L. LINER adept at seeing patterns and
Petroleum seismology is, and relationships in very confusing data,
always has been, changing very e.g. recognizing a face across a
quickly. crowded room full of strangers. Early
You might have heard whispers seismic records were like that – lots of
about exotic topics like crosswell noise, not much signal. But the signals
tomography, wavelet transforms, were there, and skilled interpreters
cluster analysis, texture segmentation could recognize them.
impedance inversion, geostatistical Some of these signal events came
estimation, etc. in straight lines across the traces;
So why in this high-tech age is others formed curves. But whatever
someone writing about something as the shape, each kind of signal showed
ancient as normal moveout? a delay from trace to trace as we
The answer involves the move away from the source – and thus
importance of understanding was known as moveout.
fundamental concepts, the natural
lead-in that normal moveout provides A Normal Example
to the juicier topic of dip moveout (to
be covered here next month), and a A shot record is the collection of
chance to do it without any equations. seismic traces generated when one Figure 2 – A land shot record from Alaska
Normal moveout has two meanings source shoots into many receivers, as (courtesy of SEG). Several kinds of events,
– it is both: shown in Figure 1. in addition to reflections, are marked on the
❒ A seismic effect. In this example, the upper black right side of the record, indicating the path
❒ A seismic processing step. and wave type from source to receiver.
By itself, the term “moveout” goes Graphics courtesy of Christopher Liner
back to the earliest days of reflection continued on next page

Advertisement
not available.

APRIL 1998
 AAPG

EXPLORER 27

Figure 3 – A seismic line is acquired by rolling the shot and receivers a certain Figure 4 – All the traces that live at the same CMP location will be processed
distance forward. This gives more than one ;trace with the same midpoint. together as a family. Ultimately, they will be added (stacked) to make one stack
trace that lives at this location. The process of Normal MoveOut (NMO) helps
prepare the traces before they are added together.

continued from previous page

line is the acquisition surface and the

lower one is the reflector. Dots below
the reflector show subsurface
reflection points. Halfway between the
source and a receiver is a point on the
ground called the midpoint. These are
shown as black dots above the
acquisition surface.
Where there is no dip, the midpoint
is directly above the reflection point.
As the offset (source to receiver
distance) increases, so does the
travel time from source to receiver.
This characteristic delay of
reflection times with increasing offset
is called normal moveout.
In Figure 2 (page 26), reflections
can be seen in real data along with
other kinds of events.
There are receivers on both sides
of the shot in this case. The right side
has been marked-up to identify
different kinds of events – direct
arrivals (p-wave, s-wave, air wave,
surface wave), head waves and (a
few) reflections. The left side is
uninterpreted.
The reflection events have a
hyperbolic shape characteristic of
normal moveout.

Making the Earth Flat

So now we know something about

normal moveout, the effect.
What about normal moveout, the
process?
For this we will use the acronym
NMO (Normal MoveOut) since for
most people this term implies the
process, not the effect.
What is NMO? The short answer is:
A seismic processing step whereby
reflection events are flattened in a
common midpoint gather in
preparation for stacking.
(If this makes sense then you can
move along to the next article.
Otherwise, read on.)
A seismic line is generated by
“rolling” the shot and receivers
forward a certain distance and firing
again. As shown in above Figure 3,
this generates a second shot record
which partially overlaps the first.
Note that six of the seven reflection
points from the blue shot were also
reflection points for the red shot. This
occurs by design and is called
Common MidPoint (CMP) shooting.
As the shots roll along there will be
many source-receiver pairs with the
same CMP location, and the CMP fold
is the total number of traces that live
at any given CMP.

See Geophysical Corner, page 28

APRIL 1998
 AAPG

28 EXPLORER
Geophysical
Corner
from previous page
CMP fold can vary from as few as
six (low-fold land 3-D) up to several
hundred (2-D marine).
The reason for gathering multifold
data is that we get redundant
information about the reflection point
down in the earth, and this
redundancy can be used to reduce
noise and create a more reliable
image. Our goal is to eventually
process all these traces as a family
and add them together (CMP stack) to
make one trace that lives at this CMP
location.
NMO is aimed at removing the
hyperbolic curvature in reflection
events. Basically, it is removing the
effect of offset. If this is done properly
then the reflection should come in at
the same time for all offsets (since we
have removed any travel time delay Figure 5 – Data from Figure 2 after Normal MoveOut Figure 6 – Data after NMO with a 25 percent stretch mute.
due to offset). processing (and air wave removal). The blue box highlights Note that the nasty events in the red box of Figure 5 have
In short, reflection events should static problems; the red box shows stretched direct arrivals been removed.
be flat after NMO. and head waves. Reflection events are flat (more or less).
Figure 5 shows the data after NMO
processing (and air wave removal). In
this case, we see the events are pretty fat (not flat) these events look after zeros. We could do this by hand, but a limit, then the data values are
well flattened by NMO, but there are a NMO. This is because NMO actually seismic line may contain many replaced by zeros in that part of the
couple of interesting areas. operates by stretching the trace – and thousands of CMP gathers. trace. Figure 6 shows the result of
The blue box shows some the shallower something is the more it It turns out we can let NMO itself allowing a 25 percent stretch in the
disturbing behavior along a flattened stretches. do the muting for us. The idea is to NMO process.
reflection. This has nothing at all to do Since our goal is to eventually keep track of how much stretch NMO
with NMO, but is related to lateral flatten all these traces and add them is doing to the trace. The stretch What About Dips?
changes in the near surface layers together to make one trace, keeping changes down the trace – biggest at
(termed a static problem). this kind of stuff would wipe out the top, smallest at the bottom. So the At this point more than one astute
The red box shows what NMO shallow reflections. It needs to go. idea of an NMO stretch mute is to set reader is saying, “Yeah, yeah, you
does to the direct arrivals. Since these We get rid of these events by a limit on how much stretch we are
were linear and not hyperbolic, NMO muting – which is nothing more than going to allow.
has not flattened them. Also, note how replacing the offending data with If the stretch gets bigger than our continued on next page

Advertisement
not available.

APRIL 1998
 AAPG

EXPLORER 29

Figure 7 – When the reflector is dipping, midpoints are not vertically above
reflection points. Compare Figure 4. NMO can handle this case, but breaks down
when multiple dips are present in the subsurface.

continued from previous page

academic types are all alike – ignoring

the real world. What about dipping
beds?”
Figure 7 gives a hint of things to
come.
The basic problem is that when the
bed dips, midpoints are not vertically
above reflection points. Furthermore,
the reflection points become unevenly
spaced along the reflector.
The fact is that NMO has a
constant dip assumption built in. If
every bed were dipping at the same
angle, say 23 degrees, it turns out we
could do NMO just fine. The real
problem comes where there are many
dips in the subsurface. In this case,
NMO acts like a dip filter –
preferentially passing some dip the
processor chooses while suppressing
others. Advertisement
This was particularly notorious in
places like the Gulf of Mexico, where
not available.
gently dipping beds often meet steep
salt domes. You could do NMO in
such a way as to get a good image of
one or the other, but not both.
This was the situation up until about
1981 or 1982. Then something new
came on the scene. It was called –
you guessed it – dip moveout.
That is our next article.

(Editor’s note: Liner is associate

professor of geosciences at the
University of Tulsa. He may be
contacted via e-mail at [email protected],
or
http://douze.utulsa.edu/~cll/ChrisLiner.
html)

Advertisement
not available.

APRIL 1998
 AAPG

64 EXPLORER
Geophysical
Geophysical Corner
Corner May, 1998

Dip MoveOut Just Isn’t ‘Normal’

(Editor’s note: The Geophysical orig offset S/R zero offset This new trace is a zero-offset
Corner is a regular column in the S R trace, and adding (stacking) all
trace NMO trace
EXPLORER and is produced by the such traces that live at this midpoint
AAPG Geophysical Integration yields a stack trace.
Committee. This month’s column is x
All the stack traces plotted side-
the second of a two-part series on by-side form the stack section,
“DMO – The Other MoveOut.”) which is raw material for post-stack
z migration processing.
Figure 1 illustrates the NMO idea.
By CHRISTOPHER L. LINER On the left is a field trace with
You can talk about Normal Earth model some arbitrary offset and one
MoveOut (NMO) all day long without reflection event. NMO assumes the
mentioning migration – but Dip reflection comes from a horizontal
MoveOut (DMO) is another matter.
NMO interface as shown in the middle
In fact, DMO started out with the figure.
cumbersome-but-descriptive name time time This is an important and
“pre-stack partial migration.” That restrictive assumption.
was in 1979, but at least two years The NMO correction adjusts
Figure 1. Normal MoveOut (NMO) is a process applied to pre-stack data. Here the
earlier there was a DMO processing observed travel time (red path) to
effect is shown on a single trace with one reflection event (left). NMO assumes the
product on the market. It was midpoint zero-offset travel time
reflection comes from a horizontal interface in the earth (center). Using a velocity
named DEVILISH, an acronym for (blue path). So after NMO, the event
function supplied by the processor, NMO adjusts the original time (red) to that which
“dipping event velocity inequality is moved up in time. Technically, we
would have been observed at the midpoint, marked S/R. The blue path is two-way
licked.” are changing time coordinate from
time down and back, which must be less than the red path time. So NMO’s job is to
You think I am making this up, raw time, t, to NMO time, tn.
move the reflection event up the trace (right). Note NMO operates on one trace at a
but it’s true.
time, which makes it inexpensive.
Before moving on, I should say * * *
something about name-dropping. Graphics courtesy of Christopher L. Liner
Unlike NMO, the development of But what if the interface is not
DMO is very recent. The people what DMO is and does, not create a This is pre-stack data. horizontal?
involved are still alive and kicking. historical who’s who. From last month’s article, we It’s tempting to think that the
We owe a debt of gratitude for know that NMO lives to remove reflector could be anywhere and still
their hard work and ingenuity, but if * * * offset from pre-stack data. In the be consistent with the observed
I mention one name I will have to computer, we adjust this trace to travel time. But this is not the case.
mention them all. So we will deal Now, back to business. simulate one that would have been Let’s say the original travel time
here with the concepts and not the In the field, a source and receiver recorded with no effect at a point
names. are located, say, 3,000 meters apart half-way between the source and
We will attempt to understand (the offset) and a trace is recorded. receiver (the midpoint). continued on next page

Advertisement not
available.

MAY 1998
 AAPG

EXPLORER 65
Figure 2. Dip MoveOut (DMO) is a
continued from previous page process that is applied after NMO. S S/R S/R S/R S/R S/R S/R S/R R
Since NMO assumes the reflection
comes from a horizontal bed, it is
picking up only one of many
possibilities. For one trace with one
is 1.0 second, and we know the reflection event, all possible travel x
velocity is 3,000 m/s. The total paths have the same length – that is,
distance traveled has got to be the distance from source to reflection z
3,000 meters. So all valid reflector point to receiver is a constant. The
positions have one thing in geometrical shape with this property zero offset zero offset
common: The total distance from is an ellipse (upper). Some of the NMO trace DMO traces
source to reflection point to receiver original path possibilities are shown
is constant, namely 3,000 meters. in red. NMO reduces travel time
Thinking back to Geometry 101, based on a horizontal reflector (blue
this is just the definition of an ellipse path), while DMO does all the other
with the source and receiver at each cases (green). So the action of DMO
focus. (lower) is to take the NMO’d event
Figure 2 shows such an ellipse. (blue) and broadcast it across DMO
Remember the goal is to remove several nearby traces (green). Since
offset and thus create a zero-offset DMO operates on several traces, it is
section. NMO gives one of many expensive. time time
possibilities, DMO gives all the rest.
In Figure 2, some of the possible
original travel paths are shown
(red), along with the single travel
path after NMO (blue) and the many
travel paths after DMO (green).
From Figure 1 we saw that NMO
is a process that takes one trace in
and gives one trace out. DMO is
different. One trace into DMO
generates many traces out – all of
which live between the original
source and receiver locations. This
is illustrated in figure 2 (lower) and
figure 3 (page 66).

* * *

Let’s consider figure 3.

We have a panel of data
containing only two spikes of
amplitude on one trace (left panel).
The other traces are there, but
empty.
NMO moves the spike up on the
same trace (middle panel). DMO
then throws the spike out along a
curve, which lives between the
source and receiver (right panel).
This is sometimes called the DMO
smile. Since it comes from a spike
or impulse on the input data, it is
also called the DMO impulse
response.
Now here is some magic. By
creating the DMO smile, all possible
dips are handled simultaneously. We
do not need to know what the dip is
in the earth – by processing all
traces with DMO the actual
reflections will emerge because they
are tangent, at some point, to DMO
smiles.
Also, DMO does not depend on
the velocity, so long as the velocity
is constant.
While not strictly true, this is one
of the things that got everyone
initially excited about DMO. And its
weak dependence on velocity is one
reason it is still so widely used.
Anyway, it is unlike NMO and
migration, which always need
velocity information, and are quite
sensitive to it. Another big selling
point is that NMO+DMO, rather than
NMO alone, passes all dips into the
stack section.
This gives more raw material for
migration to work with in creating a
final migrated image.
Since DMO spreads things out
across traces, it is much more
expensive than NMO, which only
shifts things up on one trace. Even
so, NMO+DMO is still cheaper than
pre-stack migration.
Figure 4 (page 66) gives some
representative run times for these,
and other, processes.

See DMO, next page

MAY 1998
 AAPG

66 EXPLORER
DMO
from previous page

* * *

So here is the bottom line. If

structure and velocity variation are
not too nasty in an area, we can get
away with a traditional processing
sequence:
NMO + DMO + Stack +
PostStackMigration.
In this equation, “+” means
“followed by.”
However, if things get really tough
down there (e.g., subsalt), this
sequence breaks down and fails to
give a good image. In this case we Figure 3. This numerical example illustrates the effect of NMO and DMO on a trace with two spikes, or reflection events. The
are compelled to do one grand spikes are shown on the left surrounded by a bunch of zero traces. The source and receiver location are denoted by S and R,
process called pre-stack migration. respectively. NMO shifts the spikes up in time, but only on the same trace (middle). DMO throws the NMO’d spike amplitude out
In fact, DMO was originally along a curve to handle all possible dips. This curve is called the DMO smile, or DMO ellipse, or DMO impulse response. Notice
invented to complete the following the DMO smile only lives between the original source and receiver positions.
equality under mild subsurface
conditions: Figure 4. This chart illustrates
PreStackMigration = NMO + ? + prestack mig 5104 processing time in seconds for a small
Stack + PostStackMigration. sample data set. The relative times are
poststack mig 18 important, not the size of the data set
The “?” turned out to be DMO – stack 2 or kind of machine used. Starting from
and is used world-wide every day. the bottom, the processes represent
dmo 2845 (in order) a more-or-less standard
processing flow. By far the most
nmo 22 expensive item here is DMO. But if we
(Editor’s note: Liner is associate run pre-stack migration instead, it is
decon 20
professor of geosciences at the even more expensive. Basically, the
University of Tulsa. He may be sort cos 9 pre-stack migration replaces NMO,
contacted via e-mail at DMO, stack and post-stack migration.
[email protected], or sort cmp 8 Due to the cost, pre-stack migration is
http://douze.utulsa.edu/~cll/ChrisLin a method to be used only when
er.html) filter 30 required by strong lateral velocity
variations and/or extreme structure.
gain 28

Advertisement not
available.

MAY 1998